pax_global_header00006660000000000000000000000064130371274720014520gustar00rootroot0000000000000052 comment=45eb058b12ab18ad4b38982a2443b9a83c7f2dce kafbox-2.0/000077500000000000000000000000001303712747200126335ustar00rootroot00000000000000kafbox-2.0/.gitignore000066400000000000000000000001041303712747200146160ustar00rootroot00000000000000################# ## Matlab ################# *~ *.asv *.bak *.mat kafbox-2.0/LICENSE.txt000066400000000000000000000025261303712747200144630ustar00rootroot00000000000000Copyright (c) 2012-2017, Steven Van Vaerenbergh. All rights reserved. The code is released under the FreeBSD License. Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are met: 1. Redistributions of source code must retain the above copyright notice, this list of conditions and the following disclaimer. 2. Redistributions in binary form must reproduce the above copyright notice, this list of conditions and the following disclaimer in the documentation and/or other materials provided with the distribution. THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. kafbox-2.0/README.md000066400000000000000000000257751303712747200141320ustar00rootroot00000000000000Kernel Adaptive Filtering Toolbox === A Matlab benchmarking toolbox for kernel adaptive filtering. Kernel adaptive filters are online machine learning algorithms based on kernel methods. Typical applications include time-series prediction, nonlinear adaptive filtering, tracking and online learning for nonlinear regression. This toolbox includes algorithms, demos, and tools to compare their performance.

Nonlinear system identification using kernel adaptive filtering.

Maintainer: [Steven Van Vaerenbergh](http://gtas.unican.es/people/steven), University of Cantabria, Spain. Contributors: - [Miguel Lazaro-Gredilla](http://www.tsc.uc3m.es/~miguel), Universidad Carlos III de Madrid, Spain. - [Sohan Seth](http://www.sohanseth.com/), University of Edinburgh, UK. - [Masahiro Yukawa](http://www.ykw.elec.keio.ac.jp/yukawa/), Keio University, Japan. - [Masa-aki Takizawa](http://www.ykw.elec.keio.ac.jp/member.html), Keio University, Japan. - [Osamu Toda](http://www.ykw.elec.keio.ac.jp/member.html), Keio University, Japan. - [Dominik Rzepka](http://home.agh.edu.pl/~drzepka/), AGH University of Science and Technology, Poland. - [Pantelis Bouboulis](http://bouboulis.mysch.gr/), University of Athens, Greece. - [Stephan Rhode](https://www.fast.kit.edu/lff/1011_3295.php), Karlsruher Institut für Technologie, Germany. Official web: https://github.com/steven2358/kafbox/ This toolbox is a collaborative effort: every developer wishing to contribute code or suggestions can do so. More info below. Sample run --- Figure generated by `demo/demo_sample_run.m`.

Online learning of a 1D nonlinear regression by a kernel adaptive filter on a fixed budget.

Directories included in the toolbox --- `data/` - data sets `demo/` - demos and test files `doc/` - documentation `lib/` - algorithm libraries and utilities Setup --- 1. Run `install.m` to add the toolbox folders to the path. 2. Type `savepath` to save the changes to the path. Octave / Matlab pre-2008a ------------------------- This toolbox uses the `classdef` command which is not supported in Matlab pre-2008a and not yet in Octave. The older 0.x versions of this toolbox do not use `classdef` and can therefore be used with all versions of Matlab and Octave. http://sourceforge.net/projects/kafbox/files/ Usage --- Each kernel adaptive filtering algorithm is implemented as a Matlab class. To use one, first define its options: ```matlab options = struct('nu',1E-4,'kerneltype','gauss','kernelpar',32); ``` Next, create an instance of the filter. E.g., for an instance of the original KRLS algorithm run: ```matlab kaf = krls(options); ``` One iteration of training is performed by feeding one input-output data pair to the filter: ```matlab kaf.train(x,y); ``` The outputs for one or more test inputs are evaluated as follows: ```matlab Y_test = kaf.evaluate(X_test); ``` Example: time-series prediction --- Code from `demo/demo_prediction.m` ```matlab % Demo: 1-step ahead prediction on Lorenz attractor time-series data [X,Y] = kafbox_data(struct('file','lorenz.dat','embedding',6)); % make a kernel adaptive filter object of class krls with options: % ALD threshold 1E-4, Gaussian kernel, and kernel width 32 kaf = krls(struct('nu',1E-4,'kerneltype','gauss','kernelpar',32)); %% RUN ALGORITHM N = size(X,1); Y_est = zeros(N,1); for i=1:N, if ~mod(i,floor(N/10)), fprintf('.'); end % progress indicator, 10 dots Y_est(i) = kaf.evaluate(X(i,:)); % predict the next output kaf.train(X(i,:),Y(i)); % train with one input-output pair end fprintf('\n'); SE = (Y-Y_est).^2; % test error %% OUTPUT fprintf('MSE after first 1000 samples: %.2fdB\n\n',10*log10(mean(SE(1001:end)))); ``` Result: MSE after first 1000 samples: -40.17dB Citing KAFBOX --- If you use this toolbox in your research please cite ["A Comparative Study of Kernel Adaptive Filtering Algorithms"](http://gtas.unican.es/pub/356): @conference{vanvaerenbergh2013comparative, author = {Van Vaerenbergh, Steven and Santamar{\'\i}a, Ignacio}, booktitle = {2013 IEEE Digital Signal Processing (DSP) Workshop and IEEE Signal Processing Education (SPE)}, title = {A Comparative Study of Kernel Adaptive Filtering Algorithms}, doi = {10.1109/DSP-SPE.2013.6642587}, year = {2013}, note = {Software available at \url{https://github.com/steven2358/kafbox/}} } Included algorithms --- - Kernel Recursive Least-Squares (KRLS) algorithm with approximate linear dependency criterion, as proposed in Y. Engel, S. Mannor, and R. Meir. "The kernel recursive least-squares algorithm", IEEE Transactions on Signal Processing, volume 52, no. 8, pages 2275-2285, 2004. - Sliding-Window Kernel Recursive Least-Squares (SW-KRLS), as proposed in S. Van Vaerenbergh, J. Via, and I. Santamaria. "A sliding-window kernel RLS algorithm and its application to nonlinear channel identification", 2006 IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP), Toulouse, France, 2006. - Naive Online Regularized Risk Minimization Algorithm (NORMA), as proposed in J. Kivinen, A. Smola and C. Williamson. "Online Learning with Kernels", IEEE Transactions on Signal Processing, volume 52, no. 8, pages 2165-2176, 2004. - Kernel Least-Mean-Square (KLMS), as proposed in W. Liu, P.P. Pokharel, and J.C. Principe, "The Kernel Least-Mean-Square Algorithm," IEEE Transactions on Signal Processing, vol.56, no.2, pp.543-554, Feb. 2008. - Fixed-Budget Kernel Recursive Least-Squares (FB-KRLS), as proposed in S. Van Vaerenbergh, I. Santamaria, W. Liu and J. C. Principe, "Fixed-Budget Kernel Recursive Least-Squares", 2010 IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP 2010), Dallas, Texas, U.S.A., March 2010. - Kernel Recursive Least-Squares Tracker (KRLS-T), as proposed in S. Van Vaerenbergh, M. Lazaro-Gredilla, and I. Santamaria, "Kernel Recursive Least-Squares Tracker for Time-Varying Regression," IEEE Transactions on Neural Networks and Learning Systems, vol.23, no.8, pp.1313-1326, Aug. 2012. - Quantized Kernel Least Mean Squares (QKLMS), as proposed in Chen B., Zhao S., Zhu P., Principe J.C., "Quantized Kernel Least Mean Square Algorithm," IEEE Transactions on Neural Networks and Learning Systems, vol.23, no.1, Jan. 2012, pages 22-32. - Random Fourier Feature Kernel Least Mean Square (RFF-KLMS) algorithm, as proposed in Abhishek Singh, Narendra Ahuja and Pierre Moulin, "Online learning with kernels: Overcoming the growing sum problem," 2012 IEEE International Workshop on Machine Learning for Signal Processing (MLSP), Sept. 2012. - Extended Kernel Recursive Least Squares (EX-KRLS), as proposed in W. Liu and I. Park and Y. Wang and J.C. Principe, "Extended kernel recursive least squares algorithm", IEEE Transactions on Signal Processing, volume 57, number 10, pp. 3801-3814, oct. 2009. - Gaussian-Process based estimation of the parameters of KRLS-T, as proposed in Steven Van Vaerenbergh, Ignacio Santamaria, and Miguel Lazaro-Gredilla, "Estimation of the forgetting factor in kernel recursive least squares," 2012 IEEE International Workshop on Machine Learning for Signal Processing (MLSP), Sept. 2012. - Kernel Affine Projection (KAP) algorithm with Coherence Criterion, as proposed in C. Richard, J.C.M. Bermudez, P. Honeine, "Online Prediction of Time Series Data With Kernels," IEEE Transactions on Signal Processing, vol.57, no.3, pp.1058,1067, March 2009. - Kernel Normalized Least-Mean-Square (KNLMS) algorithm with Coherence Criterion, as proposed in C. Richard, J.C.M. Bermudez, P. Honeine, "Online Prediction of Time Series Data With Kernels," IEEE Transactions on Signal Processing, vol.57, no.3, pp.1058,1067, March 2009. - Recursive Least-Squares algorithm with exponential weighting (RLS), as described in S. Haykin, "Adaptive Filtering Theory (3rd Ed.)", Prentice Hall, Chapter 13. - Multikernel Normalized Least Mean Square algorithm with Coherence-based Sparsification (MKNLMS-CS), as proposed in M. Yukawa, "Multikernel Adaptive Filtering", IEEE Transactions on Signal Processing, vol.60, no.9, pp.4672-4682, Sept. 2012. - Parallel HYperslab Projection along Affine SubSpace (PHYPASS) algorithm, as proposed in M. Takizawa and M. Yukawa, "An Efficient Data-Reusing Kernel Adaptive Filtering Algorithm Based on Parallel Hyperslab Projection Along Affine Subspace," 2013 IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP), pp.3557-3561, May 2013. - Fixed-budget kernel least mean squares (FB-KLMS) algorithm, as proposed in D. Rzepka, "Fixed-budget kernel least mean squares," 2012 IEEE 17th Conference on Emerging Technologies & Factory Automation (ETFA), Krakow, Poland, Sept. 2012. - Leaky Kernel Affine Projection Algorithm (LKAPA, including KAPA-1 and KAPA-3) and Normalized Leaky Kernel Affine Projection Algorithm (NLKAPA, including KAPA-2 and KAPA-4), as proposed in W. Liu and J.C. Principe, "Kernel Affine Projection Algorithms", EURASIP Journal on Advances in Signal Processing, Volume 2008, Article ID 784292, 12 pages. - Kernel Affine Projection Subgradient Method (KAPSM), as proposed in K. Slavakis, S. Theodoridis, and I. Yamada, "Online kernel-based classification using adaptive projection algorithms," IEEE Transactions on Signal Processing, Vol. 56, No. 7, pp. 2781-2796, 2008. - Kernel Least Mean Squares algorithm with Coherence-Sparsification criterion and L1-norm regularization (KLMS-CSL1) and with active L1-norm regularization (KLMS-CSAL1), as proposed in Wei Gao, Jie Chen, Cédric Richard, Jianguo Huang, and Rémi Flamary, "Kernel LMS algorithm with forward-backward splitting for dictionary learning," 2013 IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP 2013), Vancouver, Canada, March 2013. - Mixture Kernel Least Mean Square (MXKLMS) algorithm, as proposed in R. Pokharel, S. Seth, and J.C. Principe, "Mixture kernel least mean square," The 2013 International Joint Conference on Neural Networks (IJCNN), pp.1-7, 4-9 Aug. 2013. - Fixed budget quantized kernel least-mean-square algorithm (QKLMS-FB), as proposed in S. Zhao, B. Chen, P. Zhu, J. C. Príncipe, "Fixed budget quantized kernel least-mean-square algorithm", Signal Processing, Volume 93, Issue 9, September 2013, Pages 2759-2770. - Probabilistic Least-Mean-Squares (ProbLMS) algorithm, as proposed in J. Fernandez-Bes, V. Elvira, and S. Van Vaerenbergh, "A probabilistic least-mean-squares filter," 2015 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), Brisbane, Australia, 2015. - Discrete-time Kalman filter, as proposed in R. E. Kalman, "A New Approach to Linear Filtering and Prediction Problems", ASME Journal of Basic Engineering, 1960, 82(1), pp. 35-45. How to contribute code to the toolbox --- Option 1: email it to steven.vanvaerenbergh@unican.es Option 2: [fork the toolbox on GitHub](https://github.com/steven2358/kafbox), push your changes, then send a pull request. License --- This source code is released under the FreeBSD License. kafbox-2.0/data/000077500000000000000000000000001303712747200135445ustar00rootroot00000000000000kafbox-2.0/data/generate_channel_switch.m000066400000000000000000000047401303712747200205720ustar00rootroot00000000000000function [x_embed,y,y_ref,x_test_embed,y_test_ref,H] = ... generate_channel_switch(opt) % Generate CHANNEL_SWITCH data set: input-output data of a nonlinear % channel whose linear part is changed abruptly at a chosen point. The % nonlinear channel is a Wiener system with a randomly chosen linear part. % % Input options: "opt" is a structure with the following fields: % - N: total number of training data points % - N_switch: iteration after which the channel switch occurs % - N_test: number of test data points (before and after switch) % - sigpower: input signal power % - chlen: linear channel length % - fun: nonlinear function % - SNR: signal-to-noise ratio of additve output noise % % Outputs: % - x_embed: system input with time embedding (each datum is a row) % - y: system output % - y_ref: noiseless system output % - x_test_embed: test system input with time embedding % - y_test_ref: noiseless test system output, one column per channel % - H: channel matrix (each row is a channel impulse response) % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ %% DEFAULT PARAMETER VALUES options = struct('N',1500,'N_test',500,'N_switch',500,'chlen',5,... 'sigpower',1,'fun','tanh(x)','SNR',20); %% CUSTOM PARAMETER VALUES if nargin >= 1, for opt_name = fieldnames(opt)', if strmatch(opt_name,fieldnames(options),'exact'), options.(opt_name{1}) = opt.(opt_name{1}); end end end N = options.N; N_test = options.N_test; N_switch = options.N_switch; sigpower = options.sigpower; chlen = options.chlen; fun = options.fun; SNR = options.SNR; %% PROGRAM H = [ones(2,1) randn(2,chlen-1)]; % linear channel coefficients f = inline(fun); % Wiener system nonlinearity N_all = N+N_test+chlen-1; x_all = sqrt(sigpower)*randn(N_all,1); x_all_embed = zeros(N_all,chlen); for i = 1:chlen, x_all_embed(i:N_all,i) = x_all(1:N_all-i+1); % time-embedding end x_all_embed = x_all_embed(chlen:N_all,:); x_embed = x_all_embed(1:N,:); x_test_embed = x_all_embed(N+1:N+N_test,:); % linear filtering xp1 = x_embed(1:N_switch,:)*H(1,:)'; xp2 = x_embed(N_switch+1:N,:)*H(2,:)'; % apply nonlinearity y_ref = f([xp1;xp2]); % no noise yet % add noise noisevar = 10^(-SNR/10)*var(y_ref); noise = sqrt(noisevar)*randn(N,1); y = y_ref + noise; % get test outputs: 2 columns, no noise xp3 = x_test_embed*H(1,:)'; xp4 = x_test_embed*H(2,:)'; y_test_ref = f([xp3 xp4]); % figure;plot([xp1;xp2],y,'.') kafbox-2.0/data/kafbox_data.m000066400000000000000000000026751303712747200161770ustar00rootroot00000000000000% KAFBOX_DATA Data handler. Returns input-output data specified in the % options. % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ function [X_train,Y_train,X_test,Y_test] = kafbox_data(data_options) if isfield(data_options,'class') data_name = lower(data_options.class); elseif isfield(data_options,'name') data_name = lower(data_options.name); else error('Unknown data.'); end dataset_handle = str2func(['kafbox_data_' data_name]); [X,Y,X_test,Y_test] = feval(dataset_handle, data_options); % time embedding L = 0; if isfield(data_options,'embedding') L = data_options.embedding; X = time_embedding(X,L); end % apply offset if isfield(data_options,'offset') X = X(1+data_options.offset:end,:); Y = Y(1+data_options.offset:end); end % crop training data N = length(Y); if isfield(data_options,'N') N = min(data_options.N,N); end X_train = X(1:N,:); Y_train = Y(1:N); if isfield(data_options,'n_train') && isfield(data_options,'n_test') n_train = data_options.n_train; n_test = data_options.n_test; X_train = X(L:n_train+L-1,:); Y_train = Y(L:n_train+L-1); X_test = X(n_train+L:n_train+L+n_test-1,:); Y_test = Y(n_train+L:n_train+L+n_test-1); end function X_embedded = time_embedding(X,L) N = size(X,1); X_embedded = zeros(N,L); % L-1 first vectors are zero padded for i = 1:L, X_embedded(i:N,i) = X(1:N-i+1,1); % time embedding end kafbox-2.0/data/kafbox_data_channel_switch.m000066400000000000000000000053611303712747200212430ustar00rootroot00000000000000function [x_embed,y,x_test_embed,y_test_ref,H] = ... kafbox_data_channel_switch(opt) % KAFBOX_DATA_CHANNEL_SWITCH Data generator for channel switch data % % CHANNEL_SWITCH data set: input-output data of a nonlinear % channel whose linear part is changed abruptly at a chosen point. The % nonlinear channel is a Wiener system with a randomly chosen linear part. % % Input options: "opt" is a structure with the following fields: % - N: total number of training data points % - N_switch: iteration after which the channel switch occurs % - N_test: number of test data points (before and after switch) % - sigpower: input signal power % - chlen: linear channel length % - fun: nonlinear function % - SNR: signal-to-noise ratio of additve output noise % - H: the channels to be used (each row is a channel impulse response). % If not provided these channels are generated randomly. % % Outputs: % - x_embed: system input with time embedding (each datum is a row) % - y: system output % - y_ref: noiseless system output % - x_test_embed: test system input with time embedding % - y_test_ref: noiseless test system output, one column per channel % - H: channel matrix (each row is a channel impulse response) % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ %% DEFAULT PARAMETER VALUES options = struct('N',1500,'N_test',500,'N_switch',500,'chlen',5,... 'sigpower',1,'fun','tanh(x)','SNR',20);%,'H',0); %% CUSTOM PARAMETER VALUES if nargin >= 1, for opt_name = fieldnames(opt)', if ismember(opt_name,fieldnames(options)), options.(opt_name{1}) = opt.(opt_name{1}); end end end N = options.N; N_test = options.N_test; N_switch = options.N_switch; sigpower = options.sigpower; chlen = options.chlen; fun = options.fun; SNR = options.SNR; % H = options.H; %% PROGRAM % if numel(H)==1 % if assert(H,0) H = [ones(2,1) randn(2,chlen-1)]; % linear channel coefficients % end % end f = inline(fun); % Wiener system nonlinearity N_all = N+N_test+chlen-1; x_all = sqrt(sigpower)*randn(N_all,1); x_all_embed = zeros(N_all,chlen); for i = 1:chlen, x_all_embed(i:N_all,i) = x_all(1:N_all-i+1); % time-embedding end x_all_embed = x_all_embed(chlen:N_all,:); x_embed = x_all_embed(1:N,:); x_test_embed = x_all_embed(N+1:N+N_test,:); % linear filtering xp1 = x_embed(1:N_switch,:)*H(1,:)'; xp2 = x_embed(N_switch+1:N,:)*H(2,:)'; % apply nonlinearity y_ref = f([xp1;xp2]); % no noise yet % add noise noisevar = 10^(-SNR/10)*var(y_ref); noise = sqrt(noisevar)*randn(N,1); y = y_ref + noise; % get test outputs: 2 columns, no noise xp3 = x_test_embed*H(1,:)'; xp4 = x_test_embed*H(2,:)'; y_test_ref = f([xp3 xp4]); % figure;plot([xp1;xp2],y,'.') kafbox-2.0/data/kafbox_data_lorenz.m000066400000000000000000000005611303712747200175600ustar00rootroot00000000000000% KAFBOX_DATA_LORENZ Data loader for Lorenz data set function [X,Y,X_test,Y_test] = kafbox_data_lorenz(data_opts) data = load('lorenz.dat'); % prediction horizon horizon = 1; if isfield(data_opts,'horizon') horizon = data_opts.horizon; end % construct signal X = data(1:end-horizon); % input Y = data(1+horizon:end); % desired output X_test = []; Y_test = []; kafbox-2.0/data/kafbox_data_mg30.m000066400000000000000000000005641303712747200170200ustar00rootroot00000000000000% KAFBOX_DATA_MG30 Data loader for Mackey-Glass 30 data set function [X,Y,X_test,Y_test] = kafbox_data_mg30(data_opts) data = load('mg30.dat'); % prediction horizon horizon = 1; if isfield(data_opts,'horizon') horizon = data_opts.horizon; end % construct signal X = data(1:end-horizon); % input Y = data(1+horizon:end); % desired output X_test = []; Y_test = []; kafbox-2.0/data/lorenz.dat000066400000000000000000003027161303712747200155600ustar00rootroot0000000000000035.000000 34.766667 34.443866 34.040837 33.568386 33.038044 32.461343 31.849241 31.211716 30.557528 29.894122 29.227642 28.563023 27.904135 27.253942 26.614667 25.987949 25.374974 24.776596 24.193432 23.625936 23.074461 22.539306 22.020750 21.519082 21.034626 20.567755 20.118914 19.688631 19.277539 18.886391 18.516081 18.167662 17.842374 17.541667 17.267234 17.021041 16.805365 16.622828 16.476443 16.369649 16.306356 16.290981 16.328473 16.424340 16.584639 16.815953 17.125323 17.520127 18.007893 18.596022 19.291405 20.099918 21.025775 22.070741 23.233226 24.507284 25.881611 27.338641 28.853921 30.395928 31.926560 33.402451 34.777196 36.004420 37.041438 37.853054 38.414922 38.715874 38.758752 38.559574 38.145186 37.549863 36.811496 35.968001 35.054434 34.101069 33.132443 32.167246 31.218785 30.295801 29.403408 28.544015 27.718134 26.925046 26.163301 25.431085 24.726468 24.047563 23.392612 22.760036 22.148448 21.556646 20.983604 20.428448 19.890442 19.368966 18.863505 18.373630 17.898994 17.439325 16.994421 16.564149 16.148447 15.747331 15.360898 14.989335 14.632934 14.292103 13.967386 13.659482 13.369274 13.097856 12.846569 12.617045 12.411252 12.231553 12.080770 11.962257 11.879982 11.838623 11.843663 11.901496 12.019536 12.206317 12.471572 12.826294 13.282725 13.854270 14.555289 15.400701 16.405363 17.583145 18.945630 20.500403 22.248915 24.183977 26.287073 28.525809 30.852017 33.201165 35.493812 37.639723 39.544863 41.120823 42.295412 43.022387 43.288054 43.112947 42.547960 41.665848 40.550242 39.284741 37.944247 36.589662 35.265966 34.002969 32.817670 31.717271 30.702151 29.768404 28.909800 28.119171 27.389328 26.713597 26.086103 25.501876 24.956856 24.447835 23.972372 23.528703 23.115653 22.732554 22.379182 22.055703 21.762630 21.500791 21.271305 21.075562 20.915211 20.792142 20.708475 20.666538 20.668843 20.718054 20.816933 20.968284 21.174863 21.439273 21.763831 22.150399 22.600195 23.113566 23.689740 24.326557 25.020208 25.764976 26.553035 27.374319 28.216496 29.065096 29.903807 30.714962 31.480210 32.181348 32.801257 33.324853 33.739973 34.038079 34.214709 34.269609 34.206536 34.032767 33.758380 33.395404 32.956935 32.456322 31.906479 31.319363 30.705641 30.074516 29.433699 28.789473 28.146838 27.509683 26.880975 26.262934 25.657207 25.064998 24.487198 23.924474 23.377345 22.846245 22.331566 21.833695 21.353041 20.890060 20.445273 20.019287 19.612811 19.226675 18.861852 18.519476 18.200870 17.907569 17.641351 17.404265 17.198670 17.027268 16.893137 16.799775 16.751128 16.751617 16.806167 16.920199 17.099623 17.350787 17.680385 18.095317 18.602472 19.208426 19.919033 20.738912 21.670792 22.714756 23.867383 25.120847 26.462054 27.871943 29.325088 30.789787 32.228801 33.600846 34.862887 35.973085 36.894137 37.596556 38.061394 38.281920 38.263917 38.024560 37.590062 36.992546 36.266700 35.446722 34.563940 33.645267 32.712485 31.782215 30.866364 29.972847 29.106402 28.269388 27.462480 26.685241 25.936560 25.214974 24.518880 23.846685 23.196881 22.568094 21.959098 21.368814 20.796299 20.240735 19.701407 19.177696 18.669056 18.175010 17.695135 17.229061 16.776459 16.337043 15.910566 15.496821 15.095640 14.706900 14.330523 13.966485 13.614819 13.275626 12.949086 12.635465 12.335138 12.048602 11.776498 11.519637 11.279035 11.055941 10.851891 10.668748 10.508772 10.374685 10.269750 10.197872 10.163699 10.172742 10.231508 10.347639 10.530055 10.789095 11.136630 11.586145 12.152737 12.853005 13.704757 14.726473 15.936429 17.351376 18.984693 20.843946 22.927842 25.222734 27.698976 30.307721 32.979005 35.622179 38.129724 40.385152 42.274785 43.702032 44.601463 44.949268 44.767076 44.117737 43.094035 41.803412 40.352665 38.835936 37.327703 35.880720 34.527677 33.284936 32.156915 31.140175 30.226735 29.406501 28.668889 28.003823 27.402262 26.856431 26.359861 25.907312 25.494654 25.118724 24.777183 24.468384 24.191258 23.945216 23.730063 23.545931 23.393215 23.272528 23.184645 23.130469 23.110981 23.127199 23.180126 23.270699 23.399728 23.567829 23.775349 24.022284 24.308192 24.632099 24.992407 25.386807 25.812195 26.264615 26.739217 27.230253 27.731116 28.234420 28.732144 29.215815 29.676757 30.106365 30.496413 30.839372 31.128703 31.359117 31.526776 31.629415 31.666374 31.638553 31.548274 31.399089 31.195535 30.942870 30.646805 30.313257 29.948129 29.557136 29.145674 28.718730 28.280834 27.836048 27.387971 26.939774 26.494239 26.053810 25.620643 25.196660 24.783594 24.383038 23.996484 23.625359 23.271059 22.934981 22.618542 22.323210 22.050520 21.802095 21.579662 21.385064 21.220273 21.087399 20.988689 20.926530 20.903432 20.922019 20.984991 21.095083 21.255004 21.467359 21.734543 22.058616 22.441145 22.883028 23.384284 23.943833 24.559255 25.226561 25.939984 26.691812 27.472297 28.269674 29.070299 29.858954 30.619308 31.334539 31.988084 32.564467 33.050134 33.434211 33.709109 33.870891 33.919378 33.857961 33.693174 33.434063 33.091449 32.677160 32.203316 31.681717 31.123383 30.538232 29.934921 29.320792 28.701920 28.083221 27.468598 26.861100 26.263077 25.676328 25.102230 24.541848 23.996023 23.465447 22.950718 22.452392 21.971014 21.507152 21.061421 20.634504 20.227177 19.840325 19.474963 19.132261 18.813563 18.520417 18.254595 18.018129 17.813335 17.642848 17.509653 17.417111 17.368992 17.369490 17.423232 17.535279 17.711091 17.956475 18.277486 18.680275 19.170883 19.754938 20.437283 21.221490 22.109277 23.099834 24.189079 25.368901 26.626451 27.943602 29.296688 30.656674 31.989880 33.259338 34.426795 35.455237 36.311706 36.970041 37.413148 37.634408 37.637964 37.437849 37.056091 36.520161 35.860185 35.106351 34.286804 33.426219 32.545045 31.659340 30.781037 29.918473 29.077047 28.259880 27.468421 26.702962 25.963040 25.247745 24.555938 23.886397 23.237913 22.609343 21.999636 21.407845 20.833123 20.274715 19.731948 19.204218 18.690982 18.191744 17.706049 17.233476 16.773629 16.326138 15.890649 15.466826 15.054346 14.652899 14.262186 13.881917 13.511814 13.151605 12.801027 12.459826 12.127755 11.804574 11.490053 11.183967 10.886098 10.596238 10.314184 10.039742 9.772727 9.512960 9.260272 9.014505 8.775509 8.543146 8.317290 8.097829 7.884669 7.677731 7.476959 7.282321 7.093813 6.911467 6.735355 6.565596 6.402370 6.245925 6.096596 5.954819 5.821157 5.696322 5.581211 5.476945 5.384916 5.306847 5.244862 5.201574 5.180185 5.184612 5.219635 5.291065 5.405955 5.572825 5.801928 6.105534 6.498237 6.997245 7.622642 8.397550 9.348121 10.503239 11.893768 13.551145 15.505070 17.780018 20.390379 23.334157 26.585503 30.086906 33.742613 37.415699 40.931690 44.091423 46.694030 48.567678 49.601589 49.769954 49.138956 47.853181 46.105348 44.099421 42.018454 40.004631 38.152964 36.515433 35.110575 33.934232 32.968870 32.190477 31.573096 31.091462 30.722332 30.445007 30.241380 30.095768 29.994648 29.926390 29.881003 29.849938 29.825912 29.802779 29.775425 29.739675 29.692226 29.630572 29.552937 29.458212 29.345883 29.215957 29.068898 28.905548 28.727065 28.534850 28.330492 28.115711 27.892315 27.662152 27.427086 27.188967 26.949615 26.710802 26.474250 26.241624 26.014531 25.794522 25.583096 25.381704 25.191751 25.014605 24.851597 24.704027 24.573161 24.460236 24.366451 24.292970 24.240905 24.211314 24.205179 24.223395 24.266743 24.335865 24.431240 24.553141 24.701607 24.876402 25.076972 25.302407 25.551402 25.822226 26.112690 26.420132 26.741416 27.072936 27.410656 27.750152 28.086691 28.415319 28.730979 29.028635 29.303418 29.550759 29.766537 29.947201 30.089876 30.192445 30.253595 30.272831 30.250457 30.187521 30.085742 29.947407 29.775263 29.572404 29.342154 29.087965 28.813323 28.521672 28.216356 27.900570 27.577333 27.249475 26.919632 26.590247 26.263589 25.941764 25.626737 25.320350 25.024348 24.740396 24.470099 24.215024 23.976712 23.756693 23.556503 23.377686 23.221806 23.090447 22.985216 22.907736 22.859635 22.842535 22.858026 22.907642 22.992819 23.114851 23.274835 23.473604 23.711653 23.989056 24.305371 24.659551 25.049841 25.473695 25.927692 26.407481 26.907751 27.422243 27.943801 28.464487 28.975748 29.468639 29.934099 30.363267 30.747809 31.080260 31.354321 31.565117 31.709373 31.785507 31.793626 31.735430 31.614033 31.433725 31.199701 30.917767 30.594063 30.234814 29.846116 29.433775 29.003192 28.559295 28.106505 27.648745 27.189458 26.731652 26.277953 25.830655 25.391780 24.963128 24.546333 24.142899 23.754249 23.381758 23.026783 22.690696 22.374905 22.080881 21.810175 21.564439 21.345442 21.155083 20.995402 20.868592 20.776994 20.723104 20.709552 20.739093 20.814563 20.938841 21.114778 21.345112 21.632354 21.978649 22.385613 22.854126 23.384119 23.974321 24.622010 25.322761 26.070223 26.855953 27.669331 28.497595 29.326024 30.138293 30.917011 31.644420 32.303233 32.877538 33.353691 33.721105 33.972841 34.105935 34.121419 34.024036 33.821697 33.524750 33.145155 32.695661 32.189060 31.637583 31.052453 30.443611 29.819591 29.187516 28.553181 27.921199 27.295163 26.677826 26.071269 25.477050 24.896333 24.329998 23.778728 23.243080 22.723534 22.220544 21.734562 21.266075 20.815621 20.383812 19.971354 19.579065 19.207893 18.858943 18.533494 18.233027 17.959250 17.714132 17.499931 17.319229 17.174967 17.070476 17.009513 16.996284 17.035468 17.132215 17.292137 17.521264 17.825960 18.212799 18.688363 19.258972 19.930314 20.706970 21.591822 22.585358 23.684878 24.883653 26.170107 27.527128 28.931630 30.354542 31.761360 33.113395 34.369750 35.489952 36.437014 37.180550 37.699496 37.983971 38.035931 37.868502 37.504089 36.971644 36.303568 35.532730 34.690027 33.802657 32.893195 31.979342 31.074190 30.186806 29.322969 28.485923 27.677057 26.896495 26.143545 25.417049 24.715620 24.037810 23.382207 22.747494 22.132476 21.536086 20.957383 20.395537 19.849820 19.319585 18.804262 18.303338 17.816352 17.342885 16.882555 16.435011 15.999930 15.577017 15.165999 14.766631 14.378691 14.001983 13.636342 13.281632 12.937751 12.604640 12.282279 11.970706 11.670015 11.380373 11.102026 10.835321 10.580718 10.338814 10.110368 9.896333 9.697889 9.516491 9.353920 9.212344 9.094390 9.003234 8.942696 8.917358 8.932695 8.995225 9.112665 9.294114 9.550219 9.893348 10.337724 10.899495 11.596710 12.449112 13.477688 14.703867 16.148231 17.828624 19.757547 21.938774 24.363277 27.004749 29.815321 32.722432 35.628169 38.412497 40.941502 43.080840 44.713005 45.755257 46.173783 45.989852 45.275635 44.140614 42.712455 41.117654 39.466469 37.844411 36.310170 34.898195 33.623721 32.488386 31.485260 30.602781 29.827534 29.146025 28.545706 28.015476 27.545852 27.128936 26.758288 26.428745 26.136234 25.877596 25.650420 25.452901 25.283726 25.141960 25.026972 24.938352 24.875849 24.839320 24.828675 24.843833 24.884679 24.951024 25.042559 25.158823 25.299162 25.462690 25.648259 25.854426 26.079431 26.321173 26.577207 26.844740 27.120647 27.401496 27.683593 27.963037 28.235791 28.497769 28.744926 28.973361 29.179419 29.359791 29.511602 29.632488 29.720657 29.774920 29.794714 29.780085 29.731663 29.650609 29.538553 29.397516 29.229836 29.038081 28.824974 28.593323 28.345960 28.085687 27.815237 27.537245 27.254230 26.968579 26.682551 26.398272 26.117749 25.842875 25.575445 25.317166 25.069676 24.834552 24.613326 24.407495 24.218531 24.047890 23.897013 23.767334 23.660276 23.577251 23.519647 23.488821 23.486079 23.512655 23.569683 23.658162 23.778914 23.932541 24.119364 24.339369 24.592142 24.876803 25.191940 25.535547 25.904974 26.296885 26.707235 27.131281 27.563609 27.998209 28.428577 28.847857 29.249027 29.625097 29.969339 30.275515 30.538106 30.752502 30.915170 31.023757 31.077142 31.075429 31.019873 30.912761 30.757257 30.557214 30.316984 30.041228 29.734741 29.402303 29.048556 28.677914 28.294501 27.902116 27.504221 27.103946 26.704111 26.307250 25.915647 25.531373 25.156324 24.792252 24.440805 24.103555 23.782027 23.477726 23.192159 22.926858 22.683394 22.463395 22.268557 22.100657 21.961552 21.853188 21.777593 21.736868 21.733170 21.768687 21.845603 21.966045 22.132024 22.345353 22.607554 22.919739 23.282480 23.695664 24.158325 24.668485 25.222989 25.817357 26.445667 27.100488 27.772877 28.452466 29.127644 29.785851 30.413980 30.998865 31.527848 31.989360 32.373493 32.672482 32.881079 32.996752 33.019709 32.952737 32.800883 32.571021 32.271339 31.910823 31.498759 31.044312 30.556189 30.042397 29.510101 28.965563 28.414146 27.860364 27.307966 26.760039 26.219110 25.687249 25.166165 24.657288 24.161844 23.680915 23.215496 22.766532 22.334960 21.921738 21.527875 21.154457 20.802668 20.473813 20.169345 19.890880 19.640226 19.419403 19.230663 19.076512 18.959730 18.883380 18.850821 18.865705 18.931964 19.053776 19.235520 19.481687 19.796767 20.185087 20.650602 21.196624 21.825497 22.538203 23.333920 24.209545 25.159197 26.173774 27.240583 28.343145 29.461242 30.571271 31.646975 32.660560 33.584158 34.391538 35.059899 35.571521 35.915061 36.086292 36.088172 35.930242 35.627441 35.198548 34.664471 34.046611 33.365494 32.639753 31.885509 31.116107 30.342140 29.571673 28.810584 28.062949 27.331427 26.617610 25.922313 25.245818 24.588055 23.948743 23.327482 22.723826 22.137321 21.567539 21.014089 20.476624 19.954854 19.448539 18.957493 18.481590 18.020758 17.574990 17.144344 16.728953 16.329033 15.944892 15.576949 15.225744 14.891962 14.576456 14.280273 14.004686 13.751233 13.521758 13.318464 13.143965 13.001354 12.894271 12.826986 12.804477 12.832524 12.917795 13.067925 13.291586 13.598523 13.999546 14.506448 15.131821 15.888728 16.790181 17.848378 19.073644 20.473032 22.048590 23.795316 25.698946 27.733798 29.861064 32.028026 34.168777 36.206970 38.060849 39.650419 40.905950 41.776431 42.236247 42.288445 41.963694 41.315088 40.410040 39.321159 38.118011 36.861103 35.598593 34.365515 33.184825 32.069490 31.024919 30.051289 29.145490 28.302625 27.517072 26.783184 26.095709 25.450012 24.842162 24.268941 23.727805 23.216817 22.734588 22.280205 21.853177 21.453391 21.081070 20.736756 20.421284 20.135782 19.881661 19.660622 19.474662 19.326073 19.217456 19.151720 19.132075 19.162025 19.245338 19.386002 19.588155 19.855990 20.193616 20.604880 21.093139 21.660969 22.309831 23.039672 23.848495 24.731896 25.682629 26.690221 27.740708 28.816564 29.896883 30.957884 31.973760 32.917888 33.764300 34.489332 35.073250 35.501657 35.766487 35.866439 35.806781 35.598562 35.257366 34.801794 34.251887 33.627691 32.948100 32.230036 31.487987 30.733850 29.977015 29.224603 28.481796 27.752191 27.038149 26.341096 25.661787 25.000515 24.357266 23.731843 23.123948 22.533240 21.959377 21.402039 20.860941 20.335848 19.826575 19.332993 18.855036 18.392699 17.946047 17.515220 17.100444 16.702039 16.320433 15.956177 15.609967 15.282663 14.975315 14.689199 14.425846 14.187087 13.975098 13.792455 13.642190 13.527861 13.453621 13.424296 13.445460 13.523516 13.665761 13.880436 14.176754 14.564868 15.055783 15.661157 16.392977 17.263050 18.282278 19.459664 20.801028 22.307425 23.973318 25.784618 27.716815 29.733529 31.785904 33.813343 35.745991 37.509207 39.029850 40.243721 41.102985 41.582120 41.681012 41.424385 40.857633 40.040003 39.036691 37.911464 36.721028 35.511683 34.318187 33.164329 32.064542 31.025957 30.050464 29.136500 28.280478 27.477827 26.723716 26.013496 25.342959 24.708464 24.106967 23.536009 22.993672 22.478523 21.989565 21.526182 21.088103 20.675368 20.288306 19.927522 19.593893 19.288568 19.012975 18.768836 18.558180 18.383362 18.247084 18.152408 18.102774 18.102004 18.154301 18.264227 18.436666 18.676748 18.989749 19.380926 19.855312 20.417428 21.070927 21.818158 22.659643 23.593496 24.614804 25.715011 26.881380 28.096614 29.338751 30.581416 31.794551 32.945645 34.001468 34.930182 35.703645 36.299609 36.703506 36.909544 36.920914 36.749081 36.412283 35.933506 35.338244 34.652382 33.900440 33.104322 32.282594 31.450238 30.618772 29.796614 28.989573 28.201371 27.434145 26.688883 25.965779 25.264517 24.584475 23.924871 23.284865 22.663622 22.060346 21.474299 20.904811 20.351277 19.813157 19.289963 18.781260 18.286657 17.805800 17.338375 16.884097 16.442713 16.014002 15.597775 15.193875 14.802183 14.422617 14.055143 13.699778 13.356596 13.025743 12.707440 12.402008 12.109872 11.831591 11.567876 11.319618 11.087922 10.874150 10.679959 10.507366 10.358802 10.237197 10.146058 10.089574 10.072728 10.101418 10.182601 10.324431 10.536407 10.829507 11.216297 11.710989 12.329400 13.088791 14.007485 15.104208 16.397054 17.901955 19.630595 21.587694 23.767729 26.151273 28.701379 31.360705 34.050335 36.671410 39.110549 41.249470 42.978188 44.209827 44.893908 45.024626 44.641547 43.822190 42.668473 41.290679 39.792918 38.262878 36.766861 35.349435 34.036208 32.838060 31.755567 30.782862 29.910631 29.128209 28.424947 27.790987 27.217663 26.697628 26.224837 25.794450 25.402686 25.046684 24.724355 24.434260 24.175497 23.947606 23.750492 23.584357 23.449643 23.346979 23.277139 23.240995 23.239474 23.273507 23.343984 23.451698 23.597275 23.781117 24.003315 24.263579 24.561142 24.894680 25.262223 25.661077 26.087762 26.537965 27.006520 27.487434 27.973939 28.458609 28.933509 29.390408 29.821024 30.217299 30.571694 30.877478 31.128986 31.321836 31.453082 31.521284 31.526509 31.470239 31.355227 31.185284 30.965048 30.699732 30.394881 30.056153 29.689128 29.299166 28.891294 28.470143 28.039912 27.604363 27.166840 26.730297 26.297344 25.870287 25.451183 25.041883 24.644078 24.259339 23.889157 23.534973 23.198212 22.880305 22.582719 22.306972 22.054658 21.827459 21.627160 21.455663 21.314991 21.207294 21.134847 21.100042 21.105369 21.153393 21.246713 21.387906 21.579457 21.823662 22.122518 22.477580 22.889797 23.359324 23.885320 24.465729 25.097060 25.774196 26.490225 27.236341 28.001837 28.774206 29.539382 30.282132 30.986590 31.636922 32.218075 32.716552 33.121143 33.423534 33.618726 33.705218 33.684934 33.562903 33.346747 33.046032 32.671565 32.234713 31.746794 31.218586 30.659983 30.079769 29.485532 28.883660 28.279416 27.677053 27.079957 26.490786 25.911618 25.344070 24.789416 24.248673 23.722683 23.212176 22.717813 22.240235 21.780092 21.338072 20.914924 20.511484 20.128690 19.767612 19.429466 19.115641 18.827724 18.567523 18.337099 18.138788 17.975235 17.849420 17.764682 17.724743 17.733722 17.796134 17.916879 18.101205 18.354640 18.682881 19.091637 19.586407 20.172182 20.853061 21.631785 22.509174 23.483489 24.549752 25.699067 26.918017 28.188241 29.486295 30.783940 32.048931 33.246393 34.340748 35.298067 36.088618 36.689282 37.085484 37.272307 37.254610 37.046107 36.667601 36.144654 35.505099 34.776739 33.985501 33.154167 32.301715 31.443162 30.589794 29.749623 28.927970 28.128045 27.351495 26.598859 25.869945 25.164107 24.480450 23.817971 23.175648 22.552499 21.947610 21.360145 20.789351 20.234550 19.695135 19.170560 18.660328 18.163988 17.681124 17.211353 16.754316 16.309679 15.877127 15.456366 15.047116 14.649118 14.262128 13.885920 13.520288 13.165046 12.820029 12.485100 12.160146 11.845091 11.539893 11.244554 10.959125 10.683718 10.418511 10.163768 9.919847 9.687222 9.466505 9.258473 9.064098 8.884586 8.721427 8.576443 8.451858 8.350376 8.275269 8.230488 8.220782 8.251845 8.330478 8.464759 8.664243 8.940154 9.305576 9.775605 10.367453 11.100420 11.995700 13.075900 14.364177 15.882836 17.651248 19.682956 21.981906 24.537895 27.321550 30.279573 33.331346 36.368438 39.258663 41.855931 44.015959 45.616015 46.574794 46.867164 46.529084 45.650577 44.358546 42.794538 41.093633 39.369236 37.705651 36.157622 34.754452 33.506172 32.409815 31.454707 30.626410 29.909379 29.288573 28.750319 28.282679 27.875521 27.520413 27.210446 26.940023 26.704651 26.500746 26.325468 26.176574 26.052292 25.951221 25.872242 25.814447 25.777077 25.759470 25.761019 25.781131 25.819193 25.874545 25.946452 26.034083 26.136491 26.252597 26.381178 26.520862 26.670120 26.827270 26.990486 27.157807 27.327159 27.496375 27.663230 27.825469 27.980852 28.127190 28.262387 28.384487 28.491707 28.582481 28.655486 28.709669 28.744264 28.758803 28.753116 28.727327 28.681835 28.617301 28.534617 28.434880 28.319362 28.189474 28.046741 27.892765 27.729206 27.557750 27.380092 27.197919 27.012895 26.826650 26.640774 26.456810 26.276254 26.100550 25.931094 25.769230 25.616256 25.473423 25.341936 25.222951 25.117581 25.026888 24.951880 24.893508 24.852659 24.830143 24.826685 24.842910 24.879325 24.936306 25.014071 25.112664 25.231925 25.371475 25.530683 25.708651 25.904191 26.115810 26.341696 26.579723 26.827448 27.082131 27.340762 27.600097 27.856712 28.107064 28.347563 28.574654 28.784900 28.975071 29.142228 29.283797 29.397639 29.482098 29.536037 29.558855 29.550480 29.511351 29.442381 29.344904 29.220618 29.071518 28.899832 28.707948 28.498363 28.273620 28.036264 27.788805 27.533688 27.273272 27.009814 26.745470 26.482283 26.222196 25.967053 25.718607 25.478534 25.248440 25.029872 24.824330 24.633273 24.458129 24.300299 24.161161 24.042073 23.944368 23.869353 23.818298 23.792424 23.792889 23.820765 23.877009 23.962438 24.077685 24.223161 24.399007 24.605044 24.840717 25.105045 25.396564 25.713282 26.052641 26.411487 26.786065 27.172026 27.564472 27.958013 28.346871 28.725001 29.086247 29.424510 29.733940 30.009126 30.245273 30.438369 30.585310 30.683992 30.733352 30.733366 30.684989 30.590071 30.451227 30.271692 30.055166 29.805656 29.527335 29.224413 28.901027 28.561164 28.208596 27.846847 27.479173 27.108561 26.737736 26.369182 26.005168 25.647773 25.298917 24.960393 24.633890 24.321026 24.023368 23.742460 23.479837 23.237047 23.015666 22.817306 22.643632 22.496360 22.377265 22.288178 22.230975 22.207568 22.219881 22.269822 22.359239 22.489877 22.663306 22.880848 23.143485 23.451751 23.805620 24.204372 24.646465 25.129401 25.649608 26.202335 26.781592 27.380125 27.989468 28.600055 29.201425 29.782502 30.331960 30.838645 31.292041 31.682731 32.002827 32.246328 32.409372 32.490349 32.489881 32.410663 32.257195 32.035419 31.752320 31.415519 31.032888 30.612217 30.160954 29.686011 29.193646 28.689407 28.178129 27.663963 27.150438 26.640535 26.136767 25.641258 25.155823 24.682031 24.221275 23.774818 23.343844 22.929498 22.532918 22.155270 21.797769 21.461710 21.148491 20.859627 20.596782 20.361779 20.156624 19.983522 19.844890 19.743368 19.681825 19.663358 19.691275 19.769075 19.900403 20.088988 20.338546 20.652665 21.034639 21.487266 22.012604 22.611674 23.284134 24.027909 24.838823 25.710238 26.632748 27.593980 28.578547 29.568204 30.542273 31.478329 32.353179 33.144058 33.829960 34.392979 34.819497 35.101068 35.234878 35.223710 35.075441 34.802127 34.418834 33.942368 33.390054 32.778697 32.123795 31.439024 30.735988 30.024182 29.311113 28.602516 27.902630 27.214476 26.540119 25.880906 25.237659 24.610833 24.000640 23.407140 22.830309 22.270085 21.726406 21.199230 20.688555 20.194431 19.716969 19.256353 18.812851 18.386819 17.978725 17.589152 17.218825 16.868625 16.539616 16.233070 15.950503 15.693701 15.464773 15.266183 15.100809 14.971991 14.883588 14.840036 14.846408 14.908459 15.032670 15.226270 15.497225 15.854182 16.306357 16.863326 17.534710 18.329713 19.256490 20.321313 21.527528 22.874320 24.355314 25.957143 27.658132 29.427343 31.224284 32.999605 34.697044 36.256747 37.619850 38.733849 39.558005 40.067833 40.257733 40.141178 39.748369 39.121855 38.311050 37.366722 36.336367 35.261014 34.173580 33.098594 32.052881 31.046814 30.085747 29.171415 28.303133 27.478747 26.695348 25.949755 25.238835 24.559693 23.909764 23.286851 22.689122 22.115089 21.563575 21.033684 20.524767 20.036397 19.568351 19.120594 18.693275 18.286724 17.901457 17.538187 17.197839 16.881571 16.590798 16.327225 16.092878 15.890146 15.721826 15.591164 15.501910 15.458363 15.465418 15.528605 15.654118 15.848815 16.120194 16.476322 16.925693 17.477018 18.138888 18.919318 19.825129 20.861155 22.029275 23.327280 24.747647 26.276298 27.891528 29.563302 31.253177 32.915131 34.497477 35.945949 37.207793 38.236456 38.996195 39.465814 39.640787 39.533298 39.170159 38.589051 37.833833 36.949826 35.979798 34.961139 33.924326 32.892547 31.882187 30.903815 29.963388 29.063445 28.204170 27.384245 26.601508 25.853412 25.137345 24.450813 23.791553 23.157579 22.547192 21.958970 21.391748 20.844586 20.316748 19.807675 19.316972 18.844387 18.389809 17.953259 17.534896 17.135015 16.754066 16.392660 16.051590 15.731856 15.434687 15.161580 14.914333 14.695089 14.506384 14.351205 14.233045 14.155970 14.124684 14.144598 14.221890 14.363557 14.577450 14.872272 15.257531 15.743420 16.340606 17.059880 17.911652 18.905238 20.047909 21.343699 22.791961 24.385743 26.110094 27.940522 29.841878 31.768056 33.662892 35.462596 37.099853 38.509392 39.634414 40.432886 40.882489 40.983120 40.756327 40.241752 39.491366 38.562777 37.512902 36.393037 35.245797 34.103919 32.990534 31.920398 30.901571 29.937164 29.026911 28.168450 27.358294 26.592512 25.867179 25.178643 24.523667 23.899489 23.303823 22.734837 22.191110 21.671591 21.175559 20.702586 20.252512 19.825426 19.421651 19.041744 18.686498 18.356952 18.054403 17.780427 17.536907 17.326054 17.150444 17.013046 16.917259 16.866936 16.866417 16.920536 17.034631 17.214517 17.466435 17.796962 18.212857 18.720845 19.327310 20.037889 20.856945 21.786932 22.827636 23.975332 25.221906 26.554023 27.952456 29.391734 30.840271 32.261129 33.613540 34.855176 35.945066 36.846857 37.531997 37.982351 38.191773 38.166348 37.923235 37.488346 36.893282 36.172080 35.358251 34.482475 33.571112 32.645517 31.722008 30.812306 29.924233 29.062505 28.229507 27.425967 26.651519 25.905124 25.185388 24.490774 23.819744 23.170837 22.542717 21.934187 21.344191 20.771803 20.216215 19.676719 19.152695 18.643598 18.148942 17.668295 17.201274 16.747530 16.306757 15.878680 15.463059 15.059687 14.668393 14.289046 13.921555 13.565875 13.222016 12.890048 12.570110 12.262426 11.967311 11.685196 11.416642 11.162366 10.923271 10.700478 10.495365 10.309620 10.145293 10.004863 9.891319 9.808245 9.759929 9.751476 9.788942 9.879478 10.031478 10.254737 10.560593 10.962040 11.473795 12.112263 12.895361 13.842125 14.972020 16.303839 17.854090 19.634778 21.650527 23.895092 26.347488 28.968180 31.696114 34.447615 37.118359 39.589447 41.737957 43.451192 44.642383 45.264378 45.317593 44.849649 43.946450 42.717138 41.277028 39.732711 38.172083 36.660013 35.238753 33.931323 32.746172 31.681819 30.730785 29.882556 29.125620 28.448744 27.841684 27.295518 26.802732 26.357168 25.953905 25.589095 25.259816 24.963916 24.699884 24.466735 24.263914 24.091208 23.948683 23.836620 23.755462 23.705764 23.688147 23.703253 23.751695 23.834005 23.950585 24.101644 24.287138 24.506703 24.759586 25.044578 25.359947 25.703378 26.071921 26.461957 26.869187 27.288641 27.714721 28.141285 28.561757 28.969278 29.356895 29.717763 30.045371 30.333766 30.577765 30.773134 30.916734 31.006606 31.042003 31.023364 30.952224 30.831094 30.663290 30.452752 30.203857 29.921239 29.609619 29.273673 28.917916 28.546623 28.163778 27.773042 27.377749 26.980914 26.585254 26.193219 25.807022 25.428677 25.060038 24.702825 24.358667 24.029125 23.715720 23.419963 23.143369 22.887484 22.653896 22.444253 22.260271 22.103745 21.976550 21.880645 21.818062 21.790899 21.801297 21.851412 21.943372 22.079227 22.260879 22.489995 22.767913 23.095514 23.473092 23.900204 24.375507 24.896600 25.459869 26.060355 26.691655 27.345879 28.013686 28.684395 29.346213 29.986554 30.592470 31.151158 31.650519 32.079730 32.429769 32.693856 32.867757 32.949918 32.941425 32.845792 32.668610 32.417089 32.099561 31.724974 31.302434 30.840815 30.348466 29.833005 29.301205 28.758959 28.211292 27.662428 27.115876 26.574526 26.040752 25.516513 25.003435 24.502900 24.016103 23.544120 23.087949 22.648556 22.226908 21.824005 21.440903 21.078746 20.738781 20.422388 20.131098 19.866614 19.630839 19.425891 19.254127 19.118162 19.020882 18.965458 18.955344 18.994276 19.086242 19.235450 19.446251 19.723050 20.070167 20.491652 20.991054 21.571126 22.233479 22.978166 23.803240 24.704274 25.673905 26.701428 27.772530 28.869203 29.969938 31.050250 32.083577 33.042527 33.900438 34.633088 35.220387 35.647838 35.907549 35.998648 35.927045 35.704572 35.347652 34.875694 34.309459 33.669576 32.975360 32.244000 31.490109 30.725602 29.959798 29.199701 28.450341 27.715157 26.996350 26.295202 25.612336 24.947928 24.301867 23.673869 23.063562 22.470544 21.894417 21.334808 20.791388 20.263875 19.752038 19.255704 18.774755 18.309136 17.858852 17.423981 17.004675 16.601175 16.213815 15.843044 15.489435 15.153713 14.836772 14.539708 14.263847 14.010786 13.782435 13.581068 13.409377 13.270539 13.168280 13.106962 13.091653 13.128219 13.223401 13.384890 13.621381 13.942594 14.359244 14.882940 15.525960 16.300895 17.220083 18.294805 19.534195 20.943821 22.523956 24.267589 26.158326 28.168428 30.257375 32.371429 34.444740 36.402421 38.165777 39.659407 40.819319 41.600690 41.983629 41.975548 41.609469 40.938593 40.028421 38.948226 37.763552 36.530900 35.294942 34.088017 32.931229 31.836453 30.808604 29.847793 28.951110 28.114006 27.331267 26.597654 25.908293 25.258877 24.645749 24.065911 23.516986 22.997165 22.505142 22.040058 21.601450 21.189207 20.803540 20.444962 20.114273 19.812554 19.541170 19.301776 19.096327 18.927088 18.796645 18.707921 18.664174 18.669003 18.726329 18.840367 19.015579 19.256589 19.568072 19.954590 20.420383 20.969093 21.603428 22.324754 23.132632 24.024305 24.994168 26.033271 27.128905 28.264348 29.418871 30.568060 31.684553 32.739178 33.702502 34.546655 35.247259 35.785243 36.148277 36.331610 36.338187 36.178024 35.866938 35.424862 34.873989 34.237002 33.535589 32.789357 32.015176 31.226915 30.435483 29.649097 28.873654 28.113162 27.370155 26.646068 25.941547 25.256706 24.591319 23.944957 23.317090 22.707150 22.114580 21.538852 20.979484 20.436045 19.908157 19.395497 18.897793 18.414822 17.946413 17.492449 17.052864 16.627651 16.216868 15.820644 15.439188 15.072803 14.721900 14.387014 14.068824 13.768180 13.486131 13.223955 12.983204 12.765745 12.573817 12.410086 12.277722 12.180468 12.122733 12.109684 12.147342 12.242690 12.403763 12.639736 12.960980 13.379067 13.906708 14.557578 15.345982 16.286319 17.392270 18.675647 20.144873 21.803038 23.645609 25.657894 27.812553 30.067574 32.365299 34.633183 36.786907 38.736193 40.393111 41.681931 42.548824 42.969316 42.951578 42.534572 41.781398 40.769504 39.580093 38.288977 36.960282 35.643429 34.372927 33.170077 32.045643 31.002722 30.039336 29.150520 28.329867 27.570580 26.866132 26.210636 25.599016 25.027057 24.491370 23.989332 23.518998 23.079018 22.668563 22.287256 21.935119 21.612530 21.320194 21.059117 20.830591 20.636181 20.477718 20.357293 20.277242 20.240134 20.248749 20.306047 20.415119 20.579123 20.801195 21.084337 21.431268 21.844246 22.324850 22.873724 23.490297 24.172473 24.916316 25.715758 26.562341 27.445059 28.350320 29.262075 30.162165 31.030882 31.847767 32.592580 33.246411 33.792791 34.218720 34.515457 34.679000 34.710169 34.614297 34.400582 34.081178 33.670170 33.182539 32.633233 32.036420 31.404953 30.750057 30.081201 29.406127 28.730978 28.060497 27.398250 26.746844 26.108141 25.483434 24.873603 24.279236 23.700726 23.138342 22.592291 22.062753 21.549914 21.053992 20.575253 20.114029 19.670731 19.245866 18.840051 18.454030 18.088694 17.745101 17.424504 17.128374 16.858435 16.616694 16.405484 16.227502 16.085853 15.984094 15.926280 15.917005 15.961437 16.065340 16.235080 16.477599 16.800349 17.211176 17.718119 18.329126 19.051650 19.892105 20.855175 21.942966 23.154009 24.482177 25.915570 27.435525 29.015915 30.622951 32.215729 33.747702 35.169166 36.430688 37.487169 38.302018 38.850790 39.123590 39.125779 38.876798 38.407346 37.755488 36.962415 36.068583 35.110710 34.119902 33.120840 32.131851 31.165573 30.229925 29.329171 28.464920 27.636984 26.844052 26.084199 25.355242 24.654973 23.981299 23.332325 22.706378 22.102019 21.518026 20.953379 20.407233 19.878898 19.367825 18.873581 18.395846 17.934401 17.489124 17.059992 16.647082 16.250578 15.870782 15.508125 15.163184 14.836703 14.529619 14.243088 13.978523 13.737629 13.522457 13.335450 13.179505 13.058045 12.975088 12.935330 12.944230 13.008099 13.134175 13.330698 13.606955 13.973283 14.441013 15.022320 15.729949 16.576760 17.575055 18.735637 20.066537 21.571425 23.247692 25.084324 27.059754 29.140010 31.277616 33.411779 35.470384 37.374186 39.043169 40.404520 41.401065 41.998542 42.190055 41.996539 41.463023 40.651545 39.632406 38.475666 37.244405 35.990546 34.753254 33.559375 32.425165 31.358590 30.361664 29.432494 28.566913 27.759670 27.005247 26.298362 25.634258 25.008835 24.418686 23.861074 23.333879 22.835535 22.364967 21.921530 21.504965 21.115356 20.753105 20.418913 20.113769 19.838950 19.596016 19.386822 19.213527 19.078597 18.984818 18.935295 18.933450 18.983009 19.087968 19.252542 19.481094 19.778016 20.147583 20.593746 21.119883 21.728475 22.420736 23.196176 24.052135 24.983292 25.981207 27.033941 28.125821 29.237436 30.345922 31.425617 32.449089 33.388527 34.217397 34.912220 35.454252 35.830853 36.036346 36.072220 35.946667 35.673516 35.270747 34.758805 34.158958 33.491864 32.776495 32.029433 31.264540 30.492928 29.723143 28.961482 28.212372 27.478753 26.762430 26.064379 25.384996 24.724293 24.082040 23.457874 22.851366 22.262077 21.689584 21.133496 20.593471 20.069215 19.560489 19.067107 18.588940 18.125920 17.678038 17.245355 16.828003 16.426201 16.040259 15.670597 15.317757 14.982427 14.665461 14.367907 14.091042 13.836402 13.605834 13.401537 13.226125 13.082685 12.974852 12.906882 12.883740 12.911186 12.995857 13.145353 13.368296 13.674369 14.074305 14.579800 15.203330 15.957819 16.856124 17.910264 19.130378 20.523335 22.091017 23.828305 25.720894 27.743185 29.856608 32.008870 34.134693 36.158531 37.999553 39.578691 40.827014 41.694041 42.154301 42.210566 41.892845 41.253313 40.358362 39.279643 38.085945 36.837240 35.581396 34.353362 33.176156 32.062899 31.019197 30.045428 29.138669 28.294184 27.506485 26.770028 26.079641 25.430745 24.819450 24.242563 23.697554 23.182492 22.695986 22.237114 21.805375 21.400637 21.023105 20.673297 20.352027 20.060393 19.799784 19.571875 19.378636 19.222341 19.105574 19.031236 19.002540 19.023006 19.096438 19.226882 19.418567 19.675811 20.002895 20.403888 20.882424 21.441424 22.082757 22.806845 23.612217 24.495039 25.448646 26.463123 27.524990 28.617074 29.718630 30.805781 31.852329 32.830928 33.714577 34.478312 35.100923 35.566489 35.865509 35.995467 35.960749 35.771914 35.444457 34.997246 34.450877 33.826134 33.142730 32.418407 31.668404 30.905262 30.138889 29.376795 28.624438 27.885593 27.162716 26.457278 25.770032 25.101242 24.450849 23.818599 23.204136 22.607059 22.026965 21.463473 20.916242 20.384978 19.869437 19.369432 18.884830 18.415560 17.961612 17.523046 17.099994 16.692673 16.301393 15.926573 15.568753 15.228618 14.907018 14.604995 14.323814 14.065002 13.830386 13.622144 13.442859 13.295580 13.183893 13.111993 13.084766 13.107873 13.187833 13.332097 13.549110 13.848342 14.240274 14.736314 15.348611 16.089732 16.972153 18.007525 19.205648 20.573141 22.111784 23.816585 25.673680 27.658310 29.733206 31.847853 33.939164 35.934033 37.754045 39.322169 40.570747 41.449501 41.931949 42.018728 41.736910 41.135412 40.277555 39.232511 38.067407 36.841392 35.602236 34.385299 33.214316 32.103235 31.058480 30.081157 29.168950 28.317600 27.521975 26.776791 26.077052 25.418291 24.796687 24.209075 23.652930 23.126304 22.627773 22.156373 21.711547 21.293102 20.901172 20.536200 20.198914 19.890331 19.611746 19.364744 19.151206 18.973321 18.833597 18.734875 18.680329 18.673475 18.718155 18.818512 18.978948 19.204049 19.498481 19.866836 20.313440 20.842087 21.455724 22.156055 22.943096 23.814666 24.765868 25.788575 26.870993 27.997367 29.147916 30.299078 31.424142 32.494295 33.480075 34.353137 35.088176 35.664780 36.068971 36.294196 36.341618 36.219657 35.942865 35.530322 35.003796 34.385946 33.698758 32.962362 32.194277 31.409062 30.618295 29.830789 29.052951 28.289198 27.542384 26.814176 26.105392 25.416259 24.746624 24.096100 23.464181 22.850310 22.253927 21.674500 21.111540 20.564606 20.033313 19.517326 19.016366 18.530202 18.058656 17.601599 17.158960 16.730722 16.316933 15.917712 15.533255 15.163853 14.809899 14.471909 14.150540 13.846617 13.561157 13.295403 13.050864 12.829358 12.633065 12.464584 12.327003 12.223974 12.159799 12.139520 12.169019 12.255121 12.405687 12.629702 12.937333 13.339950 13.850068 14.481189 15.247502 16.163374 17.242582 18.497228 19.936268 21.563645 23.376052 25.360433 27.491480 29.729525 32.019383 34.290818 36.461276 38.441266 40.142302 41.486554 42.416643 42.903503 42.950362 42.591676 41.887157 40.912380 39.748227 38.471460 37.147986 35.829361 34.552199 33.339634 32.203874 31.149059 30.173903 29.273847 28.442688 27.673690 26.960299 26.296549 25.677263 25.098114 24.555608 24.047019 23.570311 23.124053 22.707342 22.319735 21.961196 21.632049 21.332949 21.064853 20.829010 20.626942 20.460444 20.331567 20.242618 20.196142 20.194904 20.241860 20.340112 20.492850 20.703269 20.974457 21.309260 21.710102 22.178780 22.716210 23.322149 23.994882 24.730908 25.524629 26.368076 27.250726 28.159423 29.078473 29.989939 30.874165 31.710532 32.478418 33.158297 33.732887 34.188218 34.514507 34.706708 34.764691 34.693003 34.500268 34.198310 33.801115 33.323772 32.781500 32.188845 31.559096 30.903928 30.233235 29.555140 28.876114 28.201163 27.534058 26.877559 26.233636 25.603656 24.988539 24.388896 23.805124 23.237486 22.686174 22.151346 21.633162 21.131809 20.647521 20.180590 19.731389 19.300377 18.888121 18.495308 18.122767 17.771485 17.442634 17.137598 16.857999 16.605735 16.383015 16.192399 16.036842 15.919741 15.844977 15.816963 15.840684 15.921721 16.066274 16.281143 16.573690 16.951734 17.423396 17.996845 18.679939 19.479735 20.401847 21.449644 22.623293 23.918678 25.326257 26.829977 28.406408 30.024304 31.644825 33.222652 34.708115 36.050359 37.201301 38.119945 38.776408 39.154936 39.255326 39.092428 38.693819 38.096124 37.340701 36.469461 35.521437 34.530437 33.523857 32.522471 31.540950 30.588779 29.671344 28.791000 27.947998 27.141247 26.368887 25.628702 24.918395 24.235756 23.578766 22.945635 22.334815 21.744995 21.175078 20.624159 20.091503 19.576526 19.078775 18.597916 18.133726 17.686089 17.254992 16.840532 16.442917 16.062483 15.699698 15.355189 15.029756 14.724400 14.440353 14.179114 13.942490 13.732639 13.552132 13.404008 13.291842 13.219822 13.192824 13.216496 13.297341 13.442783 13.661230 13.962099 14.355796 14.853628 15.467607 16.210123 17.093429 18.128899 19.326002 20.690979 22.225193 23.923226 25.770817 27.742899 29.802048 31.897835 33.967566 35.938885 37.734472 39.278675 40.505344 41.365632 41.834188 41.912290 41.627070 41.026940 40.174282 39.137091 37.981289 36.764955 35.535022 34.326280 33.162141 32.056430 31.015587 30.040822 29.129967 28.278925 27.482722 26.736214 26.034525 25.373288 24.748754 24.157820 23.597997 23.067365 22.564514 22.088484 21.638712 21.214994 20.817445 20.446485 20.102817 19.787424 19.501570 19.246805 19.024976 18.838240 18.689076 18.580301 18.515079 18.496925 18.529701 18.617596 18.765089 18.976887 19.257821 19.612718 20.046203 20.562454 21.164890 21.855782 22.635798 23.503494 24.454763 25.482283 26.575027 27.717898 28.891573 30.072666 31.234266 32.346922 33.380067 34.303810 35.090948 35.718959 36.171741 36.440805 36.525756 36.433975 36.179557 35.781690 35.262736 34.646294 33.955487 33.211640 32.433407 31.636343 30.832845 30.032353 29.241721 28.465661 27.707194 26.968062 26.249090 25.550467 24.871976 24.213152 23.573403 22.952088 22.348566 21.762227 21.192509 20.638902 20.100954 19.578264 19.070486 18.577321 18.098519 17.633873 17.183224 16.746462 16.323526 15.914411 15.519172 15.137937 14.770912 14.418394 14.080789 13.758626 13.452582 13.163501 12.892430 12.640650 12.409719 12.201515 12.018297 11.862765 11.738133 11.648212 11.597504 11.591299 11.635785 11.738161 11.906745 12.151069 12.481954 12.911530 13.453187 14.121409 14.931447 15.898767 17.038211 18.362789 19.882051 21.600010 23.512635 25.605078 27.848900 30.199789 32.596405 34.961137 37.203491 39.226504 40.935967 42.251372 43.116632 43.508167 43.438215 42.952291 42.121345 41.030599 39.767808 38.413422 37.034134 35.680117 34.385300 33.169596 32.042026 31.003901 30.051613 29.178815 28.377985 27.641462 26.962073 26.333461 25.750222 25.207918 24.703022 24.232830 23.795361 23.389262 23.013726 22.668408 22.353376 22.069052 21.816182 21.595800 21.409207 21.257951 21.143807 21.068757 21.034968 21.044758 21.100561 21.204872 21.360182 21.568888 21.833192 22.154964 22.535584 22.975764 23.475337 24.033034 24.646251 25.310821 26.020806 26.768346 27.543575 28.334645 29.127882 29.908094 30.659044 31.364070 32.006831 32.572128 33.046720 33.420063 33.684882 33.837526 33.878042 33.809980 33.639953 33.377011 33.031914 32.616373 32.142356 31.621494 31.064631 30.481520 29.880669 29.269293 28.653365 28.037722 27.426209 26.821833 26.226921 25.643261 25.072226 24.514885 23.972088 23.444538 22.932853 22.437603 21.959354 21.498695 21.056264 20.632768 20.229009 19.845899 19.484486 19.145972 18.831737 18.543367 18.282677 18.051743 17.852930 17.688921 17.562752 17.477839 17.438000 17.447474 17.510931 17.633456 17.820522 18.077920 18.411657 18.827793 19.332218 19.930344 20.626713 21.424500 22.324920 23.326553 24.424606 25.610171 26.869561 28.183819 29.528543 30.874145 32.186678 33.429293 34.564303 35.555737 36.372106 36.989040 37.391382 37.574388 37.543819 37.314895 36.910316 36.357694 35.686838 34.927265 34.106248 33.247511 32.370586 31.490717 30.619165 29.763752 28.929502 28.119286 27.334405 26.575073 25.840802 25.130693 24.443636 23.778450 23.133970 22.509097 21.902821 21.314229 20.742505 20.186922 19.646828 19.121637 18.610821 18.113897 17.630419 17.159974 16.702173 16.256651 15.823059 15.401064 14.990347 14.590599 14.201523 13.822831 13.454245 13.095494 12.746315 12.406453 12.075659 11.753693 11.440321 11.135314 10.838451 10.549516 10.268302 9.994604 9.728226 9.468976 9.216670 8.971128 8.732177 8.499649 8.273385 8.053228 7.839031 7.630654 7.427963 7.230833 7.039148 6.852801 6.671698 6.495757 6.324909 6.159104 5.998311 5.842523 5.691761 5.546077 5.405566 5.270370 5.140689 5.016793 4.899041 4.787894 4.683942 4.587930 4.500793 4.423701 4.358107 4.305811 4.269037 4.250526 4.253649 4.282539 4.342257 4.438978 4.580218 4.775089 5.034597 5.371956 5.802934 6.346182 7.023530 7.860171 8.884648 10.128508 11.625428 13.409577 15.512913 17.961123 20.767945 23.927867 27.407554 31.137098 35.003084 38.846437 42.468526 45.648387 48.171408 49.865566 50.636431 50.489568 49.531170 47.945200 45.954505 43.779162 41.604420 39.564437 37.740922 36.171499 34.861919 33.797890 32.954417 32.302225 31.811652 31.454736 31.206114 31.043263 30.946390 30.898200 30.883621 30.889572 30.904769 30.919580 30.925919 30.917156 30.888043 30.834641 30.754231 30.645217 30.507016 30.339927 30.145004 29.923909 29.678783 29.412110 29.126604 28.825102 28.510480 28.185584 27.853178 27.515907 27.176281 26.836658 26.499250 26.166127 25.839228 25.520379 25.211308 24.913666 24.629041 24.358980 24.105003 23.868618 23.651333 23.454669 23.280166 23.129388 23.003928 22.905403 22.835448 22.795708 22.787816 22.813375 22.873919 22.970878 23.105527 23.278924 23.491845 23.744697 24.037438 24.369474 24.739564 25.145723 25.585127 26.054041 26.547764 27.060607 27.585916 28.116147 28.642991 29.157568 29.650671 30.113068 30.535828 30.910671 31.230307 31.488735 31.681481 31.805761 31.860537 31.846485 31.765866 31.622315 31.420584 31.166245 30.865394 30.524373 30.149519 29.746970 29.322510 28.881471 28.428678 27.968432 27.504523 27.040262 26.578531 26.121837 25.672371 25.232063 24.802637 24.385662 23.982594 23.594817 23.223679 22.870521 22.536709 22.223653 21.932836 21.665829 21.424313 21.210090 21.025106 20.871451 20.751376 20.667289 20.621754 20.617478 20.657292 20.744109 20.880880 21.070519 21.315811 21.619291 21.983099 22.408797 22.897166 23.447964 24.059674 24.729240 25.451808 26.220505 27.026276 27.857817 28.701635 29.542274 30.362718 31.144981 31.870873 32.522884 33.085136 33.544287 33.890316 34.117081 34.222582 34.208903 34.081843 33.850295 33.525447 33.119915 32.646900 32.119450 31.549873 30.949326 30.327581 29.692932 29.052228 28.410981 27.773529 27.143214 26.522566 25.913473 25.317329 24.735165 24.167751 23.615680 23.079433 22.559434 22.056085 21.569800 21.101028 20.650278 20.218137 19.805287 19.412523 19.040776 18.691132 18.364855 18.063412 17.788501 17.542085 17.326417 17.144082 16.998029 16.891608 16.828604 16.813267 16.850333 16.945034 17.103085 17.330649 17.634260 18.020696 18.496787 19.069143 19.743784 20.525660 21.418050 22.421843 23.534716 24.750262 26.057115 27.438204 28.870264 30.323763 31.763427 33.149481 34.439669 35.591980 36.567845 37.335431 37.872532 38.168577 38.225377 38.056443 37.685032 37.141266 36.458858 35.671981 34.812681 33.909085 32.984436 32.056858 31.139651 30.241933 29.369417 28.525208 27.710528 26.925317 26.168711 25.439391 24.735832 24.056471 23.399798 22.764423 22.149090 21.552689 20.974245 20.412908 19.867935 19.338675 18.824560 18.325085 17.839804 17.368317 16.910272 16.465351 16.033275 15.613799 15.206714 14.811848 14.429068 14.058283 13.699451 13.352583 13.017752 12.695104 12.384864 12.087359 11.803025 11.532436 11.276321 11.035596 10.811401 10.605133 10.418504 10.253590 10.112901 9.999457 9.916884 9.869510 9.862485 9.901915 9.994999 10.150181 10.377302 10.687735 11.094497 11.612301 12.257511 13.047950 14.002488 15.140325 16.479870 18.037096 19.823310 21.842262 24.086680 26.534452 29.144925 31.856102 34.583757 37.223648 39.657802 41.765164 43.435758 44.586064 45.172180 45.197179 44.710235 43.797481 42.567051 41.132389 39.597839 38.049122 36.549312 35.139366 33.841513 32.663821 31.604729 30.656863 29.809923 29.052653 28.374082 27.764215 27.214353 26.717176 26.266699 25.858140 25.487781 25.152803 24.851152 24.581402 24.342648 24.134408 23.956547 23.809206 23.692742 23.607681 23.554666 23.534411 23.547659 23.595127 23.677462 23.795180 23.948608 24.137821 24.362566 24.622195 24.915585 25.241068 25.596363 25.978516 26.383859 26.807986 27.245765 27.691374 28.138378 28.579853 29.008541 29.417045 29.798057 30.144600 30.450273 30.709484 30.917660 31.071396 31.168565 31.208350 31.191213 31.118809 30.993841 30.819877 30.601153 30.342361 30.048451 29.724453 29.375323 29.005827 28.620454 28.223363 27.818354 27.408865 26.997984 26.588476 26.182817 25.783229 25.391722 25.010136 24.640173 24.283436 23.941461 23.615744 23.307772 23.019039 22.751075 22.505459 22.283833 22.087923 21.919537 21.780584 21.673066 21.599080 21.560808 21.560497 21.600439 21.682927 21.810207 21.984411 22.207476 22.481037 22.806310 23.183949 23.613887 24.095163 24.625745 25.202350 25.820286 26.473328 27.153636 27.851761 28.556731 29.256257 29.937052 30.585269 31.187040 31.729083 32.199344 32.587596 32.885966 33.089307 33.195398 33.204937 33.121341 32.950373 32.699650 32.378084 31.995318 31.561199 31.085342 30.576781 30.043742 29.493511 28.932385 28.365700 27.797899 27.232629 26.672857 26.120982 25.578947 25.048338 24.530469 24.026458 23.537290 23.063865 22.607048 22.167696 21.746698 21.344996 20.963612 20.603671 20.266425 19.953271 19.665779 19.405714 19.175056 18.976030 18.811122 18.683105 18.595057 18.550369 18.552756 18.606249 18.715168 18.884085 19.117750 19.420983 19.798524 20.254831 20.793817 21.418519 22.130701 22.930392 23.815371 24.780629 25.817847 26.914949 28.055809 29.220187 30.383997 31.519967 32.598729 33.590327 34.466038 35.200349 35.772848 36.169775 36.385005 36.420292 36.284751 35.993659 35.566774 35.026449 34.395789 33.697075 32.950594 32.173900 31.381492 30.584829 29.792570 29.010951 28.244226 27.495093 26.765087 26.054908 25.364689 24.694197 24.042984 23.410490 22.796116 22.199270 21.619390 21.055960 20.508517 19.976652 19.460011 18.958289 18.471232 17.998635 17.540341 17.096246 16.666295 16.250497 15.848920 15.461707 15.089084 14.731372 14.389001 14.062530 13.752670 13.460305 13.186525 12.932661 12.700325 12.491460 12.308392 12.153896 12.031269 11.944409 11.897907 11.897146 11.948403 12.058955 12.237180 12.492643 12.836147 13.279740 13.836634 14.521005 15.347633 16.331299 17.485903 18.823218 20.351233 22.072081 23.979592 26.056636 28.272577 30.581295 32.920442 35.212638 37.369240 39.296954 40.906943 42.125252 42.902632 43.221525 43.098341 42.580260 41.737294 40.651589 39.406521 38.077775 36.727664 35.402843 34.134749 32.941793 31.832284 30.807406 29.863787 28.995522 28.195613 27.456934 26.772805 26.137306 25.545404 24.992965 24.476707 23.994118 23.543368 23.123223 22.732959 22.372304 22.041380 21.740656 21.470926 21.233272 21.029058 20.859905 20.727687 20.634511 20.582704 20.574792 20.613462 20.701526 20.841857 21.037309 21.290618 21.604270 21.980343 22.420314 22.924837 23.493495 24.124523 24.814540 25.558269 26.348321 27.175024 28.026372 28.888105 29.743973 30.576183 31.366052 32.094828 32.744637 33.299478 33.746163 34.075094 34.280802 34.362148 34.322200 34.167774 33.908733 33.557123 33.126252 32.629831 32.081225 31.492887 30.875979 30.240162 29.593547 28.942758 28.293061 27.648549 27.012335 26.386740 25.773467 25.173753 24.588492 24.018340 23.463795 22.925258 22.403080 21.897605 21.409193 20.938244 20.485222 20.050670 19.635228 19.239649 18.864821 18.511789 18.181772 17.876192 17.596705 17.345225 17.123964 16.935464 16.782635 16.668797 16.597710 16.573617 16.601262 16.685913 16.833355 17.049864 17.342141 17.717201 18.182193 18.744141 19.409591 20.184135 21.071816 22.074400 23.190529 24.414802 25.736837 27.140423 28.602912 30.095009 31.581164 33.020699 34.369780 35.584173 36.622580 37.450174 38.041809 38.384379 38.477852 38.334801 37.978487 37.439886 36.754207 35.957478 35.083704 34.162842 33.219695 32.273604 31.338735 30.424757 29.537673 28.680690 27.854994 27.060413 26.295920 25.560025 24.851032 24.167218 23.506940 22.868688 22.251115 21.653036 21.073423 20.511386 19.966159 19.437085 18.923594 18.425197 17.941472 17.472058 17.016647 16.574983 16.146863 15.732130 15.330684 14.942480 14.567539 14.205949 13.857881 13.523597 13.203469 12.897990 12.607801 12.333713 12.076737 11.838122 11.619393 11.422403 11.249386 11.103031 10.986549 10.903771 10.859239 10.858319 10.907322 11.013627 11.185812 11.433768 11.768799 12.203672 12.752598 13.431101 14.255711 15.243437 16.410916 17.773168 19.341882 21.123163 23.114790 25.303076 27.659664 30.138749 32.675479 35.186442 37.573127 39.728907 41.549401 42.944996 43.853295 44.248587 44.145737 43.597179 42.683658 41.501154 40.147311 38.710314 37.261888 35.854653 34.522915 33.285576 32.149876 31.115061 30.175488 29.322981 28.548468 27.843027 27.198488 26.607727 26.064753 25.564691 25.103688 24.678800 24.287872 23.929420 23.602530 23.306774 23.042133 22.808938 22.607822 22.439677 22.305615 22.206939 22.145107 22.121697 22.138366 22.196807 22.298687 22.445586 22.638912 22.879811 23.169055 23.506919 23.893048 24.326307 24.804637 25.324908 25.882789 26.472655 27.087524 27.719063 28.357672 28.992642 29.612424 30.204975 30.758197 31.260425 31.700944 32.070488 32.361678 32.569356 32.690785 32.725685 32.676125 32.546263 32.341985 32.070467 31.739726 31.358175 30.934235 30.476017 29.991086 29.486311 28.967779 28.440785 27.909857 27.378818 26.850870 26.328679 25.814468 25.310102 24.817167 24.337038 23.870940 23.420000 22.985288 22.567859 22.168781 21.789167 21.430201 21.093157 20.779430 20.490552 20.228213 19.994286 19.790844 19.620178 19.484817 19.387533 19.331357 19.319574 19.355712 19.443518 19.586915 19.789936 20.056623 20.390901 20.796400 21.276237 21.832742 22.467135 23.179159 23.966669 24.825214 25.747635 26.723717 27.739965 28.779554 29.822519 30.846241 31.826258 32.737376 33.555051 34.256902 34.824207 35.243203 35.506010 35.611044 35.562861 35.371459 35.051161 34.619224 34.094404 33.495610 32.840815 32.146260 31.425990 30.691667 29.952614 29.216011 28.487181 27.769914 27.066782 26.379431 25.708825 25.055447 24.419458 23.800816 23.199361 22.614875 22.047128 21.495903 20.961012 20.442315 19.939724 19.453208 18.982802 18.528616 18.090835 17.669737 17.265696 16.879202 16.510877 16.161489 15.831981 15.523496 15.237407 14.975352 14.739279 14.531487 14.354682 14.212032 14.107230 14.044561 14.028971 14.066136 14.162520 14.325429 14.563036 14.884375 15.299273 15.818213 16.452083 17.211792 18.107702 19.148842 20.341887 21.689866 23.190641 24.835215 26.606023 28.475441 30.404830 32.344517 34.235113 36.010459 37.602259 38.946092 39.988049 40.690896 41.038497 41.037503 40.715823 40.118215 39.299959 38.319999 37.234832 36.094030 34.937733 33.795956 32.689238 31.630105 30.624848 29.675270 28.780203 27.936715 27.140999 26.388980 25.676706 25.000574 24.357438 23.744645 23.160030 22.601877 22.068882 21.560105 21.074930 20.613034 20.174361 19.759103 19.367693 19.000803 18.659350 18.344503 18.057702 17.800682 17.575492 17.384528 17.230559 17.116763 17.046752 17.024600 17.054857 17.142562 17.293220 17.512767 17.807495 18.183918 18.648592 19.207844 19.867415 20.631994 21.504644 22.486109 23.574032 24.762121 26.039319 27.389101 28.789002 30.210557 31.619789 32.978379 34.245571 35.380744 36.346433 37.111465 37.653734 37.962163 38.037500 37.891775 37.546533 37.030165 36.374827 35.613438 34.777161 33.893612 32.985850 32.072061 31.165785 30.276470 29.410202 28.570452 27.758779 26.975415 26.219741 25.490635 24.786731 24.106582 23.448772 22.811974 22.194981 21.596714 21.016220 20.452660 19.905294 19.373471 18.856610 18.354195 17.865759 17.390879 16.929168 16.480272 16.043865 15.619648 15.207345 14.806706 14.417505 14.039544 13.672650 13.316682 12.971532 12.637131 12.313454 12.000524 11.698425 11.407308 11.127404 10.859037 10.602643 10.358791 10.128205 9.911796 9.710697 9.526304 9.360328 9.214855 9.092417 8.996073 8.929508 8.897145 8.904276 8.957201 9.063395 9.231676 9.472383 9.797544 10.221026 10.758619 11.428036 12.248748 13.241593 14.428049 15.829054 17.463241 19.344473 21.478608 23.859529 26.464705 29.250774 32.150087 35.069428 37.892369 40.486493 42.715907 44.458012 45.621697 46.162667 46.091466 45.471349 44.406253 43.022271 41.447861 39.797620 38.162419 36.606178 35.167735 33.865587 32.703535 31.675896 30.771687 29.977620 29.280062 28.666180 28.124532 27.645283 27.220208 26.842572 26.506958 26.209077 25.945586 25.713924 25.512163 25.338885 25.193076 25.074032 24.981291 24.914558 24.873657 24.858472 24.868906 24.904836 24.966072 25.052315 25.163120 25.297858 25.455680 25.635482 25.835877 26.055166 26.291324 26.541985 26.804443 27.075664 27.352312 27.630782 27.907261 28.177793 28.438357 28.684962 28.913739 29.121046 29.303566 29.458389 29.583099 29.675828 29.735299 29.760842 29.752390 29.710457 29.636086 29.530794 29.396498 29.235440 29.050108 28.843155 28.617336 28.375439 28.120237 27.854444 27.580686 27.301476 27.019206 26.736139 26.454413 26.176042 25.902932 25.636885 25.379620 25.132779 24.897946 24.676654 24.470402 24.280659 24.108875 23.956487 23.824919 23.715584 23.629882 23.569189 23.534848 23.528156 23.550334 23.602511 23.685682 23.800674 23.948095 24.128289 24.341273 24.586676 24.863679 25.170944 25.506563 25.867996 26.252038 26.654793 27.071677 27.497449 27.926269 28.351806 28.767363 29.166056 29.541005 29.885555 30.193501 30.459305 30.678296 30.846831 30.962409 31.023728 31.030688 30.984331 30.886729 30.740841 30.550336 30.319406 30.052577 29.754545 29.430015 29.083584 28.719644 28.342315 27.955410 27.562413 27.166486 26.770481 26.376970 25.988271 25.606490 25.233551 24.871234 24.521208 24.185062 23.864336 23.560542 23.275192 23.009816 22.765981 22.545304 22.349469 22.180232 22.039429 21.928979 21.850878 21.807193 21.800044 21.831579 21.903940 22.019214 22.179375 22.386203 22.641195 22.945449 23.299544 23.703387 24.156067 24.655688 25.199212 25.782319 26.399284 27.042918 27.704555 28.374131 29.040358 29.690998 30.313244 30.894192 31.421382 31.883364 32.270248 32.574194 32.789776 32.914199 32.947342 32.891620 32.751695 32.534065 32.246581 31.897939 31.497197 31.053356 30.575019 30.070143 29.545890 29.008551 28.463542 27.915446 27.368090 26.824637 26.287692 25.759396 25.241522 24.735561 24.242789 23.764335 23.301232 22.854459 22.424982 22.013787 21.621903 21.250435 20.900582 20.573664 20.271141 19.994639 19.745967 19.527144 19.340417 19.188282 19.073498 18.999105 18.968425 18.985064 19.052893 19.176019 19.358727 19.605400 19.920398 20.307901 20.771695 21.314909 21.939688 22.646816 23.435278 24.301800 25.240376 26.241838 27.293522 28.379089 29.478593 30.568851 31.624166 32.617425 33.521527 34.311032 34.963886 35.463002 35.797483 35.963321 35.963449 35.807147 35.508910 35.086941 34.561515 33.953411 33.282587 32.567209 31.823034 31.063159 30.298023 29.535622 28.781824 28.040741 27.315097 26.606559 25.916028 25.243867 24.590088 23.954480 23.336712 22.736397 22.153138 21.586554 21.036303 20.502084 19.983650 19.480808 18.993420 18.521409 18.064758 17.623521 17.197823 16.787874 16.393974 16.016532 15.656077 15.313278 14.988970 14.684174 14.400134 14.138347 13.900609 13.689061 13.506240 13.355144 13.239296 13.162822 13.130525 13.147975 13.221584 13.358685 13.567595 13.857647 14.239179 14.723458 15.322508 16.048796 16.914750 17.932044 19.110618 20.457373 21.974567 23.657909 25.494495 27.460778 29.520918 31.625950 33.714303 35.714138 37.547803 39.138302 40.417127 41.332251 41.854714 41.982283 41.739196 41.171994 40.342373 39.318753 38.168319 36.950889 35.715262 34.497962 33.323833 32.207782 31.156986 30.173104 29.254196 28.396235 27.594219 26.842913 26.137328 25.472973 24.845981 24.253141 23.691873 23.160177 22.656581 22.180070 21.730041 21.306257 20.908809 20.538094 20.194798 19.879889 19.594618 19.340521 19.119430 18.933484 18.785142 18.677195 18.612776 18.595361 18.628764 18.717115 18.864824 19.076511 19.356912 19.710737 20.142483 20.656185 21.255109 21.941365 22.715469 23.575835 24.518243 25.535307 26.616003 27.745322 28.904140 30.069382 31.214574 32.310809 33.328146 34.237357 35.011879 35.629749 36.075271 36.340174 36.424065 36.334118 36.084047 35.692536 35.181385 34.573631 33.891891 33.157067 32.387495 31.598512 30.802388 30.008512 29.223746 28.452844 27.698894 26.963719 26.248222 25.552672 24.876925 24.220581 23.583107 22.963912 22.362399 21.777999 21.210185 20.658479 20.122460 19.601760 19.096062 18.605100 18.128657 17.666566 17.218707 16.785017 16.365486 15.960171 15.569194 15.192763 14.831174 14.484832 14.154264 13.840143 13.543312 13.264808 13.005905 12.768145 12.553389 12.363868 12.202247 12.071692 11.975953 11.919450 11.907369 11.945765 12.041669 12.203188 12.439595 12.761390 13.180322 13.709325 14.362357 15.154074 16.099295 17.212188 18.505119 19.987092 21.661778 23.525151 25.562863 27.747645 30.037152 32.372878 34.680832 36.874667 38.861622 40.551093 41.864878 42.747302 43.173042 43.150608 42.720426 41.947888 40.913116 39.699911 38.386244 37.037778 35.704819 34.422200 33.211140 32.082068 31.037621 30.075328 29.189760 28.374104 27.621223 26.924316 26.277285 25.674894 25.112810 24.587564 24.096475 23.637566 23.209469 22.811350 22.442836 22.103959 21.795112 21.517011 21.270675 21.057403 20.878756 20.736553 20.632851 20.569932 20.550284 20.576569 20.651588 20.778221 20.959355 21.197791 21.496117 21.856557 22.280790 22.769731 23.323288 23.940095 24.617225 25.349925 26.131365 26.952456 27.801757 28.665525 29.527917 30.371397 31.177341 31.926816 32.601517 33.184761 33.662461 34.023963 34.262661 34.376292 34.366894 34.240428 34.006124 33.675638 33.262132 32.779380 32.240978 31.659728 31.047202 30.413499 29.767152 29.115171 28.463160 27.815495 27.175514 26.545714 25.927929 25.323491 24.733358 24.158227 23.598616 23.054934 22.527533 22.016742 21.522904 21.046397 20.587657 20.147193 19.725606 19.323608 18.942037 18.581882 18.244298 17.930639 17.642479 17.381644 17.150247 16.950718 16.785850 16.658828 16.573275 16.533283 16.543445 16.608878 16.735223 16.928631 17.195714 17.543443 17.979001 18.509550 19.141904 19.882104 20.734848 21.702812 22.785827 23.979972 25.276618 26.661531 28.114151 29.607220 31.106955 32.573918 33.964724 35.234572 36.340460 37.244749 37.918582 38.344607 38.518505 38.449012 38.156427 37.669893 37.023967 36.255087 35.398464 34.485758 33.543672 32.593403 31.650774 30.726822 29.828614 28.960127 28.123073 27.317604 26.542874 25.797466 25.079692 24.387794 23.720069 23.074944 22.451002 21.846999 21.261849 20.694616 20.144497 19.610799 19.092931 18.590383 18.102720 17.629570 17.170619 16.725610 16.294337 15.876648 15.472447 15.081698 14.704432 14.340750 13.990841 13.654988 13.333585 13.027156 12.736376 12.462096 12.205376 11.967519 11.750115 11.555090 11.384767 11.241931 11.129909 11.052655 11.014856 11.022038 11.080684 11.198358 11.383830 11.647183 11.999898 12.454891 13.026462 13.730130 14.582287 15.599600 16.798101 18.191858 19.791181 21.600308 23.614625 25.817568 28.177554 30.645480 33.153556 35.616372 37.934985 40.004473 41.724572 43.012031 43.812322 44.107904 43.920714 43.307965 42.352247 41.148527 39.791245 38.364178 36.934402 35.550369 34.243091 33.029153 31.914366 30.897255 29.971960 29.130410 28.363822 27.663652 27.022129 26.432509 25.889151 25.387483 24.923914 24.495718 24.100926 23.738207 23.406783 23.106339 22.836960 22.599076 22.393411 22.220952 22.082914 21.980705 21.915901 21.890212 21.905440 21.963437 22.066050 22.215050 22.412050 22.658414 22.955134 23.302703 23.700971 24.148982 24.644813 25.185409 25.766441 26.382181 27.025430 27.687508 28.358326 29.026553 29.679887 30.305437 30.890190 31.421557 31.887942 32.279303 32.587643 32.807389 32.935618 32.972100 32.919175 32.781461 32.565444 32.278988 31.930826 31.530074 31.085795 30.606660 30.100697 29.575129 29.036302 28.489680 27.939883 27.390767 26.845518 26.306754 25.776626 25.256910 24.749097 24.254461 23.774127 23.309118 22.860408 22.428954 22.015729 21.621752 21.248117 20.896009 20.566734 20.261738 19.982632 19.731210 19.509475 19.319657 19.164236 19.045958 18.967848 18.933219 18.945669 19.009070 19.127533 19.305362 19.546967 19.856757 20.238974 20.697494 21.235561 21.855465 22.558166 23.342858 24.206502 25.143350 26.144504 27.197560 28.286412 29.391293 30.489115 31.554167 32.559189 33.476793 34.281115 34.949565 35.464442 35.814219 35.994295 36.007102 35.861556 35.571945 35.156424 34.635363 34.029741 33.359791 32.643981 31.898374 31.136344 30.368573 29.603250 28.846395 28.102228 27.373543 26.662054 25.968684 25.293804 24.637423 23.999320 23.379149 22.776508 22.190982 21.622174 21.069721 20.533308 20.012668 19.507590 19.017917 18.543552 18.084457 17.640661 17.212264 16.799445 16.402470 16.021709 15.657647 15.310904 14.982255 14.672653 14.383266 14.115501 13.871053 13.651944 13.460578 13.299800 13.172961 13.083995 13.037490 13.038780 13.094024 13.210290 13.395621 13.659088 14.010792 14.461821 15.024108 15.710178 16.532721 17.503958 18.634743 19.933358 21.403986 23.044867 24.846228 26.788173 28.838814 30.953073 33.072680 35.127877 37.041226 38.733564 40.131646 41.176426 41.830431 42.082616 41.949439 41.471804 40.708561 39.728082 38.599806 37.387286 36.143675 34.909743 33.714011 32.574266 31.499746 30.493439 29.554155 28.678196 27.860597 27.095994 26.379165 25.705356 25.070431 24.470923 23.904029 23.367554 22.859859 22.379794 21.926642 21.500071 21.100091 20.727032 20.381520 20.064468 19.777071 19.520812 19.297466 19.109109 18.958131 18.847245 18.779496 18.758259 18.787234 18.870424 19.012095 19.216710 19.488837 19.833009 20.253544 20.754306 21.338406 22.007838 22.763052 23.602475 24.522006 25.514507 26.569353 27.672106 28.804382 29.944009 31.065529 32.141113 33.141868 34.039482 34.808073 35.426034 35.877641 36.154197 36.254523 36.184739 35.957344 35.589786 35.102717 34.518213 33.858162 33.142984 32.390764 31.616783 30.833400 30.050190 29.274253 28.510605 27.762592 27.032276 26.320774 25.628542 24.955587 24.301643 23.666282 23.049003 22.449286 21.866625 21.300549 20.750634 20.216507 19.697848 19.194391 18.705921 18.232277 17.773352 17.329097 16.899524 16.484712 16.084814 15.700070 15.330816 14.977502 14.640706 14.321159 14.019769 13.737649 13.476155 13.236922 13.021914 12.833476 12.674397 12.547977 12.458107 12.409352 12.407046 12.457382 12.567512 12.745632 13.001051 13.344227 13.786749 14.341239 15.021127 15.840267 16.812332 17.949933 19.263398 20.759189 22.437938 24.292179 26.303933 28.442456 30.662587 32.904295 35.094038 37.148475 38.980703 40.508636 41.664436 42.403258 42.709373 42.598044 42.112578 41.317176 40.287368 39.100210 37.826204 36.524021 35.238242 33.999561 32.826602 31.728483 30.707483 29.761401 28.885432 28.073551 27.319440 26.617072 25.961033 25.346670 24.770130 24.228326 23.718873 23.240018 22.790558 22.369775 21.977375 21.613441 21.278401 20.972995 20.698264 20.455535 20.246418 20.072802 19.936851 19.841004 19.787968 19.780700 19.822388 19.916411 20.066286 20.275584 20.547831 20.886360 21.294136 21.773531 22.326065 22.952087 23.650443 24.418102 25.249794 26.137677 27.071081 28.036369 29.016981 29.993707 30.945214 31.848861 32.681761 33.422030 34.050121 34.550084 34.910623 35.125791 35.195228 35.123927 34.921544 34.601381 34.179172 33.671844 33.096380 32.468902 31.804005 31.114366 30.410590 29.701245 28.993020 28.290974 27.598801 26.919104 26.253642 25.603547 24.969497 24.351862 23.750809 23.166380 22.598557 22.047299 21.512573 20.994376 20.492750 20.007793 19.539671 19.088627 18.654995 18.239207 17.841812 17.463489 17.105070 16.767556 16.452151 16.160285 15.893650 15.654239 15.444388 15.266823 15.124708 15.021708 14.962033 14.950503 14.992599 15.094499 15.263111 15.506071 15.831701 16.248915 16.767047 17.395569 18.143690 19.019780 20.030625 21.180463 22.469829 23.894239 25.442783 27.096789 28.828756 30.601843 32.370205 34.080490 35.674641 37.093992 38.284315 39.201167 39.814663 40.112733 40.102162 39.807127 39.265552 38.524039 37.632402 36.638784 35.586003 34.509405 33.436137 32.385500 31.369998 30.396697 29.468611 28.585957 27.747170 26.949692 26.190521 25.466585 24.774971 24.113053 23.478547 22.869520 22.284383 21.721854 21.180937 20.660881 20.161160 19.681449 19.221609 18.781674 18.361854 17.962532 17.584274 17.227842 16.894209 16.584586 16.300446 16.043557 15.816020 15.620311 15.459325 15.336429 15.255510 15.221029 15.238067 15.312374 15.450388 15.659245 15.946752 16.321307 16.791757 17.367163 18.056447 18.867905 19.808542 20.883227 22.093657 23.437154 24.905358 26.482928 28.146431 29.863655 31.593620 33.287572 34.891175 36.347926 37.603616 38.611333 39.336257 39.759388 39.879440 39.712455 39.289204 38.650921 37.844235 36.916258 35.910572 34.864545 33.808024 32.763208 31.745340 30.763855 29.823686 28.926511 28.071819 27.257768 26.481807 25.741124 25.032927 24.354616 23.703871 23.078687 22.477375 21.898543 21.341071 20.804078 20.286900 19.789062 19.310264 18.850367 18.409385 17.987485 17.584989 17.202384 16.840336 16.499703 16.181562 15.887235 15.618320 15.376729 15.164730 14.984995 14.840652 14.735341 14.673271 14.659285 14.698910 14.798409 14.964814 15.205932 15.530319 15.947186 16.466240 17.097415 17.850463 18.734387 19.756670 20.922297 22.232546 23.683600 25.265053 26.958452 28.736103 30.560424 32.384196 34.152018 35.803183 37.275967 38.512989 39.466963 40.105891 40.416649 40.406196 40.100058 39.538422 38.770669 37.849484 36.825603 35.743917 34.641226 33.545532 32.476486 31.446572 30.462586 29.527147 28.640022 27.799210 27.001750 26.244298 25.523502 24.836229 24.179688 23.551476 22.949586 22.372389 21.818595 21.287226 20.777573 20.289177 19.821798 19.375404 18.950161 18.546429 18.164770 17.805951 17.470966 17.161050 16.877708 16.622742 16.398287 16.206848 16.051341 15.935139 15.862113 15.836678 15.863828 15.949166 16.098914 16.319899 16.619501 17.005553 17.486167 18.069474 18.763254 19.574428 20.508403 21.568252 22.753744 24.060254 25.477620 26.989073 28.570407 30.189593 31.807089 33.377044 34.849541 36.173847 37.302433 38.195282 38.823841 39.173901 39.246827 39.058867 38.638665 38.023489 37.254897 36.374618 35.421218 34.427874 33.421298 32.421611 31.442913 30.494229 29.580597 28.704126 27.864905 27.061748 26.292748 25.555674 24.848232 24.168235 23.513684 22.882815 22.274107 21.686271 21.118230 20.569097 20.038154 19.524829 19.028680 18.549386 18.086731 17.640609 17.211016 16.798057 16.401951 16.023040 15.661808 15.318891 14.995104 14.691465 14.409224 14.149901 13.915328 13.707694 13.529601 13.384126 13.274889 13.206128 13.182775 13.210544 13.296005 13.446660 13.670999 13.978520 14.379712 14.885954 15.509314 16.262211 17.156881 18.204614 19.414715 20.793143 22.340845 24.051828 25.911097 27.892696 29.958211 32.056199 34.123053 36.085759 37.866760 39.390721 40.592424 41.424526 41.863579 41.912897 41.601478 40.979159 40.109154 39.059666 37.896299 36.676451 35.446194 34.239444 33.078819 31.977501 30.941446 29.971528 29.065369 28.218754 27.426661 26.683940 25.985737 25.327720 24.706185 24.118072 23.560934 23.032893 22.532572 22.059044 21.611777 21.190592 20.795634 20.427344 20.086451 19.773964 19.491170 19.239648 19.021270 18.838222 18.693010 18.588482 18.527831 18.514598 18.552670 18.646258 18.799853 19.018164 19.306013 19.668195 20.109281 20.633368 21.243752 21.942541 22.730197 23.605013 24.562575 25.595209 26.691505 27.835971 29.008913 30.186623 31.341968 32.445413 33.466481 34.375568 35.145950 35.755757 36.189655 36.439969 36.507089 36.399069 36.130513 35.720918 35.192752 34.569543 33.874204 33.127759 32.348519 31.551694 30.749359 29.950676 29.162271 28.388681 27.632796 26.896267 26.179859 25.483727 24.807634 24.151111 23.513567 22.894368 22.292883 21.708512 21.140703 20.588959 20.052836 19.531946 19.025952 18.534565 18.057544 17.594695 17.145870 16.710970 16.289949 15.882817 15.489649 15.110591 14.745874 14.395823 14.060875 13.741598 13.438710 13.153108 12.885897 12.638430 12.412342 12.209608 12.032596 11.884132 11.767577 11.686910 11.646822 11.652820 11.711335 11.829829 12.016904 12.282395 12.637421 13.094397 13.666953 14.369726 15.217987 16.227012 17.411161 18.782565 20.349396 22.113679 24.068722 26.196335 28.464165 30.823669 33.209411 35.540461 37.724524 39.665097 41.271183 42.468281 43.208535 43.477673 43.296805 42.718442 41.817661 40.680661 39.393432 38.032787 36.660915 35.323484 34.050495 32.858794 31.755228 30.739732 29.807944 28.953212 28.168005 27.444832 26.776777 26.157777 25.582722 25.047446 24.548665 24.083888 23.651314 23.249748 22.878509 22.537371 22.226496 21.946396 21.697896 21.482106 21.300401 21.154400 21.045948 20.977101 20.950097 20.967328 21.031299 21.144574 21.309704 21.529142 21.805125 22.139538 22.533753 22.988434 23.503320 24.077000 24.706662 25.387865 26.114331 26.877788 27.667898 28.472300 29.276784 30.065637 30.822146 31.529266 32.170405 32.730274 33.195726 33.556500 33.805780 33.940513 33.961438 33.872835 33.682029 33.398714 33.034186 32.600567 32.110098 31.574558 31.004826 30.410606 29.800289 29.180936 28.558344 27.937169 27.321082 26.712932 26.114901 25.528651 24.955450 24.396277 23.851906 23.322979 22.810059 22.313674 21.834351 21.372648 20.929173 20.504610 20.099736 19.715444 19.352763 19.012876 18.697151 18.407158 18.144702 17.911850 17.710964 17.544729 17.416188 17.328771 17.286322 17.293121 17.353891 17.473790 17.658388 17.913599 18.245585 18.660595 19.164745 19.763714 20.462343 21.264140 22.170674 23.180878 24.290286 25.490259 26.767273 28.102383 29.470998 30.843100 32.184048 33.456039 34.620223 35.639332 36.480572 37.118398 37.536746 37.730343 37.704850 37.475810 37.066601 36.505755 35.824104 35.052181 34.218162 33.346504 32.457281 31.566089 30.684385 29.820057 28.978105 28.161320 27.370886 26.606890 25.868719 25.155353 24.465577 23.798118 23.151735 22.525261 21.917637 21.327906 20.755220 20.198822 19.658039 19.132271 18.620974 18.123656 17.639863 17.169176 16.711204 16.265579 15.831951 15.409989 14.999376 14.599809 14.210995 13.832655 13.464521 13.106334 12.757847 12.418823 12.089038 11.768277 11.456340 11.153038 10.858198 10.571661 10.293288 10.022957 9.760570 9.506055 9.259370 9.020507 8.789501 8.566432 8.351440 8.144734 7.946602 7.757434 7.577734 7.408150 7.249502 7.102812 6.969356 6.850706 6.748799 6.666009 6.605233 6.570003 6.564607 6.594237 6.665159 6.784909 6.962512 7.208719 7.536254 7.960069 8.497549 9.168662 9.995968 11.004407 12.220743 13.672512 15.386273 17.384968 19.684205 22.287382 25.179765 28.322047 31.644371 35.042505 38.378319 41.486834 44.191259 46.325489 47.760542 48.428627 48.337407 47.569027 46.263403 44.590879 42.722800 40.808009 38.959632 37.252150 35.725808 34.394591 33.254734 32.291995 31.487060 30.819119 30.268000 29.815266 29.444661 29.142171 28.895877 28.695717 28.533237 28.401337 28.294053 28.206370 28.134067 28.073582 28.021918 27.976550 27.935361 27.896588 27.858775 27.820736 27.781524 27.740404 27.696831 27.650428 27.600971 27.548365 27.492638 27.433919 27.372431 27.308471 27.242408 27.174663 27.105707 27.036045 26.966211 26.896760 26.828260 26.761285 26.696408 26.634199 26.575214 26.519995 26.469062 26.422909 26.382002 26.346773 26.317613 26.294873 26.278858 26.269820 26.267958 26.273413 26.286263 26.306519 26.334124 26.368948 26.410786 26.459356 26.514298 26.575175 26.641471 26.712593 26.787877 26.866592 26.947943 27.031080 27.115109 27.199098 27.282093 27.363128 27.441238 27.515476 27.584925 27.648714 27.706028 27.756127 27.798354 27.832147 27.857044 27.872697 27.878870 27.875443 27.862415 27.839896 27.808111 27.767389 27.718155 27.660928 27.596305 27.524954 27.447606 27.365043 27.278088 27.187594 27.094440 26.999518 26.903727 26.807968 26.713132 26.620102 26.529740 26.442891 26.360370 26.282962 26.211420 26.146458 26.088746 26.038912 25.997529 25.965117 25.942138 25.928984 25.925979 25.933370 25.951319 25.979901 26.019094 26.068773 26.128707 26.198552 26.277845 26.366005 26.462329 26.565992 26.676050 26.791445 26.911012 27.033491 27.157536 27.281736 27.404629 27.524728 27.640540 27.750593 27.853459 27.947783 28.032303 28.105880 28.167512 28.216355 28.251738 28.273172 28.280358 28.273184 28.251729 28.216249 28.167175 28.105094 28.030735 27.944953 27.848712 27.743065 27.629139 27.508113 27.381210 27.249675 27.114767 26.977745 26.839863 26.702357 26.566443 26.433307 26.304108 26.179966 26.061966 25.951153 25.848529 25.755051 25.671628 25.599119 25.538324 25.489987 25.454780 25.433306 25.426083 25.433540 25.456004 25.493691 25.546690 25.614957 25.698297 25.796354 25.908598 26.034319 26.172612 26.322376 26.482311 26.650919 26.826511 27.007221 27.191022 27.375750 27.559135 27.738839 27.912491 28.077737 28.232287 28.373960 28.500732 28.610782 28.702530 28.774672 28.826204 28.856438 28.865012 28.851886 28.817331 28.761915 28.686469 28.592067 28.479986 28.351670 28.208700 28.052756 27.885587 27.708979 27.524737 27.334655 27.140510 26.944040 26.746940 26.550855 26.357373 26.168032 25.984310 25.807635 25.639383 25.480879 25.333403 25.198189 25.076426 24.969257 24.877779 24.803035 24.746012 24.707634 24.688747 24.690111 24.712379 24.756087 24.821625 24.909221 25.018910 25.150513 25.303607 25.477498 25.671197 25.883398 26.112458 26.356386 26.612839 26.879134 27.152256 27.428901 27.705512 27.978343 28.243530 28.497174 28.735434 28.954622 29.151301 29.322381 29.465198 29.577586 29.657932 29.705202 29.718958 29.699340 29.647040 29.563248 29.449595 29.308077 29.140983 28.950816 28.740223 28.511928 28.268672 28.013170 27.748072 27.475932 27.199195 26.920186 26.641104 26.364027 26.090919 25.823637 25.563947 25.313531 25.074004 24.846924 24.633807 24.436131 24.255352 24.092904 23.950208 23.828672 23.729689 23.654630 23.604842 23.581629 23.586233 23.619818 23.683434 23.777984 23.904186 24.062521 24.253182 24.476020 24.730477 25.015526 25.329613 25.670600 26.035715 26.421528 26.823934 27.238166 27.658840 28.080030 28.495378 28.898239 29.281860 29.639576 29.965030 30.252387 30.496540 30.693289 30.839483 30.933107 30.973320 30.960433 30.895835 30.781868 30.621683 30.419056 30.178212 29.903645 29.599965 29.271751 28.923447 28.559278 28.183194 27.798839 27.409542 27.018324 26.627916 26.240781 25.859155 25.485073 25.120406 24.766897 24.426190 24.099861 23.789444 23.496456 23.222420 22.968881 22.737425 22.529691 22.347383 22.192279 22.066231 21.971168 21.909090 21.882051 21.892146 21.941476 22.032110 22.166031 22.345071 22.570828 22.844564 23.167093 23.538647 23.958734 24.425981 24.937982 25.491153 26.080599 26.700033 27.341729 27.996563 28.654123 29.302928 29.930741 30.524979 31.073203 31.563650 31.985786 32.330807 32.592063 32.765351 32.849052 32.844095 32.753764 32.583363 32.339790 32.031053 31.665795 31.252845 30.800846 30.317960 29.811667 29.288646 28.754736 28.214941 27.673489 27.133909 26.599126 26.071558 25.553210 25.045757 24.550629 24.069068 23.602195 23.151051 22.716643 22.299977 21.902089 21.524071 21.167100 20.832457 20.521552 20.235945 19.977367 19.747745 19.549218 19.384159 19.255192 19.165205 19.117357 19.115080 19.162065 19.262238 19.419713 19.638718 19.923494 20.278154 20.706491 21.211744 21.796304 22.461361 23.206510 24.029309 24.924833 25.885241 26.899422 27.952769 29.027153 30.101173 31.150723 32.149910 33.072314 33.892494 34.587627 35.139091 35.533790 35.765037 35.832857 35.743673 35.509429 35.146282 34.673076 34.109797 33.476201 32.790720 32.069728 31.327138 30.574302 29.820120 29.071302 28.332701 27.607662 26.898361 26.206096 25.531541 24.874939 24.236253 23.615279 23.011724 22.425260 21.855562 21.302328 20.765296 20.244253 19.739037 19.249545 18.775736 18.317633 17.875332 17.449008 17.038925 16.645446 16.269048 15.910339 15.570080 15.249205 14.948851 14.670391 14.415471 14.186052 13.984459 13.813438 13.676215 13.576564 13.518883 13.508264 13.550577 13.652534 13.821757 14.066817 14.397231 14.823418 15.356563 16.008376 16.790695 17.714902 18.791100 20.027018 21.426619 22.988436 24.703698 26.554415 28.511660 30.534437 32.569570 34.553083 36.413427 38.076638 39.473078 40.544895 41.252914 41.581516 41.540320 41.162190 40.497987 39.609277 38.560597 37.412759 36.218140 35.018246 33.843289 32.713193 31.639397 30.626911 29.676251 28.785082 27.949476 27.164816 26.426397 25.729800 25.071086 24.446884 23.854411 23.291437 22.756247 22.247580 21.764583 21.306764 20.873956 20.466289 20.084170 19.728279 19.399560 19.099230 18.828788 18.590028 18.385061 18.216327 18.086625 17.999124 17.957382 17.965352 18.027378 18.148178 18.332795 18.586527 18.914812 19.323062 19.816436 20.399545 21.076070 21.848300 22.716588 23.678735 24.729345 25.859189 27.054655 28.297380 29.564178 30.827361 32.055561 33.215091 34.271813 35.193391 35.951694 36.525046 36.899994 37.072295 37.046969 36.837387 36.463573 35.950011 35.323308 34.610037 33.835002 33.020059 32.183477 31.339790 30.499995 29.671980 28.861055 28.070502 27.302088 26.556498 25.833687 25.133156 24.454141 23.795758 23.157089 22.537241 21.935374 21.350718 20.782574 20.230314 19.693375 19.171249 18.663478 18.169648 17.689381 17.222332 16.768184 16.326649 15.897463 15.480388 15.075212 14.681748 14.299840 13.929365 13.570233 13.222397 12.885856 12.560664 12.246938 11.944867 11.654729 11.376904 11.111891 10.860333 10.623042 10.401033 10.195559 10.008157 9.840701 9.695467 9.575204 9.483222 9.423493 9.400761 9.420679 9.489946 9.616465 9.809508 10.079872 10.440027 10.904216 11.488493 12.210628 13.089840 14.146263 15.400034 16.869915 18.571301 20.513560 22.696665 25.107268 27.714540 30.466449 33.287425 36.078671 38.722366 41.090599 43.058846 44.522343 45.412134 45.706721 45.435771 44.674377 43.529298 42.121059 40.566646 38.966572 37.397971 35.913337 34.543223 33.300936 32.187636 31.196851 30.318001 29.538871 28.847212 28.231656 27.682174 27.190224 26.748734 26.351979 25.995421 25.675539 25.389667 25.135842 24.912679 24.719261 24.555040 24.419765 24.313408 24.236108 24.188117 24.169750 24.181340 24.223187 24.295517 24.398429 24.531845 24.695463 24.888696 25.110626 25.359948 25.634926 25.933348 26.252498 26.589135 26.939495 27.299306 27.663836 28.027955 28.386234 28.733063 29.062786 29.369866 29.649041 29.895493 30.105004 30.274083 30.400078 30.481241 30.516755 30.506726 30.452128 30.354716 30.216919 30.041713 29.832479 29.592875 29.326710 29.037832 28.730037 28.406994 28.072198 27.728931 27.380249 27.028974 26.677703 26.328826 25.984541 25.646880 25.317736 24.998884 24.692011 24.398737 24.120636 23.859257 23.616142 23.392838 23.190915 23.011971 22.857640 22.729596 22.629551 22.559252 22.520465 22.514963 22.544497 22.610764 22.715365 22.859748 23.045148 23.272504 23.542375 23.854836 24.209374 24.604773 25.038997 25.509091 26.011081 26.539914 27.089425 27.652360 28.220454 28.784574 29.334940 29.861403 30.353795 30.802308 31.197896 31.532665 31.800212 31.995893 32.116993 32.162776 32.134421 32.034856 31.868492 31.640912 31.358522 31.028209 30.657024 30.251913 29.819506 29.365961 28.896878 28.417248 27.931458 27.443316 26.956100 26.472623 25.995299 25.526209 25.067170 24.619788 24.185518 23.765707 23.361637 22.974562 22.605739 22.256457 21.928063 21.621984 21.339749 21.083011 20.853560 20.653345 20.484486 20.349284 20.250229 20.190002 20.171468 20.197662 20.271757 20.397023 20.576759 20.814208 21.112437 21.474190 21.901700 22.396473 22.959028 23.588606 24.282870 25.037585 25.846337 26.700286 27.588030 28.495588 29.406573 30.302563 31.163711 31.969567 32.700083 33.336719 33.863558 34.268291 34.542964 34.684392 34.694176 34.578337 34.346613 34.011529 33.587364 33.089127 32.531659 31.928920 31.293494 30.636307 29.966528 29.291618 28.617473 27.948625 27.288469 26.639478 26.003411 25.381489 24.774539 24.183115 23.607588 23.048219 22.505210 21.978743 21.469013 20.976248 20.500727 20.042797 19.602889 19.181530 18.779361 18.397155 18.035834 17.696493 17.380423 17.089143 16.824428 16.588346 16.383297 16.212050 16.077793 15.984173 15.935342 15.935999 15.991418 16.107473 16.290633 16.547933 16.886899 17.315412 17.841501 18.473028 19.217268 20.080330 21.066444 22.177081 23.409937 24.757841 26.207658 27.739343 29.325322 30.930429 32.512614 34.024606 35.416589 36.639771 37.650496 38.414358 38.909632 39.129373 39.081761 38.788580 38.282158 37.601362 36.787398 35.880093 34.915116 33.922314 32.925084 31.940587 30.980474 30.051894 29.158545 28.301651 27.480778 26.694484 25.940786 25.217494 24.522424 23.853524 23.208946 22.587068 21.986494 21.406046 20.844738 20.301756 19.776438 19.268253 18.776789 18.301740 17.842901 17.400164 16.973514 16.563042 16.168941 15.791525 15.431238 15.088669 14.764580 14.459922 14.175875 13.913872 13.675648 13.463284 13.279261 13.126521 13.008537 12.929385 12.893831 12.907412 12.976527 13.108514 13.311723 13.595559 13.970484 14.447953 15.040260 15.760244 16.620827 17.634312 18.811415 20.159962 21.683258 23.378139 25.232816 27.224717 29.318667 31.465851 33.604125 35.660200 37.554040 39.205436 40.542146 41.508364 42.071879 42.228260 42.000953 41.437159 40.600464 39.561958 38.391760 37.152437 35.895048 34.657752 33.466410 32.336423 31.275076 30.283879 29.360593 28.500837 27.699239 26.950229 26.248519 25.589370 24.968719 24.383202 23.830127 23.307420 22.813559 22.347509 21.908668 21.496812 21.112065 20.754865 20.425951 20.126349 19.857373 19.620621 19.417990 19.251673 19.124176 19.038317 18.997231 19.004364 19.063451 19.178487 19.353667 19.593303 19.901711 20.283042 20.741084 21.278986 21.898942 22.601808 23.386667 24.250360 25.187018 26.187615 27.239624 28.326830 29.429376 30.524114 31.585315 32.585756 33.498134 34.296729 34.959138 35.467882 35.811667 35.986116 35.993854 35.843943 35.550756 35.132479 34.609454 34.002593 33.332032 32.616127 31.870829 31.109407 30.342450 29.578071 28.822232 28.079107 27.351464 26.640994 25.948610 25.274680 24.619210 23.981982 23.362656 22.760832 22.176101 21.608073 21.056391 20.520744 20.000873 19.496571 19.007689 18.534135 18.075880 17.632959 17.205480 16.793633 16.397696 16.018050 15.655195 15.309767 14.982559 14.674549 14.386927 14.121131 13.878888 13.662258 13.473691 13.316080 13.192833 13.107947 13.066081 13.072646 13.133887 13.256958 13.449997 13.722160 14.083633 14.545568 15.119935 15.819250 16.656128 17.642626 18.789314 20.104040 21.590374 23.245741 25.059353 27.010116 29.064828 31.177104 33.287530 35.325582 37.213652 38.873185 40.232401 41.234502 41.844804 42.055206 41.884824 41.376541 40.590274 39.594523 38.458058 37.243240 36.001789 34.773052 33.584306 32.452349 31.385704 30.386894 29.454458 28.584576 27.772259 27.012181 26.299195 25.628631 24.996443 24.399249 23.834320 23.299529 22.793292 22.314505 21.862492 21.436950 21.037920 20.665754 20.321099 20.004887 19.718333 19.462939 19.240498 19.053109 18.903185 18.793467 18.727027 18.707275 18.737949 18.823092 18.967017 19.174238 19.449374 19.797013 20.221526 20.726828 21.316072 21.991278 22.752908 23.599371 24.526510 25.527093 26.590346 27.701632 28.842314 29.989917 31.118651 32.200341 33.205767 34.106340 34.875985 35.493013 35.941746 36.213657 36.307857 36.230829 35.995485 35.619678 35.124434 34.532134 33.864888 33.143258 32.385396 31.606585 30.819141 30.032568 29.253879 28.487997 27.738181 27.006410 26.293731 25.600537 24.926785 24.272166 23.636218 23.018412 22.418202 21.835063 21.268506 20.718090 20.183428 19.664184 19.160075 18.670871 18.196394 17.736516 17.291168 16.860336 16.444072 16.042498 15.655818 15.284327 14.928425 14.588638 14.265632 13.960240 13.673492 13.406643 13.161217 12.939046 12.742325 12.573671 12.436189 12.333549 12.270068 12.250804 12.281650 12.369437 12.522022 12.748375 13.058633 13.464108 13.977229 14.611376 15.380563 16.298925 17.379940 18.635335 20.073618 21.698221 23.505279 25.481181 27.600133 29.822133 32.091925 34.339570 36.483259 38.434723 40.107106 41.424455 42.331248 42.799963 42.834784 42.470389 41.765978 40.796036 39.640030 38.373253 37.060299 35.751703 34.483390 33.278104 32.147909 31.096966 30.124113 29.224970 28.393524 27.623226 26.907687 26.241084 25.618357 25.035273 24.488410 23.975095 23.493332 23.041716 22.619361 22.225833 21.861099 21.525483 21.219637 20.944516 20.701366 20.491711 20.317351 20.180353 20.083046 20.028013 20.018072 20.056254 20.145761 20.289911 20.492062 20.755503 21.083326 21.478245 21.942391 22.477061 23.082428 23.757219 24.498384 25.300751 26.156733 27.056092 27.985822 28.930200 29.871039 30.788186 31.660269 32.465672 33.183683 33.795718 34.286484 34.644959 34.865055 34.945876 34.891532 34.710555 34.414997 34.019330 33.539314 32.990937 32.389536 31.749149 31.082111 30.398868 29.707973 29.016211 28.328803 27.649647 26.981564 26.326534 25.685898 25.060527 24.450966 23.857538 23.280427 22.719743 22.175560 21.647956 21.137031 20.642930 20.165853 19.706075 19.263952 18.839935 18.434589 18.048605 17.682819 17.338235 17.016051 16.717681 16.444793 16.199343 15.983613 15.800255 15.652338 15.543396 15.477481 15.459209 15.493804 15.587134 15.745727 15.976763 16.288029 16.687816 17.184748 17.787517 18.504491 19.343186 20.309568 21.407180 22.636091 23.991719 25.463584 27.034129 28.677800 30.360609 32.040459 33.668465 35.191417 36.555371 37.710066 38.613654 39.236978 39.566612 39.606026 39.374607 38.904716 38.237407 37.417631 36.489801 35.494307 34.465311 33.429790 32.407604 31.412251 30.451984 29.531028 28.650716 27.810452 27.008475 26.242413 25.509674 24.807697 24.134098 23.486748 22.863803 22.263699 21.685140 21.127070 20.588646 20.069217 19.568299 19.085559 18.620809 18.173991 17.745182 17.334594 16.942582 16.569655 16.216491 15.883959 15.573142 15.285368 15.022241 14.785686 14.577988 14.401850 14.260445 14.157480 14.097259 14.084755 14.125674 14.226514 14.394613 14.638175 14.966249 15.388665 15.915879 16.558718 17.327974 18.233823 19.285027 20.487885 21.844938 23.353437 25.003657 26.777221 28.645656 30.569525 32.498518 34.372904 36.126600 37.691905 39.005541 40.015223 40.685648 41.002677 40.974731 40.631026 40.017010 39.188016 38.202499 37.116095 35.977339 34.825328 33.689149 32.588596 31.535647 30.536225 29.591920 28.701463 27.861901 27.069452 26.320098 25.609953 24.935485 24.293611 23.681737 23.097746 22.539961 22.007110 21.498275 21.012857 20.550544 20.111284 19.695271 19.302938 18.934951 18.592219 18.275902 17.987430 17.728526 17.501227 17.307915 17.151351 17.034701 16.961573 16.936038 16.962654 17.046472 17.193027 17.408300 17.698645 18.070667 18.531043 19.086255 19.742241 20.503932 21.374672 22.355531 23.444512 24.635691 25.918378 27.276362 28.687409 30.123143 31.549491 32.927813 34.216784 35.374980 36.363953 37.151439 37.714258 38.040395 38.129902 37.994432 37.655489 37.141737 36.485838 35.721364 34.880176 33.990559 33.076146 32.155586 31.242753 30.347317 29.475489 28.630792 27.814786 27.027672 26.268778 25.536925 24.830686 24.148556 23.489071 22.850858 22.232677 21.633418 21.052105 20.487881 19.939996 19.407789 18.890678 18.388144 17.899724 17.425000 16.963597 16.515171 16.079413 15.656045 15.244816 14.845505 14.457923 14.081913 13.717354 13.364165 13.022311 12.691808 12.372732 12.065226 11.769517 11.485924 11.214880 10.956949 10.712849 10.483487 10.269988 10.073736 9.896429 9.740129 9.607338 9.501071 9.424956 9.383334 9.381388 9.425277 9.522285 9.680986 9.911403 10.225166 10.635640 11.157997 11.809199 12.607828 13.573693 14.727127 16.087853 17.673309 19.496331 21.562122 23.864565 26.382109 29.073707 31.875622 34.700242 37.438185 39.964828 42.151652 43.881543 45.065545 45.657295 45.661080 45.130821 44.159912 42.864723 41.366352 39.775093 38.180431 36.647116 35.216158 33.908787 32.731536 31.681156 30.748664 29.922341 29.189744 28.538935 27.959141 27.441039 26.976808 26.560047 26.185626 25.849523 25.548647 25.280688 25.043975 24.837353 24.660089 24.511776 24.392268 24.301611 24.239988 24.207671 24.204968 24.232179 24.289552 24.377231 24.495210 24.643280 24.820980 25.027544 25.261848 25.522364 25.807113 26.113633 26.438954 26.779590 27.131546 27.490352 27.851119 28.208618 28.557385 28.891851 29.206488 29.495966 29.755316 29.980088 30.166489 30.311504 30.412975 30.469656 30.481210 30.448184 30.371940 30.254554 30.098703 29.907530 29.684518 29.433358 29.157833 28.861722 28.548716 28.222360 27.886007 27.542795 27.195638 26.847225 26.500031 26.156337 25.818247 25.487715 25.166570 24.856541 24.559275 24.276369 24.009382 23.759858 23.529337 23.319377 23.131555 22.967481 22.828802 22.717200 22.634392 22.582117 22.562126 22.576157 22.625907 22.712995 22.838908 23.004948 23.212159 23.461241 23.752462 24.085552 24.459595 24.872917 25.322979 25.806280 26.318282 26.853368 27.404837 27.964959 28.525088 29.075849 29.607379 30.109648 30.572815 30.987615 31.345748 31.640234 31.865708 32.018633 32.097397 32.102307 32.035466 31.900558 31.702561 31.447423 31.141717 30.792326 30.406151 29.989881 29.549812 29.091733 28.620860 28.141813 27.658635 27.174823 26.693392 26.216933 25.747678 25.287570 24.838319 24.401460 23.978400 23.570464 23.178932 22.805073 22.450175 22.115569 21.802659 21.512937 21.248009 21.009612 20.799626 20.620097 20.473244 20.361465 20.287346 20.253653 20.263321 20.319428 20.425161 20.583753 20.798406 21.072188 21.407895 21.807880 22.273854 22.806642 23.405911 24.069880 24.795013 25.575731 26.404158 27.269949 28.160234 29.059719 29.950980 30.814973 31.631770 32.381470 33.045258 33.606488 34.051701 34.371445 34.560806 34.619569 34.552002 34.366290 34.073707 33.687640 33.222577 32.693188 32.113549 31.496575 30.853667 30.194543 29.527230 28.858167 28.192380 27.533694 26.884943 26.248183 25.624870 25.016015 24.422315 23.844246 23.282151 22.736291 22.206891 21.694179 21.198404 20.719865 20.258919 19.816005 19.391655 18.986512 18.601344 18.237070 17.894774 17.575735 17.281455 17.013682 16.774455 16.566133 16.391437 16.253497 16.155886 16.102670 16.098442 16.148357 16.258146 16.434115 16.683116 17.012470 17.429838 17.943019 18.559648 19.286784 20.130365 21.094513 22.180696 23.386755 24.705849 26.125411 27.626229 29.181834 30.758417 32.315470 33.807323 35.185659 36.402882 37.416038 38.190777 38.704717 38.949595 38.931761 38.670903 38.197258 37.547845 36.762448 35.879967 34.935631 33.959250 32.974493 31.998967 31.044859 30.119857 29.228161 28.371429 27.549585 26.761467 26.005308 25.279074 24.580693 23.908188 23.259757 22.633803 22.028940 21.443985 20.877938 20.329961 19.799361 19.285569 18.788124 18.306666 17.840921 17.390704 16.955912 16.536529 16.132627 15.744377 15.372057 15.016065 14.676940 14.355374 14.052245 13.768642 13.505896 13.265628 13.049785 12.860704 12.701163 12.574458 12.484477 12.435783 12.433708 12.484448 12.595155 12.774024 13.030364 13.374627 13.818392 14.374261 15.055635 15.876325 16.849941 17.989008 19.303746 20.800478 22.479671 24.333670 26.344299 28.480620 30.697315 32.934262 35.117934 37.165136 38.989266 40.508676 41.656067 42.387163 42.686738 42.570422 42.081689 41.284701 40.254773 39.068634 37.796412 36.496425 35.212959 33.976489 32.805497 31.709024 30.689321 29.744189 28.868853 28.057318 27.303306 26.600826 25.944497 25.329696 24.752591 24.210114 23.699897 23.220196 22.769817 22.348047 21.954595 21.589548 21.253333 20.946692 20.670664 20.426578 20.216044 20.040953 19.903471 19.806044 19.751384 19.742462 19.782479 19.874837 20.023078 20.230808 20.501594 20.838820 21.245511 21.724113 22.276221 22.902275 23.601214 24.370101 25.203758 26.094425 27.031490 28.001348 28.987431 29.970472 30.929026 31.840287 32.681146 33.429459 34.065391 34.572705 34.939841 35.160627 35.234553 35.166528 34.966208 34.646961 34.224647 33.716352 33.139238 32.509605 31.842211 31.149876 30.443321 29.731197 29.020258 28.315598 27.620933 26.938871 26.271170 25.618949 24.982875 24.363300 23.760371 23.174115 22.604494 22.051451 21.514933 20.994921 20.491436 20.004560 19.534439 19.081294 18.645435 18.227271 17.827322 17.446237 17.084811 16.744010 16.424991 16.129135 15.858081 15.613759 15.398437 15.214764 15.065824 14.955186 14.886968 14.865886 14.897316 14.987333 15.142749 15.371119 15.680709 16.080419 16.579626 17.187930 17.914773 18.768904 19.757656 20.886022 22.155519 23.562878 25.098621 26.745668 28.478167 30.260819 32.049012 33.790070 35.425828 36.896552 38.145930 39.126533 39.804879 40.165132 40.210632 39.962890 39.458234 38.742821 37.867058 36.880440 35.827568 34.745686 33.663714 32.602455 31.575580 30.590985 29.652230 28.759841 27.912395 27.107356 26.341674 25.612189 24.915884 24.250024 23.612226 23.000471 22.413093 21.848754 21.306410 20.785277 20.284807 19.804660 19.344692 18.904941 18.485624 18.087139 17.710073 17.355211 17.023559 16.716361 16.435129 16.181677 15.958154 15.767087 15.611428 15.494598 15.420541 15.393771 15.419416 15.503260 15.651756 15.872030 16.171840 16.559482 17.043634 17.633099 18.336437 19.161449 20.114502 21.199662 22.417666 23.764726 25.231272 26.800725 28.448503 30.141479 31.838178 33.489953 35.043321 36.443461 37.638633 38.584992 39.251061 39.621053 39.696334 39.494707 39.047638 38.396011 37.585246 36.660690 35.663926 34.630369 33.588161 32.558144 31.554577 30.586256 29.657757 28.770611 27.924309 27.117098 26.346568 25.610062 24.904945 24.228757 23.579301 22.954671 22.353255 21.773718 21.214975 20.676164 20.156623 19.655863 19.173558 18.709528 18.263732 17.836270 17.427383 17.037461 16.667056 16.316894 15.987903 15.681232 15.398282 15.140745 14.910641 14.710364 14.542738 14.411066 14.319199 14.271597 14.273392 14.330453 14.449439 14.637832 14.903952 15.256918 15.706552 16.263202 16.937433 17.739584 18.679129 19.763820 20.998591 22.384222 23.915803 25.581096 27.358972 29.218175 31.116762 33.002590 34.815200 36.489299 37.959755 39.167650 40.066534 40.627766 40.843807 40.728690 40.315504 39.651430 38.791436 37.791906 36.705310 35.576579 34.441311 33.325560 32.246735 31.215102 30.235478 29.308832 28.433634 27.606913 26.825018 26.084143 25.380643 24.711225 24.073027 23.463647 22.881126 22.323917 21.790843 21.281054 20.794001 20.329394 19.887192 19.467582 19.070974 18.698004 18.349540 18.026692 17.730835 17.463626 17.227037 17.023382 16.855350 16.726047 16.639021 16.598306 16.608440 16.674491 16.802055 16.997234 17.266584 17.617015 18.055632 18.589496 19.225303 19.968936 20.824906 21.795659 22.880760 24.075988 25.372389 26.755398 28.204148 29.691132 31.182414 32.638548 34.016312 35.271273 36.360988 37.248534 37.905865 38.316442 38.476671 38.395834 38.094548 37.602027 36.952673 36.182602 35.326597 34.415859 33.476647 32.529772 31.590752 30.670393 29.775608 28.910280 28.076074 27.273129 26.500609 25.757118 25.040997 24.350516 23.684004 23.039912 22.416847 21.813582 21.229045 20.662314 20.112589 19.579184 19.061509 18.559053 18.071377 17.598103 17.138910 16.693528 16.261739 15.843373 15.438316 15.046505 14.667943 14.302699 13.950920 13.612842 13.288804 12.979264 12.684821 12.406235 12.144460 11.900675 11.676324 11.473165 11.293321 11.139349 11.014307 10.921846 10.866303 10.852805 10.887390 10.977133 11.130268 11.356316 11.666183 12.072234 12.588291 13.229541 14.012279 14.953456 16.069918 17.377284 18.888363 20.611055 22.545731 24.682187 26.996411 29.447620 31.976256 34.503803 36.935351 39.165590 41.088314 42.608538 43.655241 44.191949 44.222368 43.789293 42.966893 41.848367 40.532172 39.109997 37.658543 36.235736 34.880695 33.616202 32.452322 31.390157 30.425142 29.549600 28.754577 28.031032 27.370548 26.765693 26.210164 25.698784 25.227431 24.792921 24.392893 24.025690 23.690249 23.386018 23.112870 22.871045 22.661098 22.483852 22.340363 22.231882 22.159829 22.125752 22.131289 22.178128 22.267947 22.402356 22.582819 22.810561 23.086467 23.410960 23.783870 24.204290 24.670427 25.179457 25.727390 26.308963 26.917570 27.545245 28.182725 28.819588 29.444488 30.045483 30.610448 31.127556 31.585785 31.975431 32.288566 32.519400 32.664523 32.722989 32.696240 32.587893 32.403397 32.149617 31.834379 31.466025 31.053004 30.603534 30.125346 29.625507 29.110322 28.585298 28.055169 27.523943 26.994983 26.471092 25.954607 25.447484 24.951377 24.467715 23.997758 23.542656 23.103492 22.681321 22.277206 21.892246 21.527602 21.184524 20.864371 20.568635 20.298963 20.057175 19.845288 19.665533 19.520371 19.412507 19.344902 19.320770 19.343574 19.417005 19.544941 19.731390 19.980403 20.295948 20.681751 21.141091 21.676535 22.289638 22.980577 23.747757 24.587393 25.493094 26.455505 27.462034 28.496758 29.540535 30.571411 31.565337 32.497206 33.342169 34.077119 34.682217 35.142257 35.447705 35.595264 35.587869 35.434131 35.147318 34.744025 34.242724 33.662392 33.021335 32.336315 31.621995 30.890682 30.152316 29.414632 28.683426 27.962872 27.255845 26.564216 25.889114 25.231138 24.590524 23.967279 23.361266 22.772283 22.200099 21.644492 21.105267 20.582272 20.075403 19.584615 19.109927 18.651428 18.209283 17.783745 17.375163 16.983997 16.610832 16.256397 15.921588 15.607494 15.315426 15.046950 14.803929 14.588569 14.403465 14.251659 14.136704 14.062724 14.034489 14.057477 14.137942 14.282968 14.500500 14.799347 15.189138 15.680197 16.283329 17.009475 17.869189 18.871917 20.025040 21.332645 22.794068 24.402233 26.141933 27.988256 29.905458 31.846671 33.754838 35.565219 37.209608 38.622026 39.745293 40.537414 40.976560 41.063527 40.821036 40.290004 39.523594 38.580375 37.517916 36.387844 35.232832 34.085464 32.968599 31.896667 30.877397 29.913604 29.004769 28.148327 27.340638 26.577659 25.855386 25.170115 24.518582 23.898009 23.306108 22.741053 22.201438 21.686232 21.194738 20.726561 20.281574 19.859904 19.461921 19.088232 18.739686 18.417381 18.122683 17.857243 17.623021 17.422315 17.257792 17.132514 17.049973 17.014117 17.029368 17.100635 17.233304 17.433208 17.706559 18.059841 18.499634 19.032377 19.664034 20.399659 21.242849 22.195085 23.254970 24.417398 25.672715 27.005964 28.396334 29.816963 31.235262 32.613884 33.912435 35.089875 36.107459 36.931890 37.538253 37.912253 38.051363 37.964655 37.671335 37.198280 36.577002 35.840568 35.020911 34.146821 33.242712 32.328141 31.417896 30.522490 29.648854 28.801099 27.981235 27.189796 26.426345 25.689861 24.979013 24.292357 23.628455 22.985944 22.363581 21.760247 21.174953 20.606827 20.055105 19.519109 18.998242 18.491970 17.999815 17.521342 17.056158 16.603900 16.164237 15.736866 15.321507 14.917907 14.525838 14.145097 13.775512 13.416940 13.069271 12.732436 12.406408 12.091212 11.786931 11.493715 11.211797 10.941502 10.683267 10.437663 10.205416 9.987439 9.784867 9.599100 9.431853 9.285215 9.161721 9.064433 8.997038 8.963963 8.970496 9.022937 9.128754 9.296752 9.537253 9.862257 10.285590 10.822989 11.492088 12.312261 13.304219 14.489282 15.888197 kafbox-2.0/data/mg30.dat000066400000000000000000001277101303712747200150140ustar00rootroot000000000000000.890000 0.890000 0.890000 0.890000 0.890000 0.890000 1.100658 1.216270 1.279719 1.314540 1.333651 1.151368 0.817173 0.555632 0.384163 0.278092 0.260419 0.630732 0.928617 0.919939 0.795084 0.664468 0.750585 1.028859 1.115664 1.242806 1.304385 1.313666 1.265603 1.007113 0.723809 0.484018 0.343257 0.259997 0.396604 0.822248 0.973611 0.895334 0.753865 0.687596 0.926911 1.059449 1.129726 1.262279 1.306573 1.344875 1.143268 0.923316 0.656208 0.443266 0.308898 0.289830 0.566884 0.936870 0.991884 0.873831 0.732332 0.776238 1.033923 1.036191 1.122595 1.260039 1.322634 1.266775 1.066802 0.913502 0.651542 0.439907 0.310518 0.392358 0.677828 0.997889 1.022942 0.885922 0.786589 0.911879 1.082927 0.985737 1.065319 1.231158 1.314888 1.108931 1.009139 0.987438 0.740093 0.493814 0.410032 0.625314 0.770240 1.028086 1.097925 0.983623 1.008909 1.170402 1.207880 0.953636 0.906511 0.989617 0.830418 0.587117 0.663395 0.962690 1.042702 1.135655 1.221082 1.196538 1.265891 1.098715 0.920918 0.655640 0.512125 0.393374 0.394691 0.647967 0.980580 1.047180 0.978658 0.874088 0.941457 1.111555 1.000726 0.967238 1.110088 1.218909 1.019677 0.909743 1.008339 0.937218 0.685156 0.614388 0.896820 1.004995 1.011290 1.182616 1.195311 1.278019 1.185348 1.112186 0.878088 0.635810 0.469521 0.376108 0.424071 0.673637 0.989272 1.032561 0.931924 0.868515 0.951784 1.108238 1.003062 1.032685 1.178062 1.251209 1.024757 0.919550 0.953820 0.785127 0.555459 0.529861 0.838546 1.000425 1.157831 1.212373 1.121211 1.212940 1.205392 0.976007 0.679401 0.568248 0.513546 0.405840 0.514006 0.900456 1.032038 1.060783 0.986018 0.930453 1.112015 1.078938 0.944802 0.920929 1.052002 0.987593 0.796794 0.879376 1.067184 1.052917 0.935981 1.128149 1.265691 1.159858 0.938137 1.017140 0.949082 0.663956 0.486707 0.677178 0.886378 0.909102 1.121008 1.107558 1.108983 1.242283 1.298439 1.103804 0.838481 0.754104 0.578930 0.410252 0.364437 0.728982 1.049139 1.162681 1.072466 0.915817 0.986220 1.080378 0.826607 0.694941 0.857706 1.043770 0.887032 1.008810 1.182584 1.285154 1.182133 1.148105 1.200978 0.914246 0.621188 0.448814 0.478892 0.414431 0.584912 0.938230 0.973299 0.952296 0.925851 0.926401 1.118060 1.107152 1.133462 1.167802 1.232694 1.055575 0.831436 0.703461 0.592598 0.471214 0.454760 0.812584 1.085737 1.167220 1.115249 1.005543 1.109409 1.079422 0.807719 0.652359 0.716946 0.777134 0.653526 0.900813 1.108804 1.201069 1.306234 1.326064 1.351295 1.130455 0.817744 0.552124 0.369896 0.266640 0.261619 0.648139 0.938688 0.915495 0.781881 0.650191 0.756463 1.027504 1.110985 1.244372 1.296737 1.306385 1.257355 1.012142 0.726944 0.487215 0.349696 0.267203 0.399811 0.820676 0.974885 0.900730 0.762770 0.698528 0.932360 1.062098 1.126464 1.259175 1.311290 1.347768 1.137637 0.920816 0.658831 0.444218 0.307190 0.291597 0.575848 0.942758 0.996924 0.875874 0.732924 0.782505 1.036868 1.028034 1.112985 1.254642 1.321108 1.258901 1.069207 0.931740 0.668792 0.451192 0.318780 0.401001 0.662497 0.989917 1.030687 0.898253 0.801974 0.917816 1.094474 0.988806 1.050716 1.220169 1.305239 1.094838 0.984600 0.988058 0.757814 0.508779 0.428137 0.674670 0.821063 1.050137 1.122958 1.010924 1.053373 1.218767 1.203849 0.915643 0.840362 0.891335 0.705271 0.501020 0.655907 0.998592 1.166790 1.277957 1.222515 1.157520 1.224804 0.947799 0.645727 0.455362 0.435767 0.377806 0.519662 0.904704 0.970995 0.925758 0.877535 0.855406 1.074311 1.104838 1.149426 1.228269 1.314229 1.208530 0.936541 0.755442 0.575617 0.408467 0.310597 0.540743 0.929666 1.090938 1.032654 0.876143 0.841651 1.067248 0.964219 0.851325 0.997329 1.193997 1.148581 1.017782 1.169290 1.219613 0.933764 0.658273 0.730495 0.688747 0.511773 0.602296 0.951905 1.119994 1.246949 1.211193 1.129086 1.226596 1.043757 0.728311 0.515210 0.486545 0.430208 0.453126 0.835616 0.992419 0.983389 0.960116 0.903288 1.061320 1.113323 1.081763 1.087631 1.168744 1.142340 0.882669 0.773557 0.731200 0.639138 0.518644 0.740785 1.053628 1.218178 1.274581 1.206565 1.200233 1.193580 0.862604 0.577390 0.430315 0.405548 0.345476 0.605477 0.930485 0.946635 0.893862 0.823706 0.850965 1.081529 1.123004 1.197910 1.291297 1.359178 1.203510 0.925264 0.702108 0.496975 0.337776 0.265205 0.525534 0.921926 1.031328 0.931273 0.768024 0.756825 1.019474 0.998497 1.020553 1.192692 1.287970 1.272932 1.102074 1.070377 0.848059 0.576217 0.391905 0.412244 0.521815 0.782907 1.027417 0.983474 0.885364 0.912938 1.065072 1.106673 1.023627 1.133987 1.232726 1.155446 0.877596 0.848776 0.785332 0.587353 0.460924 0.706893 1.031019 1.211250 1.259038 1.145152 1.127590 1.178906 0.877178 0.588353 0.473565 0.521408 0.450263 0.652746 0.961570 0.985014 0.990897 0.981955 1.003574 1.151863 1.107543 1.089523 1.060611 1.077995 0.866695 0.710826 0.669132 0.692553 0.698199 0.861775 1.112056 1.213271 1.264781 1.308784 1.352109 1.161904 0.815676 0.563255 0.393490 0.281813 0.250660 0.623327 0.927348 0.927640 0.805778 0.667124 0.743217 1.025273 1.107916 1.232715 1.302070 1.309948 1.269493 1.016972 0.740420 0.496456 0.351340 0.264461 0.389277 0.810884 0.979316 0.907224 0.766171 0.693638 0.921654 1.056200 1.113650 1.250311 1.305751 1.346568 1.147958 0.943069 0.682609 0.460753 0.318378 0.292005 0.548581 0.927114 1.005351 0.892724 0.747801 0.778313 1.033873 1.027402 1.092937 1.243021 1.318102 1.269387 1.069943 0.953429 0.699318 0.471099 0.327879 0.400929 0.645764 0.977844 1.045184 0.919190 0.815878 0.920408 1.105640 0.987113 1.018541 1.194882 1.288168 1.082352 0.962892 1.012074 0.812913 0.551903 0.462909 0.729441 0.852083 1.026010 1.144589 1.056044 1.114741 1.258161 1.246063 0.939541 0.793101 0.769084 0.576688 0.408567 0.546930 0.937336 1.165522 1.228771 1.103097 1.001118 1.154501 0.959003 0.661329 0.549004 0.702333 0.690956 0.698725 1.003318 1.069513 1.140146 1.255011 1.272892 1.278271 1.056890 0.862095 0.617237 0.437864 0.320467 0.385647 0.733251 1.019461 1.019375 0.891994 0.787494 0.933903 1.063571 0.968609 1.055125 1.224105 1.304713 1.091964 1.040485 1.024649 0.771748 0.514746 0.443780 0.619671 0.702201 0.961481 1.080814 0.996977 1.033932 1.158105 1.253165 1.041884 0.955186 0.972966 0.810132 0.576306 0.537000 0.793507 0.931731 1.103911 1.199534 1.128326 1.208545 1.276155 1.121399 0.791613 0.628843 0.545017 0.402908 0.361634 0.722417 0.996594 1.076720 1.010201 0.881611 0.957712 1.100283 0.961145 0.886640 1.035296 1.175388 0.979158 0.930778 1.107180 1.129334 0.858986 0.780606 1.003316 0.953227 0.733669 0.860698 1.117469 1.184467 1.085828 1.224039 1.308774 1.137699 0.793694 0.694187 0.580646 0.415113 0.343507 0.705643 1.010313 1.129271 1.059013 0.903279 0.952513 1.095362 0.893369 0.762759 0.911706 1.104369 0.946083 0.973302 1.179178 1.286081 1.095779 0.985745 1.113680 0.904148 0.609612 0.486350 0.725580 0.729585 0.806682 1.056499 1.043270 1.114670 1.253441 1.318932 1.216231 1.009247 0.893021 0.646976 0.440688 0.325497 0.483177 0.790895 1.060838 1.056570 0.909034 0.846579 1.034218 1.068609 0.898902 0.976166 1.171669 1.188761 0.938161 1.020554 1.138828 0.926545 0.649719 0.735223 0.926375 0.782621 0.799572 1.060526 1.161570 1.250744 1.309397 1.358456 1.245473 0.942114 0.646551 0.448541 0.308592 0.236797 0.455151 0.869830 0.956185 0.855723 0.703991 0.672463 0.958860 1.074668 1.176268 1.286251 1.294339 1.324667 1.106818 0.853061 0.583444 0.402832 0.291521 0.308584 0.658146 0.963510 0.957477 0.830942 0.708379 0.815703 1.047217 1.066045 1.187062 1.290122 1.339742 1.227173 1.023491 0.805408 0.551456 0.375081 0.283129 0.429140 0.803867 1.021500 0.965006 0.812091 0.751617 0.961213 1.043716 1.015078 1.164717 1.282618 1.325055 1.103946 1.031086 0.864258 0.595919 0.397387 0.375353 0.559318 0.820278 1.058696 1.013397 0.881395 0.910576 1.095587 1.079924 0.952022 1.072576 1.208511 1.105216 0.847620 0.927138 0.954061 0.725085 0.569798 0.835846 1.082619 1.099854 1.226349 1.225442 1.265181 1.149727 0.911373 0.692954 0.497477 0.397794 0.340674 0.622401 0.975691 1.051243 0.975462 0.853006 0.887518 1.086314 0.977007 0.959456 1.112761 1.252470 1.105852 0.996940 1.081288 0.982075 0.695632 0.535228 0.714599 0.762520 0.775491 1.040552 1.092767 1.154330 1.277531 1.341600 1.281964 1.001647 0.808297 0.574139 0.387376 0.274700 0.413099 0.817201 1.041839 0.991511 0.830551 0.740651 0.959162 1.019694 0.956623 1.108654 1.250920 1.308416 1.115141 1.113703 1.013589 0.715238 0.474675 0.419564 0.503421 0.579074 0.919363 1.019601 0.938767 0.940327 0.993365 1.155236 1.081674 1.085308 1.159393 1.146953 0.941774 0.753436 0.743132 0.640579 0.543756 0.652539 0.994449 1.186079 1.259334 1.212737 1.177138 1.237778 0.949160 0.640537 0.468304 0.426796 0.360724 0.496924 0.890416 0.968601 0.927682 0.864989 0.829522 1.056245 1.108805 1.147716 1.233038 1.323988 1.248464 0.963924 0.768628 0.582299 0.407697 0.296607 0.477853 0.884839 1.073183 1.024867 0.866706 0.798525 1.029261 0.991577 0.885182 1.026416 1.212705 1.226842 1.051305 1.153888 1.170956 0.874004 0.588505 0.591387 0.608323 0.496434 0.683531 0.977036 1.041743 1.127260 1.107886 1.105332 1.195920 1.081728 0.884650 0.722039 0.692028 0.556149 0.528609 0.770272 1.063896 1.199983 1.214558 1.137751 1.187898 1.164859 0.853643 0.599905 0.511518 0.490990 0.408319 0.670072 0.972450 1.017450 1.014958 0.952812 0.981681 1.133327 1.059674 1.000773 1.023579 1.107203 0.907643 0.780857 0.835818 0.906464 0.810250 0.864148 1.118393 1.264321 1.309929 1.345395 1.387399 1.184762 0.800117 0.519753 0.353616 0.244693 0.214657 0.595109 0.886862 0.867548 0.737764 0.598490 0.673405 0.996322 1.157190 1.280555 1.285542 1.246049 1.272434 0.995248 0.673272 0.451601 0.354134 0.284825 0.401886 0.831764 0.942413 0.869191 0.761587 0.701048 0.944393 1.093719 1.191171 1.300642 1.335835 1.357439 1.114276 0.829430 0.563226 0.376658 0.263188 0.275213 0.655095 0.949897 0.929944 0.790189 0.657900 0.772097 1.030400 1.091847 1.226869 1.290590 1.311948 1.249574 1.024700 0.756956 0.510422 0.362037 0.274875 0.401117 0.807429 0.989013 0.924150 0.783368 0.716308 0.934425 1.056102 1.091264 1.231086 1.306336 1.346910 1.135330 0.957888 0.716484 0.484560 0.329698 0.306655 0.556385 0.927894 1.030408 0.921134 0.773581 0.806468 1.049985 1.008009 1.037125 1.206082 1.307985 1.240704 1.054683 1.022072 0.798393 0.538437 0.373467 0.462987 0.629479 0.908574 1.070815 0.980699 0.898587 0.994417 1.167572 1.044519 0.964424 1.103068 1.152438 0.928297 0.755994 0.904590 0.884866 0.693871 0.745210 1.048153 1.212401 1.241539 1.318066 1.324471 1.266557 0.906192 0.624785 0.428737 0.305955 0.236212 0.466376 0.872806 0.940457 0.836754 0.695146 0.664711 0.961701 1.088488 1.203959 1.299187 1.294424 1.322241 1.094612 0.814546 0.547740 0.380431 0.279727 0.308778 0.691970 0.959470 0.929787 0.801202 0.686248 0.821382 1.048626 1.098826 1.228865 1.302826 1.340886 1.221245 0.992075 0.734039 0.494512 0.341956 0.267106 0.445052 0.853032 0.999824 0.914211 0.762979 0.722060 0.970268 1.043054 1.085048 1.233334 1.302869 1.329446 1.115933 0.969218 0.722016 0.488794 0.333517 0.332705 0.568578 0.929449 1.034845 0.926754 0.789703 0.836247 1.067811 1.012023 1.030498 1.199966 1.308831 1.210559 1.020278 1.007606 0.800743 0.540773 0.383461 0.522383 0.706824 0.957628 1.099813 1.000437 0.941202 1.079059 1.212571 1.006260 0.903615 1.030981 1.001349 0.739143 0.670787 0.936730 1.014320 0.933582 1.113686 1.198852 1.280272 1.139772 1.144827 1.042013 0.755645 0.524778 0.434676 0.475379 0.509134 0.865365 1.030097 0.976235 0.951997 0.953538 1.111846 1.085043 1.034810 1.092440 1.134951 1.028854 0.807451 0.816339 0.781656 0.679024 0.644773 0.944863 1.170271 1.293476 1.337780 1.304989 1.332586 1.047836 0.691677 0.447207 0.314725 0.241634 0.302028 0.757656 0.907291 0.828344 0.701069 0.606161 0.822328 1.070720 1.210975 1.304275 1.285176 1.305192 1.191030 0.867009 0.572954 0.393295 0.304902 0.262898 0.562366 0.906659 0.916122 0.811237 0.689709 0.725973 1.014991 1.122827 1.245263 1.317522 1.324036 1.297599 1.023202 0.727161 0.482403 0.335095 0.247599 0.357970 0.794056 0.958794 0.882044 0.738438 0.655065 0.883986 1.059447 1.148115 1.274001 1.297503 1.337178 1.170675 0.919526 0.638953 0.432995 0.310548 0.274014 0.540226 0.918540 0.969317 0.858734 0.719920 0.744815 1.016965 1.063735 1.163298 1.281500 1.322132 1.293830 1.066925 0.851750 0.590047 0.401239 0.284615 0.356673 0.711035 0.995670 0.978261 0.835052 0.728652 0.877679 1.053728 1.023278 1.149888 1.273159 1.340130 1.169737 1.037480 0.879374 0.613042 0.410612 0.327895 0.486110 0.769152 1.037939 1.016089 0.871241 0.840026 1.016477 1.081783 0.964516 1.076298 1.236483 1.242182 0.974116 0.974644 0.970942 0.720151 0.495072 0.575971 0.850690 0.925008 1.123677 1.140986 1.074354 1.210352 1.270219 1.094019 0.800989 0.728975 0.620696 0.448842 0.400755 0.768640 1.059542 1.184533 1.120335 0.975953 1.050603 1.093148 0.818787 0.640949 0.733460 0.878265 0.737677 0.922265 1.116937 1.203905 1.297002 1.354389 1.377629 1.120267 0.808639 0.545810 0.365360 0.251063 0.254614 0.657514 0.938457 0.912521 0.769759 0.631848 0.750672 1.024291 1.112184 1.246922 1.287505 1.292088 1.253288 1.011447 0.723824 0.487162 0.357253 0.275701 0.405758 0.824612 0.975435 0.904845 0.772774 0.710480 0.942319 1.065659 1.125811 1.257218 1.316678 1.349121 1.127909 0.912930 0.656668 0.442116 0.303779 0.295680 0.593638 0.953193 1.001166 0.875390 0.732219 0.793473 1.041210 1.018404 1.105108 1.250200 1.320581 1.245570 1.071045 0.949352 0.684348 0.461155 0.327738 0.413749 0.648722 0.980092 1.036016 0.908668 0.819194 0.927189 1.107105 0.996559 1.041691 1.210569 1.292755 1.076708 0.952612 0.974469 0.764002 0.518145 0.447934 0.732189 0.895461 1.095107 1.154094 1.039809 1.101415 1.241358 1.133965 0.827821 0.744989 0.770844 0.589119 0.463172 0.755321 1.055882 1.227375 1.269146 1.149267 1.154169 1.158123 0.839318 0.559600 0.451964 0.488962 0.425712 0.688224 0.964453 0.964617 0.958147 0.933004 0.985729 1.136363 1.119323 1.140777 1.161577 1.190271 0.947354 0.759398 0.646936 0.562666 0.472408 0.590744 0.957517 1.135181 1.160236 1.098775 1.051448 1.182762 0.976277 0.732160 0.631515 0.691983 0.616283 0.638240 0.972485 1.120894 1.199481 1.251225 1.217374 1.271878 1.028094 0.756559 0.534364 0.418758 0.334489 0.402234 0.813904 1.004629 0.967729 0.868431 0.783962 0.974867 1.063978 1.046574 1.152831 1.281028 1.316634 1.076467 0.959527 0.810977 0.574473 0.386430 0.389921 0.659174 0.963950 1.123197 1.034977 0.893121 0.966110 1.142927 0.947063 0.815840 0.978348 1.124280 0.931734 0.868646 1.108279 1.207963 0.981090 0.913207 1.135942 1.063177 0.751354 0.701307 0.981945 0.916704 0.729424 0.972602 1.134920 1.221609 1.145717 1.265646 1.307504 1.041472 0.721331 0.590157 0.463623 0.332464 0.375536 0.797193 1.001107 1.023928 0.912192 0.793890 0.964916 1.080834 0.994385 1.046297 1.213538 1.285692 1.051254 0.968469 0.973985 0.759724 0.514269 0.483854 0.745217 0.890483 1.094180 1.150169 1.050286 1.131296 1.260140 1.147328 0.838217 0.749464 0.729551 0.543619 0.424944 0.719658 1.039014 1.212161 1.226938 1.089818 1.089925 1.147317 0.851172 0.586001 0.523987 0.630239 0.545186 0.748400 1.009318 1.031472 1.093696 1.127989 1.165506 1.191055 1.053508 0.925696 0.749593 0.641660 0.494967 0.534653 0.760691 1.053444 1.184609 1.150076 1.078951 1.163377 1.172318 0.867308 0.649030 0.628944 0.606997 0.479398 0.693228 1.001855 1.108691 1.171501 1.121654 1.111359 1.185572 0.982998 0.748531 0.611473 0.621496 0.515110 0.593498 0.941293 1.100393 1.154874 1.144239 1.101556 1.214886 1.033813 0.798542 0.632298 0.607225 0.523767 0.521765 0.864608 1.082691 1.140565 1.142037 1.071786 1.181184 1.083598 0.843106 0.668485 0.643913 0.609509 0.531037 0.813698 1.075123 1.161739 1.210586 1.156615 1.212259 1.140944 0.861923 0.630045 0.510866 0.466145 0.398765 0.683626 0.994278 1.042125 1.015279 0.933470 0.979465 1.118417 1.003248 0.947102 1.015933 1.118517 0.919992 0.854052 0.981028 1.036884 0.871752 0.881870 1.118406 1.185306 1.051021 1.091847 1.243432 1.096662 0.778243 0.683837 0.743458 0.580996 0.476958 0.810768 1.057108 1.210246 1.251132 1.138011 1.180644 1.146152 0.847468 0.576226 0.476014 0.487972 0.422926 0.689713 0.973817 0.988107 0.984606 0.940937 0.992437 1.134732 1.088710 1.076926 1.101686 1.147181 0.917659 0.767457 0.727179 0.702421 0.607956 0.722207 1.037944 1.206374 1.286491 1.286780 1.268347 1.250499 0.910933 0.606507 0.414876 0.326766 0.264656 0.484175 0.877491 0.924700 0.832911 0.717809 0.699162 0.985473 1.113295 1.229465 1.318229 1.327306 1.329501 1.062382 0.763544 0.507422 0.346006 0.249825 0.312941 0.740207 0.953592 0.894208 0.751682 0.644797 0.830719 1.049079 1.134346 1.264862 1.294664 1.325426 1.206870 0.955587 0.670509 0.452399 0.326643 0.267364 0.479950 0.885253 0.972531 0.875668 0.735688 0.720511 0.989956 1.064666 1.150285 1.275697 1.317455 1.322174 1.091513 0.875051 0.612152 0.414659 0.289825 0.326220 0.662494 0.979333 0.984384 0.847709 0.723848 0.837387 1.051052 1.025700 1.139847 1.267941 1.335128 1.204869 1.048338 0.894329 0.627762 0.422172 0.317795 0.448983 0.731276 1.022025 1.018705 0.875498 0.816449 0.974390 1.086763 0.975563 1.074410 1.237178 1.283090 1.028170 0.982813 0.970377 0.721269 0.486148 0.488605 0.757792 0.873333 1.095212 1.123364 1.023311 1.129333 1.264911 1.151139 0.859742 0.810959 0.783961 0.573435 0.436038 0.713766 1.042210 1.222303 1.260471 1.124918 1.111266 1.159734 0.850238 0.568954 0.474285 0.554850 0.484641 0.708220 0.980923 0.986283 1.010274 1.022932 1.066230 1.166840 1.104045 1.071920 0.992300 0.947170 0.732490 0.636850 0.641575 0.755443 0.912427 1.126081 1.207548 1.229265 1.282536 1.337674 1.130713 0.791006 0.566656 0.418941 0.299422 0.279110 0.675866 0.951096 0.956837 0.841240 0.705217 0.807377 1.049705 1.081935 1.190160 1.292771 1.335356 1.228401 1.001466 0.780737 0.536100 0.367506 0.279047 0.444337 0.838083 1.027505 0.958342 0.803802 0.749205 0.979144 1.029902 1.011520 1.169005 1.282821 1.314450 1.101783 1.048072 0.870166 0.597449 0.399169 0.387324 0.545303 0.793984 1.044978 1.006970 0.884912 0.912711 1.082002 1.098506 0.978846 1.090490 1.213778 1.123058 0.851700 0.886257 0.890975 0.677387 0.533026 0.793390 1.075280 1.177195 1.279990 1.223080 1.236312 1.169280 0.866995 0.594038 0.429729 0.378915 0.327809 0.620778 0.948371 0.963020 0.890334 0.796884 0.842551 1.070428 1.086292 1.167664 1.281861 1.353587 1.208629 0.970242 0.774182 0.550615 0.370481 0.282982 0.494057 0.885434 1.060569 0.984538 0.819220 0.781798 1.023464 0.995084 0.938197 1.110180 1.257058 1.258434 1.073675 1.131950 1.032064 0.723151 0.483587 0.495602 0.551233 0.571164 0.905514 1.016110 0.976699 1.022796 1.052200 1.185663 1.111697 1.078490 1.044307 0.952956 0.775189 0.621284 0.644337 0.682759 0.834539 1.071244 1.179562 1.210463 1.251014 1.322986 1.227746 0.885897 0.641326 0.475225 0.347109 0.269307 0.536666 0.917409 0.993991 0.909248 0.765100 0.753489 1.022777 1.048755 1.100253 1.242740 1.313445 1.282715 1.063084 0.918881 0.680307 0.460995 0.320976 0.389089 0.685324 1.006790 1.049261 0.915062 0.802780 0.925146 1.086593 0.952363 1.001073 1.188249 1.286039 1.077109 1.018348 1.084792 0.868732 0.585235 0.490867 0.695341 0.711489 0.855468 1.073190 1.042272 1.111298 1.235330 1.313928 1.164249 0.975105 0.872522 0.647970 0.447894 0.352131 0.569477 0.876888 1.109120 1.088491 0.938399 0.918486 1.119489 1.021447 0.813373 0.881304 1.080894 0.998061 0.822433 1.042206 1.218435 1.117303 0.938179 1.125794 1.163275 0.851674 0.626512 0.805581 0.836199 0.629949 0.784826 1.037800 1.176077 1.277177 1.316811 1.310063 1.236542 0.947625 0.640461 0.425264 0.312657 0.252111 0.450943 0.865668 0.937590 0.837047 0.704463 0.677938 0.960274 1.091429 1.207897 1.303380 1.308034 1.332315 1.097928 0.811195 0.543736 0.373621 0.270785 0.298282 0.686311 0.953765 0.922075 0.789325 0.670987 0.805749 1.044083 1.106698 1.238549 1.300925 1.331643 1.230219 0.992561 0.722893 0.485936 0.340653 0.264977 0.435438 0.849115 0.989147 0.904261 0.757011 0.712108 0.961553 1.051027 1.106753 1.247874 1.306537 1.335381 1.120401 0.942148 0.685863 0.463861 0.318874 0.317248 0.581423 0.945441 1.018311 0.900835 0.762450 0.817074 1.055779 1.012705 1.068501 1.228986 1.320081 1.228140 1.048553 0.981569 0.737853 0.495161 0.350891 0.462149 0.667515 0.975137 1.069125 0.950434 0.873652 0.993352 1.151434 0.988346 0.970400 1.139486 1.189653 0.937062 0.844107 0.993658 0.887517 0.650926 0.695172 1.017696 1.112556 1.121945 1.241512 1.244009 1.257776 0.973800 0.813649 0.600822 0.447141 0.337700 0.474133 0.871897 1.085112 1.053540 0.924148 0.843504 1.051224 1.000146 0.856570 0.941529 1.147156 1.159149 0.982055 1.124177 1.231506 1.027064 0.726341 0.781811 0.795702 0.592581 0.545593 0.901285 1.121800 1.265262 1.293704 1.190667 1.256878 1.087626 0.744855 0.490127 0.390292 0.352775 0.356923 0.755274 0.944562 0.898500 0.832315 0.758849 0.911662 1.096406 1.170934 1.274511 1.348446 1.379939 1.141937 0.855396 0.597407 0.398607 0.269572 0.254599 0.611868 0.944133 0.952699 0.814055 0.665517 0.743219 1.017689 1.064722 1.194164 1.280551 1.299252 1.267813 1.063766 0.822369 0.559704 0.394154 0.290961 0.371685 0.735987 0.990274 0.959684 0.825142 0.734974 0.897080 1.060533 1.054719 1.182859 1.291946 1.349524 1.155623 0.994755 0.800907 0.550691 0.369404 0.311419 0.521505 0.871198 1.057527 0.981875 0.826618 0.822425 1.049063 1.019439 0.956473 1.121224 1.266836 1.229719 1.011753 1.065247 0.970288 0.680014 0.464563 0.558471 0.699456 0.759820 1.035284 1.060972 1.018547 1.140597 1.249242 1.235354 1.002012 0.955597 0.836265 0.609706 0.428868 0.525951 0.800148 1.025824 1.174980 1.095938 1.005935 1.145972 1.187459 0.892635 0.703404 0.780082 0.757928 0.568218 0.703598 1.021356 1.190904 1.304321 1.290549 1.240756 1.245176 0.938776 0.620592 0.414306 0.339073 0.281996 0.466300 0.870364 0.924670 0.837035 0.735289 0.711049 0.984697 1.114512 1.229650 1.321512 1.340891 1.339076 1.065827 0.764481 0.507568 0.342093 0.243519 0.303930 0.733179 0.950190 0.890717 0.744730 0.634081 0.817706 1.045709 1.136674 1.266609 1.289364 1.315994 1.212890 0.958715 0.669725 0.452939 0.331468 0.269358 0.473099 0.880620 0.969639 0.876723 0.740051 0.719878 0.987432 1.069021 1.154272 1.278193 1.320208 1.325955 1.090833 0.867286 0.605562 0.409568 0.286046 0.322198 0.667746 0.979489 0.979402 0.841126 0.716468 0.834063 1.048259 1.029605 1.148812 1.271018 1.334548 1.208201 1.049699 0.882753 0.615833 0.415433 0.313206 0.441227 0.735583 1.020495 1.009896 0.866047 0.805377 0.966064 1.080255 0.983445 1.091686 1.248461 1.297504 1.048173 0.992542 0.949778 0.693715 0.464667 0.454574 0.709875 0.858541 1.095316 1.103467 0.989503 1.077254 1.235994 1.132373 0.865834 0.861522 0.891376 0.676752 0.510008 0.766422 1.065138 1.187333 1.285566 1.213185 1.212884 1.179937 0.867426 0.588998 0.428281 0.396548 0.340643 0.616184 0.943628 0.956712 0.894671 0.815656 0.853185 1.078069 1.100578 1.178612 1.282822 1.354480 1.197904 0.945954 0.741537 0.528741 0.357844 0.279429 0.521166 0.914946 1.056710 0.967754 0.802548 0.782772 1.032314 0.983797 0.954307 1.134432 1.268212 1.248721 1.083909 1.134194 0.992141 0.685204 0.461650 0.483567 0.528546 0.595019 0.938827 1.009431 0.961509 0.996500 1.034271 1.177536 1.093085 1.090472 1.087558 1.023756 0.818293 0.672112 0.674427 0.652635 0.708362 0.940525 1.136232 1.220378 1.256560 1.285783 1.333158 1.069744 0.753209 0.527279 0.385936 0.282246 0.312941 0.744022 0.960541 0.928246 0.804312 0.686960 0.855611 1.059490 1.109535 1.234265 1.306488 1.346598 1.192153 0.958222 0.706392 0.476760 0.329936 0.271082 0.497868 0.897046 1.006089 0.904882 0.752729 0.742908 1.005363 1.030172 1.082942 1.235033 1.305866 1.302432 1.096900 0.974587 0.722943 0.488909 0.335094 0.364084 0.588678 0.937996 1.039925 0.929083 0.806627 0.871467 1.088609 1.011926 1.024598 1.194667 1.303167 1.163478 0.981292 0.993905 0.801603 0.543589 0.404739 0.609863 0.799955 1.015671 1.133344 1.026874 1.008475 1.177561 1.221962 0.941477 0.822053 0.912677 0.785240 0.556322 0.637942 0.984143 1.161816 1.259596 1.266923 1.194536 1.254096 0.973289 0.676791 0.460688 0.390882 0.331195 0.452357 0.867568 0.968120 0.902769 0.821890 0.774356 1.009276 1.089805 1.154408 1.265479 1.343436 1.322512 1.050141 0.826941 0.593155 0.399406 0.275112 0.348720 0.743230 1.012095 0.989208 0.835758 0.716682 0.886387 1.038208 0.989767 1.124203 1.256078 1.327598 1.165434 1.084738 0.957656 0.673767 0.451416 0.358265 0.464145 0.641226 0.973240 1.023828 0.906325 0.867489 0.966164 1.131200 1.027770 1.076109 1.217181 1.254514 1.009974 0.869445 0.864947 0.676502 0.479458 0.496610 0.861126 1.095475 1.237471 1.183318 1.057546 1.171215 1.084743 0.766595 0.551248 0.572617 0.591009 0.521142 0.836301 1.020021 1.048533 1.117997 1.095743 1.180305 1.152498 1.021300 0.856089 0.727864 0.644614 0.513908 0.595803 0.874467 1.123202 1.224884 1.176969 1.131910 1.237658 1.082263 0.757083 0.558561 0.523546 0.458406 0.420633 0.792578 0.997252 1.025851 1.011957 0.930247 1.044069 1.121290 1.035424 0.975465 1.041716 1.080974 0.848333 0.784014 0.886688 0.938629 0.811853 0.960712 1.177230 1.288349 1.268858 1.315487 1.342476 1.036871 0.682692 0.462623 0.341583 0.247402 0.312449 0.768564 0.915165 0.857273 0.729189 0.626794 0.844948 1.078465 1.200114 1.304031 1.302453 1.326324 1.181249 0.864215 0.576716 0.390289 0.291619 0.255663 0.570751 0.913194 0.920325 0.805367 0.675855 0.721489 1.012141 1.113119 1.239346 1.308872 1.312153 1.292277 1.032189 0.740880 0.493350 0.346456 0.257388 0.358771 0.786798 0.964531 0.895128 0.755268 0.670323 0.887250 1.059027 1.134472 1.264844 1.303664 1.344158 1.170464 0.931554 0.657488 0.443520 0.312531 0.274108 0.531794 0.916569 0.981033 0.869950 0.726528 0.745994 1.015630 1.051107 1.140887 1.269733 1.318249 1.294271 1.076955 0.888081 0.624720 0.423748 0.297693 0.358486 0.676339 0.991276 1.000611 0.862733 0.751135 0.874737 1.063850 1.005944 1.109193 1.255221 1.333851 1.164053 1.036680 0.936699 0.672682 0.448269 0.350185 0.512490 0.734586 1.028966 1.052525 0.916684 0.887626 1.044186 1.128471 0.955142 1.005578 1.170361 1.151980 0.872903 0.881219 1.004129 0.831036 0.620235 0.785691 1.073425 1.095838 1.154447 1.241481 1.268498 1.194040 0.919014 0.774565 0.560365 0.421232 0.331803 0.580607 0.954233 1.101440 1.036383 0.895120 0.879290 1.089098 0.943879 0.834712 0.971271 1.167646 1.078263 0.990488 1.169564 1.243097 0.985513 0.743329 0.868860 0.778198 0.553639 0.561207 0.930941 1.154709 1.270191 1.271679 1.163716 1.245537 1.023305 0.697414 0.468096 0.404268 0.370467 0.419292 0.832161 0.959093 0.904069 0.852280 0.798684 0.997954 1.107973 1.169503 1.264452 1.343612 1.333001 1.043845 0.798172 0.570906 0.388234 0.267883 0.340681 0.758829 1.004858 0.971488 0.818448 0.698744 0.881923 1.035722 1.014655 1.156399 1.267192 1.330017 1.174921 1.068195 0.894676 0.619595 0.419469 0.335893 0.453675 0.711394 1.007876 1.006710 0.874326 0.832160 0.971354 1.099573 1.006809 1.103333 1.250855 1.287769 1.027577 0.939176 0.889011 0.653765 0.444784 0.460574 0.778536 0.992253 1.173648 1.121929 0.992460 1.105238 1.190806 0.928443 0.700986 0.759647 0.809505 0.619888 0.683293 1.007163 1.172303 1.293068 1.321560 1.270550 1.276181 0.979547 0.653641 0.427843 0.322587 0.258501 0.400304 0.833504 0.925223 0.833912 0.715096 0.661778 0.926517 1.096088 1.217893 1.311383 1.312616 1.343657 1.121999 0.812877 0.539643 0.367578 0.267519 0.273444 0.656501 0.935514 0.907197 0.777372 0.653263 0.767471 1.034672 1.125414 1.255415 1.302427 1.315016 1.252214 0.991289 0.701355 0.468973 0.335969 0.258935 0.415875 0.840861 0.968033 0.883129 0.742919 0.689547 0.942865 1.065265 1.146960 1.273535 1.309635 1.342244 1.128950 0.896005 0.626965 0.424180 0.297233 0.293187 0.601821 0.950824 0.978772 0.853234 0.716811 0.785848 1.037959 1.044056 1.149743 1.272878 1.327092 1.253301 1.056955 0.874805 0.610890 0.414156 0.298873 0.398692 0.718964 1.009639 1.000989 0.857096 0.769424 0.922010 1.068018 0.994987 1.108049 1.256669 1.326529 1.112444 1.023504 0.939247 0.674241 0.447964 0.384842 0.584008 0.776921 1.051835 1.065121 0.935084 0.953774 1.122277 1.140984 0.933236 0.972785 1.104420 0.984691 0.722102 0.815770 1.010450 0.907571 0.833731 1.075183 1.221871 1.207316 1.154966 1.275601 1.194947 0.856506 0.586545 0.522236 0.420194 0.329675 0.606279 0.930343 0.998152 0.974299 0.859300 0.865132 1.089901 1.090830 1.059991 1.161089 1.283454 1.145898 0.902252 0.825153 0.724658 0.521587 0.405700 0.671675 1.007038 1.199446 1.205834 1.057916 1.048165 1.159246 0.885128 0.617449 0.578136 0.723061 0.625964 0.747986 1.023545 1.075075 1.176681 1.247969 1.259464 1.234866 1.024829 0.806938 0.570078 0.426286 0.326963 0.439602 0.811764 1.033550 1.004024 0.880368 0.809830 0.999515 1.051996 0.974931 1.080602 1.239775 1.271846 1.035117 1.020782 0.978511 0.721479 0.486351 0.504105 0.707741 0.807802 1.056534 1.101672 1.021295 1.117190 1.255006 1.214625 0.936007 0.872885 0.822944 0.610247 0.433064 0.600172 0.947225 1.152138 1.244923 1.134776 1.058058 1.191837 0.990357 0.678085 0.529443 0.614998 0.563831 0.583758 0.933649 1.029831 1.074266 1.133422 1.108604 1.217900 1.110255 0.975082 0.788872 0.683734 0.561954 0.493884 0.654170 0.968699 1.156345 1.194350 1.111899 1.117017 1.222339 0.974564 0.695455 0.587459 0.599563 0.494895 0.552357 0.924170 1.060545 1.116681 1.108635 1.047450 1.181655 1.071701 0.876977 0.723741 0.729393 0.654578 0.581359 0.822234 1.093443 1.223176 1.290136 1.239903 1.279579 1.164869 0.824842 0.550411 0.400772 0.331889 0.291443 0.632122 0.928390 0.921664 0.833433 0.725875 0.794135 1.052572 1.130633 1.241602 1.325964 1.358281 1.254013 0.970244 0.696680 0.465619 0.316346 0.239234 0.428849 0.856097 0.975145 0.875853 0.719720 0.672172 0.943562 1.051236 1.137881 1.266999 1.290273 1.328762 1.135752 0.920881 0.645447 0.441860 0.314323 0.304301 0.576740 0.939994 0.988057 0.874507 0.742018 0.792674 1.043266 1.044798 1.130128 1.265268 1.331960 1.255905 1.047210 0.887825 0.632168 0.425331 0.303095 0.410382 0.726275 1.021292 1.021421 0.874702 0.784681 0.941267 1.070652 0.964130 1.064438 1.232349 1.305097 1.079951 1.033552 1.017983 0.755310 0.502687 0.445152 0.640920 0.720596 0.981995 1.081081 0.991478 1.043195 1.177009 1.252864 1.026312 0.954671 0.969808 0.785523 0.551212 0.542158 0.809425 0.938740 1.118596 1.189471 1.113548 1.208625 1.268449 1.095969 0.774477 0.639379 0.557977 0.412410 0.383617 0.761307 1.017866 1.100616 1.032559 0.905870 1.002256 1.099293 0.921765 0.824955 0.968709 1.096244 0.896833 0.926509 1.144463 1.231286 1.018506 1.015671 1.176835 1.009151 0.695451 0.627112 0.835489 0.716680 0.651318 0.964810 1.084255 1.225061 1.317297 1.299583 1.320562 1.100209 0.803836 0.531707 0.365212 0.270640 0.301612 0.686014 0.945946 0.907746 0.778413 0.667590 0.803779 1.047283 1.126377 1.255857 1.306938 1.333692 1.231345 0.973046 0.690720 0.462270 0.326037 0.255926 0.445197 0.863100 0.973441 0.878991 0.733015 0.697017 0.962908 1.060136 1.143398 1.271300 1.306747 1.332836 1.116681 0.898089 0.629672 0.427628 0.299756 0.308137 0.615215 0.958905 0.985980 0.859613 0.727764 0.806566 1.046645 1.035631 1.136599 1.267442 1.331696 1.235747 1.050284 0.890945 0.628327 0.423432 0.307817 0.422429 0.722914 1.017894 1.017118 0.872643 0.794148 0.948589 1.077824 0.975508 1.075999 1.239136 1.304025 1.068627 1.010555 0.983849 0.726235 0.484534 0.443756 0.672055 0.791038 1.044167 1.096908 0.990950 1.055820 1.212551 1.209730 0.944934 0.902404 0.932884 0.731503 0.514825 0.636557 0.952863 1.080685 1.221097 1.207405 1.148259 1.242353 1.070729 0.791370 0.557137 0.501195 0.419113 0.428652 0.780336 1.001150 1.014167 0.974475 0.896925 1.023626 1.108245 1.034818 1.020590 1.126869 1.162276 0.915112 0.835417 0.876098 0.805113 0.624615 0.745459 1.045359 1.209996 1.299877 1.323456 1.302322 1.258895 0.907498 0.602124 0.398004 0.297848 0.238169 0.470236 0.868743 0.910330 0.803489 0.675518 0.657593 0.959304 1.112648 1.242840 1.311603 1.291740 1.321090 1.078961 0.763387 0.504094 0.355311 0.267284 0.308307 0.726899 0.943411 0.891012 0.764224 0.658975 0.827216 1.056477 1.148705 1.273407 1.309724 1.336944 1.211144 0.937709 0.650855 0.435564 0.310533 0.253824 0.487177 0.889731 0.959636 0.853463 0.710405 0.701843 0.983782 1.071421 1.175848 1.287083 1.308286 1.319805 1.092947 0.848626 0.581032 0.398043 0.283779 0.319910 0.675810 0.971792 0.959270 0.825017 0.706349 0.829801 1.048531 1.058684 1.184435 1.286991 1.340052 1.214409 1.025493 0.813960 0.557956 0.379252 0.290135 0.439899 0.802522 1.025422 0.971957 0.820030 0.765966 0.971803 1.046936 1.006159 1.153505 1.279894 1.317273 1.087009 1.028206 0.881068 0.610974 0.406908 0.396210 0.590068 0.824346 1.067556 1.028483 0.901790 0.947890 1.127801 1.090546 0.934913 1.039007 1.159663 1.015123 0.770247 0.891737 0.982695 0.791801 0.703904 0.992963 1.188038 1.157475 1.230702 1.296121 1.308943 0.984452 0.706267 0.559482 0.400897 0.289533 0.394466 0.837494 1.019833 0.982253 0.841917 0.738599 0.965282 1.041737 0.999434 1.131568 1.264651 1.312474 1.101346 1.041240 0.916452 0.645193 0.431485 0.400319 0.569106 0.774044 1.048663 1.043960 0.924209 0.950353 1.109172 1.129405 0.951913 1.014548 1.131226 1.018288 0.755012 0.828525 0.963213 0.825192 0.740773 1.007735 1.195680 1.223578 1.261521 1.331976 1.318018 0.980378 0.663018 0.489383 0.345400 0.248581 0.374960 0.821329 0.956331 0.892022 0.748837 0.661011 0.915056 1.076219 1.151416 1.274567 1.302105 1.339226 1.137003 0.884858 0.619754 0.420814 0.300645 0.285702 0.605741 0.950926 0.974075 0.851556 0.714548 0.781496 1.035671 1.048235 1.156675 1.276481 1.326526 1.255728 1.057072 0.864341 0.600811 0.408036 0.295121 0.393618 0.724556 1.008369 0.993213 0.848762 0.760662 0.917139 1.063760 1.002679 1.122583 1.263844 1.331923 1.122970 1.026286 0.918400 0.651896 0.433214 0.369315 0.560895 0.781699 1.054298 1.050871 0.914899 0.923669 1.097323 1.116529 0.933264 1.002161 1.151182 1.060941 0.792409 0.876021 1.017017 0.850013 0.700209 0.953310 1.170026 1.133907 1.156734 1.267103 1.315462 1.030858 0.745012 0.656092 0.488622 0.346746 0.378863 0.811672 1.056576 1.088086 0.964498 0.827614 0.992444 1.042679 0.862186 0.873611 1.091269 1.196640 1.001000 1.090166 1.240955 1.142834 0.808309 0.749396 0.817650 0.623819 0.476162 0.759254 1.050119 1.225458 1.291007 1.181369 1.175481 1.174894 0.852039 0.561186 0.427361 0.439954 0.382765 0.643263 0.943294 0.942858 0.909488 0.865012 0.908314 1.108400 1.135221 1.200413 1.270619 1.328561 1.115132 0.854068 0.652069 0.477221 0.337472 0.312949 0.673944 0.997177 1.044582 0.931395 0.784237 0.866949 1.057403 0.955330 0.990420 1.175944 1.293156 1.151535 1.074405 1.118633 0.908799 0.613154 0.455959 0.554206 0.568973 0.715267 1.007513 1.008226 1.006575 1.066168 1.134468 1.188584 1.061320 1.040067 0.946750 0.806070 0.600908 0.594069 0.688737 0.834851 1.075359 1.191295 1.171471 1.219200 1.306649 1.211286 0.863295 0.645546 0.513275 0.381502 0.296500 0.578721 0.938822 1.028923 0.962284 0.820691 0.815487 1.057888 1.022027 1.007966 1.159344 1.286486 1.207158 1.018471 1.010167 0.861512 0.596054 0.420059 0.538530 0.727041 0.927935 1.118092 1.050515 0.990660 1.120648 1.248291 1.039181 0.855896 0.917330 0.861113 0.625235 0.551292 0.873056 1.097096 1.191223 1.274299 1.194612 1.258879 1.112843 0.823282 0.562050 0.431368 0.374659 0.351955 0.704560 0.974374 0.964274 0.894464 0.801576 0.903906 1.082890 1.079608 1.158947 1.276394 1.343463 1.133110 0.934382 0.765264 0.551715 0.372957 0.323302 0.604900 0.961831 1.101417 1.008738 0.846638 0.874896 1.086760 0.936914 0.852471 1.039560 1.216164 1.103736 1.015676 1.183915 1.185671 0.868297 0.641492 0.769454 0.691781 0.516920 0.694824 0.994591 1.157807 1.274519 1.225734 1.180644 1.227710 0.983561 0.665648 0.467403 0.426957 0.359205 0.470473 0.872259 0.966852 0.925579 0.861085 0.820803 1.039622 1.107837 1.150695 1.237526 1.327465 1.276548 0.988314 0.780140 0.584855 0.406835 0.289555 0.432945 0.847097 1.056209 1.015794 0.859216 0.770329 0.992121 1.010917 0.916687 1.054158 1.226526 1.274215 1.081234 1.133477 1.113556 0.815238 0.541504 0.502427 0.559146 0.515336 0.781991 1.003987 0.994777 1.038558 1.047528 1.128596 1.159299 1.084711 1.018339 0.921368 0.828539 0.632510 0.607354 0.686095 0.891938 1.106040 1.221930 1.213636 1.237777 1.315440 1.148050 0.797594 0.571494 0.445169 0.334059 0.295834 0.664789 0.946636 0.968503 0.878099 0.749971 0.828228 1.064330 1.084850 1.163592 1.283894 1.345618 1.212246 0.974601 0.784708 0.554711 0.375343 0.286154 0.487729 0.877630 1.060982 0.988449 0.825354 0.783186 1.021481 0.998369 0.935748 1.103806 1.254344 1.259824 1.070159 1.129027 1.039902 0.731939 0.488972 0.499799 0.560153 0.571407 0.899559 1.018560 0.981437 1.031834 1.062274 1.190614 1.115550 1.073614 1.029291 0.925871 0.751452 0.601605 0.635371 0.695275 0.878691 1.111921 1.186804 1.202287 1.247669 1.323584 1.155417 0.818907 0.607406 0.462353 0.338862 0.290825 0.651917 0.961468 1.000134 0.903822 0.763790 0.827088 1.057036 1.033988 1.093348 1.240967 1.324480 1.205732 1.023083 0.917819 0.683234 0.460874 0.339610 0.485207 0.764147 1.050473 1.074849 0.932746 0.869951 1.037853 1.098531 0.902245 0.941113 1.135533 1.154371 0.894348 0.964651 1.132617 0.983430 0.716856 0.825325 1.051233 0.905144 0.809234 1.062878 1.210906 1.151839 1.087626 1.242220 1.196897 0.872924 0.620164 0.645162 0.534805 0.400969 0.628574 0.954406 1.082232 1.130450 1.027059 0.996096 1.153295 1.035020 0.811869 0.752938 0.890499 0.774167 0.694288 0.952455 1.164881 1.278677 1.333525 1.348196 1.360480 1.060036 0.703566 0.463465 0.313139 0.225404 0.279419 0.738804 0.910202 0.836947 0.695389 0.586178 0.791520 1.055547 1.196530 1.295645 1.269402 1.279734 1.205725 0.894771 0.595343 0.411657 0.329158 0.276528 0.534515 0.901802 0.930785 0.838495 0.723478 0.736063 1.015733 1.113027 1.224993 1.316711 1.339658 1.307673 1.032899 0.750934 0.502407 0.341387 0.246241 0.347683 0.779242 0.969252 0.899087 0.749199 0.656144 0.872220 1.049501 1.120892 1.256480 1.291645 1.333634 1.182107 0.958142 0.682469 0.462458 0.329212 0.280064 0.501533 0.897451 0.988851 0.890304 0.748561 0.747201 1.009243 1.051212 1.120443 1.258355 1.319582 1.306041 1.080086 0.908644 0.653240 0.440903 0.305064 0.355040 0.659024 0.988690 1.018916 0.883519 0.763100 0.873416 1.068890 0.988977 1.070951 1.233629 1.323317 1.158647 1.041129 0.995553 0.738852 0.492513 0.378002 0.534482 0.687432 0.979720 1.070512 0.957458 0.933270 1.064040 1.189276 1.003822 0.975728 1.105455 1.058763 0.789079 0.726743 0.912698 0.862843 0.745999 0.962244 1.152001 1.263569 1.283562 1.349555 1.361230 1.063124 0.715258 0.489775 0.337483 0.236893 0.286942 0.739344 0.926238 0.870979 0.728961 0.613161 0.807376 1.056395 1.171215 1.287065 1.287753 1.304080 1.207666 0.915344 0.621283 0.421885 0.319848 0.265755 0.513073 0.898775 0.945208 0.845317 0.716906 0.720106 1.001572 1.091788 1.201546 1.302564 1.324912 1.314188 1.062751 0.797110 0.538886 0.367511 0.263113 0.331859 0.735177 0.975765 0.929922 0.785637 0.678899 0.852608 1.046662 1.085899 1.225002 1.293402 1.338775 1.197209 0.998567 0.746825 0.505456 0.350442 0.282989 0.462670 0.855985 1.010538 0.929142 0.780517 0.750799 0.990378 1.041916 1.062353 1.214945 1.304772 1.318098 1.092814 0.981477 0.756914 0.512940 0.346571 0.363259 0.595592 0.931846 1.060003 0.955703 0.823372 0.886170 1.099641 1.001193 0.974905 1.151943 1.275567 1.130437 0.958413 1.039299 0.903822 0.624398 0.472390 0.700212 0.830305 0.917347 1.122466 1.083241 1.110254 1.255682 1.323275 1.093107 0.855732 0.784633 0.590932 0.403400 0.363300 0.730258 1.049934 1.175457 1.080181 0.913851 0.988405 1.082634 0.818073 0.674650 0.844029 1.035280 0.878805 1.006130 1.172651 1.273353 1.190412 1.165232 1.216382 0.929715 0.640312 0.458769 0.464771 0.391891 0.550646 0.923459 0.978578 0.952450 0.908016 0.891198 1.097835 1.099083 1.121400 1.175000 1.263690 1.126950 0.884745 0.749772 0.622349 0.471024 0.393735 0.694038 1.025826 1.162604 1.124134 0.987246 1.017777 1.124837 0.872166 0.679682 0.735230 0.897737 0.758180 0.867718 1.107184 1.212079 1.298108 1.348400 1.382355 1.175656 0.834729 0.556400 0.374505 0.257342 0.227322 0.588253 0.909584 0.905231 0.772617 0.627988 0.693395 1.000407 1.121613 1.253657 1.292344 1.271373 1.280533 1.028153 0.724060 0.483913 0.359432 0.277666 0.373215 0.799785 0.960063 0.894821 0.772362 0.694866 0.912287 1.072422 1.148106 1.271941 1.321140 1.356610 1.149105 0.897291 0.630189 0.422519 0.292977 0.271865 0.578227 0.938860 0.973070 0.847053 0.702748 0.754961 1.023351 1.049412 1.159388 1.275368 1.315229 1.274489 1.071567 0.869385 0.602205 0.412222 0.295587 0.370645 0.697661 0.994191 0.987711 0.849275 0.749993 0.887924 1.064019 1.019742 1.134488 1.269147 1.340075 1.154661 1.021837 0.890092 0.628379 0.418789 0.338790 0.522498 0.793269 1.056621 1.036278 0.889460 0.872522 1.058901 1.081636 0.928775 1.029463 1.199689 1.164023 0.898520 0.968159 1.043509 0.811518 0.586789 0.757197 1.001878 0.940997 1.058874 1.170496 1.208729 1.222490 1.155056 1.122442 0.835893 0.632368 0.461352 0.438782 0.434495 0.729071 1.017670 1.036337 0.965665 0.920328 1.007437 1.108125 0.976726 0.994830 1.097099 1.145153 0.912360 0.895241 0.988951 0.906662 0.716059 0.801238 1.063575 1.107785 1.126273 1.255682 1.312905 1.224124 0.939221 0.785901 0.584386 0.408515 0.305779 0.522038 0.917472 1.098999 1.037982 0.877670 0.829025 1.058279 0.958624 0.836556 0.983774 1.187462 1.165040 1.032240 1.185587 1.244370 0.962040 0.668361 0.709569 0.653044 0.477056 0.541378 0.916814 1.095375 1.216028 1.164544 1.062013 1.189958 1.068999 0.769733 0.571820 0.597434 0.568481 0.521926 0.846980 1.036415 1.080360 1.133520 1.092560 1.183836 1.134958 0.966851 0.784016 0.685926 0.619797 0.507552 0.655585 0.975283 1.158452 1.223137 1.161695 1.144561 1.234055 0.974898 0.684366 0.529727 0.513657 0.427505 0.511075 0.900073 1.018243 1.025352 0.989403 0.934706 1.113578 1.087067 1.005167 0.979234 1.074583 0.996336 0.797395 0.817895 0.931267 0.915355 0.836705 1.069144 1.237892 1.293563 1.258058 1.326144 1.235694 0.866863 0.565701 0.407403 0.307509 0.244386 0.524581 0.881299 0.905885 0.814793 0.685944 0.691418 0.993322 1.134589 1.253299 1.321978 1.312019 1.313855 1.033123 0.721599 0.476093 0.333035 0.248013 0.337345 0.779794 0.945700 0.870687 0.733137 0.642453 0.863392 1.062939 1.165960 1.284698 1.298014 1.332251 1.180804 0.905464 0.618809 0.419096 0.305438 0.265495 0.540310 0.913668 0.951660 0.840977 0.705153 0.730094 1.010618 1.081679 1.193687 1.295213 1.319718 1.299727 1.060386 0.809251 0.549821 0.377118 0.271238 0.346928 0.737173 0.984727 0.944091 0.800772 0.698484 0.869868 1.049163 1.066988 1.205365 1.291584 1.344055 1.183870 1.008322 0.779801 0.530518 0.361980 0.294662 0.473219 0.845706 1.027446 0.953969 0.803011 0.776446 1.003355 1.038371 1.020713 1.177753 1.293006 1.300464 1.069067 1.015275 0.836909 0.573041 0.384458 0.412279 0.616751 0.882209 1.080201 1.008507 0.888835 0.961132 1.149072 1.047326 0.925905 1.063163 1.169002 0.982538 0.783855 0.947569 0.988528 0.768058 0.715791 1.017217 1.178829 1.098220 1.205977 1.276899 1.278226 0.946006 0.742988 0.626709 0.452910 0.326754 0.470363 0.893484 1.096825 1.076887 0.932218 0.835877 1.051201 0.983559 0.813817 0.890819 1.118888 1.142716 0.991948 1.153956 1.277476 1.115036 0.796507 0.825843 0.780518 0.554967 0.441815 0.777469 1.075511 1.236263 1.257241 1.114096 1.140051 1.127954 0.806321 0.541166 0.469993 0.538179 0.473255 0.760558 0.988379 0.976628 1.001225 0.997638 1.076128 1.152194 1.104989 1.086417 1.035661 0.988199 0.755640 0.660960 0.639877 0.703811 0.795215 1.046079 1.181774 1.223751 1.263401 1.332423 1.274666 0.931349 0.653950 0.474293 0.339806 0.251679 0.452889 0.872113 0.972724 0.892733 0.746633 0.695087 0.971680 1.067604 1.137439 1.266812 1.306046 1.326899 1.101083 0.887438 0.630400 0.428636 0.300959 0.320392 0.646592 0.977015 0.996897 0.866447 0.736437 0.833781 1.052485 1.014012 1.113341 1.255663 1.330559 1.206318 1.055685 0.935165 0.669319 0.448627 0.332094 0.453329 0.691896 1.005424 1.038443 0.904006 0.840675 0.974837 1.114060 0.979421 1.037513 1.206529 1.256718 1.000351 0.932070 0.981467 0.771836 0.531970 0.541079 0.857986 0.966380 1.127901 1.179485 1.095301 1.208324 1.220895 1.046224 0.747528 0.646926 0.571023 0.438345 0.455429 0.838061 1.055195 1.133092 1.065387 0.966875 1.097246 1.086983 0.860703 0.741248 0.848036 0.889275 0.727141 0.899403 1.137841 1.270391 1.296494 1.352016 1.379736 1.131486 0.761077 0.502201 0.341685 0.238970 0.236697 0.660716 0.903001 0.864559 0.728427 0.597107 0.725234 1.025925 1.168603 1.286140 1.283445 1.265332 1.249975 0.956443 0.644788 0.435630 0.341016 0.274129 0.447890 0.863339 0.941130 0.857191 0.741379 0.704682 0.972630 1.097330 1.201148 1.305961 1.332022 1.341945 1.087071 0.805966 0.544132 0.366705 0.259045 0.297619 0.698259 0.960090 0.922911 0.780161 0.659155 0.807302 1.039736 1.098206 1.234997 1.292081 1.323283 1.226389 1.003767 0.735457 0.496168 0.351021 0.273949 0.432057 0.839882 0.992935 0.915761 0.771879 0.724303 0.962779 1.052883 1.095040 1.237628 1.308850 1.337255 1.115668 0.948787 0.703628 0.475532 0.324069 0.322517 0.585107 0.946527 1.031563 0.915051 0.774142 0.829302 1.062904 1.000936 1.038698 1.209196 1.311857 1.211366 1.042587 1.020737 0.793086 0.533252 0.378612 0.498196 0.655666 0.926509 1.076974 0.982408 0.920995 1.033971 1.191399 1.038339 0.956595 1.085347 1.088974 0.840770 0.704630 0.889747 0.896101 0.754558 0.907674 1.126081 1.255705 1.266969 1.338076 1.374144 1.127556 0.770206 0.523846 0.364308 0.253617 kafbox-2.0/data/mimotestbed8K.dat000066400000000000000000004507741303712747200167760ustar00rootroot000000000000001.728795,0.112725 0.188405,0.479524 -1.703818,0.708624 1.747009,-0.894487 -0.876078,0.185039 -0.859322,0.616283 -2.205457,-0.556002 0.792494,-0.421710 -1.929850,0.529303 0.336898,0.176414 -3.930718,0.102619 -1.488604,0.594737 -1.104613,-0.280867 -1.639452,0.014405 2.218284,-0.258064 -1.229892,0.673396 -0.509271,0.521879 -0.539659,-0.415766 -3.343988,0.891370 -3.752091,0.142756 1.150012,-0.261101 -1.732266,1.202011 -4.233045,0.719818 -1.928931,-0.163871 0.425459,0.041457 0.955835,-0.018777 0.201315,0.440210 0.594867,-0.272271 1.140297,-0.466236 -3.248991,0.426976 1.286886,-0.794799 1.363722,-0.630612 0.029309,0.230538 -2.603082,0.212389 -2.569174,0.097720 1.624426,-0.449323 1.677096,0.154239 2.840642,-0.556179 -1.979504,0.263948 -2.366459,-0.068504 -0.932517,0.349511 -0.731887,0.709842 2.236666,-0.175966 -0.931230,0.996189 -3.121599,0.636585 -0.566206,0.413833 -2.645883,1.422138 -0.392476,0.763357 0.838078,0.261060 1.484636,-0.131521 -0.286064,-0.106961 -4.323887,0.050833 -1.288452,0.489877 2.879179,-0.126250 -1.693833,1.487115 0.114679,0.007140 1.286816,-0.655057 -1.340862,0.871798 -0.006284,-0.260389 0.705863,-0.423341 2.359005,-0.489152 -1.371804,0.297039 3.353579,-1.379547 -0.510618,0.236993 -1.295097,0.504857 -0.364428,0.236054 1.703601,-0.110159 -0.613100,0.771064 -0.881058,0.256516 -1.222944,0.615112 -0.970414,1.149550 2.394039,-0.014523 2.789575,-0.276054 0.330735,-1.014699 -1.019934,-0.562102 2.755433,-1.219853 2.597035,-0.566597 -0.260234,-0.379735 1.480497,-1.878196 2.664035,-1.244355 -0.556141,-0.215331 -0.655986,-0.402847 -0.025054,0.127938 1.806357,-0.262994 -2.224927,0.818665 -1.678422,0.420357 0.071069,0.555091 -2.493058,1.738211 1.769010,-0.339837 5.076668,-0.407564 2.633590,-1.741521 2.884426,-2.360820 2.933839,-2.096678 -2.214104,-0.405185 -0.921872,-0.071742 -0.040592,0.865986 -0.091996,1.060385 -1.088973,1.029578 1.834069,-0.513674 -0.038837,0.322131 1.549260,-1.468871 -1.188107,0.057461 3.640553,-1.276841 1.049437,-0.702938 1.371876,-1.757393 -1.802608,-0.662373 4.272046,-1.639566 0.640252,-0.279513 -3.168237,0.149050 -1.005028,0.353921 1.475853,0.308722 -1.050784,1.272359 -3.064230,0.688242 -0.307573,1.394716 -1.293464,1.999444 -2.682900,2.062957 0.543068,1.019065 0.679082,0.408717 3.349160,-1.502817 0.671271,-1.755807 -1.101112,-0.849888 -0.573014,0.003942 -1.629582,1.864219 0.107016,1.114035 -0.855683,1.703353 0.927720,-0.208299 0.833177,-0.401181 0.383268,-1.337006 -2.568656,-0.135248 -2.013417,0.951529 0.083572,1.607039 -1.514552,2.372254 4.557743,-0.317713 -3.600828,1.428302 0.352597,-0.749707 -0.527589,1.229198 -1.667776,1.161488 0.441534,0.620137 -1.764459,1.931081 1.713019,-0.682198 -1.851379,1.365505 -1.828141,0.700104 -2.655257,2.598557 3.223455,0.843330 -1.123655,1.690989 0.552082,-1.091404 -0.455305,0.928405 0.368367,-0.530832 0.165659,0.011553 0.904071,-1.256970 0.202822,-1.015579 -0.731521,-0.480789 -0.182069,0.224411 1.478914,-0.128256 1.880657,-1.363977 -0.057922,-1.707795 -1.857421,-0.419390 1.490077,-0.420418 4.976196,-0.543502 1.383850,-2.776058 -1.883420,-1.215960 0.769994,-1.428129 -0.557774,1.190544 -1.965887,0.775940 3.241503,0.213969 -6.103650,1.759196 -0.096907,-0.311440 0.636405,0.681821 -1.271028,0.844340 -2.057471,0.970363 3.282761,0.714237 0.038990,0.685335 -4.094537,-0.074828 -2.258609,1.780079 -4.711172,3.382622 -1.122498,3.317341 -0.175945,2.578918 2.147554,0.417817 -0.623818,0.202142 -2.957548,0.357688 -0.087957,1.597172 -1.599737,2.602240 -1.730316,2.134650 -0.238014,2.479705 -0.429659,2.238675 0.014630,0.788761 -2.078945,1.730711 1.665672,-0.143487 -1.493390,1.636994 0.698551,-1.164655 0.968026,-0.187822 -2.015718,-0.329028 2.006938,-1.467495 -5.352178,1.702000 0.033645,-0.475617 -2.886491,2.553522 0.213004,0.718084 -1.046943,2.245538 1.937162,-0.642327 -1.351523,1.337979 -2.173023,0.655498 1.585835,1.200978 3.215933,-0.360844 -2.772400,-1.177559 1.717312,-1.805929 0.415149,0.177873 2.097729,-2.972201 -1.568142,-0.605630 -0.652293,0.140059 -0.830731,2.830043 -0.681570,2.387563 1.130032,0.816602 0.883658,-0.565010 -0.218415,-1.751799 0.861097,-1.815850 1.386081,-1.498110 -1.095178,-1.056006 1.889472,-1.769823 -1.585113,1.340622 0.560336,-1.178412 -3.285949,2.099786 0.629491,-0.463145 2.060572,-0.204361 3.503403,-3.178175 -0.503217,-2.097693 3.638427,-2.429328 2.468798,-1.580293 -4.679223,-1.531200 -0.077250,-0.301275 0.014586,1.744232 -1.130057,1.098061 2.216515,-0.295525 1.040948,-0.760258 -0.995342,-1.555602 -0.355795,-0.083557 2.182033,0.500070 2.319463,-1.641737 1.350009,-3.915386 4.583511,-3.851638 -2.797691,-1.388782 -3.065640,0.812511 0.806023,2.103072 -0.933018,1.786338 0.856544,-1.136987 -2.781029,1.639135 -1.227732,0.958776 -1.991061,3.805365 -2.212094,3.630116 0.690312,1.999688 3.276811,-0.518852 -1.101825,-1.343559 3.329213,-1.934927 -0.096075,0.340106 1.324016,-2.698685 -0.593977,-0.176180 3.321378,-2.066687 0.114263,-0.335415 -4.454835,-0.610830 2.490398,-0.088069 -2.317255,1.668447 1.734794,-1.650997 -1.602631,1.744494 -0.527220,-0.129943 1.502117,1.192705 3.590456,-1.798041 1.968702,-4.189880 0.093338,-3.046974 0.646887,-2.412039 1.044884,-1.756206 -1.577055,-0.880003 1.468141,-1.210555 0.160831,0.377782 -1.087976,-0.896860 0.632514,0.146251 -2.817420,1.909702 0.373628,-0.186348 -4.524867,2.624270 1.000751,-0.352436 -0.449652,1.364988 -2.910949,0.340402 -0.031007,2.228942 -0.875592,2.595854 1.814299,-0.233070 2.568267,-1.257328 -1.460182,-2.083348 -2.945337,0.570975 -3.188708,4.526915 0.995173,1.966914 1.482099,-0.536493 -0.710078,-2.138547 -1.621149,0.509266 0.476424,1.941688 3.010146,-0.316100 -2.379122,-1.250871 -0.389647,0.053064 1.249574,1.395333 -2.556134,0.589405 0.200519,-0.312691 -0.684363,2.430705 -0.004029,0.308860 -0.996812,2.039529 2.099510,-0.126521 -3.341117,1.167147 -4.028741,0.551901 1.973232,2.273510 -0.120965,0.242008 2.385882,-2.977746 5.371161,-2.105135 1.707467,-4.538293 2.011097,-4.239805 -0.001963,-2.710452 -1.120917,-0.520864 -0.382792,1.783536 -0.097826,2.171102 1.200921,-0.423651 -3.989283,0.141672 -1.930267,0.885532 -1.886398,4.582832 -0.401341,3.408548 1.112333,1.129888 4.036761,-1.987734 3.627473,-3.923943 -0.224896,-2.772168 -1.779953,-0.058874 -1.453685,2.963703 1.527004,1.872312 -1.197029,1.063032 1.447459,-1.599415 -1.735876,1.351525 1.683346,-1.499362 -1.701877,1.592762 1.866853,-1.541078 0.971919,-0.610893 -0.432405,-2.711095 -0.762995,0.432216 -2.854081,3.253365 -2.975338,3.341861 -5.030205,4.532680 -2.612420,4.171322 -0.753901,3.713640 -2.215008,3.637188 0.625556,1.831570 -1.690480,1.783954 0.475195,-0.525032 -1.837534,2.024706 4.883382,-0.856773 0.166241,0.231635 0.532525,-2.709620 -1.525454,0.279819 -4.985610,1.658940 -0.327744,3.980972 1.403758,0.869143 -1.710125,-0.227868 0.747667,-0.806258 -1.008312,1.621126 -2.149161,0.933320 -1.265904,4.539505 3.709718,1.684395 0.934846,-1.359099 2.632136,-4.578707 3.558075,-3.697234 -0.769276,-2.544061 1.790258,-1.326599 0.947284,-0.812255 -0.047143,-2.699574 3.224898,-1.784637 1.679345,-2.116596 0.495812,-3.261212 -1.080908,-1.065885 -3.616869,2.087088 -0.532407,3.284072 1.538048,1.216460 0.153447,-1.185101 0.157190,-2.043078 2.126192,-1.232931 0.698394,-2.398849 1.510860,-3.086028 -1.248007,-1.058318 -0.842747,0.740630 1.192058,1.999392 -2.779975,0.439446 -5.269337,1.104596 -1.613869,4.099304 0.183860,2.779672 4.742028,-0.241923 -0.017471,-2.502546 4.244310,-2.520023 -1.368461,-0.023523 -0.826066,0.249007 -1.715366,4.200609 1.941797,1.270422 -2.128417,0.604225 3.001500,-1.274232 -0.941414,1.275532 3.099052,-1.857335 0.179909,-0.111079 -3.719083,-1.115969 -2.817207,2.422498 0.230867,3.247351 -0.961321,1.704831 2.473478,-0.535034 -4.030870,0.775758 1.127040,-1.114104 -0.087040,0.728638 0.920896,-2.141975 0.565308,-1.115848 2.120064,-3.271389 1.094112,-2.908711 0.440233,-2.752746 -3.819402,-0.537789 2.234378,0.247949 -3.215863,1.094441 -2.887400,0.309145 -0.629102,4.542489 1.532865,1.117169 0.349729,-1.417778 2.633699,-3.245221 1.917171,-2.676028 1.295381,-3.177009 0.184971,-2.850635 -0.822548,-0.585514 0.693259,0.545389 -0.697959,0.573321 -0.400804,0.219249 0.786522,1.191005 -3.705295,0.111575 1.993839,-0.193291 -0.962094,1.079563 -0.590912,-0.176834 -0.618087,2.866989 -0.766014,1.842234 2.047673,0.728637 2.720959,-2.049332 -1.411665,-2.116881 -1.218736,0.933647 -0.224019,3.027708 1.811701,0.239484 -1.185802,-1.053069 4.288329,-0.941656 1.497139,-0.840200 -3.308184,-2.616508 -1.955822,1.746042 -0.694735,4.211047 -0.215468,1.748314 -0.805252,1.447899 -2.131234,1.807163 1.757046,1.697011 0.921102,-1.380053 -2.157244,-2.314759 1.289481,0.509550 1.210797,-0.282943 0.110146,-3.321067 -0.011011,-0.949643 -0.198969,0.478580 -0.451156,0.804916 -2.052009,1.489508 -1.464705,2.819948 -2.810906,3.160286 -2.238953,3.038401 0.372313,2.474989 -0.629127,0.675145 -0.177535,0.043388 -0.321838,1.443731 -3.872555,1.051553 5.503511,0.881087 2.584808,-1.392551 -0.467790,-2.876185 -0.386279,0.384661 -0.208038,1.950129 -1.629852,0.968607 -0.217152,1.465952 -3.138286,2.413150 0.424228,1.399147 3.357550,-0.224892 0.758019,-3.225703 -1.336837,-1.566778 3.455948,0.395069 3.386775,-1.175391 -1.574090,-2.668863 -3.748941,1.639156 0.047705,3.039762 3.036909,-0.118058 1.068954,-2.945769 -2.710933,-1.730353 0.561847,0.778684 0.364199,0.617681 -0.513134,-1.320710 1.716822,0.154613 0.114140,-0.648333 -2.924996,-0.745996 -4.652331,2.486462 5.125290,2.170372 -1.278642,-0.310558 1.412019,-1.032241 1.254748,-0.521874 -2.930543,-1.846327 1.097908,0.820808 3.789240,-0.307801 1.803877,-4.181242 -1.369685,-1.155563 -0.820131,1.457944 -3.668689,3.045527 -0.714352,1.516197 -0.163091,1.219761 -1.114321,0.538147 -3.556598,1.883295 0.077348,2.608653 3.206804,-0.324550 0.856616,-2.717127 -0.643357,-1.867705 0.317333,0.460357 -3.662450,1.087384 2.166275,0.226474 -0.884636,0.035910 0.426005,-0.830228 -0.858136,0.978381 2.225385,-0.300961 2.108076,-1.209205 0.384365,-3.092558 -0.351294,-0.377278 1.122841,0.366936 2.502042,-1.058579 -0.838754,-2.174735 -2.929813,1.104542 -1.906617,3.625710 2.768517,1.627339 -1.933107,-0.978387 -0.004143,0.088692 1.017254,1.078072 0.692684,-2.070868 -2.793882,-1.263438 1.040260,1.144012 -0.699660,0.526602 1.060585,-0.681161 -0.512737,-0.055890 -3.234573,0.024602 3.113719,2.032829 1.642136,-1.436515 0.412189,-3.199173 -3.394706,-0.308732 -0.336673,2.110048 1.656389,1.100496 0.095286,-1.711695 -1.631848,-0.613016 1.854588,1.479818 2.144300,-0.904175 1.550079,-3.023256 -2.036836,-1.077862 -3.151304,1.465676 3.887532,2.775741 2.958689,-1.548491 3.162210,-2.779656 -1.029385,-1.016911 -1.814217,1.545856 4.517605,2.427647 0.081546,-1.379139 0.485731,-1.260202 -0.693212,0.199891 2.094625,0.129362 -0.534203,-0.291987 -1.643556,-0.027462 0.071279,1.792235 2.966174,-0.238054 0.722544,-1.937676 1.247519,-1.107685 -1.820498,-0.490825 -5.113664,1.634278 3.330113,2.439594 -1.919162,-1.260408 -1.033740,0.234148 -0.009263,1.957734 -0.870893,0.204198 1.763509,0.062610 0.994933,-1.222245 -1.707894,-1.706355 1.083341,1.430671 1.102827,-0.352139 -0.275632,-2.269209 -1.287701,0.785510 -2.600911,1.704715 -2.509037,2.142302 3.680388,0.888646 -1.317704,-1.176710 -0.550994,0.181118 2.965648,1.302569 -0.016697,-1.601501 0.180484,-0.863608 -3.743990,0.696538 -4.392970,1.366313 -2.085170,2.074387 1.090271,0.521584 -2.328930,-1.352787 1.593574,0.148514 -0.500590,-0.113581 -2.224426,-0.361650 -0.546323,2.445406 -2.026903,0.490168 1.744330,0.299976 -2.111163,-0.868301 1.696029,0.150463 0.602598,-0.084883 -1.364574,-1.235497 -1.015804,1.607018 -2.058931,1.384872 -0.571671,0.163142 -3.352417,0.778705 -0.994978,0.821391 -0.773795,1.086500 0.086917,0.131469 -1.310338,-0.448544 -0.293364,0.680975 0.171449,0.570796 -1.923256,-0.639188 0.919268,0.555210 -1.032646,-0.632709 -1.470580,0.135164 2.941567,1.578267 1.994547,-1.349931 2.613965,-2.152876 0.202508,-0.382742 3.155149,0.362538 5.932406,-0.210867 -0.587362,-0.831554 2.687810,1.224084 -1.499583,-0.151210 -1.397586,0.507252 -2.604235,1.626441 -0.343519,0.170480 1.422561,-0.453960 -0.323674,-1.656679 0.407558,0.016516 0.628451,0.232814 0.432354,-1.021415 0.499263,0.154776 -3.221883,-0.174085 -0.902313,1.162015 -3.200002,1.247787 -0.291625,-0.052188 2.384076,-0.134278 0.571378,-1.521053 1.902269,-0.727534 -1.930761,0.036279 -2.218849,1.284181 -1.233632,1.117206 -2.320836,-0.448377 0.543655,0.593972 -3.929985,-1.122950 -0.399420,0.049215 3.584469,0.731000 -0.158803,-1.832330 1.529457,0.081804 1.320797,-0.039212 -3.377150,-0.490429 -0.859949,1.665725 0.145683,1.158352 2.959574,-0.830678 -0.356855,-1.693591 -2.413166,0.289284 0.783974,1.692641 -0.093098,-1.069089 1.905055,-0.547421 0.985312,-0.388230 0.615780,-0.765606 -3.335975,0.405024 1.957083,1.460121 -1.714295,-0.563179 -0.884569,0.050382 1.007549,1.394422 -0.377106,-1.420099 -1.043434,-0.364168 0.423785,1.118027 -1.372784,-0.940453 -1.768537,-0.143934 -0.119139,0.791593 -4.951670,-1.135862 1.131747,0.275088 -1.850857,-1.204830 -1.882013,0.021511 -0.380839,1.083989 0.256180,-0.484247 -0.135764,-1.083652 2.942524,0.183036 2.135364,-0.335871 0.459750,-0.441938 -0.011985,0.659695 -2.336390,0.559744 -1.803559,0.903605 0.646416,0.514195 -0.024653,-1.508023 1.108276,-0.217373 0.125389,-0.175275 -1.285993,-0.142925 -0.662608,0.982896 -0.534350,-0.262684 -0.495521,-0.798945 1.554800,-0.076977 0.687814,-0.912106 1.727520,-0.156009 1.717067,0.886115 1.374614,0.708759 -2.598622,0.427348 1.207649,2.028062 -3.247054,-0.612775 -1.240982,-0.231098 1.287202,0.676149 -2.291332,-1.869278 1.688381,0.647190 -0.085813,-0.256378 -1.008670,-0.581738 -0.886544,0.883462 4.166105,0.454835 3.717750,-1.067298 1.853188,-0.208390 0.590831,0.852959 -0.532658,0.662915 1.484775,1.567786 -0.161868,-0.337091 1.871223,0.652463 -1.670407,-0.255319 -1.490378,0.444785 2.723370,0.997646 -0.612301,-1.851982 1.756875,0.515735 -2.201291,-0.270054 -0.978232,0.245617 -2.700480,0.132768 1.575559,-0.354243 1.646818,-1.130846 -1.358637,-0.804373 1.194354,1.884693 -2.342332,-0.581452 4.074008,0.681316 -0.881396,-0.707365 -0.556881,-0.089132 0.541456,0.740801 0.801988,-0.710262 0.147787,-0.221239 -2.081983,0.267725 -2.907070,0.420728 0.469676,-0.106039 2.099353,-1.012016 0.682803,-0.363304 -1.985358,0.298634 -3.234833,0.861607 -0.888688,-0.411970 -2.111468,-1.573890 -2.997942,-1.558031 0.356998,-0.677271 -1.914573,-2.030010 2.628801,0.273404 0.061002,-0.509905 0.144148,0.062782 1.110045,1.216612 -1.459638,-0.486316 2.192743,1.445373 2.671586,0.372405 0.822879,0.354769 3.265783,1.911777 0.486801,1.377273 -0.006215,1.304034 -1.477665,0.609015 -3.535797,-0.109480 3.588448,0.003622 2.563089,-0.953440 0.256742,-0.009322 -0.429791,1.112760 -0.740717,0.620491 -2.317180,-0.786512 -2.956657,-1.411157 -2.951270,-1.904982 1.302502,-1.098057 -0.184695,-1.697492 -0.734505,-0.346278 1.634300,1.131716 2.471209,-0.062920 0.672527,0.779106 0.462796,1.760213 0.092575,1.289469 -0.627182,0.460199 -2.146640,-0.278756 -0.126629,-0.525684 1.830217,-0.678098 -3.350078,-1.449815 1.836347,1.397708 2.047178,0.541676 -1.809865,-0.406996 -3.842902,0.736033 -0.273467,-0.494435 2.727909,-0.976131 1.566750,-0.212793 1.054717,1.383165 -1.493949,0.982162 3.422703,1.992009 -0.302502,-0.382815 3.038029,0.989560 -0.799673,0.116669 -4.025529,0.017505 1.428517,0.557515 -3.854962,-2.042025 -1.746821,-0.774255 -0.722085,-1.007224 -3.215795,-2.496416 -1.553987,-2.162695 -0.641745,-1.984941 -2.626974,-2.434986 -0.217012,-1.515352 -2.035223,-2.171907 2.601395,-0.438081 2.432300,-0.214418 -2.093507,-0.083762 0.247906,1.687303 -2.213050,-0.970604 0.981877,-0.077931 -3.363191,-1.410799 -0.343088,-0.420873 1.446203,0.159509 -1.554490,-1.521301 -0.186312,0.356694 0.335276,-0.176313 -1.275935,-1.422226 -0.208072,-0.237649 1.263936,-0.081347 -3.375390,-1.301153 -1.035665,-0.069511 0.173040,-0.632144 4.399917,-0.623786 1.521838,0.778181 -2.912896,0.670745 -3.549790,0.436530 0.591700,-0.418865 2.037515,-0.910454 -1.287985,-0.364400 -2.223186,0.222103 2.917049,0.005075 0.206372,-1.129290 -1.276847,-0.215515 -0.050754,0.538031 -2.602568,-1.667180 0.689387,-0.587356 -2.721089,-1.747587 0.471544,-0.086443 4.865103,0.220831 -0.705764,-0.200475 -0.506815,0.706048 -0.157359,-0.396068 -2.407772,-1.915006 0.908411,-0.345928 1.339323,-0.117576 -0.804945,0.115486 1.518052,1.934309 2.563681,1.045499 1.607195,1.987179 -2.408165,1.405535 -1.652367,0.287969 -1.422071,-1.853030 0.872606,-1.598756 2.043852,-0.450165 -0.725314,0.287357 -0.596596,0.786724 1.466926,0.408134 0.681336,0.044085 -2.212614,0.239902 -4.086657,-0.312191 -0.717810,-1.938795 1.190799,-1.502233 0.750905,-0.293171 2.404268,1.922782 1.087150,2.722890 0.576921,2.556150 -1.331914,1.245346 -0.302883,0.291481 -1.319524,-1.750539 -3.961751,-2.417116 -3.649625,-2.911016 -0.510601,-2.893658 -1.653552,-3.026388 3.064986,-0.966972 3.409806,-0.044467 -0.429979,0.981040 3.410881,2.487548 0.743740,1.429559 -0.528579,0.993324 5.006453,1.991749 1.471412,1.314442 -2.199913,0.776626 -2.584979,0.197505 2.498351,-0.539291 0.396571,-0.786174 -1.409800,0.144061 0.769379,1.093920 -1.488921,-1.304385 -0.171020,-0.596358 -1.521654,-1.697660 1.176317,-0.668630 -0.977573,-1.372921 -1.581441,-0.844292 0.373849,-0.932780 2.646473,-0.338607 -0.504480,0.307439 0.788895,1.796517 2.442842,1.464097 -1.261788,0.644754 -0.345570,0.560244 1.180801,-0.039889 0.971415,0.385092 -0.690943,0.952844 0.774622,1.594219 0.009140,0.230554 0.769041,1.085024 -2.824280,0.039784 -2.392021,-0.949224 0.095914,-1.825294 0.844616,-1.140696 2.161742,1.215993 -0.144069,1.603304 -1.505751,0.707733 -1.111515,-1.249138 -2.609930,-2.933639 -0.224106,-2.676945 0.734068,-1.404241 -0.654091,-0.837153 -0.672890,-0.432344 -0.777310,-1.541773 3.361820,-0.797576 1.414492,0.591164 2.061035,2.821416 0.610118,3.335771 -2.243968,1.394704 -0.245805,0.270954 -1.387447,-2.184271 1.751822,-0.824784 -2.188468,-1.400195 1.850003,0.737481 -0.459143,-0.271196 0.450519,0.668971 -0.671814,-0.155537 2.436630,0.650246 -0.192274,0.183510 0.241136,1.088229 2.009767,1.495276 -1.314742,0.404527 0.810346,1.163504 1.780541,0.874244 2.898089,2.236366 2.765845,3.620770 2.352179,3.768286 -3.459597,1.981788 0.209298,1.133270 -2.975252,-1.638999 -3.486134,-2.173294 -1.021838,-3.158238 -0.134587,-2.887306 -0.127883,-1.839170 -4.392712,-1.732442 2.212289,-0.370458 3.052255,0.698927 -1.374332,1.045077 2.321853,1.873269 -0.850152,-0.231386 -0.121323,-0.016379 -3.799962,-1.624973 2.438075,-0.673694 1.802224,0.679125 1.646474,2.951518 3.764421,3.924876 0.476812,3.387576 -0.858435,1.678615 -3.595124,-0.706621 2.934582,-0.880934 2.060701,0.366059 1.785676,2.835910 2.791174,3.973882 0.832975,3.635306 3.090240,3.661886 1.329858,3.489135 1.412598,3.411176 5.518586,3.509626 -0.103400,2.788602 -1.679336,0.740111 3.110652,0.253516 0.298516,0.121855 -3.369302,0.313476 -1.139902,-0.903468 0.977185,-1.431009 -0.058466,-0.714476 -4.476510,-0.425235 -4.234476,-1.733329 0.305332,-2.132102 -2.707179,-2.441164 -0.412906,-1.546828 -2.408238,-2.782992 -0.873708,-2.960352 0.094298,-2.210970 4.957928,-0.231515 -2.577365,0.185127 1.131758,0.989178 -0.978267,-0.715193 1.143950,0.526920 -1.066563,-0.397498 1.529466,0.712691 -3.497151,-0.588190 -1.459851,-1.283968 -0.008945,-2.051846 1.071985,-0.897802 -0.042244,0.511411 -1.339365,0.127668 -0.112870,-0.844317 -0.426158,-1.728682 0.865785,-0.338516 -0.131441,0.213017 -4.544594,-0.347652 -2.092499,-1.910396 -0.956769,-3.002136 -1.531517,-3.533572 -0.187478,-2.571452 -2.278136,-2.627451 -1.383008,-2.572596 -7.193101,-3.338332 0.926136,-2.193527 -3.932658,-1.799241 2.445408,0.199066 0.474626,0.453718 -2.031970,0.526416 1.280731,0.507337 -2.382698,-0.988205 0.401503,-0.149326 1.691668,0.449695 3.735471,2.349228 3.125198,3.749534 1.644258,3.699128 -0.724367,2.360739 -3.390178,-0.169500 -0.475925,-1.899779 -0.080190,-2.173544 1.539652,-0.247660 0.994315,1.617791 -2.550341,1.259399 -0.634684,-0.268528 -0.281870,-2.096531 1.294630,-0.609090 -0.444075,0.069705 -1.977497,-0.418708 1.032175,-0.634802 -1.766106,-1.242237 -1.476023,-1.435750 1.632295,-1.174210 -2.082457,-1.236681 1.141724,0.471519 -1.314878,-0.654794 -0.649995,-0.918572 0.048455,-1.582835 -1.079491,-1.733826 1.405885,-0.379947 2.002077,0.931395 0.293679,2.088471 -1.046404,1.185526 1.410101,0.586894 1.887938,1.037397 2.733010,3.219537 -2.566083,1.836729 1.969817,1.210762 -1.131468,-0.398104 2.030531,0.749882 -0.387746,0.283944 0.014382,0.430696 -0.813687,-0.643738 2.509373,0.002265 0.412340,0.825622 -0.874610,0.907916 -0.955052,-0.867344 0.690234,-1.245856 2.788480,0.628791 -0.499579,1.265268 -0.536375,0.321040 1.267203,-0.071212 -3.353639,-0.560870 3.462904,0.529391 0.804524,0.838385 -0.642107,1.338934 -4.212339,-0.510350 -1.890748,-2.584605 -1.531631,-3.190090 2.668903,-1.771012 -0.071893,-0.123340 0.417082,1.235962 -5.118107,-0.177412 1.176906,0.072924 -2.387401,-0.876010 2.514250,0.280577 -0.290138,-0.049026 -2.834537,-0.341161 0.839858,-0.846115 -0.788224,-0.935828 1.018148,0.378884 0.048700,0.419267 0.497109,1.226872 2.119617,1.791635 -0.724199,1.040888 -1.144834,-0.212409 -1.762533,-2.452944 3.769491,-1.461034 0.403301,0.412390 0.416510,1.858263 -0.907272,0.568557 -0.342922,-0.742631 -1.013127,-1.903441 4.340191,-0.647985 -1.335656,-0.424358 2.325028,0.749113 -0.930915,-0.129299 5.318243,0.763538 -0.801043,0.036759 -1.958001,-0.699441 -0.920483,-2.411673 -1.908993,-2.932815 2.120022,-1.298616 2.270425,0.773469 -2.173841,1.328145 3.209162,0.910050 -0.731458,-0.134939 -3.199616,-0.737827 -2.177241,-2.741738 -1.902334,-3.294035 -1.430350,-3.160714 3.401158,-1.536373 3.385028,1.020631 -2.714554,1.262710 -2.140945,-0.606924 -0.310938,-2.272726 2.066705,-0.808459 -2.115100,-0.121338 -2.228991,-1.154938 -1.047280,-2.768360 1.322131,-1.445040 -2.000178,-0.681822 -0.148689,-0.587049 2.033841,-0.218261 3.249743,1.765011 1.544291,2.782561 -3.066498,1.441052 1.539804,0.559003 -2.548698,-0.562604 0.282233,-0.201626 -2.806709,-1.341453 4.187979,-0.310544 0.326942,0.372813 -0.912739,0.649802 1.342492,0.420399 -4.790503,-0.390288 -4.318859,-1.262232 2.820297,-1.375293 -0.198323,-0.137299 -0.180497,0.454209 -1.835460,-1.043443 -1.257625,-1.930925 -2.953411,-2.767714 3.937472,-1.423189 3.297887,0.962257 2.261250,2.691981 -0.007825,2.260899 -2.566275,0.266998 2.286042,-0.254033 1.267410,0.857058 -0.461628,1.492295 -0.585984,0.006611 1.858086,-0.146366 1.364498,1.210680 -0.083656,1.691305 -0.604380,0.395503 -0.126143,-0.690332 -0.083737,-0.738574 -1.090447,-0.891023 3.030247,-0.277574 2.282466,1.117076 1.807490,2.805028 0.051167,2.078238 0.450419,1.381649 -0.623497,0.204171 -0.657687,-0.722021 4.066358,-0.354011 -0.953546,0.047406 -2.307585,-0.803377 -0.647818,-2.235645 -0.802904,-2.124284 -0.985460,-1.892095 -0.089426,-1.575809 -0.373019,-1.102443 -0.880069,-1.075124 -1.678088,-1.802173 -1.367531,-2.421179 2.974185,-1.028101 -0.029842,0.309748 -1.253589,0.115564 -0.203553,-1.085027 -0.451853,-1.290336 3.811543,0.071771 1.590370,1.483551 -1.448119,1.228528 3.504045,0.591183 -2.497783,-0.334032 -0.770583,-0.779710 -0.668507,-1.968473 -0.481953,-1.559676 -0.756182,-1.475804 -3.371690,-1.850320 1.509087,-0.936823 -3.388413,-0.628774 -0.067528,-0.417211 -0.730175,-1.362198 2.321614,-0.018350 -4.568423,-0.164230 -1.180190,-0.913355 -1.573087,-2.412299 0.772535,-1.203850 0.602597,0.321049 0.132800,1.191715 0.020849,0.576876 0.092811,0.181699 0.130088,0.210663 -4.259181,-0.550000 1.141355,-0.344431 -3.220304,-0.721392 -0.304216,-0.609274 1.648512,-0.396164 0.406124,1.022073 2.310303,1.813669 -1.066614,0.736738 -0.524362,-0.454564 1.584246,-0.431156 -0.801328,0.090698 -0.688759,-0.397589 3.195904,-0.398550 -0.007138,0.607313 0.859115,1.334662 0.313167,0.891184 -1.372432,0.173762 2.359298,-0.008463 0.697378,0.717128 0.780166,1.776706 0.929771,1.508625 -0.874298,0.624028 -0.629773,-0.611480 -0.727400,-1.548735 3.071766,-0.226239 0.488737,0.882674 -2.426200,0.446968 1.018111,-0.138023 0.675873,0.440659 -0.713991,0.680534 0.086474,-0.126458 -1.323174,-1.023950 1.254322,-0.108538 -1.523671,-0.498634 2.194462,0.610691 0.164314,0.368979 -1.881480,0.221423 -1.664647,-1.548050 -0.623333,-1.965102 -1.459527,-1.629394 -1.904557,-1.883882 -2.014061,-2.048581 0.128204,-1.252819 0.344324,-0.093424 -0.246005,0.325359 0.112602,0.050902 -0.012722,-0.107423 2.099166,0.768255 3.445306,1.556371 0.327486,1.766794 -3.067899,0.141586 -0.328063,-1.041114 2.236246,-0.237057 3.190986,1.450601 -0.089960,1.274866 -3.362268,-0.311704 1.472166,-0.332517 2.547072,0.802459 0.734664,1.790094 6.489219,1.779953 -0.319539,0.931300 2.777603,0.906279 -0.589294,0.090853 -4.685815,-0.520553 -2.709820,-1.799918 -2.405887,-1.983999 -0.770955,-1.471498 -2.843748,-1.565592 -3.608301,-1.532890 -2.494527,-1.798687 2.687037,-0.654550 3.121267,1.055572 -1.188084,0.761529 -0.008147,-0.231156 0.169773,-0.652712 -2.612845,-0.405081 -0.588073,-0.977114 -1.120244,-1.426217 -3.503166,-1.262804 1.565936,-0.438199 3.867726,0.926564 1.499433,2.284196 3.936716,1.653128 0.367488,1.164716 -3.082652,-0.005419 2.892519,-0.040190 -3.869298,-0.497406 0.208657,0.144618 -0.483225,-0.851386 0.466436,0.304709 0.246601,0.209129 -0.886181,-0.013087 2.704999,0.161265 1.708413,0.823778 1.858999,2.206373 -3.580344,0.400810 1.187374,0.142351 2.239714,0.424919 3.344806,2.178070 4.018763,1.441124 1.086290,1.495863 2.397929,1.262538 0.270289,1.145532 1.096548,1.197936 3.525945,1.253708 -1.029245,0.642009 3.647847,0.457504 0.215585,0.088662 -2.665322,0.066791 1.983534,-0.088110 -0.967546,-0.357447 -0.666750,-0.201109 -2.506585,-1.713339 -0.044628,-0.789986 4.569881,0.372752 -0.268585,0.603020 -0.512911,0.056535 2.770327,-0.014151 -2.450966,-0.134446 -0.796859,-0.354746 2.655360,-0.619064 -4.797628,-0.139639 0.477482,0.387791 2.974130,0.216520 2.341869,1.972672 2.055756,1.344251 1.589041,1.626945 -3.695991,-0.161223 0.286060,-0.442636 -2.762022,-1.299536 -1.261377,-0.560922 -1.254594,-1.239742 1.285214,0.145608 2.481299,1.052451 -0.509695,0.712370 0.252690,-0.114432 -1.097357,-0.955272 -4.185084,-0.751516 -1.651424,-1.034273 -0.686383,-0.776349 -0.889967,-1.052736 1.300686,0.450353 -2.317501,-0.162903 -2.936118,-0.425064 1.026005,-0.964254 -1.606478,-0.293406 -0.575920,-0.115165 0.672630,-0.434148 -0.237515,0.234918 -3.752769,-0.306282 -3.429369,-1.674007 1.158289,-0.335809 1.203212,0.972474 -1.254045,0.406491 1.393082,0.284672 0.939536,0.145235 0.854298,1.585344 2.565410,0.952048 -0.610268,-0.177902 -1.369989,-0.700114 3.851292,-0.361929 1.686982,0.646108 2.691025,1.278247 -2.606673,-0.397556 -4.240683,-1.212087 -0.186967,-1.078050 -1.005335,-0.471680 -2.438857,-0.457913 0.248081,-0.057414 0.075964,0.116418 -2.966102,-0.120684 1.586334,0.147935 -0.593974,-0.565078 1.788698,1.072150 0.534668,-0.114936 0.628646,1.471326 1.609220,0.291249 -3.750452,-0.603904 -1.663471,-0.807616 -2.094040,-1.923295 -0.886625,-1.178986 -1.799313,-0.483357 1.426035,0.434662 -2.311659,-0.252389 3.137222,0.904514 4.038764,0.129907 1.460247,1.828578 5.567947,1.113393 -0.866205,-0.250212 -3.121348,-1.035024 0.520634,-1.051255 2.115864,0.735440 2.129299,1.500561 3.024338,0.819135 -0.144427,-0.215984 2.077540,0.334357 2.488074,0.204621 2.948154,2.000233 -1.206272,-0.817689 1.449286,0.133033 -1.650493,-1.181382 -2.496523,-0.114126 0.692693,-0.171722 2.186811,0.919173 -0.475259,0.706769 -0.857426,-0.465089 0.796210,-0.135382 0.690063,0.570342 1.484061,1.198618 -1.973944,-0.680223 1.739898,0.640777 -0.746876,-1.347135 -1.127917,0.209529 2.852120,-0.507028 0.182328,0.533308 1.000652,0.860123 -1.088071,-0.987323 1.415286,0.564874 -0.803626,-1.010726 -1.198314,0.811028 -0.104300,-0.880074 -0.860783,-0.774722 -2.223739,-0.255001 1.194110,0.619406 -0.210648,-0.943084 2.052591,2.050926 2.090288,-0.044410 3.486099,1.074066 -0.169934,-0.691112 1.432240,0.238021 -0.349598,-0.983375 0.467229,0.352835 -1.265757,-1.208602 1.139406,0.968394 0.925717,0.055907 1.505373,0.890043 -0.127983,-0.211222 1.236622,0.500947 2.154362,-0.087427 -2.751885,-0.462885 -1.176668,-1.076808 -1.835388,-1.047320 0.564224,0.675170 -1.392489,-0.579630 -2.770876,-0.144082 -1.767457,-1.348931 0.195517,0.567793 2.201715,1.040065 -2.204981,0.114390 2.937699,0.957068 2.046726,-0.410077 -1.585011,0.684176 1.997254,0.211779 3.511424,0.036139 0.518056,1.220982 -0.261571,-0.445222 4.650555,0.081338 -2.167214,-1.132714 -0.160116,0.564236 1.232531,-0.027278 0.310616,0.924084 1.237999,1.173530 -0.740195,-1.049653 -0.333029,-0.141164 0.483914,-0.321206 0.020911,0.515863 -2.677932,-0.370293 -2.749235,-0.982977 0.965219,0.334971 0.057794,0.565815 -3.228521,0.406565 -0.174023,-0.554987 -1.243472,-1.402445 -2.001070,-0.191826 1.457425,0.685524 2.593405,1.238775 -0.764437,0.120307 1.353834,0.287262 -0.933452,-1.499375 3.187714,1.562557 0.557718,-0.584328 -2.241424,0.349472 -1.624635,-1.552937 1.572317,-0.395561 -2.239035,-0.249070 1.228098,1.719851 3.728361,-0.088349 -0.125812,1.251354 1.589198,0.069020 -0.489263,-1.086963 -1.005699,-0.217024 -0.535705,-1.116970 1.710500,1.174366 1.085269,0.607022 3.997543,1.968254 0.229393,0.049331 1.110167,-0.161711 -1.539780,-1.735751 -2.289440,-0.583283 2.154042,0.067030 0.019384,0.303793 -0.065148,0.823631 0.304196,-0.616239 1.299682,0.821470 1.803336,0.854837 0.331843,-0.383371 -1.754771,-1.341051 2.470510,0.122093 2.521138,-0.031361 2.037340,1.203564 1.423913,-0.112928 -0.443668,-1.424049 -0.886689,-0.955509 2.897693,-0.215773 0.116823,-0.018568 -1.196573,-0.213452 0.165809,-0.664518 -1.061195,-0.850184 1.985668,1.659368 1.626701,0.035560 -0.688963,0.980602 -0.954118,-1.102404 2.747783,-0.441570 -1.711305,-0.445273 0.783584,1.809874 -2.056669,-2.407965 1.303396,1.770770 2.593290,-0.002952 2.276973,1.058184 2.732377,-0.001177 -1.632979,-1.229760 1.364422,0.192198 0.087038,-1.629171 -0.263334,1.023470 1.243338,0.040967 -1.885589,-1.309261 -1.771393,0.035376 -3.576610,-0.821754 -0.272532,1.008640 1.713809,1.282243 1.591150,1.786404 0.464984,1.300654 -6.760308,-1.408362 -0.381784,-0.426172 2.313739,0.642559 1.727675,1.666825 -1.095072,-0.452793 2.055901,0.261628 -0.514726,-2.025724 -0.290835,0.873835 0.179699,-0.647108 0.634787,0.839702 3.238618,1.492021 -1.086158,-0.953982 -1.098326,-0.624508 -4.046075,-2.154202 0.938482,1.598929 3.978628,1.130781 -1.793950,-0.466065 0.197365,0.000683 -0.617547,-2.312518 3.216138,2.101477 2.631935,0.222490 -1.520128,-0.050239 -1.317364,-1.084859 -1.976909,-0.934755 1.226564,2.148273 -2.552410,-0.597134 1.105835,2.273833 -0.849585,-1.900688 0.958268,2.101192 2.217672,-0.461222 0.277454,0.874673 -0.796221,-1.037863 0.786734,-0.323627 -4.924944,-1.620277 -2.478175,0.239076 -0.634789,-1.937747 0.611751,1.346403 0.356828,1.049705 -3.979436,-0.148927 0.334091,0.022510 1.388250,-0.249353 0.723499,1.339593 0.301714,-0.260717 2.194701,0.060824 0.447132,-0.956637 -1.645680,-1.249161 -0.926511,-1.282367 -0.370581,-1.194307 0.787885,1.483645 -1.439778,-0.279980 -0.526536,0.381174 -2.269108,-2.406403 4.499189,1.952357 0.281480,-0.095285 -3.519884,0.501324 -2.819695,-1.926434 2.165255,0.174506 0.389678,0.728691 1.017440,1.453598 -0.910532,-1.792751 3.570096,1.190484 0.351577,-0.942990 0.199773,0.585758 -0.936356,-1.752596 0.954291,0.500500 -0.965763,-1.399853 -0.074874,1.419132 3.222795,-0.252418 -2.662079,-0.788688 0.466624,1.048692 1.506573,-0.927047 0.759712,1.508398 1.619659,-0.012840 1.774718,0.231925 -0.844950,-1.777961 -0.631592,-1.020314 1.050121,-0.088966 2.060596,1.240982 -5.286883,-0.875752 -0.558198,-0.451091 -0.197982,-1.930636 3.375980,2.024841 1.377006,0.801747 -1.276219,-1.019520 -2.425031,-1.306363 1.361454,0.088183 1.788484,1.061760 -4.217502,-0.309959 -2.861306,-1.053383 -1.960265,-2.466520 -0.360959,1.467159 -0.463883,0.655333 0.689308,1.689691 0.217401,0.188655 0.277516,0.724925 -1.503760,-1.618388 -2.551463,-0.413009 4.060890,0.354229 -3.255894,-0.779363 -0.647222,1.659841 1.481182,-0.112161 -1.164520,-0.397010 -1.485512,0.513420 3.837295,0.474872 -0.585144,-0.167813 2.342029,2.510573 -0.448049,-2.406448 4.281073,2.534655 0.565252,-1.149536 -1.579874,-0.365859 4.385277,0.460489 -1.796186,-2.094708 -0.364477,1.886136 -2.534474,-2.011174 0.427646,2.089073 0.750703,0.367549 -0.285147,1.098731 1.712817,0.826864 -2.386656,-2.381723 1.909128,1.934765 1.510411,-1.256068 -0.730811,-0.201396 1.955676,0.404689 2.140547,-0.983569 1.017392,0.774226 -0.683612,-1.602863 -3.358339,-1.299143 1.023853,0.190233 0.878444,0.527062 0.754221,1.469742 0.314078,-0.893087 -1.148853,-1.608730 -0.710318,-0.942929 0.467186,0.072277 0.880664,1.341844 -3.685246,-0.643538 2.781029,1.007557 -0.248503,-2.240368 -3.512864,0.654326 -2.250105,-2.037113 0.390564,0.564236 2.991465,1.800911 -0.002183,1.244929 2.140730,0.983972 3.295100,-0.788229 -3.186789,-1.327904 0.615831,-0.117906 -1.124260,-2.582193 -1.855979,1.053347 -1.460485,-1.070228 0.151051,1.574639 -1.305489,-0.152926 2.459347,2.598176 0.744811,-0.421754 4.590812,2.205688 -2.279889,-2.644880 -3.803895,-0.806756 -1.922610,-2.578348 -0.741735,0.905569 3.113264,2.369065 0.740510,0.967845 -0.666866,-0.076597 -1.103747,-1.445195 0.361152,-0.578763 1.634008,1.131638 1.174403,1.559273 1.423982,0.690588 1.005849,-0.863350 0.046481,-1.143846 2.214785,0.187820 -0.689323,-2.087595 1.093555,1.513264 2.085519,-0.910171 1.735970,1.155534 -1.377468,-2.059994 1.467936,0.376961 0.243988,-1.759776 -1.562718,0.635915 -1.852759,-1.391111 -2.200088,-1.671184 0.549287,2.143769 1.319925,1.343520 -1.108818,0.497382 1.122751,0.974392 -0.786927,-2.537356 0.102623,1.910248 -0.124341,-1.735948 0.025276,1.006787 1.820254,0.771404 -0.369448,-0.762046 -2.011005,-0.388714 0.335039,-0.572497 -2.242002,-1.048199 2.110670,2.860111 0.591919,-0.579761 2.857063,2.876050 0.019564,-1.365755 3.757440,0.849900 2.155986,-0.817138 1.525029,0.929187 2.235752,-0.165696 -0.811954,-2.297467 -1.578273,-1.334438 -0.551045,-1.947931 1.457097,1.585052 -1.789793,-0.479622 0.948712,2.414806 0.707466,-1.314317 1.369616,2.484963 -0.868724,-1.796293 1.526379,0.962457 -1.159946,-2.872554 -0.798587,1.207322 -0.925639,-1.514140 1.422495,2.027849 -0.697906,-0.667904 0.073987,1.590385 -1.133861,-2.276769 3.236974,2.228178 -2.793812,-2.051661 -1.288115,0.878928 0.552899,-0.745350 2.179009,1.676256 -1.624418,-0.315378 2.360258,1.236054 1.701844,-1.458341 1.848480,1.694886 0.150163,-1.253369 -0.843407,-1.543368 1.632367,0.251700 1.212353,-0.256169 -2.174517,-0.224887 -3.483373,-1.263539 1.392538,-0.455405 -0.873002,-0.288606 -1.079315,1.561218 -0.122685,-1.115350 -3.652451,-0.313490 -0.884054,0.781232 -0.694982,-0.564629 -2.414481,0.336541 -5.001800,0.263492 -2.585984,-0.109781 -0.490162,0.993356 1.157125,2.067616 2.928874,1.589663 2.181789,1.328449 1.577036,-0.219629 0.787260,-0.632426 0.031654,-1.163033 0.097480,-0.877510 -1.016554,-1.707034 0.927904,1.206974 -2.239432,-1.790039 0.382005,2.240097 -1.790829,-2.442889 -1.399913,1.307075 0.170383,-0.519108 3.500173,1.696374 -1.748112,-0.348151 1.901181,1.723942 1.163123,-1.384698 -2.737279,0.241971 -0.205135,-0.512363 -0.162268,-1.674583 2.282910,2.278274 0.572966,0.126213 -0.261772,-0.025381 1.719703,0.071843 0.207541,-1.172117 -0.815298,0.109184 -0.031700,-0.815313 3.368773,0.420572 -1.303189,-0.720951 -0.310589,0.782829 0.422914,-1.172251 0.291537,1.047365 -0.854222,-0.275186 -0.057559,0.010105 -0.307134,-0.971169 3.274475,2.006670 0.827128,-0.001074 1.321147,1.339692 -0.955458,-2.196913 0.383657,0.292177 -1.831957,-2.741086 -0.991444,0.825538 0.892764,0.113137 2.777129,1.536482 -0.337419,0.205406 1.311919,0.597299 1.233431,-1.346652 -0.511485,-0.170303 -3.818233,-1.343503 3.357955,0.195674 -0.778697,-1.000229 -2.625357,0.875439 -1.924523,-1.856942 -0.665202,0.429445 -0.374750,0.816122 1.959963,2.107968 -0.654319,-0.426656 1.418432,2.125375 -0.671171,-2.678773 1.512555,2.314200 -1.056801,-2.787661 1.497233,2.496721 -0.180891,-2.098959 -0.561589,1.208621 -4.841046,-1.907852 1.079871,1.303726 0.363452,0.031680 -4.123502,0.711486 -2.957721,-1.212204 -0.039640,-0.447902 0.365299,1.232495 3.461732,1.979654 0.125240,-0.248234 -1.913709,-1.055486 -1.285122,-1.524921 3.198377,0.902141 1.016203,1.078325 -1.872585,0.196495 -2.005371,-1.295759 -0.412311,-1.229766 0.439240,1.387525 2.642588,1.708232 -1.821896,-0.824215 -0.876542,-0.266015 5.901454,-0.554801 -3.123960,-0.694860 1.051758,2.615644 -1.778223,-3.030168 1.828692,2.915846 -0.073458,-2.192939 -2.862533,0.635919 -3.434966,-1.647216 2.434751,1.080337 -6.206515,-0.669601 2.749016,3.088171 3.702434,-1.173465 -0.953173,0.535992 1.455758,0.570483 -0.764427,-2.657793 -0.000457,1.706670 -1.664423,-2.266586 1.632058,2.131190 -5.259987,-1.837035 1.427461,3.071310 -3.408107,-2.996510 2.056432,3.081852 2.458763,-1.551025 -3.594175,0.134701 -1.909912,-1.071644 1.160213,-0.476451 -3.043589,-0.582576 3.517436,3.119557 -2.064320,-2.837075 0.866202,2.725506 -4.800727,-2.977705 0.598796,2.456581 2.987004,-0.598725 -1.716377,0.266321 -1.549118,-0.637743 1.225823,-0.467336 -2.227636,-0.487802 1.747070,2.904801 2.487252,-1.155423 -3.141794,0.156109 -1.298023,-0.890927 -0.064320,-1.260519 0.538738,1.236188 -0.935744,0.080246 1.418127,1.486735 -1.634382,-2.254892 0.322831,1.898943 1.701117,-0.721245 1.791429,1.429950 2.088425,0.584589 -0.299536,-1.518835 -0.066097,-0.604485 2.707095,-0.269311 1.546552,0.450516 1.987025,0.932211 -1.416480,-2.083229 2.405079,0.608904 -3.355331,-2.931536 -0.857795,1.491142 2.113711,-0.486230 1.588133,1.025078 1.957577,1.755945 1.369083,-0.123944 -1.916854,-1.973800 -1.116873,-1.043443 -1.204149,-1.828612 -3.186956,0.225396 -1.164210,0.241020 0.021960,0.697298 -0.539050,0.221627 -2.596611,0.109920 1.188235,0.944525 1.693147,0.753998 -2.322548,0.085987 1.584801,0.726920 1.048127,-1.200729 -1.915358,0.244870 -1.417995,-0.783015 0.430204,-0.551634 0.147415,0.590562 1.773699,1.959274 -4.951080,-1.544954 -0.959119,-0.005223 0.885673,-0.420583 0.200753,0.703614 0.564237,1.495987 0.632576,-0.476590 -0.366891,-0.562064 3.137130,0.558998 0.585398,-0.492020 -1.386828,-0.164009 0.471453,-0.072796 -1.753847,-1.931592 -1.923566,0.476255 0.002943,-0.527653 -0.997873,-0.095391 -1.834141,0.428247 0.944721,0.750728 0.591121,0.594025 -0.877257,0.517143 -0.815905,-0.716060 -0.332182,-0.990081 1.516048,1.326866 1.284829,0.772109 -1.177785,-0.462884 0.421400,0.076525 -2.869632,-2.215279 2.110132,2.214972 1.402478,-0.282884 -0.957794,0.527759 -0.557677,-0.655546 -5.414641,-1.906076 0.068286,0.720093 -0.662801,-0.492638 -1.635334,0.595665 -0.839808,-0.322818 0.523045,0.682473 1.254805,0.999843 0.124040,0.836505 -1.242610,-0.871229 -0.950355,-1.135085 -0.213752,-0.575277 -0.783183,0.267446 -1.691050,0.520959 -0.730139,-0.394113 1.484865,1.095374 -2.048905,-0.373125 -3.316525,0.531022 3.175414,-0.308787 0.456173,0.781622 0.213790,1.281216 -3.221392,-1.585075 2.426631,0.750758 0.829741,-0.577702 1.727077,1.355579 -1.523415,-1.886328 -0.973148,-0.420426 2.754781,-0.308380 -2.052691,-0.646154 -3.172352,1.067166 0.271472,-0.539479 -1.750532,-0.333474 2.304713,2.100856 -0.431279,-1.383670 -0.359322,1.203409 -3.523238,-1.578685 0.914384,1.008108 -0.316969,-0.537573 -0.213343,1.235990 -0.777472,-0.908308 -0.135353,0.445831 -1.870514,-0.505386 -4.409175,0.680024 0.496367,0.436141 0.806686,0.671007 0.945638,0.816441 -4.347393,-1.153717 0.931605,0.653423 1.547155,-0.844071 1.784441,1.439168 -3.503603,-1.380188 -2.387623,-0.743957 0.656514,-0.339399 0.600629,0.356938 -1.134711,0.320643 -1.132310,-0.612191 1.892204,0.254828 -1.467765,-0.596898 -1.011284,1.165839 -1.022309,-1.176136 -1.746372,0.034730 -1.503573,-0.007511 2.009941,1.308253 -0.959166,-0.347113 -1.246666,0.746970 -0.988554,-0.840671 3.491841,1.244051 1.303292,0.694933 2.429308,1.703756 1.706678,-0.780254 0.532312,-0.300723 -3.332035,-1.181367 -0.461514,-0.523448 0.905204,-0.148299 -0.479684,0.025296 3.586331,1.205540 0.912929,-0.947076 -4.438988,0.011401 2.289680,0.377098 -1.787438,-1.873031 1.526396,1.848181 -0.323645,-1.656253 4.719022,1.816928 0.357987,-1.117291 -0.573409,0.079688 -1.772229,-0.771246 1.315702,0.465655 0.782998,-0.260992 4.910084,1.690268 1.173871,-0.249884 2.917451,0.441672 -0.187616,-0.987692 1.206872,0.427129 2.300550,-0.718553 -1.904954,-0.309510 3.783765,0.623017 1.120771,-0.802150 1.289708,0.682095 -3.848918,-1.329318 0.803465,0.277498 -2.701513,-1.547222 3.068477,1.922116 -3.530610,-1.513701 1.488748,1.929724 -0.109539,-1.381979 0.436276,1.094834 0.946129,-0.337739 3.775639,0.018884 2.349889,0.385937 -0.104459,-0.221975 -0.625975,-0.566546 1.720356,0.040449 0.059991,-0.319830 0.557140,0.788866 2.852358,-0.109683 -1.625009,-0.450051 1.683513,0.775015 -0.021049,-1.295765 -1.389494,0.501378 2.456614,0.307851 2.642535,0.013144 4.650863,1.449127 0.563910,-0.790104 3.247999,0.364340 2.129608,0.025526 1.118950,-0.394174 -0.086849,-0.802733 -1.047919,-0.651357 0.821492,0.350268 0.363892,-0.286337 0.340798,0.392940 3.187828,0.203064 3.028130,-0.225741 -0.765933,-0.638488 -3.963878,-0.741979 4.171668,0.322810 -3.631208,-0.945738 0.341558,1.214594 0.127590,-0.761417 3.018144,1.117366 -1.532878,-0.611346 -1.866815,0.304267 -1.701300,-1.237716 -2.858190,0.122497 -1.421061,-0.403933 2.138381,0.658903 0.093704,0.086812 1.988115,1.198262 -1.439638,-1.222505 0.031744,0.561803 -0.113325,-0.601803 1.328872,0.811961 2.601024,-0.378152 3.729002,-0.032536 -2.081650,-0.795743 -0.590327,-0.242936 -2.138461,-1.344431 -0.206983,0.295191 3.490530,0.326060 1.389038,0.171584 0.289519,-0.084789 1.051249,0.158286 2.586507,0.060653 2.350579,0.426891 0.708496,0.137246 2.372074,0.259816 0.565271,-0.536587 -0.215970,-0.379576 -2.452613,-0.226045 -1.385134,-0.118188 0.283646,0.057788 1.802121,0.528984 -0.448349,-0.276953 -1.676532,0.016388 -0.463640,-0.308593 -0.068494,-0.175460 0.630796,0.288697 1.488496,0.502599 0.586241,0.347159 1.941105,0.137830 0.545196,-0.115086 -2.128549,-0.347693 -1.560114,-0.423777 -0.163778,-0.567538 -2.222773,-0.234076 0.195020,0.422113 0.378049,-0.192142 1.057928,0.157550 1.329163,-0.354498 -0.407963,-0.374477 -0.065548,0.096266 -2.189240,-0.837118 0.229491,0.548635 0.313786,-0.232635 -0.592506,0.260500 0.120959,0.186488 1.369842,0.043816 -2.577887,-0.205061 2.774390,0.823105 0.037237,-0.654105 -2.559729,0.219890 2.415359,0.145998 -2.088430,-1.090596 -0.998935,0.601194 -4.185274,-0.808715 1.436645,0.430431 -3.132937,-0.573971 -2.533486,0.612988 -5.440537,-0.194568 -1.645355,-0.421925 1.854279,0.304110 0.243805,-0.380325 -1.719675,0.088317 2.882440,0.148112 -2.342131,-0.852779 0.231314,0.788099 -0.890515,-0.481213 1.269626,0.554591 -0.575500,-0.490729 0.304053,0.615450 1.439251,-0.150674 -0.238012,-0.147639 -1.255574,0.084807 -1.394357,-0.740203 -1.381188,-0.119954 2.542536,0.159253 -0.963957,-0.338127 0.182190,0.760531 -1.381171,-0.777358 1.499059,0.561665 3.672009,-0.188019 1.946163,0.685250 -1.110804,-0.186049 -1.805880,0.508793 1.885892,-0.126029 1.963250,-0.542647 2.419339,0.720279 3.138263,0.564692 0.804736,0.345510 -1.604400,-0.013992 -1.176133,0.094447 -3.357639,-0.665740 -1.788995,-0.526768 2.526886,0.129442 -7.147379,-0.418772 -3.885526,0.279828 -1.408100,-0.543848 -0.542847,-0.186504 2.531221,0.039292 -3.517177,-0.347078 2.179910,0.820945 0.816673,-0.445238 0.571774,0.760536 -0.715004,-0.036417 0.477912,0.230469 -2.916841,-0.532014 0.657145,0.338748 -1.288386,-0.604752 -1.582028,0.280238 3.655283,-0.021381 -0.732606,-0.660024 -1.356143,0.379932 -3.664642,0.038213 -0.795571,-0.479029 2.165174,-0.050458 -3.005300,-0.379264 -1.760917,0.291955 -2.262633,-0.890142 -1.579242,-0.181695 -0.246379,-0.035603 -0.213630,-0.310280 0.291474,0.034074 -2.228681,-0.185309 0.142379,0.097659 1.857308,-0.191169 -2.849793,-0.256237 -0.946387,0.169830 1.119334,-0.279066 0.076517,-0.547359 -2.260081,0.108508 0.331136,0.335188 1.125818,-0.178788 1.137335,-0.307864 -1.747616,0.107638 -0.809000,0.519236 0.576286,-0.257489 0.016530,-0.170477 0.898894,0.337936 -1.331818,-0.078932 0.187838,0.337370 -3.533297,-0.366855 -0.689453,-0.051431 3.906644,-0.245109 -3.200807,-0.358038 0.443993,0.405590 0.438402,-0.157433 1.138872,0.016679 0.221993,0.122640 -1.188568,0.220528 -0.263327,-0.113912 -1.020512,-0.331234 1.017954,-0.043767 1.262574,-0.244032 0.025443,0.414806 2.894675,0.544381 0.857279,0.307464 0.780988,0.155087 -1.444153,0.394687 -0.025333,0.214660 0.836758,-0.179258 1.460367,0.217729 -4.133769,0.171441 -1.583368,0.279735 2.358363,-0.099201 0.792633,-0.456598 2.349895,0.216172 -2.640639,0.255858 0.742845,0.473603 -0.021743,-0.288543 0.794955,0.312156 -1.070011,-0.029036 -2.519765,0.173547 -5.142681,0.034162 0.472371,-0.452867 -2.697013,-0.524750 0.030524,-0.095399 -0.089846,-0.208057 -3.416078,-0.247749 3.098484,-0.004907 0.637674,-0.394110 -1.472859,-0.009459 -1.744960,0.254140 -0.080770,-0.524344 -3.171324,-0.539258 -0.981251,-0.292240 2.593477,-0.388967 0.467115,-0.266172 1.766607,0.442698 -0.860087,0.342064 -1.208922,0.370716 -1.040858,0.193290 -6.470413,-0.554154 -1.016328,-0.581658 0.418761,-0.498224 1.014718,-0.149711 -1.962502,-0.142434 -3.035848,0.139875 3.714767,-0.182471 -0.596328,-0.530218 2.866706,0.172584 -1.263314,0.154827 3.466993,0.184741 -0.560284,-0.078154 -3.661179,0.025892 1.301483,0.060496 -2.208602,-0.377984 -2.065224,-0.130258 -3.604613,-0.359447 0.485387,-0.547953 -4.226726,-0.512167 2.090225,-0.073621 2.236359,-0.098988 -1.065955,0.101073 -1.111673,0.576198 -0.690963,0.061785 0.470650,-0.550165 0.893777,-0.237149 -0.371588,-0.026603 1.280587,0.285289 0.638827,0.180383 -2.379399,0.321797 0.489770,0.319732 3.989550,-0.138518 -0.985363,-0.066304 -2.791252,0.286779 -1.273446,0.189385 -3.448027,-0.652198 0.136769,-0.638666 -2.284542,-0.470720 0.352851,-0.278163 -2.530384,-0.141965 7.025660,-0.380129 0.156310,-0.069895 -1.434710,0.148510 -1.053317,0.018004 -1.986259,-0.462648 1.264287,-0.656206 1.671075,-0.392408 2.120795,-0.037431 -2.187441,0.843804 -0.980493,0.573212 2.111987,-0.233629 0.668908,-0.410151 -0.899820,0.214887 -0.718783,0.529763 0.797681,-0.134359 -2.944812,-0.181422 1.350456,-0.078197 3.583893,0.048522 -2.902557,0.044727 -3.163181,0.441076 1.039369,-0.381315 -2.318535,-0.425655 -2.382519,-0.327079 2.283256,-0.262301 0.996162,-0.291210 -0.649378,0.278379 0.951257,0.400023 1.874974,0.210326 -0.767599,0.295616 -1.062483,0.491351 2.955767,-0.096119 2.049033,-0.322421 -1.239945,0.267998 0.305800,0.487959 -2.633076,0.216523 2.363984,-0.322966 0.712724,-0.024506 -1.037199,0.171540 -1.406381,0.392837 0.005853,-0.212851 -3.259420,-0.530385 -2.782790,-0.612947 0.606947,-0.560142 6.660753,-0.541400 -3.159851,0.172093 1.894668,0.210677 1.321971,0.114183 1.737003,0.137195 2.092191,0.700876 -3.656418,0.890893 -0.000836,0.314984 -3.258892,-0.006338 -2.363491,-0.433731 -2.981751,-0.536524 3.160503,-1.079424 -1.721120,-0.225992 -1.167833,-0.145141 -0.115063,-0.061615 -1.279005,-0.622904 -0.941866,-0.670309 -0.575326,-0.506350 -1.566666,-0.563473 2.244422,-0.610823 -0.880558,-0.231766 0.196360,-0.402992 -0.941881,0.334356 -1.973316,-0.228581 -2.413819,-0.364543 3.993673,-0.992286 1.787377,-0.509081 -0.013582,0.139415 -1.053482,0.590506 0.248229,0.046455 -1.117015,-0.110428 3.772672,-0.517762 -0.782981,0.206048 -1.734629,-0.180138 1.097627,-0.134921 -1.714036,-0.267525 -0.406024,-0.317895 -0.647423,-0.067807 0.633184,-0.559289 -0.554158,-0.045180 -2.158833,-0.134066 1.342644,-0.319683 -0.306270,-0.229974 -0.275083,-0.045431 1.064709,0.038517 0.015306,0.047468 -0.876199,0.204512 0.982788,-0.068534 -1.970910,0.134695 1.365894,-0.450172 0.532803,0.139769 -2.054633,0.050462 4.661012,0.063087 -1.101639,0.243956 -1.655101,-0.192923 -1.425420,-0.047383 -1.414857,-0.795693 -0.506883,-0.871429 0.140253,-0.828690 -1.326202,-0.276012 2.827820,-0.505031 1.675252,0.116332 0.042527,0.236352 0.334603,0.636379 1.379057,0.548711 -0.951452,0.532727 -0.226328,0.177431 0.695036,-0.018423 -4.277627,-0.111337 3.111571,-0.349745 -2.018062,0.437915 -1.690481,-0.227049 1.756199,-0.320345 1.135973,-0.339459 0.293283,0.166351 -2.623151,0.774281 -0.253509,0.001790 -1.353381,-0.200624 0.226853,-0.793248 -0.548694,-0.199117 2.470203,-0.471936 0.176691,0.382581 -1.296775,0.154388 1.265396,0.005180 0.483822,0.127582 -2.234878,0.110119 2.625877,-0.093043 -2.515877,0.386374 3.594643,-0.689961 -0.241413,0.405799 -3.473398,-0.188393 -2.032749,-0.041651 1.266935,-0.661123 -0.920964,-0.337929 -0.250709,-0.608913 0.868339,0.008872 -3.775131,0.044935 -0.433171,-0.399860 -0.277478,-0.063892 -1.246349,-0.606934 -2.022865,-0.492755 -0.922997,-0.858727 0.798354,-0.911559 1.871033,-0.398314 -4.103701,0.311799 2.009372,0.104182 -1.299333,0.449559 -0.231152,-0.328139 3.352889,0.018229 0.820650,0.013347 -2.893297,0.465736 0.197865,0.156124 1.481499,-0.019671 -3.637331,0.124639 -2.127857,0.068644 -2.143399,-0.423187 0.196692,-0.897950 3.091116,-0.546842 -0.463176,0.149148 4.696750,0.078227 -0.886295,0.691107 1.090168,-0.245357 -1.939756,0.586451 -0.924233,-0.631144 1.594041,-0.359579 -1.175515,-0.120177 0.663522,-0.541323 -3.500509,0.491193 2.068798,-0.512547 0.545161,0.382763 -2.002922,0.168555 0.648584,-0.070971 1.352342,0.262937 -1.304622,0.167090 1.308593,-0.013732 4.331907,0.629617 1.500810,0.088143 0.588143,0.824283 -2.642646,0.844582 0.558874,0.007243 0.633567,0.283921 3.997084,-0.037205 2.574926,0.677645 -1.746248,0.868230 -2.916324,0.187685 -2.223779,-0.516674 2.943826,-1.210149 1.692353,-0.337596 -1.516213,0.281786 0.150113,0.033088 -2.114466,0.175213 -1.596471,-1.034808 1.496918,-0.653121 0.762627,-0.129136 2.763682,-0.094549 2.275389,0.644857 0.342103,0.860429 -0.169699,0.772224 1.676567,0.153201 -0.676257,0.555478 0.184081,-0.219360 -0.261788,0.282179 0.668242,-0.304899 -0.162455,0.319102 2.851020,-0.193230 -1.046055,0.847306 3.709624,-0.180301 -1.066401,0.815637 1.220801,-0.558283 -2.432009,0.672374 -3.971248,-0.548700 0.749816,-0.441047 -0.534463,-0.129460 1.303758,-0.508906 -1.646465,0.677368 0.164975,-0.494504 -0.153959,0.240858 -0.080537,-0.366785 1.706785,-0.045928 2.335665,0.558641 -0.517894,0.614876 2.493059,0.307604 -1.894508,0.730727 -1.245510,-0.593047 2.506425,-0.340455 -1.312117,0.115238 -0.997823,-0.497700 3.389532,-0.338260 0.553586,0.211079 0.216369,0.176342 -0.390616,0.565021 2.263320,0.053970 -1.809317,0.535624 1.720822,-0.544410 -3.128409,0.689854 -3.053991,-0.423178 1.284115,-0.418170 0.288088,-0.086240 1.165384,0.070895 3.616456,0.794249 -0.589131,0.847192 -1.966217,0.069044 3.939753,-0.366703 -2.860341,0.273656 0.193258,-0.727167 -0.021670,0.242691 -0.306420,-0.220325 -0.883763,-0.181153 -0.484476,-0.595294 0.132841,-0.348064 1.118812,-0.247017 -1.691713,0.340024 -1.473096,-0.304102 -4.477178,-0.327623 -0.302357,-0.682034 0.861465,-0.550991 1.195588,0.074397 0.814718,0.145724 -0.487205,0.657563 -2.303393,0.092740 0.946430,-0.385139 2.100093,0.073771 2.438316,0.105292 -0.811391,0.723788 1.047617,0.008414 1.317994,0.333956 0.854216,0.599970 0.053663,0.651389 3.299519,0.194610 3.112054,0.854644 0.230344,0.815506 -0.990516,0.477550 -0.831824,-0.026854 2.910761,-0.362653 -1.847794,0.357425 -0.933638,-0.646750 -1.022793,-0.101765 -0.847495,-0.924608 0.458630,-0.708984 -1.395428,0.001416 1.415757,-0.615537 0.770140,0.576056 0.196385,0.142912 0.617674,0.296616 2.072460,0.391560 -1.645002,0.681648 2.414254,-0.299636 0.541875,0.741143 -1.667527,0.189861 -2.231756,-0.227426 -1.635558,-0.403845 1.729375,-0.776844 -0.554468,0.242067 -3.296842,-0.371456 3.222053,-0.473152 -3.064301,0.547283 -2.560569,-0.854184 4.217693,-0.282218 -2.158699,0.441514 -2.811964,-0.699093 2.033375,-0.335481 -0.990934,0.461314 -1.090031,-0.626954 2.588664,-0.347135 -2.365506,0.401515 0.246114,-0.669499 -2.609101,0.233318 3.772350,-0.807537 -5.837471,0.663628 -0.003466,-0.806279 3.679222,0.086037 -1.537095,0.336517 -0.697792,-0.624449 -0.770707,-0.142449 -0.583717,-0.413639 2.910170,-0.385123 1.008426,0.440214 2.798644,0.558566 -1.210817,0.669494 -0.996039,-0.437727 -1.064326,-0.354972 1.769594,-0.633701 -1.294213,0.432151 3.022755,-0.620413 -0.073252,0.812093 2.297357,-0.255845 0.732516,0.657187 1.772916,0.229154 -0.374712,0.676018 3.793161,-0.181562 0.621013,0.819729 -3.043509,0.187150 0.661982,-0.421728 -0.890326,0.473527 4.789454,-0.651802 -3.322153,0.848726 -0.532947,-0.651988 1.072654,-0.131494 1.260914,0.200394 -1.598766,0.321292 2.330642,-0.447751 2.709954,0.645745 0.806260,0.382261 1.182399,0.204622 -3.593337,0.461006 2.077396,-0.553283 -0.412557,0.691676 -3.608261,-0.270915 1.727379,-0.388903 -0.407110,0.656436 3.555801,-0.468745 0.728860,0.697063 1.650362,-0.222493 0.653116,0.276935 2.221298,0.157120 -1.153824,0.513933 0.074151,-0.612526 2.537242,0.189867 0.758152,0.230724 -0.297831,0.109474 0.024360,-0.110087 1.324597,-0.032275 -1.184568,0.491387 3.515871,-0.535097 -2.292825,0.888716 2.297315,-0.735786 2.390171,0.732432 -2.245896,0.231362 -0.138875,-0.634394 -1.458779,0.376418 1.734079,-0.585812 -0.595667,0.608357 -1.635797,-0.358902 3.470098,-0.376456 0.204787,0.653551 -3.362800,-0.099970 0.178277,-0.394838 -0.011534,0.323557 -2.093272,0.023013 -0.278703,-0.379775 -1.901074,0.207340 0.244085,-0.517124 2.222916,0.287414 0.045798,0.280106 -0.517555,-0.160245 2.502957,-0.206994 -0.421243,0.584276 0.982284,-0.645840 -4.047125,0.571919 2.040596,-0.731873 1.173351,0.692353 -3.427108,0.146988 1.266179,-0.655682 1.080641,0.572102 -2.110041,0.240343 1.409471,-0.756626 -1.143661,0.849308 -1.826080,-0.554941 1.970352,-0.231871 1.644533,0.546937 0.809208,0.201565 0.639134,0.106886 -3.884551,0.008485 -1.405845,-0.758742 1.175232,-0.194266 1.963421,0.687879 -4.818299,0.327482 -2.841239,-0.778957 4.245264,-0.256496 1.830546,0.786253 2.477919,0.277995 -1.459550,0.312032 0.626488,-0.730122 1.103548,0.458044 -0.631375,0.351695 1.675360,-0.618986 0.093386,0.610033 1.054745,-0.238337 -1.771468,0.508709 -2.089866,-0.724123 -1.388358,0.190907 -0.111656,-0.128285 -1.940346,0.235850 3.767787,-0.572005 -1.020760,0.853665 0.962166,-0.884105 2.696399,0.520951 1.082714,0.180624 -0.228912,0.077219 5.169732,-0.433436 1.810938,0.496600 -2.369886,0.084007 0.185799,-0.836410 -0.613984,0.365207 -0.917242,-0.264077 -1.743101,-0.084704 0.405167,-0.149947 0.498632,0.479650 1.936232,-0.006692 1.835464,0.096991 -0.571258,0.108786 -0.356114,-0.555871 0.355020,0.005839 -0.249070,0.351682 -1.139699,-0.213735 1.358663,-0.381628 1.677512,0.828145 0.771420,0.252718 -1.319150,-0.058771 -0.417592,-0.697919 -2.733461,0.087334 -0.246766,-0.225339 -0.175444,0.325380 -2.178403,0.117281 1.353367,-0.404817 3.488284,0.815283 -1.891729,0.239975 0.485812,-0.948403 -0.187566,0.440471 1.654678,-0.262612 -1.667153,0.438567 1.299849,-0.720633 -2.371110,0.656430 -2.377228,-0.727195 0.603058,-0.135212 1.241147,0.903077 -0.077209,0.179427 -0.467998,-0.420243 2.082558,-0.258067 -2.127857,0.602782 0.787509,-0.750236 1.687896,0.598729 2.717239,0.370395 -3.044109,0.006807 -1.190108,-1.143782 1.170666,0.016098 0.731305,0.798805 0.623702,-0.085963 2.883937,-0.160822 1.274123,0.133365 -1.761711,-0.325891 0.464451,-0.989833 1.292753,0.512385 -3.293923,0.252886 1.691704,-0.868227 1.780713,0.743674 -2.658827,0.064153 -0.502248,-1.145637 -1.194599,0.165776 -0.379892,0.027127 0.095166,0.293429 -1.634568,0.339296 1.507682,-0.506572 -1.867568,0.833210 -1.241674,-0.765859 -2.674139,0.015173 2.776339,0.356912 1.324123,1.165237 -4.323702,-0.035116 1.447377,-1.016171 -1.163610,0.881103 -0.447948,-0.722953 3.833957,0.186273 1.168522,0.673945 0.605547,-0.147440 0.840529,-0.295376 -4.870743,0.061951 -0.114560,-0.846725 3.176151,0.426335 -2.177163,0.559201 -2.675747,-0.967812 -0.891259,0.005815 3.405162,0.259275 -0.004254,0.923221 0.914004,-0.448414 -0.679284,0.360723 1.734884,-0.567839 3.583069,0.604903 0.236839,-0.086581 -1.611815,-0.859740 -1.375726,-0.601102 0.801266,0.205899 -1.672980,0.914139 0.009289,-0.490583 -0.259069,0.445883 0.907782,0.140035 1.594800,0.419142 -0.642417,-0.161531 3.452130,-0.962483 0.168897,0.530863 -2.761730,-0.493120 2.226342,-0.572797 0.130470,0.935045 0.986602,-0.420237 -0.039585,0.065416 0.290655,-0.497860 3.305969,-0.052572 -1.838201,0.075797 -2.605489,-1.062798 0.556128,0.100825 4.133592,0.751625 0.808727,0.187615 2.862192,-0.396252 -2.981339,-0.293781 0.264379,-1.138120 2.355626,0.624652 0.155192,0.354914 0.892775,-0.553730 2.953492,0.012486 -0.192357,-0.054198 -1.751091,-0.930937 -0.641921,-0.368097 1.954942,0.368497 0.819300,0.913756 -1.823053,-0.370622 1.945100,-0.863912 2.877238,0.989263 -1.084469,-0.150014 -0.722147,-1.217195 -0.328659,0.129432 3.548920,0.348477 -2.904159,0.639986 -3.976411,-1.041617 0.625995,0.325800 3.812837,1.146785 2.062737,0.109677 -3.026841,-0.339794 -0.555831,-1.091634 2.009931,0.315869 -1.941004,0.802758 0.307477,-1.077432 -2.708462,0.642528 -2.233556,-0.277782 1.636853,0.175060 -3.477633,1.068876 -0.191908,-0.607008 -0.716006,0.472369 -1.998991,0.091063 -0.947660,0.339327 -1.514377,0.508779 4.563892,0.142270 -0.853210,0.979066 -0.990520,-0.673878 1.006525,0.161747 -3.198807,0.858650 -0.647183,-0.905676 -0.238349,0.456729 1.802630,0.312709 2.750439,0.625954 2.085537,-0.300085 1.994980,-0.489318 0.368143,-0.469631 0.267911,-0.782528 2.030658,-0.294886 2.615426,0.072702 -1.484511,-0.420910 -1.228077,-1.195603 -0.247386,0.278563 -2.770107,0.527300 0.658293,0.218414 -0.308988,0.959153 5.860393,-0.394356 2.217579,0.619123 -2.155118,-0.686537 2.392808,-1.071211 3.631443,0.964998 -2.093340,-0.397302 0.644315,-1.161166 -0.526284,0.634748 -1.021519,-0.363408 -1.082698,0.096842 -2.174794,0.301921 2.241500,0.498503 1.834833,1.234307 -0.510191,-0.346621 -1.660114,-1.078487 0.002983,-0.017063 -0.473034,0.683064 3.207511,0.106699 0.857445,0.717729 -0.843555,-0.540995 -0.700647,-0.908388 0.448371,0.458819 0.744195,0.604124 1.850235,-0.104707 0.438576,-0.117807 3.291632,-1.046900 0.270413,-0.456526 -0.434187,-0.756750 3.023373,-0.443680 -0.695367,0.297593 -2.164647,-0.951749 2.108350,0.000142 -0.764333,1.113835 -1.245391,-0.853854 0.087741,0.169250 -0.734166,0.783078 0.341547,0.047490 -2.401807,0.549797 0.349021,-0.403394 -0.708641,0.865481 -1.025700,-0.258231 -0.571604,0.266382 0.775784,0.600706 -2.097193,0.541728 2.214641,-0.687983 -0.689138,0.950411 -2.452787,-0.901776 1.751302,0.108180 -2.154519,1.073718 3.498719,-0.806828 1.203518,1.010065 1.577012,-0.582398 2.356486,-0.398635 -0.939218,-1.040363 -0.672696,-1.056273 -0.991640,0.414148 1.279036,0.332540 -0.552555,0.916806 1.502466,-0.611894 0.830539,0.130763 1.418191,-0.657826 -1.743931,-0.216475 0.983362,-0.831460 0.150267,0.946753 -3.312768,-0.525078 0.129276,-0.400451 0.710307,0.936315 5.140779,0.009288 -1.977169,-0.486849 0.574863,-1.182784 4.117314,0.724040 -1.609225,-0.157353 1.505116,-1.031709 1.207174,1.035052 2.126620,-0.790023 1.054527,-0.773523 0.262582,-1.334427 -0.077302,-0.770703 1.015018,-0.205547 0.641220,0.068197 2.212967,-0.921911 2.615376,-0.002260 0.917321,-0.526206 0.696456,-0.976522 0.109355,-0.545434 -2.754226,-0.651024 1.108060,-0.006573 -3.056535,1.015926 0.339373,-0.456813 1.478511,0.793641 5.504446,-0.303915 1.540248,-0.758999 0.604359,-0.822267 1.599543,-1.123308 2.023336,-1.028466 2.588798,-1.205887 -1.746723,-0.737580 -0.898096,-1.038726 -1.458169,0.431106 0.441341,0.437957 1.887139,0.865656 -1.466753,-0.429948 -0.481884,-0.804239 -0.766668,0.588804 -1.096315,0.446393 2.404434,0.699867 4.827782,0.898988 -1.854384,-0.812782 -0.056294,-0.964677 2.902948,0.758755 -0.155599,0.254058 1.119412,-0.804861 2.160025,-0.059631 4.098785,-0.523561 -0.625359,-0.813885 -1.027706,-0.896014 -3.474154,0.239840 1.102717,0.505734 0.444059,0.987136 3.285356,-0.804296 2.323413,-0.494043 1.193069,-0.829574 -0.325563,-0.756341 -3.712319,-0.846043 -1.334324,0.325647 -2.110559,0.903839 -1.420379,0.914551 3.751914,0.763288 -4.416652,0.746636 2.119820,-0.787379 0.631021,0.807347 -0.604212,-0.808243 -1.245308,-0.509367 4.661566,0.263503 -0.900983,0.907107 -1.128527,-0.942353 -1.664089,0.521365 -2.315327,0.712901 1.404807,1.448173 0.157352,0.884578 -4.897677,-0.805091 -0.373098,-0.310460 1.598317,0.893932 -2.769286,0.067482 1.315912,-0.675378 -0.337776,1.072902 1.562290,-0.676591 1.286621,0.046706 -0.471052,-1.267482 -3.295689,-0.747934 -0.326306,0.479773 -1.462097,0.627813 1.841051,0.705064 -2.531474,0.849594 -2.222057,-0.712308 -2.877527,1.235836 -0.200606,0.924589 1.087448,0.764412 -0.535484,0.179028 -1.525165,-0.721316 -0.098162,0.140933 -0.561094,0.794807 -2.152315,0.335464 -1.911666,0.869316 0.682348,0.517537 -2.134715,0.822405 -1.178562,-0.756727 0.386538,0.576145 1.010317,0.778115 1.517951,-0.284485 -2.908798,-0.872097 0.453147,-0.740347 0.464136,0.840231 2.272586,-0.493551 0.167489,-0.706718 -1.914636,-0.882137 -0.375054,0.080734 1.167553,1.044915 -4.340945,0.142147 3.452911,-0.668974 1.556160,0.684756 0.285100,-0.835062 -2.842874,-0.905781 -0.351644,-0.090181 2.949769,1.003691 4.245955,0.123973 -2.118096,-1.272227 -1.016367,-0.661138 -0.456498,1.088457 -0.291807,0.811030 1.279544,0.364237 -0.235060,0.179367 0.947649,-0.668531 -2.453485,0.181867 -3.214748,-0.355130 -2.775844,0.720067 3.904496,1.071374 0.146454,0.498583 2.170665,-0.810703 -0.237482,-0.419174 2.664655,-0.826981 0.080329,0.353435 -0.047559,-0.964800 -3.024718,0.028085 -0.552597,0.041691 2.567450,0.628606 -0.911263,0.183053 2.761624,-0.620034 -1.573535,0.546102 -3.753133,-0.708938 -1.301584,1.036788 -0.113085,1.203476 -0.205571,0.606062 2.270562,0.070624 1.197657,0.106720 0.509475,-1.004959 -0.377393,-0.901242 -4.282631,-0.263544 0.101623,0.307980 0.483258,0.683192 -1.986375,-0.267509 -0.336423,0.124484 -0.739206,0.784666 0.110198,0.582726 -2.825942,0.497031 1.985974,0.086465 0.006940,0.794922 0.196468,-1.015057 0.612024,0.196507 -2.063411,-0.343424 1.362678,-0.258343 0.503853,0.601622 2.448111,-0.923640 1.335915,-0.873112 1.744228,-0.862334 -1.718301,-0.409818 -2.813859,-0.599566 -4.989359,0.667187 -0.030831,0.924645 -2.178939,0.474610 2.275291,0.127488 3.593787,0.708162 -2.193889,-1.398328 -2.295179,-0.531912 -0.386217,0.797915 1.780833,0.676438 -2.575234,-0.003209 -0.826786,-0.371197 2.203201,1.110091 -1.095655,0.194819 0.206617,-0.690251 -2.170980,0.625159 -1.791957,0.285354 1.886926,0.896845 0.991287,0.319776 0.010380,-0.774615 -0.191760,-0.389997 -0.648330,0.224636 3.459672,0.229688 -4.537318,0.166495 1.025572,-0.686460 -0.749781,0.794721 -0.870310,-0.621903 -2.104329,0.996366 -1.515516,0.287383 -2.179346,0.933684 3.582889,0.602963 0.234949,0.353879 3.657755,-0.887813 -0.934639,0.014241 0.985279,-0.588158 -1.920966,0.397867 2.109130,-0.381492 -0.240991,0.605302 2.213550,-0.704827 0.295594,0.240769 2.348941,-0.982510 1.675148,-0.192756 0.195421,-0.753034 -4.112431,-0.503664 -1.610295,0.559925 2.830097,0.957511 -3.378145,0.191236 -1.860564,-0.848363 1.613155,1.456609 -0.594348,0.129939 1.417948,-0.710891 -3.562555,0.271732 1.314543,-0.426854 1.544944,0.440881 -0.789000,-1.075231 0.421841,-0.174466 -0.518948,0.508217 3.119246,-0.272915 -2.195123,0.053056 -2.437619,-0.650923 4.940513,0.925149 3.523367,0.182024 0.421110,-0.728976 0.251474,-0.584640 -1.439296,-0.139247 -1.949734,0.068356 0.679084,0.848085 -2.077165,0.285294 -2.916146,-0.468516 -0.087036,0.693087 -2.926196,0.439461 -1.749095,-0.145004 0.817654,1.106220 0.754035,0.107172 0.880173,-0.653539 -0.739256,-0.314157 0.428257,0.011703 1.692734,0.512357 0.349025,-0.754110 2.367714,-0.273976 2.882656,-0.159145 1.238716,-0.710149 -0.469409,-0.537561 -0.541461,-0.132576 0.450231,0.595173 1.674438,0.135769 -4.497130,-0.650852 -4.089990,0.010425 1.570537,1.308001 -0.734701,0.071962 1.367510,-0.583390 -4.696424,0.200659 -1.642881,-0.394193 3.928557,1.048715 2.176777,-0.169657 0.623086,-1.009571 -0.880935,-0.434615 -1.221495,0.071233 -3.761158,0.524133 -1.142991,0.452413 1.908015,0.361456 0.498735,0.024935 0.243901,-0.982314 2.939869,0.037607 0.191142,-0.102666 -3.559238,-0.547887 -0.142312,0.504847 2.034900,0.634732 -1.611245,-0.468699 1.579140,-0.140194 -0.848947,0.344732 -0.859010,-0.605688 0.338884,0.700462 0.551770,0.029510 -1.299508,-0.539957 1.646022,0.121938 0.412172,0.339052 -0.433647,-0.701121 0.244093,0.267202 -3.427129,0.153118 0.802418,-0.063085 -0.823753,0.223517 -2.188280,-0.470436 -0.433145,0.895743 -0.506655,0.175699 0.948985,0.046273 -0.549890,-0.154145 0.082389,-0.372624 -2.853124,0.380997 2.067663,0.112798 0.100403,0.190680 0.738297,-0.780099 1.192438,0.329784 -0.062408,-0.441619 -0.910717,-0.128670 -1.496459,0.289653 -1.754770,0.602792 2.652556,0.443955 0.973313,-0.111433 -0.082867,-0.998415 0.979348,0.171308 2.480116,0.165960 -0.403610,-0.526040 1.025522,0.102343 -0.004280,0.369801 1.453992,-0.350700 -0.743939,0.035040 -2.095146,-0.253234 1.375224,0.766119 0.427954,-0.080101 -0.414713,-0.677907 2.552272,0.394738 1.789742,-0.012714 0.247922,-0.700595 0.340040,0.102734 -0.534143,0.148726 -0.348316,0.132489 -1.495265,0.186685 -0.140416,0.187894 -1.597978,0.020655 3.110866,-0.015655 -2.641514,-0.055577 0.670099,-0.389298 -1.890679,0.379330 0.980368,-0.240991 -1.526686,0.157873 0.263263,-0.352455 2.343405,0.537347 2.752376,-0.661979 3.298349,-0.204183 0.872012,0.033837 -1.216381,-0.074158 -1.344438,0.249934 1.491969,0.697423 -1.635209,-0.441872 -1.654837,-0.379445 1.890619,0.747567 1.825959,-0.177581 1.374870,-0.780948 0.503584,0.010893 1.689840,0.233974 -1.439922,-0.019203 1.028715,0.262173 -1.933981,0.281825 3.091864,-0.215571 0.863388,0.466196 3.896298,-0.432323 0.518406,0.242455 -0.646122,-0.064070 0.356455,0.450305 -0.712269,0.167176 -1.016233,-0.294477 -2.049555,0.126465 -1.104138,-0.108484 -3.124664,-0.200948 -1.200784,-0.237518 1.901851,0.069092 -0.305550,-0.496474 -3.135555,-0.491635 -0.329039,0.493786 0.634823,0.122746 0.836046,-0.280669 -1.132302,-0.383104 -1.804995,0.076730 -1.036325,0.307008 1.144459,-0.020725 -0.034270,-0.416488 -4.074012,-0.499949 1.165380,0.563989 -0.216645,-0.072678 -3.536575,-0.552039 0.243790,0.441012 2.241323,0.049007 -2.134641,-0.630648 -1.189464,0.083958 0.284009,0.500696 -1.889812,-0.528135 -1.078496,-0.355738 -3.619392,0.057263 -0.553200,-0.350914 1.652166,-0.143549 -4.514567,-0.612593 -0.941475,-0.248561 -2.772590,0.266093 -0.927777,-0.379235 -1.661385,-0.381579 -4.155130,-0.482438 0.821497,-0.075304 1.843545,-0.174220 -2.525575,-0.595540 1.544014,0.490303 -3.765286,0.092135 1.628941,-0.158424 1.035643,0.407047 -1.624454,-0.537102 -3.372650,0.295028 -0.888803,0.153276 -0.854142,-0.457271 0.173387,-0.349944 -2.402719,-0.459460 3.811865,-0.060524 -0.529569,-0.098173 3.795429,-0.093031 -3.180336,0.196763 0.757872,0.032330 1.121454,0.402764 0.547803,-0.266307 2.511057,0.500029 0.433199,0.450363 -4.267105,0.081977 -2.145089,0.278140 0.494795,0.088578 -0.467859,-0.524149 0.917817,-0.183003 -0.250549,0.120617 -4.801555,-0.219453 2.346085,0.406331 -0.452780,-0.301207 -2.306406,-0.421735 0.083478,0.318341 0.433805,-0.178650 -1.300452,-0.504512 -4.623793,-0.130492 -3.117816,-0.124443 -2.576355,-0.720519 -1.090883,-0.765216 -1.081288,-0.643459 -2.834514,-0.767360 -1.223916,-0.571026 -1.733289,-0.705026 1.861729,-0.356213 -2.098336,-0.627351 -1.470595,-0.388065 0.372136,0.212336 2.350458,-0.259192 -0.292216,-0.210024 0.537682,0.410528 0.978081,0.529387 -1.681839,-0.097651 -1.730866,0.104217 -0.598606,-0.094316 -1.268711,-0.694071 -0.554415,-0.538310 -2.977313,-0.650711 -6.035644,-0.645006 0.207270,-0.381332 0.797705,-0.361454 5.653706,-0.040524 -1.921213,0.163112 1.850328,0.600974 -5.110741,0.032957 -1.223969,-0.329201 4.237228,0.176575 1.808240,-0.273457 -0.036196,-0.034484 -1.983169,0.426122 -0.412363,0.143026 2.259738,-0.210029 -2.770695,-0.555003 3.838362,0.265723 -1.247252,0.009027 1.792609,0.036942 0.877046,0.432408 3.581934,0.389985 -1.180292,0.415383 0.824946,0.607132 -1.750070,-0.028997 0.425324,-0.205499 2.480918,0.187419 -1.518606,-0.160851 2.875997,0.476935 3.546587,0.394764 -3.208811,-0.149575 1.563734,0.738130 0.571906,0.304653 -1.403764,-0.014673 1.946519,0.551394 -0.784531,-0.137649 -2.452515,-0.338235 0.333777,-0.057041 -2.325288,-0.854066 -0.733075,-0.725688 -0.810839,-0.611355 2.280757,-0.405334 1.271553,-0.106913 1.960673,0.767883 -0.342304,0.741582 -1.554468,0.438620 0.115279,0.084741 0.501472,-0.408720 4.423890,0.160456 3.942066,0.848220 0.539801,1.151991 1.383087,1.342130 0.972949,1.019176 2.066707,1.069867 2.467966,1.011665 -0.039139,1.027216 -0.895529,0.640937 0.534406,0.261820 0.721074,-0.079431 -3.184671,-0.117201 2.109527,0.335536 2.518417,0.221495 2.006944,0.770756 0.983886,1.356624 -1.888078,0.647555 4.043038,0.675669 0.589316,0.179894 2.624233,0.829808 -2.536388,0.699641 -0.522386,0.093579 2.683558,-0.184226 3.007507,0.194024 -0.165699,0.696870 -0.929732,0.539884 -2.334170,-0.430648 3.984747,-0.717923 1.024471,-0.138715 2.315813,0.615442 -1.875478,0.647649 -0.953612,0.244129 -1.847117,-0.606940 -0.954009,-1.344936 -2.673319,-1.224386 -0.529601,-1.089266 0.251728,-0.746899 1.026821,-0.238480 0.327549,0.535192 2.226797,0.886106 -2.371748,0.527238 1.858284,0.374760 -1.962049,-0.090378 0.394442,0.037304 0.377580,-0.066869 -0.580646,-0.062077 -4.534454,-0.201566 -1.941630,-0.983585 -1.724027,-1.750329 0.925034,-1.037386 -1.055364,-0.768412 -2.364644,-0.463987 -1.782999,-0.736491 -0.134633,-1.230707 0.838762,-0.656894 0.866068,0.099557 -0.709928,0.473131 -0.033724,0.355059 -1.332093,-0.591555 1.523301,-0.296539 3.154369,0.169733 0.149091,0.861379 -1.479515,0.615192 1.178425,0.002248 -3.477596,-0.345819 0.160113,-0.329617 0.115812,-0.423778 0.148519,-0.077133 -3.927164,-0.079307 -3.429235,-0.739929 -1.863797,-1.700331 -2.116530,-2.106964 0.868179,-1.115797 -1.760117,-0.784156 3.014207,-0.022159 2.356997,0.459987 1.388150,1.025904 3.986713,1.824091 -0.005196,1.632156 0.745494,1.170426 4.286449,1.065704 -1.769810,0.897511 -0.257819,0.245124 1.221483,-0.333971 -2.516190,-0.151906 -1.166840,-0.392308 1.267704,-0.579479 -0.986269,-0.503208 4.535639,0.185690 -4.194096,-0.017147 0.533669,-0.070409 -0.383427,-0.261635 -1.556613,-0.390831 1.408860,-0.406829 -1.204047,-0.507705 -0.423410,-0.192359 -1.990966,-0.964710 3.453477,-0.924132 -3.766101,-0.351604 2.186090,0.225257 -1.567214,-0.276206 0.727655,0.146120 -2.088692,-0.399224 0.746060,-0.260027 1.040362,0.110574 -2.057865,0.483500 2.510626,0.379487 0.788878,0.363303 -1.475797,0.787347 3.472866,0.423868 -1.563606,-0.118551 0.003739,-0.011624 1.417530,-0.247668 1.565037,1.046463 2.323692,1.947657 -3.510085,1.242456 0.105657,0.122480 -1.377676,-0.874229 2.297901,-0.496427 -0.575985,0.007278 1.244225,0.676281 -0.093408,0.377828 -0.193213,0.413860 -0.783830,-0.276916 1.436398,-0.458166 1.820820,0.529243 2.356799,1.479705 -0.467328,1.759664 1.212707,0.929116 2.380574,0.850852 -1.506107,1.171218 -1.767047,-0.238253 0.403948,-1.413292 -1.448993,-0.828976 -1.734073,-1.032907 -3.473395,-1.674315 2.221579,-1.511085 -0.769199,-0.341042 -4.716993,-0.276143 -0.933309,-1.371563 -0.200374,-1.855979 -0.063492,-0.925163 -2.503633,-0.697608 2.102007,-0.721044 -1.070460,-0.336620 -0.978726,-0.292878 3.462467,-0.822476 2.761370,0.248665 0.318502,1.860497 -0.868642,1.042289 -1.049327,-0.613227 0.763118,-1.102952 3.445671,0.272083 -0.757695,1.207169 2.231393,0.686302 -1.556943,0.234209 2.586200,0.345226 -1.189742,-0.142960 0.506734,0.177779 -0.443166,-0.445209 -0.583138,-0.300245 3.033381,-0.377451 0.238733,0.659142 -0.708513,0.891915 0.660094,-0.017879 1.454268,0.379523 0.358085,1.389351 2.038293,1.763424 -1.194947,1.178497 -3.684536,-0.291425 1.949918,-1.191975 -0.095145,0.013025 2.022156,0.887996 -2.123984,0.620140 -4.263941,-0.579695 0.928220,-1.418036 1.268366,0.152181 -1.876091,1.084551 0.163779,-0.078865 1.982872,-0.348487 -1.164843,0.855690 0.858206,0.143974 0.406576,0.156159 2.305060,1.232124 -1.592549,1.167469 1.930201,0.190960 2.631654,0.627591 -3.117920,1.391560 4.837580,0.231729 -3.536349,-0.151155 -2.453813,-0.640644 -1.391402,-2.268156 -0.246256,-1.838342 1.189658,-0.866446 -3.057386,0.354147 2.093411,0.083842 -0.953033,-0.179737 2.581851,0.483220 2.272408,1.085550 0.852214,2.021511 0.769477,2.233716 -1.674792,1.073970 -0.734982,-0.908877 2.492214,-0.979405 2.262368,1.093480 -2.189238,1.831701 -0.250813,-0.324306 -1.243908,-1.461063 1.743377,-0.900370 -2.438929,-0.096069 -2.290244,-0.686294 -0.402455,-2.012562 0.980511,-1.151502 0.184272,0.559651 -0.035701,0.900054 0.930185,0.471210 -4.151130,0.378662 -4.052601,-0.708523 -0.607125,-2.146299 -1.351672,-2.179910 0.444111,-1.566411 -1.784256,-0.767114 1.645387,-0.350248 1.998304,0.700029 0.609550,2.500671 3.233072,1.918430 0.021523,1.854192 -1.747249,0.631584 -1.453336,-1.542261 3.592156,-1.540327 1.483334,0.649869 1.190150,2.768546 -0.232766,1.798972 1.987457,0.929109 -0.042344,0.982146 -1.359844,0.559712 0.512586,-0.899314 0.150030,-0.135976 -1.323591,0.335311 -1.004602,-1.088149 0.545320,-1.475271 2.766540,0.209524 -0.416621,1.634685 0.569847,0.675429 -1.608602,-0.044001 2.137148,-0.266805 -0.638373,0.177260 2.887020,0.508879 2.732132,1.299818 -0.885245,2.077532 1.704531,0.369020 -1.054723,0.297584 1.349840,0.218160 0.221901,0.555167 1.076253,1.274862 -0.713703,1.016689 0.617899,0.140603 1.928993,0.517828 -2.296569,1.548956 3.540521,0.107909 0.246394,0.494122 -0.538262,1.248785 -0.684783,-0.506445 1.544711,-0.835217 1.989266,0.948680 0.331134,2.181471 1.976824,1.924312 2.347263,2.203040 -2.206261,1.896994 -3.452108,-0.772645 1.320623,-2.190793 2.210958,0.528278 1.521602,2.848393 -2.157833,1.851245 -0.226349,-0.609585 0.791401,-1.085167 -1.688532,0.371536 1.640704,-0.325169 2.194247,0.851833 -1.542743,2.150793 2.083262,0.205834 -0.668363,0.171658 0.506189,0.283864 -1.298627,-0.190433 -2.350417,-1.234309 -1.911195,-2.716401 4.648082,-1.878244 -3.442226,0.009748 -1.425055,-0.906345 -1.682936,-2.636737 0.303478,-2.067004 -1.892076,-0.972789 1.260404,-0.615099 -1.335411,-0.145159 1.255444,-0.158196 1.218929,0.759520 -2.844482,1.714228 1.134135,-0.291662 4.686372,0.616485 3.480431,2.918150 3.560315,2.940934 0.594910,2.727172 0.041107,1.812247 -0.477174,0.507081 -2.317826,-0.913830 1.775601,-1.443272 -0.593988,0.177478 -0.597507,-0.083689 -4.852369,-1.585423 -0.760049,-2.327367 -0.113382,-2.031171 -0.133400,-0.901055 0.210081,-0.229733 0.548245,0.405311 -0.523741,0.721708 -2.580340,-0.607741 -0.366102,-2.153100 -1.716046,-1.832687 0.848825,-1.413275 1.998934,0.435044 3.618727,2.599891 0.577792,2.741853 2.108819,2.182818 1.395198,2.347834 -1.831207,1.918031 1.852071,-0.197203 -1.589682,0.032858 0.244925,-0.298050 1.770813,-0.155483 1.400821,1.915382 -3.503064,1.813147 -3.601753,-1.103370 6.084084,-2.057903 -1.602467,0.198341 0.935802,-0.311454 1.995970,0.677794 -1.934492,1.922365 0.829094,-0.334896 -0.724233,0.027571 1.388218,0.067446 -0.471921,0.690877 3.331294,0.488941 -2.836924,0.621306 5.045630,-0.189117 -0.837411,0.541659 -2.143166,-0.552409 1.573304,-1.575377 1.845163,0.780069 -2.560290,2.029226 3.197090,-0.142910 -2.897231,0.147000 1.417279,-0.357736 0.240150,0.639488 -0.533700,1.054355 -0.869404,-0.715628 -2.973629,-2.008461 1.423504,-1.822806 -5.017376,-0.011837 -0.297419,-0.856251 -3.419694,-1.814254 -1.788785,-2.242066 -2.950418,-2.871910 1.212349,-2.102193 -0.649303,0.110004 -1.385675,-0.508339 -1.546949,-2.269582 -2.435436,-2.678610 2.685798,-1.937072 3.270506,1.194543 0.565400,2.658891 -1.611221,1.287697 -0.221036,-1.031020 0.314486,-1.081008 -2.455408,-0.230152 -2.164949,-1.673850 -0.375547,-2.555598 1.589318,-1.077995 -2.186622,0.731713 -2.736670,-1.157436 1.066511,-1.945829 -1.818516,-0.050469 0.277940,-0.748364 -1.196439,-0.726548 0.514631,-0.821652 2.344145,0.429123 1.004263,2.254490 1.502630,2.473953 -1.229840,1.744440 -2.587088,-0.944153 -0.127807,-2.261479 -0.411589,-1.272973 -0.650260,-1.073463 -1.453072,-1.570640 -0.005358,-1.952262 5.022468,-0.526129 0.336714,1.689876 0.071756,1.422590 -3.193819,0.063985 0.653744,-1.288889 -2.211593,-0.382349 -0.695956,-1.409244 3.533765,-1.368123 0.735944,1.097677 2.227602,2.265857 -3.439174,1.654042 1.111338,-0.352584 2.903415,0.915600 -2.441377,1.826005 -2.606165,-1.059263 -2.947011,-2.812654 1.337692,-2.277952 1.363057,0.764152 -0.198634,2.056911 0.158563,0.560048 1.971559,0.454446 -0.489062,1.421820 -0.561812,-0.095939 1.391824,-0.659709 -0.211265,0.850912 0.270353,0.521891 0.339689,0.444371 0.434901,0.878515 1.048430,1.247192 0.701474,1.833067 -2.595722,1.280198 -1.395249,-1.369868 0.357347,-2.005531 0.969202,0.011677 0.761423,1.744500 -2.142409,1.235111 -3.153691,-1.357403 2.420732,-1.937667 1.634350,0.864266 1.850801,2.459322 -0.229369,2.348169 1.832058,0.973264 2.535221,1.772417 3.582746,2.659598 -2.251095,2.148673 -0.565259,-0.907533 0.672295,-1.299729 -0.365479,0.337681 0.489859,0.007102 -0.273346,0.346291 0.332268,0.027435 3.888832,0.729754 0.612352,1.946945 2.106723,2.074413 -1.660858,1.674275 -0.916685,-1.009465 -0.163171,-1.861184 0.579912,-0.582957 1.094811,1.006719 2.541023,1.997870 3.195308,2.509260 3.343908,2.737074 0.003740,2.487285 -0.107363,0.809430 1.448535,0.377328 -0.173206,1.203839 0.955128,0.684912 1.623416,1.513159 -0.198879,1.862730 -0.549624,0.402346 -1.186595,-1.027462 1.637393,-1.231491 -1.052549,0.482508 2.967870,-0.213747 -4.455350,0.600728 1.028789,-0.542171 0.771100,0.824114 0.875574,1.654533 -0.244993,1.591881 2.376870,0.572570 -0.934008,1.017749 0.321368,-0.292599 0.629224,0.333555 1.961328,1.282020 -4.058099,1.445078 0.465075,-0.569263 0.354336,0.101315 0.389561,0.897979 2.093094,1.264631 0.567922,1.990446 3.674849,1.767224 -0.000804,1.958858 2.158418,1.203063 -3.822610,1.162686 -0.620638,-0.960209 -0.766488,-1.564255 -1.386707,-1.901366 2.257693,-1.469684 3.139490,1.139287 -1.729779,1.625622 1.694528,-0.278877 1.186511,1.054099 -1.012130,1.411987 -0.910983,-0.750456 0.330864,-1.297638 -0.711607,-0.482245 -3.455218,-1.078278 0.511324,-1.624643 1.197729,0.158628 0.364688,1.457542 -0.015179,1.190306 -0.078520,0.376336 -0.078548,-0.063624 1.557197,0.165351 2.333331,1.577730 -0.739476,1.764959 -0.842343,-0.353348 1.527670,-0.869255 1.955073,1.035349 1.602459,2.079056 0.614264,2.049507 -0.146586,1.359713 1.454160,0.485747 0.594334,1.446096 1.207647,1.496465 -2.200522,1.194164 2.818778,-0.335348 -0.905897,0.633558 0.910216,-0.362364 4.232254,0.985445 1.324045,1.937416 1.362459,2.197562 2.034892,2.105485 -1.836400,1.488623 0.825664,-0.346598 -0.740954,0.261658 0.950429,-0.430681 1.357242,1.013237 0.017853,1.497264 -0.280519,0.665002 2.556428,0.149681 -1.551101,0.857932 -1.766977,-1.003510 1.499172,-1.258648 -1.434993,0.255688 -1.780055,-1.119482 -1.431939,-1.809231 -2.513798,-2.312264 -1.419402,-2.451924 1.581690,-1.488575 -0.155600,0.467526 1.978417,0.428124 -6.454806,1.021382 0.792758,-0.536230 -1.658583,0.230766 0.946751,-0.764004 -2.710224,0.265879 -0.482169,-1.063650 -1.300170,-1.216542 -1.643957,-1.735891 0.781105,-1.549744 2.481779,0.450462 0.255304,1.428042 2.745677,1.142310 -1.302043,1.153535 1.153768,-0.139595 1.462779,1.066445 3.016793,1.621716 1.839349,1.986327 -2.965053,1.605538 1.119375,-0.138390 -0.759275,0.487591 0.389201,-0.485468 -1.993507,0.019389 -1.652123,-1.463863 -0.383762,-1.470478 -1.444623,-1.277488 -0.295481,-1.498303 0.304110,-0.666361 -0.417286,-0.097650 -1.122874,-0.657526 1.137507,-0.776279 0.423290,0.507099 2.073824,0.843675 2.494242,1.816479 1.342716,1.769050 1.012803,1.922854 -2.376188,1.239437 0.995354,-0.292703 -1.101431,0.292407 -1.287374,-0.996153 1.443362,-0.789832 4.103421,0.657133 -0.450950,1.049783 -0.009188,0.254480 2.413999,0.400163 -1.965810,0.609052 1.864271,-0.351370 0.104875,0.869706 -1.402579,-0.022223 -0.298201,-0.681140 3.435574,-0.316047 -0.860020,0.413898 -0.311812,-0.484476 1.748394,-0.059164 -3.305585,0.390570 2.818831,-0.422875 2.963684,1.241530 -2.845403,0.861115 -0.147240,-0.474763 0.035499,-0.253566 3.657724,-0.005526 0.324042,1.067187 1.973309,0.844723 -1.741777,1.008695 0.367595,-0.358533 2.594280,0.551689 0.970227,1.016974 3.178892,1.451574 -5.022872,1.151401 0.906989,-0.268087 0.136002,0.612583 -0.019988,0.076435 -2.567153,-0.037698 -4.251829,-1.158312 1.513069,-0.911139 -1.397154,0.001431 -2.821060,-0.983184 1.867672,-0.632030 0.570883,0.353944 -1.852732,0.229015 -1.724024,-0.628870 4.825176,-0.656302 -2.293730,-0.024356 1.630827,-0.512645 0.436546,0.937196 3.329828,0.532104 1.199481,1.563020 -0.229564,1.064915 0.410414,0.551551 1.124347,0.640860 -1.691030,0.483900 1.513751,-0.228024 -0.657955,0.461371 -1.898503,-0.746933 -0.054421,-0.657778 -1.964974,-0.870816 -3.198524,-1.220027 -1.529566,-1.110498 -3.069314,-1.592476 -1.853393,-1.749685 -0.573622,-1.391494 2.879500,-0.783176 -2.037518,0.019988 1.880443,-0.352033 1.072567,0.940911 2.524077,0.796005 -0.139251,1.123221 2.569849,0.644394 -0.572802,0.886992 1.947442,0.073941 0.401564,0.933814 -1.770355,0.135710 3.481966,0.001317 1.112738,0.752345 1.556692,0.698034 -1.853883,1.010621 0.622910,-0.032626 0.362522,0.492902 0.071969,0.072699 -0.298368,0.235131 -0.758198,-0.180160 1.189681,-0.255929 1.060756,0.409544 -0.332768,0.375012 -0.258526,0.257608 -0.229833,-0.023244 0.123309,-0.262410 3.268740,0.162507 0.472279,0.673428 0.477897,0.688678 -2.456349,0.502427 1.538681,-0.084123 1.774985,0.715706 -1.152110,0.130662 0.146020,0.090628 1.606156,0.437218 -1.609132,-0.029930 -2.213975,-0.419141 0.716606,-0.165615 -1.292042,-0.476665 -1.083590,-0.827811 -2.127893,-0.408863 0.343646,-0.741609 3.435672,-0.004016 -2.501703,-0.029687 -0.775506,-0.331490 2.145981,0.109713 -2.306331,-0.241121 -2.140993,-0.752245 -1.172367,-0.370288 2.867424,-0.644124 3.120876,0.166920 -3.295846,0.095711 -1.663287,-0.056596 -1.727111,-0.348748 -0.400515,-0.978002 2.336574,-0.438014 1.155252,0.143347 -3.392735,0.092293 2.008378,0.296055 -0.570732,0.356643 3.199943,-0.155241 -1.158806,0.469074 -1.330516,-0.514017 -4.143003,-0.345385 2.080118,-0.526243 0.876435,-0.009862 -1.271452,-0.231422 2.737661,0.364110 0.034803,0.486551 -1.556905,-0.380448 0.757997,0.210065 -0.812000,-0.025724 -1.343276,-0.602488 -3.307515,-0.206076 0.194587,-0.489388 0.240277,-0.337908 -0.356286,-0.326752 2.407236,0.283099 3.354165,0.390567 -0.022946,0.110134 1.647893,0.673026 -2.283508,0.318257 -1.794405,-0.277683 -0.223587,0.042013 -1.689069,-0.711669 -0.546814,-0.594931 1.986138,-0.059451 -2.887411,-0.408738 1.659994,-0.093151 -0.658845,0.281109 -3.107872,-0.627826 0.023037,0.168252 3.010880,-0.076939 -0.549179,-0.322785 -2.850668,-0.171873 0.391412,0.302672 -0.645121,-0.272313 -0.620344,-0.505683 -0.329447,-0.022187 -0.053675,-0.213077 -1.260546,-0.390703 -0.375804,-0.254135 1.789100,0.094956 -0.793633,-0.409429 1.475506,0.013809 0.666147,0.431700 1.438775,0.004174 2.998897,0.523397 -0.252021,0.214213 -0.048871,0.356150 -0.976468,0.199543 2.087043,0.083112 1.438523,0.112320 0.584723,-0.140861 -4.330720,0.128445 0.863667,0.436638 -1.211954,-0.008998 -4.567086,-0.490386 3.503573,0.476854 -0.253764,-0.398095 0.916390,-0.219666 -1.395314,0.142563 0.774953,0.111150 0.838316,0.315395 0.513955,-0.238715 -0.729498,0.068034 -1.038021,0.136552 2.101233,0.406482 -1.094278,-0.403644 0.308163,-0.175439 3.614859,0.436450 -0.301874,-0.447501 0.502292,0.142129 -0.303342,0.245382 -2.315322,-0.225355 4.763431,0.466978 -2.937397,-0.334104 -0.859806,-0.332400 -2.779110,0.361638 0.195899,-0.140577 -0.408529,-0.297637 -0.046737,-0.296521 3.635713,0.279384 -3.660917,-0.503668 -0.266809,0.097835 -1.982141,0.314738 0.762392,-0.012846 -0.540617,-0.237418 2.887660,-0.120024 1.296951,0.111619 0.244267,-0.388349 0.873829,0.253286 -0.379888,0.017824 -0.877395,0.003603 -2.151150,0.260488 4.513098,0.351963 -0.951246,-0.555820 2.640746,-0.009309 -2.485337,-0.058538 -1.267696,-0.123531 -1.962942,0.561291 2.641609,0.214432 -1.850464,-0.646739 -1.232907,-0.192558 -0.109288,0.669718 1.438383,-0.045874 0.441039,-0.512143 -0.267317,-0.316041 -0.850892,0.259091 -1.351347,0.351989 1.653363,0.480336 0.436960,-0.427318 -0.793846,-0.630697 -1.078783,0.386174 1.163850,0.668924 -0.643999,-0.556576 0.722691,-0.014054 2.884156,0.143161 -1.253090,-0.821069 -3.455040,0.327047 -2.390430,0.774578 1.401212,0.431767 -0.835300,-0.656239 0.941967,-0.015646 -1.010369,-0.039859 3.776586,0.119171 -0.711168,-0.214388 -1.923279,-0.192775 -2.124275,0.782707 0.909661,0.545617 0.169874,-0.469059 1.213713,-0.401285 4.550078,-0.191415 -1.012904,-0.698327 -2.541892,0.428705 0.156666,0.982404 -0.068354,-0.143694 1.147586,-0.185968 -0.434217,-0.501143 -0.006822,0.083921 -3.459036,0.177039 1.612700,0.590791 4.318436,-0.285511 3.403914,-0.988166 1.055507,-0.516048 3.471187,-0.318833 1.073928,-0.449228 -2.653946,-0.625287 3.859973,0.901464 -2.865097,-0.337019 0.700099,0.299860 4.891289,0.163132 -1.225514,-1.142421 1.372105,0.616129 -2.080101,-0.230341 1.142096,0.317524 0.169329,-0.128950 -0.438651,-0.569520 -0.894827,0.425257 1.316125,0.787538 -0.371206,-0.611904 2.121006,-0.030696 1.480656,-0.388681 1.214888,-0.923270 1.578629,-0.577020 1.436266,-0.620229 -0.161334,-0.711629 0.199469,0.152534 0.718464,0.138857 2.164074,-0.510369 0.469063,-0.876675 -1.142167,-0.597218 -0.397228,0.786829 -3.337158,0.584409 4.231999,0.876404 2.879758,-0.646633 -1.118382,-1.496818 -0.076869,0.668592 -1.203654,0.478555 -0.074667,0.606002 1.779369,0.320116 -1.937748,-0.970882 -0.945636,0.539103 -1.977291,1.100678 -3.762396,0.827117 -0.230578,1.104776 -0.593770,0.292642 2.644237,0.419285 -0.321415,-0.911977 1.284282,-0.099491 0.247139,-0.458401 2.458079,-0.515313 0.899811,-0.816548 -2.133694,-1.109860 -0.183901,0.920736 1.134572,0.662968 1.532256,-0.890002 2.281906,-1.217126 0.006911,-1.007520 -4.768914,-0.252963 0.563196,1.273245 1.720359,-0.125526 -0.440675,-1.275138 -0.147266,0.197171 0.637955,0.483691 3.513395,-0.575114 -0.184236,-1.319791 -1.781791,-0.073628 3.333081,1.202948 -1.157133,-0.841335 2.968826,0.402895 0.115772,-0.684204 0.863783,-0.590428 1.452670,-0.657184 4.872012,-1.188138 -0.815132,-1.308178 0.636432,0.557431 0.710767,-0.246654 2.066811,-1.017801 -0.264876,-1.192581 1.598017,0.007878 -0.847488,-0.787157 -1.296909,0.373605 3.949158,1.293633 -0.399529,-0.969973 -1.599914,-0.085232 3.088075,1.323291 0.807840,-0.911098 -2.122931,-1.345287 0.655019,1.057244 1.285722,-0.178687 -1.703769,-1.508311 -1.559108,0.836474 1.652387,1.756930 1.494935,-0.928270 -1.337443,-1.590145 0.906784,0.744092 1.604471,-0.461280 2.036367,-1.575838 2.929512,-1.586794 -0.424612,-1.494488 0.141867,0.185780 0.325955,0.100195 -1.623314,-0.440875 -1.142675,1.164463 2.101083,1.569397 1.712286,-1.050138 -2.073328,-1.695318 1.593346,0.947515 -2.530952,-0.690376 -0.626627,0.845378 0.013760,1.177548 0.020401,0.167370 0.947523,-0.090241 -1.608290,-0.985831 -1.292977,1.009125 0.699305,1.700944 -0.180783,-0.423401 -0.395135,-0.109737 0.455026,0.647417 2.589806,-0.453019 -2.858405,-1.646770 -1.893770,1.064386 -3.349687,1.898704 0.641799,1.745791 0.909710,-0.463294 2.460072,-1.483154 2.593730,-1.811754 -0.259929,-1.763026 -0.416491,0.002580 2.642435,0.695147 0.606907,-1.257479 1.239907,-1.465360 1.859425,-1.565236 1.412113,-1.886277 1.333327,-1.769542 1.965361,-1.795945 1.539536,-1.642885 -2.831074,-1.776022 -0.787022,1.033763 1.102276,1.437071 -5.056630,-0.862453 0.218352,1.148857 0.688508,0.175404 2.318500,-1.033081 2.310818,-1.888282 0.228598,-2.179568 -3.064201,-0.641084 -3.128788,1.596630 2.075315,1.725818 -1.456826,-0.918732 -0.215663,0.837727 4.936122,0.665959 3.959651,-1.549687 2.149561,-2.134523 -2.663209,-2.124893 -1.029565,1.027068 2.425420,1.603190 0.761252,-1.142616 0.098039,-1.463091 -1.789540,-0.545540 -0.478013,1.505746 -1.524786,1.369693 -0.099202,1.723782 1.429132,0.564676 -1.273435,-1.259871 0.843177,0.794569 -1.221761,-0.589422 -0.821027,1.081909 1.178396,1.287872 -0.821238,-0.807325 1.550241,0.549884 0.116101,-1.204814 1.596512,-0.719116 -0.742362,-1.544889 -0.248137,0.512712 1.954319,0.640003 -1.207686,-1.339793 0.470663,0.851902 0.845838,-0.438546 -0.155047,-1.301771 -2.457699,-0.013633 0.805683,1.502527 -2.299227,-0.376090 1.030620,1.170874 1.391123,-0.963049 1.570406,-2.038300 1.528408,-2.519404 -0.122042,-2.165004 2.391501,-0.645975 -1.305128,-1.576794 -1.989452,1.094698 -1.421772,2.058373 -0.467751,2.491189 -0.010586,1.520298 3.872006,0.270136 -1.157211,-1.397623 0.191809,0.808248 3.376579,-0.112328 0.453489,-1.982101 0.279347,-1.348567 -0.764882,-0.765685 -1.157672,0.971244 0.040147,1.891902 -2.729876,0.735240 1.701657,1.733715 0.628824,-1.137246 1.261130,-1.565716 -3.163487,-1.870929 1.600144,0.927657 -1.942420,-0.862643 -0.825509,1.432137 -1.827893,1.962706 2.875507,2.219897 0.302268,-1.158079 -0.365744,-0.882671 -0.666582,0.390834 -1.573249,1.365154 3.170632,2.020487 -3.315707,-0.907529 0.913223,1.171446 0.315974,-0.825558 -2.605635,-0.518955 1.074380,1.399531 4.063355,-1.096575 -0.494333,-1.744086 -0.212902,0.110447 4.416994,0.151181 -2.017376,-1.546420 -0.268936,1.139504 0.120466,0.926787 -2.404290,0.441260 0.337721,1.523173 -0.340359,0.170029 -0.112153,0.638482 -1.438246,0.520502 -0.678715,1.805550 -2.148510,2.027264 0.600776,2.393856 0.559954,-0.352214 -1.800776,-0.636446 0.168869,1.200058 3.869258,-0.052437 -1.239498,-1.501934 2.956217,0.602894 0.850565,-1.833658 0.744058,-2.009775 1.960909,-2.151810 -1.749665,-2.074135 -1.036582,1.045928 1.376578,1.577548 1.030888,-1.089308 -0.247282,-1.816738 2.264244,-0.715906 -2.221556,-1.760292 -2.327754,1.448260 0.694778,2.260234 1.276474,-0.287295 -0.112820,-1.603016 -1.098709,-0.344314 -2.956371,1.673415 2.329564,2.342428 1.032931,-1.198693 2.600021,-1.901823 0.393655,-2.998546 -0.771466,-1.404957 -0.059869,0.584698 -0.975071,0.714131 0.375270,1.601613 -1.230985,0.139396 2.329633,1.236937 2.666667,-1.641699 -0.667845,-2.266690 -1.128914,0.267446 -3.954996,1.644003 1.865028,2.362555 -0.161151,-0.720603 -0.710313,0.307559 -2.341781,1.315359 2.898338,2.055870 0.848067,-1.256825 -0.064538,-1.607990 -1.251769,-0.614111 -0.932156,1.588104 -2.365586,2.329189 -1.770197,2.917499 3.848521,2.311823 -0.258566,-0.606768 -1.431164,0.565550 0.828435,1.198376 4.381174,-0.914516 -0.073336,-1.964048 0.052291,-1.022980 -1.412090,-0.192714 -2.156037,2.048655 -0.063214,2.445946 1.516101,1.017346 2.386283,-1.755461 -0.055931,-2.746462 2.337993,-1.319933 1.301775,-2.823936 2.040984,-2.880823 -1.307209,-2.852529 -1.717194,0.844899 0.236726,1.679425 -0.424818,0.867147 2.649983,0.686926 -0.350413,-1.600524 -2.814464,0.529610 2.811313,1.215406 -0.366436,-0.783464 -0.367067,0.677524 -3.337573,0.707887 -0.322855,2.461119 -0.294302,1.539912 -0.077951,1.446574 -0.417285,0.691665 -0.111037,1.011777 -0.434306,0.733928 -1.082232,1.374664 -3.029141,2.493682 -0.501029,3.028452 2.967427,1.719791 -1.466181,-0.603725 2.691275,0.834762 -3.252880,-1.373649 -2.730104,2.182179 -0.843182,2.526710 -4.285394,3.197920 0.199142,3.020049 -1.866877,1.340748 0.610480,2.110687 0.334982,-0.536174 -1.848832,0.024301 3.839529,0.824896 -1.576858,-0.889243 1.206350,1.097863 0.601022,-1.763650 1.286907,-1.636503 -0.187715,-2.254047 -2.791567,-0.242715 -0.106899,1.468329 0.830592,0.674429 0.365405,-0.665754 0.588110,-1.489333 0.143260,-1.748439 0.248163,-1.059712 -1.023394,-0.518452 -2.625551,1.530600 -2.053262,2.761602 3.331158,2.432764 0.976094,-0.892184 2.458383,-2.274509 3.056066,-3.707219 3.425580,-3.597919 -1.434589,-2.535773 2.571106,-0.412124 -5.721395,-1.270177 1.655917,1.150366 -2.896346,-1.013579 0.395852,1.644812 1.318462,-1.013958 0.319519,-1.530517 1.817390,-2.105077 -1.191576,-2.186596 -0.939442,0.875744 0.568270,1.018191 0.968622,-0.211744 -1.857182,-0.673748 -1.317834,1.251092 0.932795,1.463248 -0.230693,0.190766 -0.020338,0.379127 -3.536090,0.293627 -3.991823,2.571497 -1.597459,3.114053 3.706065,2.624878 1.883254,-0.866034 1.888309,-2.407617 -0.356789,-2.862271 1.227636,-1.409000 0.115440,-1.967869 -0.522515,-0.348890 -3.788269,0.795789 -1.170355,2.760773 -0.143891,2.485956 1.136963,1.150092 0.917536,-1.098719 0.505206,-2.375169 -1.591511,-1.577434 -1.477110,1.168001 -1.942449,2.853973 1.371364,2.423745 -0.708412,0.362055 -2.372605,1.681207 2.294013,1.020584 1.322330,-0.887561 -0.845211,-1.330171 -0.898328,0.168603 0.162920,1.234930 -0.386304,1.080880 -4.293064,1.738010 2.509802,1.193138 -1.291402,-0.092629 4.546783,0.865174 1.238858,-2.446165 1.007928,-2.013604 -2.316034,-2.158417 1.890506,-0.318844 -1.652708,-0.781583 2.143675,1.072843 -1.855118,-1.002721 -1.473263,2.225231 -2.294622,2.479649 0.164841,3.341585 2.525684,0.678052 3.688537,-1.735370 3.570612,-3.424189 1.446115,-3.741004 -2.118436,-2.723821 1.335922,-0.563325 -1.573532,-0.690982 2.711687,1.032949 -0.606572,-1.329917 -2.950950,1.616402 1.509595,0.845787 -1.391709,0.367619 0.603802,1.250313 -1.372941,-0.285558 -3.050384,2.555196 1.186884,1.373398 -0.666475,0.587702 2.930317,0.464560 -0.299415,-1.700458 -0.567524,0.468847 0.065225,0.292108 -0.977304,1.103572 -0.617933,2.058974 -0.302159,1.758162 1.951817,0.989890 3.906858,-1.595887 3.431463,-2.936735 -2.146794,-2.825190 -1.271223,0.430852 -0.122045,1.653013 -1.218626,2.437064 -4.512026,2.920735 -1.261500,3.025288 -0.584552,3.248904 0.821130,1.940023 -1.489846,0.751219 1.668176,0.617209 2.551904,-2.032649 -0.789741,-1.312492 -0.041513,-0.612885 -0.059254,-0.162704 -0.998555,1.292390 -1.308933,1.943062 -1.010062,2.735143 -4.293457,3.499183 -1.529011,3.546915 2.067387,2.217093 -4.974867,0.853905 0.074362,1.549241 2.222650,-0.777331 -3.237135,0.180006 3.068523,-0.080090 1.060489,-2.136348 -0.259772,-0.872524 3.561070,-1.929980 -1.683410,-1.302558 2.066230,0.122511 0.402489,-1.919748 3.141451,-1.268909 -2.308436,-1.902359 1.579300,-0.384541 -0.866711,-0.914584 1.536582,0.385502 3.686559,-2.388109 -2.280315,-0.794912 0.674182,-0.608679 3.715037,-1.760797 4.789654,-2.179240 0.881536,-3.373904 2.917023,-3.545137 -2.232788,-2.251470 0.909476,-0.675710 -0.495742,-0.805183 2.111818,0.177673 1.080525,-2.267974 -0.018546,-1.729886 -2.448080,-0.736821 6.176812,-0.122787 -1.415014,0.195368 2.195464,0.486618 -2.065494,-0.674863 0.326437,1.113551 2.322258,-1.350468 -0.626207,-0.391980 -0.534090,0.066862 3.483198,-0.536627 1.076206,-0.980542 -0.085417,-1.540969 -1.636043,-0.726152 2.129375,-0.230816 1.045580,-1.009544 0.479843,-1.456720 -2.558446,-1.325547 -1.916225,0.970713 4.899866,0.809997 3.477294,-0.531326 1.376610,-1.967319 0.400894,-3.287869 -0.286519,-1.948407 0.032981,-0.806871 -1.282368,0.576301 -1.287231,2.134723 -0.023153,1.879228 1.941038,0.743030 1.877391,-1.373873 0.762517,-2.541786 1.354879,-3.343024 -1.332409,-1.783739 1.061832,-0.875128 0.937165,-1.811569 2.617609,-1.728932 -2.046717,-1.798255 1.443737,-0.965070 0.397107,-1.620015 0.130441,-0.449567 -0.747904,-0.718773 -1.555081,1.162209 1.333404,0.764538 -2.140932,1.216549 2.072160,0.455499 -0.364824,-0.916119 -2.579313,1.409259 -1.799514,1.593083 -1.260110,3.041848 -0.902079,3.371963 3.740367,1.553960 2.436109,-0.790515 0.519608,-1.917938 0.744174,-2.937957 3.561484,-3.013657 -4.317300,-1.267918 2.017870,-1.197613 0.481931,-1.602639 0.443504,-0.836371 -1.148399,-1.009948 -0.254444,0.425836 -1.364390,1.361200 1.405150,1.177812 3.240350,-1.010349 -3.557783,-0.123359 -0.679935,0.188844 -1.684301,1.578249 0.630498,1.970104 -0.115260,0.824190 -0.879194,1.162072 -0.586993,0.901659 1.553894,0.435823 1.911735,-0.891075 -1.288826,-0.504681 1.540775,-1.468460 -0.868958,-0.478487 -0.985473,1.520659 2.383187,-0.247500 2.793696,-0.799211 8.047335,-2.064102 -0.236024,-2.618667 -1.904595,-0.549685 -1.367807,1.101049 1.739557,1.046309 0.312511,0.323396 2.765413,-1.196576 1.395637,-2.957550 2.953236,-2.977742 -0.717622,-2.051716 0.657389,-1.529723 -0.316326,-0.685810 2.451410,-0.585093 -0.132989,-1.179935 -1.106503,0.385783 -3.932829,1.099061 1.550564,0.894209 2.426518,-0.442102 -2.708559,0.159770 -2.185711,0.550226 -0.999400,1.725826 0.818663,1.691072 -0.747877,1.326079 -0.586494,1.249557 1.184880,-0.402874 -1.691171,0.996982 -1.397893,1.597351 -1.434261,1.818399 2.744254,1.309816 -2.002669,1.589373 -0.813717,1.568228 1.356608,-0.291811 -2.946041,1.426977 -3.099428,1.811974 1.125107,0.270016 3.809351,-0.199310 1.862473,-1.679653 3.833730,-3.203439 1.008187,-3.130918 -0.588141,-1.720124 2.298299,-1.839210 -0.724279,-0.486645 1.407821,-0.675346 1.200250,-1.979919 0.175335,-1.078836 -0.587894,-0.932062 -1.274582,0.509636 2.228971,-0.010212 1.292017,-0.627213 2.215544,-1.774551 3.870163,-3.266297 -0.104911,-2.163058 -3.749167,-0.365078 -6.031495,1.025769 -1.201848,2.511094 -1.860788,3.092479 1.841908,1.192262 -1.675326,1.534429 2.831933,-0.245259 0.057557,-0.933171 -0.598136,0.536758 -1.072397,0.352534 0.420011,0.544260 -1.212854,1.471519 -0.278656,1.247103 -3.513351,1.517879 -0.280632,1.766179 -1.859841,2.046461 -0.780351,2.307583 1.823640,0.381009 -2.383883,1.496247 -1.556878,1.505845 -2.745973,1.619942 0.879797,1.331689 2.727558,-0.217608 -1.569564,0.549992 -0.231676,-0.339087 -0.817995,0.719482 0.780881,0.587063 -1.272774,1.029444 -1.364271,1.494083 1.562960,-0.294140 0.478099,0.080162 -0.580483,0.307451 1.499476,-1.520521 -1.210064,0.361725 -1.560776,1.205595 -1.261051,1.276976 0.894595,0.898712 1.137684,0.017181 2.262392,-1.345770 -1.385418,-0.648998 2.864650,-1.659876 -0.481850,-0.162094 -1.120744,1.063685 -0.363465,0.413509 2.280087,-0.104799 -1.074474,1.034222 -0.776539,0.700543 1.726673,-0.798921 -1.730193,1.344566 -4.121138,1.331823 3.845895,-0.596029 -2.441254,1.780074 -0.270948,0.797113 -5.108185,1.118559 0.185152,1.100074 -2.632182,1.789864 -3.283952,1.966053 1.331461,-0.024266 0.186313,0.705218 2.971156,-0.963268 -1.472548,-0.121036 -0.486396,0.277355 1.300871,-0.773148 0.456102,-0.063348 0.104071,-0.508389 -3.787346,0.247648 -1.249420,0.924091 -3.194652,1.722455 1.706749,0.350346 -1.468119,1.412535 0.479884,-0.129481 1.451008,-1.115749 1.977610,-1.051967 -2.022824,-0.094083 1.063067,-1.431597 0.680658,-0.848610 0.709081,-0.702695 -1.139324,-0.173270 -0.873343,0.217245 0.977277,-0.443015 -0.962421,1.516134 -2.141724,1.149655 1.492476,-0.712663 2.929723,-0.450226 1.117873,-0.998817 1.014411,-2.013894 1.827742,-2.067457 1.312949,-1.872143 0.920691,-1.580396 -2.283046,-0.088336 0.224212,-0.877887 -1.504164,1.103016 -1.533148,1.516627 -1.018527,1.157157 -2.623618,1.955461 -3.538664,2.220849 2.456430,-0.201744 2.639000,-0.294543 -1.069534,0.637956 -1.804868,0.121026 0.047105,0.078480 -0.080917,1.011105 0.936466,-0.374819 1.293523,-0.977735 2.106319,-1.350048 -2.088740,0.267400 -0.324628,-0.466851 -3.053666,1.280290 -0.204677,0.764246 0.229514,0.418130 1.034636,-0.296426 1.884904,-0.990659 1.483411,-1.396524 0.544063,-1.526010 -1.380290,0.135272 -2.421601,0.544821 2.805907,-0.819728 -1.031907,1.357460 2.619113,-0.982559 -1.225789,0.787010 1.106731,-1.052340 0.276377,-0.304034 1.439490,-1.059807 -1.686102,0.557935 -2.007378,0.461694 2.436205,-0.941297 -3.502195,1.710165 2.639048,-1.156075 -1.442648,0.943072 -0.723817,0.473751 0.242400,-0.207651 0.553912,0.008508 -3.575384,1.135248 -3.752600,0.751319 -5.001261,1.125146 -0.730595,1.406555 0.491677,0.109572 3.238573,-0.462669 -1.043469,0.834617 1.720112,-1.467352 -2.595013,1.162122 2.459812,-1.361024 1.336246,-0.590061 -0.606588,0.449110 -2.739079,0.492172 -0.902917,0.573794 -1.282609,1.033410 -5.894721,1.131213 5.128271,-0.650273 -3.150481,1.547876 0.869529,-1.077281 -2.112855,1.195649 -4.675531,0.846060 0.104680,-0.120055 0.529485,-0.095194 1.169740,-0.642405 -1.520671,0.912668 2.610068,-1.455485 -1.732512,1.148442 -3.094443,0.640987 -3.015036,1.105550 0.401397,-0.185711 2.138826,-0.380353 -1.323997,1.041096 2.194076,-1.368584 -4.418928,1.177600 -0.161181,-0.167355 0.717665,-0.437684 1.529028,-0.651705 -0.808421,0.742107 3.558580,-1.342571 -2.382831,1.373364 0.545802,-1.026107 -1.349678,1.282905 -0.417240,0.129826 -0.442395,0.482391 1.224781,-0.812522 -3.355688,1.121590 0.497793,-0.934889 0.523581,-0.214751 -2.038110,0.926086 -2.153041,0.554893 -1.014309,0.639783 1.786817,-1.019957 2.019153,-0.574087 0.168196,-0.082901 -1.911663,0.857528 -0.434922,0.090900 -0.532497,0.548328 -0.164914,0.053401 -0.476630,0.623259 2.786132,-1.039894 0.750051,-0.178973 -1.737682,0.834667 -0.718034,0.127804 -1.890169,0.871268 0.406560,-0.388789 -2.577805,1.405638 1.456560,-1.327316 2.051788,-0.315620 0.225857,-0.130485 -2.612352,0.947348 -2.545031,0.707569 -0.334231,0.232176 0.369965,-0.261262 -0.766433,0.767274 -3.042026,0.640537 -0.409697,0.139953 1.565960,-0.946010 -2.217189,1.053005 0.635827,-1.023715 2.398560,-0.483889 4.108384,-0.799213 2.220386,-0.738591 -0.964976,0.897192 1.263386,-1.001514 -0.780860,1.216207 0.852611,-1.044510 1.702503,-0.357496 -2.775201,0.994413 3.252048,-1.229827 -2.205695,1.491466 -0.194506,-0.462312 0.351933,-0.070534 -1.467912,0.850368 5.557528,-1.301255 -1.033467,1.225482 0.259468,-0.897211 1.756360,-0.386823 -2.620716,1.020819 0.126139,-0.399027 0.245015,0.249017 2.820964,-1.071917 1.900477,-0.432552 4.772682,-0.626316 -3.267204,1.216486 0.415484,-0.696121 0.263805,0.298589 -0.913632,0.388523 1.286267,-1.032978 -0.243277,0.765846 -1.416060,0.386343 -1.405489,0.754295 0.663689,-0.795460 2.008618,-0.780023 0.677690,-0.459241 1.293630,-0.526726 0.443520,0.283195 -0.223249,0.470362 -5.229061,1.000494 -0.275860,-0.017980 -2.688959,0.769181 2.613964,-1.515348 -3.255433,1.234155 -1.258412,-0.170111 -4.443423,0.817324 -1.041843,0.059478 -2.901274,0.125131 1.056882,-1.267277 -1.321647,0.734641 -0.906804,-0.273473 2.095916,-0.779087 -0.897375,0.459891 -1.017081,-0.049222 -2.928477,0.992559 -1.314737,0.166059 0.128055,-0.885124 3.849934,-1.197962 -1.522114,0.809547 1.612008,-1.145914 0.742842,0.374110 1.465470,-1.069931 0.246147,0.617659 -2.901742,1.002755 -0.522932,0.365221 1.433729,-0.717904 -2.521559,0.851800 3.682774,-1.446947 -0.575300,1.214595 0.323446,-1.019731 1.599920,-0.062192 -3.135584,0.869331 -0.082358,-0.353890 -0.895285,0.851620 0.226721,-0.921019 4.596325,-0.735863 -1.034760,0.627669 2.327127,-1.144627 1.590443,0.210989 2.505576,-0.638479 -0.632782,1.237869 2.071248,-0.336231 -1.167350,1.569418 1.554459,-1.299898 1.077040,0.365273 -0.236389,0.120381 -3.819903,1.076800 0.088796,-0.017620 0.046665,-0.132133 -0.388392,-0.267715 -0.288420,0.092644 -1.038931,0.418518 -1.973386,0.105696 2.246575,-1.127805 -1.016936,0.326604 -0.625142,-0.367380 1.376362,-0.223646 -1.048984,0.510668 1.837946,-1.054838 -3.547996,1.664884 1.225269,-1.482708 1.415005,0.202795 1.505890,-0.629307 1.900390,0.333681 -2.437732,1.626271 3.028800,-0.632188 0.095916,0.951732 -2.443080,0.041895 -3.468270,0.904211 -0.595366,0.063615 -2.357193,-0.316919 -1.625099,-0.540761 0.400539,-1.004908 1.552261,-1.153143 -1.811278,0.311289 2.441700,-0.655510 -0.621938,1.200455 3.066307,-1.442255 1.345434,0.466548 0.168579,-0.084234 -1.116831,1.286470 0.284472,0.162415 0.277253,-0.026272 -1.225019,0.061673 1.817176,-0.689499 1.333626,-0.058418 1.425416,-0.671085 1.511610,0.450927 1.628587,0.474642 0.365258,0.759357 -1.091922,1.204144 -3.296496,0.961918 5.987984,-0.656659 -1.929677,0.031411 -1.167573,-0.501113 2.460441,-0.187618 0.634528,-0.705290 1.977113,-0.662029 3.077875,0.489061 1.419333,0.224146 -0.681213,1.366052 -1.731849,1.067115 0.454496,0.128273 1.244134,-0.924403 -0.410422,-0.198273 -0.207308,0.455655 0.576654,0.266795 2.393919,-0.647430 3.389290,-0.245997 -0.787106,1.078528 2.456952,0.238138 -0.508450,1.127306 1.015680,-1.024894 -0.255300,1.053173 -3.018803,0.105827 -0.362171,0.447417 -1.351658,-0.026298 -3.141361,-0.439186 -0.958398,-0.358978 1.014808,-1.322473 0.092395,-0.768688 -0.931325,-0.163199 0.226284,0.343381 1.687207,-0.390127 1.170547,-0.694843 3.041796,-0.000142 4.071561,0.509194 -2.986929,1.329700 -3.499726,1.038881 -0.302706,0.552470 2.440501,-1.392026 -0.797503,-0.397728 -0.586168,0.061822 -2.648594,0.942263 1.298661,-1.054976 -1.514053,-0.025616 -0.272162,-0.660187 -0.634477,0.772104 0.845168,-1.158912 0.079761,-0.139290 -0.454898,-0.180200 -0.082695,0.486441 -1.229802,0.239035 0.406393,-0.593040 -3.415928,0.039897 0.343241,-0.878053 3.068513,-0.135676 -1.415170,-0.377782 -0.376727,0.365353 3.536836,0.290884 -0.218775,-0.460265 1.253250,-0.534915 -2.299143,1.142517 3.372707,-0.300598 0.582318,0.367118 -0.116303,-0.555146 0.786715,0.779099 -2.069172,0.664454 1.423870,-0.205673 1.344475,0.145226 1.000042,-0.815728 -2.938705,1.232371 -0.047139,0.770656 2.646428,0.190549 -1.412993,-0.653795 -5.785129,0.334591 -2.601547,0.859119 -2.066938,-0.716073 -0.757218,-0.845126 0.839456,-1.132294 4.035134,-1.161916 -1.324161,-0.262431 -1.033907,0.808942 -1.130592,0.943852 -1.362567,-0.657566 -2.267646,-0.496830 -0.447654,-0.707281 2.471535,-1.144686 -1.969216,-0.884511 2.938920,-0.183550 -1.138654,0.615713 0.287335,-0.663147 -0.430728,0.756806 -0.596454,-0.455920 0.673803,0.079981 -2.596447,-0.175446 0.568621,-0.396846 1.303074,0.147404 1.860449,-0.957552 -1.107988,0.680863 -1.927826,1.173299 -1.758459,0.831814 1.919435,-0.975514 1.263292,-1.226072 2.167161,-0.932920 0.901931,0.729063 2.375369,0.667309 -2.068956,0.895613 -0.211009,0.946781 2.038399,0.559571 0.735207,-1.115905 -0.576765,-0.010217 -0.723839,1.140065 -1.628144,0.609042 -2.238834,-0.278237 0.316331,-0.524063 0.365705,-1.049574 0.345170,-0.857463 -2.574159,0.277808 3.005059,0.635930 -0.094500,-0.065008 1.502557,-0.613562 -0.751038,0.701479 -1.153525,0.337313 -1.855656,0.894681 -2.248247,-0.329235 -3.592446,-0.512118 0.145371,-0.761617 -0.764024,-0.995375 -0.337662,-0.642471 -1.020655,0.118931 -0.155332,-0.495074 2.485900,-0.391000 0.947101,-1.010349 -2.807269,-0.137167 -0.453865,1.174805 3.907759,0.483135 1.789211,-1.321513 -2.346684,-0.414668 0.011310,1.328705 2.519205,0.466414 0.148014,-1.212247 -0.329331,0.231768 -0.751078,0.933614 0.764810,0.200394 0.811513,-0.551720 2.893643,-0.648702 0.158616,0.211051 -5.538172,0.644169 0.896690,1.057032 3.465165,-0.352204 -1.951433,-0.836327 -0.968927,1.127101 1.471547,0.684410 -0.511965,-1.043083 -1.513656,-0.250777 0.291816,0.802923 0.648668,-0.619951 2.172691,-0.887799 -0.078511,-0.284431 2.491917,0.735124 0.036956,0.348838 0.466344,0.124684 -0.137261,0.513868 -0.524424,0.135801 -0.635728,0.504876 -1.728824,-0.071658 -2.298902,-0.175436 -2.395436,-0.325013 -1.047052,-0.392172 0.652079,-0.421165 -0.616614,-1.166790 -0.384401,-0.152725 0.224604,0.576663 1.697833,-0.474020 3.215711,-0.671430 0.703241,0.077390 0.881493,0.506652 1.659945,0.592974 -1.767687,-0.097557 -1.012510,1.052370 1.338105,1.079921 1.535140,-1.183168 3.906392,-0.793600 -2.405321,0.013272 -2.117449,1.228451 -1.481439,1.067613 0.296525,-0.162338 2.559243,-0.867375 -3.342882,-1.303502 0.460781,1.030047 0.442164,0.208641 0.018591,-0.774745 0.066739,0.375093 -0.700983,0.142496 -1.637000,0.129731 -0.280375,0.338028 -0.669761,-0.517299 -2.443504,-0.474817 -1.672827,0.038953 0.861616,0.030608 1.992178,-1.005266 1.877561,-0.847563 -1.416026,-0.384739 -5.010341,1.283920 1.586224,1.457940 -2.191342,-1.684856 1.737509,0.548764 0.872812,-0.241681 0.766534,-0.786958 -4.015816,-0.132353 3.622817,1.359169 2.335008,-0.120696 2.561575,-0.916376 0.144735,0.314943 -0.669423,0.487589 -0.568350,0.812923 -0.815055,0.210401 -0.287345,0.132241 1.055974,-0.028180 1.838845,-0.875165 -3.912982,-1.300366 -0.302723,1.395795 -4.684968,0.338253 2.271533,0.393376 2.106056,-0.909413 -1.008532,-1.266154 1.681773,1.706110 0.921737,-0.288848 4.049887,-0.203419 4.043007,-0.255728 -1.587497,-0.270337 0.751844,1.666199 -1.453522,-0.534743 3.186023,0.726403 1.913307,-0.422164 4.468269,-0.824644 1.752756,0.200525 1.714955,0.222688 -1.707716,-0.189765 -2.276039,1.084250 1.742450,1.463098 2.450481,-1.114708 -1.172673,-1.352225 -2.995139,1.152150 1.209367,1.766304 -2.432229,-1.595448 -2.488965,0.142823 -2.065794,1.075165 -1.209629,0.407779 2.310441,0.817687 3.381768,-1.219605 -2.599945,-1.113634 0.994360,1.637367 -1.509760,-0.807284 0.925744,0.983810 2.941361,-0.152795 -1.651485,-2.028333 -1.685668,0.893466 -0.726819,1.137471 -0.081319,0.383110 1.211346,0.044857 -4.060909,-1.456182 -2.679036,0.559588 1.110835,1.338165 -0.840365,-1.489684 0.931962,0.906667 -0.906598,-1.019326 -2.490894,0.298532 -4.086077,0.784105 1.252553,0.951239 0.075971,-1.442352 -2.188764,-0.866601 2.148244,1.544550 -0.114904,-1.192132 1.609265,0.576846 1.511393,-0.404764 -1.533999,-1.403643 1.327299,1.703336 -2.589220,-1.369977 -2.658970,0.661439 -0.322417,0.686822 -3.900300,-0.535577 2.717421,1.243480 -1.699900,-1.926545 -2.445202,0.392575 2.170122,1.752951 1.773561,-0.960193 -1.856105,-1.548463 -1.239052,1.190630 -2.536156,0.553435 2.657304,1.031560 -1.267955,-1.817273 -2.424103,0.484669 -1.083036,0.841051 -0.133751,0.421344 -0.566530,-0.452186 1.208882,0.836905 -2.358650,-1.606599 -0.766699,1.049370 -0.207483,0.614475 3.190929,0.477165 3.536222,-0.766861 -2.150594,-1.676587 2.256632,1.965343 1.122499,-0.848038 -2.519098,-1.753318 -0.476358,1.234660 -0.234863,0.427860 -0.804477,-0.181239 -2.192874,0.248111 -0.078077,1.395490 4.604850,0.426109 1.204016,-1.176937 -0.650897,-1.211248 -0.283950,0.764598 -1.972409,0.008152 2.349928,1.712551 -2.425482,-2.193821 -4.146297,0.914491 1.078141,1.538146 1.150634,-0.691574 1.329172,-0.188345 3.124889,-0.942576 0.733206,-1.231384 -0.159668,-0.858983 1.516707,0.710344 -0.750405,-1.725363 2.502493,1.642883 4.519434,-0.865023 -0.609850,-1.966915 0.621042,1.132349 -0.784783,-1.080948 -0.784390,0.697345 -1.048156,0.393234 0.088230,1.369022 0.410379,0.114835 -1.856165,-1.129374 2.281162,2.033115 0.407001,-1.445237 -2.375806,-1.135671 -1.236278,0.555797 2.963033,2.176444 -1.587403,-2.099150 0.613300,1.822479 1.839744,-0.420296 1.226270,-1.226027 1.354030,-1.036753 -5.278538,-2.314362 -0.171571,1.203921 -2.508544,-0.813411 -3.573384,1.120753 -2.646075,0.708810 -0.449086,0.904637 2.340802,1.592975 0.034126,-1.345117 0.790465,0.383119 0.961887,-0.900672 -0.252187,-1.046290 4.536668,0.857245 -3.318401,-2.452030 -2.397721,1.123263 -0.742836,0.220904 4.197260,1.903904 -0.639852,-1.532553 -0.714319,0.650864 -1.908739,-0.303061 -1.438481,0.984375 -2.289307,1.114013 -1.936202,1.145640 0.728261,1.798594 -0.692920,-1.143352 -1.471433,0.661120 -1.298442,0.767178 0.021228,1.510304 -2.659112,-0.448014 -1.501977,1.470964 0.077323,1.479820 -1.148647,-0.427695 0.485191,1.968657 -2.771580,-1.676809 0.921434,2.365730 -2.264635,-2.187614 -0.327062,1.691372 1.298509,0.793706 2.284797,-0.498504 0.758589,-1.138504 1.747274,-1.162341 -0.626686,-2.400953 -1.092901,0.133654 1.219975,1.465025 -0.576323,-1.412516 -2.548358,0.627419 0.300316,1.473186 0.129065,-0.300772 -3.162420,-0.659182 -2.786883,0.926306 1.499321,2.893421 -1.730377,-1.886718 3.244627,2.819527 2.492491,-1.524037 1.802760,-0.411075 -0.472249,-2.774940 -1.268674,-0.467520 -0.182886,0.509134 0.137614,0.764568 1.327016,0.788906 -0.731624,-1.901498 -1.255910,0.398851 -0.554919,1.136141 2.291540,1.892751 0.671495,-0.995299 2.337769,-0.056794 -1.040618,-2.900459 -3.293064,0.275462 -2.320416,0.817952 -2.360199,1.251749 -1.055545,2.199458 -0.353831,2.093544 -3.000043,0.350605 -1.802251,1.723032 -1.914375,0.784078 2.186874,3.496379 -2.064174,-1.904951 0.167908,2.525936 2.914462,0.181502 -0.522358,-1.986467 1.184529,1.048398 0.305361,-1.772081 1.593289,0.570045 2.495870,-1.718120 3.034788,-1.719411 -0.539244,-3.097739 2.033031,0.803200 -0.370501,-2.704259 -1.053934,0.247739 -2.445300,0.292571 -0.754487,2.013503 0.478966,2.430952 -1.107676,-0.698737 1.744402,2.461842 0.517854,-1.452303 -2.827218,-1.257261 1.866462,1.827831 -2.205041,-3.087332 -0.793643,2.065737 -1.342994,0.090826 1.177734,3.340626 2.391792,-0.204353 -1.062274,-2.213435 -2.225989,-0.057852 -0.458422,0.363497 -1.368833,0.986647 -1.564281,1.532523 1.405972,3.062509 2.498024,0.194375 -0.035718,-1.194924 -2.646184,-1.623096 -1.303435,0.094107 -0.900692,1.209719 2.624992,3.308126 -0.426282,-1.355197 2.056352,2.139387 2.117838,-1.692300 3.054048,-1.001478 2.549840,-1.853410 0.453366,-2.607275 0.771101,-1.722558 -3.268311,-3.057988 -0.343699,1.020514 2.259158,0.932197 0.026157,-0.787053 0.510364,0.715534 2.514216,-1.006436 0.276327,-2.142376 0.565637,-0.405878 -1.283443,-2.712319 -1.538496,0.377186 1.714358,2.035136 2.968290,0.041316 -2.146750,-2.324499 -0.466767,0.371296 -1.182848,-0.777381 1.733859,3.335231 -3.163517,-2.119757 -0.378403,2.174744 -0.891522,-0.088145 1.573547,3.292542 -3.013563,-2.031287 -1.936194,1.621891 -3.070462,1.016751 -0.644817,2.995405 -0.732118,1.871391 -4.739334,1.853361 0.680367,3.364141 -1.090231,-0.711061 -0.505266,2.155763 -1.333306,-0.209162 1.750040,3.352005 -5.204725,-2.159259 0.375710,3.106859 -0.637739,-1.683136 -0.011050,1.949783 1.632011,0.551256 0.334018,-1.077963 -0.342544,-1.197105 -0.293831,-0.817150 0.745289,0.650760 -1.857947,-1.627190 4.162507,2.808057 -1.196411,-3.464994 2.154504,2.988287 0.441662,-2.640115 -2.257189,-0.869023 -4.708811,0.039955 0.729476,2.587865 1.034681,1.027297 -1.198229,-0.982606 3.663344,1.443963 -3.148242,-3.472823 4.096746,2.995462 1.360065,-2.051655 -0.949060,-1.695593 -1.340733,-0.776810 2.511661,1.524443 0.800571,-0.292219 0.145007,-0.870674 -0.577500,-2.097775 -2.151890,-0.672523 -1.267599,1.456058 -3.160278,2.196314 1.903163,4.012601 -0.153958,-0.729196 -3.150630,0.378215 -2.266388,0.235908 -0.792209,1.521280 -2.632329,1.727021 -2.001361,2.332603 -1.678419,2.425693 -1.078759,2.468885 -1.718025,2.400685 2.950135,4.072440 2.195281,0.404058 1.260790,-0.357348 0.186996,-3.299427 2.036514,-1.196724 2.015799,-2.275781 -0.831244,-2.839452 0.726009,0.274150 -0.502099,-2.361640 -2.091196,0.839179 3.416606,1.867635 0.012516,-0.979317 -1.816615,0.132621 -0.364594,0.041662 -3.557415,-0.094708 -2.392847,2.202543 -0.606996,2.535824 2.549802,3.614549 1.173351,0.264442 -1.149149,-2.213222 -3.893843,-1.007178 0.613941,1.174370 1.823988,1.219895 0.926136,0.171869 -2.298860,-3.367790 -2.952578,-0.768693 1.010785,1.291001 -2.242564,-0.263856 0.723703,3.078339 -1.573666,-3.172461 -0.699935,2.310303 -2.799673,-0.300478 0.081367,3.388652 -0.384149,0.853648 -0.898246,1.525179 0.563591,1.720351 1.876825,0.718600 -0.050718,-0.850072 -1.778673,-1.818282 -1.828096,-0.122267 -1.767443,0.882362 2.286674,4.000639 1.240318,0.852332 1.559607,0.783322 3.732416,-2.687657 -2.756336,-4.036171 -0.483050,-0.126548 1.658677,0.137055 1.262088,0.457059 -2.488855,-2.135165 3.099560,0.894156 -1.739633,-3.423746 1.572944,3.447729 -3.882858,-3.384396 0.258615,3.176111 -0.414885,-1.998559 2.027427,3.049780 0.560234,-1.468663 1.083215,0.409763 4.356788,-2.136010 1.307548,-2.253456 2.254152,-2.458618 0.724179,-3.051577 1.757154,-1.712501 -0.411947,-3.634978 0.301121,-0.380107 -1.807280,-2.431927 -5.043286,1.227715 -0.629114,0.878713 0.227434,2.996522 -0.867500,0.818643 1.755941,2.872801 -1.134949,-2.129787 -2.419402,1.127924 1.174104,1.254673 -0.433806,-0.089621 0.512432,1.868382 0.570188,-1.559517 1.378916,0.131809 -1.964140,-3.141735 0.535839,0.978968 2.367532,-0.730996 0.987795,0.525285 -1.077417,-2.963033 3.257332,0.225437 2.739685,-1.543216 -2.872589,-1.539717 1.334655,0.250817 2.274677,-1.616064 -0.618668,-1.110990 -2.703284,-1.660710 1.804665,1.104085 1.850289,0.483423 0.403906,0.178107 -0.542366,-2.841388 -0.244498,-1.161605 2.096163,0.700546 2.145677,0.218962 -1.171693,-2.046400 -0.526728,-0.988205 1.962057,0.179228 0.040823,0.037454 -0.833216,-0.249112 2.095422,0.345867 3.143586,-0.481688 1.316860,0.087979 1.996311,-2.604097 2.119434,-2.696017 1.639517,-2.550038 0.540879,-2.871191 1.587540,-2.023814 -1.492039,-3.801949 -2.396633,-0.647555 -4.571291,-0.033375 -0.672931,2.945535 -2.636295,2.211945 -3.530368,2.611670 -2.410993,2.109002 -0.416460,3.221934 -0.857987,2.152868 3.297774,3.482598 -1.022925,-1.193607 -2.211174,1.031321 -2.228080,-0.240975 1.379727,3.739343 -0.777122,0.192244 2.538997,2.273731 -4.099045,-3.094315 1.598370,2.701833 0.751093,-1.964280 3.176846,1.289287 -1.143751,-3.617177 0.130314,-0.058223 -0.561421,-1.414810 2.011505,2.834889 0.559385,-0.891224 -3.430370,-1.175332 2.304349,0.330689 0.600094,-1.293908 -1.063495,0.226879 -4.703658,-1.486865 1.199890,1.448593 -2.501445,-0.318214 1.822003,2.954566 5.965899,-1.248593 1.604691,0.666321 -0.186952,-3.317575 1.298942,-1.238757 -1.903437,-3.306818 -0.156609,1.064934 1.934258,-0.184235 -2.187074,-0.366464 1.694964,1.840713 0.214277,-1.715241 -1.870235,0.178400 2.505313,0.375116 -1.499780,-1.900987 -0.500998,2.081656 1.102305,0.109807 0.285717,0.791933 0.900769,0.717213 -1.665926,-2.949472 0.729656,0.691450 0.226434,-0.971350 -0.201607,0.817310 -4.761585,-1.369671 0.317394,1.150658 -1.375502,-0.186033 2.335276,3.390672 -0.698465,-1.849226 -1.355519,1.129030 1.953684,0.485631 -1.033752,-0.376294 1.326048,2.701282 -0.833344,-2.409095 -0.845899,0.937005 -0.163479,-0.525979 -0.815159,1.053397 0.697766,2.204914 -2.216014,-0.761762 0.228800,1.852242 1.416661,-0.119770 1.605827,0.958008 -3.587305,-2.328645 0.775214,-0.371919 -2.868954,-2.564625 2.420720,2.863551 -0.508168,-1.676726 -2.777774,1.141501 1.939219,0.523016 1.975733,0.816565 3.012082,1.034370 4.705097,-2.276693 -3.647192,-3.461040 -0.432806,-1.211171 -3.512468,-0.913104 0.798455,3.002162 1.695664,1.060571 -0.068137,0.014751 -2.471861,-2.019448 1.020526,0.230245 3.766524,0.423640 0.568588,0.515543 -2.387726,-2.807424 5.166125,-0.130708 -1.336696,-2.129193 2.488628,2.806586 -0.924034,-2.185889 2.517409,1.956722 0.877484,-1.630720 -1.118263,-0.641382 2.173878,-0.368292 0.852553,-0.985187 0.610558,0.575413 -0.813977,-2.353477 1.417153,-0.077135 1.086839,-0.686082 1.324834,0.386970 -1.147006,-2.837836 -0.787124,-1.247164 -3.559055,-0.929727 -0.294121,2.047128 -0.386806,1.372427 -3.087573,1.263859 -0.313152,1.157107 2.812711,1.821962 1.897613,1.260110 0.345559,-0.627032 -0.133006,-2.007990 0.942195,-0.952679 0.250161,-0.728716 -2.612032,-0.969069 0.909849,0.133175 0.689115,-0.227558 -1.455824,0.028518 -0.381534,-0.779018 -0.929107,-0.468614 -0.332504,1.844806 -0.631988,0.663451 -1.005289,1.088703 -1.800710,0.608211 -0.124894,1.474247 -0.396675,1.294306 -1.321095,1.025469 -3.260422,0.366717 2.962832,1.984349 1.399583,1.457073 0.464479,0.534380 -2.622314,-2.726827 0.364119,-0.461024 0.971570,0.217267 0.861421,0.938238 0.103731,-1.079263 -1.569149,-1.930902 -0.204245,-0.523242 0.156288,0.492340 2.463436,1.622218 2.874991,0.276599 -0.238920,-1.283777 2.496480,-0.908473 2.077733,-0.991708 -3.149396,-1.198264 -1.780683,-1.565926 -0.428606,-0.532774 1.278772,2.489976 0.612721,1.238830 -0.871313,-0.503058 -3.585597,-1.519769 0.040109,0.196039 0.025097,1.257004 -2.224376,0.530651 -0.873188,-0.022248 1.716388,1.190440 -3.521926,0.404002 1.754678,1.377713 1.771598,-0.858427 2.287572,0.994080 0.682690,-1.883071 -2.494198,-2.175645 2.007963,-0.473020 0.995509,-0.382797 2.361426,1.068062 3.335391,-1.221148 -0.260973,-1.700916 -1.364081,-1.914845 -2.280319,-1.259250 -1.672463,0.445613 1.833124,2.615836 -1.891702,0.631805 1.295114,1.502854 -1.764585,-1.326255 -0.352592,1.292568 0.496509,-0.313474 0.958984,1.548548 -0.037237,-0.072789 1.428974,-0.289172 0.637228,-0.916149 -3.908939,-1.033957 0.799448,-0.531690 -0.111655,-0.269774 -1.110862,0.365350 2.197186,0.057554 -2.331982,-0.241370 -0.500458,0.330986 -0.635897,-0.960692 -1.471375,1.095151 3.305334,1.248151 -1.117010,0.241889 1.401825,1.201467 0.810863,-0.811576 1.041724,0.882119 -0.560927,-1.744018 -0.563506,-1.169204 -3.809500,-1.117360 -1.009652,0.554896 -2.448697,0.619806 -3.612630,0.897842 0.488324,1.836392 3.209796,1.789722 1.492176,1.234554 -3.713106,-1.523922 1.656669,-0.610430 -0.528561,-0.656420 -2.145854,0.196354 1.256883,-0.164124 -0.203976,0.557803 3.334462,1.071189 -3.407431,-1.810358 1.293544,0.670967 0.463253,-0.503193 -0.258237,0.450494 2.331969,-0.516058 0.331702,-0.137401 2.624520,0.415127 0.224797,-0.724490 -2.705216,-1.077548 1.039392,-0.398722 -0.155901,-0.073996 -0.866589,0.381366 2.029295,-0.277193 -0.442117,0.256255 1.814395,0.817300 0.511005,-0.718063 0.584831,0.371048 1.525307,-0.446941 0.421050,-0.127246 -0.318613,-0.840485 4.457926,-0.384109 -0.500785,-0.403863 0.067030,0.134697 -1.770900,-1.283320 0.732288,0.440586 -1.863321,-0.278112 1.398872,1.201524 -1.596429,-0.904220 0.364825,0.716116 0.287303,-0.213248 -3.360795,0.214137 1.637659,0.189013 -1.572329,-0.189694 -1.941289,0.191425 -0.013367,-0.650878 1.211111,1.508483 -4.240214,0.333591 0.189957,-0.455009 -1.302433,-0.932285 -0.946008,0.327638 0.986983,0.215034 -0.747128,0.550675 -0.691067,0.169582 0.896514,-0.186603 -1.309504,0.337810 2.068207,0.875275 -4.422080,-0.563748 -0.279169,-0.121809 0.744190,-0.196034 1.114111,1.239500 0.170343,0.407020 0.857523,-0.324946 -0.886138,-0.794150 1.473671,0.156014 -1.297447,-0.746357 3.693720,0.835779 3.589012,0.024079 -2.905271,0.038896 0.317898,-0.899342 1.166338,-0.159661 -1.657433,0.414055 0.226626,-0.560871 1.795002,0.038208 1.873603,1.130473 -1.435312,-0.412450 0.362273,-0.901395 0.870187,0.065934 3.965819,1.114414 0.963342,-0.061669 0.195740,-0.475664 -0.594391,-0.849611 0.969532,-0.243473 -2.767718,-0.182393 0.918170,0.286730 0.427367,0.210331 3.090686,1.002244 3.511423,-0.053322 0.659269,0.100109 1.696668,-0.659666 -3.437656,-1.418991 2.997727,-0.438244 0.404141,0.199921 -2.935344,0.050741 -0.756293,-1.133849 0.450663,0.072603 0.691187,1.152512 0.817087,0.827537 0.686088,0.165648 -0.762023,-0.430823 -4.197366,-1.194037 1.355112,-0.029319 0.038036,0.952446 2.286260,0.862565 0.860333,0.168481 -4.247077,-0.336723 -0.540544,-1.282958 -0.760495,-0.629013 0.591026,0.597985 4.569737,1.033161 -2.298382,0.135113 1.109771,-0.527303 -1.712106,-0.605483 -0.287794,-0.237828 -0.097858,-0.306532 -1.919061,0.141552 2.320450,0.352465 -0.471046,0.227020 0.097074,0.221092 -0.453401,-0.530710 -4.183758,-0.397975 2.913864,-0.004337 -3.081998,0.406999 -2.740085,-0.569782 -0.714765,-1.170415 3.944524,0.915551 -2.717942,0.397083 -0.065750,-0.277009 -3.047231,-0.751906 -1.365715,-0.549168 -3.105731,-0.465219 1.449091,0.142590 0.177595,1.030092 3.757342,0.806118 1.017175,0.747280 0.061103,0.380803 0.300260,-0.301823 2.632921,0.263674 3.229774,0.539497 -2.177963,-0.370990 1.050284,-0.516234 -1.342648,-0.178220 2.929718,0.178410 2.033369,0.685626 -1.325123,0.238462 -2.626174,-1.176909 1.852083,-0.096764 -2.472534,0.362069 -0.070645,-0.269526 2.825616,0.173720 -4.635533,0.565412 -1.037576,-1.025742 -1.888375,-1.053698 1.974391,0.451213 1.622498,1.386293 -0.272490,0.648511 0.395840,-0.549925 -1.952997,-0.496169 0.135214,-0.311114 5.714619,0.479519 0.367554,0.873320 -3.557656,-0.551953 0.296158,-0.664123 -1.915698,-0.222241 -1.767090,-0.461885 1.189658,-0.068120 0.470986,0.916389 0.246188,0.459557 -0.367295,-0.413366 2.041841,0.100737 -0.425005,0.297596 0.368676,-0.121766 -1.042821,-0.241802 1.128458,0.043089 2.643770,0.861693 -1.086157,0.036986 0.856301,-0.515150 -1.352501,-0.001861 -4.306221,-0.567146 -0.943087,-1.092911 0.052740,-0.397190 -1.533681,-0.161398 1.084925,-0.144017 0.498156,0.715277 -1.223858,-0.079259 0.557207,-0.401200 2.903996,0.734368 -0.045666,0.689204 -0.760349,-0.292014 1.699117,-0.063791 0.020476,0.377744 0.577502,0.143482 -1.081157,-0.158468 -0.452929,-0.477364 0.338098,-0.295963 0.636209,0.620261 1.750278,0.649486 1.299491,1.168817 -0.255171,0.261712 -0.457156,-0.531298 -2.369518,-0.685088 -1.493572,-1.015742 -4.993868,-0.798357 0.904997,-0.417102 -0.358608,0.435266 -0.936122,-0.391560 3.728601,-0.127286 1.074373,1.147692 -1.753331,-0.060698 2.529228,-0.480300 -2.085222,0.019667 -2.177514,-0.877055 -1.574252,-0.821234 -0.365149,-0.623682 -1.807875,-0.507813 0.420828,-0.289279 2.504395,0.661033 2.960981,1.155026 1.002691,0.713222 -0.249480,0.141685 3.113155,0.184988 -0.653124,0.505422 2.948019,-0.056774 3.117629,1.034265 -1.652271,-0.038276 -1.300601,-0.967919 -0.271403,-0.531992 -1.738939,-0.252343 2.314676,0.183074 -3.282835,0.248602 2.032002,-0.227380 0.399544,0.856641 2.004007,0.513630 0.783666,1.166203 -2.160271,0.161179 -1.664953,-1.170096 1.692886,0.043568 -0.955974,0.513041 -1.335979,-0.854703 0.299707,-0.454175 -1.332845,-0.015653 4.375240,-0.168719 -0.104475,0.708088 -0.876321,-0.322415 -0.582759,-0.598141 -1.723723,-0.844270 -2.016283,-0.932887 4.061390,-0.157775 -0.467095,0.574159 2.465708,0.005003 -2.745159,0.378091 -1.734554,-1.047718 -0.750823,-0.616569 -1.092089,-0.488365 -0.742633,-0.875945 1.303272,0.013017 4.690060,0.920866 -0.631174,0.602749 1.310526,-0.042319 0.607190,0.794544 -2.141364,-0.192227 2.387074,-0.279831 0.127648,0.693440 0.301844,0.240261 1.364708,0.514578 -0.361593,0.701788 0.359845,-0.093941 0.580594,0.431407 1.050345,0.799154 -0.687938,0.129893 -0.373148,-0.499665 -2.963645,-0.636152 -0.499748,-0.702756 1.721727,-0.088821 -2.887022,0.195380 0.169970,-0.684894 0.924622,0.461404 -0.512356,0.294607 -3.905546,-0.731378 0.987417,-0.404947 -1.151851,0.161840 -2.457680,-1.125080 -0.886954,-0.790361 -2.564441,-0.614400 0.317006,-0.624933 -2.517160,-0.298611 -1.026818,-1.130622 1.259994,-0.321110 -0.307135,0.371392 3.479335,0.178564 4.714132,0.929000 -1.182490,0.080390 -0.417269,-0.572041 -1.129818,-0.507446 1.374693,-0.194570 -0.508828,0.314880 -1.029367,-0.617952 3.543732,-0.224184 -0.319327,0.456543 -0.629919,-0.471112 -1.641074,-0.646603 -0.139647,-0.762944 1.031881,0.155423 -0.536764,0.286026 0.054839,-0.317541 -0.129478,0.026970 -1.051534,-0.433415 -4.881017,-0.784712 3.290793,-0.169835 0.637433,0.809220 -0.359375,0.193600 -1.393668,-0.453985 -0.342315,-0.560712 2.551169,0.165893 -1.925246,0.153549 -2.435743,-0.979533 -0.855406,-0.721001 0.243693,-0.629520 -2.272052,-0.337254 -1.773132,-1.196859 0.361707,-0.125039 -2.985497,-0.177182 -3.930865,-0.962082 -3.204857,-0.854834 -2.336920,-0.913657 0.841377,-0.553512 -3.725438,-0.005031 -3.661349,-1.035383 -0.462693,-0.934522 -2.531645,-0.804920 0.024318,-0.532795 0.718185,0.290793 -1.519653,-0.129925 -0.347483,-0.630901 2.006133,0.301775 -1.431250,0.141439 -1.382562,-0.945501 0.538828,-0.129591 1.443878,0.586320 1.898367,0.993910 -3.544209,0.244616 -0.844703,-0.795483 -1.892719,-0.847691 -1.569707,-1.009323 1.405504,-0.189893 -0.874992,0.082062 -1.105675,-0.792239 -5.749460,-0.734858 2.447161,-0.454361 -1.527528,0.187872 0.923444,-0.150615 3.080548,0.790058 0.887154,0.725636 0.885055,0.702592 -1.113542,-0.049484 1.180427,0.053902 -0.147522,0.319532 -1.753647,-0.448207 1.649161,-0.063987 -2.359767,0.083880 1.064986,-0.171705 0.497537,0.586168 -1.869685,-0.163089 0.041760,-0.368054 3.615611,0.347999 -2.844527,0.059587 -2.411368,-0.756569 1.431830,0.001054 -0.420065,0.204946 -0.089499,-0.212086 -3.960223,-0.425944 0.551295,-0.307234 -0.261195,0.134048 -2.029709,-0.511867 1.515067,-0.065101 1.180309,0.672654 -1.464138,0.107838 -0.391558,-0.540486 -1.973591,-0.540115 -1.336593,-0.709986 -1.516120,-0.712970 0.127551,-0.323429 -1.884816,-0.361545 0.986036,-0.078529 -2.500426,-0.153314 2.258255,0.009861 -0.216434,0.365277 5.534132,0.353016 0.585327,0.681932 -2.703264,-0.063382 -0.693047,-0.751468 -2.042898,-0.669901 -1.136929,-0.759266 -0.260941,-0.457792 -0.765622,-0.422009 -1.249860,-0.462725 0.021059,-0.377805 -2.909363,-0.451146 -3.685752,-0.946202 -0.881088,-0.865879 1.928148,0.248323 0.780880,0.756624 2.616416,0.772770 0.166559,0.513158 -0.363307,-0.041020 2.932739,0.198185 0.696638,0.688992 3.340033,0.582757 -0.885707,0.089572 -0.457334,-0.488416 -0.318881,-0.409555 2.160896,0.373989 -2.688293,0.114839 1.833129,0.098602 -0.050547,0.239843 3.089365,0.465582 -2.677048,-0.113055 -2.861690,-0.806623 -0.149419,-0.417361 0.559056,0.159524 -2.329625,-0.111251 1.426862,0.035704 2.031278,0.563098 0.868009,0.874121 -1.872759,-0.264735 2.279797,0.037982 -0.209209,0.173191 -1.292347,-0.314995 0.114149,-0.479686 -3.374956,-0.364947 1.097950,0.062505 -1.158936,-0.134516 2.514572,0.339500 1.660103,0.568014 0.619207,0.758707 -2.677333,-0.265510 2.372021,-0.030362 2.607403,0.691274 3.101381,1.035279 1.297825,0.504813 2.763592,0.377618 -0.682596,0.039611 0.923994,-0.051789 0.250099,0.153061 -2.967233,-0.242695 4.250164,-0.003235 1.069010,0.564795 -0.308246,0.336598 -1.657041,-0.649332 -0.304703,-0.553862 1.629582,0.275621 2.566075,0.777000 -1.475048,0.037418 0.370859,-0.224910 1.420457,0.239229 -1.373709,0.012152 -0.088268,-0.350637 -2.410430,-0.654315 0.559055,0.233662 -1.151369,-0.210084 -2.497775,-0.379825 -2.106293,-0.843835 -0.422251,-0.252437 -0.349490,-0.030372 -1.480318,-0.229529 -3.093119,-0.530124 -0.684175,-0.359420 0.437303,0.290932 0.707662,0.562463 0.559962,0.542513 -3.332434,-0.335238 0.087865,-0.356475 5.851703,0.272306 1.899792,0.846805 1.017649,0.493183 4.282881,0.334811 1.653102,0.308891 -0.818719,-0.411820 0.238617,-0.356179 -2.611207,-0.663503 2.734337,0.347539 0.022302,0.178844 1.961557,0.734760 5.203186,0.248600 0.923605,0.534705 -1.405351,-0.424261 0.955470,-0.145851 0.526897,0.061790 -2.038638,0.003010 0.898861,-0.145565 5.544540,0.341063 -0.674232,0.382773 -2.326976,-0.713547 -1.943002,-0.933599 1.280744,0.327681 1.886980,0.844677 -3.604396,0.090248 1.721407,-0.152107 -1.068925,-0.556239 1.097287,0.616320 2.144056,0.104438 1.015844,0.804140 2.158030,0.218248 0.311289,-0.123261 -2.212408,-0.664772 0.420642,-0.380696 -0.595251,-0.329111 -1.520253,0.030414 0.318335,-0.189869 1.512362,0.611092 -1.900470,0.184433 -0.182175,-0.235459 -1.364542,-0.794710 1.025003,0.579999 2.242268,0.505194 -1.623839,0.040283 2.066109,0.020313 -1.237901,-0.685713 3.915025,0.736515 -0.004211,-0.424853 2.782362,0.795104 0.062080,-0.390576 -1.376562,-0.370326 -0.212564,-0.578723 -0.585047,-0.243957 -3.868038,0.048750 -2.337645,-0.343464 0.466803,0.401352 2.887914,0.909883 0.931618,0.857807 1.543356,0.161675 3.794625,-0.138455 2.949827,0.001239 -1.789952,-0.850329 0.542479,-0.422901 1.897884,-0.116281 1.416337,0.815708 -0.304326,-0.405198 -1.512720,-0.915326 0.539140,-0.257819 1.837194,0.536330 0.474741,0.643261 -2.011510,-0.681222 3.734987,-0.158101 -0.288847,-0.331949 1.402042,0.845167 2.041646,-0.258606 1.777976,0.476726 -2.412107,-0.970241 1.392368,-0.123851 -2.828241,-0.871450 3.158770,0.927336 -1.595780,-0.889357 2.155623,0.993343 -1.315008,-1.029426 -0.120427,0.555985 1.249352,-0.222602 3.068363,0.903242 -1.479944,-0.385446 -0.209948,-0.529731 -2.136014,-0.982725 -0.993242,0.359232 -0.431721,0.108200 -0.791506,0.484448 2.408071,0.947872 -0.898835,-0.056491 -0.865645,0.192762 -0.747132,-0.760211 0.798265,0.847350 -1.296852,0.110263 -1.189088,0.193859 -0.876620,-0.594581 0.488221,0.893920 1.973471,0.885024 0.530294,0.539725 -0.414954,-0.587714 -4.392441,-1.027566 2.274607,0.176971 1.143538,0.515883 -0.027778,0.542672 1.445331,-0.349707 -2.458323,-1.052588 -0.505756,-0.081780 0.587272,-0.322483 -1.228562,0.320895 0.417113,0.552399 0.283349,-0.312340 0.251422,0.628206 2.239550,0.010761 1.136896,-0.097129 1.437481,-0.066334 1.903701,-0.612599 -0.415024,-1.051789 -0.367261,-0.804123 1.128948,-0.369076 0.019185,0.160616 -1.548273,-0.141557 -1.307194,-0.772846 0.752733,0.488248 -2.480031,0.427974 -1.979617,0.499403 0.117348,-0.134807 -2.852712,0.378147 -1.767343,0.691200 -2.138880,-0.124259 -0.831433,0.893290 1.731806,1.372237 1.371526,0.951177 2.877515,0.601621 4.076215,-0.676098 2.497158,-0.589417 3.313064,-0.918685 -3.378666,-1.559376 2.095334,-0.398068 -2.117082,-1.267400 -0.922764,0.729931 -1.025464,-0.894722 0.377124,1.215636 1.614274,0.907022 1.099692,0.652313 -0.106877,-0.717152 -0.786301,-1.112981 4.775043,-0.177132 -0.420944,-0.297339 0.072825,0.672166 -1.406923,-1.165750 -1.801421,0.256154 -3.095564,0.013994 -1.167269,0.842994 -1.691158,0.862693 -1.266265,1.054021 0.360568,1.269758 -1.040196,0.616838 -0.017631,0.979528 0.146513,-0.126904 -2.945640,0.285485 -0.596516,0.165562 -0.093382,0.156948 -2.312797,0.614878 0.873817,1.105509 4.094623,0.295834 -1.460583,0.212780 -1.186457,-0.797508 -0.109123,-0.582537 1.714930,1.334049 -1.260105,0.235870 -1.572837,-0.012257 -0.108803,-0.501892 -2.583345,0.806193 -0.984441,0.998757 -0.921609,0.279293 -0.969857,1.158816 -0.677452,0.976418 0.230494,1.057206 -2.184684,0.349284 -0.788626,0.735049 -2.387844,-0.008180 2.114948,1.770436 -0.559862,0.153233 -1.004488,0.790095 -1.007437,-0.920985 -0.717672,0.708063 1.854114,1.465584 -0.939920,0.150749 3.656253,1.032412 1.674828,-0.993880 -1.399640,-0.075386 -1.828110,-1.129814 -3.015274,-0.590283 -0.205687,1.473936 1.351957,1.487285 0.326061,0.770611 0.320996,0.045195 1.855231,-0.771162 -1.264573,-1.062035 0.094930,-0.146806 0.780525,-0.710657 -3.322151,0.314997 2.992069,0.519578 -2.177885,-1.132271 0.129580,1.326998 -1.450882,-1.242168 1.404761,1.651857 3.787831,-0.078580 -3.343309,0.038640 -3.928850,-0.971779 -0.816057,-0.712022 -1.309750,1.381788 2.263333,1.949192 2.416843,0.726394 -1.918670,-0.030187 1.903493,-0.661163 0.025608,-1.327853 -2.516675,0.525066 2.651012,-0.291344 -3.637375,-0.663274 4.911396,1.599186 1.320683,-1.138369 -4.642559,0.191334 -2.334780,-1.038846 1.402699,-0.010758 0.024806,0.903619 2.650822,1.025688 -0.579054,-1.637279 0.494103,0.072567 0.207010,-1.176628 2.215652,0.704969 1.368610,-0.843493 4.505346,-0.537955 0.889497,-1.558898 0.424509,-1.586454 0.807076,-1.203481 -1.177425,-1.480262 -1.922382,-0.419699 2.243807,0.211388 0.887153,0.847685 -0.770661,0.334443 -0.918726,-1.125687 4.593248,-0.012209 -1.654736,0.085789 4.393463,1.307989 1.545171,-1.274488 -5.018471,0.026946 1.209514,-0.529589 -2.620526,-1.511670 -1.672253,1.268040 1.586648,0.084823 -2.173007,0.726301 0.043230,1.399648 0.944862,-0.674335 -0.398664,0.588653 -0.786125,-0.211971 -0.279497,-0.399661 -2.058100,0.605423 -2.273557,1.206077 -0.399694,1.026159 -1.097760,1.369693 3.125732,1.904209 2.112919,0.280306 3.030686,0.417074 -0.699509,-2.069888 -0.398117,-1.050420 -1.551287,-1.112239 -0.207395,1.300660 -0.348786,0.734753 -1.178960,1.094725 3.182912,1.042950 0.939331,0.212003 2.222530,0.514231 -1.274944,-2.138600 -1.374953,-0.804188 -3.834651,-0.488160 -1.662089,1.580164 1.737478,2.061294 -0.235413,0.835292 -2.678407,0.704542 0.832684,-0.303811 -0.848751,0.056325 1.099780,1.384421 0.131968,-1.049012 -2.798485,0.283884 1.507732,-0.228883 0.150299,-0.399395 2.310448,1.186363 3.398020,-1.502377 0.222409,-0.932264 2.974807,-1.421282 -0.563751,-1.919462 -0.813693,-0.384203 -0.386669,-0.640090 0.483820,1.107198 1.855958,0.809587 0.176759,-0.187138 0.980413,-0.677121 2.886843,-1.337178 -2.221327,-1.189511 -0.934977,-0.710064 -3.363141,-0.507451 1.748039,2.142291 0.957807,0.706201 0.562640,0.673429 -0.326609,-1.737725 0.658724,-0.738924 3.353410,-0.717382 1.829571,-0.172161 -0.384823,-1.697763 4.164192,-1.160388 2.915002,-1.441737 -0.971153,-0.677917 -0.289908,-1.050664 0.859111,-0.670804 -0.616144,0.438248 -0.145412,0.420283 -2.728645,-0.656324 2.382187,1.466541 -1.825282,-0.093149 1.755473,1.620371 2.278661,-1.241769 -3.267363,-0.005513 -0.130635,-0.802675 -1.356125,-0.656845 -1.657439,1.593398 -0.975765,1.012167 -0.635328,1.778250 -1.914752,1.375476 1.590401,1.809528 -3.088613,0.150487 -1.343719,1.289995 -1.824021,-0.403921 2.349125,2.169287 -2.059420,0.479578 0.439962,1.364129 0.560437,-0.904107 -0.236046,0.617107 -2.169461,-0.626855 1.572196,0.291955 0.391025,0.104291 0.191197,0.535643 -1.113844,-1.385671 -3.558402,-0.034178 0.710753,0.854296 -0.790935,0.752900 -1.997602,1.217064 0.455987,0.314992 2.395492,0.919305 3.341749,0.435542 -3.266328,-1.635765 1.664725,-0.456243 -0.703075,-1.556285 -0.163078,1.094932 -0.100591,-0.682049 3.210156,0.914829 1.445761,-0.403320 1.708752,-0.795879 2.452446,-1.947954 -0.733293,-2.018036 1.827474,-0.956033 -0.980275,-1.533064 0.774283,0.743296 0.560744,-1.146866 1.421355,0.216514 -0.240478,-1.526900 -1.468781,-0.784588 -0.279185,-0.287563 0.743492,0.845845 -0.500229,0.595487 1.772381,0.363289 -0.026851,-1.122804 0.115716,0.053723 3.931160,-0.955604 -2.583081,-0.789193 -1.359693,0.087102 1.291418,-0.043959 -0.283583,0.770512 0.776761,0.744807 0.537266,-1.179986 2.400443,-0.090011 -0.181271,-1.514074 2.429700,-0.525470 0.339826,-1.540354 -1.378360,-0.425597 -2.317896,-0.580202 -2.617545,0.620717 1.731483,1.979382 1.485335,0.784425 -0.531649,-0.067754 -2.543100,-1.031764 -1.501045,-0.071001 1.405134,1.514056 0.297228,0.740557 0.821201,0.177526 0.706693,-1.392785 1.347108,-0.999825 -1.295512,-1.531041 -0.971415,-0.422453 4.317080,0.054624 -0.282445,0.514891 1.186631,0.378544 -3.520992,-1.434248 -1.008151,0.552989 1.453593,0.148395 2.521656,0.671375 0.255426,-0.360838 -1.475234,-1.464423 -0.866052,-0.440560 2.420078,0.682252 3.456167,0.413782 3.646386,-0.201147 2.152332,-2.056060 -0.258874,-1.994383 3.612583,-1.067817 -0.534474,-1.382450 0.879648,0.090255 -3.532419,-1.147346 1.053356,0.933542 2.464194,-0.722415 -1.602856,-0.177333 1.044080,-0.657092 1.909221,-0.908426 1.587734,0.072361 1.178136,-1.813801 0.650411,-1.816606 -3.468605,-1.574733 1.432070,-0.251838 0.638807,-0.677270 0.581672,-0.033179 4.814194,-1.401358 0.638365,-1.369305 2.801381,-1.307721 0.966939,-1.768797 1.247089,-1.463723 -1.508939,-1.698672 0.983091,-0.496419 -0.609126,-0.991306 2.004095,0.818066 -1.597031,-1.038667 -0.917374,0.585846 -1.790854,-0.067783 1.092367,1.686056 0.071200,0.505853 2.069238,0.565500 3.233937,-1.318670 5.202524,-1.071063 -1.936172,-1.838079 -1.101024,-0.554424 -0.832552,0.274168 1.839779,1.553911 -0.433650,0.300845 2.209008,0.432649 1.225600,-1.218755 -1.354897,-0.711367 1.112106,-0.494421 2.438550,-0.793338 2.577538,-0.123618 0.514664,-1.641348 -1.882971,-1.463157 3.670715,-0.353105 0.060570,-0.574491 -2.886347,0.333338 1.116975,-0.104809 -1.920214,-0.080230 -1.333211,1.079662 1.939485,0.440751 -4.185425,0.432062 -1.351166,0.819009 -2.300316,0.518841 0.431872,1.704561 -1.451468,0.821722 1.398136,0.995189 1.956430,-0.504467 1.567370,-0.512418 -0.094496,-1.548535 2.124732,-0.948545 3.809173,-1.223598 0.320223,-1.127202 0.443845,-1.312691 0.955249,-1.099241 1.316525,-0.961270 -0.345948,-1.146449 -2.065752,-0.623983 -2.554202,0.174209 -1.658681,1.441367 2.079257,1.520956 -0.594218,0.421133 4.226330,0.501630 -0.367687,-0.856401 0.777707,0.054404 1.800591,-0.916208 3.557702,-0.613325 1.621393,-1.527820 -1.562223,-1.554874 4.175439,-0.495007 1.157516,-0.809388 -2.093008,-0.489620 0.956311,-0.362245 -1.550139,-0.293364 -0.060575,0.748137 -3.362190,0.026535 0.445289,1.175014 1.493163,0.155798 2.606197,-0.446791 -1.143756,-1.174454 -4.247409,-0.254637 -3.705350,0.417438 1.478902,1.391159 -1.170454,0.378404 -0.814398,0.554672 -1.333867,0.503923 2.191981,1.157315 2.664920,0.023608 3.540414,-0.703382 -0.626422,-1.473429 -1.033903,-0.534315 3.774941,0.235008 -1.587899,-0.005878 -0.131721,0.385704 -2.796245,0.270418 -2.837560,1.165343 0.438184,1.151347 -2.099480,0.682246 -1.292849,0.927133 2.440776,0.693837 -0.504790,0.204459 1.021857,-0.091040 0.896349,-0.700874 -1.099936,-0.552721 1.506852,-0.281383 4.815823,-0.559020 -2.355404,-0.647165 -3.111527,-0.005046 -0.704504,0.932705 -2.962769,1.404781 -1.403215,1.325295 -1.098863,1.268284 0.053637,1.247847 0.669862,0.584917 0.846445,-0.216764 0.861877,-0.830779 -2.277782,-0.797032 3.426854,0.014953 0.189502,-0.354975 0.688575,-0.140479 3.933880,-0.844278 0.948435,-0.975370 -0.149476,-1.057392 -0.304002,-0.488274 2.701322,-0.128273 2.726832,-0.575633 -3.906525,-0.869246 1.227853,-0.289601 -2.710134,0.015072 -3.837990,0.527155 -2.133800,0.924566 -2.632962,1.460257 0.524164,1.167736 1.064019,0.262482 4.690770,-0.658225 2.765149,-1.279007 2.363627,-1.485755 0.941794,-1.395677 -1.825516,-0.954926 -1.620330,0.222355 0.326312,0.830589 1.356565,0.360330 1.610276,-0.565808 -1.235824,-0.895273 0.428818,-0.161397 -4.091539,0.303805 -4.518229,0.688505 2.139252,0.837894 3.013382,-0.119378 0.627534,-0.985804 0.784831,-1.110808 -0.706278,-0.691404 1.004996,-0.185438 0.536033,-0.385328 -0.801131,-0.555635 1.737899,-0.146530 5.260541,-0.422370 -1.375951,-0.757109 0.674482,-0.145592 4.619836,-0.140021 -1.767006,-0.733190 -1.426603,0.154217 4.502380,0.822994 -1.602701,0.099968 -1.761329,0.343010 -2.250279,1.204797 0.170999,1.112457 -0.574035,0.580486 -2.809609,0.443156 0.511362,0.758512 -1.919418,0.485926 1.086863,0.354892 -1.869710,0.347528 0.818901,0.121619 -2.555765,0.379018 -4.932382,0.431710 3.637619,0.905943 3.533804,-0.355431 -2.316873,-0.837166 -5.130240,0.267848 -1.796877,1.053705 -1.190212,1.266967 -3.575153,1.159318 -0.304876,1.048530 2.991600,0.448874 -4.690538,-0.270669 1.169958,-0.120373 1.284736,-0.219485 1.142672,-1.143929 -2.058173,-0.618548 -4.510279,0.368176 -0.235222,1.314180 -0.702872,0.688728 0.437308,0.334358 -0.466380,0.129959 3.032654,-0.196233 0.309745,-0.354288 4.105593,-0.974171 -3.507025,-0.395448 0.825313,-0.085677 0.015299,0.387389 -0.484222,-0.471079 -0.338839,0.399965 2.433604,0.205825 -0.810998,-0.299065 1.823670,-0.539889 -0.099598,0.054922 3.095239,-0.740911 -1.593586,-0.113980 -0.116491,0.020405 0.269112,0.768145 0.405594,-0.245919 -0.700519,-0.239858 -2.649295,0.402035 -3.471504,1.128594 -0.883255,1.075453 0.160133,0.683628 0.107503,0.066956 0.286825,-0.381439 -3.412963,0.082307 4.095370,0.206249 -1.855023,0.113465 0.812456,-0.350638 0.640811,0.148291 1.356423,-1.251871 -0.911876,-0.456437 -0.870045,0.245472 -0.458643,1.201030 -0.314487,0.499232 -0.841029,0.371623 0.947607,0.290111 -0.717310,0.039737 -1.113882,-0.050896 1.426835,0.721466 0.490911,-0.342952 -2.095280,-0.769718 -0.876343,0.700121 0.441952,0.874794 0.566387,-0.295654 -1.021763,-0.507013 0.184581,0.229044 0.604118,0.280066 1.086273,-0.800284 -3.667716,-0.419485 -0.265022,0.414239 -1.775484,1.267751 -1.470982,0.555666 -1.924392,1.166200 0.261784,0.578674 -3.460153,0.283051 -3.225507,0.289169 -1.722252,1.535066 2.315452,0.458231 0.772014,-0.552879 4.601866,-1.316095 -2.479051,-0.268979 -1.552391,0.492616 -1.559585,1.736273 0.993216,0.365149 -2.494769,-0.059683 -1.428383,0.039045 6.135125,1.120926 -3.559853,-0.209088 0.644410,-0.482650 0.473972,0.563677 0.196302,-0.948256 -0.626437,0.071672 2.068451,0.140243 1.690692,-0.369789 3.488290,-1.650338 -0.000209,-0.615071 0.218480,-0.125322 -1.476063,0.477966 0.195676,0.328616 0.967344,0.281687 0.726903,-0.976477 -0.428256,-0.549990 1.485870,-0.018748 -0.840292,0.136456 -1.488435,-0.070511 -0.295814,1.473884 -0.258926,0.390684 -0.652834,0.006072 -0.885533,0.485033 1.183654,0.322031 -2.101015,-0.069934 1.560840,-0.629350 2.408598,0.187651 -2.211309,-1.181873 -0.804451,0.619021 0.970464,1.150820 -1.009819,-0.325061 -1.061985,-0.120125 -1.087909,1.603471 2.294204,0.135143 -3.481114,-0.287019 -2.980322,-0.134534 -0.365067,1.571141 -0.517172,0.299479 -1.091676,0.077502 0.549196,0.337583 2.870156,-0.431399 2.870729,-1.422444 0.245568,-0.828643 -1.453567,0.137254 0.351267,0.756175 3.084525,0.183113 -2.620389,-1.044748 -0.302276,0.287963 -0.120524,1.372586 -2.445135,-0.036885 -3.689056,0.706638 0.306966,0.853274 4.048398,-0.417742 0.033626,-1.239821 1.877045,-0.643754 -0.407285,0.234282 2.433866,-0.630389 4.388791,-0.040294 0.553352,-1.511529 4.705312,-0.496868 2.218723,-0.487237 -1.381597,-0.474453 -0.816532,0.816403 -0.621781,1.568764 -1.734859,0.331970 0.261380,0.306195 -2.549574,0.314047 2.591888,-0.436017 1.546469,-0.111643 -0.191619,-1.648428 1.845529,-0.050719 0.135763,0.142945 -0.096079,-0.726619 -1.165104,0.885745 0.618408,0.379246 -2.594956,0.404864 1.554868,-0.621499 0.294669,0.322429 -0.139082,-1.338931 -0.530011,0.678470 -2.531220,0.890202 0.073121,0.618792 -0.402133,0.260825 -2.794355,-0.177299 -2.027669,0.980526 1.485685,0.495875 -1.381818,-0.384800 -1.200011,-0.455227 -0.186910,1.678411 -0.885764,0.132185 1.739563,-0.423882 2.362235,-0.518418 0.485139,-1.750615 -2.702299,0.235510 2.449662,0.734532 3.304483,0.029525 1.634018,-2.170079 3.333056,-0.643654 1.454645,-0.589312 -0.092028,-0.339257 -1.909520,0.548216 -1.188037,1.493744 0.386271,0.795488 0.623593,-0.612505 4.005549,-1.407622 -0.182892,-0.531072 -0.695337,0.146197 -0.145097,1.195750 -1.130636,0.525022 -2.916256,0.350415 -1.013170,1.181488 3.566312,0.072252 -1.492448,-0.636067 1.702851,-0.877054 -0.544072,0.820239 -1.132881,-0.698144 1.292105,1.150574 2.361022,-0.640929 3.892536,-2.046988 -4.239383,-0.119976 -3.610437,1.100178 -2.253633,2.297361 1.370922,0.065182 -1.044253,-0.432212 -3.096147,-0.548260 -1.895252,1.885043 -6.556904,0.803409 -1.865985,0.600054 -1.113670,0.567652 0.571144,0.037342 -1.190899,-0.249449 -2.362854,-0.195451 1.619345,1.029530 -1.294828,-0.355329 -0.708777,-0.608221 -1.277364,1.762601 -0.814424,0.328639 0.365115,0.245990 2.779214,-0.785804 0.565314,-1.337233 -0.204657,-0.480799 -1.587826,0.860818 2.585551,0.744656 0.457405,-0.351299 3.081170,-1.893544 -3.118530,0.352962 -1.348989,0.603454 2.499365,1.492406 -4.649784,-0.481959 -3.117124,-0.386938 1.064433,1.578574 -0.825595,-0.458074 -1.480506,-0.530614 1.077822,1.339572 0.853211,-0.729668 1.416180,-1.703042 3.765178,-0.539467 -0.197753,-0.379716 -1.271929,0.596438 -0.645209,1.622643 0.187870,0.546849 0.105491,-0.460069 -0.336238,-0.537684 0.056469,0.246855 0.852286,0.009544 -1.152077,-0.411267 -0.723103,0.367115 0.146068,1.385855 1.082939,-0.592433 0.463481,-1.193558 -2.218370,-0.220983 0.647621,0.771492 2.794482,0.264874 -0.883188,-1.480190 0.072527,0.328785 1.889928,0.887963 2.797394,-1.374006 0.334428,-1.051402 -2.552433,0.344546 3.379596,0.909336 -0.553170,0.382010 -0.788004,-0.767976 -2.140245,1.809510 -0.258997,0.593762 3.417718,-0.151701 -0.967329,-0.945831 0.657067,-0.522630 1.929123,0.585302 -1.703836,-1.097852 1.311714,0.227813 0.646106,0.600155 1.735844,-1.717001 1.614096,-0.431276 -2.073427,-0.289834 2.400317,0.443561 2.957794,0.406522 -1.264967,-1.458691 1.034751,0.494295 -2.111223,1.071075 0.156513,-0.565056 1.021840,0.755815 2.201940,-1.452848 2.612882,-1.210185 -5.411359,-0.137266 -3.068884,1.045069 -3.611698,2.165787 0.202147,0.269498 -1.210143,-0.112526 2.365621,-0.742174 -1.558914,0.368509 1.846077,-0.948477 2.083636,0.448046 -0.388087,-1.765712 -3.011976,0.701508 -2.190220,1.877888 1.041796,0.501920 -2.447814,-0.417589 1.098904,-0.877658 -1.389152,0.950128 0.332443,-0.727922 -1.354207,1.126556 2.281270,-0.676424 -1.940124,0.417335 -2.192286,-0.349457 -0.742233,2.066947 -0.562189,0.268468 0.549572,-0.246558 4.274644,-0.893199 2.859621,-1.587621 2.182422,-1.037319 -0.550160,-0.128917 0.114937,0.479515 0.527473,0.766132 0.145308,-0.797618 -1.310484,0.012781 1.301453,0.558696 -1.452792,0.376571 -3.605018,-0.352254 2.549187,1.442722 3.412230,-1.321235 1.473347,-2.219586 -1.973918,-0.006853 -1.606943,1.072682 -2.016525,2.094751 -2.181634,0.624573 2.110035,0.091160 0.847202,-0.971883 -4.122146,-1.236776 -2.196055,1.138086 0.703969,1.491131 2.811710,-0.947450 1.802599,-2.029964 -2.834644,-0.337104 1.226202,0.540355 1.327596,0.472644 0.567958,-1.976977 -2.518011,0.168123 -0.255708,0.894535 -0.060834,1.021579 -0.682665,-0.393267 -1.438977,0.449301 1.810127,0.536778 0.283719,-0.584721 1.423403,-1.668279 -2.997732,0.311299 0.375314,0.251194 -0.014294,0.990595 1.940152,-1.178761 -0.965095,-0.091986 -0.532850,0.074262 -3.145109,1.729881 1.118126,0.093768 -1.345078,-0.132318 -1.553666,-0.317070 -3.275153,1.821301 -0.548275,0.625118 -0.905959,0.389147 -0.306014,0.048833 -0.580041,0.347401 0.989560,-0.266025 2.788485,-0.749147 3.031684,-1.530093 1.964643,-0.906013 1.287681,-0.632836 -2.277735,0.004986 1.073129,0.593373 5.562479,0.314791 -0.903481,-1.486738 0.147942,0.403764 -0.460554,1.096642 0.865340,-0.497510 -3.982680,0.176082 -0.393731,0.084480 1.949503,0.923898 1.352156,-1.360405 -1.950279,-1.039577 -0.804649,1.051035 2.171792,1.293118 -1.074208,-0.667487 3.280370,-0.719383 -1.045633,0.714744 -0.318543,-0.516625 2.483324,1.006530 -1.000252,-0.606713 2.111374,-0.482429 -1.122699,0.700328 0.643705,-0.681567 -2.686892,0.981811 -1.092165,-0.008181 -2.485964,1.450229 -2.458906,0.607686 0.822610,0.369376 1.373570,-0.826759 2.166162,-1.583466 2.658087,-0.849432 3.551862,-0.715530 1.946952,-0.783139 -3.635386,-0.119402 2.888913,0.404770 3.313160,0.305025 -0.901984,-1.472745 2.003970,0.299859 -1.141114,0.677473 -2.654737,-0.184498 -0.404651,1.807561 -1.993889,0.429144 -1.380319,0.513633 1.407693,0.580278 -0.458268,-0.576347 -0.450705,-0.565880 2.496647,0.646968 2.114329,-0.801636 -1.003184,-1.196784 -1.349386,0.746157 0.152349,1.320733 -2.440266,0.133682 -1.308380,0.200572 0.838372,0.898615 1.338572,-0.764524 -0.155575,-1.344979 -0.893747,0.223938 -4.091327,1.122048 -0.993764,0.977873 2.641300,0.262414 3.782413,-1.076506 1.623435,-1.682069 2.331396,-0.871791 -0.027619,-0.510802 -1.449317,0.315654 0.453159,0.974797 -0.474710,0.250980 -2.850391,-0.163115 -2.353190,1.359213 0.294378,0.713074 -0.680390,-0.110974 -0.368123,-0.178027 1.708800,0.370151 -0.022473,-0.627739 -0.893893,-0.403632 1.638410,0.590033 0.337252,-0.292808 2.418896,-1.164313 0.689385,-0.370670 -1.336051,-0.354357 -1.869178,0.975282 -0.765195,1.403304 1.426054,0.191622 0.926558,-0.975703 4.048125,-1.328968 -0.395189,-0.384861 2.086001,-0.143968 -0.026552,0.040077 0.378316,-0.656200 -0.785384,0.387598 -0.894796,0.455442 -0.231439,1.040051 0.810063,-0.002365 -0.924065,-0.454703 -0.338129,0.071471 -1.739055,1.022376 2.568309,0.197465 0.103692,-0.325281 1.964085,-1.196436 0.649704,-0.264668 3.078106,-0.922259 -0.725588,-0.348155 -1.259802,0.331104 2.175184,0.908855 -2.218215,-0.307478 1.581480,-0.434193 3.081078,-0.018514 1.282704,-1.747821 5.660092,-0.810565 -1.016885,-0.396713 -2.836877,0.675998 -2.924691,1.677510 -0.277336,0.903428 -1.714753,0.376315 3.283319,0.091979 -0.193237,-0.323167 -3.440411,-0.451713 0.870063,1.043601 0.727495,-0.175795 -1.680264,-0.889138 3.408870,0.273863 -1.640364,0.226697 -0.340263,-0.083800 0.039828,1.044720 4.090042,-0.466229 1.206361,-1.053661 -0.837098,-0.825144 -1.841016,0.649879 -0.857015,1.377363 -0.517105,0.579287 0.377236,0.159181 -3.611901,-0.020396 -1.750479,0.594932 0.324891,1.013045 1.004576,-0.308258 3.125279,-0.963687 -2.485826,-0.592467 -0.991201,0.547114 -0.036273,1.057292 3.178991,-0.178275 -1.149675,-0.664478 -0.163698,0.127135 -0.508238,0.820878 -0.609696,0.183703 -2.995717,0.602957 0.880408,0.603355 -2.334832,-0.032385 0.880198,-0.089415 3.810545,-0.050549 1.724164,-1.314627 -1.291386,-0.502480 -0.164159,0.481395 -1.277643,0.803877 -3.338615,0.539974 -1.527112,1.029462 3.193554,0.387253 -1.422672,-0.194689 0.904752,-0.196121 1.544546,0.008886 -0.214829,-0.908072 -0.997566,0.052112 -0.135606,0.853648 0.340292,0.291051 1.854481,-0.540023 2.492288,-0.948106 1.564974,-1.182505 -0.784359,-0.470827 2.584952,0.062111 0.388029,-0.132209 -1.898228,-0.564683 -2.076440,0.877723 -2.209875,1.351476 0.294215,0.687333 0.059453,0.025268 -2.024128,-0.151072 0.540774,0.580152 1.953090,-0.115683 -0.739510,-0.941973 -1.533419,0.219652 -0.866517,1.223069 1.615648,0.500062 -0.205290,-0.475578 -0.259563,-0.275437 2.246486,0.181710 -1.241685,-0.352145 -2.052128,0.217603 2.022316,0.789193 2.528808,-0.355932 -0.971117,-0.804549 2.871841,-0.033487 0.205963,-0.235994 -0.060788,-0.541391 2.789854,-0.079005 -1.004361,-0.442319 2.256000,-0.083049 -1.634171,-0.067961 3.283467,-0.104778 0.312906,-0.195284 -0.151690,-0.567397 0.460311,0.023455 4.991997,-0.284750 2.579566,-0.807944 -0.379862,-0.929219 0.899064,-0.149228 1.581978,-0.319294 -1.122114,-0.750141 -0.862899,0.241079 -1.092564,0.771564 -1.669169,0.757640 0.873764,0.556134 2.214050,-0.336132 -2.505275,-0.772884 -1.647610,0.538327 -0.900376,1.000384 0.689921,0.627267 -0.313536,-0.101639 2.179052,-0.240540 1.841048,-0.505333 -0.701744,-0.663358 -3.150719,0.241188 -2.351712,0.927952 1.478393,0.734676 0.868786,-0.287423 -0.692745,-0.629330 -1.162618,0.279206 2.019849,0.595415 0.186462,-0.277043 1.595585,-0.548360 0.209439,-0.591931 -1.264667,-0.246586 0.230677,0.378154 -1.947275,0.243522 2.365901,0.313806 0.104525,-0.304822 -0.281903,-0.356196 -1.880448,0.194004 -1.950711,0.847915 -1.491038,0.915407 -2.567009,1.029813 -1.118703,0.843031 0.759662,0.523474 -0.179717,-0.076232 0.419010,-0.135365 -5.576154,-0.038943 -1.002465,0.526645 -0.040431,0.558162 4.199202,0.145438 -1.162679,-0.370256 0.111852,0.155901 0.847003,0.051557 2.096435,-0.479579 -0.935401,-0.737682 -0.174468,0.098013 0.851618,0.154080 1.765162,-0.335928 -0.550185,-0.564601 -0.587069,0.014074 -2.421311,0.505305 -2.230237,0.893005 0.014002,0.876666 -3.860155,0.378703 -0.335906,0.508429 -4.222365,0.647585 2.984463,0.622215 -1.066821,-0.117325 1.369440,0.161808 1.689033,-0.400071 4.265947,-0.611688 0.570161,-0.941814 -0.187710,-0.636012 0.867588,-0.215553 -1.544903,-0.296936 -1.672407,0.483853 1.313290,0.625287 2.084098,-0.144355 -3.562337,-0.594489 0.396275,0.093692 1.157786,-0.118860 -1.273333,-0.284132 1.100254,0.105254 -1.864296,-0.168692 3.734366,0.267298 0.980850,-0.567461 -2.811127,-0.414029 -0.844713,0.170889 2.227322,0.353255 -2.097752,-0.051794 0.614932,0.279148 0.620736,-0.257146 -0.809679,-0.133168 -3.937450,0.257277 3.112494,0.424411 5.148230,-0.269431 -0.398943,-0.457122 2.436936,-0.299877 1.738879,-0.668814 1.220637,-0.721199 -0.352015,-0.650752 -0.863383,-0.140383 -0.192690,0.215678 -1.240026,0.342764 -4.545557,0.586161 1.093102,0.572249 -0.032206,-0.004946 -2.830128,0.089129 0.411106,0.259095 2.699512,-0.044464 0.476964,-0.437831 -3.168242,-0.380415 2.622140,0.154818 -2.183000,-0.273953 -3.076953,0.453596 2.847099,0.494042 -1.177378,-0.051820 -0.392465,0.436515 0.849951,0.145572 3.384163,-0.199602 -3.552477,-0.320685 0.542348,0.031484 -0.991583,-0.216265 1.522398,0.402257 -4.046061,-0.355647 -2.809588,0.530473 4.756234,0.461585 0.985028,-0.112349 -0.653622,-0.248035 -1.961967,-0.079509 -2.951266,0.468808 -0.638448,0.951577 -0.584919,0.652713 2.211697,0.509307 1.533908,-0.239712 3.434532,-0.620652 5.159159,-0.808767 0.009727,-0.849704 -0.624175,-0.437168 0.268125,-0.016477 -0.076411,-0.010634 kafbox-2.0/data/santafe.dat000066400000000000000000000757151303712747200156760ustar00rootroot0000000000000086 141 95 41 22 21 32 72 138 111 48 23 19 27 59 129 129 58 27 19 24 46 112 144 73 30 20 19 37 92 152 93 36 20 18 29 71 146 117 46 23 18 22 52 128 142 62 26 17 19 37 100 158 86 32 17 17 27 72 154 118 43 20 15 21 47 128 150 63 24 16 17 33 92 166 95 33 16 14 22 58 149 137 50 20 15 16 35 106 169 82 28 15 14 23 65 160 133 45 18 14 16 36 111 176 80 26 14 13 20 60 163 140 45 18 12 14 30 98 185 93 28 14 12 17 46 145 167 58 19 11 12 21 68 183 132 38 15 11 13 26 93 202 100 27 12 10 13 30 114 206 82 21 11 9 10 30 120 215 77 20 10 8 9 21 95 234 99 22 9 8 7 9 35 176 215 46 11 7 5 4 3 4 19 105 125 55 30 28 41 71 98 79 49 34 33 47 75 95 76 48 34 35 49 78 96 75 47 34 34 49 81 99 76 46 32 33 49 83 101 74 44 32 33 50 85 103 73 43 30 31 49 85 106 74 43 31 31 49 87 109 76 41 29 30 47 87 110 76 41 28 30 47 88 114 79 41 28 28 45 86 117 82 42 26 26 43 84 119 85 42 26 25 39 80 121 90 43 26 24 36 76 123 96 46 26 23 35 72 123 102 50 27 22 31 64 121 110 53 27 21 28 57 117 120 59 28 20 25 49 109 128 68 31 20 23 43 98 135 77 34 22 22 36 86 139 91 38 21 19 30 73 137 107 44 23 19 26 59 129 125 55 24 18 22 47 114 140 68 28 18 19 36 93 150 88 34 19 18 28 72 147 112 43 20 16 23 53 131 137 58 24 16 19 38 104 156 80 30 17 16 28 75 156 112 40 18 15 21 51 132 145 59 22 15 17 32 96 164 89 30 16 14 24 62 153 130 46 19 13 17 37 113 165 75 25 14 13 24 68 164 123 40 17 13 16 38 118 170 71 24 13 12 22 65 167 129 40 16 12 14 33 107 182 81 24 12 11 17 51 156 155 49 17 11 11 23 77 189 115 32 13 9 12 30 107 197 82 23 10 9 12 36 134 192 64 18 10 8 13 37 148 195 58 16 9 8 10 29 132 220 67 15 8 6 7 14 66 237 137 25 9 6 5 4 5 25 165 233 51 12 7 7 7 18 80 201 109 29 12 10 14 38 128 169 63 20 11 11 23 74 173 114 34 14 11 14 37 122 172 67 21 11 11 20 64 169 131 39 13 9 12 28 100 188 88 25 11 10 14 41 141 176 57 17 10 10 16 56 175 149 41 14 8 9 18 67 196 130 32 11 8 7 17 66 207 131 30 10 7 6 12 48 193 172 39 11 7 6 6 16 86 255 107 19 8 5 4 3 2 2 7 23 8 6 8 16 33 57 65 56 46 46 54 64 66 59 49 46 48 58 67 67 60 50 47 51 60 69 68 58 50 46 52 61 71 68 58 48 46 53 64 72 68 56 47 45 53 65 73 67 54 46 46 55 67 74 66 54 44 46 55 70 76 66 51 45 46 58 73 77 65 52 44 46 59 75 78 64 49 43 47 61 78 79 63 47 43 47 63 80 79 61 46 42 48 66 84 80 60 46 42 50 68 86 80 59 43 41 50 71 88 79 56 41 41 51 73 91 78 55 41 40 52 75 94 78 54 39 40 52 79 95 76 50 37 38 52 80 97 76 50 36 37 54 83 98 74 47 34 37 53 85 100 74 45 34 35 54 87 103 72 43 33 35 53 88 103 73 43 32 33 53 91 105 72 41 31 33 54 91 108 72 41 30 32 51 93 111 74 41 29 31 51 93 113 75 40 29 30 49 92 118 78 41 28 29 46 90 120 82 42 29 28 45 88 124 86 43 27 27 41 83 126 91 45 27 26 38 78 127 99 47 27 24 35 72 126 106 52 28 23 32 65 123 116 56 29 22 28 56 117 126 64 30 22 27 49 107 133 74 32 21 23 41 95 140 84 36 21 21 35 80 141 99 42 22 20 29 67 135 117 49 24 18 25 52 122 134 62 26 18 20 41 104 148 79 32 18 19 32 82 151 100 39 19 17 25 62 140 128 52 22 17 21 44 116 150 71 27 18 18 32 88 159 98 36 19 16 25 61 146 132 52 21 17 19 41 113 161 76 28 16 16 28 77 163 114 39 19 15 20 47 132 156 62 24 15 16 30 86 173 102 34 16 13 19 50 141 154 57 21 13 14 27 87 179 103 33 15 12 18 45 136 166 61 20 12 13 23 72 178 122 36 14 11 14 32 110 188 80 23 12 10 15 45 151 169 53 18 12 9 17 60 182 143 38 13 9 9 19 71 201 126 30 12 9 8 18 70 211 131 30 11 8 8 13 47 192 179 40 13 9 7 7 14 71 255 131 24 9 7 7 5 4 3 8 20 13 10 13 20 31 42 57 68 70 59 48 46 51 61 70 67 57 48 47 53 65 72 67 57 47 47 54 67 73 66 54 48 47 55 68 74 65 54 46 48 57 70 75 65 52 45 47 58 72 74 64 50 44 47 60 74 76 63 48 44 48 61 77 77 61 48 43 48 64 79 77 60 47 42 49 67 82 77 58 45 42 51 69 84 77 56 43 42 52 72 86 77 56 42 42 54 74 88 76 54 41 42 55 78 90 75 52 41 41 56 81 92 74 51 39 42 57 83 94 72 49 38 42 58 86 95 70 47 36 40 59 89 97 70 44 36 40 59 90 97 68 43 35 38 59 92 99 68 42 33 38 59 95 100 67 41 31 36 59 97 101 65 39 31 36 58 98 103 65 38 29 35 59 100 106 65 36 29 33 57 100 109 66 36 28 32 56 100 112 68 37 27 30 54 100 116 71 37 27 30 51 98 120 74 37 26 28 48 95 124 79 39 27 27 44 91 128 85 41 26 26 41 85 130 91 44 26 25 37 77 131 101 47 27 24 34 68 129 112 53 28 23 30 61 123 123 60 30 22 26 52 113 133 68 31 21 24 44 101 140 81 34 21 21 36 86 144 95 39 22 19 30 71 140 114 48 23 19 25 55 127 133 60 26 18 21 42 107 147 76 31 18 19 32 84 152 99 39 19 17 26 62 143 127 51 23 17 21 44 118 152 71 27 16 18 33 88 160 99 37 19 17 24 61 146 135 52 23 15 19 40 111 162 77 29 17 16 27 75 163 118 42 20 16 20 46 130 160 65 24 15 16 29 83 172 108 36 18 14 20 47 136 161 62 23 15 14 27 81 178 112 36 17 13 16 40 128 173 67 22 13 12 21 64 173 134 39 15 11 13 29 98 192 92 27 12 11 15 39 137 184 63 20 11 10 15 50 167 163 46 16 9 9 16 56 187 152 39 13 8 9 15 52 189 165 40 14 8 8 11 31 145 225 61 15 9 7 7 8 31 176 231 48 12 8 6 5 4 6 26 118 131 67 35 28 38 62 96 93 59 37 31 41 65 98 92 58 36 31 41 66 99 92 56 35 32 40 68 102 93 57 35 30 39 67 104 94 55 33 29 38 67 105 95 55 32 28 37 66 108 97 55 32 27 36 65 108 100 55 31 26 33 64 109 103 56 31 24 31 60 108 108 59 31 24 30 57 108 113 62 31 22 28 54 105 119 65 33 23 27 49 100 124 71 35 23 25 44 95 129 80 36 24 24 40 88 133 89 39 23 22 35 78 133 99 44 24 22 30 68 130 112 50 24 21 26 58 123 125 59 27 19 23 47 110 137 69 30 19 21 39 94 145 84 34 19 18 32 76 144 105 41 21 17 25 59 133 125 52 23 16 20 44 115 144 67 26 16 18 34 90 154 91 33 17 15 26 66 146 120 45 19 14 19 45 122 146 64 24 15 16 31 89 160 92 32 16 14 23 61 147 129 47 20 14 17 39 113 162 74 27 14 15 26 74 164 115 39 18 14 19 44 128 158 64 23 14 15 27 81 171 109 36 16 13 18 44 133 164 63 22 13 13 24 75 175 118 35 15 11 15 36 121 179 72 23 12 11 18 57 166 145 44 15 10 12 24 86 194 105 28 13 10 12 32 117 195 76 21 11 9 13 37 142 188 59 17 9 7 12 38 154 190 53 14 8 7 10 29 135 218 64 16 8 6 7 14 65 237 139 25 8 6 4 4 5 19 126 252 70 16 9 7 10 29 114 191 78 23 13 12 20 58 156 138 46 17 12 16 36 110 173 78 25 13 13 22 62 161 133 44 16 12 15 33 107 178 81 25 13 11 18 55 157 147 47 16 11 12 26 89 187 101 28 13 9 14 37 129 180 65 20 11 10 16 53 167 155 43 14 9 9 19 68 193 128 32 13 8 9 20 77 208 116 28 10 8 8 16 68 214 130 28 9 6 6 9 37 176 199 45 11 7 5 4 7 33 188 221 42 9 6 5 3 3 6 39 145 129 54 27 23 32 64 107 97 53 30 26 36 66 107 96 53 31 26 35 66 108 98 54 30 25 33 64 110 101 54 30 24 32 62 109 104 57 30 24 30 59 108 109 59 30 23 29 55 106 115 62 31 22 26 51 103 120 68 32 22 24 46 97 126 74 35 22 22 40 89 129 82 36 21 21 35 81 132 92 40 22 20 31 71 130 103 45 22 19 27 60 125 116 52 24 18 24 51 115 131 63 27 17 21 41 101 140 77 31 18 19 34 84 145 94 37 20 17 28 67 140 116 47 22 17 22 53 123 137 60 24 16 19 38 101 151 80 31 17 16 28 77 152 107 39 19 15 22 54 136 136 55 22 16 17 37 104 157 78 29 16 15 26 73 157 113 40 18 14 20 47 130 149 60 22 13 15 29 91 167 94 31 14 12 20 54 147 139 50 19 13 15 32 100 172 85 27 14 11 20 57 155 139 46 17 12 14 31 99 179 88 27 13 11 18 51 151 152 50 18 11 12 25 83 185 107 32 14 12 14 37 125 182 70 21 12 11 17 51 164 157 48 15 10 10 19 68 192 130 35 12 9 10 21 80 206 114 27 12 9 9 19 77 216 119 27 11 7 7 12 51 200 168 35 11 7 5 5 14 73 255 123 21 8 6 4 4 2 3 18 71 64 43 35 37 48 64 76 69 53 42 41 51 68 77 69 53 43 44 54 71 79 68 51 42 43 56 73 80 67 51 42 44 56 76 80 67 48 41 43 58 78 82 64 46 39 44 59 81 83 63 45 38 44 62 84 83 62 43 38 44 64 86 83 59 41 37 44 65 89 83 57 40 35 44 67 92 83 56 38 35 45 70 94 84 55 37 34 45 72 97 83 53 36 34 44 73 99 83 51 34 33 46 75 101 83 50 34 31 45 76 104 84 50 33 30 43 77 107 85 48 31 30 43 77 108 86 48 30 29 42 76 111 87 48 30 27 40 76 113 90 48 29 28 39 73 115 93 49 28 25 36 71 116 97 50 28 25 35 68 116 102 52 30 24 32 64 116 108 56 29 23 30 59 114 115 61 30 22 27 53 108 121 64 32 22 26 47 102 129 73 33 22 24 42 93 134 82 37 22 22 37 84 136 94 41 23 22 32 73 134 108 47 24 20 27 60 127 123 56 27 19 24 49 114 136 68 29 20 21 39 97 147 84 34 19 19 32 79 146 103 40 21 17 25 60 135 125 53 23 16 21 44 115 147 69 27 18 19 33 89 156 93 36 19 16 25 65 145 124 48 21 15 20 45 120 152 68 26 16 17 31 87 163 99 35 17 15 23 58 144 138 53 21 15 17 36 106 166 81 28 16 15 24 67 160 125 43 19 14 18 40 118 167 72 25 14 14 24 70 167 123 40 17 14 16 38 119 174 73 25 14 13 21 64 167 134 41 17 12 14 30 103 185 88 26 15 10 15 46 148 166 55 18 11 11 20 67 184 131 36 13 10 11 24 90 200 101 26 12 9 11 27 110 207 84 21 10 8 10 27 114 216 80 20 10 8 9 20 88 232 106 23 10 5 5 9 32 165 224 51 12 7 6 4 4 4 16 104 159 65 28 22 31 61 106 98 57 31 27 36 65 103 96 54 32 27 36 64 104 97 55 33 27 36 64 106 100 57 32 26 35 63 107 102 57 32 25 33 60 108 106 59 32 25 31 58 107 111 61 32 24 30 55 105 115 64 32 23 27 49 101 121 69 33 23 25 46 96 125 75 35 22 24 41 89 129 83 38 23 22 36 81 132 93 41 23 21 32 72 131 104 46 24 20 28 62 126 117 53 25 19 25 53 117 129 64 28 19 22 43 102 140 77 32 19 20 35 87 145 94 38 20 19 29 69 141 114 46 22 17 24 54 126 133 59 25 17 21 41 106 150 77 31 18 18 31 80 152 102 38 19 16 24 59 139 130 52 22 15 19 42 112 154 73 28 15 16 28 80 158 105 36 18 15 22 54 138 142 54 22 14 16 34 101 164 85 29 16 14 23 64 157 127 45 18 14 17 39 116 166 73 25 15 14 24 70 165 121 39 17 13 16 39 120 170 71 24 14 13 22 67 168 129 41 16 12 16 35 110 183 81 25 13 12 18 53 157 153 49 17 11 12 23 80 190 113 32 14 10 13 31 111 195 80 22 11 9 13 39 142 185 60 18 9 9 13 43 160 180 51 15 8 9 12 38 158 195 55 15 8 6 9 23 110 239 83 17 8 6 5 7 24 142 253 62 13 6 5 4 3 4 21 116 165 74 29 19 24 45 97 122 71 34 22 26 46 95 121 74 36 23 25 44 91 124 80 37 24 24 40 87 127 86 38 23 23 36 78 129 94 43 24 21 32 71 127 103 47 24 20 29 63 124 115 53 26 20 25 54 117 125 61 28 19 22 45 104 136 73 30 19 20 37 89 142 87 36 21 19 31 75 141 104 42 21 18 25 60 131 124 53 23 17 21 47 115 140 68 27 17 18 35 93 152 88 33 18 17 27 70 146 114 43 19 15 22 50 127 140 59 24 16 18 36 98 157 84 30 16 15 26 69 153 117 43 19 14 19 45 125 149 63 23 14 16 31 89 164 95 33 16 13 21 56 145 139 51 20 13 16 33 101 169 83 28 14 13 21 60 156 133 45 18 12 14 33 107 173 82 25 13 12 19 58 158 140 44 17 11 13 30 96 182 91 28 13 11 17 45 144 163 56 18 11 11 21 70 182 124 35 14 10 11 27 100 195 89 24 11 10 12 35 131 190 66 19 9 8 12 40 151 184 54 15 9 9 11 37 153 195 55 15 9 6 8 24 116 233 78 17 8 6 6 8 34 179 220 46 11 8 6 4 4 3 7 39 72 46 34 31 38 53 74 80 65 46 38 41 56 76 80 63 46 39 44 60 79 82 62 45 38 44 62 82 81 61 43 38 45 63 85 81 59 41 37 45 66 88 82 57 40 37 46 68 90 81 55 39 36 45 70 93 81 53 38 35 46 72 95 81 52 36 35 46 74 98 81 50 34 34 46 76 100 81 50 34 34 46 77 102 82 48 33 32 46 79 105 82 47 32 31 45 79 107 83 47 31 30 44 80 111 86 47 31 30 44 80 113 87 47 29 29 43 79 115 89 47 29 27 40 78 117 91 48 28 26 39 75 118 97 49 28 25 35 70 119 101 51 29 24 33 65 118 107 55 28 23 30 61 114 114 59 29 22 28 54 110 122 64 32 22 26 48 104 128 72 33 22 23 41 94 134 81 35 21 22 36 83 136 93 40 22 20 32 71 135 107 46 23 20 27 59 127 122 56 25 19 24 49 115 137 68 28 18 21 39 96 146 84 34 19 19 32 77 146 106 41 21 17 25 59 135 128 54 24 17 21 44 113 148 72 28 18 19 33 88 156 97 36 19 16 25 63 144 127 50 22 16 20 43 117 153 71 27 16 16 29 83 160 102 36 17 15 22 55 140 140 54 22 14 16 34 101 166 85 29 15 14 23 63 155 129 45 18 14 17 37 112 169 76 26 13 14 23 67 163 127 41 17 13 16 35 113 175 77 25 13 12 20 60 163 140 46 18 12 14 29 98 185 93 27 14 12 17 43 142 171 59 19 12 12 20 63 179 136 38 15 11 11 24 87 200 106 27 13 9 11 27 105 207 86 23 11 8 11 27 111 215 82 21 10 9 9 20 90 230 103 24 10 8 7 10 38 182 206 45 12 7 5 5 4 10 58 246 138 27 10 8 9 22 84 189 105 29 14 10 17 43 132 157 60 22 13 14 28 87 171 99 32 15 13 19 50 142 150 54 19 12 13 29 89 177 97 30 14 12 17 48 142 158 56 19 12 12 25 78 180 112 34 15 11 15 37 122 181 71 22 12 11 18 55 166 151 46 16 11 11 22 77 193 116 31 12 9 11 26 100 203 91 24 10 9 11 28 114 210 80 21 10 8 9 25 106 223 86 20 9 7 6 14 65 226 139 26 10 6 6 5 13 74 255 123 20 8 5 4 3 3 13 85 191 99 32 16 16 30 75 137 98 41 21 19 29 68 131 108 48 23 19 27 59 123 120 55 25 18 23 50 113 132 66 28 19 21 41 97 141 79 32 19 19 33 83 144 97 38 20 17 27 64 137 117 48 22 16 22 50 122 137 62 25 17 19 37 98 151 82 31 18 17 28 75 151 109 41 19 16 23 54 132 137 56 23 16 19 37 104 157 80 29 16 16 27 72 155 113 40 18 14 20 47 129 149 61 23 14 15 31 91 166 94 32 15 14 21 57 147 138 50 20 12 15 33 103 170 84 28 15 13 21 59 156 136 45 18 13 15 33 104 177 84 26 14 12 19 55 155 146 48 18 11 14 28 91 184 98 29 13 10 15 40 136 172 61 19 11 11 19 62 176 138 39 15 11 12 23 86 197 105 28 12 10 13 29 109 203 84 22 11 9 12 29 120 209 76 19 10 8 9 25 107 225 86 20 9 7 7 14 58 222 153 31 10 7 5 5 8 41 219 193 34 10 6 5 4 4 11 70 184 111 36 18 18 30 69 131 104 44 22 20 29 65 126 111 51 25 20 26 57 120 122 59 26 19 23 48 109 132 68 30 19 22 40 96 139 82 33 20 19 34 81 142 96 39 21 18 27 66 137 116 49 23 18 23 52 123 134 61 26 17 19 40 103 148 80 31 18 18 31 80 150 102 39 20 15 24 59 138 129 51 22 16 19 42 112 152 72 27 16 17 29 82 158 102 36 17 15 22 55 140 137 53 22 15 17 37 106 162 80 28 15 14 25 69 158 119 41 17 13 18 42 123 159 66 23 14 14 26 78 168 110 36 15 12 18 44 132 159 61 22 14 13 25 78 174 112 35 15 12 15 40 128 170 65 21 12 12 21 65 172 129 39 15 11 13 30 104 187 85 25 12 10 15 43 145 170 56 18 11 10 18 60 180 142 39 14 9 10 21 75 200 119 30 12 9 9 21 82 212 111 26 11 8 7 17 70 217 130 28 10 8 6 10 35 165 214 51 12 7 5 5 6 20 126 255 71 15 7 5 4 4 8 46 170 145 46 18 14 21 48 120 136 60 26 17 20 41 106 143 72 29 18 18 34 87 146 92 35 18 17 27 68 141 114 45 21 17 22 51 125 135 59 24 16 19 39 102 152 80 30 17 16 28 76 152 107 40 19 15 22 53 134 138 56 22 14 19 36 103 160 80 29 16 15 26 71 156 117 41 18 14 19 44 126 153 63 24 14 15 29 86 167 100 34 15 14 20 52 141 148 55 21 14 15 30 92 173 94 30 15 12 19 51 147 150 53 19 12 14 28 89 180 99 30 14 11 16 44 139 165 59 21 11 12 22 71 179 123 36 14 10 12 31 106 191 84 24 12 9 14 41 145 177 57 17 10 10 16 54 174 155 42 14 9 9 16 60 191 145 36 13 9 8 14 55 194 157 37 12 8 7 9 33 152 217 55 14 7 5 5 8 36 193 212 41 11 7 5 4 3 6 34 154 151 55 22 17 24 52 112 121 61 28 21 25 51 107 124 66 31 21 24 44 100 130 74 34 21 22 39 91 133 84 36 21 21 34 79 136 95 42 22 19 29 67 132 111 47 23 18 25 56 122 125 57 25 18 22 45 109 138 70 29 18 18 35 91 146 88 35 19 18 29 72 143 109 43 20 16 23 55 130 132 56 24 16 19 40 107 150 75 28 16 17 30 81 155 101 37 18 15 24 57 138 132 53 22 15 18 38 109 157 75 28 16 15 27 76 159 111 38 18 14 20 49 132 149 59 22 14 15 31 92 168 93 31 15 13 21 57 149 140 51 20 13 15 33 101 172 86 27 15 13 20 57 155 140 47 18 12 14 30 98 179 89 27 14 11 17 49 148 155 52 17 11 11 24 80 184 111 33 14 11 14 34 120 186 73 22 12 10 16 48 159 162 49 16 10 10 18 64 187 137 36 14 9 10 20 74 204 122 29 11 9 8 17 70 212 130 29 12 8 7 12 45 189 182 40 12 8 6 5 13 64 250 141 24 9 7 5 4 4 4 26 103 106 59 35 31 42 67 91 80 53 37 35 46 72 93 79 52 37 36 48 74 95 78 50 36 36 49 77 96 78 49 34 35 48 78 98 77 47 33 34 49 80 101 76 44 32 33 49 83 103 76 45 32 32 49 85 104 78 43 30 31 47 85 108 77 43 30 31 47 85 110 79 42 28 29 46 85 113 80 42 29 28 44 84 116 83 42 28 27 42 83 119 87 44 28 27 40 79 121 91 46 28 26 37 75 122 97 48 27 24 35 71 121 103 51 28 24 32 64 119 111 55 28 22 29 58 115 120 60 31 22 25 50 108 127 68 31 21 24 45 100 133 78 34 21 22 37 87 138 89 38 22 20 32 75 137 104 45 23 19 26 61 129 120 53 24 17 23 49 116 135 64 27 18 20 39 98 147 82 33 18 17 31 78 147 104 41 20 17 25 59 135 128 52 23 16 20 43 113 149 70 28 17 18 32 86 156 98 36 19 16 24 60 143 130 50 21 17 19 42 115 156 73 26 16 16 28 80 160 108 38 18 15 21 52 137 147 58 22 13 16 32 95 169 91 30 16 14 21 57 151 139 49 19 13 15 32 102 174 86 29 15 12 21 57 156 141 49 18 13 15 30 97 181 91 28 14 12 17 48 146 159 54 18 12 12 23 76 184 116 34 14 10 13 33 114 190 79 24 12 10 15 44 150 174 55 17 11 10 17 56 179 153 41 14 10 9 17 60 194 146 37 14 9 8 14 52 192 165 39 13 8 7 10 28 135 235 67 16 9 8 6 7 20 125 255 73 14 7 5 3 4 6 30 141 161 62 24 18 24 49 107 126 67 31 22 24 46 101 127 72 33 21 23 43 93 131 81 35 22 22 37 84 134 90 40 23 20 33 74 134 103 45 23 20 28 64 128 116 52 24 19 25 52 118 131 62 28 18 22 42 103 140 75 31 19 19 34 86 145 94 38 20 18 27 67 140 115 46 22 16 22 51 125 136 61 25 17 20 39 102 152 81 31 18 17 29 76 152 108 39 20 17 23 56 134 137 56 24 16 19 37 104 159 80 29 16 15 26 72 156 115 40 18 14 19 47 129 150 62 23 14 15 30 88 167 96 32 16 14 21 54 145 143 52 19 12 13 30 97 171 88 28 14 12 19 55 151 144 49 17 12 14 29 94 178 91 28 13 11 17 49 146 157 54 18 11 12 23 78 182 112 32 14 10 14 33 117 185 74 22 12 10 15 47 156 164 50 16 10 11 19 63 186 138 36 13 9 9 20 74 203 123 30 11 9 9 18 71 213 128 30 10 8 8 13 47 193 176 40 12 7 6 6 14 72 255 128 23 9 6 5 4 3 3 13 57 53 34 29 33 42 61 78 75 56 41 38 46 64 80 75 56 42 39 48 66 83 74 55 41 40 49 70 84 73 53 40 39 51 72 86 72 51 39 38 51 74 89 72 49 38 38 53 78 89 71 47 35 38 54 80 92 70 46 35 38 56 84 94 69 43 34 37 56 87 95 68 43 33 36 57 89 97 67 42 33 37 57 92 99 67 41 32 36 57 93 100 67 39 31 36 58 96 103 66 39 31 35 58 97 105 67 38 29 33 57 98 107 67 37 28 34 56 99 112 69 38 27 32 53 98 114 71 37 27 29 51 97 118 73 37 27 28 48 94 123 78 39 25 27 45 91 125 83 40 25 25 40 86 128 89 42 25 24 36 78 128 97 46 25 23 33 71 127 106 49 27 22 31 63 124 117 56 27 22 27 55 116 128 64 30 21 25 46 105 137 75 33 21 22 39 91 142 89 37 21 21 32 76 142 107 45 23 19 27 61 132 125 55 25 18 23 48 116 143 69 29 19 21 38 95 151 89 36 20 19 29 73 148 115 44 21 17 24 54 130 139 60 25 17 19 38 102 158 84 32 17 17 27 73 154 115 41 19 16 20 48 129 150 62 24 15 18 32 92 165 94 32 18 14 23 59 148 136 50 20 14 16 36 106 168 81 27 15 14 22 64 160 130 43 18 14 16 36 112 174 78 25 15 13 21 61 162 137 45 18 12 14 31 102 182 87 27 14 12 18 49 150 160 54 19 12 12 23 74 185 120 34 14 11 13 30 106 196 85 25 12 11 14 38 137 190 64 19 10 10 14 41 156 183 53 16 9 9 11 38 157 194 54 15 9 7 10 24 116 235 79 17 9 6 5 8 30 165 233 49 12 7 5 4 3 2 10 59 102 68 42 34 39 58 79 81 61 44 38 43 61 81 81 61 43 38 46 64 84 81 58 42 38 46 66 86 81 57 43 38 47 69 89 80 56 38 36 47 71 91 80 54 39 35 47 73 94 80 52 37 36 49 75 97 79 51 36 36 48 78 99 78 49 35 35 49 79 101 78 48 33 35 48 81 104 78 46 33 32 48 83 106 78 45 32 32 47 83 109 80 45 30 31 46 84 111 81 44 30 30 45 84 113 83 44 29 29 44 82 116 85 44 28 27 42 80 119 89 45 28 26 39 77 121 93 47 28 25 37 73 121 99 49 27 25 34 69 121 106 52 29 23 30 62 118 113 57 29 22 29 56 114 121 63 31 23 26 50 107 129 70 32 22 24 43 97 135 80 35 21 21 37 86 139 92 40 22 21 31 73 137 107 45 23 19 26 60 128 125 55 25 18 23 48 114 139 70 29 19 20 37 95 148 86 34 19 18 30 75 147 109 43 20 17 23 56 132 133 55 24 16 20 41 107 152 76 29 17 17 30 80 155 104 37 18 15 22 56 138 136 54 22 16 18 38 106 160 80 29 15 15 26 72 158 117 41 18 14 19 46 128 155 64 24 15 15 29 85 169 104 33 16 14 20 49 139 152 58 21 13 14 28 88 175 100 31 15 12 18 46 140 160 57 20 13 13 25 78 179 112 32 14 11 14 36 122 179 70 22 12 11 18 56 165 148 44 16 10 11 22 78 193 113 30 13 10 11 27 104 203 88 23 10 8 11 30 119 207 75 20 10 8 10 26 116 220 80 19 9 8 8 17 74 234 124 25 9 6 5 6 16 95 255 96 18 8 5 5 3 3 9 59 137 94 45 29 31 47 79 97 73 44 32 35 52 83 97 72 44 33 35 53 86 99 71 43 32 34 54 87 100 69 42 31 34 54 89 101 70 41 30 33 53 90 104 69 39 28 32 52 91 106 70 39 29 32 52 92 110 71 38 27 30 50 93 112 73 38 26 28 47 91 116 74 38 26 28 45 89 119 79 40 26 26 43 86 122 83 41 25 25 39 81 124 89 43 25 24 36 77 125 97 47 25 23 33 70 125 104 50 25 22 30 62 122 115 56 28 21 27 55 117 125 62 29 20 24 47 105 134 73 32 21 22 39 93 140 85 36 21 20 34 79 140 101 42 22 19 27 64 134 119 50 23 18 23 50 119 136 64 26 17 20 39 100 148 81 31 18 18 30 78 150 105 40 20 16 23 57 135 130 53 22 16 19 41 110 153 75 29 16 17 29 80 157 103 37 18 14 21 54 138 138 55 22 15 17 36 103 163 81 29 15 14 24 67 158 123 43 17 13 18 42 121 161 69 24 14 14 25 75 167 114 36 16 12 17 43 128 164 65 23 12 13 24 73 173 118 36 15 11 15 37 120 176 71 24 13 11 20 59 166 141 43 15 10 11 25 91 191 98 27 13 10 13 36 128 187 67 21 10 9 15 45 159 169 50 15 9 9 16 53 179 160 41 13 8 8 14 48 181 171 42 13 8 7 10 29 138 225 63 14 7 5 5 8 33 183 221 44 10 7 5 4 3 3 16 81 106 65 41 34 42 61 82 78 58 42 37 46 64 84 78 57 42 38 47 68 86 77 56 41 39 49 71 89 77 54 39 38 48 73 91 76 51 38 37 50 75 92 76 50 36 36 51 78 95 75 48 35 35 51 80 96 74 46 33 35 51 82 99 73 45 33 34 52 84 100 73 43 31 34 51 87 103 72 43 31 34 52 88 105 74 42 31 33 52 90 108 74 41 29 32 50 90 110 75 40 28 29 48 90 113 76 41 28 29 47 89 117 79 41 27 29 45 88 119 83 41 27 27 42 84 122 87 43 27 25 39 80 124 93 45 25 24 35 74 124 99 47 27 23 32 67 123 107 52 27 22 29 60 120 118 57 29 21 26 53 112 126 65 31 21 24 45 101 134 75 32 20 21 38 90 139 88 36 21 20 32 76 139 104 43 22 19 26 62 131 122 53 24 17 23 48 117 138 66 27 18 19 37 96 149 84 34 18 18 30 75 148 109 41 20 17 23 56 133 134 56 24 15 19 40 107 154 77 29 17 16 28 77 156 108 39 18 15 22 53 135 142 57 22 15 16 35 100 163 85 30 16 14 23 64 155 127 44 19 12 17 39 116 164 72 25 14 14 24 72 166 119 39 16 12 17 39 122 168 69 23 13 13 22 69 169 124 38 16 11 14 34 114 180 78 24 13 11 17 54 160 148 47 17 11 12 23 81 190 109 30 12 9 12 32 115 194 77 22 10 9 13 39 144 185 59 17 10 8 14 43 160 182 51 15 9 9 11 36 151 203 56 15 9 7 8 18 92 246 101 20 9 6 5 5 12 74 255 123 20 8 5 5 4 4 12 66 129 86 44 30 32 50 83 96 69 41 32 36 54 87 95 67 41 32 36 57 89 97 66 41 32 36 57 91 98 66 39 31 34 58 94 100 64 38 30 35 57 95 101 64 37 28 33 57 96 103 65 36 27 32 56 97 106 65 36 26 32 54 99 110 67 36 26 30 52 98 114 69 35 27 29 50 96 118 72 37 25 27 46 94 121 77 37 25 25 43 89 125 82 39 25 25 40 84 127 89 41 24 24 36 77 129 98 45 26 23 31 68 128 108 51 26 21 28 60 122 120 57 27 21 26 52 113 131 66 30 20 22 42 101 139 77 33 20 21 36 85 144 93 38 20 18 29 69 138 112 47 23 17 24 55 126 132 58 25 16 20 41 106 148 76 30 17 17 31 82 152 99 36 18 16 25 60 142 127 51 21 15 19 42 116 152 70 27 16 17 30 84 159 100 35 17 14 21 56 141 136 52 22 14 17 36 105 164 82 29 16 14 24 67 158 123 42 17 13 17 40 120 164 71 24 13 13 24 71 168 119 38 16 12 15 39 120 170 70 24 12 13 20 65 169 131 40 16 11 14 31 106 184 85 24 12 11 16 47 151 163 52 18 11 10 20 68 184 128 35 13 9 10 24 90 202 100 26 11 8 10 26 106 209 84 21 9 7 9 25 103 222 89 21 10 7 7 15 67 226 132 26 9 7 6 6 17 95 255 97 17 7 6 4 4 2 5 21 52 35 29 31 39 52 62 68 65 55 48 46 54 63 69 65 55 48 48 56 66 71 65 55 48 49 57 68 72 64 53 48 50 59 69 72 63 52 47 50 61 71 73 62 51 47 50 63 73 72 60 49 46 51 63 75 72 59 48 45 51 65 76 72 57 47 45 52 68 79 72 57 46 45 53 69 80 71 54 43 44 55 72 82 70 52 43 44 57 75 84 67 50 42 44 58 78 85 68 49 41 44 60 82 86 67 48 40 44 61 84 87 65 46 39 45 63 87 87 64 44 39 44 65 89 88 62 42 37 45 66 92 88 61 40 36 45 68 95 88 58 39 35 43 69 96 89 57 37 34 43 70 100 90 55 36 32 43 71 103 90 55 35 32 41 72 104 90 53 34 30 41 72 108 91 53 33 29 40 72 109 93 53 32 29 39 70 111 97 53 32 27 37 70 113 100 55 32 27 36 67 113 104 55 31 25 32 63 113 108 58 31 24 32 59 111 113 62 32 25 29 55 108 121 66 33 24 28 50 103 126 72 34 23 26 45 97 132 81 37 24 24 38 87 135 90 40 23 22 33 75 135 102 46 24 21 29 64 131 117 53 26 20 26 53 120 131 61 28 19 21 43 104 142 76 32 19 20 35 86 147 94 37 20 17 28 68 141 117 46 22 18 23 51 124 139 62 26 16 19 37 99 153 82 31 17 17 28 72 152 113 42 20 15 21 51 131 144 61 24 16 18 35 98 162 88 31 16 15 24 65 153 127 45 20 15 19 41 118 162 72 26 15 15 26 75 166 115 39 18 14 18 43 128 163 66 24 14 14 25 76 173 116 36 16 12 15 39 124 172 69 22 13 12 20 64 170 134 40 16 11 12 29 97 190 91 26 12 10 14 41 138 178 61 19 10 10 16 53 171 155 43 15 10 9 17 63 193 139 35 12 9 8 16 61 202 143 33 12 8 6 11 41 178 190 45 13 8 6 7 13 65 249 140 25 9 7 5 5 3 2 2 1 0 0 0 0 0 0 0 4 14 46 95 89 50 31 29 43 72 92 73 46 32 33 48 76 94 74 46 33 34 49 80 96 73 44 32 33 50 82 99 73 43 31 33 50 84 101 72 42 30 32 50 86 103 73 42 29 30 49 87 105 73 41 28 30 49 88 109 74 40 28 29 46 88 113 76 41 28 29 46 88 115 79 42 27 28 44 87 119 83 42 27 27 42 85 123 89 45 28 26 39 80 125 94 47 28 26 37 75 126 101 50 28 25 35 70 125 110 54 29 25 32 62 122 120 61 32 24 29 56 115 129 67 32 23 26 48 104 137 78 36 23 25 41 93 142 90 38 24 22 35 78 142 104 45 24 20 28 63 133 124 54 25 20 24 50 119 140 67 30 20 21 38 98 151 86 34 20 19 31 76 151 111 43 21 18 24 56 134 137 57 24 17 21 40 107 157 81 31 18 17 29 77 158 113 41 20 17 23 53 135 149 61 25 18 19 36 98 169 92 33 19 17 25 63 155 137 51 22 16 19 39 111 173 82 30 18 17 25 67 164 134 46 22 15 19 37 112 180 84 28 17 14 22 60 163 146 49 20 14 16 30 95 189 99 31 16 13 18 41 137 178 64 22 14 13 20 59 176 150 44 18 12 12 22 77 199 123 33 14 11 11 23 86 213 112 27 12 10 10 21 77 221 125 28 12 10 8 13 43 186 195 44 14 9 8 7 10 37 196 217 43 12 8 8 6 5 7 18 54 52 40 37 41 51 65 74 72 60 50 47 54 66 76 72 59 49 48 56 69 79 72 58 49 48 57 71 79 72 56 48 49 58 73 80 70 56 47 48 60 75 82 68 54 45 48 60 78 81 67 51 44 47 62 80 83 65 49 43 48 63 83 82 64 47 42 48 66 86 82 62 45 41 48 68 87 82 60 44 40 49 71 90 82 57 42 40 50 73 92 82 57 41 38 50 75 95 82 54 40 38 51 77 98 82 54 39 38 52 80 100 81 52 38 37 52 82 102 81 50 36 37 53 84 105 80 50 35 36 51 86 107 80 47 34 34 51 87 110 81 47 33 33 50 88 113 81 46 32 32 49 88 115 83 46 31 31 47 86 117 85 45 30 30 43 84 119 87 45 29 29 41 80 122 91 47 28 27 39 77 123 97 48 27 24 35 72 123 103 50 28 24 32 66 121 111 55 28 22 30 59 118 119 60 29 22 27 53 111 128 68 32 22 25 46 101 136 77 35 23 24 39 90 141 91 40 24 22 34 76 140 107 46 23 21 29 63 133 124 56 27 19 24 51 118 141 68 30 20 22 39 98 151 88 35 20 19 31 77 150 111 43 22 18 24 57 134 136 58 25 18 20 41 107 155 79 30 18 18 29 78 157 110 39 20 16 22 52 135 144 58 23 15 17 34 99 166 89 31 17 15 23 63 154 130 46 19 13 17 37 113 169 76 28 14 14 24 67 164 126 41 18 13 16 37 116 175 76 25 14 12 22 62 165 137 44 18 12 15 30 100 187 91 28 14 13 17 46 146 172 58 20 13 13 20 65 182 139 39 16 11 12 24 85 203 111 29 13 9 12 26 98 213 97 25 11 9 9 23 92 224 104 24 10 7 8 14 55 214 160 34 11 7 6 5 12 61 252 147 25 8 7 6 4 4 4 14 36 23 21 25 34 48 58 63 64 59 52 49 52 57 64 64 60 54 51 54 60 67 65 60 53 51 55 62 68 66 60 53 51 56 63 69 67 58 53 51 56 65 70 65 58 52 52 58 67 71 66 56 51 51 59 68 72 65 56 50 51 60 71 73 65 56 50 52 62 72 74 64 53 48 52 63 73 73 62 52 48 53 64 76 73 62 50 48 54 66 77 73 59 48 46 54 68 78 73 58 47 46 54 70 80 72 56 46 47 56 73 82 71 55 44 46 57 75 82 70 52 44 45 58 77 83 67 51 41 46 60 80 84 67 48 40 45 61 82 85 65 47 40 45 63 85 85 62 44 39 45 65 88 85 61 42 38 45 67 90 86 60 42 37 45 68 94 87 58 40 36 46 71 96 86 57 38 36 46 73 99 87 54 37 35 46 73 100 86 54 36 34 46 75 104 87 52 34 33 44 76 106 87 51 33 31 44 75 109 88 50 32 30 43 76 110 90 50 32 29 41 75 113 92 50 30 28 39 73 115 96 50 30 27 37 71 116 101 52 30 26 35 67 116 104 55 30 24 32 63 115 110 57 30 24 30 58 113 118 62 31 23 28 53 108 124 67 32 23 26 47 101 131 76 35 22 24 41 91 136 85 38 22 23 35 81 137 98 42 24 20 31 69 134 113 50 25 20 27 57 126 129 60 27 19 23 46 109 142 73 31 19 20 37 91 149 91 37 20 18 29 71 145 115 45 22 18 23 53 128 138 60 25 17 19 38 103 156 83 31 18 17 29 74 154 113 42 19 15 21 51 131 146 60 24 15 17 33 97 164 89 31 16 15 23 62 152 131 47 20 14 18 39 113 166 76 26 15 13 24 69 165 124 41 17 12 16 38 118 172 73 25 14 13 22 65 168 131 41 17 12 15 32 107 184 85 26 13 12 17 49 153 160 52 18 11 12 23 73 188 123 35 14 11 12 27 98 201 92 25 11 9 11 30 120 204 75 22 10 8 10 29 123 213 73 18 9 8 9 20 94 236 99 20 9 6 6 8 30 163 229 52 12 7 5 4 3 3 7 42 53 25 21 24 39 70 89 72 47 37 38 52 73 83 66 47 37 40 54 77 84 66 46 38 40 57 80 86 65 44 37 41 58 82 87 63 43 36 41 59 86 88 63 42 36 41 61 89 88 60 40 33 40 63 91 89 59 38 33 41 64 94 89 58 36 32 39 65 97 90 56 36 31 40 67 100 92 56 35 30 39 66 102 92 55 34 29 39 67 104 95 54 33 29 36 66 106 97 54 32 27 36 66 108 100 56 33 27 35 64 110 105 57 32 25 33 61 110 108 60 32 25 31 57 109 113 62 32 24 29 54 105 119 67 33 24 27 49 101 125 72 35 23 25 44 95 130 80 36 23 23 39 87 133 89 40 23 22 34 77 134 99 44 23 20 29 65 132 113 50 25 20 26 55 122 128 60 27 20 22 45 107 139 73 31 19 20 36 90 148 91 36 20 19 29 72 144 112 45 22 18 24 55 130 137 60 25 18 21 41 106 152 79 31 17 17 29 79 156 107 38 19 16 23 56 138 138 56 24 15 19 39 106 162 81 29 17 16 27 71 159 119 42 19 15 19 45 127 158 65 24 14 15 28 82 170 108 35 16 13 18 47 136 158 60 21 13 14 27 83 178 108 34 15 12 17 42 132 169 63 21 13 12 22 68 177 129 39 15 11 13 30 104 191 87 25 12 10 14 41 143 178 59 18 11 10 17 53 174 158 44 14 10 9 16 59 192 148 37 13 8 9 14 52 191 164 39 12 8 7 9 28 137 232 65 15 7 6 5 6 22 134 255 67 14 8 6 4 4 4 8 29 33 23 23 28 38 54 71 76 63 47 41 44 56 72 74 62 47 41 47 61 77 75 60 46 41 48 63 78 75 58 44 40 49 65 81 76 57 43 40 49 68 83 74 54 41 40 50 70 85 73 53 39 39 51 73 87 73 51 38 39 52 75 88 73 49 37 38 53 79 91 72 47 36 38 54 82 93 70 46 35 38 56 85 94 69 44 35 37 56 88 96 69 42 33 37 57 90 97 68 41 32 37 58 94 100 67 41 32 36 58 95 102 66 39 30 35 57 96 104 66 38 30 34 57 98 107 68 38 28 33 55 99 110 68 37 27 32 54 99 113 70 38 26 30 51 96 116 72 37 25 28 49 95 120 76 37 25 26 44 91 124 81 38 24 24 40 86 128 87 41 24 24 37 80 129 95 44 24 23 33 72 128 105 48 25 22 29 62 125 116 55 27 21 26 54 117 128 63 29 20 24 45 105 138 75 33 21 21 37 90 144 91 36 20 20 32 75 143 110 45 23 19 26 59 132 129 56 26 19 22 45 113 146 72 29 18 19 34 88 154 96 36 19 17 26 65 146 124 48 21 16 20 46 121 151 68 26 16 17 32 89 162 98 35 18 15 23 57 145 136 50 21 14 17 36 107 165 80 28 15 14 24 68 161 125 42 19 13 17 39 119 168 72 24 14 13 24 68 168 125 40 17 12 15 36 115 178 76 25 13 12 20 57 164 145 46 18 12 13 27 90 190 101 29 13 11 14 37 128 188 69 21 12 10 16 48 161 169 51 17 10 9 16 55 182 158 42 14 9 8 14 51 186 169 42 14 8 8 9 31 146 224 60 15 8 7 7 9 36 190 215 42 11 7 5 4 4 3 5 16 11 7 10 17 28 38 46 54 63 66 59 51 46 50 57 65 66 59 52 47 51 58 66 67 58 52 48 52 61 68 67 59 51 48 52 63 70 68 58 50 48 54 65 71 67 56 48 47 55 67 74 68 55 48 49 57 69 73 67 54 47 48 59 71 74 66 53 47 49 60 72 77 65 52 46 50 62 75 77 63 51 45 50 64 78 78 63 49 44 50 66 79 76 61 47 44 52 68 82 76 60 46 43 53 69 84 77 57 45 43 54 73 87 76 56 43 42 55 76 88 75 54 42 42 55 78 90 73 51 40 41 56 80 92 73 50 39 41 58 84 93 71 47 38 41 58 86 94 69 46 37 40 59 89 95 68 44 35 39 60 91 96 67 42 34 38 60 94 99 66 41 33 38 61 96 100 66 40 31 37 61 98 102 65 38 31 37 61 100 104 65 37 30 36 60 102 107 65 37 29 36 59 103 110 67 37 28 32 57 102 114 69 37 27 32 54 102 118 71 37 27 29 51 99 122 76 38 26 28 47 95 126 79 39 25 27 43 90 130 88 41 25 25 39 81 132 95 44 26 24 34 74 131 106 50 27 21 30 64 127 117 55 27 20 27 55 118 129 64 29 21 24 46 105 140 76 32 20 21 37 90 146 91 38 21 20 31 74 143 111 46 22 19 25 57 132 131 57 26 18 22 43 111 149 74 30 19 19 33 87 155 99 37 20 17 25 62 144 129 51 23 16 20 44 117 156 71 28 17 17 30 83 162 105 36 18 15 22 54 140 144 55 23 15 17 34 100 169 88 30 16 14 22 62 155 135 48 19 14 16 35 108 175 82 27 15 14 21 61 161 139 45 18 12 15 31 99 183 90 28 13 12 17 47 148 162 55 19 11 12 22 73 185 123 35 14 10 12 28 102 198 89 24 12 10 12 35 129 197 68 20 10 9 12 36 142 198 61 18 10 9 10 30 130 221 70 18 9 8 8 14 69 238 136 26 10 7 6 5 8 37 209 200 39 11 8 5 6 11 52 181 149 42 16 11 15 35 111 174 76 25 13 13 24 68 164 124 41 17 13 17 39 120 167 71 24 13 13 22 69 169 124 39 16 12 14 36 115 178 74 24 12 10 19 57 164 144 45 17 11 12 26 90 191 101 29 12 11 14 35 126 189 70 20 11 9 16 46 158 174 53 17 10 9 15 51 177 165 44 14 9 9 13 45 175 182 45 13 8 7 9 25 120 239 77 18 9 7 6 6 21 123 255 74 15 8 6 5 5 4 5 16 19 11 13 19 31 46 66 76 67 51 43 45 56 70 73 64 50 44 47 59 74 76 61 49 42 48 62 75 75 61 47 42 49 63 78 75 59 45 42 49 66 80 75 57 44 41 51 68 83 74 55 42 40 52 72 85 73 53 41 40 51 73 86 73 51 40 40 53 75 89 72 50 39 41 54 79 90 71 48 37 40 55 82 92 69 45 36 39 56 86 94 68 45 36 40 57 89 95 68 43 34 39 58 90 97 67 42 34 39 60 94 98 66 40 32 38 61 95 100 66 40 33 37 59 98 102 65 38 30 36 61 100 104 65 37 29 35 60 100 kafbox-2.0/demo/000077500000000000000000000000001303712747200135575ustar00rootroot00000000000000kafbox-2.0/demo/.gitignore000066400000000000000000000000741303712747200155500ustar00rootroot00000000000000################# ## Figures ################# *.png *.pdf kafbox-2.0/demo/demo_parameter_estimation_lorenz.m000066400000000000000000000037221303712747200225520ustar00rootroot00000000000000% DEMO_PARAMETER_ESTIMATION_LORENZ Estimation of the parameters of the % KRLS-T algorithm for predicting the Lorenz attractor time-series. % % This demo estimates the optimal parameters of the KRLS-T algorithm when % applied to predict the Lorenz attractor time-series. The estimated % parameters are: forgetting factor lambda, regularization c and Gaussian % kernel width. Kernels other than the Gaussian can be used by modifying % kafbox_parameter_estimation.m. % % Author: Steven Van Vaerenbergh, 2013. % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ close all clear %% PARAMETERS horizon = 1; % prediction horizon embedding = 6; % time-embedding N = 500; % number of data %% PROGRAM tic fprintf('Loading Lorenz attractor time-series...\n') [X,Y] = kafbox_data(struct('name','lorenz','horizon',horizon,... 'embedding',embedding,'N',N)); fprintf('Estimating KRLS-T parameters for %d-step prediction...\n\n',... horizon) [sigma_est,reg_est,lambda_est] = kafbox_parameter_estimation(X,Y); fprintf('Running KRLS-T with estimated parameters...\n') Y_est = zeros(N,1); kaf = krlst(struct('lambda',lambda_est,'M',100,'sn2',reg_est,... 'kerneltype','gauss','kernelpar',sigma_est)); for i=1:N, if ~mod(i,floor(N/10)), fprintf('.'); end % progress indicator, 10 dots Y_est(i) = kaf.evaluate(X(i,:)); % predict the next output kaf.train(X(i,:),Y(i)); % train with one input-output pair end fprintf('\n'); SE = (Y-Y_est).^2; % square error toc %% OUTPUT fprintf('\n'); fprintf(' Estimated\n'); fprintf('sigma: %.4f\n',sigma_est) fprintf('c: %e\n',reg_est) fprintf('lambda: %.4f\n\n',lambda_est) fprintf('Average MSE after first 100 samples: %.2fdB\n\n',... 10*log10(mean(SE(101:end)))); figure; hold all; plot(Y); plot(Y_est); legend({'original','prediction'},'Location','SE'); title(sprintf('%d-step ahead prediction %s on Lorenz time series',... horizon,upper(class(kaf)))); kafbox-2.0/demo/demo_parameter_estimation_nonstationary.m000066400000000000000000000031151303712747200241450ustar00rootroot00000000000000% DEMO_PARAMETER_ESTIMATION_NONSTATIONARY Estimation of the parameters of % the KRLS-T algorithm in a non-statiory setting. % % This demo generates data as y = ft(x) where ft is a time-varying % function. Then it estimates the optimal parameters of the KRLS-T. The % estimated parameters are: forgetting factor lambda, regularization c and % Gaussian kernel width. Kernels other than the Gaussian can be used by % modifying kafbox_parameter_estimation.m. % % Author: Steven Van Vaerenbergh, 2013. % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ close all clear %% PARAMETERS N = 500; % number of data sigma = 1; % spatial kernel width lambda = .995; % forgetting factor (temporal kernel lambda) c = 1E-2; % AWGN variance %% PROGRAM tic fprintf('\nGenerating time-varying data with spatio-temporal\n'); fprintf('AR(1) covariance...\n'); X = randn(N,1); % 1D input data time_ind = 0:N-1; T = toeplitz(time_ind)/2; Kt = lambda.^T; K = kernel(X,X,'gauss',sigma); Kn = c*eye(N); K = Kt.*K + Kn; Z = randn(N,1); % GP generator M = zeros(N,1); % output data mean Y = chol(K)'*Z + M; % output data fprintf('Estimating parameters of KRLS-T for these data...\n\n'); [sigma_est,c_est,lambda_est] = kafbox_parameter_estimation(X,Y); toc %% OUTPUT fprintf('\n'); fprintf(' True Estimated\n'); fprintf('sigma: %.4f %.4f\n',sigma,sigma_est) fprintf('c: %.4f %.4f\n',c,c_est) fprintf('lambda: %.4f %.4f\n\n',lambda,lambda_est) fprintf('\n'); % plot data figure; plot3(time_ind,X(:,1),Y,'+') xlabel('time'); ylabel('x'); zlabel('y'); kafbox-2.0/demo/demo_prediction.m000066400000000000000000000015141303712747200171020ustar00rootroot00000000000000% 1-step ahead prediction on Lorenz attractor time-series data % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ [X,Y] = kafbox_data(struct('name','Lorenz','embedding',6)); % make a kernel adaptive filter object of class krls with options: % ALD threshold 1E-4, Gaussian kernel, and kernel width 32 kaf = krls(struct('nu',1E-4,'kerneltype','gauss','kernelpar',32)); %% RUN ALGORITHM N = size(X,1); Y_est = zeros(N,1); for i=1:N, if ~mod(i,floor(N/10)), fprintf('.'); end % progress indicator, 10 dots Y_est(i) = kaf.evaluate(X(i,:)); % predict the next output kaf.train(X(i,:),Y(i)); % train with one input-output pair end fprintf('\n'); SE = (Y-Y_est).^2; % test error %% OUTPUT fprintf('MSE after first 1000 samples: %.2fdB\n\n',10*log10(mean(SE(1001:end)))); kafbox-2.0/demo/demo_prediction_kstep.m000066400000000000000000000022261303712747200203110ustar00rootroot00000000000000% Demo: k-step ahead prediction on Lorenz attractor time-series data clear close all; %% PARAMETERS k = 10; % prediction horizon % load data using helper function [X,Y] = kafbox_data(struct('name','Lorenz','horizon',k,... 'embedding',6,'N',5000)); % Make a kernel adaptive filter object kaf = krlst(struct('lambda',1,'M',100,'sn2',1E-6,'kerneltype','gauss','kernelpar',32)); % kaf = qklms(struct('eta',0.5,'epsu',.1,'kerneltype','gauss','kernelpar',32)); %% RUN ALGORITHM N = size(X,1); Y_est = zeros(N,1); for i=k:N, if ~mod(i,floor(N/10)), fprintf('.'); end % progress indicator, 10 dots Y_est(i) = kaf.evaluate(X(i,:)); % predict the next output % Train on input-output data. Use the pair from (horizon-1) steps ago % since the newest output available at step i is Y(i-horizon+1). kaf.train(X(i-k+1,:),Y(i-k+1)); end fprintf('\n'); SE = (Y-Y_est).^2; % test error %% OUTPUT fprintf('MSE after first 1000 samples: %.2fdB\n\n',... 10*log10(mean(SE(1001:end)))); figure; hold all; plot(Y); plot(Y_est); legend('original','prediction'); title(sprintf('%d-step ahead prediction %s on Lorenz time series',... k,upper(class(kaf)))); kafbox-2.0/demo/demo_prediction_kstep_split.m000066400000000000000000000032711303712747200215250ustar00rootroot00000000000000% Demo: k-step ahead prediction on Lorenz attractor time-series data. % This script trains and tests on separate splits of the data. clear close all; %% PARAMETERS k = 10; % prediction horizon N_train = 1000; % number of training data N_test = 1000; % number of test data, max 5000 - N_train seed = 1; % random seed for reproducibility % load data using helper function [X,Y] = kafbox_data(struct('name','Lorenz','horizon',k,'embedding',6,'N',5000)); % Make a kernel adaptive filter object kaf = krlst(struct('lambda',1,'M',100,'sn2',1E-4,'kerneltype','gauss','kernelpar',32)); % kaf = qklms(struct('eta',0.1,'epsu',.1,'kerneltype','gauss','kernelpar',32)); %% RUN ALGORITHM % init random generator rng('default') rng(seed); % corrupt outputs with noise Y = Y + 3*randn(size(Y)); % split X_train = X(1:N_train,:); Y_train = Y(1:N_train); X_test = X(N_train+1:N_train+N_test,:); Y_test = Y(N_train+1:N_train+N_test); Y_test_MSE = zeros(N_train,1); for i=1:N_train, % train on data set 1 if ~mod(i,floor(N_train/10)), fprintf('.'); end % progress indicator, 10 dots kaf.train(X(i,:),Y(i)); % train with one input-output pair % test on data set 2 Y_test_est = kaf.evaluate(X_test); % store MSE SE = (Y_test-Y_test_est).^2; % out-of-sample error Y_test_MSE(i) = mean(SE); end fprintf('\n'); %% OUTPUT % learning curve figure plot(10*log10(Y_test_MSE)) title('Learning curve') % errors for final prediction figure; hold all; plot(Y_test); plot(Y_test_est); legend('original','prediction'); title(sprintf('%d-step ahead prediction result for %s on test set after %d training steps',... k,upper(class(kaf)),N_train)); fprintf('Final MSE: %.2f\n',Y_test_MSE(N_train)); kafbox-2.0/demo/demo_prediction_mackey_glass.m000066400000000000000000000041201303712747200216200ustar00rootroot00000000000000% 1-step ahead prediction on Mackey Glass time series. % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ close all; clear %% PARAMETERS h = 1; % prediction horizon L = 10; % embedding n_train = 200; % train data n_test = 100; % test data sigma = 1; % kernel parameter lms_lr = 0.2; % lms learning rate klms_lr = 0.2; % klms learning rate krls_nu = 1E-4; % krls precision parameter % select algorithms i=1; % algos{i} = lms(struct('mu',lms_lr)); i=i+1; algos{i} = klms(struct('M',Inf,'kerneltype','gauss','kernelpar',sigma,'eta',klms_lr)); i=i+1; % algos{i} = krls(struct('nu',krls_nu,'kerneltype','gauss','kernelpar',sigma)); i=i+1; %% PREPARE DATA % load data [X,y] = kafbox_data(struct('name','mg30','embedding',L,'horizon',h,... 'N',n_train+n_test)); % split data X_train = X(1:n_train,:); y_train = y(1:n_train); X_test = X(n_train+1:n_train+n_test,:); y_test = y(n_train+1:n_train+n_test); %% RUN ALGORITHMS n_algos = length(algos); MSE = zeros(n_train,n_algos); y_est_final = zeros(n_test,n_algos); titles = cell(n_algos,1); for j=1:n_algos, kaf = algos{j}; titles{j} = upper(class(kaf)); % store algorithm name fprintf('Training %s',titles{j}) for i=1:n_train, if ~mod(i,floor(n_train/10)), fprintf('.'); end % progress indicator kaf.train(X_train(i,:),y_train(i)); % train with one input-output pair y_est = kaf.evaluate(X_test); % evaluate on test data MSE(i,j) = mean((y_test-y_est).^2); % test error end fprintf('\n'); y_est_final(:,j) = y_est; end %% OUTPUT figure; plot(10*log10(MSE)); title('Learning curves') legend(titles); xlabel('iteration'); ylabel('MSE (db)'); fprintf('Final MSE: %.2fdB\n\n',10*log10(MSE(end))); figure; hold all; plot(y_test,'LineWidth',2); titles2 = {'original'}; line_styles = {'--','-.'}; for j=1:n_algos, plot(y_est_final(:,j),'LineWidth',2,'LineStyle',line_styles{j}); end titles2(2:n_algos+1) = titles; legend(titles2); title(sprintf('%d-step ahead prediction on Mackey Glass time series',h)); % export_fig('test.pdf') kafbox-2.0/demo/demo_profiler_prediction_lorenz.m000066400000000000000000000071001303712747200223720ustar00rootroot00000000000000% Demo: demonstration of the kernel adaptive filter algorithm profiler. % Compares the cost vs prediction error tradeoffs and convergence speeds % for several algorithms on the Lorenz data set. clear close all %% PARAMETERS output_dir_default = 'profiler_output'; % data and algorithm setup data.name = 'Lorenz'; data.N = 10000; % number of data points data.embedding = 6; % time embedding data.offset = 50; % apply offset per simulation sim_opts.numsim = 5; % 10 minutes per simulation on an Intel Pentium Core2 Duo sim_opts.error_measure = 'MSE'; i=0; % initialize setups %% KRLS i=i+1; algorithms{i}.name = 'KRLS'; algorithms{i}.class = 'krls'; algorithms{i}.figstyle = struct('color',[.75 0 .75],'marker','^'); algorithms{i}.options = struct('sweep_par','nu','sweep_val',[1E-4 2E-4 1E-3 .01 .05 .1],... 'kerneltype','gauss','kernelpar',32); %% QKLMS i=i+1; algorithms{i}.name = 'QKLMS'; algorithms{i}.class = 'qklms'; algorithms{i}.figstyle = struct('color',[1 0 0],'marker','o'); algorithms{i}.options = struct('eta',0.5,'sweep_par','epsu','sweep_val',[1 2 5 10 12 15 18],... 'kerneltype','gauss','kernelpar',32); %% SW-KRLS i=i+1; algorithms{i}.name = 'SW-KRLS'; algorithms{i}.class = 'swkrls'; algorithms{i}.figstyle = struct('color',[0 .75 .75],'marker','s'); algorithms{i}.options = struct('c',1E-6,'sweep_par','M','sweep_val',[1 2 3 4 5 10 20 50 100 200],... 'kerneltype','gauss','kernelpar',32); %% FB-KRLS i=i+1; algorithms{i}.name = 'FB-KRLS'; algorithms{i}.class = 'fbkrls'; algorithms{i}.figstyle = struct('color',[0 .5 0],'marker','d'); algorithms{i}.options = struct('lambda',1E-6,'sweep_par','M','sweep_val',[10 15 20 25 30 50 100],... 'kerneltype','gauss','kernelpar',32); %% KRLS-T i=i+1; algorithms{i}.name = 'KRLS-T'; algorithms{i}.class = 'krlst'; algorithms{i}.figstyle = struct('color',[0 0 1],'marker','+'); algorithms{i}.options = struct('sn2',1E-6,'lambda',1,'sweep_par','M','sweep_val',[10 15 17 20 24 30 50 100 200],... 'kerneltype','gauss','kernelpar',32); %% PROGRAM fprintf('Running profiler for %d algorithms on %s data.\n',i,data.name); % output_dir = input('Folder for storing results (ENTER = subfolder here): ','s'); % if isempty(output_dir), % output_dir = output_dir_default; % fprintf('Using default folder "%s" for storing results.\n', output_dir_default); % end output_dir = fullfile(mfilename('fullpath'),'..','results'); t1 = tic; [data,algorithms,results] = kafbox_profiler(data,sim_opts,algorithms,output_dir); t2 = toc(t1); fprintf('Elapsed time: %d seconds\n',ceil(t2)); %% OUTPUT mse_curves = kafbox_profiler_msecurves(results); resinds = [1,1;2,1;3,10;4,7;5,9]; % result indices [f0,h0] = kafbox_profiler_plotconvergence(algorithms,mse_curves,resinds); [f1,h1] = kafbox_profiler_plotresults(algorithms,mse_curves,results,{'bytes','flops'}); object_handle = get(h1); legend(object_handle.String,'Location','NW'); % move legend [f2,h2] = kafbox_profiler_plotresults(algorithms,mse_curves,results,{'ssmse','flops'}); [f3,h3] = kafbox_profiler_plotresults(algorithms,mse_curves,results,{'ssmse','bytes'}); [f4,h4] = kafbox_profiler_plotresults(algorithms,mse_curves,results,{'ssmse','ssplus'},1); object_handle = get(h4); legend(object_handle.String,'Location','SW'); % move legend [f5,h5] = kafbox_profiler_plotresults(algorithms,mse_curves,results,{'ssmse','timeto'},-20); object_handle = get(h5); legend(object_handle.String,'Location','NW'); % move legend [f6,h6] = kafbox_profiler_plotresults(algorithms,mse_curves,results,{'ssmse','timetopct'},90); object_handle = get(h6); legend(object_handle.String,'Location','SW'); % move legend kafbox-2.0/demo/demo_reconverge_all.m000066400000000000000000000051001303712747200177240ustar00rootroot00000000000000% Demo: learn a nonlinear channel with an abrupt change. Run and compare % all algorithms using their default parameters. % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ close all clear rng('default'); rng(1); % for reproducibility %% PARAMETERS N = 1500; % number of training data N_switch = 500; % iteration after which the channel switch occurs N_test = 100; % number of test data SNR = 30; % SNR in dB sigpower = 0.5^2; % input signal power %% PROGRAM w = who; for i=1:length(w), % copy all parameters to option structure eval(sprintf('opt.%s = %s;',w{i},w{i})) end % generate data [X,y,y_ref,X_test,y_test] = generate_channel_switch(opt); % get list of kernel adaptive filters in 'lib' folder fdir = fileparts(which('kafbox_template.m')); files = dir(fullfile(fdir,'*.m')); [~,algorithms] = cellfun(@fileparts, {files.name}, 'UniformOutput',0); for i=length(algorithms):-1:1, if ~exist(algorithms{i},'class'), algorithms(i) = []; % remove files that do not represent classes end end % perform online learning for each algorithm fprintf('\n') num_alg = length(algorithms); titles = cell(num_alg,1); MSE = zeros(N,num_alg); MSE_final = zeros(1,num_alg); for algo_ind=1:num_alg t1 = tic; algorithm = algorithms{algo_ind}; fprintf('%2d. %9s: ',algo_ind,upper(algorithm)); titles{algo_ind} = strrep(upper(algorithm),'_','\_'); kaf = feval(algorithm); for i=1:N, if ~mod(i,floor(N/10)), fprintf('.'); end y_est = kaf.evaluate(X_test); if i<=N_switch MSE(i,algo_ind) = mean((y_test(:,1)-y_est).^2); else MSE(i,algo_ind) = mean((y_test(:,2)-y_est).^2); end kaf.train(X(i,:),y(i)); end MSE_final(algo_ind) = mean(MSE(N-500:N,algo_ind)); fprintf(' %.2fs. Final MSE=%.2fdB\n',toc(t1),... 10*log10(MSE_final(algo_ind))) end fprintf('\n'); %% OUTPUT % plot results in different "leagues" [MSE_final_sorted,ind] = sort(MSE_final,'descend'); num_fig = ceil(num_alg/5); remaining = num_alg; for fig_ind=num_fig:-1:1 figure; hold all rm = rem(remaining,5); num_in_league = (rm==0)*5 + rm; % plot the results for the num_in_league worst results league_inds = num_alg-remaining+num_in_league:-1:num_alg-remaining+1; for i=league_inds, plot(10*log10(MSE(:,ind(i))),'LineWidth',1) end title(sprintf('League %d',fig_ind)) legend(titles(ind(league_inds))) axis([0 N 5*floor(min(10*log10(MSE(:)))/5) 0]); remaining = remaining - num_in_league; end kafbox-2.0/demo/demo_regression_2d.m000066400000000000000000000014401303712747200175050ustar00rootroot00000000000000% Wobbly field regression demo. Also used in lib/test/unit_test.m % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox close all clear %% PARAMETERS algorithm = 'krlst'; % algorithm class (choose from lib/ folder) opts = struct(); % algorithm options go here (kernel type, parameters, etc) %% PROGRAM kaf = feval(algorithm, opts); % generate some data c = 5; N = 1000; x = rand(N,2)*c; y = sin(3*x(:,1)).*cos(x(:,1)+x(:,2)); fprintf('Training') for i=1:N, if ~mod(i,floor(N/10)), fprintf('.'); end kaf.train(x(i,:),y(i)); % y_test = kaf.evaluate(x(i+1,:)); end fprintf('\n') %% OUTPUT [x1,x2] = meshgrid(0:.2:c, 0:.2:c); yt = kaf.evaluate([x1(:) x2(:)]); z = reshape(yt,size(x1,1),size(x2,1)); figure; surf(x1,x2,z); kafbox-2.0/demo/demo_sample_run.m000066400000000000000000000103531303712747200171100ustar00rootroot00000000000000% Demonstration of the basic operation of a kernel adaptive filtering % algorithm with fixed budget. This script generates several images and % saves them to disk. % % KRLS-T is chosen as the example algorithm. Inspired by % http://www.tsc.uc3m.es/~miguel/MLG/adjuntos/slidesKRLST.pdf % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ close all clear rng('default'); rng(11); %% PARAMETERS n = 10; % number of training data n_star = 201; % number of points to evaluate ell = 2; % length scale sigvar = 1; % signal variance noisevar = .02; % noise variance budget = 5; % KRLS-T budget savefig = 0; manual = 0; %% GENERATE DATA fprintf('Generating data...\n'); covfunc = {'kafbox_covSum', {'kafbox_covSEiso','kafbox_covNoise'}}; loghyper = [log(ell); log(sigvar); log(noisevar)]; x = (1:n)' + 0.2*randn(n,1); y = chol(feval(covfunc{:}, loghyper, x))'*randn(n,1); % Cholesky decomp. x_star = linspace(0,n+2,n_star)'; covfunc_no_noise = {'kafbox_covSEiso'}; loghyper_no_noise = [log(ell); log(sigvar)]; %% RUN ALGORITHM kaf = krlst_split(struct('kerneltype','gauss','kernelpar',ell,... 'sn2',noisevar/sigvar,'M',budget+1)); fprintf('Running KRLS-T\n'); t1 = tic; for i=1:n, if ~mod(i,floor(n/10)), fprintf('.'); end Qold = kaf.Q; kaf.train(x(i),y(i)); for stage = 1:2 switch stage case 1 % stage 1: after including a sample stage_string = 'after inclusion'; fname_i = sprintf('fig/kafbox_sample_run_%02da',i); case 2 % stage 2: after pruning if length(kaf.mu) > budget kaf = kaf.prune(Qold,1); stage_string = 'after pruning '; fname_i = sprintf('fig/kafbox_sample_run_%02db',i); end end y_est = kaf.evaluate(x(1:i)); [y_star_est,y_star_var] = kaf.evaluate(x_star); y_star_var = y_star_var/kaf.s02*sigvar; % fix scale x_dict = kaf.dict; y_dict = kaf.evaluate(x_dict); papersize = [8 4]; figure(1); clf set(gcf,'Position',[500 400 papersize(1)*100 papersize(2)*100]) hold all f = [y_star_est+2*sqrt(y_star_var); flipud(y_star_est-2*sqrt(y_star_var))]; p0 = fill([x_star; flipud(x_star)], f, [7 7 7]/8,'EdgeColor','none'); p1 = plot(x_star,y_star_est,'r','LineWidth',1.5); p2 = plot(x_dict,y_dict,'ro','LineWidth',2,'MarkerSize',8); p3 = plot(x(1:i),y(1:i),'bx','LineWidth',2,'MarkerSize',8); set(gca,'box','on'); set(gca, 'FontSize', 14); set(gca, 'FontName', 'Helvetica'); axis([0 n+2 floor(min(y))-1 ceil(max(y))+1]) legend([p3,p1,p2],'data','prediction','dictionary bases') title(sprintf('KRLS-T iteration %d: %s',i,stage_string)); set(gca, 'Position', get(gca, 'OuterPosition') - ... get(gca, 'TightInset') * [-1 0 1 0; 0 -1 0 1; 0 0 1 0; 0 0 0 1]); set(gcf, 'PaperUnits', 'inches'); set(gcf, 'PaperSize', papersize); set(gcf, 'PaperPositionMode', 'manual'); set(gcf, 'PaperPosition', [0 0 papersize(1) papersize(2)]); set(gcf, 'renderer', 'painters'); drawnow if manual keyboard; %#ok end if savefig if ~exist('fig','file') mkdir('fig'); end % write individual png print('-dpng','-r100',fname_i); print('-dpdf','-r100',fname_i); % write animated gif fname = 'fig/kafbox_sample_run.gif'; frame = getframe(1); im = frame2im(frame); [imind,cm] = rgb2ind(im,256); if i==1 && stage==1 imwrite(imind,cm,fname,'gif','Loopcount',inf,'DelayTime',1.5); else if stage==2 && i==n && kaf.pruned imwrite(imind,cm,fname,'gif','WriteMode','append','DelayTime',2.5); elseif ~(stage==2 && ~kaf.pruned) imwrite(imind,cm,fname,'gif','WriteMode','append','DelayTime',1.5); end end end end end fprintf('\n') kafbox-2.0/demo/demo_sinc.m000066400000000000000000000020351303712747200156750ustar00rootroot00000000000000% Demo: learn a sinc. Run one algorithm using its default parameters. % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ close all clear rng('default'); rng(1) %% PARAMETERS N = 1000; % number of training data N_test = 500; % number of test data SNR = 20; % SNR in dB algorithm = 'krls'; %% GENERATE DATA x = randn(N,1); x_test = linspace(min(x),max(x),N_test)'; y_ref = sinc([x;x_test]); y = y_ref + sqrt(10^(-SNR/10)*var(y_ref))*randn(N+N_test,1); y_test = y_ref(N+1:N+N_test); %% RUN ALGORITHM fprintf('%s: ',upper(algorithm)); Y_est = zeros(N_test,1); kaf = feval(algorithm); t1 = tic; for i=1:N, if ~mod(i,floor(N/10)), fprintf('.'); end kaf.train(x(i),y(i)); end y_est = kaf.evaluate(x_test); MSE = mean((y_test-y_est).^2); %% OUTPUT fprintf(' %.2fs. MSE=%3.2fdB\n',toc(t1),10*log10(MSE)) figure; hold all plot(x,y(1:N),'.') plot(x_test,y_est,'LineWidth',2) legend({'data',strrep(upper(algorithm),'_','-')}) axis([min(x)-0.5 max(x)+0.5 min(y)-0.5 max(y)+0.5]); kafbox-2.0/demo/demo_sinc_all.m000066400000000000000000000044041303712747200165270ustar00rootroot00000000000000% Demo: learn a sinc. Run and compare all algorithms using their default % parameters. % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ close all clear rng('default'); rng(1) %% PARAMETERS N = 1000; % number of training data N_test = 500; % number of test data SNR = 20; % SNR in dB %% PROGRAM % generate data x = randn(N,1); x_test = linspace(min(x),max(x),N_test)'; y_ref = sinc([x;x_test]); y = y_ref + sqrt(10^(-SNR/10)*var(y_ref))*randn(N+N_test,1); y_test = y_ref(N+1:N+N_test); % get list of kernel adaptive filters in 'lib' folder fdir = fileparts(which('kafbox_template.m')); files = dir(fullfile(fdir,'*.m')); [~,algorithms] = cellfun(@fileparts, {files.name}, 'UniformOutput',0); for i=length(algorithms):-1:1, if ~exist(algorithms{i},'class'), algorithms(i) = []; % remove files that do not represent classes end end % perform online learning for each algorithm fprintf('\n') num_alg = length(algorithms); titles = cell(num_alg,1); MSE = zeros(num_alg,1); Y_est = zeros(N_test,num_alg); for algo_ind=1:num_alg t1 = tic; algorithm = algorithms{algo_ind}; fprintf('%2d. %9s: ',algo_ind,upper(algorithm)); titles{algo_ind} = strrep(upper(algorithm),'_','\_'); kaf = feval(algorithm); for i=1:N, if ~mod(i,floor(N/10)), fprintf('.'); end kaf.train(x(i),y(i)); end y_est = kaf.evaluate(x_test); Y_est(:,algo_ind) = y_est; MSE(algo_ind) = mean((y_test-y_est).^2); fprintf(' %.2fs. MSE=%3.2fdB\n',toc(t1),10*log10(MSE(algo_ind))) end %% OUTPUT % plot results in different "leagues" [MSE_sorted,ind] = sort(MSE,'descend'); num_fig = ceil(num_alg/5); remaining = num_alg; titles{num_alg+1} = 'data'; for fig_ind=num_fig:-1:1 figure; hold all plot(x,y(1:N),'.') rm = rem(remaining,5); num_in_league = (rm==0)*5 + rm; % plot the results for the num_in_league worst results league_inds = num_alg-remaining+num_in_league:-1:num_alg-remaining+1; for i=league_inds, plot(x_test,Y_est(:,ind(i)),'LineWidth',2) end title(sprintf('League %d',fig_ind)) legend(titles([num_alg+1; ind(league_inds)])) axis([min(x)-0.5 max(x)+0.5 min(y)-0.5 max(y)+0.5]); remaining = remaining - num_in_league; end kafbox-2.0/demo/demo_template_prediction.m000066400000000000000000000123541303712747200210010ustar00rootroot00000000000000% Time-series prediction with a kernel adaptive filtering algorithm. % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ close all; clear; %% PARAMETERS % Instructions: 1. Uncomment one datafile and one kaf algorithm; 2. Execute. datafile = 'lorenz.dat'; embedding = 6; horizon = 1; % 10000 points % kaf = krlst(struct('jitter',1E-6,'lambda',1,'M',100,'sn2',1E-6,'kerneltype','gauss','kernelpar',32)); % achieves -56.82 dB % kaf = fbkrls(struct('lambda',1E-6,'M',100,'kerneltype','gauss','kernelpar',32)); % achieves -55.00 dB % kaf = rls(struct('lambda',.99,'c',1E-6)); % achieves -48.16 dB % kaf = kap(struct('mu0',0.995,'eta',0.95,'eps',1E-6,'p',8,'kerneltype','gauss','kernelpar',32)); % achieves -40.40 dB kaf = krls(struct('nu',1E-4,'kerneltype','gauss','kernelpar',32)); % achieves -40.17 dB % kaf = swkrls(struct('c',1E-6,'M',100,'kerneltype','gauss','kernelpar',32)); % achieves -37.85dB % kaf = kalman(struct('R',1E-4)); % achieves -33.41 dB % kaf = nlms(struct('mu',1.99,'eps',1E-6)); % achieves -24.58 dB % kaf = exkrls(struct('alphaf',1,'beta',.99,'lambda',1E-6,'q',1E-6,'M',500,'kerneltype','gauss','kernelpar',32)); % achieves -22.48 dB % kaf = problms(struct('sigma2_n',1E-6,'sigma2_d',1E-2)); % achieves -15.86 dB % kaf = phypass(struct('mu',0.6,'s',15,'sigma',0.95,'p',8,'omega',1/8,'kerneltype','gauss','kernelpar',32)); % achieves -15.05 dB % kaf = qklms(struct('eta',0.6,'epsu',2,'kerneltype','gauss','kernelpar',32)); % achieves -10.73 dB % kaf = qklms_fb(struct('M',300,'eta',0.6,'epsu',2,'kerneltype','gauss','kernelpar',32)); % achieves -10.73 dB % kaf = rffklms(struct('mu',.9,'D',1000,'kerneltype','gauss','kernelpar',32)); % achieves -10.86 dB % kaf = nlkapa(struct('eta',.09,'eps',1E-8,'lambda',1E-6,'M',1000,'P',20,'kerneltype','gauss','kernelpar',32)); % achieves -7.89 dB % kaf = lkapa(struct('eta',.09,'lambda',1E-6,'M',1000,'P',20,'kerneltype','gauss','kernelpar',32)); % achieves -7.62 dB % kaf = lms(struct('mu',1E-4)); % achieves -5.98 dB % kaf = klms(struct('eta',0.1,'M',5000,'kerneltype','gauss','kernelpar',32)); % achieves -3.07 dB % kaf = memory_cell(); % achieves -2.37 dB % kaf = fbklms(struct('M',10,'nu',.1,'eta',.4,'kernelpar',35)); % achieves -1.56 dB % kaf = knlms(struct('mu0',0.95,'eta',0.5,'eps',1E-6,'kerneltype','gauss','kernelpar',32)); % achieves -1.32 dB % kaf = mxklms(struct('mu',1,'eta',0.8,'M',5,'kerneltype','gauss','kernelpars',.5:.5:2)); % achieves ??? dB % kaf = mknlms_cs(struct('mu0',0.9,'eta',0.5,'rho',1E-6,'kerneltype','gauss','kernelpars',[32 33])); % achieves -1.26 dB % kaf = klms_csl1(struct('mu0',0.95,'eta',0.1,'lambda',5E-4,'kerneltype','gauss','kernelpar',32)); % achieves -1.41 dB % kaf = klms_csal1(struct('mu0',0.95,'eta',0.1,'lambda',5E-4,'kerneltype','gauss','kernelpar',32)); % achieves 3.31 dB % kaf = norma(struct('lambda',1E-4,'tau',500,'eta',0.1,'kerneltype','gauss','kernelpar',32)); % achieves 10.96 dB % kaf = kapsm(struct('M',200,'Delta',10,'mu',1.8,'Q',10,'loss','l2','loss_param',2,'kerneltype','gauss','kernelpar',32)); % achieves 21.29 dB % kaf = kapsm(struct('epsilon',10^(-10),'Delta',5,... % 'thresh1',0.1,'thresh2',0.1,'Q',200,... % 'loss_params',2,'loss_type','l2',... % 'kerneltype','gauss','kernelpar',.2,... % 'sparse_flag',0,'sparse_params',5)); % datafile = 'mg30.dat'; L = 11; horizon = 1; % 5000 points % kaf = krls(struct('nu',5E-3,'kerneltype','gauss','kernelpar',.6)); % achieves -45.61 dB % kaf = krlst(struct('lambda',1,'M',200,'sn2',1E-5,'kerneltype','gauss','kernelpar',.6)); % achieves -44.26 dB % kaf = fbkrls(struct('lambda',1E-6,'M',200,'kerneltype','gauss','kernelpar',.6)); % achieves -41.09 dB % kaf = swkrls(struct('c',1E-6,'M',200,'kerneltype','gauss','kernelpar',.6)); % achieves -35.43 dB % kaf = norma(struct('lambda',1E-2,'tau',500,'eta',0.5,'kerneltype','gauss','kernelpar',.6)); % achieves -20.08 dB % datafile = 'santafe.dat'; L = 10; horizon = 1; % Santa Fe chaotic laser time-series (data set A) % kaf = krlst(struct('lambda',1,'M',100,'sn2',1E-5,'kerneltype','gauss','kernelpar',50)); % achieves 15.83 dB % kaf = krls(struct('nu',9E-1,'kerneltype','gauss','kernelpar',50)); % achieves 22.06 dB % kaf = fbkrls(struct('lambda',1E-5,'M',150,'kerneltype','gauss','kernelpar',50)); % achieves 23.68 dB % kaf = rls(struct('lambda',.999,'c',1E-4)); % achieves 26.33 dB % kaf = nlms(struct('mu',.1,'eps',1E-6)); % achieves 27.30 dB % kaf = knlmscc(struct('mu0',.1,'eta',0.5,'eps',1E-6,'kerneltype','gauss','kernelpar',50)); % achieves 27.36 dB %% PROGRAM tic [X,Y] = kafbox_data(struct('name','lorenz','horizon',horizon,... 'embedding',embedding)); N = size(X,1); fprintf('Running prediction algorithm %s',upper(class(kaf))) Y_est = zeros(N,1); for i=1:N, if ~mod(i,floor(N/10)), fprintf('.'); end Y_est(i) = kaf.evaluate(X(i,:)); % make prediction kaf.train(X(i,:),Y(i)); % train end fprintf('\n'); SE = (Y-Y_est).^2; % test error toc %% OUTPUT fprintf('MSE after first 1000: %.2fdB\n\n',10*log10(mean(SE(1001:end)))); figure; plot(10*log10(SE)); xlabel('samples'); ylabel('squared error (dB)'); title(sprintf('%s on %s',strrep(upper(class(kaf)),'_','\_'),datafile)); figure; hold all; plot(Y); plot(Y_est); legend('original','prediction'); title(sprintf('%s on %s',strrep(upper(class(kaf)),'_','\_'),datafile)); kafbox-2.0/demo/demo_template_system_identification.m000066400000000000000000000030011303712747200232230ustar00rootroot00000000000000% Online system identification with a kernel adaptive filtering algorithm. % Author: Steven Van Vaerenbergh, 2013 % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ close all; clear; %% PARAMETERS % Instructions: 1. Uncomment one datafile and one algorithm; 2. Execute. datafile = 'mimotestbed8K.dat'; L = 4; % kaf = krlst(struct('lambda',.995,'M',100,'sn2',0.015,'kerneltype','gauss','kernelpar',3.1)); % achieves -10.41 dB kaf = swkrls(struct('c',0.015,'M',70,'kerneltype','gauss','kernelpar',3.1)); % achieves -9.13 dB % kaf = norma(struct('lambda',1E-4,'tau',500,'eta',0.4,'kerneltype','gauss','kernelpar',3.1)); % achieves -8.58 dB %% PROGRAM tic data = load(datafile); % input and output data are 2 columns N = size(data,1); X = zeros(N,L); for i = 1:L, X(i:N,i) = data(1:N-i+1); end % time-embedding Y = data(:,2); % desired output fprintf('Running system identification algorithm') Y_est = zeros(N,1); for i=1:N, if ~mod(i,floor(N/10)), fprintf('.'); end Y_est(i) = kaf.evaluate(X(i,:)); % make prediction kaf.train(X(i,:),Y(i)); % train end fprintf('\n'); SE = (Y-Y_est).^2; % test error toc %% OUTPUT fprintf('MSE after first 1000: %.2fdB\n\n',10*log10(mean(SE(1001:end)))); figure; plot(10*log10(SE)); xlabel('samples'); ylabel('squared error (dB)'); title(sprintf('%s on %s',upper(class(kaf)),datafile)); figure; hold all; plot(Y); plot(Y_est); legend('original','prediction'); title(sprintf('%s on %s',upper(class(kaf)),datafile)); kafbox-2.0/demo/fig/000077500000000000000000000000001303712747200143245ustar00rootroot00000000000000kafbox-2.0/demo/fig/kafbox_sample_run.gif000066400000000000000000005460711303712747200205270ustar00rootroot00000000000000GIF89a cFFBBBttt\\\ꖖd 333kkkCLLL%%%沲44ccc+++{{{:::ꛛoo RRR^^,秧""靝,,tRRssKK衡dd1188歭H! NETSCAPE2.0!, (&(! (.8*H$`C \( C]2Hj=BHN"I^&8sֲP~(U xB "0DhȦ65R&C\I$UL6uʝK|p~pު@hW߻C#|*@A3!v&N)fl1KmKwe'=v0(!Aѽ xv&dtQuIw!#RН@{3ߛ>! "x$SͿ}!}'R vX!M#.E%)]HB!pH-HOSA<MgejP iWq"@+$atZQ j "GSTS T6Eo!PBK[ O̲F`b3D0 >FqD IxϦ_UT&!)dx%T(!' Cpe6B(?޳h eHH :J1O }]0p 9Wz\ *QmlQmd -0:߳qŏѦ+P*6 爦63Ā6IYS <%aʄ`l;" otm |`"fӃD.#"Uz}d0p)Nl;䱑m񿘝AT .ZHĐ06?߄u]g &DV˖օhh{v+2$o~P+} {/qAT:g ~!{GR-iAY&!{+a%"LC]p戣$}Y!(ǿ"%QĿDM $i]PL"'&K/(x|OD˅C !Kҡg+`}ω<h0(`P" np$`;(BPakI`sj%/)HcA:n1xDR|of:2IFyZYE쐗Ȼ؋K_)$nu!\].Jvfh9efˋsT)͠L8d}<9 h IZ3HiZ1~Ѝ."d; QU皂H,xjNf/!f 0YA$Eyԉ#AʈLU|2YRYҞ5kҬ H!lﾡ҃+@ǠJDPBŇ1[13ƅ0N;Cq;P1@SβLBk%8 |2 0p6_;PO5#t[6կu'WݙZ׻N=av|g}boA;uDZ tN&!xK "^ʓm 0/k RaGȻWok>=CO;w/rKx>=G?ۣ<5˾n@Fqֱdu@?OLx P Di18ATx;yiG~X w~+ҷQ%Ѐ;U{Pqy@$r*d|Q  0 |WXqpFZiZ_ - /Hh P`7P,sCVH$Cx E Gb |{%=&6Iw+ _x a8z w' |QF1&PR]z'~{8^x iHR1V *!-|+,HH yhIFTyhu؋؊rC;s.X 8r13WUh<8XxXX7y} #;Ǝ娎`gP%)f I掐 ٘e ّiafwsQ*,ْ Is"|e 1pČ&:<ٓ>  gO $5d:R?R9TYVy z@,Pz+s@Ϙo#pL9lٖnp樕f{p@xoA Yy9|Xom GyI 3.`7RPc7}Æ.p (2l`)d BʹٜXil<9YyYH2 @5 ,Jr I(~啄#:(P!3]X `+Н7(e -h d:il0ڡ=% ])J$y-2^9\,_$niV)6ZFzH*=ФNPR:PrPO^YR_y7k*|.Izxz+/BX ac7h3 \JHg(zzɧ榁6{S9H^i+zj|ZZ#j k,+&qjņ&POuH}wqؚں|00ڕ:+)"ŚyT D9,P <ꋔG; ɧ " $-T9؉o2:c(:%mu*Я(t8;$[ȧEy8M౼&<۳F{˧lB 8HR8'IrM @ S۵^I68hI喵r;ʶ涙@|۷iෆ{z =p7! CI- ۱{ h K uI_\  ( ;˖Fڸ7˘z+0? / ;pp[Iк`Xk ;˭+ Oi&^ٛ}g63F 1S}^Fb3ܠ["ȻP*oAP鋡~쉺IIw&y: %}wXaf+o3*9;Q癞]O| b^\;A0 *-J.3j.]M&X4<'Gy^w,z"1U4tk , _<ɔ+†@pJhzӒ~Dn?\a,&)$h`!DD`*!]ܒP\$ɘ:8:E<e@8%Bxq2~L<!-%`H1 |T <ƜK;O8鴮\P8ͪJ͝L0fE+0nUhQ<$#] RI .븼 Ê8za0gU]"ELqX#C%^DgLFԃ(ȕЖ^k 2͡58z#ԅhdz3hGrH(@cåPؔҞ֒=np[֐+iyTȥBM( #+g1(M6W2(YϑMپQi/R4k3 pg.(8, ؝, B~ ==͐ YfLPxޮ] &Hf66& NYGqQw7ri>$&$WIA.>(%^6```7^F9TIW.>Ay2~T^G >ABz:@V>dΜXT\n_yN^py;>@ =|>r.t 8/芾茮?}?~.D84nz*>W*$~Ӕㄮ~馎7^>0@^E  雞><>엎bnjzlA& ~겾赾^:\~"B>j6^=M\><*n*OZ_>/G  EP P+-Q pPlzn@],?Hߖ4@/RKs94 bK{NE@Pz/pO7Л4{_&+"0h?^O|CO)/?ftoOJC?-60O/Y _/-LkQ) tߡj? e׮lO*ݷ=oh 8HXh(Hx)9IYiyy$ 3@: : +;K[k{ (<@h\ 4DTJMz .>NΞ\ SM} 0@xڹe0! e@ /U{P0ƍ;zUP0]"KÈ/&R4eQ_ʙ4kڼ-evt3(${/>L:}JSOZ]U(y-K4رd2յJטaڽ7/Zm'\zDFQ{QAŌoK@Z|73`ǂ!]l9>zda@|(q H4ޜzA ]p7Q *!ԩ۷\^z=x^ıb>mΌwxgt,P |56pهhMxx ^ )"),Y` ]Ab5c2ci'8ˆBH89g\Hv-> eR"V^eZne^~9H)d$EdfX>8evމgzg}j=("Oj> iN 頩sih- jc",D )":jzZ 髾~JkK;-\@ t>NKmGp@ pBsA ՎKn%\{ o R'o _ދ/,|/ /l8m OLq?rqĚUqnfq&/(").\ʉZ0ߌb4\9 tV;o~2ky{9袟^᥃z>.N{B|O+D O| ;}bp$oُlB />wz?mֿ7 V _X*Pg8@ X0€P; e5(^| 삷Ӡ V0v'la"py.aTRHІh KHH.BL$:r'iJZ2|l&7 N*(GR^K*:fI#Rtew ^0bd2fO4sI͟QAFxpJ,%7 \8r.VcwsDglu '? +(P"-"tSJ8T(,? є"PH`}tdiN.b'h Ԡ uW/B ; Q ըJUR:լju < rud-kJՂxledjָuS,NV|_gͿ vL` *]c Y46_'kjvݬg?K΂vhKԆmkUvw-mo[6v qq"w̽r vҭuCjwE#w ^e7-"Vw"/{^wڕ/}]w/\x0Zv )hS 5 ft0mQ$"zO<`[` L`3e+=k^z1]z 2N?$KP#3BVٔ]zh8 47r<35B"eye|>泡suh* i+5r/Zcz4?}MOzԾ5O]JSzՌT5_GWzu.oWYz5[[{O5=9c#{uS6-^aC{ڝ6jkc{ۥ6jo{ܭ7ks{ݵF/mIg{7m{{&7m|R7F3g/8@ǣmԘ ɉ.  9Os:c]j9#VV .Sw\L.N#'KoK$.fA 嫋S 7|co;vEt} NwW!qo~w7#ڊ_<xfC>Ȟ^=I׃:4k>Һ{{>?s>l9ʟ2Cgl3 Xdbv?k;%aZ%5Hvu/]/~#g~GK&NaE8d+bO؀H  *ȁv(,.<%hdɶ9 M>y63D'CHEhGIKȄMOQ(SHUhWY[ȅ]_GteVE6}Acf do|=ug utx|_2=r=Ӈ1pȈ舏'A~MMôpؤ)!K(0P;0H"QJN@Irr^qH!PiOFw N관 QTOOߴIqDbEOhI-,=3PE<܈遏`I"wwNM}XJMSe=ޣG=r%PӸVr,׋hȤ>d#XF>: _"Ph!"BKuDswd4)P1wgk$?!g⓿1@i:::dd ##RRR{秧;;RRtt:%霜KKu44蠠ii-r沲00Ht H*\ȰÇ#JHŋ3jȱǏ CIɓ(S\ɲ˗0cʜI͛8sɳϟ@ JѣH*]ʴӧPJJիXjʵׯ`ÊKV)ݙpʝ+4 & |@a.H(Åb( r^ 3rFh`_AsM"mͻ]+} pe( ,qw1B @HH9]s PCk`?~EcǘI+ơwj1fwC[&A%蟅BdH@v(!y~e$](,wW!}adc~d9 aPB w5]5yv:e @uucu9* ЭhNnR 9bB:$B~(70 .x:SEc B`5$aft@( h 9!^Գ|X${ Jwq`CbGLҞ#9uYubԄ n+H /ARExedq e|_nTPpcjԐ:rP" %4Sg ჽtq mpKVބcy XgJ}@F |1ٟBrUr@Ԍǁ D?.4ƙdA0Ȉ0304aMH ӀK^%SKLh0 2Z4O bs @[jF.{['7Տ3V0Qɳ!͌H6nwoғ2G4āfI3~/E jhyC6~4L'A`lfҔ5eNwAҁ1V*\  P'*X'<'ė @!bKb}! SwzG~{;{Ho]hx~W [U a 8J70"h3z 1؁.0%( 8H4?*-D8!hHX: -r1EXT|7YȄMȃN;k!gdxvKXAC0"r *^bTX~ȇ\@)$~-(p pWx!'+4~,;_{}؉(Ƈ4871Ã3"R`=8O[=KX^'~q_cTEh"%BuƈȨGx{3@iq0VHhx8819XHR@)=? IzB؏A#qj}1@]D%qUْ*QHrv_ q2SD6AnNI)*Y,YPה*-pmQٕc8B (!P7`zƕ^x+w"P#pmhv)u{B. wyy{x)P{E9){miW{i{Af.f~92qʑyHA o[sɏpٛi4 a B%ٙI|Yyؙɜ™ĩw 2oA1z\W)؝I9ɞA !La)wvٞ\EZz8A66{rmk؆ZƘ10 /@:jZfH雉#@S)郠6z۹ :ʣ;ڣ@BDE:FHv4NPR:TS H#-=pzQH> =*+x#Ym0ʱY;ɝ1(wpA[I>H:jaکJXLRE*gʙq;yec!-2,Ѣyxps(ʨpQȚ٣9GJ Ժa7Dںڭz衜΢ܔH"NiHJʫ+᫼9ZiY[9?j+A!$ )*d ))K8(3 )V*F3.tӊ눱ie,sQaX#Y(.LM+ J%S+# W6"+Pyc0!,jћIO+* 2Ӂ0 (5! +#L"ʉrJ;0p;j o{v;ɰ +,U۪ F3YCc;Bk+!i;)ٹ/Aě@Jśxɣ _$D:ỹ )&˰ wlɽm t{˝Jy{ۉr%`{!{7z̊[| krWA \훰!L~%.#|<8 9*'|Dp7C,ĢidkMܫ)6=\VLGzAڃ-!$+&vyj|#̣VŚZě]x0->0AƽPƄ\ڹ{7?P:7p6~3ܜg,hȜ ;p <," sXl/%*h )>o0xFE4El" E30 2 Yz5؜IHȥl+(Q+Dy%|~3Ȉ81fpyx[ϗX!"dja<|ѻ<ơLʃ?"1B,-§u9E2;H6!J;"ԶPUP4\|:_(PzH-* Pav(#*&Ezi8љ(Մ][,Wͤ$]; [Gڸ-֑-!H 6& 2^]&}KKPȊș 鈳6kC+al0@M[hT4yaѽZ(Kmƶʹ۹ѝ !ǦLuMM&l$(3KtM"; LHƂ06+ Ī>wۜM}Q[<ˑ}*˸;|k.)&')Xl`%i?jWfn!' E\vV^{0 ,+(/m%?Y]';g.՛)~iL wyBD$[rZNEޞ^N1l15w V]r.u"ɭ78@s~Bp 2n~裾?nۣWz0 țzF۾-l~}zＹ3p an$׮ܾe-:zB ;ZC~ 7O*,_j23û /z5 <`c>`2Hh0@?H?9w ,o"S ?1` DPB>QD-^Ę!C=~RH%MDR]\SL5męU3$"(`@UT^ŚUV]~Vb{)K$UڴYu/śW^-_"W`…w  \<__iXfΝ=cYX%Km A`˥/ƝsKڼu\$PV' M|8 ٴ(\vݽK{^ ^DV A"= ^~/@"`>"o>B /0C @H/[!vnܰE;gmۋE:jnk$H#Dqn" ʕJ-/4\zʣ6DGqK5dߖ;2S0: ,}3,6!$jOBDA'RKqPa :!Q!D(A!RUwU_%.d1:BPV_VLgpX[S`46+YsqŘ.6\qnXlg"ڊ1ȅW(,]ޚ %{J=f2)촷`|EwԳ_#D8cWTcG·.N7&G X$ cgy!En}cBps>)kwTfBHg7J)`DQ gk:l6[`z1:\ylJ[B E'=g\ղ;rBjM*%@"եAoL _u!Q&toOPO}gҬ;!'>ǝJ]u4}g>{=T *>zz'>[уo_\5!Pa7w\lЀC\0X_$6zA1iZD0LUGM` eؕ3`BjZZ7J:@i%rBJ .C%&]|U0 Dda/`48cG>яd 9HBҐDd"HF6BOXƇ4SA&5IE5l+q8&H?`6HT,2KY֒e.qK]e09La4f2Le6s? (E3vҚK *xGqP%3)a g<9OzӞ\$$wI00\#^M ~$$<5Tj= w0CR0@/|s(9PcE';!3? PT 68 s2(!%PRw3@\ ;`йTlC9P 1mݜwd *V:TQf#Zְڵ$H* '*5,)SW}+aظ.v98}@@AR<ղr Q٦r֣,fw(쯱lpi\VŪnV6 mBRenxA0BWg)XGkE}mډ׾3)/Dk^i]o|<זoB^zB]{9XW/u0 wXa"nU,+[̛1k|c7.:vpVCF2~l 'ɥ+2d*lRnr<+c9[sņ3|f6_*jk<(yssl8V3gB)Ђn=C#:ы4F4+}iN3m5Nz;@Qԯ~UfLTԭ&5ueq]{3fzؽ3v<2|ϖ}X{$]mr+w˽nkY& o:4fwuס&@{o,4Ip/*I8W|UQ~'fw#oq+$"rI~rW+sr\)m~s\19}Cvэ~tZKywtG]ꦦzխ~Kg][uc'{>g]k?svoYs{ݟ|w]C{^' x'Ëg<xG>ݓ1-g̗X|-yЋX &}Mz^ff~}cV 8FC^7(~Q7͑Fޡ9UORIb@.}|wB hߎOC[|yx(k#5~7  #p8܃dApп: 4/ԙp"A,!4BDC7;'(4>ד*+2*-.,-010$3D4,3T6t7ʃ9:6<.=?|>A$BDDCTFtDcGIԺHKDLN|MPE>Q4SRDUdERVXWZE[]D\_F&`$b$A_4dTƏetFglgbiF^kZmFVP sD=@@'Ly {2zB@~ȄG* @Ht<Z@yȋ1ȏ)ɔ H@xHr8Ii@;A3 țX"ȠDJ]䝨ɇ?@"ȟʮ^>@ʳlHt˯DƷKIƹ˻)˽=˿LH4L7 T̯KtLkǔ̖ɴLLlQMA 4M2+TM#KtMkה,ٴMˋM*N 4)TγIt<ΥkNӁNR>2I@dOЬl<Ϫ0OͩX ϣ1jC¨D1 "PaIPfdIi  Qftx1GU77Ot%ўxxO%R "UM%8!$5(ҩ+,R./S1u25SD455eSt78Sͤ:;S=u>S@5A%T4CDUTdFGTIuJԳLMTOEPU$RSMȋ^دK%\c :}9х p s s=O}H3]JZHZ>B ]m>U`=ȇ腺U웻tR >[sM@m9_ !}>S`%heZJ;^ \K+}Ed[cݖS`_.&c?]:XPR[E0Z%IszJjB`#Jֺ]P `va; `Bza3˸֕&@5 L* hkګ(.I*.&+_cb`kdb}aȝ=:5 -Ybxc?c 5@t; 6Va/:8+elBe˸NFҍb>3Ab.fⷚ_(:e2:e>:@ˀۯ6ب_9{BYˀ**]F;Bva X˕O @gvZzg{նM:%$&毳a倃pL2m.a+hJ.$p޼R۞ƭ!, cFFBBBkkkC\Ty秧\\\ ꖖ333{{{䋋~~LLL%%%44tttccc+++oo΢>::: ^^}dd{##RRRtt靝%,,QQJJ衡du11殮Fi88׏`×H H*\ȰÇ#JHŋ3jȱǏ CIɓ(S\ɲ˗0cʜI͛8sɳϟ@ JѣH*]ʴӧPJJիXjʵׯ`ÊKV$ *ݩBpʝ+ * @|Hq0H((| ,^3 X AsM"mͻ]/ pe) pw1B HH9]s CkYu` %8\3Ǝ1wW`> ]2,wW5B*T_oAaA*\p̅R"UwA'ܕFa}ad~?j\A܌ `Aj TPk. .\v.L.G.p&zT_",@`P_wi@'Z `B7)#0BV@ )ફ5:ji±.˧ ^+|Ă:]r ,A{6W@9Z8)Am Ad%Cȗw5Xغȗ*w&I}y]x@,\*_ua/ jp~rQ 2. u2 @d$ehB@J0Hvf5DJujs} a@8mF%"  I!s&E "X3Jw9p sJ9uYYu^(`+B(M ;ȧ/>b`ߠ@H.ĵ i enԐ9>P2;̀p#X޶6ŧn_nH>b*W<"( `  0y u %<^ƄK*eWL6KH2S~6SctI<ĒK&9EF?mRfe-|fCIV*M4MẎTt;L :qDCfЗ1#5ՙJ @SHBB:u Q#gAILJ-+mOPlj;(>ITZ&%xPܢASFBt!hBr:X@175Q#/XT &Ay,>L{:9|jL_LMZ4 0w׻{5r%"K>E*F+Lm5ڔ-^dE $Qu%7I`/( u AKZnr0]ԺH.b4UЀfRN-k;96 b{kRֵpEnI<ت]Ďf I+B#5fsRC4J%SL--!-~MOX,VBV|T;*TfVPZ87'x%L2# @11DsÕIv+5S@\:4!#[Z$L91=]05!P؁ \wO\}*/J4 #,:+5i^z% @JR ۫Th!>EMHe3">6w\!A-nDN|`uMzR} R0fd(INO.(pUd'Nqϸ7QG`'Wr`(GfN8Ϲw@ЇNHOҗ;PԧN[KHRϳ6_ 5v}#.A#s#bpw3`qFu7M2y!@9{C@#HaTy/2{L6H˲{Gz?/H{|w_q賘>Cb}{HOaOgݻg?HV5aPq 8| x ؀ Az``Q%0(q؂]Qy `2H2x;v rb.8X25x4}{8D8w6WL񄴧Hb|qJXQC3: \@B8|gZ-*$~-R)Q !S}FG'S"~bT(`)f8;0ՄQbG4cT؋t(Y%%pE.PQ0R2VQ◌HFhЌb@i3jSfÍUqḏ4g8WT֣a?xȏF#(A05 [ ^*zI/_2 qTiZ,Y%AQncKFP#Qp m'^zrE;kTjaQ#0\kyad-QrzUy٘9Y|0ٙɘSٖ z{в19\)xb %99YƩ&͉V)w8]qٝiɩ4 уA Y]\9Yx^.t#p ڠъNJxuwR jvgsC$&͖kH; q1FX0ayx0CC!P9v$VzXZ\-p4Ң s*7X惞ѳ*);0jA5!-<@2aYx5prYyzt}MzRZzj!0کB(ei%iT"nڄ٢rriTieKi ^d,Yx`s':Cڬ:v- :ZzZ_*.@[ 'FPZ81[ ԟ䉎 jŠRhA7%ۅdF FԵđ۲(p""0k[@H 4a!^bkcp==CLX+Xh 5l3MP-XUMN%M9I-ckĪˠDנ >AܩuO+< 8sYX`*]k^}9i4ceDh"`LUM;Q۠*A)Ϋ"{KN~3@m?ԑab*=;vްpACV&23jӝ\|2 0A̓IK7L-B<+) / y$ublMP ^zDkq qP)ݔ1.H}QZ~6z5QMN}RJ3C^L5Ѭl`.2 EY>&s jn|m1X;0syAr~ Qqz n4`++/[q~^>s.Z A6^'긞xڴ>+@@:@`쐙p>.z>>4YPiE@No.w=N@Gi2n  5`6JNinO[f ?$OIA` 쾘RpI?U<";x^C`n6쌹P(O>?T'ũ Eߩ>GX3?rSlP !1oHӇ?%{qO_zMof?O*O0P _bܠCoeiaq{ G]̊_HE [L_//ߠ+J@_B *=okl6@@#DPB >QDQ x8 谑NJ-])L5ęSN=}TPEETҝm>eg$U"QQ$I(UHj ZmݾW\u:M׮ňf[n-1Wbƍ?YxBF ƲMFZj֭ZlӵKAhl4++F2[pōGvl#:`#F1sݽr Ûu:"Àǟ_~?,h(+cA1s C~+"$DOD )DA$ İ 2jSGtnE pq (F @:z,(1v\J d1) H=@F3WViiWq\r=a u?Znd=y`e%5FMbz"X'2׉.{{rݻ#S܈dP|7g6q3Y,$S{K$4N# R04$ T r`vM,YVt LB"8X_ӽo6&#XCNXD.Jы@a4 y_RZbF:%ɪ3">- uO^0 $d\H"ґ+?="}d&]- #BFp82Ҕs|d*CAi2e,eyIMҖdZ!$e"G0Lb*rTe2sG\6󖯔e49ͥ !- ()cv}#Lrt\0Rzӝ g<@QTwh@óf "?Pw ԡj` Q2 R 3>hGURԥ/7]$LuȘ8jNy:TF jRIT*ӨHUjTT~RjV+TUJZ+X:}5g*Yժ8խX]k\ַ5r+jW&5ϠW²E^ X"ֱR,c%VV?lfUjY~F;;Z|6](iU;*ֵ]mlOcֶmdhk[v3]n{;S׸r.q+6,]nt\*Ů?\7}'w׼oz_R/D^-Ŧw_FWoQ®TO;u` }\ v0̄&1h@"0U`qx@ 6-m :c)qeac?F‚!8Oܒ*$(`Ba(wYR^T`b#[NlD0=^s榾rƳT;s9Ҟw4tH|hF/$]t%}G4ғ+Kgӛv~==jڑ&M]GTԫb]jXPֳvuxk\Zeu_vuld7UرlhCKvv?umlc[^v[n$Mdnt[bwwƷ덟xw|o9w郻& x>{Gc~q;:xEnq ` Z葷|%g YrˌIcq=Gm5#h">u.QMCr_?'h`GdWͼ4>vmWNw?4򕳌 w{a1Y.a?/7oflc@ȧG=KV:zޭi[Ozkȁ?|W|H/~*/?}곯>}'~~Gϟ~+~S}?w|@$D@y;dn[@d{Y @O F?$5D';dA[AƓy{؀ йʫ9=DB|#5$Bg[x)^sЀ-1$5湢;B2TC ,,65CsxC;>=&2,cHL>E#3#cJD+ C4NT=DDES7D+DTEkH^\1C_$FzbDƍ:ddwZfSzh8jFlFnGlp$rDǺtdǴXvG8[Szs|wG{~vȁLǂ4rDȄ DždntȇƈjȊƋfȍLƎbɐő$YdX<ɖQpatɚddIIpCğ4ʛ$h J^<ʧ4XC${B|dOx\.z > Ou;O#LZ]|+@[Tв2PzгPP мIP  U *Q}ȹҳO 0*}Q|*Z3TPQ"5#мx{;/**U*+ *, ʛ1R<@$253E4U5e6u789:;<=>?@T5}K+CÄ`=8|O2k\T<0D$ET4ddccc +++ʔ霜$$%EEBBBq]]:RRRtt秧ꕕJJ44Cii衡似iH11=t沲y H*\ȰÇ#JHŋ3jȱǏ CIɓ(S\ɲ˗0cʜI͛8sɳϟ@ JѣH*]ʴӧPJJիXjʵׯ`ÊKV.*…pʝ+ |hb.H( >(t 0^(3\҆fj 4GѤ/8۸kKWBw#0l ="~sT q_(:˟?rw^GeqwW]&|B+A/THr݅az\I-En$ыh'iwI_r@: pp uDHq Q9E=x %wI9ИɐySs(L$r,(tB)5p Wsd6{8P&w0Uil4xWH Bؐ*]JiB &(j` y«r:*j jk l6@۬@ZijC|mJ% [.)[!| $A! 4  _M@`:"_0\l&:\.`5@Z*ؓ\X *wqe0E't]WWz4lUo?`BЃy||e|@C  0O?P13N4I&g!@2 }#?ki2A j(ƮXS 5\@&P1 :+Z~ 9 —Rbvh UĢvJ ԀsƯɈ{ ^CKGzf*bz:5&2o,iG0Z #AJ L%N$ƚ Y؟Nd`ZO!I  إN#L;^ƃ)9ٶH.3\Ar%$ya1,ݜ.CX 1Rb\ sxY 4(F8T@DS4xԠ <kcU2lg$@S #XN  p?^V)%4)!QbvH:0mSudL[8@jk)_QVXIji+1.4N/' B؜ e&ͨ?xT?Ry! Ǩiʳ5% QxF$rP׫c*ks 8 .{^jINBYR#/9QCn"x x+ U qD -QC!U],D8by],W =hnIJKxA:Y8`,v@d ^:eЪ&<:pN!jM%$dk~ U&2BC`pLP}O<!@?8AW|m4O &hay?HzUo%Ͻyaï|' o%@[ԗ*?yo}%BO}{䏿O~#!iLW#[v7W؀XٷyW Z-g"X%H~)8:88~(('1U/C({6!h'8 x(xG.0*h8hP(~8Ahwp!thpx a`fdž7hHi(oG0w(r8\HdXQ!'^h[ Ȅl񇈊QH?hq8shXhS ,ׇ8XU!H৊xghwshEGd!8$8h̸x}B{xj9~q/43/XY8hX Yx;{8q:<Ϣ `j ‘{89[x8 4 )7XQ&0s:~΂TB)D)y{,6859T970;IHb25 ROHb 􏡨M90O)3YtY)(PVR&~\cdKbkm>f(u9Y%1\I| (Ò0FA:fFHiK+٘MqKsY_H!eV5.< Bə[h(1i)qj}1,A]QЉ9DQxعٟjxY)5~[!# :,0󜋩i)ڡ.Y 6ms% "*j2چww) (F:Ѣ%3zHADiKTzդBZD*K\ڥMXg#`&W Yj^rj`J*X 1-q |moz9:(ȦCǧ}mXzׇh owJZj}j{QzB*zsxHѪ9`o.sڬAD8LJ{:ꢔ ZI aSqHJ~&xZZ0iJIej cw?z@ZZH=51T{%rpl(9p*$g𵣪$ `ǰ);qu>HJL۴N@` 4 ̃.@7iụ.@)= jqaeƁ,! T4Qxq<;>pjDu0;[;F (P屜> 6u H5RK6 Bz:r ;@Kʸ;[HkțʻۼΛQ+5`[p&F^YAe9I۹6Q7ػ4FJ{ڙJ0$"AXgbSL=`(k2 ~+~Õ˜PcC)KKaQ21Ww,Z٩#L;3AӚ"0/:pt7j%J<;#k[ r *!&PL _kkж/-!ÙxZpZ*Ǝ\"Z` =7;əv+bł̛^`Ȳ e^ qƊ/:GÇA9'D!,0ijlm=E`KyP.-RMn):Z\m ]Ah֒sԖ}!0prLmצy3P;3p'̮5QyA@M}d]!,֝mum Fٛ rMDr*ڐ{=!kS-'k*2/g3>H,#xA@$@4`!\d;sc߷BA4m'I@pm;] (F]mGZDe{5p&ӵ ku{E&X1od;+wdZٹ![۵_kH߫LIg^ B]C =ڑ 81D5r,h|ۺ2?qh9;-MK#BR炶9}ɒׇl;IR = y:F:+P+;gYOכa-56q RHٻ i>R؛8ZjJ*@^4i$V_d鍘H"e"JkKB$!B!0;yn>PI0 ZB$@>Nڦ m$LB&|(?3dʩY[otB5 %C RcլG~ ~" Nn#/?x-{}H\CsB8Vf)V|WF? 3dpĹUo@8}eĔ? -|H pgkfqwڕ{9vVy},#=8Yi\ ;яCR!B?߯. ܟwaO_}?h]>(QH2DB >QD-^ĘQD<~RG#3f9yH"WPjp&<@C_Œ͏.t agR6< 5zDBGH_T%!΢L5ZmAر^7W`… FXb+HxASL\cCd`ApABMFЂO #Եm}Z/c"=s E_#[ "Kqwo> | $ܥG^Ž/|@f);yAwd(/@QB$fT m@P;& jBƲh;6({/‰4D1Eٛˬ&H" Dtwx?z2I%KpE'ջ,CB`" o<΀2M5t/H* & 0GShI=\Y!3$P \3Ql:E2"DIO3d-GG%1^tD M:)@  @/-5Xa!u ؖ&c!<z駧z>{!C۞eM}݃XgHqd_|QsU%Џa"J r,πIW¦PЁ\=o$z 7"*kB~2@ @ Aq\Ї?3(; 1¸E?MЋ@aDc4(! B FA -aF6̐jtj\& 7F1~0.$6|Tf)4ya$UtG[e,eXP${[B1z($کp!A 9HV IXd G'fVeHݑb0)dAA"feMf>A- RNz"r̡UFH1`'o.Q09% zxD2::Qޓ/"~@ ȉ0``?;ZC%yKe: #r-gER#SfҷֵWY_xq?vD!$؆prm ] dJ,:TCU b-uAH.Pj3 Gmn5!_Cr)QAjօF9ƳWl\ qG6i^ih>Vљ3LҞNj@BW9(h)A 0` ȶk=3)i@q0@ 8\8@D)r+|hUZ֍/Jg7^Ksǩ _ٽ:ު6Dؤm}6\J5}!1 D[4>A3h^n o{Tq6 #gt vDHKAH0 SG:KC~K(<ÁI[ǎ$ɒ4ɓŒ0 -H2ɇHƉ )@IZ$Ȑʟʠ$JɄl ^tIpH Z̼T%ȝʮ,Iʟ8􈇴G̫QKqȒl 5_L˶*xK<|˿a0ɾħRLo|Ŏ#QdlH̔IC1 8 LlSMLei_ܬ6|* $TNS1NX|\NlCNNNA$N|?TMETNMHEU# TITE0˦Q%NZS!UCSu lTVսqX\]X_mZp~9 ՏS`eVʁligehmem= k-lMMn> BUitr6Uw] vwzu6y+3}|z׀%K؃]LE؆K5u؉,$}،M-؏Hْ-N 5YEYPٕ]ٖuCٚ=Yٚ\mYY٠ ڡ5ڥ5-Z]Z}ZZZZZZ|Z[|֯ ۲U-۴G2E[u۵Ҷ}۹[[ۺϻ۾ Y8C%@D%A=-\ee E4\U]Zq^MQ]ڽSOd hإ]ݘ{: ̨TERޞ|[]h}wb5V CMEfSM eQϙxH_}T_LfQ`| ` &aFa<f\d|a&̻wb% &bq"Fb:$fbY&by(a*Q,b5.0& 2Fc84fX6x8c˜:ctUl>V| cU`I `4̾hO^U~Ykeρ]UmRtkF{bl[mH'ߎ&nwpY_ pV g lV5G47]VFl&wq,q0nqqq#/ }#?rfrg''Mh(wr+ r, Ô=&M-d1` P%17ss9/ 0}m$2r:3' Ѐt>KaFwGHIJKLMNOPuMEfZp>ce ^uLfvHUGs6оuC_m^/}C'vQ/Ofwgha1DDd\ jݔtCkW,jOľTwm_ PER4d\Dž\*ͲkfkEWQh= z_{I}%xK`\m̵nkܬZSw PLB]yx mtyCxƏFKX]ƜLȽvePtHGyzzİ䏠G|]kg6KF7 Mq.lլe Hm^L8`jL7˯IP5/JoxTl0oIE6|H`[`gL%`ol^ɮ촷^wU\}7˧l~|^u]^g\˜VL *˨LC)mM+VD 2l!Ĉ'RHQ  3Z#ȐEQȔ*W`B&YҬiKLpr.LiQ(2m*[p)֬[U(ְb_j\T(ѵrk] hs.LhlÊWjh7! Sܫ7C0c}r.M3YN〄MV8au?W0C-z yŁ 4F ݜ2֯c}=۹wE&~7A=|܄NLZAPFzUZz&/0x!j(!`xfx")"-"18#5x#9#=#A 9$Ey$r!, cLLL>댌kkkd333 [{{{~~\\\SSBBB,,%%%殮ddtttccc ++++DDʔ$$:::霜%oo]]RRR::tt=秧JJuꖖ44hhC蠠9 oF-יs111沲tH] H*\ȰÇ#JHŋ3jȱǏ CIɓ(S\ɲ˗0cʜI͛8sɳϟ@ JѣH*]ʴӧPJJիXjʵׯ`ÊKV$ -٢Epʝ+ |X" ,H( <( 1^3ӨE a_ AsM"mͻ]; pe+ pw1 HH9]s CkYx틮<EcǘI+ƭwW _vwA [M@pP t!̅R#p ,H\FܵB"ntэ'PywI`sy@BdpW eDHr qe5D`$pWLޥ@lT'Cw &ICyn 䠄%qPIq( d!L xCxY 7tmJ(|DgZujyJ +Ub5j ) 3*P0 Jj3ju:e@YC $@BY@5$ר\[w[{_`ho 4 f _堯`N+#_p\|,'Xɩu`3$`Y@@]΂Ĺrx"WFm!/AR fU*p¬|\5 vUEr A=ȍiY͗ FW!P4hh 5tY@-~ 1@~|ǥ)k9#T9U0yw1]%Q @(P]  jyb~oQUF%P@+:B872  xC";zoe<1P 803f I@7~ @lrAxP-9uk]1H;ZPG닢ҾVgF  7~ A@ 6E1~S6J P~H7O+*H|D A E&qt4Y<'SҔR rLޗ@d&HZޤ}Az X$/AĭEC,:{AL-Ǚ M4ƹBjƃ%|Ci}){DMa3 _ĀXEK9٦ 9_ҖMhK~?S0?Dq2yMO&IIqwz'A∆8_2 Ƃ!dK]HS %=^$MzƦt%>7Y(c:BU3j({VbIOm(х,Sq ]g `@c! OxG`uU3wB:y4Ls@xWtu& m/"Զ6C]ov mi;dȎMzel~ۖ[RO UqsZ'Nq2k7񎃥GNĀ*WD`$xgNR(H@ЇNHOҗ;PԧN[XϺַ{`EغPgG>(} köHwQ{w!9@Jw?G4>'@g<7w00P1NckjG}!B `Sw3{g~[|Kt̫%[ml,s\DŽ|O9wzDܢq\Ȕ,oxŢn >;ȕʥ{ɎŃ{+ʮlɧiA +ɟʺĤ y9|̰lK 24{|ߙ;0=#z%ԝ ?;_jfDAKh-x%@yi5lB!4a!q@n%~Ƈ o<# |->>> M,-P40mDtɏACh1ZRyQi !Q:6~>~f=x'.а=0H.?L{[$]ٶܛ !3N}T,x4Ch& '.?|=/b,c' M]c*ڹ<ް[e<* =61o"?ĒaR[61C]&Uj/JƠ w T8ң E-n0ZeLZѣDJgp>=pɷ2@³C3l-8#NIi+6RLڱ6-آǻI~4t$Ox\2ҵmƝ'}cF ɻ8-0C%tUi\vy%FQÉ+[^If%)^FHG_(j`dAR; ',/l0PXJaNL&)C_Q0":ϡ&{N;"4/H+6 XJ+s0F-jÚJ$a:#L<2TL;ǒ vL"!P􁶓p KN%тZ3T2IG7645Tx'xG>yG^v7 ]qCYʴ&ܠw=|U'R;hp@b?Dww:kҽEW!ACpI$ ~"> @^F L!C>s,_MZ$Ca e8û (<pЪl ATfijNy$v yCF@.ьgCT"r _ b\]oR2)2gJ,hQ(4IO|иINІCa֬2 E?V$䆏wSR'piL"M,WAfzT’'!Df2IJ`)Ap@ zbW*pE:Uė[d U=){ʄg<Pp e;H&jAs3e@䜡J$M' O@<5Q1+z6rbǟ9W}4A !H)O&Ґ0Q;H%Ig~Q8X JNcԫ_ Pv-,17DpIۦc+UX]5]^y,ՒwհEX#'TU6@5A,@$v92=af+`%V{wׁ֓ IĶ֫kȃVz!8A S)iTrt@yBș61Z* '?msqFnj+1+Uu0`qoC < ZbEwQ~2XP֫wů~Zc H o6:LS]0A|8>.+̗ cp$rKbS~$_9{9"~E[BYbV~莭cMI᪑Ŝ8$S ۉ dDn@|nZFiA=RZQY"!e|^ 燌p̪IK| AZeN3rS2 !wgF$5Jq|t]m'j~-7!Lx]f[[?9zn]>Zvϯ%M"Ji D);tYT'8 B E5# :q[2 φݗM冁 Áj7~f] RU|p\y@d<mf2H+C.Mz#|Z#Iww)Wy$Lz *$blwB87:9ŀ2dEH!}k֭uX[d:&u}:gU 4#iFwPiְ[l+[ٿR9l,RYj+Ee2Hu{>t 6̳ʣ̔}UsAcq"mM^vMf6I:gFs4_8EKCakIe=;< 2;p.`?@h:#ƒVr1;"Ih0@*@ ay $ųty { B |ȻK%)&߳>69;.tL&؞c踕,3DCA5;"Eҍ&4@A+3ͱp 2CDDCp}*r%9A:*Ʊ 3L\=>|8E9ȂېD?>HK(S0DOPd.:KkEҿ h ؑ̃ +@\]0B(1aܱ'4!3vZ>h @ A<ymԲb'DC\s`!3N H/9wx*EGqoF;+ Lf[,  Ⱐ 耴wӥHG-C$‘ĝ5|Ɠ\ XȢpalI(B@7 b*LD 4IރPIXJ{*Ɍ伎l3 4!Iʒ"s  ( ˇDKyBKmJWJ#Jɠ˺ɤrǁ L54͚,LOӴ LLDŽMǔؤٜMB P$ MϤMáɠ ɴڴNۤ% ̿L䌭OTMT#AcN销 ȋԚ, OM4OA1O豁$ O\,fQMa45AHYsHI@$%5Ɂ PPʹѠD];Q UYLE$͐&ōORp U*R&uTQ5*/Q,1.3ESSN2e2/M85m:MS9SS;ӯ3=A]-S\FBUT9CD]HE>uT2TIT+mJKOEMԓTP5QET<=V=TUU,,UW9XY]\eU^0_@]dub5V VeVfe7dkmEf֚VlVmnr=mc#WseW t gmyHtW1[Wz{{ |׀u~XEXzׂuAyM؇=W_euXXw%؉}؎X:X%Yِ'fӒuY0YXEO2OiٗYX)BUɲ١]ғY4 q!%ڦ؝=x AuګUXZOPZ-ڭz ۴5Y%[32[U۸DCm۾HMۿ][[qN[e5ƥ܈[ȝU \\ҭ ݀#5֝=M]fYץ]ؕ]݅Y]+]^=5]5^D^ޛۥ^^ -8_!_e_}Չm_^_]~1߈ TF`_v` Zmv`.;=` ` $֍ a 6`favaܵfa !"6 FbU%&'*NS*-ص!P\.6c)SX d]3c-A&AG:1>:d\K E֕A?i) TdFbAXIKed X}H UQ$f2ehOfZ&OW>^[[͵aO1 ش2&gNJ%]h氅0n[ɁK gr9t.YmdugyE{=}g4* 6h-VMvl臖艶h{h9h 6i-+ViJvlj闖i 陶iwi\; 6)Vj IvjjkꩶՋ j6"56^He2fnkdK뿶큢 gFmNJ^>lȶl&tl PFmఘ弥HkN~c> l7uKߦFn - mmpl虀͆HFgn q!Xxn>nv^7Fgv_   {'qu6WkfqbqVKgdDn!lfj0dFh#r`*+Ol` r mPa>s P2~^ (sھ$gmpI7t(/FNqI#EtKnLgts'(?WSk*u+guXY[?\^_aOb7vFdegvugohvjkvmnvpq'w5stWwev7wwyz|}wWۀ fi%5xJxG툟O҅?)'%7Nux>*Fғ `ryЊ6dy z0бRѢiG裷 uzz3Ae3Я_w-J?'7GWgwLJ|t^FplcFܤ`rgm|k`nn8tkmNʴWدkPzǓt\ljWP0F1F"Q P ~\٣~v7 G@HdHT5n7 s?rυ Xo٣&dc7 $h(h „H(B'Rh"ƌ5j Ѣ$l,i$ʔs B%̘2g"H h(ғ9z !†FLj*¥M@ȑ(S8C,ִjxiprצQܼI#TJAzDA 93FIB51抏#O6*'o&t E:# }q ۺ1v$Yаݨ+D!i֦7&nsAՠ@BAؙs7Xc h +А"C@ɁDn}7=ut}F0PB + fTE-](`P M"a 9Z*Jm`~06Wh^?ε@uy( t]ix/VyI , [5 =њj-AbgRE v'O])i9Uk& OA )@h$f-L!4*jbRI0wP SG*S G t+LlL0*1kWfdh2pYj[eK碤 @ {Rd6/G~nVe3P(^C~*lo4 0co @ 6qW\| _q) BSϾl}Mܔ;O) 7\tAT ;1rB7 2sS=Q9 4 ME`6q=7u}7y7}7 >8g@!, cLLL;;댌kkkoo333[ {{{\\\SS:::,,%%%殮>tttpTuIu%(#FG ePLPO@TI~9%AٺEQM>ǁXB(<\ԨpA&cPʮe,h ikiǁʼnBm//l4+S.Pa_0883G_r@tSW >ɩˆ:&hH:ԇw|bkgyE>ew MU?yE>7Ko`0z\tֿ{pt}\Џ}Oؿ {>O\z7lS߷OG!'@Ώ7wrX}r&gu eAg{ #[*0h "{+{ P@8@XaGfC{ <DCThH„ PWScY[M`؆=|ShB(C#:;I')s.$~--; xRQqxEVxA'؎1FdFFˡ?T^*T[1ّ.H2(2 75 \*^y)!$H2˕`2" 3yN971ol"6LYONaXD n崙f\9?^i F> diytis?kY:%nwY |?r_ x٘=QQxć%瘚]xG|鷙IyW 3/qQAzah;p>dt'!3КqIQWHG)gYfHh %~Bֹ:ȝ[99+yQɞ)jC-*Y,muxl `ч;`+>1`&aR9z P㹟(Т.02:0Ms- q=)7`{YI׳. *>@jUq-#0< N !zn5)P)Bq)vi0pr:tZv:Jp(09XJ ЗÜdLv 2#`DWs$ɦlJ0Zzj))`vPyQ$Ry\Ȍ} hZWڬk8>I$31W"źQIy"33Yꪞꩅ ,A:#(3īxBXRZb1j) ڮ(D|Z3ѐ #<2#`MaX ;#}rO3!%y+!%PM!FQ*0-,벟;"-K8uB=G:bEYCah M*,۲T[ DFA2:o qsymˮ۩kɗ ڸY r'A벞HǹPhK[ꊝkj:~}!u ;;ׇ5PGߛATFȻʻ;u{֫8qwJX ;kI {uKY\+rK \+| p65  k; @,IJlp#9S5Po:5p 419پ`ᙽH`Ppn|P`f̚oM1pptvH/hG?J)Ge;`ŚNDɛ:"`Cr020hZÖڻbOfh`Vh/0R`F,CǷ[˱LGŀ axKڤ3`'ȉ @5$ Y[J]Zܣ! WtZ擠}zmBJ 7hPMj/AdrOڝI} EEy;><$@00`Q@}M:N<`_m}G\M{QN+;{* +I+Tݼ> 0K皒&Mqs}j `4 dNFJ Ϯ. ܭbjk  ]36>~~>yPH]!@D^=HSQD$ć^ ?jb_EQr7'.M_}0@ls%ܖ 8P(*BO0! bCqh0 ( PP 9RG! b)'RdRޞ2J)J+2K-K/,0@6 끲0&N;j 5%0` ЎD B(4sPK4ϽTK2SO?5TQ􉪷@)KXԪ$ <;o+M=L ƺhȀB)5 sD,bCtj6[lM[$7\B)MR5V`¬^]K hhCsNjlE#Y/iŽ/FIE8bH+H/8c?8GTBWlBxСׄ4^)_/=.z~ zaBM|⧯:kIe2, b y߶o}mbЋT 5v˯,pV[q#|r+rNe67jB86%pXoH @rzq=}G>ys>'*S s"m@zMO4׿}_~azbh#$v;Dߞu`"cy?b珂P>VS[B{>SLxT[B6%7BpT3J$*VBCړ?0_/T&ozsЦ%WX0hhB&6щ;AaQTn vW r LD@?MB3% g YyB'юwls /qIse,b ÷{1;dE"GJVҒ ui,cȺ4j Ah9r);UCbi 9Y%u˧eu:A- &/d\)SRG&"(ѮejrW9|IHQқܚ/I-AG+ SX-Ռ^3;, f9&٠ H8hGE*xN uP=" 6! ֠P7S`MjZ R\C"2}!nGڑ?+uElIڪeRNIɫ-dEe4X 5AkbE&bihqŜ!6`Y)ZpWByhַ͎-H(FhS;Sؔ`Fi;̳-_]K7]kY@q,1.sր,6Td.B[^kj]8ZaƼS`&/9X vdA'dB2{ + B$(  +B0e&C'>+),L(HB|0 2 ̐B4Λ hlàB8)aAFC"C>D?L? !P  6t-tCGD=QVC < BLDNDPEQ^ODZ xC8lEqgEa3\$n F!Fm;CŚ ppfUwWLCZ,oj=M$ 0ʍ4K4GBÔwtȜL|M1DM,M͝ ΉM4NhNdNLN+F|ƋN0[gN$ ϏT;OOl^KӛOOO|P%O0EP=+P;u[jPΊP LP! Qz5щQ*eQ/JZQyѴj@īMC sL%ҭё:)t"uR=ҁY!}+(R(3R,ӢR^S"` 41ե:S5dJr/#sqQBEIQEկA CME)K1L?N TLT͗RS UUbUXU[qUA^VUaE֞,OduֱZ֫9gVҌVqjVxpmn`rqeHr5u&t oejuwUxN~{ל&~M؁%0؃E؄}e؈.؆ض؋،؎ثؐ5Gǜ M=YjMٖٖ֗}Ypl֛vYr}Y ZsX-uuؓ BȤDW_h ($lҩ[|V_Bv ڠ`ٰرX#,Z[eU_r T=WE--h\UBHDť\жz@TWdTG &\յ '4T4_+\}LZ^襀UuL(q^[^ E 5_,U_Lu_ylV_:_  6+V`zKv`-k`ۊ ` `W `/a 6a)V[Iva-ak Ε ! !&#f #F%F]a& ؀ Zߍ,&%M10  B,&6&s`c7c :>faA^?&dY9."BfdY* HTFd@ڸQvdhTG]XdcRv`^BVNW e_䩠D8d`V渜)h]Q^j]dkZngTp&g_r&g%FVu>vvxy{|~~>߂6EޅfhuVވh芾݋h܎^ܑ&i5链۔Vieۗi],di(n+&i7^mi>⓺h `Vv,.SP&tj8tdLfFd@5Dc%n]T@Uij(ΤV@ (ls>츆6jkl ^,n~am6m~EԶfmvu6ؖmpmjvmd&n^5VnXeNnQI><4%F+Uvvo"oÞ'0.o H0oo:p/pXgv0Lpp ^ iFSp 77q=q oqqou4[fvqqh}qx%P#_&%_ _ (}kwrO,dž&E^/!273G4W5g6w789:;G5!PBAO6].~.^(fJag~/#VR7SGTWuM~ǦMHG GxFѢp?& vuY D@d@WڬGvVj1wlv f: R`/~&,ZZng MwIJ M[Uj~G5V =wGfpD% Jk04xڥ{CD|lUTjvŭeRgd@bA*ex&TyxxX㾥V[ƨkwhH4)4I.Ry7y[YFEy4pNglPV lJwETfB3u7DH휬ϊ5,DUV|\yoİkCBwEuFwü_7{ YF\CfkyÏhZg|ר ? 8 x *P@!, cLLL;;댌kkknn333[ {{{\\\~~SS:::,,%%%殮>tttcccdd+++ <霜## m%DDBBB]]RRRttꖖu秧JJ{44ii蠠CtiבH, q11_沲yݒ=y H*\ȰÇ#JHŋ3jȱǏ CIɓ(S\ɲ˗0cʜI͛8sɳϟ@ JѣH*]ʴӧPJJիXjʵׯ`ÊK$ ,Cpʝ+T |XB .P( … >( 3^@F3^"!_ A+M۸Kp_Wb5 w'0\@D]{h" 5jo >g+0hWDGAA[)Ag0Dx]$^V]0̡wpKuha|0(H t_\A7p@5"pp uADeHq$5 ayPv 1dDS]=Tɢ iOVgQhw(cr+Cw hN0[m1pY s`$x DUWD B:P"d$j!DB $$Tj` yݪb)q^0jAƠ |1*-kn֮wJ=Bp'^ ;^B5 ABv:$Pʦȗ, @ǁmgqŐ Dp"B!W62l1o<ޥ妧EEX{+]ρLS<Vt-6X5A‘=A"Du`"wQ  C @Ejq <+p wu0 TyW1_*-N@w=.\Lgm  x0{tF}^zn"|5o* @=Wհ`5']+lAC+$u@0[I5^yJT jeTm7+C7'Sh T'dCȦTQ@WB DWN F."@׽*߻AtTTi $ A"E8}5t1Р_Z}:Tj`A2!dtlw? 6 ϭ=щ o;]Nz.ɔy'?/ۛbT$jvA2'Įmj@D749$ǴQ! /R:B6: t*Ԡ`AO8 1Ҕ3=<ƕbB~k $DyHHP K,cR]n֐Vԑ#ņ>ۄGG^$$e{R}$/ xz J$5%<]Rx1 Vq ޶6+Ck1pR S6LuM@H}AdBT kζ,J*|G8%ATE|֫TKE \>V c|bȲI5@BRj!0d![ԻQfMbk=~"cWeЀM&R'{"AҶZm=H D n۝ )7XwSN(-F)"s;+3 AN[)'A/{'BE(OW0gN8Ϲw@ЇNHO:/wz&UTPK_:#[CA=o7u6x~.ua۹a;\$ʯCdP0pC+DI~Y`*;Wbu+G9jtoIyو"fvy-=?{ѐm:D0c8*=@g:X^ Yw@zP ڠ  ;C,?(2y%uWI,)=`C١2$ɟ :Fj YL R:TZVzTI3$ZL2gH"#'\:9c4gB,+b wrD Mh@Zzz*1p0&'E(%dHJiA v Zvs?򨁒&v%gbya¦y2!.@|B2jAWIz_!: jʟ K3F|b:Bױ1᫐YsD ʏ> α:"H<= IhZ/W$pB'!"p& [ڰ%R#Q"i3$(4Sʚ 35+Ji?y0еR;Zt'A~jٟz0v)lnv2Biw5;+QQa {~s˸!mK۹| `VI{ wz;b ۻۻ˻{ +yIP5`Z)4[Yٻ˺Hi+ @jZpȫi8X::д˽x{|Kl YARSj)f;VÛ/ѿ>lI6̟<۪A& <Z5PP:5pٜz@d ?H4,`Q+B\TkuQzQWd4YYKfR>)({'V9ԅ,!pC2& c-,k=tg\þ9e?|ĭI)ETs})ϫKG1 ̫Էʁ} ]smR|Ǿȫ aK{Lgijg":Ąf"!070zEXϋG |H>@MNRǝ-HFkݺKM]aͭʜi:ZҼ1z0@n7E1$ Hߗ !rKݓ }ϊ ilƝ `-0P`Q!0G@ϝҽ+U <`-(28VW)dMXv EKrWYv'dV>X^4. ۍÓ$.#-8(v.R>T ">ݵ PY)"&焮-n?Eùͼ 9W)"dM+r .%"u2s0c/b?]Q4.q೼ϻ&-`Lv:"ðKۭH@/.˶1 1{[\`?xOm][n"cOK/`-+ . L0-p0w=|p >@߿񏺅7pGO\u"b kQPD-^H@F=~RH%MD%F%A# <(FPEYDz + [@B TR0ZqĤ2_fKحc46ZlIW\uz$qJ=HK {k۵w?Pj-4N9 0D8ٲ[.IVFhGI/\pÍ_$=%o͝+YtG<0I*(a :*sݿ/(׉g)|؏*`(PҲoBmB 7/?1DG$DOD1EWdE(*@ d7(zA"䰢DR*S@s8+ ٫#A+I' 4t3N8ZC63O=OT`FZzH; ZHYJRKW )@!I \?}PTAJ`" TV[oe U\wըҴ .C< iF(rKv#(ˢU%'tuS[,*2K5^yy-^5|շ_~7`xNaA" "'!قJͪZ#&U rBa&t֙j޵,fwgG(OE,,-,kc#9 3`dBS rS^\NS'z(߆49瞓nn{o|=K"Z<3l׀vmjl:i-hےr"Z6zf-j_=v;:u"hȵRv3mrG꘾GW`CF>IOق3i`m_=|'G=Zc*|ى;~e2[2x\S3sI¹5Ё'YkW Ԣ!t)짱i sVU5?0YZ*N5*`uCM0$ȼO{8+5 #!^=D _{iy8F2|Ӊ~MH[-oO5=j@{8HBr(>Kۥ4N+˽%)d'I*F DDe*ŇHZDt@VU0{$.WRD/-Jf63̥ '4`&j hTYփ5+z&]'C $9Oz)YVֲi)KWl噧tB抾ʙsӢ=1eu.y4]$J+b#4oxj(ujƋ7$FuS=i$;D!sli FrƘ0)SvC>5xc5 =$?0@^E#Y#?/0P) =կakIA Wa%F/Jb]/bIZ Wv"^VrQ@lz IZJ3K[zֶ-tH E`IP:"Z$JήQ6_z[4"@i#ЀuV մPs\D&;z^XU^3%Y "@. K,ptѥ6{-7mQDq:fB;̸+pA4N qj Aee#gRv]D/[hRs:Vkjb+d]jg9 O^pZ4 XJP`FPe;TO.c͸p0+v ."6vs4۪6ol:3d34 T`Y1۾wր{Fwl0\(Y584!J5>8"V5h ͲeUuK4ZQj`ÞNpge7~70oxWo αN+@Y=AC[ _A@L]㣆>>:,Lbzs%oQMBѴ9GvtTtEDohUϒS-ba<ۖ`yss`׉z=z[Qȗ7 rXB)G_;ppx#${`՟0l%lj.WS 7aK8yb5c@Ui/з) sf=Ah? h&(Ө H 4)U>Ha$BDSA &- ,{B C%"B?$[>({AT"94 '*H?4{Z)7$BD4#,? ?C&  KDBfRO\D;l/P 0I(h 0D4#PE!cD>,C ; @Uþb= WKEN̡[F\E.=<Œ8Ca b4r4G ;nT[vdHrEXDXƧ{G3�n=AB#s,t|GwȇH;4G Yǐ~hm4_LTb=ՊȚLɸkI'``I0TʑFQ3OyJb}QQu'@buRfkA},L1&0(RUS1E$2E3M75e tS8ӹ9m:>U\ܴ͓ ]UU]Dӭ]u=]:]]|ݐ]\ ]}-E^S^ue =^}^U[ ^5]-mEߋ _5}__)E3R`65EvUf `D` .^]-&Ef vΑ V 4JᰲaY' BfF{$uCm+ S@LA$̴b,c {b܈)4N6ͭX4fc77^[BJh ;d2N%D'-V8Z XC[EIb1ڰ ؀ K&eRveDĒ0W'&XKD_ @\&a&^Q.fvhNOhjFNjlKlnInpFprDr6tBtVv>Avvx?xz=z|7|~2~v-n(6$V!vNe.>茾>萶钖 6锾 V閖 vH~h霞h~# %ne6XA8df>ꨖ{cЈĢ꭮Af&Jɭ.뵎jêk&AT듎  ڽ&l کd[.leeЁleD̮'Zvlf&픘hïkֶ Xeվ:mnFnVvnn{nwVos6ooFfokvog.\~R'pE6W;fw1 & a0mbq+8O@q` q֨@)D ?j:l$FC6rvdnEֈ&,/۰ڰs!GZTsos+;h=s>@s6DEgtuG_HJKtM/NtPQ'5SWTWeVWʕYwZ\]_`%bcGvUefwvhiv@E5n3o Nrj6fqWtKpgHݖ9vEtww w~_w0qxk~,_x/ HfOx/bxdx ywxh 7?ʑ?苯x/ hXBGwoy/ hImpgwz~^kil0qwiulKL{OI>mrOiZހnC6i@gwLJO sF_J XJ[؏AA I"&#Fq}4oqާ 0n[bO\L,#Pc+|jbbGwj6?jXbT+~XÍhƷ „ 2l!Ĉ' LѣB 7ra "#ʔ*U P +gҬyPF \"L  l-JNWd$Y 0jF'kC+`k,(в= s#EY! N%Q.w*o׍Hh@q3o*2E+؀DhHCTkUӆcnjic]x!eaP6 <e&H ũ'Qa R-q6yGgW|TYguxw;LX{ރf}mv9Xa`@ӕHA= az+%s Va0E]x$zD!&$ZLS))$W'\Z KD]i4$NB@ nzz5X~Z߄} +xta)iJC BS#ZU2@3Н[х ]HJ:T tjk(8묎ɮ% b: v хkS[ز= G0r[xWYD o߅k ?4>*A/x}p댌kkk_333 {{{3\\\}}WKK,,BBB%%%͛歭tttddccc^^ o+++##ʔ;;霜%:::ooSSRRRzttDDCꕕ秧~~{44Shhj蠠幹;k*H_Ùt~11[<ט~ H*\ȰÇ#JHŋ3jȱǏ CIɓ(S\ɲ˗0cʜI͛8sɳϟ@ JѣH*]ʴӧPJJիXjʵׯ`ÊKV0 pʝ+ |xbA ,H( >( 23^3QӨQ_ء@sM"mͻ]7 pe/ pw1 HH9]s Ck)<EcǘI+ƽw_vw [Q@p l!̅R%⊏`F !}adc~GYV_Wyq:ߵ]Y;S5>2MBwB!P7hh` @%tR@`]ko5C-*_# BAmPgNU@958|INjAV~0 Tǟ9V] j}ζe3+ű`.j] B@MC"~N t@2 l  }pH$`gVͣN6%r(p3o*یJ?B%\ZwQ@ 1{폂` @6  8 GA.QğyPlbBE>T>Yup&NJ@`Yi UBBR41vR 5o$<@6;lKDM@M& G ML*o± =( `ځE/ ;#+T@s/8b6!\@Ji=#M !lCWi#T1)TSJM.$q~s%h)F :h<2SB$ݸ;Ks KqDCg 9ʃ Hu=ODdG$azf)yÕЀ)ĞyCdp@m?JՑܒ`d4hN-$= DLM!<))f'oH|!p=(|)'u tlI0ersT8 Ʈ{ͪ%(Km@E<ՅĀJRhcF $xx,T MH-IRsA)-ȯ Ih Uskb.-|P܏?Ng+ TwzSS۝Y \MtyH]ӚD[Ҫ`գVJ rRbG"Aåp[JgI,/Nj/Yt%ě5HqMEh,FV"{1?@s JmPz⁼.1_.+h1%<o.•vݩ%*Hdn%h݊/yŗDqIYm}Ưgx^o͉-20m i: DW`F5%C4:LZ˄hWIv]Ys! J^ >vKM;PN  X"D^Hmn{ .pN*7:hMz /d~r#NԚ;QA [?!x {,T 6(OW0gN8Ϲw@ЇNHO}~ %TԥWEvPݖmXQ{4ZR*t`O~]S۽7< x~7e/q%߰7@ ^@P(X H@?' 6xX:x, -W7xRz)HK?N";e1aHxM{K؃j#~ޤ/0axcZAaXH1{[QYI 0jч8np(8t;8o2_Z@2mĆXшx>h43.rGh(!ӆ؋k=؄X2&~!^cl#%X9Ȋ @ia:SYf@ x#ȅ+MEL?c㏴x9h(B75Zi4 ia*5~a["A*@+i897yن\x;?Y5Y;F 7lwRE 7J'VyVSᕫw|#P&`ٖ - xpizr0zP|yY8GoG}3G{/1 9}9f~+yy-H51T],GI(lB8,WIP ()!9OV*^wI䩓4e#C]W^륏g+>3]w&yQ=jٝ %K 9HICڡBp 0I#-<)7yzX@(h5 P!-:; 2 /)0S SiK0^`b:dL (YQ`%gTr1 5h6dg"- Ž/Mڝ)*Q'L0ZzQTjRtP`yىHڪډ4A ^ya3!Bv%hhJ>(3ʙj!ѭ⚙/.QBPD?2%F:p 6Z~êiJ%:i+cY+20Tsdӑ *x*$! ,/q$PD=jţ%k>۝!:RrbC.h(;(DsjT8Zz:y@iZ!pڤ+cKi9z~⚸^IyuY Ghf7Mښ;yuDKF [w[Fy뫲;~ۺ!˻zL1P˫gHkˠYzLP[I.! {˝Z 7_ڼ`Zyș5\9 ܹM V`~^ѥ}< |ܝ\^z^ 1@؟h 3>b0Ӟw;i۪)>DK dؒ^;B] SaK .*{W72: > &b9$iЋM7 P'.i"kA,dW^GZ]{:EE:8M(- }ݭʵRݿ,`;ۼR{d-7,sڶ5p/l Y1O,s}_XXzq`@%QnD`W48@Q.D4 m ^s[r )PpY@y{P-jҗ@9%x@ѡ"(mM׎RE,q;_ = ]+pK:aJkͺP핋qamUVP&`ϸ.DD`֣kq8 (i4nvt\ ƉKN褫nOx q/x>`rҭ D0p|nYΊS_*>`W`aW[V^mhw&o40&W]x)jk*+\Vk:ZJ%˼^*M,[.b Z'W` P:o/yq 2@PB  Tb+`q .(!H+=*4RJ-]SL0Oĉ",`1ÁAZ(9>U*%RTl h('kEVZjW\u:P ti7aG%Izц΅ĊLkm-_Ɯ`SzQG L⡫T4. _ L' 3",R1$H#e!k&t2J(|J)J-K/; z|$h0P#+ȪH;3 Bca1ԬG~,‹8H;-2LI#4LK'RL7Ք?-3CQ ڌX5F_Eҵ"1F c@\'eYg"QWz! ij4.co'l[YX+BEȺ6޳]n7_ C-3ڔK=[RDw""޽0aR౼+uPbUW9ַd[6MA lK(v7Of tsEڲ = )ĻLC"N`:ödcViF7+^# *re{ #/m? \/V;'= -oz/sMv+7c7֒ bc}rkˀ`#RR R4 OkLc\DM܈\gaO@y 2A fw'N78E7M]4d$L;"sBt+a@ Rɜg#b@jbx2؍g7Z˷e@mfyT9Z(VψcnS 9LYBD浈cSaX0W9Nl eYŷ} @y |(D;|FӴ pawim Gn}O{W Aq2-`5(8Hjx7ݙG]*ĥ! v<ڂ-Z( hЪW6"}WT7pNЄ==6=p(޵,8練BuT ⢈]r1Wh\܈+-=?*AJoS[lS.$uIvKNި{v߬tܓZ# Mn;w t=rR[{|K]/3fy=S^ ;g0"'yɛK wUu!>u,Ew=N7l]RRp-u[$;9B%0 ^Yk9N27~Bb/Zs>y%AP > QP%Q Q U3;LQjё|QbRK!ER*R)<$u\Rl'ĉ*RĻRL #1. ?= `#o$6e03=)?5D?72Sԃ 05AE=m̳؁i>Bu1T3d;}KeH5;R5 @P5:M:NTQEՙ!ՠc#XW]HNTZc#? SU`[%ֲh3U+e%cJVVfVXtgm]kdVnqE^*;tuWXWs wWjzW }eם׀5&Xy*؃eXKت׆Xx}،؉XkIZ،=؍ͥ؏پؑEYG,;ٔuY\Y܍i}YtY YxٝZ~١٠5Zk٣5ڤeZuZ-ڨZZn\ګmڬZZ|۰۱ƭ5ۣE[A)۵%ٶ۸[ۺۻe۽ۺP8C1ۿ \p10>M\]ܹp10BH׳}[+ElTe]#] (ƂT]}]] xٽD@BT^}%Q3 eUVX5 ^^m]^ ; ^u$\(_ ^]Ú _WXxK>` "}O-`m_`]x ZK `=[6 S_AKJH FH "F"6$vE$V&au(fK&*6@*,>,.=.0v:02Na(64^E2V6F)6v8"8:N:<<>>@@BB6DfDVF6) UEG1 < P2 JU/0xUfZ~c e[9^ דaഐ]&en=Hd6_f暠s W`NxJk&bn ^U%H?a.w^;kՅ{0D_qU|hyq1 @f`׃ew腎>x`_ae^ Ufhai^鑮6UFfiKvn䘖D.@<꠮&86nV4f.㧆0⪶,ꬮ(n$&.F Vv6P]k&5xfi=(kfh=[\&l{6lxNI gp욆2v;jf ^q.n.?S&׾l@g_Wkm^.n^ 6`ۦnvnn&5nVoeNooofp%GUwwp p p7q7qzEwgqtuqej'7%5> gO7qqHr_X(& g#!)!*G +". s1'N3 RCFsyf.s3 "\cls:w<9 hi(%s9?tЁ5M^EDGBL7g# 6SGTWUgVwWXYZ[\]^_`am^o6N>_kfh0[vi_hml_vJS rh@uwwxyǴ[; _AMÅDC CPӕw3W }N.<$t\ȕ5Cn llR@ 'wU;y{?Ǎ%y]ͳm$\]l&Dd܋WANE6 :ETIܼ֭AޝoF n5@%`nQ(^ǒ' +v Nbz.P _D0횀*fZr쁐Q(` fM488³]_x[{`v &9v/rG>SЧ}g+KvPAGBK p:gZݡ{AUnW_ m$%څzu "U_ʱ>ڃ/zEAX*]uG3#P"2$Q5JYU^)W[z%a9&ey&i&m&q9'uy'y'}!, LLLd>댌kkk_333 {{{3\\\}}WKKBBB,,%%%秧tttddcccSS+++ o##;;%:::幹霜oo]]RRRuttCCCꕕ歭~~{44Shh衡j͈;kH_*t沲~11zݎ[<~ H*\ȰÇ#JHŋ3jȱǏ CIɓ(S\ɲ˗0cʜI͛8sɳϟ@ JѣH*]ʴӧPJJիXjʵׯ`ÊK/Xⅈpʝ+ @|hB *P(B <( 6^ȠF3Q@ӨQ[a@+M۸KPWB w'0\@@]{h H75o >g+0h[@EA [M}` h !y'R@% ∏`F v]"("' |I_ry؈ ]64tW@-%AL9 Y Uiw)qCjRYQ[2Ftip$Ypʭ@6$0A 2B0e*X w@]}2Pys|@.I)>ze |QP@(8 %pjCa0C i`3yZ`AVjAΠ |03x mޚ뮭n.wJ?p͗!gA*Ƭ 5В Y_@e'w p $+vY;\ up=A#WV2.5ńj@%{љݬOwWwL@- f2-WLA•?`6B"Du`+zW!]5H sG ]|Khn3B%]#`ެ ڂAm]P7NQ0ٷw.]{AZ@lrrGGRsWg_. #uT9,ض;6 P Fe: w柙!~/ v S9Яo O (L w9ݥH9 xh P Q@=P 3=ݥ0[0 i,}AsPLi 8ŝlͻO Rь $EtHdX*Nk A^юA^63N'w}8$< Xŗ !VD_} I8t,(e6?DH5 9 ȄıOe`f)8oq%)ЙgxZP^zkgR4?eo 'NvK QSt*"!)+9)zLYu&A@H]I'R NBP,ifCAy<$ zȑ1LgORך΃6 BP)$eD" 5-D91t6dL 33*|k).9@e'AĿlR 5#*;;JT)ޒp%%lq mFok΃a 5Ni3V^T#)`<%2&t[M1 vO:l*p g[]hBT4<Rj|/ei^25ج zK.FK@C,!lIݖ : U8#eA\yt<̶.KOLe5EWkjf*ȝ|)# @u(i{MFȗ8wA0Uj%v =׮.7%Z@Vg*]!"xW 2@#$ю~IF Yh "w>ԨJ3{B%F,`RNT$&Q~N68Wݓ%2TSnЂ" I 't5@BNbiηr`{yr(%H;'N[ϸ7{ GN(OW"An['\l6Su%g4t`EQc{ZtT=WMֵ~P)~2mNO2`@I(t=v;%R i2xߝNKvz{_W7y}kŝG|OO,ջf^aOk[Tgqߐ]|\#O^ү<}F$߾p~O` ??&0o kG8Il'G2#f&Wǀf\e:˷*=׀"8Xbx(}+g.q599KV!@`3T׃1/#~,-PƁ!P` cS1U8LqA&.~!ɲ5~[$pytV=!322p EcRr8-0w0%d243p$H`PB?iQ9b5Th23P(Ҹ, <A>QVd\QqӸ5сY)f/@?QAM tȁ(8iH2ugs"")H6j!iyV6b(Vw(9Vo,&Aw4(o7YM9%L>=C9PRzMyh%[%0DbOɂ[ x7le96Q #my8wH1!lTpx/?8Bp1-P t ck@!TmJٙXNas pIvMtq+?!G24{qRU@'CH?`= 2@6C.w0yuP Y6KPy9!?-3B /P7[؅aF?X(Z%`Ò扞u:0 6<ڣ>@>yVy"Ĝqzj5J p[ۘIX ò"θ*Y N_zhj1LЦnpr:p0L&y~8ȡcDSy*ڎ_!1*zɐa1}!q[f:(ya#ӥ.{zg" +2Azph926aa)hI"1!1!X*yRښS* + "1~1q"P"D:?Zj1֚!@*+ W 9bA/;- Cъ2ѱ` " m `!jB ?[AC;yS Ugʔ L:>﹯f쨗DGR*r[!;YspvKٷ;xk{{myE˸,ฐ[ :W˔v빞k{{*hc{ZɎf+鸓K_ Z1H໽K ֺ.;S[*J+;z;;7xL kY{黼[{ػ;y+K+~l< =:2X P + P ;v1KP::1pɭ ̻TXk|G;z;BViIQ|5)1)0cMX/䃜v m3(?mE!I:Q ,pB5̵[:|A@5ϻ{l*IꅏJ:`bh%A ! C%@`m(Rl'v|Ljm ڠxz|F~[ _;QHMJ+JQ*VJ̥uo-<T`Yo IҊ* Xf9DU3Ϝ㗯k<[[ȈL%1z /Ť8aF{BW Q PPjA y0=s -GÀR=P=@wxF]F;аȑ<я Q 3ڪe,(Qad,`ĊHZ.!6e")t9;m'?)Cm+.p0 MzH<'L)\ͿxNs-,2@*Vьe-plE3"̈ A b{}}mw;;:|E# ٓͣEwXD` =ݖmԙzܣ̣nשܸ˸*j+Mٕ}ΌBME=#z+ KZ+5H9!cg{ʫ8劳k!P)T jcZn' HpH{)p<^L"==^> Ǽyۻ iU{jIZ6 #bmY.ɤꊵf5=9^l~JoسW= Jn>>t[- Q4;.yIŴǫ)(nt҈&ʫT9Y P۽ޮ>:qI<])N$@/&< . П {˸D/__R+u[##^yě!+>,/}nHV9޿^*EE S/7p..GYt ' ?Ѯ= 0oonj¨ ̍f!Ȁ.Q@B (`C,$2d@@F=~RH%MDR-]SL5męSN=ibPb  PB$P%:8cA1RpTV@W~Vذ\-2(b +\ 8P@nYToڮ}E+ta[oGȰMxyŜYfZhҥM3]{Ҕ $]:)RuÇ^ō?5o"xy LiB}gr =61"_>ǟ߬W鴋h (خ&+,- "@DBRM?RBV… 5*2^$H# 2 W @P Z MH/0hJ)" cx(߾4ВR=+QCEtQGQI4RJ/4J7ŔS*A,?h-JLU3ĕ!+ 2Lt1:<x Z/Ym-RSTUbPpnjJ7Ѐ E ٕ56(A@@P6"#@?v2]IWX%ĸtLC}9V*|F".,$Xw D6&5CjbJ`)pvGȠqjH e t3qvYBn9[jí(L4!][U%N])},JSY~9SƑgXEm|-18cFfBMfU/ֽEmkx+e ƕ`f]dvS34!k&K٥vry'bVBS{5Pm(AK:2n1RWJ4ɀRvG_|ydHJCY*m OX"*ʘKf;zGz}9=WI;0F /+ɽw,Qy!l V0pwo;哦i.ُ<髾>>("S? dÿ$>7-@T=A1k p*'@  ~2Ad£cAx/ #d*$~qA“-x)x-x(m ʵ& Lˑd.K>̂Xo܈(̢QijhDA-, ;A+dE9EfG8#!M$΁B[[4D2׬>ؔ4"/­/@E<9xK$JƔ ibNrԃ֋N@P2hm O)9OM1({lٴ-N2 |cp;CdC\PqA \@ïLSS7Qk"Z&ɔO @<61ʦ)ݞ&ףR#YQP e *R*e6M$.-徖#S(J&ET̕È5uB5|S-A:A,R R -sQ COO%8ZMG-HT1Mc1aTTPUOX[SS =keE@$-֋AA%ڭ-M @בN'ʉO P}#p}4YۢUY >0V].};VcFMP]a >؜MWMۤOO}Ur#H9G}mBuȻ8ؒ]ͭڣ1A][T]^Xtݝrc:f ^XQ.)dFȭ}ݖM 0sT8F)DL/cp:Tu W__n]eڙ|^deZyt] %On &"k19\* j$afa@BnaAa޹ L,9%n.b67Ѐ+_&'Ёh+cb-V;% 1vc&b3?%3 h7c9<d>c>>&bE'6d+9@cF7GĐdJe+cLF+z5V$~;WB6ePdW[Ge:ed\f|*]V^`FfHfb*cddfZ ki*alnJm֦nr+Y"gsfQCtΦunywj_z{q瀮R{gFR ׃NƢefʑg~~hhsh"ޚhikv6idiFnmyi鏞i$ihih%֏.jNinj~~jpjjjjjjjF\N}jn~Jmk&h2ZT뺎k.lO }TĦ럎/0H%lꬎM|&jOt˞z݈lI綻n6n&V:hάʘx܎n.9Pl.+N䍘c TVe ( h_oFnCV`o&o3&, .> ҁ8Ip ppGqgɪq&'jq'qo7!qp/"W"7$U$W&?T&w(gS(*Q*,GP,.'N.0K02I274wF4W6qv8oS6:5:GO<@?%@Bq>7D?sDWF_sFwHsĖJEBLLNNPQ ~fQW;O _ZGZT_'ʁ d uev2R:|LoZk_0go7|@nFEtjhTGuwȡoUpЁD{/OLIGuOLnGbw@n@{rx|vgyyey[_yQzM'GzIWwwzE7zAz={9w7{5G7g{1w{-{)w{%W:|/m>|7ȁo|7h7GxQ` |y'dҟy%3CwWwvG΁wnoydOI]׈n\}7|_ecWd5V¯w7Nv^x}W?Ox Gv7w,h B2l!Ĉ'Rh"ƌ7r#Ȑ"G,i$ʔ*.r%̘2gҬi&Μ:wsKA-j(ҤJ2m4PQRj*֬Zr%9׮bǒ-k,ڴjd-ܸrҭ-Fv/z-l0ĊK,L1Ȓ'S,rc˚7s(戡?.m4ꂣYn5lϫΎm6Wk37iFX|8ʗ9/W8:ҫ;x o< O=ǯo>/?\d2t0`~:x=8 %`20ᄦEև vQ *plሤx֋0 %Qe㌝H֏c9B / *$P!C*XZY]uɕc hP }՚lnYYʼnc7X`%h\ }' :(z(*(:(J:)Zz)j)z)Rڐ * *q@P+c PB{,KBAAJ;-Nj OI-z[R ߚ{.1@.;/{/ꛞ  iyAi ` "6ٯoe$0 ktC 0% `@ ?@ ng5LB{z6-Pey9Pd>ur+l%0l˙F`tP @FNX}I@OԋvAU}f\Г}k@M#|tx"pmg(đvP8cpCy5ݲf/\ٝɌ❵piz-یQzT:#{A^P kAA˟vC k@ 7Pf,Tπ*@g"H]r%>Y־ [뷯XlA/D2pB@ [f癦A(<# \XT{H-F} "B景4 A, ,LӖuf!ط NWdhhP" Bי ,  -Y  D!TCu&hʝe P"V*8p8AΈus87;20HHV{ 7cSqJ6DNF- @G PDɊo#d%3hAjPvq&mRJQeGt@  ./0W$ ?t2sJ9*+aӘKVžf2AƅI͔O \e-OBd&>`l"_kkk+++ 2oo,, p歭BBBCC霜]]ttRRRuddꕕ>KKzzH岲衡44iiTTjΖ iS幹zC9tyd&11( H*\ȰÇ#JHŋ3jȱǏ CIɓ(S\ɲ˗0cʜI͛8sɳϟ@ JѣH*]ʴӧPJJիXjʵׯ`ÊKV1,Äpʝ+# |pѠA0hH( <(l!24^X3҂LHءAsM"mͻ]G pe. pw14HH9]s (Ckõ?EcǘI+ƹ€-wz1fw֖|uW 0{ }8]#tpy p F$vH|՘ߎA(biSds1Cd@C.5paPtBt|PV9_v |E@ tCp+B&`Bf@#@ .ꮥ:X꫱ĩ[ñ.+9h{@C|庾{ ]^ hA2`7Q f _yh0p\^]S~|aNjAzT8V] j5źS_+`h]'BlC$[H<4'h 2 l~uЪ~yT"u`icͼ(`ZB򥽁tP ZbQ1[$ dsA99j G*d~9 b٘s H8GLbn0G8a69mZAPE" 'b0 !zltB±rofTeGkw1$9Cxo $Tܐi(hZ2'MyE8Pb F,H N2cbdf<N$u1'a9--K f`rhB;=L4B?EeI| uOpsđ8oІ7&w1(:03HSNLIs5Lw0o(%Jh)i֓ 4T(R:]H1mbiUв$4+J䐟bt{JͿL}%OcGKL^`n9G*dPt,OnD0c>;E"P$K5cLCN-e<:F9Xe )ݓƺ[*yNJ&ŬQK Dۥ'uIZ)2@"UWD@9,`DAؘ׈u{K{x"d nV.+#Ws_ൔu&*lɥa :%ҫ&_q0m|9@($C*٘ye煷*Ih&Q ]@>q,UpNH-h=KatbmJr Sqg\Հp,*p/ ͧ[tqt RU҇N dz.flT9Pim6"x8/ZUĀmD&>mg/"Զ6CSiv js7dȎMz%(~ۖ4O8kْ{ N؅{@ȂaXVX;e^؆M{aHk<~9D/wXyhY熆#d'1$~-.q<H~8Q(&;s"~QuYxL)GX Hpx=882m=RhX*8M\<T8W3Ĵ&2PF;6M# R5QTWx7H8 @G(tFʁI(`z;Sf`H1c=H@[Aa.*ak}@Q45\%XY2'aq*`~\2>@;S3Tn5Q)/INɔX %mdWY7U%`BigYr "##`sᗉǕ - Y|1 |` I}y;ј_p§}6Ǚq.@Y)= R&!zIaȊZߴ͋\G ` ]l[A#ӋԄ蝣q,h;MVޥ݇} J;]QkP߷>/P0M6W;++12;DA8$A@a’M!sCs9ejεD)ff,Ynzd9^՗4̃^0>^+ MM޿Bm ) \}ѼS{P("W(]k#ME u>s{; "rC\ޞn\\eGI.֝R.۷})p -ט^_x[2$ĢqԾ')qGnB >,AnG߽֫n$֩+Qp+*'4Q]ϱ_;r.aI䣞+Ͽ_:r6/:xviqPۇ +j߼Mj[;94]MzvF +qCPx<@)߿R/#wBsbܮ ?ĆF_~9^T.Qn i  KuYkM @3P_|QZ@OoS~t0 cmMo<ǟP<,=]{/*(X>QD+6QCY8H`f0J-]SL5męSN=}T̍E'Yae D#b`pǏ 3.xB @j4bVZ3npĆV0t%a}#℈߉S,!^;lYdʕ-_Ɯئ< sr=""*5̵m\qC6DbM{-Gqo_Ǟ]v/Ӿ -L{Eׯ!fH{^_q_pT=j"!> 'B (<Ƌ;a)2E0|h?BÇ.G&۬!*(/*E%daHÜ8 Ɔv"Ӻ!H /3P,'WqTZpB23HJ3M|8Jrp /?tM 6P';(T83/E5UU4Cu*b B:ԯU(|؊F<TDLx6[m "Aon%2h;jq!VLMkVUg%b1/@B`!xFw}ە~#DO!e1'\6[w߇dxׄX!*ոe%bV"$Wyܗ;luM%~c=5-L6XW3j"9ĒKe~nפ#(ޗ2|((lNH.3mE4&58 B|#Qxԧ,ѥ6β9!Ã@vT\ C B+ aOWUuNs#Æ8Ag 'S*@! ]W L?1%ZY8:SgP5#,gT:*+q@P2VOi,F+)dLfj[X5Dk "v'Ԫ^=ܠgu wu'[ծ! UX/Ɖ$9jҺob~QV@ҤKSu iErWlKEKjYwح<jHT~9ZJ`*UE$ B1uҔE D]\b\Bl*U\`pWp!+VIkr@OzXfm0TV!_  yA hr c0@zTS!Ⰱ:dq:,#0 ",w8 :(w%4r?-ĭi…(?+w*/FދC2 å//0Al<9ßDEĚz :iܾ]B;əD@X<zɠ(ZE%ʈH=΋{ɽٍ|ʯƏ,[YJ˯ȱɲ;-;(1l˙HZ:3K(ـKE˿<ɷ|Ƹ7ɨqʭ*؁hL4L ̏LKQM:$1̞I1'LHєM2L2ŞM\ؼٔ/uNڀ$`MesMt˨L-܌J:N $M<[I=;P|Gє O'OwPOJ@="pe! zɊ6bPL\A J' BO "O3ZжIόIQ7HBQ@Mҧ(!(#m &mQ:<| E|LPSlư0%/5YDӲ҉0N]ˆ pa 1SP ćQˬK̴NdPuX =ԧ,7TMj\N)(ix;?;lOTIAdA5yUiѲIԕ]e'M%` VU[ ]N! oЀppf%g֖U=jJw] k֬K-9[e9lAL)"ʬV!$XJ,yK5UJu7;,H;CafFn(8Aά8e $Vhz b T-߬ 7N-W]͋8h Ѐ3+>N amg bHcC6 BuVed{ i5fFjTU&#]ٜWf"d%khp$\ خ&#UefGZWL2恕h3v.Wq%W(EmV(PkɶUW~00E^أ _fmb =Nl.>; m΍ֳFnua=plVkhpo 6.bez6^֛\eo~l탓#a=J<9̓fz.L@c XV(0 )UK’<  &+me(q ͸ppg!'')7wN\&Op'U,9c!_'vQ3g-Ϗ%rasQm1rr A;-pvC8s)-Y $8ϊ0&X1+s6 A't֘q.PF6D5<%@./&.,,1I@PtYTUoYLouMO> uBu$Gch^pfuiuLegv.gMigڱv@ko@_v`7ugkXvs9&w.F`xws>o jqwwɞO[xoxHxAUnhOBvv'NoxGwoɉyDpyRH zxy GzOWpz?IPnVW_zzoʸ-(g7$*{D{!p!(:0{{sy"!hpg|zh6!|wA=h 5l} 07o|L 0} |gzoq n{.S}}؟z%W~&@~?~'٨~/Pᓡ–xOjޓhB ƒ&2l!Ĉ'Rh"ƌ7r#Ȑ"G,i$ʊVV%̘2gҬi&Μ4]LB7"pᢁA2m)ԨRRjǗ]Ǔ/g17=u;龽Ev~ *]%x`Z J8!5xS Aцz!!֡R"x")FC,"1ؔx#9T#A _mCy$:ВI:$59\UZy%Eb%].7^9&cY&iYؙk&fj'u9gIa޹'e' jIzz(A(y(vY` .h0z~=`¦%~z*?꧁b#~7`κ+W 7@z,g@䪔 JkhZشb@=D׊;wɠ  ,;/r4ЂK6D-,E4 LKف:1k1tUl!|2V24215 73934=B}tF#4J3NC=RS}5VcZs5^=b}fvjvn=wr}v7zw~>~8#83C>9S~9Y!u~s @`ze?aDC0|?4L@2!h`%^ r C(&8 34OZЃ 2ģ)yb:S4!KN(e1X X(Z.M T j!D Zƨ#r&.p|u'FBFV .VՃuC WuY][yPpKuKO2q:< 6l-bɪ{eX,>욪\ɟd!pƹ?.B@ek[!lqu' |~Bi32)IwD^Fƺ#.F.{hݏ¢0͎{dĦI|M@ž d=E _pqh]P?MyX@!1C*D'Eɿ6n@qDWzk= H0aԪ. dL17q5"*Za? 5Z|.ǹ,`- upԼ8 B}3E3ю~4+!, LLLÛc;;닋tttt(jΖ 11H H*\ȰÇ#JHŋ3jȱǏ CIɓ(S\ɲ˗0cʜI͛8sɳϟ@ JѣH*]ʴӧPJJիXjʵׯ`ÊK 0* Cpʝ+# |p@@.hP(Ņ >(d!26^8F3 THa_A+M۸Kp_WB@w'0@\@(]{h!uo >g+0h] B֖|uנ 0xz! ]#x@x p ]04@8ZpT#4#B7 w]DMB9В Ip@%XpShg:fy%c1)3  \p d Ő[* 3\j: x"u1: \p0Ǘ(hSB : (;J 3P*0 v*),hʩ:&e3l(0Bj[뭹ޅ+A: N;5pY`DH.d|mCH% %p`YGpGe`3$2^0 =A/ &2+5*^@󅥥{Fȃi `-`,u|c`n]*1@ _3Ɨ4@_qksD]|Chn3Cu]0^<VA/8'aH~!R ;]$4njS<P': lx"{a{8!T(|*0@F58Hj(F:wk_{*yAWE 0(q٬s 5*-TEQHb_Sř]1zܢBXDgZQH6p3i :̤&0؈~Sj7HRh`I/!& *Cf% 0A%Q$˷XvR5ұG4BJDQ"%@2&4-(XT5?(Nr4ę|e Ξ%9pFDxjcc.D1X.B`!=H*>WwiIL $*=#z:/b.;%ЯHw`݅TȓJ%=Ad_Hf2&e z[^ SX3HdXVvsc;8 |EL _CJҔ>C}K"4C@ih",|"WDލF!^z+MlĵU Pkd;܉TĪ>;*h4Ng}r=쀆~ \ $)3!܄P~ LȂ{`m T;JU[I@q GN(OW0gN8Ϲw@Ї~@E1=NOK$0z֧/.;0-Z}`;|oC$p pM.}6 !@qL`&$Hxi|Ch 0kj<\4yK=H kA=YTzk{}AR{=ځO#\2o#$ѧO꿜!Ώǂ=O?䏏FO{_'?_A!~8h:px xPHaPfPqb^x"؀!4h&x ~0bx3C9;,.NX765'98Z\xYфOPyRH!;m`jqy1x-Ȇu!&!~qI] w\؉G68aPs2y eFϔ|Ȅ8!SXJP"&~!Ia|6eD$vzs!H1GpQE(=aPhɘȌ58bD<0?h![1Q؏5p7k*2CXP"PiYA$v% %# ْx:(.p8,(@jWD)C=y&6,PyvEpIYd90* P)^Ty ,hI]=UIYh{i9AA(n7|tٗ@ڣO uu'JF%Yy:bf,)",A r. !6Z8Y?jlڦnpڦZ:{2@d2UpyAa*[U9!w0 å+1jZ31_c=2{" )zdy!ay!~j:Z"J%DUWhI_qʨyJZJ爫bF©ZYEi"2fc:ӊjѮ' J2 ~fQ*@Bѡy1'pj!ѯ&{ʆ"-:r9*'Qp*Ɒq$(J?y12@9fWD[I{`N{z[KKblZ!ʶvk( y+qs۶9Kg1 }ٸ;}hy˸᧹_waR+xO!0 CI[0# }AZ8(Y:xڥkPZ`ܛ|;:<ʼ%%(UӾ.1+1(j"PP>0$|:JI )k=p+`NB/6? @>ŷ:z:*UxIbVcÜ >[(/}Rw{n3 @nK<3SI:A0\BM'Ȇ<҄ ak!tJk]C$Ę`|$ؔt%3`j3zTػc%[*~ B<Hђ:֒{ F K@g 4Bi 8q=mE3$ݘ=:,d&Ԓ=ׯ{} ;Fǝ}km(BU]òA$"~S5ud}ҦujPTjЇc$Δ$ng˿J<=X97^0 p; k 5Ӑˡ˱ V>ōM|D=;Y8dj  O+vjWݑ{(6CC6 §ќ `ūPylq ;5{xj\&"ĬMʞ,*.+^>u*URat޹v>s<=zՂ kqɞfQ 69-1(Ȟ&(^{`{@^qB01 փ] ?Jn$P: q{Q8(qb;(Ǩ}?}7.Vُmߎd.݈g$Ƕ(?=?Փnݽ*?ϓ.84qwYuI&<N;~S_^]a=qk0nQ@tv/23/aI"'K<*i;W`[@6(R?xoK\?-RO7_:@ ax,øԏv 6OK_^w Avǻ^1A@ ɍxBN PF)v! L.vࢂ > |I@L5męSN=}TPEETHUE4Ɨ-',!EC\C a`U*Y4K6Rmݾٴc!]oHAP!}%3QZV3 Wdʕ-_ƜaRF)>L1PJ1rYn@hD`aaSb?VKrZ jv:Bi™#l V2,.z@DT*>H_}Sȳ7 akϽ/gl^vEn Z7޽5ۨjʒ!bh{P%2 ]B-n|7|jg{w㟎ݩ6+:b"`mo;W#h ?^h}W^ر<++vߨe@kUrDv:ЁAB'@}9փ @8N8`q utYGD֐N"} a(%$ X l-q!`VD*i Qؔ (\ גŨav @⓪Υm_zV8G˳cHQ^\ n&:NQ(VFґ@8H"fZJ+7 t#X.?0ȏ2"JV0P6) bY+"pfVTR2]%ˠ]\\c֤JGH41 @`^\&DVNzD&Pyq iAm^ZH i4/RA`h$`-YC5Z-?xܧ\*e8`6"m A.ы:3+3Mއ(-l4q◆`Q7L՞MPnږT~.%#*["Djm~޵4-Hd(J=DZ`ȚPx%XZ ,@l<2e򓣘V[=0)Ƀsve\Rb hAh1yw[t$s`<9 x@i#ϮT=qӜ3>0hd5Z9m A^$5H*ٮpY'~o H>S/Pȃ~j&6Pe:1}XzGHSthR^!j4!: 9z?=;b)&wMl̉ڛ_aK5QA޾8qt#R,~0^WA?yĵ>Msu G[|Ca繯Ч`L}z*{I Cn~yo[E$}&0P`9K4c;4d?\&,bj i@j ؁ ۈ<ԯ!A'؁ssA:;T0C7޲?#D0| 5.' L !£j0DC@sKt(3|:ļgIBBC'2:AL5$ĥC"ALG9C>C9GMA$47\D"MS4Ij,@SDYU 6-XY_|B [V|?nE`dF1D% ƺ2BkIFZfF`)pE\Fl4 c6Lr4l̼4 vdDyxF((o<+;qq&7~dH&Rb|qG;ʀ)x,XȆH8Di<}_1ȆD<1H#H}|I`=P\GHAY$WɗIٛIcl'1Ȍl`(ɧEc,$IIa ˏJ\>(ȰyqʷŠDKN+A YhDp#̊ġ܊ Ф lLTC6 L ? QA"ErdȠ(̻#ՂjMgF-+ lMN<Ȝ/MxA ptɎX!c0NYӬ)h;բ4׌ɱ!Ѭ'xϖ u$/!O4 (5N` eKI7V  9AqC)(]Ke ؼ# >@ۼ PQlkk"JR @Q!K+}Z;STy)JkTTu 2 TTHOe@NP"U⩤d^ V hWUdQЄ 15U=SSLA%T,5-MV$RG!kV2iKQMo=OO3#g-Ut= K6LkUxW ‘|Q|}HJWO)UݓC+.5D{Ze1=-OiX“X:X ؋C}%#h#?+Adٖs3q>0RWؤ@֞G\J!ڏݢ|AO@!EOZ3"]%mؓ¦E [vY(R YVtچSιXe7[[!zXKOC= % M7MM~X=G`-x\MOk\\徐A)?e\[ݧ]H]f-i]U]vܫݢHFK۬[uXҵ^P Ȯ*\2 ٭Q]ލ%_̼Õ_UXX]#uߤ5L q;Eף5VZTE]}l.ZAORL nFhVkk\!2!iݛ} e hDXE+%G nYم9:eBlyx .TNd?j_UeecJ(E" t!f`kJ2fnDȳ  WVL#NQUN4Yw6CD''~e茖cfMg?熞ggjG UQebI}idђP.画a6O,N `hBdq 鏚6tԣ~-л }j=b 3+krĐ  ;Nk6"%`i&Vet&g; ;hAkD. #NJ'Vl ™ :!̾쏉Y|9^$&4 pמmxꦋ"- A\>j^-5pNuj(5ZW^<[Vhx86o^z8H]ܴ4(w-BӉ`b pəۥ؈ϭLʤʏ^6_gVm}PpZ괬܁Fq qɝKvhq#7<$TJ+w pk5_so-J>IUI/.s* 4r̆#wJ#%& sAB_`ȏxr!1'ٵIt& epRK<'Z3wTt QWʭNuug u^oa`\'1vW!`fatiib (p1! enkgh7Y>m7˵\w Zr-gdq/rsoX1oWI#w_o}O8~x(54d\/yo=Zs=nN=gWyvv=zg:a? _z }g!2{GuW '0%>دWQ'tP|?E}}pڍ˂@{7O+Ͽj p HoBj#َtꗻ'? p ,h „ 2l(Ĉ'RhE7B"F"G,i$ʔ*Wl%̘2gҬi&L:wpЧ@<-ʰΕ!SPadҨRRj*֬ZriAbkQ]1.5… ҭk.޼z%-ھږA\*&Sf2̚7s|ӰNdG.ʪ1ZSTШh3ܺw}ƒngͷ#\B:TD7ڷs>xAu jG@ݶӯh˃F;}E g * *'PX E%MU!8"'^qyaH  ы%#=O*Vxcv9RT #M:dL@Eݒ&e %]zDgXehWz%Ih&mՄd*"jdyg|F't|z( 4 l%:)n`'e)YB'D)*!Cu;z`V=랻Z,*,UN,J;-H-j;u-ߎ{.W.6 /k// |0S0 ;K<14okm{q<2\2)c{- ,/}5֐GU5aLJ& .T NJ6m9 0m}f[@7hgdx >xVSfRu8Zk6Ƞmu{.Uz=[uB9Ӵ s;K_ <ޮ;Ky[wkv{uߋots?s뻯r?rۯq/qp'@pW@z%pTW(AqEpV/hepDV?(Xp&U OBJp.4T _(C7p6R o'>KD&jD4 <%.nu:+"eDLq]'$9XQ"(1$JH$2vmGT@q#E0ʱ$ p'?Pyc )` Y$CR{#IX؀#Pߺ'INҒ4)I$v|b)YҲ%.s]򲗾%0)a<&2e2|&4 ǚF:.9.9:Q6Ù.G"8. ("ʜ:㩭CM <9M'@5A=(BЅ2{2@z'@48T4% jyhDdRPe T` )nTȈ=zJZT6MmL] .~IHLzBɩNyʓ@$5E$5< & }8A; L<]kXʤU"ekZ4%kgDzvMOLLL%%%ddtttnnccc<+++岲 ]] lʔ,,:::uttꖖSSRRR歭BB(霜JJ)%%zz£dr任衡44 fiiTTF삂 1i|ߖ"-Ct^”2y11V{h}(j͏`ÑH H*\ȰÇ#JHŋ3jȱǏ CIɓ(S\ɲ˗0cʜI͛8sɳϟ@ JѣH*]ʴӧPJJիXjʵׯ`ÊKV2*)Cpʝ+T  @|"A 2H(A @(t 8^@3ro9&}qFts].^~2J!Z$ gt@v54:p?lE_LjI+wvw [I@p `!ˁR"=@p pW щF8@Z@<%T$' 4A7Pw]H[%])ЖB 8 wQf]C8yi ) 'e&:א`%A5W#`Ari_ C 8@*uJ1Ł0j<}p@>B((k` AyqƚìZl: <ۭ@%;*"C|%{ױ[]**^"iAJj)BA%CWw,ȗ2*O[ 倐6RcY"Tse7y9re>0qFD`@uȒ,_P';AS,UvH 1n>y|pd.d C Hp (,/>6|倃 we|@y&o&4B]~PH:G>y139f: P;Rr>w肖~zwN`m*P.sɊNl4. rY?$ RHhCu@KRC8nfN"` (ykCH2( M@ F/l]@2|G^(~POR :2V-<݁2QJBpC[AL *TB_pH Rׯ0\CIW ISLH@ZD308<U(Ǚ c2SJ^RFI)H)U W|9Arl&$gІ7%TNeS^,CL$EHkZ[@XhZAǩgDa-hSB$g7G12 5SxRMrœRz2a=m8 C ҭj3yCJVRahpƢ^ :dt(E!<(7.(h\N _TBKJ@N"7Jݕ&J }po;)]К-G~y4 :p~ bM4Y'e[pG.So[ &绖]'3 @bອsg/7ji-,9 wKאwM6ê{"'x\BbƱV]5&uaLpxk$V$c,c <ʬk HKsc\Y t/Mbp7,Q]^Z42d-/psBB\S9AQ\[i\Cj51V_PxϻWZ'/!u؃/tvzr E74Li Xr&_ uis!u~l0-&Idmկ!&A 2h$޶Mlo!.U2sOCz\$ЁUe) 1聻;Q<([Kbf 9XW( PӔ0b89K8Ղ#@ЇNHOҗ;PԧN[XϺַ{`Nv_$1&]vϭ툂{~k܃w G<jxOgC8P@|5?1 rA}%vKL: b@>>7={~.7A~o|H~;?WHO|ķr O?Οw?eu_?gr|xrU GgXedŀ؁p|!؀x{)P)`+؂`(8EGЃ0(+؃#HhNr9؄Z=Ȃ@?H}ߒ%N؅gAȂaXVhq^؆Q}aHl~830`wyhYnxL1}qHZB-1q!HSAaXGQ(1'>Z~QQJNWh(p>8&5h!q>K؋Ҙ1c%X3s7p;L"Ս:O\Ә9H&FJ;uf!-Έ討I;(ChN5:2c=?I[1/a*ák}q@^Qu)iɑ4 (*9~\_3 kXJI3Y_1 aєKy&(rTi-X9JaﳕyEC'VfIp)h#qٗg7.q}~@ɗ}!g\d~Y<_(}q~6Hq1V 2hii gEIH(™tut!1J֛hx IHS(2щiʸ)xi|9j'Y,FvHC0  |hFF]o2 HI ":$D09jM-a=- :0{ሐxIA2ݘ;^(آRd۩@s^VX`b:dZf:Q"fVU"ib%x: taKi ?ؑQ0t4gaɥ mGQ0Zzz))unPy1QxHڪzPJ*|qM%H#Bv&hy{JC#Z3 ڭ"j " e~qiG8#R:b^S ѭ ڞ3d(#)s' a6y23qi2u&׊3گ;=KA!"y+a(L1 B*@j"-0;:ɓ7:cBJiZB+DK_;*G4Yi[U qy7ACazKx K[ۆsaۄ|)wt[۫[Z ;[z{#]˨빠K|[+x˹鸵˨K{[Njػ +۽+ ;G拕[;[~" A0 a.+ah¾K{ݫ{d[e}g*Y Q%+(3 {, \o,B|'l)<6|Ө+ϫKVK~'`^ "p튅OGI,[Ĉ[Z̫V̫T܃&o,+~_ AWAq2là&?I;׍i+-C zhC<A,Ÿ0Y p܅N OLzŲǴ},PKճ"2XЌ^㤳8h31ga @ge;鉖;'!2J6 ) ̵|̱u\dpQ. ?H|\`LQ"£V" @hh}@;Qn:n*1Q0j:&0HjSEB!w / YݪdžT X@4Y=; T`G07\ *z %RZOak7Vw=0N#-j:j혀cwL N RBYZ_ЛV 6Wd"gآQdD);.T"%~='qF-]O"*"dݠ8}kK ;`_B N [M"Pa|bL_ e@"|ƪ1yxةêpR9%ޏ)$c~v{3 dK*p_xlmTL >^;L ׋1hٳ_IceG{vpp>r.gXlJB5$xKc Y\]s$̰2 O5[;:9=a~`c% WxђGa6&N5j!ƞ } p=5̱3pMF~ ~KmB.7-]BX#MDy~}6j$7#RBt6Nyv|^|AŻ*bԌjUڎMFFs $! gHk\z’`sN.Q݃pp [M;.^Ka4e#D*ArJ]`,P}`^\2.+ͱ$-r;q\Ȟ|?"PA.caN=Kr/<I{xwIw^bߋ˶{=`/ZM{rCQb*^Ѝ?٧/?DA`B VpbD-^Xq" B IJdH@J-]SL5męSK;}TP=gũH]&qU\3~[XeCAX1 rw-EacȅD4wYƍ?>Bdʕ-_ƜYfΝ=Z4fNmrp(,T,D0um߿B\x^np 40  +oaa 8pzݿ'k TkTk#J:)3/AWl/ꈆ a :"#BPkWdE_t> ,^m6A,` ,+E`DҬڰ?j\h ˰ 3:&J2M5ds~6oGj4(< }k34+=źz6p "ҪK//T+D=yHPQG%Ԡd)Ηq  Ͻ(""=4u͌R/"Et!Xp,I33L!8\6QQIU *%Ln!8 "3v]cv76تuȉH p%W:'h7x'[6Nh0B@ 8x aEP HZ KKh. Nn9֯ׄ0E+bRCz4/FhӉv~4O k{%eҘ5bȭJy H-j^X!ؙ. fmȼ<~`G=xop׌Nc!/U}cB%s!DBls$=Ljw3R޷M}}精vܡVkky{}ZBa=EIljaM.ӹ-lYQ"3qn_CLг  d5ЅAN$ HP^db6D-hPV0ЊWdL sC9 !sU׫ %rZCg1I ";,яI+$Lz #9i[vD{ev>8GZ:N?Ҕ6ѢIU_V5Y펋 j\vo_Ɋ-ANLf$[vu,dم.mFh>C((@!^M a#)9R>&ѬXF? uiP*4}#;R,KڣgFhϚ3v`@;,O8cKbPs8kboB,9ZF+DWI{8.4#k3nm91I4իL^\(M"%㸬PvJU6pb ,Dž؀l#}հ AǪ6x*4X*R#5ӋZߊ֨I*BԪίr#Y[z+OZd9T:Y,xV8ؚ"$ 5$P.f+ "{gT[UKf3Y8.dx 5"]8"dA%D]`k;`SmwUe] V^"\l}F_#1Y &1g<%pKcRQQlxU1R R0H?d&ߪ 38x9"(e !BndӡAz4YAO]opv@̨0g-W"hT%qagf΋R^*iN_fQƅ1klq d3i;=9HPYz8xV3͟)V1|Dg5\dB w9pG& AܸP)7oTH &DiیdA=pKf.AMNCq rVy3@Kg-'׉ r0ƠWA<7r!p<1#3&U哞 C$ѥ.'2yF;w.^p  |8aqT `ݢzsB?/%Ƽe>(? -"EHĢ]Ĉb cPƈ:Ɯ\ @`FctFCU 1M4rlТ d! \4?daE`@X{<ſТ8@ABi}$,Nt3T<3G *I{b8 Ô$A4ɢɞ\@/4IDJVTGɤ̕.9JCl#JSʯ$dJjE/JJ,j6TLIKJtC8,KL8l=TKt4G|K$ʑD$Gd9LIt C@Kʑ\A@=.ؚ,aTLجKB)DڼAMt|J`zK$ MGt,NM|dJܤHD9NFJ O,OS\^Φ' rOTNOlOtͳ"5OVDTа|?DI$ P5꼙]% .HL&*DN|REuQ)OQ K,Qrͱ,Ґ,Q2Q\,{ zғ.X3߬g!9/8%_RTR R7|X$FDTB S>Hiw2C B-TDUT5-$L|̯1r KFiM(UK%UFԳSVx\}hP]WS͑Xxҕ6ցƠ쩯kUpgŷhTkUlEVVLpqT 7ȵm̐= 1  y]e#]\;#E \E]Se]u* 4)K|-:]=եxҿ^Xw ԟ^:ߒ,a#= 7_}DBUBN,]Js1^:e}3)YÍۉ )+^&`M\j-߭1au@} ޴p;<+JU D.bXD`]#_$&nbPC V&)/X1c!,)n6 Vaˠ`c;c69{Mk6>"DH CE;~딃M+eMdBnKL!p@2 GRឱ5@"5s-F߯N̢4ÉGR(f}EccF+ DW$%&0|1Dm潰sHc,f> c e\ HGy0x @[eVR紽^j.sŔ0h^h $h㪮J1"b_|άlnES,i5p ;-f좬vL'rO0~چ1J>!D8ZjN9!anBK[]e jj^X 'X㡶]Jj/ነeI@d]@ QYi&f3٣/ԉXV|l͈*&sh`#hY xmؒX=m6*6#+bn.tUNދϮܦˍ Pgmྉ noضaX̥648nozmsDrf FvjCvoTw`ۑj 83HkMgq7^p5+~-^'QwNq I* oJҶ~Kߎʝ~#pfiq)r#U\-yQ$ƒԎr-s%7Lb3RIr&K:rQ_56PO27t 4NH6r#8mrӾBt9rEF/Hw1~OU_u GC,>gxs3R^b'"zZ玱QkAuug'i?@MJP^/ϩ#^kN/owum5ju7Z=pƶjy7`GvӁ>2 o>SyGrٯItKy֔j f/gJKBonWys`䩷A]yw/iєwVDNrp`r'Eozfw'n&nLـ0?g7ҞAr+{ՅxdYKI2e0  ,li)֬Z:(`H!/b+ʴjײm-ܸrMy4Q]PݪS(7UJ4ߦO !Q0OĆSmE1>4ٺG.m4ԪQ\ -ӆi5᫐[^BȆ> ]_{ ! FҧSn:뼳 xTqӯo>~ۛ"aRHp h[oB]  ] 8ȗj!rV pHYy um *xxR ]!9# KNđH>5QMB(D|Bg%X8&eyfD f5MCT5 /ZT E)p-iܝy `AİAb9)Z@4 C,1jM51YheHdƜ%VbP(饽+کLmFU PS%b;@_9IZP-@*2m{."eʔL.XQQA 5yʂ@ .ٞH.A@(|P ;0U@ML,AI(y:T( HA5͍15@Pf)hp20,H;p9$ 1Ђ7޼5]N"X̳B05tU0C ̩XUQd 5^/A T4ƥJ"+LEt|;;E>:5O"4FАۤ4U@ CFSq9rSd+[Mh_lJy^*T;0.x=+I}d?0J3??Ob U ~0W UpL- TE= T$>1=MS,~16] S4-mR<1} !QD> %WPL~2 )Nf*ViM]>n-=ܗ4vn &h~sg .7fgñzm (xM(8 I:Ғp`@Ѓ.\ǔrӭP(]iJpxSwё,e3~6-iS־6ms6-q>7Q(Fq3T~:gU]$`0tM`t209P~n1'9 #<@ ҃ N'0aU\kv:u%1@o&[^_zƮC]9 02[^pڙ5 ] m!| Gl'gd,f}^{kxuS@AZ?lj8r FLc5L F6n6vĥ"_3N/v 0^8;֌mހ?V @!f$.fșF@CŸI.OH$KK$LƤL$MD@!, BBBÛc;;댌kkk%V{[{\\\ 333{{{zzJJ##秧:LLL%%%ddtttccc2+++ m歭,,nn]]:::uꕕ~~0VRRR人SSBtt霜BBäɖ&衡}44ii岲TTd삂Ffʚri*;zԎ_Øy,11=tHh(` H*\ȰÇ#JHŋ3jȱǏ CIɓ(S\ɲ˗0cʜI͛8sɳϟ@ JѣH*]ʴӧPJJիXjʵׯ`ÊK 10Cpʝ+  |x@ ,P( >(l 9^F3 Crp/&MA%H+}!@nEBz.8[=%ZZkg+h_4ߠEAA[-Ag3DHq݉zLXaH&eFawq'44@;%~#&t#B6w]D @@O6teB[䀉pׂYiweuimNY\2Vxycp(YPpɱ . [ xBpe-<78G 8]}: Aw@2)N |m`@,HA)~ ($tj`+7ꥯn`@ZP7_-P ^z 믱n/wJA f5pu-`DL.T|iC8| p vi+.) á>A'W2%1<e{$+ i ]|L@W<bm6X%-wPEC"5_<=%_wt,PBҿrB! o$& wݐC XW px5=`kZ8䒋H@n8*  N|wƵ󢓎k#zUM,<2P3P 6݆8 &A@{"PPhCk{kHl*M?O!y{\ ƌOy=K"&P1 iH$%ZokOL/ +0(NZ)[pp|QX} (da5giA*W.seS6 bS;fؙ.z@`+ݼWȘ4N8.אĠg BL15jO"02RdMY5Mb[+*]N5 U80lh[ۧ0 PD^;*[Z^g}=c⏺8rXx%h@F!0Bxh"a$vEi#2A-/Y!9:9i&(Ah7I=HiGIzj՗T)KzMY&&6 yeuUHxqE9n9@( `)n dY !-c ximyzr1lxyYC{1o~>חq/1 9Rűx&BQ>pnr$)BxY3(s!ɚٚQ!,PmhVՖQG&Yy8ɩnwHyKYvI+2PڣP*eJO? !yċ :huRڡD07i?,?,6y}xWP?)E5J/6(P#j ,uizX*gu@^`b:dO DcJ2!_$}g!i:Ic %,,ґ.ArǠ4UYڨJguDZzz!JB@jbTP`yt:BXy: '#-AZjĊᷔ77&$7(GvassZJEp"z6!zڮ*ʮ ǬE5*Z ~Qөpjɤ׉Z Úz;*J#VL}_YQby"0s3%Xd1 W jzǬ< ; R ihJq%%tjaEУJk]ʓ {YE+m*m[|j!Eb+I)4(!BIzѨj{*;9|̚3ol k먓ʹ[nv[ ;j z{BK›{#ИfǺ*kK؛۹[{˾qKӋ{ q;#7ثXʹ\W: ꛗ[Zl$!B+|'0aù bi %ĵKe AX 1M ;YB\ɺa_`/89Z0* a"×K, P zl]W!@P]! 5) QX[XZ?ߙ(xp{ 'z᥉aHYi1[]f-e/賞w'n,EP g(%4C6$@Ƣ!ĩ8u\zP>̳͠4lm'|^*2ꇄ1shAD*%GZH BnxE,؈5j10!`!,0zɰtXwI?kT`Z 88J%pB Ϧ qz=([ȷ|"[Muщ$l*/N0)I|#XMEMK- )՛kE JHۮ %J8pg@(4]!pdJ4S8ps *'l_*ZzO:J^F"avAM\9(P,jjۙґp= ďPP6މ{;*)ލ:Β ٗZbKݏ}JD(4a`ķ%&&"ꬾ%5>j΅[܉ZYKP5@ݝ6ؐοP  0Mo[QʅL\1',[(K#J&`n*%.y!䞘 bY[+~WZmbɭ^:$0# `!ȭa!n;o'ƨe&SK,28Z÷0[LQ8X~+ѱe&PBIM@>ۻ^җyv~G:}<;cNiv5m\}lbj1Y+K\`< ~0B‘IM738djgPO4vS|pmk+ L:r~\50{|_m.Ԟ2j̭.1r05s5p*a:/> Xd$_)KJ| diū݅z>$_P؈$@^,֖>?kA,ppy/>?\%PO"8؃GKӋ, ^`;!o&?#?p`]'ॎL `[p+e϶JGu 2|\Og#ɏNgw?<*6;4i9vbpK@ DPB >QD ĘQF!)R8PP@d/=СaJ-1bx)K+1̨GGTI3h͠,ETRQ4UT :JG(,s 8|p0TfMmYVWnU4CrHl3m`iAй-_Yf2Ɔ &B rI)B(kµ1vƝm)ҌD@Gu!'|xg . (*0ȃ ƞsWݽ^x͟G^z+nh0Q5:d6Ӯ@ 0#!I2ЫL0*t&tz`pbb D1EWd1""$PX* jȩ0( (2)!lD*39\P)B4:.Á$L3ϤEfdnT`#K  l:p%4I 8$Jp`0> zB1SO?m\02ac%H .H@=JV;<"XOI5mX"4 m Wi675"Uı#hR:Zy+3 /p`#De+'2nt&`u(ۄxMIU]?+Zc4~ A52 XA}kw啘E@+$Vo9gnHaƇMIH$AR5ΓRr6wiٶL|:l!7%'Hi60*{0ȭ/)_'l_2I VP+i~*AR8܅ka8@ְ"1ahrWBix} "%&r8)p6b3C dAQ0 ] &ej"uF-v$žPE>bA:{^PB6LĚ46e*+ q.Y`E?F$pL2L\c sS<5##".D+u#L>ɒPҘ+%Dzի j2 ^n,$ %fҵKI B02̮[2ȷd*VB xf4Бp9Q3Plgia|9\R)lRp&`,z~--7AaYSDI9] 18bD L &OLS1oՔ? =(ؒ'%Z!tIT[5&,_kPLיo *401dkb;Vmݕ0pFh:FVаȡAG ZUlj؅a60J\}i{t-o$.: x1İBj",Hh \Y+aK̻)c:"I0(5n]n|S\m0Z"mj;% ֫/eFoD_ٗare__ /Lf:~ַE0`ԫ<ێU/\Е& CAHPLU4Hz\%ךb$ {-B c*Ou ;*G|c Kځ̯.OX"YjYe:ÅsIg,@VNt" #AEh+DX`.bHQPΗʝ/a-u: ;- JP c(m%5d qf |18@i 7-F9"ˎ&$G GPk6-Usmf!'u"_9n>WP%\MYL\%u[c4(Π6`GfAvws8Mk2 x2ε^sw5#[peͻ[Ƚ4@eNSrWYAm(:Gx˻پ>gR0)L VriW/FN-Tv ^gXϑqf,2_kcLQr0+nF^-ξJCUgÜJA^fa mxudtn5bq` ~R89O N{*C@*DA6;< ;dQ+A!€ ??R+#D )) )%$, EˈC h( iߩ  0p@GEyŖE0̝b40x gĎc켎8D޻6m(%оĚ0DqdYǧ)ɛ{v3YGA{ǁ }DFt: HC#hƈԥyHH2ӏH H4I$H#PTAɎdɶ BIT JFH`cȗâ\9 ʜd +!Jȣ+X 8LKJPP>l,[˩KˁpKc˪PTMѿN<@ѣ N/ɀ{?ڊT\҈Q¡>F{BSDϢKF}$k0TI5J ~Qk HU 1ATEP>RʭԆP]ռI€ae=cU"]I hFiUS%0QoP}@ `Vg=QtN"4ꃖwxMsWE)87UxsGŤUl}=؃WeBtҟ[GXXX MwMMzXY!|A t#]|K~Bmو- ڠٚmEH,Fz0 1Oțe>A.4ͬ89uiԡ(XZou; Z-Ye +ٴ@eMی* 8C[ThY%X۸>Ѥ$MI =8Q((_M @&uP8ScM;VLޮ;J0=`@m"UUȅK \Y!Q »y` 4^ º^їv`.9.ci^$0hb'K[+Oq+"X+$13`1؀ !``^K.__#@b#IY;4'"d]=\>(ͣ=~JV"+?zẠ[>8/VdI$#?Vne#tJZ>+,Ƚ0fÉxx@Fk\pa+Q @3Kg7=lFxyg|>afIvLG \ X8x*v?*5V-9"ό\NRw~g'D63f~@ &A%P3yT@!]DNԃhfi6`r޵ M6O N/fĆXUV@ =^VA4%ZJ봒a2͊W 꾮^Y֥SH(ېfR;,HlIM0p9..ΉAmݖ.m5_KJ4~E>0ȉmޅ ?myj `Ůn/㌵l*ÎOe ~n-o Q"f$^'+;<&*l﵈inLv6\DYZD8h]bI cU@&൮ofqn%'q.o>@Q76dۑew1rɂB~ԗ%/L>lr"ٞ)&v P"ѣ3GV %/rA?pB/td{is GZۨը$a G. t=7s-GY_tdV_G ltK+/(qzr__^Bp1ɛuQw_$,XjNyvh_LݡrcOBdwЍ:Dq'w{AovLVwp9yz/Yr Nu#?dfx(b?rx/xh)ջQuq|'Ԋx%ĕzpKeW,] /zy{?$5￐x*I LyMqJ~s=&$LM` ?XK: N z{eWYxl7ɠtwȬo}AVÿ́zGyx oGo! , „0!Ĉh8##F":d8p"Ȑ"GtB*W$ㅇ%^4h͜:w'РB-j(ҤJ2m)ԨP2"#Z#zXbF A-DAyYENR/.li¶cIȒ'#* @}z)BgvFఀc@&B)< D&y00@X:8iw괁+}d{,BG*IB P1!A"@)(HGhn+${38\]*.ʻ"Q.'DPa`E 8vp9  ,gk1>;Q* m9J8+-*c`g.P%3С~Q)Fra -_tF |5\=#u]{5C+$EHt *TQނ8Xq/ZÜݸuRq 6Nu:eoQE57pc"8.0xQ:뭛J8x8@[pv{pU8ziC5:K?q uA ‘R5spO>G%v"U3gK``bJ<`W0D _d4 k鄀?C(s+󐅂HC330L`򰇅3QH xBC31ц|();TpņG #(RflZ/#(ǯJi$ήF0m s# 9V=,J!Efl|DȝLd MЏ$(dMR!dJ%CUNg$_RCVҲ"+Yb̖p ,1&0Ldp# 2iRztnL͛T| 6em<'jNʰD'<u+q*@Х3AB#*Ѣ4TD3Q? >A8#5)Lc걎vEOPȴ> JMy,`4;Oԯ/˥R>^$PV*X%@tQ 6ֵIiP^m+^ H9 bT=z֦7.D Z ҩ"Or0 1/+Fgv-,iS쐰H]-lcӖ-nD٭o+ܻ:=.^V}m VԽeV.[ ZU=oR͋T}F FT@S9 SS>00 R ~0( 3R- Q= P$>qL*z}r`c pJYئ4n]d< ; v YJM+etV2˹'COJ̺aX=3Мf1N#r'p(<1'Ɂ<,kY{ 4<ȁ (چvіFi〙QԦ>5SUծ~5c-YӺֶ5s]׾5-aHEMPlj.MujIK-n^*SдK%aQ@]ʡʈiV7k:ǻ7}H0y pBz8;(b D9zZUt~Z/vrB+a *m@,J>y7Y~%Ol3=h19N8BU@ Z^h@͒у};t~(JϺ>hYݪ:-hVu) odJį;8& wDgd9}?Gר O_}_̟Hȕ8]Q@ y J_4 qI`H ``uӵKԡ v ]̰H"  j ^QPu<] ^8fyd L XJa4 Dqb]؀f]_g}! aJ4!2 *]š K T5"KDJL") *8ȍ`-Nv T`Stb %Kib,&"r-ޢ-b@bx*N&z42j  ΀6N 7n zㆀ% 4@:v̼ $?&B.$C6C>$DF_!, BBBÛc0V秧kkkڏW%V{{ \\\333{{{~~##KK닋=LLL%%%ddtttnnccc;+++ 霜ʔr,,]]:::ntt峳;;RRRBBSSꕕ§̕)zz歭{44衡人eTTiiFd3"C)큁^šh uٙtԌj11[~~,H` H*\ȰÇ#JHŋ3jȱǏ CIɓ(S\ɲ˗0cʜI͛8sɳϟ@ JѣH*]ʴӧPJJիXjʵׯ`ÊK# 0 5 Cpʝk3  | 0P( >(| r5^ G3<]9ܸ˃FGrzhokM)%/꿕 4 7$"!w=-#$ ؏=oGϿFy t~6>;4wWvm{UV dsD@ 87wP m$]1vWaFaDiB _d t- %z ]54t @-T&A9 AYq[Di]DApThC}BtGF*WI@=Ys88ܕTBn1t] }p P>9v @ež1A;y^w+yv'pB6@ 1 ++ff@ +hJ. g*[,AZ,S:@iJ9A½BAB9 8% %t`6``?p ݐb bIU5c`"DteFu`7 ϕ^(jQ'{*Mu%_.l5 7#nxR±=A@DD{V"4v`<4i*_8wuz1 @uy@~}~W& 5~~-A>i@vMIP~WnPA@; `]{zo \~ j>#@t'1L.[Hzv*y VJ@6Gۋ>o )ANU_tAA9 '@>٬GӠ C|+b T. 8EPP৊3ILlCA"Ȑ[wV6"s _T p5hhna JcCgJ q Y?Efmq4$"03@&J"~e,P$A.Y*LJb6f6 |rX9V1`-!9*A fxZp eh<$Cu҇:Z8SNN&q3gNGXbD 2ƝDyLez!;E %.PN{201* ƵnvR Sy!8 P#?XFzҎZ$^E= YC{t(f 3c_rd]Mk *xF`#2 -#S Đ==hJS7![x\uүH!DaB F8pdsPz~-Zjl@('٬=KO:&QPNA'qQTxլHnf%x ;sG)A;S?RJ|r>61pxP \-tCU/u 0Zԡ%+-ŵ@@oeƘc)f%&o2ЍRMo4Pji: fb:,z`Go୘""J﫝me/nS V:q-!N?E(ei"yoC,+Q{4Er(W kFgXYkV԰SgLhz_-*80ꎻ`E6wjKJVL^@:iفIJHr |!Hʒ t.t֏` cl-J|auB{۟A DlhdH `i?{!HfR )aACo{OpTu'NT# b|[ vnL"OW  egNs5Ϲw^S<Ї>DOҗR9ԧN[XϺַ{`NhOpN] I{?_' G#5H'R%o@2t3h_yƇ^N_Ӏ>Gg7~ 9T|DJp>>3Ϲ2 /?\Qwr[ouw8(v  :7yh  1!AWs X$\#('W%+0(A$x:@h>p"|䗃;g|PP8Ph-'H'-/Ȅ`G?8PX .2p}A(@dCkхqSQ~dxR(f-<;Y|}XA#RB!`jQmGH~׉'MC#~5CcC(x(nh،!gH-B4`!@Gv&m34-OMXC[fX1؏<1H!+<̡-xxyNfXp70`> %ʘo. =!*l}A_H01VQq/.15Hvva`1SAMFYH`_g aaѕ1aٖ# oNzpY@r<&8`|֗{y؀;' o9) B0!Wy} qz0pZ)G1ٚ'~9$W}y隶},I-5Qq{H qNtu9+AeP"ٹނuQ Iy3yy8靻 zG繗ĸܘ)U9$1DCs{͂np0 BYiq0p0p$XF4Z6z8:zE07C16 0'#ӝ0*DНA9'?jד1,ЦmC<tZvzxJKw~:P]T"%~ѝZ 6 ٢ #DU,zz*3ڪ wEZzz=*7J(ɐԈ"30flyQ[ !Daʪ879bC%P+egҎzC A2jx1[ xֺ(yZdEMiꊭ*9*av &',.(۪gIk<p˱Zܺ*a/X;1Ky*j°NEQ9b=QWf´yO Q['k᪸w~(i(ʙy4лQ[t+ɻWW!a;[ͻ~c'a"+[˽۽tY.ʷ ۺϫۿJ"HS癿꫿ Y˾˿|,tLy\ۚ{*t 8:©~ B A|a -ĤKz;|°YPPFi1.Az,I0{yq;|]D _|)q<fh+ܾ P VLۺ< P} pȟ5Qb qȯ˼zguPx{\=,s͗+p~Ɇ8CZ ü0!i$'<ĠڣzIl\t,DE(Qb=1ET7(؊a@ZڶqP{]([j0"!IMZ'] Qq6+gWpW@B=BIϰkɸ\,2PѪEUȐè"~tXدuV=Ro2H ZR@\P2 +؅!03p{,l*8zY=%U}#qÚa Rب*CERǚ-+רZ)HU]H܍Ӆ]Սg@8]}rϚѵ$)D*Z-.9+&`>&`AQ n"ݷ o|mg݋]m ~J-rLI<С?}CkA iIa?q\S s%0Hн"@ٮj,8 _~'/ej0+ҳd Lx.5~6P~qL )P~GP\n /)@nTpDLKվ|j+x+U9F#(zؒUJzMS|L; )ܾJs_"q/ 1-@tCIgҝ؁_W`~V:}ϫV88Pu^q/\ΔQsn?؃M3p=J@%I"DAO{yIZsW…+g~'xJm3;4xe3?*π-PC DPB 2AăIY A=>$RH%MDRJ-]yDL5męSgə*{A ';dJHtUǝU^Ū!ʼn;J(d̞E1T $H7vԸ O JXf6P<2-[iӧ}=7DZ4|2 ZhW|PVlݲv\xōG\r͝?]tWF^E2`s̓1#ٰrA߾ im $cΆ=#n[av3,L&B /|ϻ" e(~`T8©K?0F6V,a/ZC$r hE"+dȯdHdJ+RÕ6"c5 Twݍe\u"&}娂,MBo9g F aKĆODb m9"ed@ofꗍ5 ̝i&Oy$}; 24˒Nk #A}W6 @3;rɕ=ۤZ]va cH'WUB2!M}!o!X\ R^K #tf Э W`)WhD?ħ&Ʒ"١5C*@lf'@T VA8LY`1O3$φw`*KqzWTa1v=oi =q[Rgk[zv :MDžLA)#Jī.𾈯$`i&<'ua dsAJB@3#LԥZϮZ3ҍE^\k٨^NF] .>G=arG،`Sk{[j|^ i>=`npyM}ꟗzg4χڐM0I}QbTYȠo^^V>G~HhOO]ݜ|oI[̞Mؗ`A|0{5?Xcgy}3$p&S~;8 \@`1b3?BɼZ;M[ /2@80 Ȁ$<@Lj/8ٿx, #d @jAt4 p;+A3A "$V9BX8 )B8"'A dF8 `! -D1$|*AF$ DC57;9$*AB,03L+L0Rt2ĩ`J/QC@6`4K  jRƤ80si5/Dgl.Fp nd+AǍaƏ (ߘ~Z$87G Gij{SȆ,DqŒ:tB#Ȍ 4H.2GJHȐ$IHH<ǒ,cɗ4ILh@ ɜDEL|Ə Iɠɂ$H \I,dʗ4JljDH{Jqʫ,ɨ\d&Iԙːɨ?9,K!˶HLC̙Kd<.NC .IStO8L PO)QPK|}#Pt }O ̵ϱF#P9 QO^P<јК?EG$GH<mX$-YOMEr=؝v ؀(`=9{J:ER"3 5p]P +t<nSIekbal  |Hc=4P`PCf'[1ĉ79[bP#_`yҍP(XC@di,@X@n@ޚd$.h[7R3YBA.3@ MTO X.YNOFXdMBQBq^(ݪZT] 8 /f0Sf=ErޒOs`b2e%UJO)d gѥpvngT<BhY^hOv=Uy8Fhsi^}]"f毘J.F$ާ%x0cS%j0t5x;]%"*Qz vD[gXiArNQA.jh6;)b)P귆5^LGKkOnVlW:XViQgnl6=kS+/!RŶ>Ξ- "h&h bm׎l#!/3g^b#IٸQ:mVEjkS:o*gHmdlÜTy_ƗMSoD'HQTF :ooqn̞[n1onk7!`\Ha$!h*V>p Op7jN\ @(qpWjgzN^Z3ژqLo_o,o7*"WN $]70,l.kd6vڮ0#Aʡ5I-˞&o|Qn$4seF>!;tȹv@U92JGto")uqQ`"x!$&.+_#w$E)rJIi/Pu#NZ Z?qV9 _v=)h)+Go( mHuCIso'Cn& Xyg ~C,w X{v B(wfAO`諾g;gw"C*Z4w*ssyv "؀)#) 4A(z]4_|ިy'Eȕ&x رd"ϺO蚠A{l%2Q^{~ƽ+(ch H06r` @D檎|&yS*e&2v9V.B= bԯ qJ[ҕ0<[dO[6aygٷPUmU{]đw/nc'3I"~v>4zo(-Ӵy*KJ~ X„ B`C 'RP7r#H'Bz``9<$%̘2gҬi&Μ:w'РBŊ5@@p$:E 1)֬Z"B <,׆l‰48A[z/.Q!R#1BRlXiPPV+Ukxc^96ڎOӮm6ܺwx2;iʷ&k|ضg,A$Ǔ/o]{552b7cwuPǸki5K`=7e&\b6QoU,ܲ=8Ox'~sK>9P/> FS9Ǥ[4": Ez/7rꩻ~;m쌗\{ ?|Hi><ěnyEs+{:n,9ߛSt{>^??b?(@o< OS2( R'V r b z& |/'|! H.a-[ sC+1 I!}b@ Vƒ N|be& x9h;(r"g #KlӨ9.x j@x@#VǺ9XqvGE&S:Z+c)K.Ep%-E\򲗓/)̿rse7)NYsF*\,!+Y\+$II]$s b+ 1xA@}/|+ҷ 8ܚ A1 A րAkG_A#es`<5ϕ[o5r'; h0lG6,cd7NIl84ʞc 0TB2l#&x0"sI&|0'2MMᵯ~ JT2 DpJd{._ `"vqdlЊ<#Y5sd:pXUlp4$";]h%0@PY P$'@Y5ʎkt$)u+3 럕ql输 mXJ\>;p 5EozF%%%ddtttccc;+++峳 ^^,, muٔ:::nn霜SSRRR)歭CCttꕕ§KKӉ{ߌi衡44%ii人H%3TTd큁rC)_É~11t[,` H*\ȰÇ#JHŋ3jȱǏ CIɓ(S\ɲ˗0cʜI͛8sɳϟ@ JѣH*]ʴӧPJJիXjʵׯ`ÊK# /45pʝ{  |*P(A >(l :^F3<]9Ը˃FGrZhtsm.^[90 MXCq,R~~sp`v5Ԑt+mM} >z z]$Z _QZA/P@ReHg@ܕ,wp]"%'|I] $%#B 'w]D[! @W:%Bc' A yiw&iy2V~&( %Ay 9u} A @UJ} 0,|Aĸ @`] 5*P +* kTl >HjK짨{lz @#ڲ/|WU~A&sMZopaQvh烉m| Bq XwWӕ X !f`]>V%B7W3BVCݽ\]e,_<5MU5_D sC䃴|}-|]ogW)AAEޒM[xLg ǂ$|+-A_uk=V/0_m|@lN䃖Hg4KA`])]3͹:/'Ar&n mspse;/,@+E`\~IP0C~XD@h ى _@P ʟo"5 7LB^2=$ 0P!kJPp|k3OAUg|Ol"ů~PbQȈQ**6F2;[O]wRH4+<o&F LM-8dT;ljxqC<. S76#A8ɡgb!M(kQ|75=QNH6ĦqJ R@jrC`*CNbf*& e.HHvRBi 32Bb/M3 6 4NķKll锟,JP|Kt>ӗy "9TԴM6cί55yb@r'C HTK1)JOG磕 Hu;pҘr=(q*KyJX% 32T:|t !#I6zGPLfr2f_xU G @S= έ\5w\تf:z^ZnJ+z˪jWptke8 NՂ80S1 _Ig̉\,Ch3}Kώ܀}R5`s7hZ \8w>.Ļ$F8W+7@N.]}Χ?"G81t|:0!j 7gHjT PpCcms9X4nVǪCL#_*C&TNY0l$>puAlnrWfȭs͐X)p z$4 oj[YYSQq v8j>NK@jLwRvtNO@4RxRxQϸSwo WtE(OS,0N0ł#$8Ϲw@ЇNHOҗ;PԧN[XϺַu_$)x 5vB{~6Z|S;#wgC.} xO >O H'Yn$?S<]$!-^.KUoy)a/ɲ`6H2|~q?F/;?'@O;ӷjsOOc|O A]gxW 'r Xy5QqQz&P&`%xpW (2XPp*(%xHBW.g| 3xC7h:9胃z0HHxPtz;h[P1#Xf7NX;~ P0` x8^-/p$~. 0PqPRNQKȄA'\"~2<((iX88!7>BRc7d[Q˗Ψxх](v%t4,%GJHBPmXqx@H! F(0,x戎5=FE|AWt/^6!*i/@q]yD)rяVY1qz I9kRv $3$Cyd6ٔ\Alт ADFZىX,x[YT#VShirY #csؕXA yHQɐ)Izv|WYcq0QIRyᚋAo[s7)x5 !ᛱ x>G)3q,@mb/ji`GܩHHǨI{wٝGٞy~Y-D',ֆP s +fMCR4 ie@p(*,ڢ*J JS-1A,7{(B9A)E17)5 ̓)*0,Х] B[֡:!dZ ze*lڦnpn1Y'6!s&%% FKsf 2#0Y;z#*3aeGQ:Z Qv"mPy!y"6KzYiJ59%$NqQvRvه:E)z3ú 'z凭oz }8HYG]J&ܚ ڑJ ۱!ᇭUGwZ$}  ~05Aj<YX:۬"kF۞I $} O .+4!R1q$=j!ES[g >kz":aM I#dOk}j~R@ؓ?)Q4ĸ",ҳfJj"˷ڞK Qem {ۻ͇wzK˞yЫ sتHsGu[K:1۾+7ۡ} {ڿ;{')ܻE [uYɽz̼"<ЇѾK; \4ܼ"jQ ?tYWcGK*L; ,,|&Lr:ћË.aZy+Zs!Xqlm _H0Yhj,y\x P \ɐ+Õqk *qRx/) I,Q\/l |̾"q,ՠtaxlAf9f{tx2pq5,EprAϲ<,` ̳.\L Ip=J0Q |p|@ʈK/`S(MʣP%SjU7d*fnO%<<nX!ϙ:kXKX2Y-YmM  $Kmz|~!A`*+/JKU+[z m?x*0P J%$&Q8 ;@vqkgXP$ bBP\X_M`Y\\60 }˛MW\l*q1a"3(rR,*ݦҏI}h~ ›<R`-M 9Re*pR@۟+AomȬ`H5hCXguᘪ*D%  Boȡߊc ӫ\à=m^- -ڽ*Vk `J0pŬLMLya 㰩!M"O'^#pފɨH8:>G:R,@]= z7 LnOշ]1٩m l* F$3;#y1g b^ydm,Qbv*%f%Cq--[q1Zj~-Р{ɥ-(.MО R0NUWop$ɸa\d3+j 0ŋϼQ!RP!C ;~8QǁLDRJ-]SL5USN4wS(L/@ʑM>8TTU 1F |-EF f\ɰ`Vm^Q#LJ! yU^TX_9}IB%=?vdDZEƜYI AW8j᎝ I4Tn-0f޽\^7mDq@>'kV2eՇ_f"@]Ȅ ޞޡ+q3j)Wjs ̞{b0@$@D0A0h#1k@h6@O=#7 "ܯA$ Qan+mzQѮ'F!HS1JBnΉ!4}lwv$ўk+mbs'Vioμ=EbtTHb˜vasu=qt+5 f}r۽zyC=%MYz:2F# ˭3 4 }[jy el1jtC`j[Cʧ?o-[2c3) KEp 0F/PG [+  bbY(є1 L P!a1P8%oOςRgDIPgHh;HS6* G=' pݢjy,l#sD6M VT`G< qf<9R+שJ0#7!%G^`dN Ot~vK(n&f6CЫEizFpP^nry s*[A=fԧIhVI5 "+XcRE S7N4EDEtӲEzrOŊP٥\zVJ"  h{X[PI?tضV$W gW5iZ."lhiUw1H} XRСdE=;1˂ &{mpcEڎ6+P s` 5vO_ Hmv΀R`Z[׼|".Ԣ /) xjپ"d햎B54hE˷VFG@m|7A2>ıI\ 誦[P5ȗe3geL– x;sPAxpȂ JkALRV7K+g.5ԕ|f G\UHm/o , DO .3xuV u3hsRri=fMXЃ&A< Xlc]5kV Ҙv an6ޒ>,4 +8TdsX#3hSwħBi !X̆򭞊dIÝsmNxVu)9O @qHmRvB62^E:p!t߅xm{}vʗv{Q`cBwʞ&ȼ!J ɬÿ0mvC Qn8V6HڕeKܚlw?wI?_[A \o ]_y?Ń*Ӂé2ko9p+B˹0?urrʸS@ 3 ?!@@` )BS?AB B"2\C>9 `%RA=#A+C5Ȟ'D HCq-@= B=RD쁁@77 QԉADD:X+E`B {4EĶ"e,@HH`VKFВFf**pD@7Ɠm IFPǪhƬ(D4LC|Gp{ ٹw$bF,W Hvŷ#s'hH&FCnԢ}4\~s#ÓxxLI-!<ˆ+;L&'VD@HX'݂ƁZr 0&:+L l#OIKH C- LG$H%0 +Iqtr5oIļ,<Ȥ̐[<L r6Ma L ҄8MQtͤ\ ּݬ  s4ɿMK,$CMTN<Ɍ't@"N$@N?ͅbL\OMBRQ8qO\,Z;O ,ρ(PwlcMcS9"S+c,!ͤ5d7؜pBфD6ZS_Vd (.LEFA)b" mі'kG~Uܠ j<ҭxdu` o\&noeFFp^p?2( Gpq7@nsmmW]ӀcVjw>!'" Fq{iKFop޿#NxZLh J(r?Dq&goLƈFn5g6󻄡`vA4]ݖ"q-v,h „ 2l!Ĉ'Rh"ƌ7r#ȁFB@$JCPBȅ2gҬi&Μ:Q^`B 4Mj jrPJeMQ@EZӐZr+ذbǒ-ћ=[ʀ-ܸKBXRȈt-Ww X2n1Ȓ'SI2+B\3hv_ƼIt$fЩCG `,l6ܺw}9gQ@Y/hLKk'P7ڷsN WDR::ϧ~zD)t\n>SDS(^{dAZvQSteql)(Vj!Vp#DZJĉ)&ah "/8ل = bE!I*$ f32gUC}YPFp&\`_GqI -} $yg~OTUv%\CXQ8" פFښ/`),C t'z*c]>0#'QB*K6(B :,zN@CtmedCz܃~k_}@g{/6D\ `  l $Bd &=`! [<2Ђ b  O1D1fzkpe<  ,aI+4\a QipDV0,<5J?D 4qˍ8hBs4 f+l6\hC=9啋L \̙ M0zٲ̈́I3~N7d뾻FV_}T[ zJ |;m.;3팻ۼ{=nS- /p&A+:C?DKXe?pGpO! =ߑ$ `\p}7H=EA@` DS CNbѰt~`sva-GN!qdu#dw+aDp' ⨊-Z2U/=oNpaM9qro3.gPTDtQ.eDCHELDzi|XH"$#C)|'kj\rqD>IY*-vHW[%-o)aꆓ(խr M~l@]=+ZEBU+zHr>A ʊ´5_$ׂu+`I@kmAx@25J0YF*eC+|H BRv}-Ԧ7)0@:;WHU8łټi6-[낷 EK󲷽{+_P$h~Wos V>h 3V ~_ UpT- TL= T$>1DMT,~q=] S416mS<(} !7RD>%PL~r ){OT嬌-Mm͝}f&` <5V+$9#u>sDZ0=àmʝA!+ ynУ9i Mx~H`=Cr=4ChU)F7ˬ5s-j4 <(:n5qTCd55R־6ms6-q>7ӭu~7-y[oӧ}3AX&.Dȼps^|Mg3hD9.rmB#?9S򕳼.Chd9nyJ91 &%;AdN 7gA/*vAKN ,c=6O@p@ 41 yΡEމMm 5] _hVnq_a  ` %  vA6 _  jz-ܼIFދᾤ@ 8mI)߁L !ڼ<"U`lƘE!!͜P% qEu ,eȐ-$t"!b"FD1M]b@WNq9iܲb @la('D)f%U| VNV! $ 1ͮl/D0!FL4̱9*aha,c 0rx#8ޢk9Ѣ0Ġ61. h4> @䕌ҀB Cb}DD@dd$K$NN$OO$G;kafbox-2.0/demo/krlst_split.m000066400000000000000000000140161303712747200163110ustar00rootroot00000000000000% Kernel Recursive Least-Squares Tracker algorithm % % S. Van Vaerenbergh, M. Lazaro-Gredilla, and I. Santamaria, "Kernel % Recursive Least-Squares Tracker for Time-Varying Regression," IEEE % Transactions on Neural Networks and Learning Systems, vol. 23, no. 8, pp. % 1313-1326, Aug. 2012, http://dx.doi.org/10.1109/TNNLS.2012.2200500 % % Comment: using back-to-the-prior forgetting. This version uses a separate % pruning function that can be triggered from outside, for demonstration % purposes. % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ classdef krlst_split < handle properties (GetAccess = 'public', SetAccess = 'private') lambda = .999; % forgetting factor sn2 = 1E-2; % noise to signal ratio M = 100; % dictionary size kerneltype = 'gauss'; % kernel type kernelpar = 1; % kernel parameter jitter = 1E-6; % jitter noise to avoid roundoff error end properties (GetAccess = 'public', SetAccess = 'private') dict = []; % dictionary Q = []; % inverse kernel matrix mu = []; % posterior mean Sigma = []; % posterior covariance nums02ML = 0; dens02ML = 0; s02 = 0; % signal power, adaptively estimated pruned = false; % flag reduced = false; % flag end methods function kaf = krlst_split(parameters) % constructor if (nargin > 0) % copy valid parameters for fn = fieldnames(parameters)', if ismember(fn,fieldnames(kaf)), kaf.(fn{1}) = parameters.(fn{1}); end end end end function [mean_test, var_test] = evaluate(kaf,x) % evaluate if size(kaf.dict,1)>0 k = kernel(kaf.dict,x,kaf.kerneltype,kaf.kernelpar); q = kaf.Q*k; mean_test = q'*kaf.mu; % predictive mean if nargout>1 ktt = kernel(x,x,[kaf.kerneltype '-diag'],kaf.kernelpar); sf2 = ktt + kaf.jitter + sum(k.*((kaf.Q*kaf.Sigma*kaf.Q-kaf.Q)*k),1)'; sf2(sf2<0) = 0; var_test = kaf.s02*(kaf.sn2 + sf2); % predictive variance end else mean_test = zeros(size(x,1),1); % prior var_test = nan(size(x,1),1); % signal scale is unknown end end function train(kaf,x,y) % train the algorithm m = size(kaf.Sigma,1); if m<1 % initialize k = kernel(x,x,kaf.kerneltype,kaf.kernelpar); k = k + kaf.jitter; kaf.Q = 1/k; kaf.mu = y*k/(k+kaf.sn2); kaf.Sigma = k - k^2/(k+kaf.sn2); kaf.dict = x; % dictionary bases kaf.nums02ML = y^2/(k+kaf.sn2); kaf.dens02ML = 1; kaf.s02 = kaf.nums02ML / kaf.dens02ML; else % forget K = kernel(kaf.dict,kaf.dict,kaf.kerneltype,kaf.kernelpar); kaf.Sigma = kaf.lambda*kaf.Sigma + (1-kaf.lambda)*K; % forget kaf.mu = sqrt(kaf.lambda)*kaf.mu; % forget % predict k = kernel([kaf.dict; x],x,kaf.kerneltype,kaf.kernelpar); kt = k(1:end-1); ktt = k(end) + kaf.jitter; q = kaf.Q*kt; y_mean = q'*kaf.mu; % predictive mean gamma2 = ktt - kt'*q; gamma2(gamma2<0)=0; % projection uncertainty h = kaf.Sigma*q; sf2 = gamma2 + q'*h; sf2(sf2<0)=0; % noiseless prediction variance sy2 = kaf.sn2 + sf2; % y_var = s02*sy2; % predictive variance % include new sample and add a basis Qold = kaf.Q; p = [q; -1]; kaf.Q = [kaf.Q zeros(m,1);zeros(1,m) 0] + 1/gamma2*(p*p'); p = [h; sf2]; kaf.mu = [kaf.mu; y_mean] + (y - y_mean)/sy2*p; % posterior mean kaf.Sigma = [kaf.Sigma h; h' sf2] - 1/sy2*(p*p'); % posterior covariance m = m + 1; kaf.dict = [kaf.dict; x]; % estimate s02 via ML kaf.nums02ML = kaf.nums02ML + kaf.lambda*(y - y_mean)^2/sy2; kaf.dens02ML = kaf.dens02ML + kaf.lambda; kaf.s02 = kaf.nums02ML/kaf.dens02ML; kaf.pruned = false; % delete a basis if necessary if (m>kaf.M || gamma2 end criterion = [ones(1,m-1) 0]; else % MSE pruning criterion errors = (kaf.Q*kaf.mu)./diag(kaf.Q); criterion = abs(errors); end [~, r] = min(criterion); % remove element r, which incurs in the minimum error smaller = 1:m; smaller(r) = []; if r == m, % if we must remove the element we just added, perform reduced update instead kaf.Q = Qold; kaf.reduced = true; else Qs = kaf.Q(smaller, r); qs = kaf.Q(r,r); kaf.Q = kaf.Q(smaller, smaller); kaf.Q = kaf.Q - (Qs*Qs')/qs; kaf.reduced = false; end kaf.mu = kaf.mu(smaller); kaf.Sigma = kaf.Sigma(smaller, smaller); kaf.dict = kaf.dict(smaller,:); kaf.pruned = true; end end end kafbox-2.0/demo/literature/000077500000000000000000000000001303712747200157375ustar00rootroot00000000000000kafbox-2.0/demo/literature/liu2010kernel/000077500000000000000000000000001303712747200202345ustar00rootroot00000000000000kafbox-2.0/demo/literature/liu2010kernel/fig2_12.m000066400000000000000000000037571303712747200215570ustar00rootroot00000000000000% Reproduces figure 2.12 from "Kernel Adaptive Filtering: A Comprehensive % Introduction". % % Comparison of the performance of LMS and KLMS in nonlinear channel % equalization. Execution time: < 1 minute (Intel Pentium Core2 Duo). % % Weifeng Liu, Jose C. Principe and Simon Haykin, "Kernel Adaptive % Filtering: A Comprehensive Introduction", Wiley, 2010. % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab % https://github.com/steven2358/kafbox/ clear close all %% PARAMETERS embedding = 5; % input embedding delay = 2; % equalization delay N_tr = 1000; % number of training data N_te = 50; % number of testing data setups{1} = lms(struct('mu',.01)); setups{2} = klms(struct('eta',0.2,'M',N_tr,'kerneltype','gauss','kernelpar',1)); %% PREPARE DATA fprintf('Fig. 2.12 from "Kernel Adaptive Filtering: A Comprehensive\n'); fprintf('Introduction".\n') u = randn(N_tr+N_te+embedding-1,1)>0; u = 2*u-1; % binary input z = u + 0.5*[0;u(1:end-1)]; % output of linear channel ns = 0.4*randn(length(u),1); % channel noise y = z - 0.9*z.^2 + ns; % output of nonlinear channel X_all = zeros(N_tr+N_te,embedding); % time-embedding for k=1:embedding, X_all(:,k) = y(k:N_tr+N_te+k-1); end X_tr = X_all(1:N_tr,:); % training input data X_te = X_all(N_tr+1:N_tr+N_te,:); % test input data T_tr = u(delay:delay+N_tr-1); % training desired output T_te = u(delay+N_tr:delay+N_tr+N_te-1); % training desired output %% RUN ALGORITHMS num_setup = length(setups); MSE = zeros(N_tr,num_setup); for setup_ind = 1:num_setup, kaf = setups{setup_ind}; for n=1:N_tr if ~mod(n,floor(N_tr/10)), fprintf('.'); end % progress indicator, 10 dots t_te = kaf.evaluate(X_te); % test on test set err = T_te - t_te; MSE(n,setup_ind) = mean(err.^2); kaf.train(X_tr(n,:),T_tr(n)); % train with one input-output pair end fprintf('\n'); end %% OUTPUT figure plot(MSE,'LineWidth',2) legend('LMS','KLMS') xlabel('iteration') ylabel('testing MSE') kafbox-2.0/demo/literature/liu2010kernel/liu2010kernel.bib000066400000000000000000000003211303712747200232030ustar00rootroot00000000000000@book{liu2010kernel, title={Kernel Adaptive Filtering: A Comprehensive Introduction}, publisher={Wiley}, author={Liu, Weifeng and Pr\'incipe, Jos\'e C. and Haykin, Simon}, year={2010}, pages={209} } kafbox-2.0/demo/literature/richard2009online/000077500000000000000000000000001303712747200210735ustar00rootroot00000000000000kafbox-2.0/demo/literature/richard2009online/fig1.m000066400000000000000000000040661303712747200221050ustar00rootroot00000000000000% Partially reproduces figure 1 from "Online Prediction of Time Series Data % With Kernels". (3 algorithms, 25 MC simulations.) % % Learning curves for KNLMS, NORMA, and KRLS on a nonlinear system. % Execution time: 2.5 minutes (Intel Pentium Core2 Duo). % % C. Richard, J.C.M. Bermudez, and P. Honeine, "Online Prediction of Time % Series Data With Kernels," IEEE Transactions on Signal Processing, % vol. 57, no. 3, pp. 1058-1067, March 2009, % http://dx.doi.org/10.1109/TSP.2008.2009895 % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab % https://github.com/steven2358/kafbox/ clear close all %% PARAMETERS N = 3000; % number of training data points ktype = 'gauss'; kpar = 1/sqrt(2*3.73); setups{1} = norma(struct('lambda',0.98,'tau',38,'eta',1,'tcoeff',-1/2,'kerneltype',ktype,'kernelpar',kpar)); setups{2} = knlms(struct('eta',.09,'eps',0.03,'mu0',0.5,'kerneltype',ktype,'kernelpar',kpar)); setups{3} = krls(struct('nu',.6,'kerneltype',ktype,'kernelpar',kpar)); numsim = 25; %% RUN ALGORITHMS t1 = tic; fprintf('Fig. 1 from "Online Prediction of Time Series Data With Kernels".\n'); num_setup = length(setups); MSE = zeros(N,num_setup); titles = cell(num_setup,1); for sim_ind = 1:numsim, fprintf('SIM %d:\n',sim_ind) % Generate the data [X,y,yref] = generate_doddbench(N); for setup_ind = 1:num_setup, kaf = setups{setup_ind}; titles{setup_ind} = upper(class(kaf)); for n=1:N if ~mod(n,floor(N/10)), fprintf('.'); end % progress indicator, 10 dots y_est = kaf.evaluate(X(n,:)); % test on test set err = yref(n) - y_est; MSE(n,setup_ind) = MSE(n,setup_ind) + err.^2/numsim; kaf.train(X(n,:),y(n)); % train with one input-output pair end fprintf('\n'); end end toc(t1) %% OUTPUT % MSE smoothing by moving average for visualization MSE_smooth = filter(1/20*ones(20,1),1,MSE); figure plot(10*log10(MSE_smooth)); title('Learning curves') grid on xlabel('iteration') ylabel('MSE (dB)') legend(titles) kafbox-2.0/demo/literature/richard2009online/fig2.m000066400000000000000000000043771303712747200221130ustar00rootroot00000000000000% Partially reproduces figure 2 from "Online Prediction of Time Series Data % With Kernels". (5 algorithms, 20 MC simulations.) % % Learning curves for KAP, KNLMS, NORMA and KRLS on a nonlinear system. % Execution time: 3.5 minutes (Intel Pentium Core2 Duo). % % C. Richard, J.C.M. Bermudez, and P. Honeine, "Online Prediction of Time % Series Data With Kernels," IEEE Transactions on Signal Processing, % vol. 57, no. 3, pp. 1058-1067, March 2009, % http://dx.doi.org/10.1109/TSP.2008.2009895 % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab % https://github.com/steven2358/kafbox/ clear close all %% PARAMETERS N = 3000; % number of training data points ktype = 'laplace'; kpar = 0.35; setups{1} = norma(struct('lambda',0.09,'tau',35,'eta',0.09,'tcoeff',-1/2,'kerneltype',ktype,'kernelpar',kpar)); setups{2} = knlms(struct('eta',.01,'eps',9E-4,'mu0',0.3,'kerneltype',ktype,'kernelpar',kpar)); setups{3} = kap(struct('p',2,'eta',.009,'eps',7e-2,'mu0',0.3,'kerneltype',ktype,'kernelpar',kpar)); setups{4} = kap(struct('p',3,'eta',.01,'eps',7e-2,'mu0',0.3,'kerneltype',ktype,'kernelpar',kpar)); setups{5} = krls(struct('nu',.7,'kerneltype',ktype,'kernelpar',kpar)); numsim = 20; %% RUN ALGORITHMS t1 = tic; fprintf('Fig. 2 from "Online Prediction of Time Series Data With Kernels".\n'); num_setup = length(setups); MSE = zeros(N,num_setup); titles = cell(num_setup,1); for sim_ind = 1:numsim, fprintf('SIM %d:\n',sim_ind) % Generate the data [X,y,yref] = generate_richardbench(N); for setup_ind = 1:num_setup, kaf = setups{setup_ind}; titles{setup_ind} = upper(class(kaf)); for n=1:N if ~mod(n,floor(N/10)), fprintf('.'); end % progress indicator, 10 dots y_est = kaf.evaluate(X(n,:)); % test on test set err = yref(n) - y_est; MSE(n,setup_ind) = MSE(n,setup_ind) + err.^2/numsim; kaf.train(X(n,:),y(n)); % train with one input-output pair end fprintf('\n'); end end toc(t1) %% OUTPUT % MSE smoothing by moving average for visualization MSE_smooth = filter(1/20*ones(20,1),1,MSE); figure plot(10*log10(MSE_smooth)); title('Learning curves') grid on xlabel('iteration') ylabel('MSE (dB)') legend(titles) kafbox-2.0/demo/literature/richard2009online/generate_doddbench.m000066400000000000000000000017351303712747200250430ustar00rootroot00000000000000function [v,d,dref] = generate_doddbench(N) % Generate DODDBENCH signal. % % Benchmark signal introduced in Dodd, T.J., Kadirkamanathan, V. and % Harrison, R.F., "Function estimation in Hilbert space using sequential % projections," Proc. of the IFAC Conf. on Intelligent Control Systems and % Signal Processing, 113-118, 2003. % % Comment: copyright Cedric Richard, http://cedric-richard.fr/ % % Input: N: number of data points % % Outputs: v: input sequence (2-dimensional sequence [v(:,1);v(:,2)]) % d: noisy desired output (1-dimensional sequence) % dref: noise-free desired output % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ dref = zeros(1,N+2); dref(1:2)=[0.1 0.1]; for t=3:N+2, dref(t) = (0.8-0.5*exp(-dref(t-1)^2))*dref(t-1) - ... (0.3+0.9*exp(-dref(t-1)^2))*dref(t-2)+0.1*sin(pi*dref(t-1)); end d = dref + 0.1*randn(1,N+2); v = [d(1:N); d(2:N+1)]'; d(1:2)=[]; dref(1:2)=[]; kafbox-2.0/demo/literature/richard2009online/generate_richardbench.m000066400000000000000000000015761303712747200255500ustar00rootroot00000000000000function [u,d,dref] = generate_richardbench(N) % Generate RICHARDBENCH signal. % % Benchmark signal introduced in C. Richard, J.C.M. Bermudez, P. Honeine, % "Online Prediction of Time Series Data With Kernels," IEEE Transactions % on Signal Processing, vol.57, no.3, pp.1058,1067, March 2009. % % Comment: copyright Cedric Richard, http://cedric-richard.fr/ % % Input: N: number of data points % % Outputs: u: input sequence (1-dimensional sequence [u(:,1)]) % d: noisy desired output (1-dimensional sequence) % dref: noise-free desired output % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ v = zeros(1,N+1); v(1) = 0.5; u = 0.25*randn(N+1,1); dref = zeros(1,N+1); for t=2:N+1, v(t) = 1.1*exp(-abs(v(t-1)))+u(t); dref(t) = v(t)^2; end d = dref+randn(1,N+1); d(1) = []; dref(1) = []; u(1) = []; kafbox-2.0/demo/literature/richard2009online/richard2009online.bib000066400000000000000000000004601303712747200247050ustar00rootroot00000000000000@ARTICLE{richard2009online, author={Richard, C{\'e}dric and Bermudez, Jos{\'e} Carlos M and Honeine, Paul}, journal=IEEE Transactions on Signal Processing}, title={Online Prediction of Time Series Data With Kernels}, year={2009}, month=mar, volume={57}, number={3}, pages={1058--1067}, } kafbox-2.0/demo/literature/vanvaerenbergh2006sliding/000077500000000000000000000000001303712747200226165ustar00rootroot00000000000000kafbox-2.0/demo/literature/vanvaerenbergh2006sliding/fig2.m000066400000000000000000000051061303712747200236250ustar00rootroot00000000000000% Reproduces an experiment similar to the one from figure 2 in "A % sliding-window kernel RLS algorithm and its application to nonlinear % channel identification". % % MSE performance of sliding-window kernel RLS for two different window % sizes. Execution time: < 1 minute (Intel Pentium Core2 Duo). % % S. Van Vaerenbergh, J. Via, and I. Santamaria, "A Sliding-Window Kernel % RLS Algorithm and Its Application to Nonlinear Channel Identification," % 2006 IEEE International Conference on Acoustics, Speech and Signal % Processing (ICASSP), May 2006, % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab % https://github.com/steven2358/kafbox/ clear close all %% PARAMETERS opt.chlen = 4; % channel length, and input embedding opt.N = 1500; % number of training data opt.N_switch = 500; % iteration after which the channel switch occurs opt.N_test = 100; % number of test data opt.SNR = 20; % SNR in dB num_sim = 10; setups{1} = struct('M',75,'c',1E-2,'kerneltype','gauss','kernelpar',5); setups{2} = struct('M',150,'c',1E-2,'kerneltype','gauss','kernelpar',5); titles = {'SW-KRLS, M=75','SW-KRLS, M=150'}; %% PREPARE DATA fprintf('Fig. 2 from "A sliding-window kernel RLS algorithm and its \n'); fprintf('application to nonlinear channel identification".\n'); % perform online learning for each algorithm fprintf('\n') N = opt.N; N_switch = opt.N_switch; num_alg = length(setups); MSE = zeros(N,num_alg); MSE_final = zeros(1,num_alg); for sim_ind = 1:num_sim, % generate data [X,y,y_ref,X_test,y_test] = generate_channel_switch(opt); fprintf('SIM %d:\n',sim_ind) for algo_ind = 1:num_alg t1 = tic; kaf = swkrls(setups{algo_ind}); fprintf('%9s M=%3d: ',upper(class(kaf)),kaf.M); mse = zeros(N,1); for i=1:N, if ~mod(i,floor(N/10)), fprintf('.'); end y_est = kaf.evaluate(X_test); if i<=N_switch mse(i) = mean((y_test(:,1)-y_est).^2); else mse(i) = mean((y_test(:,2)-y_est).^2); end kaf.train(X(i,:),y(i)); end MSE(:,algo_ind) = MSE(:,algo_ind) + mse/num_sim; mse_final = mean(mse(N-500:N)); fprintf(' %.2fs. Final MSE=%3.2fdB\n',toc(t1),... 10*log10(mse_final)) end fprintf('\n'); end fprintf('\n'); %% OUTPUT figure; hold all for algo_ind=1:num_alg plot(10*log10(MSE(:,algo_ind)),'LineWidth',1) axis([0 N 5*floor(min(10*log10(MSE(:)))/5) 0]); end xlabel('iteration') ylabel('MSE (dB)') legend(titles) kafbox-2.0/demo/literature/vanvaerenbergh2006sliding/vanvaerenbergh2006sliding.bib000066400000000000000000000006621303712747200301570ustar00rootroot00000000000000@inproceedings{vanvaerenbergh2006sliding, title={A sliding-window kernel {RLS} algorithm and its application to nonlinear channel identification}, author={Van Vaerenbergh, Steven and V\'ia, Javier and Santamar\'ia, Ignacio}, booktitle={2006 IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP)}, year = {2006}, address = {Toulouse, France}, month = may, pages = {789--792}, volume = 5, } kafbox-2.0/demo/literature/vanvaerenbergh2012kernel/000077500000000000000000000000001303712747200224425ustar00rootroot00000000000000kafbox-2.0/demo/literature/vanvaerenbergh2012kernel/fig3.m000066400000000000000000000064701303712747200234570ustar00rootroot00000000000000% Partially reproduces figure 3 from "Kernel Recursive Least-Squares % Tracker for Time-Varying Regression". (Only 3 algorithms, only 1 MC % simulation.) % % MSE performance comparison of different tracking algorithms on a % communications channel that shows an abrupt change at iteration 500. % Execution time: < 1 minute (Intel Pentium Core2 Duo). % % S. Van Vaerenbergh, M. Lazaro-Gredilla, and I. Santamaria, "Kernel % Recursive Least-Squares Tracker for Time-Varying Regression," IEEE % Transactions on Neural Networks and Learning Systems, vol. 23, no. 8, pp. % 1313-1326, Aug. 2012, http://dx.doi.org/10.1109/TNNLS.2012.2200500 % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab % https://github.com/steven2358/kafbox/ clear close all %% PARAMETERS N = 1500; % number of training data points N_test = 500; % number of test data points Nswitch = 500; % switch from model 1 to model 2 after Nswitch iterations B1 = [1.0000, -0.3817, -0.1411, 0.5789, 0.191]; % model 1 linear filter B2 = [1.0000, -0.0870, 0.9852, -0.2826, -0.1711]; % model 2 linear filter f = @(x) tanh(x); % Wiener system nonlinearity SNR = 20; % SNR in dB embedding = 5; % time-embedding setups{1} = swkrls(struct('c',1E-2,'M',50,'kerneltype','gauss','kernelpar',1)); setups{2} = krlst(struct('lambda',.999,'M',50,'sn2',1E-2,'kerneltype','gauss','kernelpar',1)); setups{3} = norma(struct('lambda',1E-2,'tau',1500,'eta',0.1,'kerneltype','gauss','kernelpar',1)); % large tau -> very slow %% PREPARE DATA fprintf('Fig. 3 from "Kernel Recursive Least-Squares Tracker for\n'); fprintf('Time-Varying Regression".\n') % generate Gaussian input data s s = rand(N+N_test,1); s_mem = zeros(N+N_test,embedding); for i = 1:embedding, s_mem(i:N+N_test,i) = s(1:N+N_test-i+1); % time-embedding end s = s_mem(1:N+N_test,:); % input data, stored in columns s_train = s_mem(1:N,:); % input train data, stored in columns s_test = s_mem(N+1:N+N_test,:); % input test data, stored in columns % generate internal data x and output y X1 = s_mem(1:Nswitch,:)*B1'; X2 = s_mem(Nswitch+1:N,:)*B2'; X = [X1;X2]; Y_nn = f(X); % noiseless Y vary = var(Y_nn); noisevar = 10^(-SNR/10)*vary; noise = sqrt(noisevar)*randn(N,1); Y = Y_nn + noise; % noisy output data X_test1 = s_mem(N+1:N+N_test,:)*B1'; X_test2 = s_mem(N+1:N+N_test,:)*B2'; noise_test1 = sqrt(noisevar)*randn(N_test,1); noise_test2 = sqrt(noisevar)*randn(N_test,1); Y_test1 = f(X_test1) + noise_test1; % noisy output test data, model 1 Y_test2 = f(X_test2) + noise_test2; % noisy output test data, model 2 %% RUN ALGORITHMS t1 = tic; num_setup = length(setups); MSE = zeros(N,num_setup); titles = cell(num_setup,1); for setup_ind=1:length(setups) t2 = tic; kaf = setups{setup_ind}; titles{setup_ind} = upper(class(kaf)); fprintf('%s\t',titles{setup_ind}); for n=1:N, if ~mod(n,round(N/10)), fprintf('.'); end if n<=Nswitch, Y_test = Y_test1; else Y_test = Y_test2; end Y_est = kaf.evaluate(s_test); % test on test set err = Y_test - Y_est; MSE(n,setup_ind) = mean(err.^2); kaf.train(s_train(n,:),Y(n)); % train with one input-output pair end fprintf(' %.2f seconds\n',toc(t2)); end toc(t1) %% OUTPUT figure plot(10*log10(MSE),'LineWidth',2); grid on xlabel('iteration') ylabel('MSE (dB)') legend(titles) kafbox-2.0/demo/literature/vanvaerenbergh2012kernel/vanvaerenbergh2012kernel.bib000066400000000000000000000006601303712747200276250ustar00rootroot00000000000000@ARTICLE{vanvaerenbergh2012kernel, title={Kernel Recursive Least-Squares Tracker for Time-Varying Regression}, author={Van Vaerenbergh, Steven and L{\'a}zaro-Gredilla, Miguel and Santamar{\'i}a, Ignacio}, journal={IEEE Transactions on Neural Networks and Learning Systems}, year={2012}, month=aug, volume={23}, number={8}, pages={1313--1326}, keywords={}, doi={10.1109/TNNLS.2012.2200500}, ISSN={2162-237X}, } kafbox-2.0/demo/literature/yukawa2012multikernel/000077500000000000000000000000001303712747200220215ustar00rootroot00000000000000kafbox-2.0/demo/literature/yukawa2012multikernel/fig1a.m000066400000000000000000000053151303712747200231720ustar00rootroot00000000000000% Almost reproduces figure 1a from "Multikernel Adaptive Filtering". % % Comparison of the performance of LMS, KNLMS and MKNLMS-CS in nonlinear % channel equalization. Execution time: < 1 minute (Intel Pentium Core2 % Duo). % % Masahiro Yukawa, "Multikernel Adaptive Filtering", IEEE Transactions on % Signal Processing, vol.60, no.9, pp.4672-4682, Sept. 2012. % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab % https://github.com/steven2358/kafbox/ clear close all %% PARAMETERS N_tr = 3000; % number of training data N_te = 500; % number of testing data SNR = 20; embedding = 5; % input embedding delay = 2; % equalization delay mk_thresh = 0.36; % threshold value from paper % mk_thresh = 0.9; % alternative value mk_kernelpars = [.2 2]; % kernel parameter values from paper % mk_kernelpars = [.2 .5 1 2 5 10 20 50]; % alternative values % remark 1 from paper determines mu0 setups{1} = lms(struct('mu',.01)); setups{2} = knlms(struct('mu0',mk_thresh^(.2/min(mk_kernelpars)),'eta',0.2,'eps',3E-2,'kerneltype','gauss','kernelpar',1/sqrt(.2))); setups{3} = knlms(struct('mu0',mk_thresh^(.5/min(mk_kernelpars)),'eta',0.2,'eps',3E-2,'kerneltype','gauss','kernelpar',1/sqrt(.5))); setups{4} = mknlms_cs(struct('delta',mk_thresh,'eta',0.2,'rho',6E-2,'kerneltype','gauss','kernelpars',1./sqrt(mk_kernelpars))); %% PREPARE DATA fprintf('Fig. 1a from "Multikernel Adaptive Filtering".\n') u = randn(N_tr+N_te+embedding-1,1)>0; u = 2*u-1; % binary input z = u + 0.5*[0;u(1:end-1)]; % output of linear channel varz = var(z); noisevar = 10^(-SNR/10)*varz; ns = sqrt(noisevar)*randn(length(u),1); % channel noise y = z - 0.9*z.^2 + ns; % output of nonlinear channel X_all = zeros(N_tr+N_te,embedding); % time-embedding for k=1:embedding, X_all(:,k) = y(k:N_tr+N_te+k-1); end X_tr = X_all(1:N_tr,:); % training input data X_te = X_all(N_tr+1:N_tr+N_te,:); % test input data T_tr = u(delay:delay+N_tr-1); % training desired output T_te = u(delay+N_tr:delay+N_tr+N_te-1); % training desired output %% RUN ALGORITHMS tic num_setup = length(setups); MSE = zeros(N_tr,num_setup); for setup_ind = 1:num_setup, kaf = setups{setup_ind}; for n=1:N_tr if ~mod(n,floor(N_tr/10)), fprintf('.'); end % progress indicator, 10 dots t_te = kaf.evaluate(X_te); % test on test set err = T_te - t_te; MSE(n,setup_ind) = mean(err.^2); kaf.train(X_tr(n,:),T_tr(n)); % train with one input-output pair end if setup_ind == 1 fprintf('\n'); else fprintf(' Dict. size: %d\n',size(kaf.dict,1)); end end toc %% OUTPUT figure semilogy(MSE,'LineWidth',2); grid on legend('LMS','KNLMS 1','KNLMS 2','MKNLMS-CS') xlabel('Iteration number') ylabel('MSE') kafbox-2.0/demo/literature/yukawa2012multikernel/yukawa2012multikernel.bib000066400000000000000000000004461303712747200265650ustar00rootroot00000000000000@article{yukawa2012multikernel, author = {Yukawa, Masahiro}, journal = {IEEE Transactions on Signal Processing}, title = {Multikernel Adaptive Filtering}, year = {2012}, volume = {60}, number = {9}, pages = {4672-4682}, doi = {10.1109/TSP.2012.2200889}, ISSN = {1053-587X}, } kafbox-2.0/demo/run_all_demos.m000066400000000000000000000033261303712747200165640ustar00rootroot00000000000000% Script to run all demos consecutively. % % Author: Steven Van Vaerenbergh, 2015. % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox clc close all clear % get list of demo scripts fdir = fileparts(which('run_all_demos.m')); files_demo = dir(fullfile(fdir,'demo_*.m')); files = files_demo; folders = repmat({fdir},length(files),1); % get list of literature scripts folders_literature = dir(fullfile(fdir,'literature','*2*')); for i=1:length(folders_literature); folder_i = fullfile(fdir,'literature',folders_literature(i).name); files_i = dir(fullfile(folder_i,'fig*.m')); folders_i = repmat({folder_i},length(files_i),1); files = [files; files_i]; %#ok folders = [folders; folders_i]; %#ok end [~,files] = cellfun(@fileparts, {files.name}, 'UniformOutput',0); t1 = tic; fprintf('\n') for i=1:length(files) close all clear eval save(fullfile(tempdir,'temp.mat'),'i','folders','files','fdir',... 't1'); % memory map try % run script cd(folders{i}) fname_demo = files{i}; fprintf('\nRunning %s\n',fname_demo); eval(fname_demo); catch err % return to demo folder load(fullfile(tempdir,'temp.mat')); cd(fdir); % report error me = err.stack(1); error(['Error in ',... '',... '%s at %d\n',... '%s'],... me.file, me.line, me.name, me.line,... err.message); end load(fullfile(tempdir,'temp.mat')); cd(fdir); end delete(fullfile(tempdir,'temp.mat')); toc(t1) close all fprintf('\nAll tests finished.\n') kafbox-2.0/doc/000077500000000000000000000000001303712747200134005ustar00rootroot00000000000000kafbox-2.0/doc/.gitignore000066400000000000000000000005041303712747200153670ustar00rootroot00000000000000################# ## Latex ################# *.acn *.acr *.alg *.aux *.backup *.bbl *.blg *.brf *.dvi *.fdb_latexmk *.fls *.glg *.glo *.gls *.idx *.ilg *.ind *.ist *.loa *.lof *.log *.lol *.lot *.maf *.mtc *.mtc0 *.nav *.nlo *.out *.pdfsync *.ps *.snm *.synctex *.synctex.gz *.tdo *.toc *.vrb *.xdy *.dpth *.auxlock *.md5 kafbox-2.0/doc/extra.bib000066400000000000000000000064011303712747200152020ustar00rootroot00000000000000@STRING{ieee-mlsp = {IEEE International Workshop on Machine Learning for Signal Processing (MLSP)}} @STRING{ieee-icassp = {IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP)}} @STRING{ieee-ijcnn = {IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP)}} @INPROCEEDINGS{singh2012random, author = {Singh, Abhishek and Ahuja, Narendra and Moulin, Pierre}, booktitle = {2012 } # ieee-mlsp, title = {Online learning with kernels: Overcoming the growing sum problem}, year = {2012}, month = sep, pages = {1-6}, doi = {10.1109/MLSP.2012.6349811}, } @INPROCEEDINGS{takizawa2013efficient, author = {Takizawa, M.-A. and Yukawa, M.}, booktitle = {2013 } # ieee-icassp, title = {An efficient data-reusing kernel adaptive filtering algorithm based on Parallel HYperslab Projection along Affine Subspaces}, year = {2013}, month = may, pages = {3557-3561}, doi = {10.1109/ICASSP.2013.6638320} } @INPROCEEDINGS{gao2013kernel, author = {Wei Gao and Jie Chen and Richard, C. and Jianguo Huang and Flamary, R.}, booktitle = {2013 } # ieee-icassp, title = {Kernel LMS algorithm with forward-backward splitting for dictionary learning}, year = {2013}, month = may, pages = {5735-5739}, doi = {10.1109/ICASSP.2013.6638763} } @INPROCEEDINGS{pokharel2013mixture, author = {Pokharel, R. and Seth, S. and Principe, J.C.}, booktitle = {2013 } # ieee-ijcnn, title = {Mixture kernel least mean square}, year = {2013}, month = aug, pages = {1-7}, doi = {10.1109/IJCNN.2013.6706867} } @misc{wiki:online_machine_learning, author = "Wikipedia", title = "Online machine learning --- Wikipedia{,} The Free Encyclopedia", year = "2014", url = "http://en.wikipedia.org/w/index.php?title=Online_machine_learning&oldid=610152751", note = "[Online; accessed 26-May-2014]" } @article{wilson2014best, title = {Best practices for scientific computing}, author = {Wilson, Greg and Aruliah, DA and Brown, C Titus and Hong, Neil P Chue and Davis, Matt and Guy, Richard T and Haddock, Steven HD and Huff, Kathryn D and Mitchell, Ian M and Plumbley, Mark D and others}, journal = {PLoS biology}, volume = {12}, number = {1}, pages = {e1001745}, year = {2014}, publisher = {Public Library of Science} } @incollection{blum1998line, title = {On-line algorithms in machine learning}, booktitle = {Online Algorithms: The State of the Art}, chapter = 14, author = {Blum, Avrim}, year = {1998}, publisher = {Springer Berlin Heidelberg}, series = {Lecture Notes in Computer Science}, volume={1442}, editor = {Fiat, Amos and Woeginger, Gerhard J.}, doi = {10.1007/BFb0029575}, pages={306-325} } @INPROCEEDINGS{fernandez2015probabilistic, author={Fern\'andez-Bes, Jesus and Elvira, V\'ictor and Van Vaerenbergh, Steven}, booktitle = {2015 } # ieee-icassp, title={A probabilistic least-mean-squares filter}, year={2015}, month = apr, pages={2199--2203}, doi={10.1109/ICASSP.2015.7178361}, ISSN={1520-6149} } @article{kalman1960new, title={A new approach to linear filtering and prediction problems}, author={Kalman, Rudolph Emil}, journal={Journal of basic Engineering}, volume={82}, number={1}, pages={35--45}, year={1960}, publisher={American Society of Mechanical Engineers} } kafbox-2.0/doc/fig/000077500000000000000000000000001303712747200141455ustar00rootroot00000000000000kafbox-2.0/doc/fig/kafbox_logo.pdf000066400000000000000000000023161303712747200171340ustar00rootroot00000000000000%PDF-1.5 % 3 0 obj << /Length 4 0 R /Filter /FlateDecode >> stream xN@E ;Y4HHEHJgƱxD [zv^gz_y68ؼ$iٲW6eNK0u7[$^Er{b>FE=k\]NT ;Rf26laN :'hϧ.3txvhD5p#`˕β4Y 6ZZac3N $!ldP, iD>`B>׎h aҡF3C'AST'v9PYZk%t%9 l@L' ɲ ԣIbeU< endstream endobj 4 0 obj 333 endobj 2 0 obj << /ExtGState << /a0 << /CA 1 /ca 1 >> >> >> endobj 5 0 obj << /Type /Page /Parent 1 0 R /MediaBox [ 0 0 51.200001 51.200001 ] /Contents 3 0 R /Group << /Type /Group /S /Transparency /I true /CS /DeviceRGB >> /Resources 2 0 R >> endobj 1 0 obj << /Type /Pages /Kids [ 5 0 R ] /Count 1 >> endobj 6 0 obj << /Creator (cairo 1.14.1 (http://cairographics.org)) /Producer (cairo 1.14.1 (http://cairographics.org)) >> endobj 7 0 obj << /Type /Catalog /Pages 1 0 R >> endobj xref 0 8 0000000000 65535 f 0000000745 00000 n 0000000447 00000 n 0000000015 00000 n 0000000425 00000 n 0000000519 00000 n 0000000810 00000 n 0000000937 00000 n trailer << /Size 8 /Root 7 0 R /Info 6 0 R >> startxref 989 %%EOF kafbox-2.0/doc/fig/unican_logo.pdf000066400000000000000000000074631303712747200171470ustar00rootroot00000000000000%PDF-1.4 % 3 0 obj << /Length 4 0 R /Filter /FlateDecode >> stream xmYI,)y8`4,ioa#r -A ^On?fq%>Esܿs~śauOm?wOOh宐=[[h>#8fޮ} ! On )=?ǻg~̔/ O@y V8(|Î 1fS{´}םSJ4X˓KSc|\9Ru_pR6̘8,7wfnO6;Ag#.>P baX qA?LE ^Ցd೸PNR]%١a T#G1HB0s=0hTDVZ5Id8ou_&o_|歁 g*|32"{ C8ꞇST(oGccPks/Toœ)_pi #ϛ*ዎ'/lOiQoL8xW|ӝ/B!MP?npq-鸶=9FER Ū'jf҅b(W!C.#R {."g "(6`=:+>qV f9֗~b( #.ģt:&Ԍ"CZen o4N6&rˎ"JZaC"|}rdY@1& HfcEH`v59;wV;v6iM4w.AAFb5(:tVCb,\QR^Oj+D 00VZr˓4FR)*If"a!֙QBZ_Ut!YeӚa%4 Y֝܋R[vYH1\YᕢE{qX -F$6vpPzúNrӭ;@hM5Wސ("sú$RcK\ߊHr.cg5Y"P5DX9jeT֔ %_JlnheW.1\w;+"Jz[Z?dޣtPʖ.!B he9Ukəb6Ɲz/v$U1 Yi5\/\hAC.iF, [u{,kqJVgS1]|_ ls|%E `]/!s"v` ?"ײK66<yr)5?v(k5 \6QݤF}쑪M/ ӚmX{᣾ 1{ ce6]U#iJYL 15sUY/ʖ jP6]B4qnba띮ZLn7ޗhl8Ϊ-!#] ,8-vh(wioHɮECQ4o 8NxzZ8[`g%hL*=L&# $` t#E$3K39Ȯ HF.$)l0.C~(UjbʲvPj!mmΘ:BFa8Lʵi *n-H '`voBl{97|cӇVMp.50k(2Fqr]5I8#I !JuG7B#J&< 7*ѓ06$l&5KST|,w4)01˖ڲF$Q~a]HTª4:0S KAvK>\ .V &Q%AةQ,75746$2a 'i4ZJ[~1|u@;JL?`]7UIL&yDž-M8w^OHy8wòɷdZ{+8.|$N¡ƞ0/6iΑs;t y7^G =طf0ٴC7t*P^ovq^~mn*=OMt 1oj.~]k|?AMI]l 6Z>b\.03=J$ɧ/g8I=/a84jնNPS-OŚ ]~-Dy9&!)0pvXZCq|:{w7(SaRxYiUGwqcXE!zD~ \[%޴rMv J+! ":"_^,KLi,> >> >> endobj 5 0 obj << /Type /Page /Parent 1 0 R /MediaBox [ 0 0 141.399994 141.399994 ] /Contents 3 0 R /Group << /Type /Group /S /Transparency /CS /DeviceRGB >> /Resources 2 0 R >> endobj 1 0 obj << /Type /Pages /Kids [ 5 0 R ] /Count 1 >> endobj 6 0 obj << /Creator (cairo 1.11.2 (http://cairographics.org)) /Producer (cairo 1.11.2 (http://cairographics.org)) >> endobj 7 0 obj << /Type /Catalog /Pages 1 0 R >> endobj xref 0 8 0000000000 65535 f 0000003405 00000 n 0000003119 00000 n 0000000015 00000 n 0000003096 00000 n 0000003191 00000 n 0000003470 00000 n 0000003597 00000 n trailer << /Size 8 /Root 7 0 R /Info 6 0 R >> startxref 3649 %%EOF kafbox-2.0/doc/kaf.bib000066400000000000000000000400231303712747200146160ustar00rootroot00000000000000@STRING{nips = {Advances in Neural Information Processing Systems}} @STRING{ieee-tsp = {IEEE Transactions on Signal Processing}} @STRING{ieee-spm = {IEEE Signal Processing Magazine}} @STRING{ieee-tnn = {IEEE Transactions on Neural Networks}} @STRING{ieee-tnnls = {IEEE Transactions on Neural Networks and Learning Systems}} @STRING{ieee-mlsp = {IEEE International Workshop on Machine Learning for Signal Processing (MLSP)}} @STRING{ieee-icassp = {IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP)}} @article{aronszajn1950theory, title = {Theory of reproducing kernels}, author = {Aronszajn, Nachman}, journal = {Transactions of the {American} mathematical society}, volume = {68}, number = {3}, pages = {337--404}, year = {1950} } @book{rasmussen2006gaussian, author = {Rasmussen, Carl Edward and Williams, Christopher K. I.}, title = {{G}aussian Processes for Machine Learning}, year = {2006}, publisher = {MIT Press}, pages = {272} } @article{csato2002sparseonline, title = {Sparse Online {Gaussian} Processes}, volume = {14}, number = {3}, journal = {Neural Computation}, publisher = {Neural computing research group}, author = {Csat\'{o}, Lehel and Opper, Manfred}, year = {2002}, pages = {641-668} } @incollection{csato2002sparserepresentation, title = {Sparse representation for {Gaussian} process models}, author = {Csat{\'o}, Lehel and Opper, Manfred}, booktitle = nips # { 13}, pages = {444-450}, isbn = {0262122413}, year = {2001}, publisher = {MIT Press}, } @article{kulhavy1993forgetting, title = {On a general concept of forgetting}, author = {Kulhav\'y, Rudolf and Zarrop, Martin B.}, journal = {International Journal of Control}, volume = {58}, issue = {4}, pages = {905--924}, year = {1993} } @book{sayed2003fundamentals, title = {Fundamentals of Adaptive Filtering}, author = {Sayed, Ali H.}, isbn = {0471461261}, year = {2003}, publisher = {Wiley-IEEE Press} } @article{engel2004kernel, title = {The Kernel recursive least squares algorithm}, author = {Engel, Yaakov and Mannor, Shie and Meir, Ron}, year = {2004}, month = aug, volume = {52}, number = {8}, journal = ieee-tsp, pages = {2275--2285} } @book{liu2010kernel, title = {Kernel Adaptive Filtering: A Comprehensive Introduction}, publisher = {Wiley}, author = {Liu, Weifeng and Pr\'incipe, Jos\'e C. and Haykin, Simon}, year = {2010}, pages = {209} } @article{dekruif2003pruning, title = {Pruning error minimization in least squares support vector machines}, author = {De Kruif, Bas J. and De Vries, Theo J. A.}, journal = ieee-tnn, volume = {14}, number = {3}, pages = {696--702}, issn = {1045-9227}, year = {2003}, publisher = {IEEE} } @article{kuh2007comments, title = {Comments on ``{Pruning} error minimization in least squares support vector machines''}, author = {Kuh, Anthony and De Wilde, Philippe}, journal = ieee-tnn, volume = {18}, number = {2}, pages = {606--609}, issn = {1045-9227}, year = {2007}, publisher = {IEEE} } @inproceedings{vanvaerenbergh2010fixed, author = {Van Vaerenbergh, Steven and Santamar\'ia, Ignacio and Liu, Weifeng and Pr\'incipe, Jos\'e C.}, title = {Fixed-Budget Kernel Recursive Least-Squares}, booktitle = {2010 } # ieee-icassp, year = {2010}, address = {Dallas, TX, USA}, month = apr, } @incollection{dekel2006forgetron, author = {Dekel, Ofer and Shalev-Shwartz, Shai and Singer, Yoram}, title = {The Forgetron: A Kernel-Based Perceptron on a Fixed Budget}, booktitle = nips # { 18}, editor = {Y. Weiss and B. Sch\"{o}lkopf and J. Platt}, publisher = {MIT Press}, address = {Cambridge, MA}, pages = {259--266}, year = {2006} } @inproceedings{vanvaerenbergh2006sliding, title = {A sliding-window kernel {RLS} algorithm and its application to nonlinear channel identification}, author = {Van Vaerenbergh, Steven and V\'ia, Javier and Santamar\'ia, Ignacio}, booktitle = {2006 } # ieee-icassp, year = {2006}, address = {Toulouse, France}, month = may, pages = {789--792}, volume = 5, } @book{haykin2001adaptive, title = {Adaptive Filter Theory (4th Edition)}, publisher = {Prentice Hall}, year = {2001}, author = {Haykin, Simon}, month = sep, howpublished = {Hardcover}, isbn = {0130901261} } @article{kivinen2004online, author = {Kivinen, Jyrki and Smola, Alexander J. and Williamson, Robert C.}, title = {Online learning with kernels}, journal = ieee-tsp, year = {2004}, volume = {52}, pages = {2165--2176}, number = {8}, month = aug, doi = {10.1109/TSP.2004.830991} } @article{liu2009information, title = {An information theoretic approach of designing sparse kernel adaptive filters}, author = {Liu, Weifeng and Park, Il and Pr{\'i}ncipe, Jos\'{e} C.}, journal = ieee-tnn, volume = {20}, number = {12}, pages = {1950--1961}, issn = {1045-9227}, year = {2009}, publisher = {IEEE} } @article{slavakis2008sliding, title = {Sliding window generalized kernel affine projection algorithm using projection mappings}, author = {Slavakis, Konstantinos and Theodoridis, Sergios}, journal = {EURASIP Journal on Advances in Signal Processing}, volume = {16}, year = {2008}, publisher = {Hindawi Publishing Corporation} } @incollection{williams2001using, author = {Christopher Williams and Matthias Seeger}, title = {Using the {N}ystr\"{o}m Method to Speed Up Kernel Machines}, booktitle = nips, editor = {Leen, Todd K. and Dietterich, Thomas G. and Tresp, Volker}, vol = {13}, year = {2000}, pages = {682--688}, publisher = {MIT Press}, address = {Cambridge, MA}, } @article{lecun1990optimal, title = {Optimal brain damage}, author = {Le Cun, Yann and Denker, John S. and Solla, Sara A.}, journal = NIPS, volume = {2}, editor = {Touretzky, David S.}, number = {1}, pages = {598--605}, year = {1990}, publisher = {Morgan Kaufmann}, address = {San Mateo, CA}, } @article{sayed1994state, title = {A state-space approach to adaptive RLS filtering}, author = {Sayed, Ali H. and Kailath, Thomas}, journal = {IEEE Signal Processing Magazine}, volume = {11}, number = {3}, pages = {18--60}, issn = {1053-5888}, year = {1994}, publisher = {IEEE} } @book{scholkopf2001learning, title = {Learning with Kernels: Support Vector Machines, Regularization, Optimization, and Beyond}, publisher = {The MIT Press}, year = {2001}, author = {Sch\"{o}lkopf, Bernhard and Smola, Alexander J.}, month = dec, isbn = {0262194759}, } @book{shawetaylor2004kernel, title = {Kernel Methods for Pattern Analysis}, publisher = {Cambridge University Press}, year = {2004}, author = {Shawe-Taylor, John and Cristianini, Nello}, month = jun, isbn = {0521813972}, } @article{liu2008kernel, author = {Liu, Weifeng and Pokharel, Puskal P. and Pr\'incipe, Jos\'e C.}, title = {The Kernel Least-Mean-Square Algorithm}, journal = ieee-tsp, year = {2008}, volume = {56}, pages = {543-554}, number = {2}, date = {2008-04-09}, ee = {http://dx.doi.org/10.1109/TSP.2007.907881} } @article{huang2005generalized, author = {Huang, Guang-Bin and Saratchandran, Paramasivan and Sundararajan, Narashiman}, title = {A generalized growing and pruning {RBF} ({GGAP}-{RBF}) neural network for function approximation}, journal = ieee-tnn, year = {2005}, volume = {16}, pages = {57-67}, number = {1}, month = jan, doi = {10.1109/TNN.2004.836241}, issn = {1045-9227} } @article{vanvaerenbergh2007nonlinear, author = {Van Vaerenbergh, Steven and V\'ia, J. and Santamar\'ia, Ignacio}, title = {Nonlinear System Identification using a New Sliding-Window Kernel {RLS} Algorithm}, journal = {Journal of Communications}, year = {2007}, volume = {2}, pages = {1-8}, number = {3}, month = may, issn = {1796-2021}, timestamp = {2007.04.30} } @article{liu2009extended, author = {Liu, Weifeng and Park, Il and Wang, Yiwen and Pr\'incipe, Jos\'e C.}, journal = ieee-tsp, title = {Extended Kernel Recursive Least Squares Algorithm}, year = {2009}, month = oct, volume = {57}, number = {10}, pages = {3801 -3814}, doi = {10.1109/TSP.2009.2022007}, ISSN = {1053-587X} } @inproceedings{lazaro2011bayesian, author = {L\'azaro-Gredilla, Miguel and Van Vaerenbergh, Steven and Santamar\'ia, Ignacio}, booktitle = {2011 } # ieee-mlsp, title = {A {Bayesian} approach to tracking with kernel recursive least-squares}, year = {2011}, month = sep, volume = {}, number = {}, pages = {1-6}, doi = {10.1109/MLSP.2011.6064585}, ISSN = {1551-2541}, } @article{vanvaerenbergh2012kernel, title = {Kernel Recursive Least-Squares Tracker for Time-Varying Regression}, author = {Van Vaerenbergh, Steven and L{\'a}zaro-Gredilla, Miguel and Santamar{\'i}a, Ignacio}, journal = ieee-tnnls, year = {2012}, month = aug, volume = {23}, number = {8}, pages = {1313--1326}, keywords = {}, doi = {10.1109/TNNLS.2012.2200500}, ISSN = {2162-237X}, } @book{vapnik1995nature, title = {The Nature of Statistical Learning Theory}, publisher = {Springer-Verlag New York, Inc.}, year = {1995}, author = {Vladimir N. Vapnik}, address = {New York, NY, USA} } @article{chen2012quantized, author = {Chen, Badong and Zhao, Songlin and Zhu, Pingping and Pr\'incipe, Jos\'e C.}, journal = ieee-tnnls, title = {Quantized Kernel Least Mean Square Algorithm}, year = {2012}, month = jan, volume = {23}, number = {1}, pages = {22--32}, } @article{richard2009online, author = {Richard, C{\'e}dric and Bermudez, Jos{\'e} Carlos M. and Honeine, Paul}, journal = ieee-tsp, title = {Online Prediction of Time Series Data With Kernels}, year = {2009}, month = mar, volume = {57}, number = {3}, pages = {1058--1067}, } @inproceedings{kuh2001adaptive, title = {Adaptive kernel methods for {CDMA} systems}, author = {Kuh, Anthony}, booktitle = {International Joint Conference on Neural Networks (IJCNN 2001)}, volume = {4}, pages = {2404--2409}, year = {2001}, organization = {IEEE} } @article{theodoridis2011adaptive, author = {Theodoridis, Sergios and Slavakis, Konstantinos and Yamada, Isao}, journal = ieee-spm, title = {Adaptive Learning in a World of Projections}, year = {2011}, month = jan, volume = {28}, number = {1}, pages = {97 -123}, doi = {10.1109/MSP.2010.938752}, ISSN = {1053-5888}, } @inproceedings{vanvaerenbergh2012estimation, author = {Van Vaerenbergh, Steven and Santamar{\'\i}a, Ignacio and L{\'a}zaro-Gredilla, Miguel}, title = {Estimation of the Forgetting Factor in Kernel Recursive Least Squares}, booktitle = {2012 } # ieee-mlsp, year = {2012}, month = sep, doi = {10.1109/MLSP.2012.6349749}, } @inproceedings{richard2012closed, author = {Richard, C\'edric. and Bermudez, Jose-Carlos M.}, booktitle = {2012 Asilomar Conference on Signals, Systems \& Computers ({ASILOMAR})}, title = {Closed-form conditions for convergence of the {Gaussian} kernel-least-mean-square algorithm}, year = {2012}, pages = {1797-1801}, doi = {10.1109/ACSSC.2012.6489344}, ISSN = {1058-6393}, } @article{parreira2012stochastic, author = {Parreira, Wemerson D. and Bermudez, Jos{\'E} Carlos M. and Richard, C\'edric and Tourneret, Jean-Yves}, journal = ieee-tsp, title = {Stochastic Behavior Analysis of the {Gaussian} Kernel Least-Mean-Square Algorithm}, year = {2012}, volume = {60}, number = {5}, pages = {2208-2222}, doi = {10.1109/TSP.2012.2186132}, ISSN = {1053-587X}, } @article{yukawa2012multikernel, author = {Yukawa, Masahiro}, journal = ieee-tsp, title = {Multikernel Adaptive Filtering}, year = {2012}, volume = {60}, number = {9}, pages = {4672-4682}, doi = {10.1109/TSP.2012.2200889}, ISSN = {1053-587X}, } @inproceedings{vanvaerenbergh2013comparative, author = {Van Vaerenbergh, Steven and Santamar{\'\i}a, Ignacio}, booktitle = {2013 IEEE Digital Signal Processing (DSP) Workshop and IEEE Signal Processing Education (SPE)}, title = {A Comparative Study of Kernel Adaptive Filtering Algorithms}, year = {2013}, address = {Napa, CA, USA}, month = jul, doi={10.1109/DSP-SPE.2013.6642587}, } @article{slavakis2008online, author = {Slavakis, Konstantinos and Theodoridis, Sergios and Yamada, Isao}, journal= ieee-tsp, title = {Online Kernel-Based Classification Using Adaptive Projection Algorithms}, year = {2008}, volume = {56}, number = {7}, pages = {2781-2796}, doi = {10.1109/TSP.2008.917376}, ISSN = {1053-587X}, } @inproceedings{pang2013low, author = {Pang, Yeyong and Wang, Shaojun and Peng, Yu and Fraser, Nicholas J. and Leong, Philip H.W.}, booktitle = {2013 International Conference on Field-Programmable Technology (FPT)}, title = {A low latency kernel recursive least squares processor using FPGA technology}, year = {2013}, month = {Dec}, pages = {144-151}, doi = {10.1109/FPT.2013.6718345}, } @article{zhao2013fixed, title = {Fixed budget quantized kernel least-mean-square algorithm}, journal = {Signal Processing}, volume = {93}, number = {9}, pages = {2759--2770}, year = {2013}, ISSN = {0165-1684}, doi = {10.1016/j.sigpro.2013.02.012}, author = {Zhao, S. and Chen, B. and Zhu, P. and Pr\'incipe, Jos\'e C.}, } @article{liu2009information, author = {Liu, W. and Park, I. and Principe, J.C.}, journal = ieee-tnn, title = {An Information Theoretic Approach of Designing Sparse Kernel Adaptive Filters}, year = {2009}, volume = {20}, number = {12}, pages = {1950-1961}, doi = {10.1109/TNN.2009.2033676}, ISSN = {1045-9227}, month = {Dec}, } @inbook{vanvaerenbergh2014online, title = {Online Regression with Kernels}, booktitle = {Regularization, Optimization, Kernels, and Support Vector Machines}, number = {Machine Learning \& Pattern Recognition Series}, year = {2014}, pages = {477-501}, publisher = {Chapman and Hall/CRC}, organization = {Chapman and Hall/CRC}, chapter = {21}, address = {New York}, author = {Van Vaerenbergh, Steven and Santamar\'ia, Ignacio}, ISBN = {9781482241396}, } @article{tuia2014explicit, title = {Explicit recursive and adaptive filtering in reproducing kernel Hilbert spaces}, author = {Tuia, Devis and Mu{\~n}oz-Mar{\'\i}, Jordi and Rojo-{\'A}lvarez, Jos{\'e} Luis and Mart{\'\i}nez-Ram{\'o}n, Manel and Camps-Valls, Gustavo}, journal = ieee-tnnls, volume = {25}, number = {7}, pages = {1413--1419}, year = {2014}, doi = {10.1109/TNNLS.2013.2293871}, ISSN = {2162-237X}, month = jul, publisher={IEEE} } @article{li2016kernel, title = {The Kernel Adaptive Autoregressive-Moving-Average Algorithm}, author = {Li, Kan and Pr{\'\i}ncipe, Jos{\'e} C}, journal = ieee-tnnls, volume = {27}, number = {2}, pages = {334--346}, year = {2016}, doi = {10.1109/TNNLS.2015.2418323}, ISSN = {2162-237X}, month = feb, publisher = {IEEE} } @inproceedings{takeuchi2016better, title = {A better metric in kernel adaptive filtering}, author = {Takeuchi, Airi and Yukawa, Masahiro and M{\"u}ller, Klaus-Robert}, booktitle = {Proceedings of the 24th European Signal Processing Conference (EUSIPCO-2016)}, address = {Budapest, Hungary}, month = sep, year = {2016} } @article{yukawa2016does, title = {Why Does a Hilbertian Metric Work Efficiently in Online Learning With Kernels?}, author = {Yukawa, Masahiro and M{\"u}ller, Klaus-Robert}, journal = {IEEE Signal Processing Letters}, volume = {23}, number = {10}, pages = {1424--1428}, year = {2016}, doi = {10.1109/TNNLS.2015.2418323}, ISSN = {2162-237X}, month = feb, publisher = {IEEE} } @inproceedings{rzepka2012fixed, title = {Fixed-budget kernel least mean squares}, author = {Rzepka, Dominik}, booktitle = {Proceedings of 2012 IEEE 17th International Conference on Emerging Technologies \& Factory Automation (ETFA 2012)}, pages = {1--4}, year = {2012}, organization = {IEEE} } @inproceedings{vanvaerenbergh2016relationship, author = {Van Vaerenbergh, Steven and Fernandez-Bes, Jesus and Elvira, V{\'\i}ctor}, booktitle = {2016 } # ieee-mlsp, title = {On the Relationship between Online {Gaussian} Process Regression and Kernel Least Mean Squares Algorithms}, year = {2016}, publisher = {IEEE}, address = {Salerno, Italy}, month = sep } kafbox-2.0/doc/kaf_diagram.pdf000066400000000000000000001611011303712747200163200ustar00rootroot00000000000000%PDF-1.5 % 4 0 obj << /S /GoTo /D [5 0 R /Fit] >> endobj 8 0 obj << /Length 1634 /Filter /FlateDecode >> stream xYn7W07 A,DN1'!:$D6f{8ܾ!4 XX9J%~^ rNRE'.;<H07M Ɗ'ă~|/B#)+=Eb&Zdx#~= %LDA}Geͨ;F~-%aJR~TIjg A5\~+v/QnwKZ;14BQ&:JOTLdĠAY٤:jX`&h_>ݾz}>c>~ڀK E|\Pېc[]2[3EN(A>C u(N;㲎 I0kb/^%STWLЂ|1<-y/u~4oY9=u3d5v0c&1bCڤRUgY j'N ݗ ͏P_䭭,$ uZd%줜N(˩-G.A |߮zZjt㣃T8nLYdp&d*OAMu.-o'Pǿ)2bUQb/%녬6`5:0J뵅z] #FZXk&$.3qAM}>v$`]Z/J~M7Hc@G@'|l8B|?Є9VUܻ4(.e f:}jL2f(i"P] Ɉ7짵aVV):*;}Hl2r_:ؓrkn"UEƐl=җL\rEiI3$W8EX5T@S>N z1KKi+fR 풴''S|i fs@vIkɊy'?V2@j+>9V* F}W@Z(:(:XˀCzIYp_aـj5.[dƀdR >~2ۮ-g5ÍV#n8Zjȶ"Ę 'n1Pԁ\zW6Й,{>0,MDBOYLvĽbk 'T6JzoL&؂(mCtjQK*&eg]{]&xb!,IFJ7ୌDQk'5@FNtvǚ>w쑚7)4볰LJ;XM]޺oإ&/]uy{njQWִ Q>(!!VǣfKIէ>{i(a "{B{,$}x_3#neLr0?Ll[gti>_8 endstream endobj 5 0 obj << /Type /Page /Contents 8 0 R /Resources 7 0 R /MediaBox [0 0 297.075 142.185] /Parent 15 0 R >> endobj 7 0 obj << /ColorSpace 3 0 R /Pattern 2 0 R /ExtGState 1 0 R /Font << /F34 9 0 R /F10 10 0 R /F8 11 0 R /F11 12 0 R /F14 13 0 R /F52 14 0 R >> /ProcSet [ /PDF /Text ] >> endobj 1 0 obj <<>> endobj 2 0 obj <<>> endobj 3 0 obj << /pgfprgb [/Pattern /DeviceRGB] >> endobj 22 0 obj [760.6] endobj 23 0 obj [777.8 277.8 777.8 500 777.8 500 777.8 777.8 777.8 777.8 777.8 777.8 777.8 1000 500 500 777.8 777.8 777.8 777.8 777.8 777.8 777.8 777.8 777.8 777.8 777.8 777.8 1000 1000 777.8 777.8 1000 1000 500 500 1000 1000 1000 777.8 1000 1000 611.1 611.1 1000 1000 1000 777.8 275 1000 666.7 666.7 888.9 888.9 0 0 555.6 555.6 666.7 500 722.2 722.2 777.8 777.8 611.1 798.5 656.8 526.5 771.4 527.8 718.7 594.9 844.5 544.5 677.8 762 689.7 1200.9 820.5 796.1 695.6 816.7 847.5 605.6 544.6 625.8 612.8 987.8 713.3] endobj 24 0 obj [576.2 583.3 602.6 494 437.5 570 517 571.4 437.2 540.3 595.8 625.7 651.4 622.5 466.3 591.4 828.1 517 362.8 654.2 1000 1000 1000 1000 277.8 277.8 500 500 500 500 500 500 500 500 500 500 500 500 277.8 277.8 777.8 500 777.8 500 530.9 750 758.5 714.7 827.9 738.2 643.1 786.3 831.3 439.6 554.5 849.3 680.6 970.1 803.5 762.8 642 790.6 759.3 613.2 584.4 682.8 583.3 944.4 828.5 580.6 682.6 388.9 388.9 388.9 1000 1000 416.7 528.6 429.2 432.8 520.5 465.6 489.6 477 576.2 344.5 411.8 520.6 298.4 878 600.2 484.7 503.1 446.4 451.2 468.8 361.1 572.5 484.7 715.9 571.5 490.3] endobj 25 0 obj [388.9 388.9 500 777.8 277.8 333.3 277.8 500 500 500 500 500 500 500 500 500 500 500 277.8 277.8 277.8 777.8 472.2 472.2 777.8 750 708.3 722.2 763.9 680.6 652.8 784.7 750 361.1 513.9 777.8 625 916.7 750 777.8 680.6 777.8 736.1 555.6 722.2 750 750 1027.8 750 750 611.1 277.8 500 277.8 500 277.8 277.8 500 555.6 444.4 555.6 444.4 305.6 500 555.6 277.8 305.6 527.8 277.8 833.3 555.6 500 555.6 527.8 391.7 394.4 388.9 555.6 527.8 722.2 527.8 527.8] endobj 26 0 obj [706.2] endobj 27 0 obj [606.9] endobj 28 0 obj << /Length1 1392 /Length2 6122 /Length3 0 /Length 7064 /Filter /FlateDecode >> stream xڍt46ѣA<э{щ> cѢG5.5z-H.] kI޼Z߷fg{k}_{ac3Up@ATp4HPQ4 $HfE  L HFcl`40@@AHo ) (CPlJw_$+a HH H=7LE{0 0DC!hvF%n(>I 8(`7j|l3#!C(L'0C m;X/9  GPA|h4;a(&v遲U}a uGPP/`ڬwPBAh)C{L} }rpt7C=8_>]1fV]YRnp=AH$ؗ3zI{~>8 0GDA/?{"<@| S{.Տ[N*yWeVM8yo=npVd>G)v.pO3]m57%7_$6~3TPHk$ozM-S^>wOMwLU Q-QhːOlyvY,7м\{>(r~2i&pl3_=8Wf$jKgNx퐢Oq&ͤ_ѫylqw%kб"dkQm@Lt{L5~alEjI嬽E6Y РKVFD3lwڷw۸h%b9\j 8_ QMq:{a^A/*Ϣg߫c^k.n_$|ijדv]lw:&iXF ^ %Ջ}̣!Hx[iʹy8tXla[eVmo("Ht ̐lIĆdm$s tkັٵ T:؁[hm@v&ae` ' =E烹]G8vPU,X/!; )ZAz{ H cR #|[&cښ6#.cK^ZSLo!{& a7O%-x6_o*7H4"`66\nմWpR\9(zc\+nm7ߝ.WYn>9*X{ENiNA;G_=eLa%o$6?nd I)|#R󈫵q{zS2z0E96Y{х;9amG죭,P3yO )P^.mjۋӈO]L_*!IR| ^+b[u$. _*pzYzwЬΩm2,*J-FN9WOF?<&) $A$jwƈ{a=Cw*78\Ҏ)d3k5lZ[DבSvdsrmbZus# Gckrz'*p^_3PJewD%JyQ(7`UBP|ta?ST`c56@OL'CWBl{Ȧ.l8I %T%P.:?-qK S[%`}&yG"F/ԭ35E93*=EqKg*DZ}9Yu8wwn0>ոU(!ScR.ɨ9og~*1/S/ hg ႖>TTبGY[B(aHRc/u+geed 9BIü|GIJj z10 H,PL SMiH;'- /VɊhP>.B69;2/|4P.AtbltH5 9~QNl28ٵ:K'}ybfo0iټn2?[SHC a ® 0Y:Zu91d>}Xm'<8bF2S\65% ({BZ{n1qnʜsɸxRA<c/Wx][ ěB"nDWA5Y ͕Y_x>3dS*B[V2>BZaV|˰4Gc):᫨l:.gk"o8539#̒01ߦjoq~yS x8SV_,.Ok18hc)k(!_1è\gyNgQ1UJ%qH""0%)N=Vi!M97@c*񁸰HiۈrN=BBxM RsZ[^4l*kQm T<4um%lQBI&AęvSGȯ."]Jt Lj4IP/Wm/UR>IzvgzpϛeiSTB ŋwi{}}թ6ޒt3cf&?/ڌt˪FuFz%\I<+f˶Z>.(55\kѧ S ANU"O5vc3̷;'.4@iebAOm  K(ԇ2-}}%./T|3! eyR#[1u^Ox3K~ %7C=rǪ^DHT(}`.}]wB[i9eqF'éoTO6㢔)KH>M'zEɵe9}@"GƏ|϶Eb mgIؖ/PE-uMR(k?-Mo.JJ][&VTAt[F o~3#  \'w.um6+Ŧޠ2m^)D}N>s7)4$Sh)5`1 whCLkeNݮ.ـԗCwb6yņoe>ܽ|[%=T[xylE6M}Y<էղ2-d?GލqXbg1 Nўc*h?Z2}GQz?)m*5!zB/N@FB_H8ql'PyIfA6Eвъg%hJ ( S@b*fmQǨS?-;Sz1Q1T0aoA/dav;ʁxN…yBsjXl)SR[Ik$ةѴěT:ϩ^.0/KG<:z;;ʣZm3vCO `pg-} I'M^u{8X ѹJVMg4-CZefwiRm}e>35WRmq wW[4S|T̽r ?p{v'򤎶uV.6gUM+%x4ʒh< ,Juzn !.JFlOgصr 'ѻ"gmZJPJub=-MZ{HU.y̕|?ѭ WOP\f>kr)L߷w Ln_HҸ*-[BLwpxΎHY7wںǾOj:lcICL|rX-YG%#f$ެwHm {rz,.gG7񅜗 3Q/+u'ObH\׷w p̐|@UZl>:L;jqɱ*oF>/kiX}DHV{b͸kzb')p7^,e42ZdPѲvamgqeŧ]/=vI9,ݼW0QQAK'A MXm,`!.\"l]uMĊW'jrHΟ В"Pw({ nvDLU(2 |rj[/vd`ybwAڒk;x,h"bhݢ՗܋ $ƣm]3Tֳ]O-V+z.#J;T4M-$^JDl#o"qc R>֣;ie .b6Y3jFo%OoR4QZXh*H˿Y_ΐ=p"KbԧVEEiNa n$h3{FbN8xj!^Tb2>K[x# =I?/TQH븇%CK?W<)(v?BQSZk r9'vMmP* ]^Z/ >̼13a@EŶ͒/i/Iri m 篳l@YV,Ye4bsr1`^zw8o(pe\:A%bNªCJcNf̡/&7HS y3-rNu5jy6H+yLfS|UT|/JjڃimR򠍨I)R3jrr`w&FW368p \xn[m.8;췢iopR@wB:<`kTw}v_=xEdͧ}>=JWI:q1!K{)s̓8b|˜Tԡ#4`ܰ6Uv{S/讶S%%Lpuw?;nVPVY]avKYvaƎDGO+5lR/!,_^c{-O{X2l z5VLXM1 IǑϷw9>Q c:>]-9p>z*)@XԜH?]^3QRT )z3 ).[6_i5%T;[لΊ !Zpv!N~2Gj!C|х*xcD,u'TI:JY4x@BaUK<.-%cD\丢?Aqp('Fvf:;u2|[qAΊ\Ƴ?q;H&򄆧 a( "uk[6hK{)!"`":|mN=IwVlɖ&ΟIQ2#WbI A JUE]6㮛\\ɚr%r> endobj 30 0 obj << /Length1 1497 /Length2 6918 /Length3 0 /Length 7921 /Filter /FlateDecode >> stream xڍT6t,ݝ , ,JttJ7H))"HH;{53u5YY)pw4( P@?(HnCB@(]?JH()D%Ab@ @GJ0{?@E+>HG' Pp"a;@va:BC8EW Ni'4!) vCÑ\/ `EAP{o]ojD#'ꯀ!FB+ uGaRCLw6@u @;w!d0wC}`+͏F`W\v 0P$ F`9 .fw{%"}>e ܻuq{c9~Ӱ0vy>Tpw~s}0[ [18Kނ?)]=W?qο8 7#'hr1+PS_=qߨYwG@@0*j8% s>`_<, b݃`^.( QEDoY>;n{'ED`$C0ky;I`8H߃` ¨R @_  F  m b}uP$0,IPo("\~Zŷ1"qj&7RdMSYI2͙zjܥq8}^#|Ǭl0us|k2y 4>GL݄r}zf8T4u;kri%L[6D毺=]Nz%yFUM<048O66X3d[REjQYvKR,*?;;}NtvX8SŜlS@|Ϙ(ԊȂHfIO&Lb*fddt 0N^$omk(UxhrvB_sO Zg8ɖKQ z2.6?~y٩1QO\?U|$Yr_b=/mwؾ)H^eQ1I޵sOjVs))ѻ⪩|I]EK^}`~; PyGD&Z&\Z\mج(2%zC 6h]ޛ?0}|{Ay~vF *w9r%ֆJOd̮7 &V$"gC[۶\|o! +DoΚ m{ t[v8K᪉540KA5:YQw<7m _(.Wr/mG+^ h9PܶR.!?'Ǭ$0!)ƺ`žV&_$8c* ۛ6DDG/c2E܌wZ8) Šϖi<. |,$I0(m6:hr uF탂B^FZA`c}7>sNoյ[͘G{i[()X&( j#Z. ܳ1yw{nXT;C4E1)uO+ilGzU<ۓN@к*[l;jF_e};TٶsQKo .pYRwx:BΒRkIÔv[N{$D//JFX13loS;.~^+/{r\8IwHE~ ap-ߒWv߄,OidU&]X:N_ $gt"]ύǯ}{7px3*s}5xA#^gc|cƭQU -liQo&B:uuzn9TKZ&窽t:Î KF.wZ<&g<̃ . a i  zl* .U8,!RbD'42>5qFw"RZF2A2u #Ɓ!Z]MCϰϤd'Q$?YM{rwķb&XUccZmj'K<%O`kjK1䜡q3غsRiA0E#Z~t^8 ~[hȋ'z.J@|80ٷ}ug8ja7k7qn#j9VbS=Gٴq5ۈM:2:EW('8傢AáF&W">ZAk?gзMd|vMioZu6^,F#CItLL_ec·خ31+HP jt]}c߆K3dÕ8pJBf&-6|"VإO@;{C2gZȰ{ATl>XyK{|9?rE9NU_d[eiϑUx;lŊbJ(L-Ns1䍏PDIGYmɜ4K?(]ߴ]Ѻ_ hSpsqD7/mR2;GՏhb|d+q7/wJ=X K*D,mWR+! J_w_ u{O[adrvYRqkvsQ˙&͕D顝Gqlʠmt8lVUj2ra\j˼-9<%;D+7C~0hI&NjΝ+w~鐚HS(R٧J<{[%\ zi˫TqqevgW02a9.n뙴҆E}q4r|1y2tdP:4%9!|upgZׅO>2Ge־"*ql! \ c;~`;"QQx]b[T&9 ,W1ئKIJd-嵓FVho]q=1S cd%^u!)PO{t)Sev'1i(`̦J|u[ 2X2I]5hx~Rdf`EXi6F_-. !/UE&.Fu^좓ݭc}f6ŃAkc;Y m|M/>AiyS %i5 .XQ'PZzT|r%(CHHMr[Kg {g[;e4bnOɉswlà]WIymԘYpue Vq)/nQL;52J?^&6/ȝmNޖJnGbyTO@_n*9>k8{ I~(#ρ]:lf_NH_, dA}e'j@BEyzSq7&IpKJ)W"[^>VKS9(Ak_q&fzٴtQp /8y6XXZI`%vZf0fv@Q)hЕ eJ3}+GN~[gQߥ}hx,p#rŐQs]d$dBubkƶ-@O+  JK>ᰔAb{^RhD5MؗogLBY_[KEHvz Ѩ@{*zd4~mGjxbeཿM(2?TJЦN8 i |JYLѲ\,K?䣵p*%m~)2u$##dU sBN@aźk0 HIWX˲i`SNIꃝօ$yÂsMb#uK\|ޔqOc9eiiVEWE{o~$U.n}|՜eĶs%N#,_Ś/c+z/MzgDe>YjN~-|(R2ٸjsOS"ul.?ʯ4dĂuޥRA/2Wg#jCQ;z<>mP ebgIgJl6dn)xҫ$6"^XXxĽ}my Q=Hg5qpxo-(|fZbÚd I԰]&쁁~TfР<Ǧ|wՉ|p4s=bo_<4bM6Gt O^Pkѭ9"f4g. q+2G|9Qj5{ɣQ>.O7% 5S[Z.QvK:->JE)vgk-%< ^w N:9*LHi|S!{VFQ3<.}?<}KWѦ ;j0¥ ιa>!6Y\Cb\RL׎L 7pk ]ę61u8i^4-`|cM#㏎c'5 Q&m^XUCz(EB_WP;5k/(Fmg 7m9Mתoﺜ2Z4t Eh'sHg&_5jEGbk&lg+)#v",<9(cK6Ftw +P/uGV]28lt;Ո[zguBml3(-9'/6((Ծ/䭭“.۟MۜOERKCBuQL"jܶr TuH{?;=At;.a\cu+ZkRYhx6(v7?f1]&I8WԯѺ/z{a-2z'K^-4ugE1BgA^6` dV,%V8;mj0Zws9wT璑S}A+1oomG܌"Lyvҿ[iGON4Bf&o]+Wѓ㦽r 7OS.ΜuX" >)3G|UuGnqWT9~9(ٜொ:eZTI;R;b-(1=|;C߉idg0Z#~w+9>"=oYĭeRVmaёiQ y:blPb^Ki?Ғ\Q<&LWߍeBv *%NnX99l=lVkn8@ϒh9- vKτ-%bJJń / h?N\uqȽ&9xrzi&CBa㌨ngjV2zD1e MVVaɍmˤolO\*Ze}NۛjM9ǗH(p^86qbux,~1AXtʼn ,i<Ğ+a*:۶N$p%1}L6#S0hgݤH X޸/ ux,|=jfKNnN<"mf[?[%9pߞ3V>3ķ_[\xJ63=>?2p != ԳP Tͽ0(9Tb˚OA1 ;(Cw;/~;hp؛#ZCM< Z3VͦNxǢԵgdqQC¶Lva>>{H dqI&нYRdhJ"v:2'{: 41"fk>6mC~EN,OY?5.R:_.2"&]a<12 Ml.!ERlr/!莞rdh֒ՙ@E] fJM,"|gG QYwIM]3¥eΡ\Ȅ,'@i}@gw'$DW?I6b,J1ZRc3'sGW2q6qINn(vV݃Uw%m)dL$'h{/SD/leR{lR/ <&O3OIsx@u"_|~wJ \vc Zx:}`UۜǢ_ G4 -5BhS{^~6?SӞ/C4a]'`Gcf7Bfؕa3In 5RxWVKァ wNҺC/H@\ $5ɚ|#MW>SMoQE1B+|>mQ؊ M!ɓ8;;4t* gȪ3Y>hwXY/ڐ(5i܏O!֧NC֠ E Ri+ endstream endobj 31 0 obj << /Type /FontDescriptor /FontName /NDYMBJ+CMMI10 /Flags 4 /FontBBox [-32 -250 1048 750] /Ascent 694 /CapHeight 683 /Descent -194 /ItalicAngle -14 /StemV 72 /XHeight 431 /CharSet (/comma/e/f/kappa/r/y) /FontFile 30 0 R >> endobj 32 0 obj << /Length1 1407 /Length2 6067 /Length3 0 /Length 7026 /Filter /FlateDecode >> stream xڍtTk.)H 1t ݩ CCH ݍ4" " ){9kfw޻빯2 a*H(ji@ A>H4` 07w8!`ڧA,HE%@ $w"MH̝tq٣c~p@9`qqQyg A ({3z"0@B0ϿZpH٣P.^^^|gw> ' an0/m32> oEyA` !=` q!$kIu70?] B.a;:*|(oqrG!?] х.3.닃+VUqGoO⎖ r0}eiK"ćM% ekw?BИH7_kwΏ#W_zo:m< M#!!mt^kcRc>I/ Y*s` NmWgV}yU1ZOcN%3O4c$+ܗ(reu[ƮH^3F^"ڦU]II/nj0\>f^q{ydv:TAWE^Ak{a(Rez%!*\"ϭ{~T9Mr&ǯelW 9Ǭ{Iv?u*̔R Bi7nG]aТ5vЭJ=WT@UNezisX( @,SPH C?f6.ێ)GO{?UODm=6ࢎTtP<y:8 ፐҚmeC/.xi6q0 e>^ 'N+^|b]~zj㷒.l')z/4DW[v}eHq{J `h5^|ĉb]]T%rJ$~U+8<ܿ[SQF]| dRKݑ ɝ*_ʅ3.0GC? ̞W-Fsא-8 zl@Rg K~J̼\ۄ?m۵s8ױ/9M|/W Ƴ5Rвa]{ҟ^[& `] ?-Fc %zyc)^ЊEukn;{s3+(\\fGW]b,7cBĩ1?Ԓ1D0釹h/Jͮ;{zPßA'>nlk|4xYt]o6:IAfj z<"Jܶ6䲇2|h2ט6ӂqj/[Նui: N 3bVȥ 84͞켤`ea Bˈ^MXVǍk&tw)o~+ܾtUbNqU [ J SQw`+RRRX|k\mɻb7_p{~Vdc3/[XE1#"m۴?!TPu]l=!mX/(!^:|)4>PU8N^9݈)0aE:)A3"iGB 2X}œ#'_**G=c9NUIڷfh5^qS&96Κt#.1A2?V|YVr/3`go]S/K#j*HKZ0Xc Js\VOJ90Ug%cdsn'Wɯu- -j\jq!^R=[Y1TSCςaCWEV1b+S&JP|Kgwl^K JAI 0',#9wy%>,H&-v z3ܻ[ %z$^-@*/ؓ(ZI=`]#SBϤjӌȰ?n?XĞ袷r Z{^oC%~ `^׽:u}]t= ndDJג oъ FԽuQ,'mJkOmi` Ȇ-KcWۨwz *$Mpz$H$ [bQZ/ ?EPor?X-7]:rZXK7c%P|#+xs){C$Fb.vٓH ]nZ&a~폞wfZ|[וɅ),ȋ/GZNߘ R/?JIz (`8j_|}!#Q>ΛouбYER :U]FgB+pR=O]Tf’hrfձķOѫ[۳,oӢ]jhCkg"L6̚_M{w´u8<c@p#wܼw/~nnk.bk)$C43>7/\ W#[f/,&e $R;F6$)kEbx%Q>w] Cl0NƉx ;`ʿvE%έ % Cۘ#%/]f\bdʲmDuR:'}yF7V\"ߨRl*ζs>Y]:[3꤯'?(}pUAݺQ2vG1}+QO [dm!S!ɷ6ZqG UEʝdO'cv;ubDKy303yյ~{eN1]-֬k1Hןe`[ ^w:]svq'Yb~NzF@'Ew oV}H;*}e#x=8,T4󝾗c%Yw$ȊH+xѠ):IϭA^60U #[3ʫW/͛t M[} W͊Ir|¥wwǫg 9%HYQnxpQ'fJ«fueH#CJAkJFmS`-Nw{@^~'Rns5*0!}ג} ~AJxg Aņ'2jMmH&ۭ4@: בW ̚]n>s)Sa%c煛:X跹en&d4vi5b ӶsEuUiz1@ 9>W &8!8ppͫr@,Pă}03 jyNWeוӟ-su}Fזp\oMiqOj͝4c*Q,WrKyk'+:H Ħ {cD -kJ\ػ9|0bdo^_R;&d1SA~NEl;\]P" P>!V_mΠTDkh觀7zlF [>J']Nh}U4S}/6kUjg)@RSkP|(;; ~`QÉb@zOlm!KI=gw)Jihui!2!;s='xn^77Ҳߢ9ߨlB [N+歵ţs⪽Fgh6=;4k]žpLg-5I$ IQ:@QÔ'5ݻFvfm!k;qoLFɥl?nuFvd L5b5wdīX#$>UH~ w$P&R~IOA4s.UQߡ1OK,})&'HuboK U&KE߄'+hNXɌZ\> 5I:7=Y{>!=]\~O1"wŁb}YeŮvl̩6soljMT!F=S8I,R ά>síh]D:4+MkHr@ܱ ӊHџ:ydHRxGgR:1!9xꏻ ~ad.&%RMn9ԓ|@N 4I3xeEC[h6{]paf^Ls8Vwc9]?+znxJYVA/.ShO)QmV!yuZp0ϒzAIι~}aLSA[9=1qkU-Lc9f,,-QVM:E2G0 Qy!!Q=VGs`1"6N r 察8sH?rm t}xch kbhKv#e'=_Ry=jzvTr$vM1 - (\F÷]CPu41ja{W:牜'(ߦk0cN +^SR=hݘœTxMl%J9~pz4uyzbs7{Gs'@ʦb|ەٞQ_ljh?9es,,/[>:Oɰ';s؏.q SNʓU Aj&Ʉ>|e qz&Gf9Gax> endobj 34 0 obj << /Length1 1369 /Length2 5981 /Length3 0 /Length 6911 /Filter /FlateDecode >> stream xڍwuTTm6Hs萘AI f`JJJB%DP$Dii %D >5k9{k}_GR\uAb1xqX 266Є0XZ  Xh?5b C@6O@c1u_4 rJy%0bqJ6%X $ ġ\F0(/;DP0(00(P|J+^JPO ,UUD G 燀H&PO$@ψ%FB/` 6 u;DOBa~'Ca0¸HIb}|t5(_}`8GERW=`ZXOOu>m#\|`1AHE%yEh"@",XA^ZQ @xV^A/7EH @ BPHCx/"$@G :@iV _ KbЁ퍫s' !,HAyKȿ s5:U4 Ÿ,7C/`%4`>/ @=Q b%6?eEw%DJXuqXO?G0?O?ԇ)1᠁ h ,!'H,kbܠܠp޿K[D]rjR_|_Jc/~fAoZic35 ,' G7F dw2eh3,s׹ so j֜P[O8Bm4; kNJ n~ ] |OW}>( Fxհ(N?AM\F™:];F4qj ;jt,$wmX;N48dUſJtؾͯ0';Y­,4 9& 7<^0V,';(B)4DMhzRTyMZCPE'ye׬;%⦹F6>%cH^?[ftBmةpSChKQv&FbII1#"6OwiRXij3YaXSz]tZZ՗C';|ML¡+h=BZM5)vPml^ ]_28D>m$wUIR>VFY4&?Dȡ>f kMC2rś 9@~Ԡ͓AuF͝h-zok\r6WuyFج)[+}d]}C8Ѝ0@jf#,dw~(#P7[\$'ԯ5jge\c>ݞzIr&Z8NȉsiNֲ2/M^]$|N 3^7$8^Tpѹ25I|DY .S$i X t]o ,zNd曓$gJcʹJI|R ?=JUu\d%~Ђqw %gyjP13)r/el=l/ߊH%/%FzDtNsa~\nYF O>QE2=:֔) ͆+/B=aQQI۳i#Vns wXa8SV )$4?XBqolh[?߻]d|efNmk_f2(12giPBdR_:J<͉=+''Wg| O`[M]5lrف$t*ۤNFR0-owLUȿ6x5vvm&%.+` Ě{b ͛ՕKO}ckRezYz ԕ*.pG_0ug;l/HO!H)̐ۉo%VЅE:߼^᛺F|oQKwq}5T+|[+,4${sxhj{w%6ò\9wkAuӮ3ߥ44ܟTjް+! Ezu _%>u9 l?DBxSp`yVC5뮖)bH4RUw!u 4gRL"g? ע.T2pNQt}{NchC(U~52k5 ,+DŽ:Sg p/4zݦej3a+j*ڡBCc+ !C[i07{sU Z[\MTͤe9mٝTٯz:7$|F:ױz>TXE'azsclv{n%*n Jwvw+Y.>QZ]PAQ<]%Bw OeoW MF? *vryi,u,v;904J!:h$BiBm2|~w*k/)I /NytLZ>*hlޤhw)];5۸X՘((u'$ٔ~Y89Fu^n/ۻQE7#(3/I0;f{-o֬+'hU')ŞmzR?8ӲtĩV,YM,J; xÃ+ytbng&WnR$!RhBE>KbT6cKZߘ&^9Ibǀ _;_عSWR^e}wqk~Ϟd -%˹£b«?ȆNSLh#qE#*wLU$}*ұ#QqnUe|z$8;EQ注: }4DMb=ڒܬ@,ݔoO}K(@;`abȣ+8!ܴl!TM\¢} aO^ڥ!8J iՔC{kߵ#<ʔ]Lf޴x I ]PR7ZG")T4,x42AMI"ѿ"AUKg-ڟgƾF;>g~ bѻ{h5MsZҗupȬ-7o\LfΧ ۃ'?LLn@Hw̴A@jfac9bw1a7 =^)@Figf,DYOGV(É@HW 7µm/5EԠ+9l5ekCB7,@z Xp_2dw&!̀vl?2>V=@k zo'B j9?7 *q?42$8³F} )JmB+UQY&tb4(51&A%3GG!@ ;_̸7){i>(]]_r3EGôyBz)XVڽl3,u)j:Q{ob{ Z*Z\oUt8bW3] oll-KigU"Wr~haAh$+sSuHzݥˍ+xJ%9QRnC/0=)Ԛk'i^ap3bxz>WNֲ\>TĿFn22%5Ͱ֢22wK1UР*֊Խv$#7`,g+f1]3MA"H-mPI$aFϊ֣+wgA Z_/kD÷RK0eR)%Gn yrx칲f] *Cb$Q>D,TAktaN+.]uH!φn7ɹcM6Dow!n4d}&Nq3vD=侹2^Dob\QZr3M~E%ץZa-*xN˔qfq/NxySnk}*PcpErشbZK%BH'^Оru]qaق$;}ǥYx>WZ^k&{"&Jw=~0Ph nCvRӾ Zg:'0)KU3?~V@!y:;*Ɉ.ym }6xƕr'hEIB_  H!QB6ea6ǥ~oNDfױSE5*_ Kn&4âjqLl'U4FLr3MQDi57ծg7U;_β?iMsٱy>&->!C˂WB׹Be[[ %4>+9ƻیݞD%Up)!a/_"`aJZWz+cݾslRX5)%`bQu:QVZjNZz56v@U%UYR`oɇDŽ."6йk4>[< {gXJOHfR>u]z5q$ #sH 68N/쪋xK0Gf[$qp_HJ9e~R[;h۞y~K"BJ2-^GD%1 Ӱ9DB(_^<9Cg^|Zۗ DfFs̐_NYu KʘFV@y} Yevň >, NNAe\i]yFGsFpqmTO]!+/k/_9hU\Ŋ\Dhkx4Rĝze:~έ o$SU{Dq9"0$Jt6"5_vR}=QVÌYUCi'X1 7aybd*%{th{Շ]?*Xgfh eZ.Yl XAa`YַUVdEbIRתi^4 u`X΋W$yNk탖pKiNtCd Z5)vf[.sd&iH6MMio7I_PƑc H2*,(8r)FywWF^>d2ϱEo,&qztKKa^u޼ gq]Bh_t\%x7n4idӘ,6ǏshK2^Pq% ګ8=`,c# endstream endobj 35 0 obj << /Type /FontDescriptor /FontName /WRVPXU+CMMIB10 /Flags 4 /FontBBox [-15 -250 1216 750] /Ascent 694 /CapHeight 686 /Descent -194 /ItalicAngle -14 /StemV 113 /XHeight 444 /CharSet (/alpha) /FontFile 34 0 R >> endobj 36 0 obj << /Length1 1723 /Length2 12010 /Length3 0 /Length 13115 /Filter /FlateDecode >> stream xڍPk-whKM-H@pw $ n#3sfޫ_koZJu- k'K.R`ssj iZ] ,)W䕓8.  ;;]_Ni 5@J99{l a`bp1ip, 5@ x a;YӓэV 4n@W5*cEhہⵜl @+^-@Wkp2@KY/fg`pilaelm6  @MVaXPppsz9XX*@VB`ZY!nn ?Jd)ˀ`I\VWg߁<"ݝt w*8[ '^Vvlv)~`Zd|Au @ ppAV%FWh~m+ `:{?>~2y/k'/4_[&)efp82vWG@^_ux=  ##:27faz^?M/߄d3)-A+;u-T^zVYh rw_u=$>F, hX5C«{6d gBq{՟">wH puFym+rn5 &6N(t&' ,lV W+wGWH:|| (`s|u6?k @N9^w` ?,_Co5Ggw8{M5Jyl׬=9_x QV!k.@/ҼP}}Xg Oq=O,KkB6\%>Yݑa_x15Y^$DO/dȤ,~O.~`[)湸czy}XدUB}fyo\:Ko=GHa!Cz}9{u=;B 8p3ngJӭȐ {tW0U`xu`ѫE"9m !gf-H3[1ɱFO2\WބkJWfz$;촑\kO#]E7emMDZpnR~1R[O2vF1+di1.=e¾!$9 {GIOy~W7vnb-8xbX~K;?/"1ϊ`ϔ_O6̆VoVJ{S('m,)V[1(d=խiK)Ԡ=e;l >ܾ4;rҽ; ˏ rSK)ٙE>іtlJ*ydP/^RW:[2Y.%$}Yߡvf-6kڬcI3$m&TCq)w_3@}::ʅ 161:_I2gCpԈ\츢rub&ܗml{f7u+J.h i(ȓs5$1DZʅ)eN(: ;ŁmCbvqRisSacm>OJ`(bֲ/K)C)B*z}yPu'{'~ϑ$_Ʋǐ]IIEW!<~0:e(M^"vxn1k}|=/l(La"sKԟfyF;/ԅ9.E P˘B-JG\:RP koB+rZ }n5nJLւg_jb7*/TTr ثF<|y7So{0.T\D"7)JJ? +ZDIJty0`(Gr*8<W5$lu #).qڦXD(3;Tq'Eh{[߃^9"mlCP)ߊpdf6RM'V?ӳ#y&ZX9h|Yz~BXӋ߅Rzfg..!I'IN>w4'w7A}ߦ<ַ8M*z75fOb v# HHC޼7!]2Y0Be0օ']9ϙϵ9r>e?EvFTa0>q*{X9J膨T87ETU;ZҲ/}{o/?y#7ۡV]3_\WQKRm_B*Gp&Aw uQ,D *Ws) OaLR9PI g~&MkC oPe^iǍ$I\g>[6/F.or+LI~ç5#XǪVwT=5@E]=IR]DUsYz&6~3~qGl lEz6a2T-\ ѸnoZtm+м;w$eP@dw_a͙|n7f6\S>D:cxfsp +9Va&0 Z:EU"Wkp -淏E`cרl+C{#msEE6ũd7do->} w`p^[l>K1z <v-n.-$;)pc|FB.<('U!2:pE).PUQP=%K,ڒlx#d.~KH~H\v(;h5[J4<JUx&`!ݑ-s{P|$rQY/Z hF܅Z)LVHyѩ*ЎTz~ *<`RٸӋ`[tGfl04l%Uf(\͡ g|ߝҝtgRT9g-S[F'!"l=Yô ㇸ/]陼el*j´Rgnl׋e4xfj)>H8M?;Xy3B]?feP߫^Ɣ@ mn[?W<@'Uyx#\Iъco "B_>?Q ÔO_ m!> -}I2*c|̳gϮX4jw Hg?**"1LfT7*6l˴\?1W-4AZ(ɇn}>SXm/N&Q#npmzHwsL\KqHRRVHzg`MKp[7O4,S׵㳙tఛ1pZ2޳62ٌxyRlK|̍)1ێj!Wڢh%u%soO@[$Wx.LzQU22Ej%aSlJĭQTͭw|xDǯB+ro p$,]>nST!rX T8r)I>]642.`ƀl%˹7 RhE;z0b<{b4%M4%-b`-XϣRZ {V@Xf_ٿOnrSN4hbw^CA~sPTݎ84Ǻd@K|W1x5 Sq" ϢNM7[75 _MjH3'RZL]a-1TJTa~4 5Uy1~+1՛-3q:z!(GR >V@5+%R=*m>&\U0! BP_δbaiD(Uh>{[T &HMʧ cy6Qn>f6 KGqs ZMEg]͟xV'DZK-|h3+i4$V`%%O*ߞt CgˆUp_ZTKPz{aͱ,:O PybH9R2C%WU}@o68Kxlg4@k?6H)y&4<Oav+tY?~a[#>df s^N,']8Lds/jl A,V1GNz_U.\JW&+\~0d5 yOu{_f{ ~:w84E ^<4s{"u;K/OwӚ\=m1s5t'$q$(ש޷PcT >"|vDr?c)ohLsE>cJ9"M%BXsSW1e+ /l3YgՓ}Р#+6Nt蒃#!_2~ռԗ~ǣZ9w3Lۆ%LY͸sM:jpك3ӾM }Y?ʝtҵO%B6]W32. ,{UPnE4#2kekrDb5@ 'KE"3.vEEpl RQVR>ͮprZ6W,`h7CK .?;.>8:=mqLOUg}EY.]$0~pUxQ_3LyINr-KcjcޘOwxw[.->(?kGLGbŨu3,;V™EynL\`S?l,vݿM {5WTMܿm~4I.+\0tQ:g}9 @)NJ=f(L'UKQJGh qBT߹=PdSX52T/s\fed/[->eZ$C3t2%CE4o-mH"W{4ZX+5i/㥧bP7wֺrLnxUrEl._{ᆯcVWICѸn*zv*Q/ /[]i9V}LR)ט!!CT'OI)B1T3J30donLgaɈaQґ~hBgy_4&?gkp Y yQ7Sp]Rh V[b6C? ~W{9"s]!2܂.ʸ}`} RV *]=<ƸNU[B~v$E *~{pcݦn~!}s_pbl*K{$ϵP9D>}nZL~T2^O^MzouLx0Ȕ8p}險y&UaVF>Qw=11 cFWuan}AR1+k:նUI-/q9T}Y_lvM-V{"T2żO 7Ut5).#pMV"1g Ga@8\`>d)q Vb!ZX+S['29=*sƹ8)MPK~b7R/;-a$cH2z;'Fv>AA_gϩ02(=k/gğWRDl?8AЮQiTsT[j=UM,Nyv(aJ\CpZ*$D Yocbj4"".*QoPKHC(J[ˮl_6 SVw_a:ǰH*Z̧&*!?bN@:l /lםnl:TUDM]ϏRAU?K4fFfVEE/q:3oߗA4$QsOB\"[4rVK j5 HnIFng.n싍+1~pZpGM)XƠ }syCLO3R#thi|lR| 9 ̺AqO[!( C mlZtqۓ%%cP~nǶjc[ lڧ\{"dW$L 'f%,_M61<32'Wqd*)821(ypVQtїոĆ,U*䋋Sbk7r5p Hb^JA 1A2`wƒ^&O!7١H9_S%dsE→}C?6I8)(InB}#Hf&*i4CPY_ÍEdj}?ZR*%Bs] ;(0COB83/eŶrzWiI7~FPXwV>}.{xLLhkmBI7+u H DJ܁e/Jx}yBPО #21Un>s7A#~dBq2pctd%1qt7bx:7hoOE` m΀ƌ*$G1DDGbXJ%nջj&egp4b_B,Iظ"yZL{]b-CmXH6/e6PBbV i ECP.y&Vu P4/g[VF~j? XxSRz<t'S?~&#^v_ΙݔgG[mQ^Q%N1r 4GpP قP"hքqvGD- ("\v%${f&"B_M~ i#F|w֪ʕ̆L`Oe01WeHrR,^$Jx|B*5ے  8NxԫEفka-(t-81nCF|"(},6:ܷ(lX$'=9e1 m-0Ie&T<Yv7wMuV骵vyӲp~}b;#L6#?  p^QŹ MH])Xc g fT_G†b؜}0BhZ2˖+wn:Z/3Uc07 Bhu&!Lit >[ӺX^-Wj^,霛+UukLPj)&R5 UrkցbI<}ӽSeku&My06XyBWZ N٥\E*j{ғ1܍~ܨgP۰2?JPK:!rk_{"11J]1FuVB;SU$rNYS8 S )NW+U<:o# 4}W:yQ 26r$S&)k!MDNt IxA1q֍8VeCec쌤-X|P4Άz>Ϭ9_%#;lRȪSG" w%x7`Q6#}!;ӆYA0~YܡU o 7 'l>RQw BS&R=+g~;:"URY"o_pPߞfE1S9>^KutK;k=䤈^ME%|uw 98xx oLMXf|D/t适V텴zVlTs&9k^P4r:^`1^ J/c3GΑ8Q/0iu1`i4ǿXTF _xz/|zxpy%T4or59&Gu3iߑ-78l-YeTܷ9_!bmE}=fl!zTb׻`+rĘgռȹhH] t͂ "yNԑ(~ɰE"|Pp5-hN`di;F3s (IQk Wۭ9[D,òn?͖q6ZAg&3D1Ո]8]Z-h-bM^I*gG8-` Q4uWߝMOz\!O&Hp7i KYawXhW{<$[l}UmnW>a&s ]'X9W-1z@ /QyIKv\&˶7LZ;[Zx(u$jQs:[DrvXS c %EQl-cݖH5";"­FGG4FDu2늼Z*ؾ,jxd1j NCr莃zjeB"cn~h??p*ۦJ[Z{CܚH B[>/1\:L\iR#EAgP\ױg.3qgj 1Q!>*7q# vU6E:[l3/5] -e2ּs#zxmA8ؾP̆-! vEǾ·5@w]>&~(%'fdf 3/(p)K%PhÖbw/L[poWJիW4vC=U0 S[;s,gPz& 瓊/mGK".#@3 Y̩vdڱvpP'zj8j]{Odі ٩ 8Xm+{2)o.ӅMo*Nc m{Fſ1FbRv{#:OuQקFe9  C.P"kr!=TAnj#s?}nĬ$|Jvv8iE\IRe$9u`J):|>rL7|qJXl63ͺ؞țzl)#@ȦzZ5gYoٔF-Дݪ# 8 :7na4G|<Сߧ7#2?VS{x$Du7 %3sRjUSVbC4ת168 kVHj#}96$u0fM %{9QRImem'/Vj5So0W=a?T lTv-0ݼ=9&=Pi eOG\ӊka*D XXFI_d8 ћ |sR7Ofgf8k}Š)5-okOV?˔ ھR8Ax}\TEUG-V_lܒ4Plh1c6R'V>:~&lX؊i>2> P*{Hƍ.JLtWF.<]ԛL _`{o~E(s{WTj g&j:;%y8ƥԁo%GIy!{. 2 ]M~r@yXG7PQ9݅pZQmIuwuDAe]6'g7ܨZ̍z0{S\m)raj4~OIui~l IZtd$i-Zb;d?]\.45} H{(?,2#ͅhC {m>:z Fs)J9TsgfUi ps047j~o+rB=!%a -|4Qޤ_OE))b ۹gt\l(Wzݳgy#z"/CPuN>כjvp2\\GiB!QuHlI0KSz;uJޅwJ%B:As[n#hԄ}zv5޳[py)u; VRCĞ)CoE%듣#)@% j+\>| 7}zvMb{;ba^ Sڽq+ endstream endobj 37 0 obj << /Type /FontDescriptor /FontName /EWANWD+CMR10 /Flags 4 /FontBBox [-40 -250 1009 750] /Ascent 694 /CapHeight 683 /Descent -194 /ItalicAngle 0 /StemV 69 /XHeight 431 /CharSet (/A/N/a/circumflex/d/e/g/h/i/l/m/n/o/p/parenleft/parenright/plus/r/s/t/v/y) /FontFile 36 0 R >> endobj 38 0 obj << /Length1 1478 /Length2 6631 /Length3 0 /Length 7630 /Filter /FlateDecode >> stream xڍx4gCأB$*VJQޭYԮUETQTii>=9'^\纓v=C~=T@R%mCs BF04ꁂ!RC Fct`4Q<@H\ ;"=`/@[ٕ0'g4ߏ.7$));A6 uTC$E+}g4]JP[@z8rahg:~Aݠ a HG7 (0„x"Lu@Y/>@ @n`/ p]U-F8rQHL< 1* 0w4J(+ *%F:2 Wp]Hoߒ# 1" %DAC,;!pB @{xB!pA{ A5/30$#|``?FEE_H/)@ 1( yk?_5H_ 0fpYn+ 1|C+(y/eO sn0 =јFb6߮пZto _zJuЃ!qa`jau`aV⊹[P6A1* `_B 0(UoH4&pDz@P$!4Gb2"j @ D1E:`L0ce1!o>a0-6P(pa syqQioe+:Ѽ?J1Y ?KmꡢC5SG <c>VsX*GCN^;ANO?'3YZr^o,BmYH%gd, ' y]{ر,las+;FرsŎ8oKKor jֈUGPQ=`IJyЌP=S*OKd+vXƱ_>]8О4(褣|[&zL^1F7Abږhgڧl5XgkşҲss-MqCm6'h/d%d+GM_OPгSi谸zViv7qc'vD5h֙׷Ԭ!z܄GE ]9NK4kR`M0&-0"3йe.4MFBzX}g2!-2iANjG7o|ihq$(D`elW-kKV4 >]WahKQ*[,ѩS2uZ 3fLQo<=Vd +W6%)- PE[:N[\RxZ;C($PbDc,Bٱp.pϒ-%R謗qQ8DU⻗fbo|uM2(hED'̼-4S9ҺZj]ZTO'LQrvwx=jG_]e2RI=.9lM>Α*PY5KIxo$XPRE8 u;[pc,GVTmZWEliVMųh܎{9WiS_2=UyޖRL6~hW'28NZ׉n~:YTzMbaPdEdԻҩ9u2#dK+X+5eL2PJ/Wv&#!tCqg]Q齐EQ193o8lQ9V>':5(8z0nY}E]@~/S=4"e=GfXKNVSyfĽHhd@u5"dQ#hOmz_V]*;I05d'';6Nogf)\XC 9r*.~!o aHCЋ? ۼu%+CsR~\ηT]t~C"N= e=f#ɱ5<-aZ0qm͆;S_{pa~:Ӑ@:a;0PA8SLQ| iNPP7>y^Na-7590$?˘#q]#?W딇`}vfG}9`έQl|2y16 .ϲ>ݖ#>8GS2Ē~Fyؖ켭L4軟zp!vZ"}{`kYҕȗ!~~u#@&U\*ZV/+1jeUpMGSJe|j"#fڀn޴t~z-F+`' T9po0l9Q\"J7ߢ S.2 Ai2E4Ͽj˺!wQr`q[RFR$SɸKKA-> ^{P/Dpe8Mziq' <(>H$._<8GES/Etk Y8X&x#۰oļ|0Z(Sðo&kKv#KgtY{ Ud28Eɟy{BSo&nfI߂[ܷgg:R{3VH}>ޱQ][I~haBly(+7Q=I݇zWSd~ea~>FeI~)eCWZK(Px5OvUZGhնlVFo(3=]7^3ƬHd|mZzOۣ]@5}nEflG*mԉ.2r\5ŪN" ewV;?!jyR?9(qBϑݭY|CX@@Ae4ƻTG3S!uYI!2կYhn6h_X#tph=[z.k- eX~ NMbuDI}"W+L9:T~h(KW Hn &͟ F'xx 3e<6LIWy,5+GWw1OP !j_ؗmu?x'XҝŶ T lϫ73dIF_X7A>([P>ѹi fzޜKuAg;khԵWstCza]DɐH_%j7;xn5\du˙sf-WC{ql)!kW5M?9.{.\Yff4jɍƓgF)T'%pm('w#(,tӌ?ӝvn ;Nݰ 'C c}8cAr!7:q֚յ {\5v+JY:sS-Jު*+-ѓt5?gd{)CI1d.GnA^Mf-\AN;8ΣKS'|5搇{']-:_; bBq)YsVsȱE^]7z]x[[>GS'a&- T5mB//?8iW =vͱʎ).8+ /bCiM uVՙA)k-ޡ6Yt$|Tޥ+/˹N>m[fWiP;[8+PӜs`ق1lDpdDΙƋdIE^yie}ej"xփpw< M%iW'7NB R8hR3O5?iM''O=Yw/?xu]}Uz\/ ]0YUX(}AS: H:<]9)||`:WJLwI 6/7(J~Th6R^[E^\UI<b|wp_ Q~蕚Ӛ1^[ȧqjc-UlK=%.B. ]_/D|9ȶ~kJ;DF"<薁eDR˽M![}jyWwI­s!\RЄ:u Lf(H-R T '}@󑝾f7Ǟ9)$'A}׳[^@n^9!ђ(Yꖔ\:u>Xw 2@"fYRW\6ҵ{Ĭ3!S u)U}җs:ZHE0Y,Q=/hN4*I^# M~szCX2%*c2':Q ^˞ݝaR 2 1tk%A\7JÆ^gT:- mG7#}N#w(]صd<<~D"(̋TZL~f݄ZK!2R7{w N&: Y*Gg~ҭu3V*s7_F{j5@9^=NˀG 7ؙ /^U 5zf@E/xjQǜu_3>nG鿤?=5w?${a-=fҡ, ۺZ?V[Mh ))@M>IM/Mޤv뛶= ս:$£MrI4Z0??N+'~'՜6.(bXcE5&M0GmL*5%РóҨ0)sINƈy# 3c/%t3rn6emY϶F ;ޝS9$tP &<7 vu_ol'%&8?2.. ܤxC顨j9aE& Kb&#A|%3v&3֙1zE]fF1Iͪ!D'.n#7re&CEii>߆Xz.RB+umuh\%Si_]>#¿@Nt~_r/z[7[xDY_l’BO̶nIq ÊΏpex]wAiW[c}K{v023)fbΚvԜv2X:.x56o9>:ڧqKPp "%p$o_咽ul0nۃ|yN6u)&bs)7=*dmSey;5o+uE֧xE6 (7P4+;gn^g ?N!6Whp\~AVA](5Y)pvwn~Kr sMzxRJGhdXlP.YW9dkb>Gs:^x@բn/go)6K;1ͳ/p%^-J1~'wZӾx @Ӕ,›jn[=%z̻?n3`Y">c(NhܣUb:lwUXȤA!֯rFAUmbE[=68E kv6#2fQpTFivujr-v.Ϥ6}EA"}8nO۹[w&.Ml6%22bTgTM endstream endobj 39 0 obj << /Type /FontDescriptor /FontName /JVIAWK+CMSY10 /Flags 4 /FontBBox [-29 -960 1116 775] /Ascent 750 /CapHeight 683 /Descent -194 /ItalicAngle -14 /StemV 40 /XHeight 431 /CharSet (/H/X/arrowright/minus/periodcentered) /FontFile 38 0 R >> endobj 9 0 obj << /Type /Font /Subtype /Type1 /BaseFont /GWMOLN+CMBX10 /FontDescriptor 29 0 R /FirstChar 120 /LastChar 120 /Widths 27 0 R >> endobj 12 0 obj << /Type /Font /Subtype /Type1 /BaseFont /NDYMBJ+CMMI10 /FontDescriptor 31 0 R /FirstChar 20 /LastChar 121 /Widths 24 0 R >> endobj 10 0 obj << /Type /Font /Subtype /Type1 /BaseFont /FRUNXE+CMMI7 /FontDescriptor 33 0 R /FirstChar 110 /LastChar 110 /Widths 26 0 R >> endobj 14 0 obj << /Type /Font /Subtype /Type1 /BaseFont /WRVPXU+CMMIB10 /FontDescriptor 35 0 R /FirstChar 11 /LastChar 11 /Widths 22 0 R >> endobj 11 0 obj << /Type /Font /Subtype /Type1 /BaseFont /EWANWD+CMR10 /FontDescriptor 37 0 R /FirstChar 40 /LastChar 121 /Widths 25 0 R >> endobj 13 0 obj << /Type /Font /Subtype /Type1 /BaseFont /JVIAWK+CMSY10 /FontDescriptor 39 0 R /FirstChar 0 /LastChar 88 /Widths 23 0 R >> endobj 15 0 obj << /Type /Pages /Count 1 /Kids [5 0 R] >> endobj 40 0 obj << /Type /Catalog /Pages 15 0 R /PageMode/UseOutlines/PageLabels<>1<>2<>]>> /OpenAction 4 0 R >> endobj 41 0 obj << /Author()/Title()/Subject()/Creator(LaTeX with hyperref package)/Producer(pdfTeX-1.40.14)/Keywords() /CreationDate (D:20160802195454+02'00') /ModDate (D:20160802195454+02'00') /Trapped /False /PTEX.Fullbanner (This is MiKTeX-pdfTeX 2.9.4902 (1.40.14)) >> endobj xref 0 42 0000000006 65535 f 0000002065 00000 n 0000002085 00000 n 0000002105 00000 n 0000000015 00000 n 0000001773 00000 n 0000000016 00000 f 0000001886 00000 n 0000000061 00000 n 0000055609 00000 n 0000055891 00000 n 0000056173 00000 n 0000055750 00000 n 0000056313 00000 n 0000056032 00000 n 0000056452 00000 n 0000000017 00000 f 0000000018 00000 f 0000000019 00000 f 0000000020 00000 f 0000000021 00000 f 0000000000 00000 f 0000002158 00000 n 0000002182 00000 n 0000002695 00000 n 0000003275 00000 n 0000003736 00000 n 0000003760 00000 n 0000003784 00000 n 0000010967 00000 n 0000011186 00000 n 0000019226 00000 n 0000019464 00000 n 0000026609 00000 n 0000026827 00000 n 0000033857 00000 n 0000034083 00000 n 0000047318 00000 n 0000047606 00000 n 0000055355 00000 n 0000056510 00000 n 0000056650 00000 n trailer << /Size 42 /Root 40 0 R /Info 41 0 R /ID [<4EB1612E17F7AEFA6D4B2D6DFB66B6EB> <4EB1612E17F7AEFA6D4B2D6DFB66B6EB>] >> startxref 56924 %%EOF kafbox-2.0/doc/kaf_diagram.png000066400000000000000000000057211303712747200163400ustar00rootroot00000000000000PNG  IHDR:' pHYs IDATxߏVǿ d]%hILC  '*CRJDw%B+(wCBPRN"P^$zfrˌsu}ϵBEVz8rTꪣŚ@5`Wn]y{h,Z]m GKV hW\+:P?IOV|+ ;He5u"r.XغsP]ɡ<.";9z$A#R֕QW՗@m궟 r.߷]1m;]0(]b^7*6sA9H rNHrfDeUhu [R ZTuzx=c(Оe' H-S9,P=dȌhլ>-PU@.<^a6ղ( h"`g[T,St_DlT+-A&؂`ӭEUA\  0;]S@ N@ 6uI!]X#"v!c% 2сtѵ>Vt]af9wPiVƾ@ Uܭ&iPqXUD|_:[u]*(ϳb5V gC13 F9NʆuQq?ƹf+Ƕtסbs:sqϳV? ;{,@DA;[{hm +𙿧8|lEA[oxC@MO ug6gM2-tL='TJTx/bA N$s0"esJDtp`Cɑ@' `&0>;^j,Q3(~zehM?_MdaᘽO?xL#h ~M1zƃŠ3yy c ȋi80Zqr\ƕSi@dHGz;ņv{8G1Ww(^N )xP;Z.Z'}_97'xk`jb bqOjWj9xA;Xu :{gi`T{&Д]HD$Ty B .qaIr}\EQ^K%aF5rYuDE$H$H$H$H$H$H$H$H$H$};,PiK<,P߹Pe~ @&P7C3vuTcXi9bXP09_ I bw pGRGN:!"I UbcІE$Q tNE*zHM/LwEsp]Otӗ#=< b8he?+U\7P]@= 3@ [iyEĂEh5HZ,ҷIG<ʟ}?q>1G·FꎅN>0L}nIֲv:Xpx_iG[ŠPӛ72h׽̊A[~A)ryھw@]?opP :-3eC ry]?LZ!3Ah=2+Iw @\@D v#n _"R&4z*%΁[5䤏<6Z@٤2U_H7b4YS2Q4h.: SPN/e{2U]Ra(R Uc(NXԺ.c@huJ0@.P)'Pe'*GuʉIIIIIIIIIIIP Ra'M ~_~-H$H$h@66Xdӳ9 bH -(S5I-p@1ʪX@o'2wP8+G@?+dn9/3G^27N}v [1N=-?6^ZS˟~Ӧ?YLO<Lu[ >&2RPG  !. & 9\>4;AxPW^[Pd\?DAD69 `oH*lK%+46@/[.A9ލD .[)@.p1ߦ~nV­9ʲ*R60YU7GJ+h"朩3?XO0;_>Bo66‡ѻ]WX|kͺ5M}lx]Ľန@и;4?:P3@cWh)Uc85"-QzިK|:xTJXߞd!}{GOĎwGdwiY@n4LdȮ:L$|~re Y'/qtTURFf뒢 QHvjWIENDB`kafbox-2.0/doc/kafbox_doc.pdf000066400000000000000000010012451303712747200161750ustar00rootroot00000000000000%PDF-1.5 % 4 0 obj << /S /GoTo /D (chapter.1) >> endobj 7 0 obj (Introduction) endobj 8 0 obj << /S /GoTo /D (section.1.1) >> endobj 11 0 obj (Goals of the toolbox) endobj 12 0 obj << /S /GoTo /D (chapter.2) >> endobj 15 0 obj (First use) endobj 16 0 obj << /S /GoTo /D (section.2.1) >> endobj 19 0 obj (Installation) endobj 20 0 obj << /S /GoTo /D (section.2.2) >> endobj 23 0 obj (Directories included in the toolbox) endobj 24 0 obj << /S /GoTo /D (section.2.3) >> endobj 27 0 obj (Versions and compatibility) endobj 28 0 obj << /S /GoTo /D (chapter.3) >> endobj 31 0 obj (Kernel adaptive filtering) endobj 32 0 obj << /S /GoTo /D (section.3.1) >> endobj 35 0 obj (Framework for online kernel methods) endobj 36 0 obj << /S /GoTo /D (section.3.2) >> endobj 39 0 obj (Kernel adaptive filtering algorithms) endobj 40 0 obj << /S /GoTo /D (section.3.3) >> endobj 43 0 obj (Algorithm code structure) endobj 44 0 obj << /S /GoTo /D (section.3.4) >> endobj 47 0 obj (Template for kernel adaptive filtering algorithms) endobj 48 0 obj << /S /GoTo /D (section.3.5) >> endobj 51 0 obj (Performing online learning) endobj 52 0 obj << /S /GoTo /D (section.3.6) >> endobj 55 0 obj (Example learning script: time-series prediction) endobj 56 0 obj << /S /GoTo /D (chapter.4) >> endobj 59 0 obj (Algorithm profiler) endobj 60 0 obj << /S /GoTo /D (chapter.5) >> endobj 63 0 obj (List of included algorithms) endobj 64 0 obj << /S /GoTo /D (chapter.6) >> endobj 67 0 obj (About) endobj 68 0 obj << /S /GoTo /D [69 0 R /Fit] >> endobj 73 0 obj << /Length 510 /Filter /FlateDecode >> stream xڅMo@{%﵏)m*88&])ƭgv'(53cFvgn%ׄ[*V5Zmj*K ߌE)L=n(`3L1ߚ{RjE%8ӚWhشɿD7K\Ma,җԷZv J0n/9qS,R2kfLߥ 1R6^l^ ݝ+xny}8[;qpC G08-*[\ߎ}n?'\O G?,bu[Z?$.'/Z endstream endobj 69 0 obj << /Type /Page /Contents 73 0 R /Resources 72 0 R /MediaBox [0 0 612 792] /Parent 79 0 R /Group 78 0 R >> endobj 70 0 obj << /Type /XObject /Subtype /Form /FormType 1 /PTEX.FileName (./fig/kafbox_logo.pdf) /PTEX.PageNumber 1 /PTEX.InfoDict 80 0 R /BBox [0 0 51.200001 51.200001] /Group 78 0 R /Resources << /ExtGState << /a0 << /CA 1 /ca 1 >> >>>> /Length 333 /Filter /FlateDecode >> stream xN@E ;Y4HHEHJgƱxD [zv^gz_y68ؼ$iٲW6eNK0u7[$^Er{b>FE=k\]NT ;Rf26laN :'hϧ.3txvhD5p#`˕β4Y 6ZZac3N $!ldP, iD>`B>׎h aҡF3C'AST'v9PYZk%t%9 l@L' ɲ ԣIbeU< endstream endobj 80 0 obj << /Creator (cairo 1.14.1 \(http://cairographics.org\)) /Producer (cairo 1.14.1 \(http://cairographics.org\)) >> endobj 78 0 obj << /Type /Group /S /Transparency /I true /CS /DeviceRGB >> endobj 71 0 obj << /Type /XObject /Subtype /Form /FormType 1 /PTEX.FileName (./fig/unican_logo.pdf) /PTEX.PageNumber 1 /PTEX.InfoDict 81 0 R /BBox [0 0 141.399994 141.399994] /Group << /Type /Group /S /Transparency /CS /DeviceRGB >> /Resources << /ExtGState << /a0 << /CA 1 /ca 1 >> >>>> /Length 3004 /Filter /FlateDecode >> stream xmYI,)y8`4,ioa#r -A ^On?fq%>Esܿs~śauOm?wOOh宐=[[h>#8fޮ} ! On )=?ǻg~̔/ O@y V8(|Î 1fS{´}םSJ4X˓KSc|\9Ru_pR6̘8,7wfnO6;Ag#.>P baX qA?LE ^Ցd೸PNR]%١a T#G1HB0s=0hTDVZ5Id8ou_&o_|歁 g*|32"{ C8ꞇST(oGccPks/Toœ)_pi #ϛ*ዎ'/lOiQoL8xW|ӝ/B!MP?npq-鸶=9FER Ū'jf҅b(W!C.#R {."g "(6`=:+>qV f9֗~b( #.ģt:&Ԍ"CZen o4N6&rˎ"JZaC"|}rdY@1& HfcEH`v59;wV;v6iM4w.AAFb5(:tVCb,\QR^Oj+D 00VZr˓4FR)*If"a!֙QBZ_Ut!YeӚa%4 Y֝܋R[vYH1\YᕢE{qX -F$6vpPzúNrӭ;@hM5Wސ("sú$RcK\ߊHr.cg5Y"P5DX9jeT֔ %_JlnheW.1\w;+"Jz[Z?dޣtPʖ.!B he9Ukəb6Ɲz/v$U1 Yi5\/\hAC.iF, [u{,kqJVgS1]|_ ls|%E `]/!s"v` ?"ײK66<yr)5?v(k5 \6QݤF}쑪M/ ӚmX{᣾ 1{ ce6]U#iJYL 15sUY/ʖ jP6]B4qnba띮ZLn7ޗhl8Ϊ-!#] ,8-vh(wioHɮECQ4o 8NxzZ8[`g%hL*=L&# $` t#E$3K39Ȯ HF.$)l0.C~(UjbʲvPj!mmΘ:BFa8Lʵi *n-H '`voBl{97|cӇVMp.50k(2Fqr]5I8#I !JuG7B#J&< 7*ѓ06$l&5KST|,w4)01˖ڲF$Q~a]HTª4:0S KAvK>\ .V &Q%AةQ,75746$2a 'i4ZJ[~1|u@;JL?`]7UIL&yDž-M8w^OHy8wòɷdZ{+8.|$N¡ƞ0/6iΑs;t y7^G =طf0ٴC7t*P^ovq^~mn*=OMt 1oj.~]k|?AMI]l 6Z>b\.03=J$ɧ/g8I=/a84jնNPS-OŚ ]~-Dy9&!)0pvXZCq|:{w7(SaRxYiUGwqcXE!zD~ \[%޴rMv J+! ":"_^,KLi,> endobj 74 0 obj << /D [69 0 R /XYZ 132.768 705.06 null] >> endobj 75 0 obj << /D [69 0 R /XYZ 133.768 667.198 null] >> endobj 72 0 obj << /ColorSpace 3 0 R /Pattern 2 0 R /ExtGState 1 0 R /Font << /F15 76 0 R /F16 77 0 R >> /XObject << /Im1 70 0 R /Im2 71 0 R >> /ProcSet [ /PDF /Text ] >> endobj 84 0 obj << /Length 510 /Filter /FlateDecode >> stream xڅTMs0 hf{t3NoeC7ʈ%a\%=?Kz2yn_!ͽ*ꓗi /0^}~fZxgd$qX%3Asp?)=HeH1T ~@A5ɢ_ˉO D:sy!*MB/R誢n1U\dq.ڡ s\aSl-jn0Ҟmݚ`4emnHjsCGb^= ա^9,rX.PMiQˑ`PLE.FGZ9AVq`gpKɓ–NؒH', R/`ُ {G)Zp8dIrMO3F췺(g s]U(ZZjϗ"Ls{4}4̈́_yG~ Axc$b5<-p^/ XS{Z(2O endstream endobj 83 0 obj << /Type /Page /Contents 84 0 R /Resources 82 0 R /MediaBox [0 0 612 792] /Parent 79 0 R >> endobj 82 0 obj << /ColorSpace 3 0 R /Pattern 2 0 R /ExtGState 1 0 R /Font << /F35 85 0 R /F8 86 0 R >> /ProcSet [ /PDF /Text ] >> endobj 105 0 obj << /Length 716 /Filter /FlateDecode >> stream xVMs0W(l-m Sddʿgi I((i' NOp\N^1P+H2Ic:JdV%W]ߥ g2_g P$E^H*}N2IVf(rO3XՐ*khH7'!y!DLC2)E} q!wg Q\54S<YxQ!X&ʈ7U?lRRFFDXjՓAOK4#+g^ Bqhd3)^r*DZ}3tv!k endstream endobj 104 0 obj << /Type /Page /Contents 105 0 R /Resources 103 0 R /MediaBox [0 0 612 792] /Parent 79 0 R /Annots [ 87 0 R 88 0 R 89 0 R 90 0 R 91 0 R 92 0 R 93 0 R 94 0 R 95 0 R 96 0 R 97 0 R 98 0 R 99 0 R 100 0 R 101 0 R 102 0 R ] >> endobj 87 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [132.772 514.77 212.874 523.681] /A << /S /GoTo /D (chapter.1) >> >> endobj 88 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [147.716 502.814 261.581 511.725] /A << /S /GoTo /D (section.1.1) >> >> endobj 89 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [132.772 480.897 193.747 489.808] /A << /S /GoTo /D (chapter.2) >> >> endobj 90 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [147.716 468.941 222.215 477.852] /A << /S /GoTo /D (section.2.1) >> >> endobj 91 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [147.716 456.986 324.249 465.897] /A << /S /GoTo /D (section.2.2) >> >> endobj 92 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [147.716 443.094 288.992 453.942] /A << /S /GoTo /D (section.2.3) >> >> endobj 93 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [132.772 421.176 272.83 432.024] /A << /S /GoTo /D (chapter.3) >> >> endobj 94 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [147.716 411.158 335.097 420.069] /A << /S /GoTo /D (section.3.1) >> >> endobj 95 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [147.716 397.266 328.289 408.114] /A << /S /GoTo /D (section.3.2) >> >> endobj 96 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [147.716 385.311 282.904 396.159] /A << /S /GoTo /D (section.3.3) >> >> endobj 97 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [147.716 373.355 384.772 384.204] /A << /S /GoTo /D (section.3.4) >> >> endobj 98 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [147.716 361.4 288.522 372.248] /A << /S /GoTo /D (section.3.5) >> >> endobj 99 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [147.716 349.445 376.996 360.293] /A << /S /GoTo /D (section.3.6) >> >> endobj 100 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [132.772 327.527 241.427 338.375] /A << /S /GoTo /D (chapter.4) >> >> endobj 101 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [132.772 305.609 284.608 316.458] /A << /S /GoTo /D (chapter.5) >> >> endobj 102 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [132.772 285.629 181.603 294.54] /A << /S /GoTo /D (chapter.6) >> >> endobj 106 0 obj << /D [104 0 R /XYZ 133.768 667.198 null] >> endobj 103 0 obj << /ColorSpace 3 0 R /Pattern 2 0 R /ExtGState 1 0 R /Font << /F37 107 0 R /F35 85 0 R /F8 86 0 R >> /ProcSet [ /PDF /Text ] >> endobj 111 0 obj << /Length 505 /Filter /FlateDecode >> stream x}S0 +|L1 MփgwY;r˴3pYIz9Fv,ggi$T5--ҏxs0KY % (_VS*H0F-2b 醃'Ud ee[1? 'Hm-h&ss-71S+Oȝl?xߦ7M p J b%IS>湒}ي%zAF9 <tpzg{9)_"5tcɸaS/7*W" jGocb鼾oSyo]{߶:?M(tPGo= aūmŐYbW%W{{5> endobj 108 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [431.597 366.479 438.57 377.327] /A << /S /GoTo /D (chapter.5) >> >> endobj 112 0 obj << /D [110 0 R /XYZ 132.768 705.06 null] >> endobj 5 0 obj << /D [110 0 R /XYZ 133.768 667.198 null] >> endobj 9 0 obj << /D [110 0 R /XYZ 133.768 492.852 null] >> endobj 113 0 obj << /D [110 0 R /XYZ 133.768 447.176 null] >> endobj 114 0 obj << /D [110 0 R /XYZ 133.768 415.295 null] >> endobj 109 0 obj << /ColorSpace 3 0 R /Pattern 2 0 R /ExtGState 1 0 R /Font << /F37 107 0 R /F8 86 0 R >> /ProcSet [ /PDF /Text ] >> endobj 118 0 obj << /Length 953 /Filter /FlateDecode >> stream xڥVK8 ϯ>X˖[EKXV:v`+D9tI "}a} qJgv1!ʋ(+8XmkG֌I*T_h&*sf4JeI &'ɜ $B<d p’1]m;澈JRšr\2L(΅$T}j6.8}8 OM mXt:B@ޔQT>kѤ8ϙ׳\sNi:\Ґ&+'I6~2Rً1Ke-

> endobj 115 0 obj << /Type /Annot /Border[0 0 1]/H/I/C[0 1 1] /Rect [132.772 213.482 409.733 223.644] /Subtype/Link/A<> >> endobj 119 0 obj << /D [117 0 R /XYZ 132.768 705.06 null] >> endobj 13 0 obj << /D [117 0 R /XYZ 133.768 667.198 null] >> endobj 17 0 obj << /D [117 0 R /XYZ 133.768 492.852 null] >> endobj 120 0 obj << /D [117 0 R /XYZ 133.768 469.038 null] >> endobj 122 0 obj << /D [117 0 R /XYZ 133.768 437.158 null] >> endobj 21 0 obj << /D [117 0 R /XYZ 133.768 404.267 null] >> endobj 25 0 obj << /D [117 0 R /XYZ 133.768 289.725 null] >> endobj 116 0 obj << /ColorSpace 3 0 R /Pattern 2 0 R /ExtGState 1 0 R /Font << /F37 107 0 R /F8 86 0 R /F50 121 0 R /F14 123 0 R >> /ProcSet [ /PDF /Text ] >> endobj 132 0 obj << /Length 1818 /Filter /FlateDecode >> stream xڥXK [UӴ%%8ɦ*LuG͙fQxN[uH>Pi.y˝ʅ?֢ծ(wy"mFU)q[;ZTІֻ#l?LY+?Ku *( %EgTڧa өéIQ8Ɖލ|2^ߟg^ggvb iƵBFpÁS@~錧jt0e-_oXlDU_Pka}#`́ѧoBz (5F oVօ@nfcn20tf.eq< AO~2Ox.i}lcB':'T]&/)4e1ފ"P8S!K/c^ĬhQ&jQPʕIB5G˜ b8]bfXtŰId0Z"[:g9GlL*oFN oiKDCZ*Bwn“ת0չ2չcȧa?=|=&UT'܏C0TkP_;S@nr^R=NA?< j4e1fReG0,'kͱwc $޹|Ӌ,u#MVpz[H!ެK٩ Ocr"wꖒU΃x-to&t;3WC9ĂSMPXBIaTMTH'LZ2uI($nၖ@H(⇄P՗s#tvMAr# 'EBX;6##~~^ Cx) Rx!d6ɼe9T9l<oܙ!LR g1Pܻ2qBY'\::w瑶5>FnOtւ~7>ynm:u [JFj=nY'=AFfqlMW̜)egAOasDnp Dy# $jݐЏD*MenU r[tם[mk͎ӊq*j4_E/WaI3,x+0!;v=%;z$E$Dzs-3ཌྷɍ.\%\ mqN ;bi\B\b> endobj 124 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[0 1 0] /Rect [393.86 440.147 421.174 449.058] /A << /S /GoTo /D (cite.blum1998line) >> >> endobj 125 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [223.986 313.863 238.709 324.711] /A << /S /GoTo /D (figure.3.1) >> >> endobj 129 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[0 1 0] /Rect [135.54 184.293 171.018 193.093] /A << /S /GoTo /D (cite.kivinen2004online) >> >> endobj 133 0 obj << /D [131 0 R /XYZ 132.768 705.06 null] >> endobj 29 0 obj << /D [131 0 R /XYZ 133.768 667.198 null] >> endobj 33 0 obj << /D [131 0 R /XYZ 133.768 488.032 null] >> endobj 135 0 obj << /D [131 0 R /XYZ 133.768 427.591 null] >> endobj 137 0 obj << /D [131 0 R /XYZ 133.768 407.321 null] >> endobj 138 0 obj << /D [131 0 R /XYZ 133.768 388.988 null] >> endobj 139 0 obj << /D [131 0 R /XYZ 133.768 358.146 null] >> endobj 37 0 obj << /D [131 0 R /XYZ 133.768 168.471 null] >> endobj 130 0 obj << /ColorSpace 3 0 R /Pattern 2 0 R /ExtGState 1 0 R /Font << /F37 107 0 R /F8 86 0 R /F11 134 0 R /F35 85 0 R /F10 136 0 R /F14 123 0 R >> /ProcSet [ /PDF /Text ] >> endobj 146 0 obj << /Length 1325 /Filter /FlateDecode >> stream xڽWKs6WHD0"6ΤIi[Lj>T],(QtiO>]&MDofog9Ke%\FU;;{~jvvQD%+b#.%˗E%%es|anVd|<~mŕr3U`En sD6k0_Vj0}@jQL^Ƶ`ۂsVfA_}2^q!)yIGZ&EJ4!jnzkupkUTV6,@C?(Lp񻏺q y+/e8jZ;n"llȴF!ި-Sm,X +IlEp<6YzO`}"2M_]uiט6R[1GˤHMGZ  MQ7{U2sU*ґO4>)k;Y=Znv"5 1 襁RО$b*/&#TRCD_`hyTKg|4)C;W"_x,`")/IMtf "X#\}gGKV#mɃ;R.K{&T`\?~ 0E<"{5MMH|qe]Z#و =UlABoB}`JUTv,qbF@]Yu?pe={@w Jnmq>H XZZIk~mnnB\( Ism!<|{Nʸòu9&VVu cMJ_z&A 0Ӧxy"ڊPS  9:"%oeM2 jp<Pnjl҇yg$Ks`a pf\iW[$ }oU41c"V;6:b{(,sZ54 Ju[ ;SV$;?H;k1rt<AzhVDl)Cd> endobj 126 0 obj << /Interpolate false /Type /XObject /Subtype /Form /FormType 1 /PTEX.FileName (./kaf_diagram.pdf) /PTEX.PageNumber 1 /PTEX.InfoDict 150 0 R /BBox [0 0 297.075 142.185] /Resources << /ColorSpace << /pgfprgb [/Pattern/DeviceRGB] >>/Pattern << >>/ExtGState << >>/Font << /F34 153 0 R /F10 156 0 R /F8 159 0 R /F11 162 0 R /F14 165 0 R /F52 168 0 R >> /ProcSet [ /PDF /Text ] >> /Length 1634 /Filter /FlateDecode >> stream xYn7W07 A,DN1'!:$D6f{8ܾ!4 XX9J%~^ rNRE'.;<H07M Ɗ'ă~|/B#)+=Eb&Zdx#~= %LDA}Geͨ;F~-%aJR~TIjg A5\~+v/QnwKZ;14BQ&:JOTLdĠAY٤:jX`&h_>ݾz}>c>~ڀK E|\Pېc[]2[3EN(A>C u(N;㲎 I0kb/^%STWLЂ|1<-y/u~4oY9=u3d5v0c&1bCڤRUgY j'N ݗ ͏P_䭭,$ uZd%줜N(˩-G.A |߮zZjt㣃T8nLYdp&d*OAMu.-o'Pǿ)2bUQb/%녬6`5:0J뵅z] #FZXk&$.3qAM}>v$`]Z/J~M7Hc@G@'|l8B|?Є9VUܻ4(.e f:}jL2f(i"P] Ɉ7짵aVV):*;}Hl2r_:ؓrkn"UEƐl=җL\rEiI3$W8EX5T@S>N z1KKi+fR 풴''S|i fs@vIkɊy'?V2@j+>9V* F}W@Z(:(:XˀCzIYp_aـj5.[dƀdR >~2ۮ-g5ÍV#n8Zjȶ"Ę 'n1Pԁ\zW6Й,{>0,MDBOYLvĽbk 'T6JzoL&؂(mCtjQK*&eg]{]&xb!,IFJ7ୌDQk'5@FNtvǚ>w쑚7)4볰LJ;XM]޺oإ&/]uy{njQWִ Q>(!!VǣfKIէ>{i(a "{B{,$}x_3#neLr0?Ll[gti>_8 endstream endobj 150 0 obj << /Author () /Title () /Subject () /Creator (LaTeX with hyperref package) /Producer (pdfTeX-1.40.14) /Keywords () /CreationDate (D:20160802195454+02'00') /ModDate (D:20160802195454+02'00') /Trapped /False /PTEX.Fullbanner (This is MiKTeX-pdfTeX 2.9.4902 \(1.40.14\)) >> endobj 153 0 obj << /Type /Font /Subtype /Type1 /FirstChar 120 /LastChar 120 /Widths 169 0 R /FontDescriptor 151 0 R /BaseFont 170 0 R /Encoding 152 0 R >> endobj 156 0 obj << /Type /Font /Subtype /Type1 /FirstChar 110 /LastChar 110 /Widths 171 0 R /FontDescriptor 154 0 R /BaseFont 172 0 R /Encoding 155 0 R >> endobj 159 0 obj << /Type /Font /Subtype /Type1 /FirstChar 40 /LastChar 121 /Widths 173 0 R /FontDescriptor 157 0 R /BaseFont 174 0 R /Encoding 158 0 R >> endobj 162 0 obj << /Type /Font /Subtype /Type1 /FirstChar 20 /LastChar 121 /Widths 175 0 R /FontDescriptor 160 0 R /BaseFont 176 0 R /Encoding 161 0 R >> endobj 165 0 obj << /Type /Font /Subtype /Type1 /FirstChar 0 /LastChar 88 /Widths 177 0 R /FontDescriptor 163 0 R /BaseFont 178 0 R /Encoding 164 0 R >> endobj 168 0 obj << /Type /Font /Subtype /Type1 /FirstChar 11 /LastChar 11 /Widths 179 0 R /FontDescriptor 166 0 R /BaseFont 180 0 R /Encoding 167 0 R >> endobj 169 0 obj [ 606.9] endobj 171 0 obj [ 706.2] endobj 173 0 obj [ 388.9 388.9 500 777.8 277.8 333.3 277.8 500 500 500 500 500 500 500 500 500 500 500 277.8 277.8 277.8 777.8 472.2 472.2 777.8 750 708.3 722.2 763.9 680.6 652.8 784.7 750 361.1 513.9 777.8 625 916.7 750 777.8 680.6 777.8 736.1 555.6 722.2 750 750 1027.8 750 750 611.1 277.8 500 277.8 500 277.8 277.8 500 555.6 444.4 555.6 444.4 305.6 500 555.6 277.8 305.6 527.8 277.8 833.3 555.6 500 555.6 527.8 391.7 394.4 388.9 555.6 527.8 722.2 527.8 527.8] endobj 175 0 obj [ 576.2 583.3 602.6 494 437.5 570 517 571.4 437.2 540.3 595.8 625.7 651.4 622.5 466.3 591.4 828.1 517 362.8 654.2 1000 1000 1000 1000 277.8 277.8 500 500 500 500 500 500 500 500 500 500 500 500 277.8 277.8 777.8 500 777.8 500 530.9 750 758.5 714.7 827.9 738.2 643.1 786.3 831.3 439.6 554.5 849.3 680.6 970.1 803.5 762.8 642 790.6 759.3 613.2 584.4 682.8 583.3 944.4 828.5 580.6 682.6 388.9 388.9 388.9 1000 1000 416.7 528.6 429.2 432.8 520.5 465.6 489.6 477 576.2 344.5 411.8 520.6 298.4 878 600.2 484.7 503.1 446.4 451.2 468.8 361.1 572.5 484.7 715.9 571.5 490.3] endobj 177 0 obj [ 777.8 277.8 777.8 500 777.8 500 777.8 777.8 777.8 777.8 777.8 777.8 777.8 1000 500 500 777.8 777.8 777.8 777.8 777.8 777.8 777.8 777.8 777.8 777.8 777.8 777.8 1000 1000 777.8 777.8 1000 1000 500 500 1000 1000 1000 777.8 1000 1000 611.1 611.1 1000 1000 1000 777.8 275 1000 666.7 666.7 888.9 888.9 0 0 555.6 555.6 666.7 500 722.2 722.2 777.8 777.8 611.1 798.5 656.8 526.5 771.4 527.8 718.7 594.9 844.5 544.5 677.8 762 689.7 1200.9 820.5 796.1 695.6 816.7 847.5 605.6 544.6 625.8 612.8 987.8 713.3] endobj 179 0 obj [ 760.6] endobj 167 0 obj << /Type /Encoding /Differences [11/alpha] >> endobj 164 0 obj << /Type /Encoding /Differences [0/minus/periodcentered 33/arrowright 72/H 88/X] >> endobj 161 0 obj << /Type /Encoding /Differences [20/kappa 59/comma 101/e/f 114/r 121/y] >> endobj 158 0 obj << /Type /Encoding /Differences [40/parenleft/parenright 43/plus 65/A 78/N 94/circumflex 97/a 100/d/e 103/g/h/i 108/l/m/n/o/p 114/r/s/t 118/v 121/y] >> endobj 155 0 obj << /Type /Encoding /Differences [110/n] >> endobj 152 0 obj << /Type /Encoding /Differences [120/x] >> endobj 128 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[0 1 0] /Rect [135.54 490.144 175.03 498.944] /A << /S /GoTo /D (cite.parreira2012stochastic) >> >> endobj 143 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [326.11 264.545 333.084 275.394] /A << /S /GoTo /D (chapter.4) >> >> endobj 147 0 obj << /D [145 0 R /XYZ 132.768 705.06 null] >> endobj 141 0 obj << /D [145 0 R /XYZ 186.565 515.05 null] >> endobj 41 0 obj << /D [145 0 R /XYZ 133.768 238.573 null] >> endobj 144 0 obj << /ColorSpace 3 0 R /Pattern 2 0 R /ExtGState 1 0 R /Font << /F8 86 0 R /F14 123 0 R /F11 134 0 R /F7 148 0 R /F37 107 0 R /F50 121 0 R >> /XObject << /Im3 126 0 R >> /ProcSet [ /PDF /Text ] >> endobj 185 0 obj << /Length 2303 /Filter /FlateDecode >> stream x[s_L &ɻř4ѴM"!=TH>]GPIon@``DwW7[̂%`2 0$\"By0Ƀч_7o1D6F$1,Ly݌pXtmY.а> k0M"zZeC-۷[p@0EmxPiN;e{ 4UQ5iQ" M';c ;M !"1UنL꫕VIiiTg7rN5iW@W>=~pm4KBfl+-j4 ʽ-i+TTzpy nRT}yeE. %3!év@UZϚbչu9KܨQLu4˺Qo lZ7]ki,e+w~ W.{ca9hݘGQx^hkc #ǺԨ4OP+k`uXm1-{qٰⱦ܈2Ի}&}-Gyi*rr[hfuFoEuVA-r^5I'[]6joH\]SiZvv i#MZWi@h+(f(a v6p[O-jaXI85pj@O%IȔ6_BH=Y\ =kP=VZؑ[]{˲zm{5c|cQ%QP"LN?#0N'>=3{íH9ZiSSMUr|ʶ` ]3=^*+Sg}вnlas_׬6|ZJmqY15cݓ*܅Fq\/ GFkr[uk=9m]꫹6s,z;'\TmW e[+#Z7=.<:(@>.CjQIaLZYBPLE 0pF '#bj  :<>p_uuվ[Bd3`hkk;T@`! D %΀4f1 _?`JH `i~@4cpp8hT0Mxon>0)#lJ08. Qܣ!޳: (irEv0[W'x)D8n(Z_Qr =38牖vO>ZD97͊ޱ}1gBڣt@xI<,ph]jģtW~8\1iW򇡝Ǡc6ק8” D˯'->jx"87t҉r Hć/m/ .6v+Բ.~krkF_X_!٣/zS_` )H1,4E 2"_ 4UVswׇ_D?US4oi׭N&W^'??I"Q 6$Б$A"b?I#G;ٓ嫀YQZa6:i#GL'':+s"y%cd R KGπ1×5Z6@!^S"^~Lg'E5`0p:,#_G`!k}^yn[n@l{_Ǭ1& 1E Awvso8m'A'&[/=fP9~]&.]2$X P%M&t|> K;vD8#X^/;w݇!|9G16]\B6\-TzXU LV٬u awq, a/4P<1hnIذcLDX<9[?7gX,$N s;"ϳ}ęxu޴=78gue?>zbq0b}8D endstream endobj 184 0 obj << /Type /Page /Contents 185 0 R /Resources 183 0 R /MediaBox [0 0 612 792] /Parent 149 0 R /Annots [ 182 0 R ] >> endobj 182 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[0 1 0] /Rect [172.346 504.807 214.577 514.732] /A << /S /GoTo /D (cite.wilson2014best) >> >> endobj 186 0 obj << /D [184 0 R /XYZ 132.768 705.06 null] >> endobj 187 0 obj << /D [184 0 R /XYZ 133.768 595.522 null] >> endobj 188 0 obj << /D [184 0 R /XYZ 133.768 575.597 null] >> endobj 45 0 obj << /D [184 0 R /XYZ 133.768 427.029 null] >> endobj 189 0 obj << /D [184 0 R /XYZ 133.768 375.32 null] >> endobj 190 0 obj << /D [184 0 R /XYZ 133.768 375.32 null] >> endobj 192 0 obj << /D [184 0 R /XYZ 133.768 365.856 null] >> endobj 193 0 obj << /D [184 0 R /XYZ 133.768 356.391 null] >> endobj 194 0 obj << /D [184 0 R /XYZ 133.768 346.927 null] >> endobj 195 0 obj << /D [184 0 R /XYZ 133.768 337.462 null] >> endobj 196 0 obj << /D [184 0 R /XYZ 133.768 327.998 null] >> endobj 197 0 obj << /D [184 0 R /XYZ 133.768 318.533 null] >> endobj 198 0 obj << /D [184 0 R /XYZ 133.768 309.069 null] >> endobj 199 0 obj << /D [184 0 R /XYZ 133.768 299.604 null] >> endobj 200 0 obj << /D [184 0 R /XYZ 133.768 290.14 null] >> endobj 201 0 obj << /D [184 0 R /XYZ 133.768 280.675 null] >> endobj 202 0 obj << /D [184 0 R /XYZ 133.768 271.211 null] >> endobj 203 0 obj << /D [184 0 R /XYZ 133.768 261.746 null] >> endobj 204 0 obj << /D [184 0 R /XYZ 133.768 252.282 null] >> endobj 205 0 obj << /D [184 0 R /XYZ 133.768 242.817 null] >> endobj 206 0 obj << /D [184 0 R /XYZ 133.768 233.353 null] >> endobj 207 0 obj << /D [184 0 R /XYZ 133.768 223.888 null] >> endobj 208 0 obj << /D [184 0 R /XYZ 133.768 214.424 null] >> endobj 209 0 obj << /D [184 0 R /XYZ 133.768 204.959 null] >> endobj 210 0 obj << /D [184 0 R /XYZ 133.768 195.495 null] >> endobj 211 0 obj << /D [184 0 R /XYZ 133.768 186.03 null] >> endobj 212 0 obj << /D [184 0 R /XYZ 133.768 176.566 null] >> endobj 213 0 obj << /D [184 0 R /XYZ 133.768 167.101 null] >> endobj 214 0 obj << /D [184 0 R /XYZ 133.768 157.637 null] >> endobj 215 0 obj << /D [184 0 R /XYZ 133.768 148.172 null] >> endobj 216 0 obj << /D [184 0 R /XYZ 133.768 138.708 null] >> endobj 217 0 obj << /D [184 0 R /XYZ 133.768 129.243 null] >> endobj 183 0 obj << /ColorSpace 3 0 R /Pattern 2 0 R /ExtGState 1 0 R /Font << /F14 123 0 R /F8 86 0 R /F50 121 0 R /F7 148 0 R /F37 107 0 R /F17 191 0 R >> /ProcSet [ /PDF /Text ] >> endobj 221 0 obj << /Length 1949 /Filter /FlateDecode >> stream x[mo_/EBKТ׻k[r(ie(v;.%v\K׸a ;3,$HH %Œ>ЄrZDI Bd>kpݔ&6IٿJL"|_^:H|P*ssj`\&c45 yN[DT/{3EHHRdE^L7 O^}i 4loƢscfev+ůiGx8B4(=Nq+d8M,Ƽ2wE:QWӖqB $q p<ۜ'4J-?nyvϮld,}]eFZ)cq 4^`]2axWBtPK>A,T4"Xu GX <챈·AKk=vd "xC0ލjiTޥ%`yqKʘqx4Sqy7} !'T !:yy@m.ud?r4RbSz*)긪-V ⸞QJθC1Wcs­zb6Uٺ[MDR'')(7j8R!j"o]^5 GaɭFfj{"+@| 4qpx<(_\iïκ\{ q#%Sy rȓ`\> /<*qa 9`F1q7NFAucNo"v'h8j AtAGї7oHDɋγd(TB1G _>, z4 L8[c6\O!,zHrSp ZƍS eX<3x> Z*hF|.r(ȋ]l\5KWN qCEp,X-{+>(NLƙ';}g\8'}F9mQ$sQ^3z`hn7: H`8^/Ƭ91YyW/Q#1ӦzΏXC `k-_ЊpX3$3XLpr)E܌knU^8__^3 ª$iLҒ|'̤8Ljt&vȶus=ڕ~b̫P6W"kW4CkeY>|ULf|bxڢfYoiƖHYnSnx)4kmkt7`Ģjs2"mOr`6ð{U_u^GRyb7d;&ҷ31!m\n3r7>od>f4pv߱Ϳ_~t5R`¦53v13 S^TuVCf{zgD bӷp^OJ16?fӨ/۳]*oZM_q> endobj 222 0 obj << /D [220 0 R /XYZ 132.768 705.06 null] >> endobj 223 0 obj << /D [220 0 R /XYZ 133.768 666.7 null] >> endobj 224 0 obj << /D [220 0 R /XYZ 133.768 657.235 null] >> endobj 225 0 obj << /D [220 0 R /XYZ 133.768 647.771 null] >> endobj 226 0 obj << /D [220 0 R /XYZ 133.768 638.306 null] >> endobj 227 0 obj << /D [220 0 R /XYZ 133.768 628.842 null] >> endobj 228 0 obj << /D [220 0 R /XYZ 133.768 619.377 null] >> endobj 229 0 obj << /D [220 0 R /XYZ 133.768 609.913 null] >> endobj 230 0 obj << /D [220 0 R /XYZ 133.768 600.448 null] >> endobj 231 0 obj << /D [220 0 R /XYZ 133.768 590.984 null] >> endobj 232 0 obj << /D [220 0 R /XYZ 133.768 581.519 null] >> endobj 233 0 obj << /D [220 0 R /XYZ 133.768 572.055 null] >> endobj 234 0 obj << /D [220 0 R /XYZ 133.768 562.59 null] >> endobj 235 0 obj << /D [220 0 R /XYZ 133.768 553.126 null] >> endobj 236 0 obj << /D [220 0 R /XYZ 133.768 543.661 null] >> endobj 237 0 obj << /D [220 0 R /XYZ 133.768 534.197 null] >> endobj 238 0 obj << /D [220 0 R /XYZ 133.768 524.732 null] >> endobj 239 0 obj << /D [220 0 R /XYZ 133.768 515.268 null] >> endobj 240 0 obj << /D [220 0 R /XYZ 133.768 505.803 null] >> endobj 241 0 obj << /D [220 0 R /XYZ 133.768 496.339 null] >> endobj 242 0 obj << /D [220 0 R /XYZ 133.768 486.874 null] >> endobj 243 0 obj << /D [220 0 R /XYZ 133.768 477.41 null] >> endobj 244 0 obj << /D [220 0 R /XYZ 133.768 467.945 null] >> endobj 245 0 obj << /D [220 0 R /XYZ 133.768 458.481 null] >> endobj 246 0 obj << /D [220 0 R /XYZ 133.768 449.016 null] >> endobj 247 0 obj << /D [220 0 R /XYZ 133.768 439.552 null] >> endobj 248 0 obj << /D [220 0 R /XYZ 133.768 430.087 null] >> endobj 249 0 obj << /D [220 0 R /XYZ 133.768 420.623 null] >> endobj 250 0 obj << /D [220 0 R /XYZ 133.768 411.158 null] >> endobj 251 0 obj << /D [220 0 R /XYZ 133.768 401.694 null] >> endobj 252 0 obj << /D [220 0 R /XYZ 133.768 392.229 null] >> endobj 253 0 obj << /D [220 0 R /XYZ 133.768 382.765 null] >> endobj 254 0 obj << /D [220 0 R /XYZ 133.768 373.3 null] >> endobj 255 0 obj << /D [220 0 R /XYZ 133.768 363.836 null] >> endobj 256 0 obj << /D [220 0 R /XYZ 133.768 354.371 null] >> endobj 257 0 obj << /D [220 0 R /XYZ 133.768 344.907 null] >> endobj 258 0 obj << /D [220 0 R /XYZ 133.768 335.442 null] >> endobj 259 0 obj << /D [220 0 R /XYZ 133.768 325.978 null] >> endobj 260 0 obj << /D [220 0 R /XYZ 133.768 316.513 null] >> endobj 261 0 obj << /D [220 0 R /XYZ 133.768 307.049 null] >> endobj 262 0 obj << /D [220 0 R /XYZ 133.768 297.584 null] >> endobj 49 0 obj << /D [220 0 R /XYZ 133.768 279.702 null] >> endobj 263 0 obj << /D [220 0 R /XYZ 133.768 223.316 null] >> endobj 264 0 obj << /D [220 0 R /XYZ 133.768 225.253 null] >> endobj 265 0 obj << /D [220 0 R /XYZ 133.768 178.231 null] >> endobj 266 0 obj << /D [220 0 R /XYZ 133.768 178.231 null] >> endobj 267 0 obj << /D [220 0 R /XYZ 133.768 131.208 null] >> endobj 268 0 obj << /D [220 0 R /XYZ 133.768 131.208 null] >> endobj 219 0 obj << /ColorSpace 3 0 R /Pattern 2 0 R /ExtGState 1 0 R /Font << /F17 191 0 R /F50 121 0 R /F37 107 0 R /F8 86 0 R >> /ProcSet [ /PDF /Text ] >> endobj 271 0 obj << /Length 1669 /Filter /FlateDecode >> stream x[YSH~WZe7sl!%쑐]T %"K^I@6~Gd0Ec;wZ/[/A "qaJA .Pއ<QJe1,iMY=2KNE}%7?`ڌPqC:8c -[)7m cМ{Y[ߡ @rN 4xV]!W<'# RVK*=̀2Ay7 4a c<. ,S>ɢy3bBh g4=*a 6BQ^ιᙷAK="╝y?53,J+c \3Lu6n0W%^[t`]U;xxP-9ud&̋`nkN<𧶈,PG5EO RmI؛Dk5gn}IA?bW`%?~ ncm|Y30[q5ְܚ $mQ{>EteQ\3-M4lhŋ~B/"LK܂Q YS/A8\k$u{psy$~SVɚc!)~;QM=xG nyGIV*zه,2q,׼.'ej Hk*6ܭ-eCYˉNLnn p_ftz':C=J.ƴΕcC@MHFLi@9id 9Y6&4 k9oЅTٷ*UlV9'l`3Zqwh]Ϛeǭkjp6H`8_j8ς<5aq7CmdC[ysv,p2qMA %}`¤mՆJ܍S͆znQ(^ R]\͏Wv(Em^zecgv堤O%H !0@8PyNŖ:YK-lZK˽>udnLEiuK}dQ9v=jFt]OS1_,bq +9jS7b; LJV˅=մAq2>]f4hiU"^ kk Yᣤy`5>V`S&ۛwލ{$QB8Fy3B@ozS{#n& jn&ǘ7~"kg1҇p&?y ~G85Fa nݎӳAf_o7;{w&7 {7#@IoD8;g"r endstream endobj 270 0 obj << /Type /Page /Contents 271 0 R /Resources 269 0 R /MediaBox [0 0 612 792] /Parent 149 0 R >> endobj 272 0 obj << /D [270 0 R /XYZ 132.768 705.06 null] >> endobj 273 0 obj << /D [270 0 R /XYZ 133.768 649.321 null] >> endobj 274 0 obj << /D [270 0 R /XYZ 133.768 651.258 null] >> endobj 53 0 obj << /D [270 0 R /XYZ 133.768 614.725 null] >> endobj 275 0 obj << /D [270 0 R /XYZ 133.768 566.603 null] >> endobj 276 0 obj << /D [270 0 R /XYZ 133.768 566.603 null] >> endobj 277 0 obj << /D [270 0 R /XYZ 133.768 557.139 null] >> endobj 278 0 obj << /D [270 0 R /XYZ 133.768 547.674 null] >> endobj 279 0 obj << /D [270 0 R /XYZ 133.768 538.21 null] >> endobj 280 0 obj << /D [270 0 R /XYZ 133.768 528.745 null] >> endobj 281 0 obj << /D [270 0 R /XYZ 133.768 519.281 null] >> endobj 282 0 obj << /D [270 0 R /XYZ 133.768 509.816 null] >> endobj 283 0 obj << /D [270 0 R /XYZ 133.768 500.352 null] >> endobj 284 0 obj << /D [270 0 R /XYZ 133.768 490.887 null] >> endobj 285 0 obj << /D [270 0 R /XYZ 133.768 481.423 null] >> endobj 286 0 obj << /D [270 0 R /XYZ 133.768 471.958 null] >> endobj 287 0 obj << /D [270 0 R /XYZ 133.768 462.493 null] >> endobj 288 0 obj << /D [270 0 R /XYZ 133.768 453.029 null] >> endobj 289 0 obj << /D [270 0 R /XYZ 133.768 443.564 null] >> endobj 290 0 obj << /D [270 0 R /XYZ 133.768 434.1 null] >> endobj 291 0 obj << /D [270 0 R /XYZ 133.768 424.635 null] >> endobj 292 0 obj << /D [270 0 R /XYZ 133.768 415.171 null] >> endobj 293 0 obj << /D [270 0 R /XYZ 133.768 405.706 null] >> endobj 294 0 obj << /D [270 0 R /XYZ 133.768 396.242 null] >> endobj 295 0 obj << /D [270 0 R /XYZ 133.768 386.777 null] >> endobj 296 0 obj << /D [270 0 R /XYZ 133.768 377.313 null] >> endobj 297 0 obj << /D [270 0 R /XYZ 133.768 367.848 null] >> endobj 298 0 obj << /D [270 0 R /XYZ 133.768 358.384 null] >> endobj 299 0 obj << /D [270 0 R /XYZ 133.768 348.919 null] >> endobj 269 0 obj << /ColorSpace 3 0 R /Pattern 2 0 R /ExtGState 1 0 R /Font << /F8 86 0 R /F50 121 0 R /F37 107 0 R /F17 191 0 R >> /ProcSet [ /PDF /Text ] >> endobj 302 0 obj << /Length 406 /Filter /FlateDecode >> stream xmRn Wą Ӧmҵ0ʨAIj{sZDˆC>%dY$  IYNV(yEQ y3QLnhm7l;ebp[Y̔Yu*X^% ypXIgAGIOg0sİ4:F6*<\zJ+@2]U˩@ И2GmF9*­'.F*ow] ~i{=-r[m`> w.?3AJ!1;'9nԨ!6lC mzX!z { )C_>{gcL54Q7z+IrDF2~Ћr_W y=U endstream endobj 301 0 obj << /Type /Page /Contents 302 0 R /Resources 300 0 R /MediaBox [0 0 612 792] /Parent 149 0 R >> endobj 303 0 obj << /D [301 0 R /XYZ 132.768 705.06 null] >> endobj 57 0 obj << /D [301 0 R /XYZ 133.768 667.198 null] >> endobj 300 0 obj << /ColorSpace 3 0 R /Pattern 2 0 R /ExtGState 1 0 R /Font << /F37 107 0 R /F8 86 0 R /F50 121 0 R >> /ProcSet [ /PDF /Text ] >> endobj 323 0 obj << /Length 1083 /Filter /FlateDecode >> stream xڵWvH+ )j&>c(cdц6XF1OUW#RwuTn=Z13هc'yĮO?Zˋ3|IvG _s 1酡j^>S)x>[/*2]60;v2: cS.VXfk,*:adsAe/QE3|ݺ&dAT'|"G'"F?6} v evńâwvqV-O0# ah~+(|'( HLACizEZ_ ?ZFh9M3.?Cm Gr |Yw{LJ@ qew ᆷk~*r` >kr^*p$gZ% C4Izt-4JzEQ60':P_8ԉO@:o3QN _'I6LsLߍq'̒Oҳ̒^f=-N혲 Ϙ) 38A&|IO aġ@y]` :/] ^7*A%t`Ֆh0vMHo6^]*Y/9 ;rb]zi!Jũ&_%5LR{Z/ySUzSZFȫMUUwY. i/BeGK BcS\:;7|/T Bo1N zvH1K_E WmJl<ίbwz ܺn纇Ybf[E $HSӃfxLGO/-d=n KsE9X~ubE8j|wM4n?öuPnG 6e P,P7Yŏ4gPܕn\m+-ȃNpс*>WD/!3u" fknOQ{q"^_F{v-hy9{JI endstream endobj 322 0 obj << /Type /Page /Contents 323 0 R /Resources 321 0 R /MediaBox [0 0 612 792] /Parent 149 0 R /Annots [ 304 0 R 305 0 R 306 0 R 307 0 R 308 0 R 309 0 R 310 0 R 311 0 R 312 0 R 313 0 R 314 0 R 315 0 R 316 0 R 317 0 R 318 0 R ] >> endobj 304 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[0 1 0] /Rect [194.485 446.027 231.485 454.828] /A << /S /GoTo /D (cite.engel2004kernel) >> >> endobj 305 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[0 1 0] /Rect [430.794 426.102 470.7 434.902] /A << /S /GoTo /D (cite.vanvaerenbergh2006sliding) >> >> endobj 306 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[0 1 0] /Rect [449.058 406.177 484.536 414.977] /A << /S /GoTo /D (cite.kivinen2004online) >> >> endobj 307 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[0 1 0] /Rect [319.406 386.252 351.148 395.052] /A << /S /GoTo /D (cite.liu2008kernel) >> >> endobj 308 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[0 1 0] /Rect [418.617 366.326 464.058 375.127] /A << /S /GoTo /D (cite.vanvaerenbergh2010fixed) >> >> endobj 309 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[0 1 0] /Rect [385.74 346.401 432.219 355.201] /A << /S /GoTo /D (cite.vanvaerenbergh2012kernel) >> >> endobj 310 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[0 1 0] /Rect [378.96 326.476 417.068 335.276] /A << /S /GoTo /D (cite.chen2012quantized) >> >> endobj 311 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[0 1 0] /Rect [188.148 294.595 222.242 303.396] /A << /S /GoTo /D (cite.singh2012random) >> >> endobj 312 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[0 1 0] /Rect [401.459 274.67 443.441 283.47] /A << /S /GoTo /D (cite.liu2009extended) >> >> endobj 313 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[0 1 0] /Rect [438.97 254.745 485.449 263.545] /A << /S /GoTo /D (cite.vanvaerenbergh2012estimation) >> >> endobj 314 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[0 1 0] /Rect [453.403 234.819 487.22 243.62] /A << /S /GoTo /D (cite.richard2009online) >> >> endobj 315 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[0 1 0] /Rect [225.259 202.939 259.076 211.739] /A << /S /GoTo /D (cite.richard2009online) >> >> endobj 316 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[0 1 0] /Rect [457.913 181.077 485.366 191.814] /A << /S /GoTo /D (cite.sayed2003fundamentals) >> >> endobj 317 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[0 1 0] /Rect [322.616 151.133 352.006 160.044] /A << /S /GoTo /D (cite.yukawa2012multikernel) >> >> endobj 318 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[0 1 0] /Rect [188.148 119.253 214.77 128.053] /A << /S /GoTo /D (cite.takizawa2013efficient) >> >> endobj 324 0 obj << /D [322 0 R /XYZ 132.768 705.06 null] >> endobj 61 0 obj << /D [322 0 R /XYZ 133.768 667.198 null] >> endobj 321 0 obj << /ColorSpace 3 0 R /Pattern 2 0 R /ExtGState 1 0 R /Font << /F37 107 0 R /F8 86 0 R /F14 123 0 R >> /ProcSet [ /PDF /Text ] >> endobj 346 0 obj << /Length 732 /Filter /FlateDecode >> stream xڵVMS0+|,嘤fRaJim>b )NYvWY'޳4Ł5pYo7ƾК}ع}GX27l՚"جX)ց]2ІJ9;ZFoKGhX9iּۨAfؿua{B)tï>{IijpFT1[-[sK) NBGղܭx(@P O#r/u/5/eFAd#1؈eIf ŝgөBJvjO Q뚮8?u欽\U_ *뫨oL0sN*䃮Fo05ޱUKn8@T݁/aC eUP|jQ5m) Z 8@.ᗜO>Cа(s>#RsVbȩ0,FXA% D ,d$a8~]?1h\QUW  贘?Nςk%4nհ.kR HFS;H2"NG2_lJ# %oZT+UBB TGg9uw|Ͳf-rCnAٲZCP=ண#J=I/k%W endstream endobj 345 0 obj << /Type /Page /Contents 346 0 R /Resources 344 0 R /MediaBox [0 0 612 792] /Parent 348 0 R /Annots [ 319 0 R 320 0 R 338 0 R 339 0 R 340 0 R 341 0 R 342 0 R 343 0 R ] >> endobj 319 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[0 1 0] /Rect [437.989 656.239 466.134 665.039] /A << /S /GoTo /D (cite.rzepka2012fixed) >> >> endobj 320 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[0 1 0] /Rect [359.699 612.403 391.441 621.204] /A << /S /GoTo /D (cite.liu2008kernel) >> >> endobj 338 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[0 1 0] /Rect [410.703 592.478 442.86 601.279] /A << /S /GoTo /D (cite.slavakis2008online) >> >> endobj 339 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[0 1 0] /Rect [298.042 548.643 338.958 558.568] /A << /S /GoTo /D (cite.gao2013kernel) >> >> endobj 340 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[0 1 0] /Rect [417.427 528.717 448.478 537.518] /A << /S /GoTo /D (cite.pokharel2013mixture) >> >> endobj 341 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[0 1 0] /Rect [160.446 496.837 198.553 505.637] /A << /S /GoTo /D (cite.zhao2013fixed) >> >> endobj 342 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[0 1 0] /Rect [408.406 476.912 455.645 485.712] /A << /S /GoTo /D (cite.fernandez2015probabilistic) >> >> endobj 343 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[0 1 0] /Rect [283.291 456.986 310.744 465.897] /A << /S /GoTo /D (cite.kalman1960new) >> >> endobj 347 0 obj << /D [345 0 R /XYZ 132.768 705.06 null] >> endobj 344 0 obj << /ColorSpace 3 0 R /Pattern 2 0 R /ExtGState 1 0 R /Font << /F14 123 0 R /F8 86 0 R /F7 148 0 R >> /ProcSet [ /PDF /Text ] >> endobj 365 0 obj << /Length 1528 /Filter /FlateDecode >> stream xڵWMs6WVjƂ9vtI#'N$"FPARr?]d$ړI`b߾}ڣޫ3ꞗwgx%qDݽ8'IzQ_bJ=$ɗ$ipaFR e~ݵh<7dqB7IidW~,yGkÄ1Cu'=1Fdp_ /q$oTjm &)4H FSqLC8ޔ$'rXhSUu5PID!˷_'J9 9ŞU&<[K݌;$'wI20fq]W-ѝPa[  [D$Y.K7`6񅵩Nھl `;q,p["O3nkZ\AC[i,?"чtŽ$>ZqޝRnźO]DsN2ol݊tZf줟1' 3$iO?K1'!$~vx{/~cHn7bR0ԟu% sK4hJ3꿄D($XvWu9Eϐ#YXDؘhяRP"iE W@ Ĺo"BR  G ~E[f#ܪug?7+QTeQ,Ԉ\apo,nc27a;qTl$fa7FؼXܯzu*- 3|1QF˺s뇙>`m3ܹ OoF` 6dYO_ӍJ1;BmiU- XϜS_˭Lm-iж!3k~+g`>?ӈ>J+V X\KSQQn&+[KJ;9_O.;(t"fA[_VM䂑 i*b?Ɨ2|je bns;:o5&S(6 ܮ,0¨UBѭ6tV4a5+>Aro.?яe NUˢ3􀃹Tھ\+q"fQ nrP=Z4C׵EE+&KLFGeX9NisP0R+s4vڡu|h?ި1!a7-wga~Y$LoY}BL-ޘ`д›i%?g {AA&`G8 ?U~'h=\sc(t-fVN=?#1 "Թ<@ endstream endobj 364 0 obj << /Type /Page /Contents 365 0 R /Resources 363 0 R /MediaBox [0 0 612 792] /Parent 348 0 R /Annots [ 356 0 R 357 0 R 358 0 R 359 0 R 360 0 R 362 0 R 361 0 R ] >> endobj 356 0 obj << /Type /Annot /Border[0 0 1]/H/I/C[0 1 1] /Rect [231.9 383.657 449.086 394.394] /Subtype/Link/A<> >> endobj 357 0 obj << /Type /Annot /Border[0 0 1]/H/I/C[0 1 1] /Rect [212.39 369.154 399.688 379.629] /Subtype/Link/A<> >> endobj 358 0 obj << /Type /Annot /Border[0 0 1]/H/I/C[0 1 1] /Rect [188.48 280.689 411.643 291.082] /Subtype/Link/A<> >> endobj 359 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[0 1 0] /Rect [235.277 251.437 267.711 260.237] /A << /S /GoTo /D (cite.vanvaerenbergh2013comparative) >> >> endobj 360 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [261.477 232.576 267.939 242.678] /A << /S /GoTo /D (Hfootnote.1) >> >> endobj 362 0 obj << /Type /Annot /Border[0 0 1]/H/I/C[0 1 1] /Rect [161.553 199.269 348.851 211.224] /Subtype/Link/A<> >> endobj 361 0 obj << /Type /Annot /Border[0 0 1]/H/I/C[0 1 1] /Rect [148.015 115.417 346.073 126.914] /Subtype/Link/A<> >> endobj 366 0 obj << /D [364 0 R /XYZ 132.768 705.06 null] >> endobj 65 0 obj << /D [364 0 R /XYZ 133.768 667.198 null] >> endobj 368 0 obj << /D [364 0 R /XYZ 149.011 128.717 null] >> endobj 363 0 obj << /ColorSpace 3 0 R /Pattern 2 0 R /ExtGState 1 0 R /Font << /F37 107 0 R /F14 123 0 R /F35 85 0 R /F8 86 0 R /F50 121 0 R /F7 148 0 R /F18 367 0 R >> /ProcSet [ /PDF /Text ] >> endobj 372 0 obj << /Length 1894 /Filter /FlateDecode >> stream xڥXKs8 WHV*7'M_8ζ݃b+&JrʱcX$@Rh,~ū3ecZ\^B&i` eʖۅUuY/!oѪhtV=H83y)]歛hvK'6pe(z- _Ⲅo%ՙ06Nݜ^/TePWFfr> gv4A'tRs ҄pamO6욠M 'ɆT %Hyq5>4?~鋎/|6/ZMgeQhǁ¤K"ݏA _sB1Ju@ PxSIb ZڢIVh$.}SB]fY8b̖ޚo_N<@²| ĉdˢv1l Bz=ȿ.'`6|]ncUs6ȿsBm%zzz*pYϋ̅?6y?If݄#uQ\au=VE^Өۂ۴Hrs ~;beh-s@.eSw>~˞ bKJ H܅StzŪ;4/Z &wn񻔀+/?'}\\giׇ82?Thظgd0Qrx^tVaEKy֚^Һχ=Cqu[_IaG bMI[ \X= ʯ0U匤Cԡ?2P ]q7 cLaC  lčj8#H­ދc#t3.NJ5'wLϑGd-snD. ЈcD5dTm6l|m6 nv¿rj >4Xe~O6P߰gt ,~W ?`z7l*+U^CSF$Cv;5d?o艗WsSw\AWsB3q1ٴǺ!ez `eUb7LYR+Ab$mHd ^X_FI M8+ojd 0ovRm*KzQy\]IpKYUe U_M6iF"Uj nQIꍒ5IO7}F"L -pRt d݄Ƹ_fMikoxPW=9ȵ .x\2iv!EXH~uEJRk5}X^{Z}tEx̵|/]V?n7T<ɂ;l$_ endstream endobj 371 0 obj << /Type /Page /Contents 372 0 R /Resources 370 0 R /MediaBox [0 0 612 792] /Parent 348 0 R >> endobj 373 0 obj << /D [371 0 R /XYZ 132.768 705.06 null] >> endobj 374 0 obj << /D [371 0 R /XYZ 133.768 667.198 null] >> endobj 140 0 obj << /D [371 0 R /XYZ 133.768 536.351 null] >> endobj 330 0 obj << /D [371 0 R /XYZ 133.768 480.56 null] >> endobj 325 0 obj << /D [371 0 R /XYZ 133.768 424.769 null] >> endobj 354 0 obj << /D [371 0 R /XYZ 133.768 380.934 null] >> endobj 351 0 obj << /D [371 0 R /XYZ 133.768 325.143 null] >> endobj 355 0 obj << /D [371 0 R /XYZ 133.768 257.397 null] >> endobj 142 0 obj << /D [371 0 R /XYZ 133.768 213.561 null] >> endobj 327 0 obj << /D [371 0 R /XYZ 133.768 169.726 null] >> endobj 370 0 obj << /ColorSpace 3 0 R /Pattern 2 0 R /ExtGState 1 0 R /Font << /F37 107 0 R /F8 86 0 R /F58 375 0 R /F7 148 0 R >> /ProcSet [ /PDF /Text ] >> endobj 378 0 obj << /Length 2436 /Filter /FlateDecode >> stream xڵX[s۸~z&"ҾݤM7fjO;v]ڦ lrO_N") '|?x\2$A,'Bx#HMg-Ox/| .cqfXOg2u֞v$dSK/}C-H6P,-nhEej_THfI*VS!M8/#ޠD.7XZ>"p2"H"iΒ`j#Ve[W'&QnӺaiekf p.c$&O\]\\*VM#Z&+ Ĭ,ny.RQE1~9,/dKٺΈ}La:J)/<񻊹@qȞU,BnٔwƊ$$F&X(;ܾ|\y _mK ?$ͪ(t02Ld^Ńd2ۛ^U+nzҮ PJeKRޤ̇Fs5 V&M'NM|UA+bt׍{( m':;l6_m=sqo&,-Anv %DX/@W$i#g)y,P-PGBF=- ԰5lTmR \6xgoS@$1ɾ7-l@}VDA8  ? 1 uBRuQS#'pH޺قDxy27QvX ghaGڧ 5@`DŒLzu>" q$*1]1šx6( ;o׭w[68UCHAGoRf($d$XT|nЇ ȟ6'7ea¥I w$rPC_yRk&x6-Xmq!HWi #ɣ=$%qA%o@hIG`jt$PD>ȢʑQ,6tnҊ1?Hڔ0r~tIՐ8Q R$&Jyh"TGlGx^ \~Ńkl)H1=T8O%s])4D!#cn E_ȄB0E湣L|yٿk|KBA\Ca+u1uks:sHr,K+2+<+v9h`G,Ő x"Yn-۪}.%ATk>$.,4f3"Փ#ki1D;/G˶4woX5/B+ #$ S J}PZvl5Sl,ԾcMR'ZI܁nw ?1,5 z`)(JXBPC( A@ eIǹxt*MzKExvQ> endobj 379 0 obj << /D [377 0 R /XYZ 132.768 705.06 null] >> endobj 332 0 obj << /D [377 0 R /XYZ 133.768 669.191 null] >> endobj 181 0 obj << /D [377 0 R /XYZ 133.768 625.355 null] >> endobj 352 0 obj << /D [377 0 R /XYZ 133.768 569.564 null] >> endobj 334 0 obj << /D [377 0 R /XYZ 133.768 513.773 null] >> endobj 349 0 obj << /D [377 0 R /XYZ 133.768 469.938 null] >> endobj 331 0 obj << /D [377 0 R /XYZ 133.768 414.147 null] >> endobj 335 0 obj << /D [377 0 R /XYZ 133.768 358.356 null] >> endobj 350 0 obj << /D [377 0 R /XYZ 133.768 326.476 null] >> endobj 337 0 obj << /D [377 0 R /XYZ 133.768 270.685 null] >> endobj 329 0 obj << /D [377 0 R /XYZ 133.768 202.939 null] >> endobj 369 0 obj << /D [377 0 R /XYZ 133.768 147.148 null] >> endobj 376 0 obj << /ColorSpace 3 0 R /Pattern 2 0 R /ExtGState 1 0 R /Font << /F8 86 0 R /F58 375 0 R >> /ProcSet [ /PDF /Text ] >> endobj 382 0 obj << /Length 1642 /Filter /FlateDecode >> stream xڍWv6}W:$oщTHcT$u%郍m0\ܹrׯof^_dj0{$ e$JPDr0[ >S,L>ILAC)mOLNeX5LN\^^qϼybqshilY GLܧKit8/n 숢Y׍S?mgZG!ܠD,d\@>"|T\|Ii} d/wCY" G?+7ܖ4gԘ'S 9pr $3~;cl>sL%cnr䠕2*{]e &uc<ֱ+z*ZfMC<7eE}[;}U`08y[Ֆ@9tyfW2iMtpS xU+]1ܙTsϻc D'8*A;6-(K)~X! SQ5ִcgmNNu=!lHtSS!t(> g{/q,2G <2 rs^dR`D/tXsb$7ӣDQȷza@Bj/$t]ٜ˞8z wLQ"46!M MOdW=Y0%m[] ;Dn.7[,J ;RhR˲\T8oϬ׹{ƕ(C L<|2 3+0HH(3b։rD)qlaF݌]oz@;792rTs4@Q8 "JXR!N'e>e4'tκwYf5N:]mGNUwc.M ^fսcGT(q0 ƿZ*|ae,bI5@}LN3?u0c(O'wK ~~_ʁYj+׈^n}~|w;@z?v`LSf _4t}r {P_`IIdȓh{BБl*_]+!KW|ٳ endstream endobj 381 0 obj << /Type /Page /Contents 382 0 R /Resources 380 0 R /MediaBox [0 0 612 792] /Parent 348 0 R >> endobj 383 0 obj << /D [381 0 R /XYZ 132.768 705.06 null] >> endobj 333 0 obj << /D [381 0 R /XYZ 133.768 637.31 null] >> endobj 328 0 obj << /D [381 0 R /XYZ 133.768 581.519 null] >> endobj 326 0 obj << /D [381 0 R /XYZ 133.768 525.729 null] >> endobj 218 0 obj << /D [381 0 R /XYZ 133.768 457.983 null] >> endobj 336 0 obj << /D [381 0 R /XYZ 133.768 402.192 null] >> endobj 353 0 obj << /D [381 0 R /XYZ 133.768 370.311 null] >> endobj 380 0 obj << /ColorSpace 3 0 R /Pattern 2 0 R /ExtGState 1 0 R /Font << /F58 375 0 R /F8 86 0 R /F7 148 0 R >> /ProcSet [ /PDF /Text ] >> endobj 1 0 obj <<>> endobj 2 0 obj <<>> endobj 3 0 obj << /pgfprgb [/Pattern /DeviceRGB] >> endobj 384 0 obj [766.7 306.7 408.9 408.9 511.1 766.7 306.7 357.8 306.7 511.1 511.1 511.1 511.1 511.1 511.1 511.1 511.1 511.1 511.1 511.1 306.7 306.7 306.7 766.7 511.1 511.1 766.7 743.3 703.9 715.6 755 678.3 652.8 773.6 743.3 385.6 525 768.9 627.2 896.7 743.3 766.7 678.3 766.7 729.4 562.2 715.6 743.3 743.3 998.9 743.3 743.3 613.3 306.7 514.4 306.7 511.1 306.7 306.7 511.1 460 460 511.1 460 306.7 460 511.1 306.7 306.7 460 255.6 817.8 562.2 511.1 511.1 460 421.7 408.9 332.2 536.7 460 664.4 463.9 485.6] endobj 385 0 obj [611.1] endobj 386 0 obj [531.3 531.3 531.3 531.3 531.3 531.3 531.3 531.3 531.3 531.3] endobj 387 0 obj [877 323.4 384.9 323.4 569.5 569.5 569.5 569.5] endobj 388 0 obj [706.2] endobj 389 0 obj [277.8 777.8 500 777.8 500 530.9 750 758.5 714.7 827.9 738.2 643.1 786.3 831.3 439.6 554.5 849.3 680.6 970.1 803.5 762.8 642 790.6 759.3 613.2 584.4 682.8 583.3 944.4 828.5 580.6 682.6 388.9 388.9 388.9 1000 1000 416.7 528.6 429.2 432.8 520.5 465.6 489.6 477 576.2 344.5 411.8 520.6 298.4 878 600.2 484.7 503.1 446.4 451.2 468.8 361.1 572.5 484.7 715.9 571.5 490.3] endobj 390 0 obj [277.8 777.8 500 777.8 500 777.8 777.8 777.8 777.8 777.8 777.8 777.8 1000 500 500 777.8 777.8 777.8 777.8 777.8 777.8 777.8 777.8 777.8 777.8 777.8 777.8 1000 1000 777.8 777.8 1000 1000 500 500 1000 1000 1000 777.8 1000 1000 611.1 611.1 1000 1000 1000 777.8 275 1000 666.7 666.7 888.9 888.9 0 0 555.6 555.6 666.7 500 722.2 722.2 777.8 777.8 611.1 798.5 656.8 526.5 771.4 527.8 718.7 594.9 844.5 544.5 677.8 762 689.7 1200.9 820.5 796.1] endobj 392 0 obj [600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 0 0 0 600 600 600 600 600 600 600 600 600 600 600 0 0 0 0 0 0 600 600 600 600 600 600] endobj 393 0 obj [625 625 937.5 937.5 312.5 343.7 562.5 562.5 562.5 562.5 562.5 849.5 500 574.1 812.5 875 562.5 1018.5 1143.5 875 312.5 342.6 581 937.5 562.5 937.5 875 312.5 437.5 437.5 562.5 875 312.5 375 312.5 562.5 562.5 562.5 562.5 562.5 562.5 562.5 562.5 562.5 562.5 562.5 312.5 312.5 342.6 875 531.2 531.2 875 849.5 799.8 812.5 862.3 738.4 707.2 884.3 879.6 419 581 880.8 675.9 1067.1 879.6 844.9 768.5 844.9 839.1 625 782.4 864.6 849.5 1162 849.5 849.5 687.5 312.5 581 312.5 562.5 312.5 312.5 546.9 625 500 625 513.3 343.7 562.5 625 312.5 343.7 593.7 312.5 937.5 625 562.5 625 593.7 459.5 443.8 437.5 625 593.7 812.5 593.7 593.7] endobj 394 0 obj [583.3 555.6 555.6 833.3 833.3 277.8 305.6 500 500 500 500 500 750 444.4 500 722.2 777.8 500 902.8 1013.9 777.8 277.8 277.8 500 833.3 500 833.3 777.8 277.8 388.9 388.9 500 777.8 277.8 333.3 277.8 500 500 500 500 500 500 500 500 500 500 500 277.8 277.8 277.8 777.8 472.2 472.2 777.8 750 708.3 722.2 763.9 680.6 652.8 784.7 750 361.1 513.9 777.8 625 916.7 750 777.8 680.6 777.8 736.1 555.6 722.2 750 750 1027.8 750 750 611.1 277.8 500 277.8 500 277.8 277.8 500 555.6 444.4 555.6 444.4 305.6 500 555.6 277.8 305.6 527.8 277.8 833.3 555.6 500 555.6 527.8 391.7 394.4 388.9 555.6 527.8 722.2 527.8 527.8 444.4 500 1000 500 500 500] endobj 395 0 obj [638.9 638.9 958.3 958.3 319.4 351.4 575 575 575 575 575 869.4 511.1 597.2 830.6 894.4 575 1041.7 1169.4 894.4 319.4 350 602.8 958.3 575 958.3 894.4 319.4 447.2 447.2 575 894.4 319.4 383.3 319.4 575 575 575 575 575 575 575 575 575 575 575 319.4 319.4 350 894.4 543.1 543.1 894.4 869.4 818.1 830.6 881.9 755.6 723.6 904.2 900 436.1 594.4 901.4 691.7 1091.7 900 863.9 786.1 863.9 862.5 638.9 800 884.7 869.4 1188.9 869.4 869.4 702.8 319.4 602.8 319.4 575 319.4 319.4 559 638.9 511.1 638.9 527.1 351.4 575 638.9 319.4 351.4 606.9 319.4 958.3 638.9 575 638.9 606.9 473.6 453.6 447.2 638.9 606.9 830.6 606.9 606.9 511.1 575] endobj 396 0 obj [272 326.4 272 489.6 489.6 489.6 489.6 489.6 489.6 489.6 489.6 489.6 489.6 489.6 272 272 272 761.6 462.4 462.4 761.6 734 693.4 707.2 747.8 666.2 639 768.3 734 353.2 503 761.2 611.8 897.2 734 761.6 666.2 761.6 720.6 544 707.2 734 734 1006 734 734 598.4 272 489.6 272 489.6 272 272 489.6 544 435.2 544 435.2 299.2 489.6 544 272 299.2 516.8 272 816 544 489.6 544 516.8 380.8 386.2 380.8 544 516.8 707.2 516.8 516.8] endobj 397 0 obj [693.3 654.3 667.6 706.6 628.2 602.1 726.3 693.3 327.6 471.5 719.4 576 850 693.3 719.8 628.2 719.8 680.5 510.9 667.6 693.3 693.3 954.5 693.3 693.3 563.1 249.6 458.6 249.6 458.6 249.6 249.6 458.6 510.9 406.4 510.9 406.4 275.8 458.6 510.9 249.6 275.8 484.7 249.6 772.1 510.9 458.6 510.9 484.7 354.1 359.4 354.1 510.9 484.7 667.6 484.7] endobj 398 0 obj << /Length1 2007 /Length2 15498 /Length3 0 /Length 16722 /Filter /FlateDecode >> stream xڍP[ P\S<@pw@qww/Z܊wSw;J9n֭+P39]YXb:,v&6JJ +W[Z g+{X;2 뻡=@`cca t2(2A.^V@cJ `f jr2 MuS+PX:13{xx0\-hV5 d+eԘ(V.R;zAw) x?.PvX_ _{Ede3heeo0\=]@{ .@w-Ё)QU=blbeWѼYLdW|V {16>AVfaȬioͻ daa OSKrdK#= 9:|_ 02u,Aw豼+ '-fRԔwU9x|9l,VV.?|e N 3@/4]obzvV^x\7-Ptxk 9ؚ_+}D-l[F+)+O_rͿb`dea?2y>\go}yHI{S@gg{'}@O1ᯎrqE q ^7 Yb0A,Y0+AE<*Yz/} sAE:SGW]Y̠?=],a?Lޑ՟(8;9OԬ#|O|e?p}2ῐ9G8Y#qCtyw;G~33sGs+A6usv~v; 2EX^t0l]NqRpr_;g# l mmvЦhHڮ$͍ ɋq[#lX{jǓQ~icц"xBF '_@6n9|'7B{~iφ/㡋\_fc4) Lrp`\0=Qon0HNbK|tb׫4\z(tq o0g|Sp|K7r#W>'.k:`ЍIѓale:&i ,6O!)7! xx!'RO7Ks[715 NW? s%pFZ:l|EI{vBORm%Goୀ:̷NѓGeD0;rDC4Qo++{R3gY61aM,;BMiG3R™a/«1N8NAjQ21sLFف{W(Fͤd {*UA?#"3S]u7|9SZ9jQNV'S$b:̭NFuyYCVn鉬m.-#&H3 ܙz]=>T%H~--3 cb!lO ($n@% #@Zh,~c#thٷ|Y}l'LP:5qqݓ/jЁ,Mq'&r v&k{% }`ay㕑܂~7_ UI"ߢej?U?=Uf<2?}CK:@՚}@qkǻöyqɪ{ԂW-l( 3:A%Y|U2^;@T=-DѦr.ل&͏ bcebUovh]ıF\ :Z{#(0.rZ,3GE`1MH[WƮ4͘I5yZK~# `\' p=YcI$>ى oWO-sgl /9-7Hh5^6Zc5a@uϱ45Hf.gshsЙ frTJ{}Bm0AT%l!ö}Џ h!d܃ N; nM /f(V}kQaE;|V￷zط +`Pk8Uoà8Y{^cөtM,d0ܐ4[H(Qgqc\'&ֵ@9oIy(IGQGb.@Ң+~ I7xSnJ|ܐ cFb鰱LJIiQRѱF3`9l[k\Xn78i/:&0DvFXmWE[s[%<͊;7\2 浻L+6y%b+zJN 䴜pqft{F"ˍ[3S=RQKhdbGwG# JAbD>Q)?y5a?/%59ԟܠ['SuS܆]RWq8[vȴ] D oqG<|e |B5lǁDH`].B5\OtOT F{ELJH|@k xȿ)1-rL1F \Vgͼݐ] px5z3$7Ë$ &w2X/W rDd{ 0ѠZO kaۗ(NdEx>|d`G>E;QZ񽎔IrU8T(&Oz/S g1rX^qQ;+RT AF L#U-4~nMq~]$11߮6~}MeԐ'Ȥ2Nu|׼U-+m9`?n(OcnjB_l#Ti/q!*A۪x3 Gp0a%x.$G3% !O 2lS|éJzqJ1_͌Hعlg/tCN S\GzH hZ{ Z4.* ";.~7GRGQ8'5ΩN?IjlĻ`zH]W.pWj/ b#6 QzR!,^Wܒj,$co3` * )GI7y9fq1WGuaV+߱W_0=;lnEYni&i/lhK!=f&%hk`-8 _֥(QS;]q<3?3LU}*ba2H۵&ݍ*>@}Z}{!`Nk9f[f+3H;iOg|`ܽCiTQKճmxW?ALfH,Boq--{B ֒3(cYw1P֤T‚zA]^=aѥq>EoE:R(M?0>r]~f&>X{L,UWxp@\1l $& ѭ> e ꌨkfv:_/k-~s5r-f)8y.,*]ěsPr&;&hR96QX+SZp>Ȓha^ p|"K&R.]~ ^b +gx17"n~URv x 'OBZTbpI|!_j]qeh##ߛ\dݶϖD˭4yNo !d]Ai+⸡/ \"|`X[̩5Per;6 ANjM;F봽xļEcU?a ]bt n.r1o?^@`ޡ|:H185VSJ" mp\ZgIyCUW(&/yTNa /x?s)"f ~r9ˆlEMtiHY+$q7F'pэ(]1GԄ(qzq¿h)j-WL9`q Q!-U\ A_ dA}]EO#奪~6~nXA VMO֜'I>xl٣𣶥~>g [A)%8AqiL`MɏNp]aå- 6 5 !&8ED .L o鴔]W- YB"kGu} ̍ $V7Nٱev MEoip4u̍+ &vΖժWrߵ4P_ WZ,2u٢}}DLTh6PW2.[n$&5ȉA]Ʉy׮T_і"F=E:^'bX_@CGq!jz6Se1oXgp!d_P ,um.>ཞ8ߝ,_iA+ŋ)jzɘk%dPjO)(8\:jL:e+/f;53>'z!ئqSڙ'!R5:lWXΆT#>e@ĞN>ӥ)!Pd2؃3rm~ Oa[ntq|DZ%.&XsŨ FmtWwKin AZn}  }1]‰!i~U]zusKo߉Oͨ $$C!-> zz-Q)tJK%iVLq$MF\7-LBLcmlhq8,c^daAc]IcH'xw C˿oͱ]BEL +S9OłX? W_-f˭8%T/4|;ʓmАļQ<  VM>ᅯ徥6V{."CZlPFyB3[=~# h @8mj>x_fcO}^ #"Ri}Е :Rl4O7JG0+Lr'4Jw][DUhR S^n+$5f+2~n&IoⳘVY-w+FRxrmw4&;x70N!AgFtE~NHzBiU!n%z1$kĩMv,R9iҶZAIc~:8Fۇ6PszqB1Ʃcr&)/X1Hߓc~R{*I$d$8kͽr\ghSOOj^7x9 0hh BT㑊u %e<Ǹż;SP/ 9iUWD3цo18 0R7qNXw_($.(kh(!ןa0}` xF;%mvkRcUeAl+e#^F4ٟ!(F֫jh~}YUj]*!LƄtSpqׇ54{"YyRcmqR 0vXH?a$}2[fs$@ 4j\<O6+76ԛٽlS検XȑޞbF HEi߳5%hj*(Y_. /Uv9s*c |OZob<;TprM%E{Bv\vTa-7ϼͬU|4IWzLd㔢>=%ѱ& eϸcꇭ-`.q(N'4wZ"&%xF<0rm>iQh2Muoml3 QxBr1|R;aSH\S-nPeD!$EǼfZ>F&NN_Xp!e Z YXS7x' bt@QpL[ᕬ(@F|iA ]]Vg~e`*GK>C|?ٛpU\ჼc| yqLKx!}yKb4 n3o:|xs_5^]:z}ER M* 躳"ku(A?hiqMp :I#Z(26I ͨ,~҆ ulђ蠗 @u0萘Gk Y\Mx3 f]!~uW2F, ~<hKg$)`־sbKr]Y<0"32h X/Y܁j 4!E=xR+ْ1KC7:~ oϼrDD륯.2Hl;D=o7z'i -oәt\rruB}]I,D#%{ <ߧЉ4Y".V7 1,k %=#g19wN9C϶lX) Q)A lZݾ!SLd*#oĭ3t=rIjNNSMXh,0HT+ʆ?R'7%?]SqלiF[ G!B`cnRR'+5xMs $o0_{a L`@2HTP]2 Jv4 Ss u n8 J8/0*e;)WV}eKټf}6 %mE7lQL=\-uۉA$b$`ax3B5 XسmݦʱlF&>p'MHI!7QXRej"Tat/- y70h):/ S]&]/yݦ׫#fGQԀ#̧{3YTs=UjrT@p斓[[zېB2Yh\2 v|[ V&dhϒ dfdc@iYtv?ןm5mb3w*=;% gJGtߡm2G3(Pwd1@"s= %'HfON_% {AƓ3xm}vx`,\J4CX:^ hQ f{EʏwTtYF^~v8ungj fTO;JZ3T`&9R$W) x ,(h`ubR~00J]YS7O*閱H7ҳ)jh18Fcvt,j栅 jQ,Vj<ۢ7i2T-(`E&gOD̺A3aʦQɫ>f[(MD߱tuOPvz%+m/ͪ *I#W/.)T\W/\"P隸!o WG|jEWǍPaheޔl (o|J8|Ocegԡ |2mE%'%P^!An3PO#MvDl6vNԷ&N~W"3OݷD.g+$J)pwV1!A H[U5t S؇&5oock:[JͪLK.{ ]AFI돲jr"׽H,.ǥ3pȆ,X]ŗ%6eG[m9LiZIѝ̱b7 bKb|+df:H|`AtH8$_U=a-f~qwo)]Nn=3Ƨ֑n H[ϱQ[bK))i4GfGkn|?wPxˬDWBʥ'X( - nKvtC?,J=R;媧KVqErx."m oǑu.QOU;HkjX/ȿ=mRLg= 2}z%I,HTb/9.3pY#=RzJ 2y[r%*Aba0"Y4W JF-Zc.#|'bɤH6QU2يS>^av/m-Zcr5B!)e +yn6iFA\x{[xqK at""O49>ufKo ֎!_3 d:`3ёW#;н?PaiJ岸kD$L}5%2)!Z)t7jG:愽"(¬f5 ;+no9w;c`-H5c@K$t`%:]~BݕFn1g|Ħ3lJ2k>)сlM)cYi.f#Dt$"+vf-CB/Sr.ܢd8uaM snAGKWXBVl~7FtRy.?'NBn\/|I1c+i{㈁(U9ϴm}澌Ii!-`wm4<8H}&Z`L;4[[QHΥHHKϙS*hUJNXHj֦$RrMbuKDoI!KRt[Gȑ)NU.J>6YW:&Rf5ǹ Tj!B@G$ ޸om4_Eɑ r夝)\<^h㐐ܒmjש'v?Ȝߺ*ndBw8IE=f F)cdjRHE7sF>wJ~^ eR??J䲕gؔ|!kujr:'BA20T u1,o!V_)?g-dD$K` c"z Q7`0RFet)j tB>.oVaKfo^cc<E(q2*Sɮe/eTƯo{´~/.gaaODMty6JMmL:m,ѧraypz Cӗ! n֗dtіoIYfl'a0l nǰcz*d պEtk&wHeuEPX80ja)e>^sa!.`Mk w.͜Qvm^N/iCaPÍY۶[ 辱EZAb7?ƆF'>] 6RC-c/\.tG`uEwq^6lXυ˒IP|z3ל%Wə {]VԾnl.x 7t !$D>O!77`֭5pd0Dѥ<M:?N Isd)=FԺA نI|e(/bVc$Kz:O#!~lYnSc< ggU/H7$G ğyȶVM!$mzt8 ^QTW]F*U׼Zz6)оf}}aMj`HPF8!}(DBQ8-@I5jk52ԍؚ4ϱU6e4s͑_.y|kT9`=6 Qa^Sךi F=c gxU߾3~ T-qLI,Z!R'\CBTAЮSVGRa4pd#?Tˈ۱$">'Fz|@`i/B.cIKCRs^ )oqDcW6k\}Cǫ߁,*j1[&1k5kJmlw 5Z 8aMf-9q8ik y~*:~N2/T;H_ep_bSq_.¡s#/ X 룛s]tcq/.p옊xxc,L \]`1]v5cIz;N+nԗu'"?u{D3m&td|gTNQgp 9_*,.pkVf'ίf1ͼS=;jҚٖmC, JBI{ )cU߄`Dh ڻJ6]P0lΊСAgc7eqШ`zt51Hv=XNHAݭ,.LC 2%Rp_+:`n_cۭXn{xTYDkUTIj%__OڞN ݬhr uiBtC8d\#!!|ӂ*"tj-tDS#:93]ْ%N"{hf 8plO,J<jvұw5ҚYX& (qxM$ %/E^7QYE!D\ʖr{R:xl`Ά}r["U!f6 'O-RN[4Ќ?BZ<<4m|VerG.վMҍMʍ!;/+(띙p' 6|lzDj,N(nJA(Qf䊎s$**piY';C 1"IO D!"6N)XJ+bb@Vgn71JF< *8HxJ0(EDA|xn:$-Gn^ u&.>n8P0F^bQ\f#-߫Ĥglۿ7x- U:9tj-1ÿ‰m -:dgtVa0Dk bRW3$*/7x䰒B2bȒ϶?ugGm]ˇB]īn kpSoW:L ׇ G?λ6~jv7Xhlg|9"2un.q}~ W $|"ӧU$'ֵAQg:եL ^oF!Oz5R~2;m,^(Fo\ݑA],ӨQ\ Jԑ UriGm0zg7\O׮m'ؼƖ&'ZgfՔ Ha,>.wkM5j>o~sW,˭3s<~aK4s"}8IANJFf,RBNXB0M)ޟ.h`l 6qBn`E&ܹd83MWEtP]$UBb[(+괨zKf9,*~KjF 9BZ;X}c1I:0@R2`|!ivSBGs>s^"τbRfjLkQ }tJalXZ+#ň]sR_5݊wvŵ^/)p\aպ>jU'߆4UR4pRxr"V{CvM0Mq^MMAoi{%l:;t*l+6aypyETYx,j5/}D.6I8︤bPI&x&F:10. ߯ (c,JBzv梓:zR3 *!{Tjz*hmՅ ]@;Fd+F7 b=B-nĽ޽쬹㙩W\%p ?n'YlkUN2ԧF*@Sq[/=&f@k'k#=#;@Q\ k Կ nW 3뿝 ml̬Mf@#5@/C}K}g}3K}K?_ fak#A++_  ?Fõq2662-P\6"?2#] MJf [- 8;<_023tM̬aD?I1~ ~}0#kK?1 տ)W) ` aa008l,#o:[8UGSg BX26 t-zzÏqߔuoE"Ne׷2t:9~l.X_SUVW?vm4s1sə9k\%Wk,ͬr6f]-z.Cc&V?S [e,}{{}7XQbdax0|)Z8~>yma:QV_!VtBN@'1D &O'_Eb)A*EAA| m,?_]33Bf1q? Bn!3fQ?9??B},t6}~|{( h^0|T. 9t?"8Q$w499მ?7z]$c77t }C},ߟ h4oc`^PɏB73CLAd@Qaǟ0ԍ#L~m㤩*9NE7Fajvq M> ҷ}W;O_ & ,;'vD^QךѠ= VIiH-߂Yl9,"HG|hJKWۻ̉w*XH<Mƨ9R%FNll ,|[)2D A: mHvUK#[#XGH;SA? I,3۱Pj _tsa^1f-Szb-ܒA&..bqxO}hpkbBo%3^I!baSL.H(ES$G6$:{d~-p?~`hT J5|#fT̩Ԍ?0@- f(1w>xTtܑQQxyK4RyXH!I5.3vtCu|;Y [z蔨7Tog"?qoyםiPUKK-In֖ `ʴօ>^V[w#gDY~kK]d" b(7cE­c=zNQwrۃѕ2NRy맑1?8Z胢i)V4gc`3=Յ8lA?ෆOݫk@.L!443DO1|fe/qÙ[ki}{0 eAZp3ИCW ZW]/~2jB}24s\+>;24#glxn^j1rAPe (4hTIVf U Ex5JG&ط4@!BrR CXqc9yL;uݢd9p 0Xs՘!ZXP \« AV *io@ZVFjNR[dI8;˩WGËg;v(\;s@лf4 ('mm+2Q~IО i9"Z^WBE ͎jk aZ6fޜ~rn9__xJVM7CA4Kp%Y5ԠGL+N)>ALS85]NŷP_Wb.4h|]ZC4mƼ r=iY 5 s ztӯdr7IJO¡C4mp7F?.Lʜph$g{9c{$0dYRJ vOk5jʗR'4ݥv 84euG2Y/wJEۀ t_H:5]@[YhoǛ8G3e6:-KBG1ad+ d -+čR.LTEsĪ:ș]CWGj^X.pmIJugEwMUO*N ݡIJ/Qz{q;O8ݠ5v7Y@u/,g=/ݴBAxxLK $ h0)j&5 !̖иn%UP%1/nBZҵ9/#[2h',9kOR`H*z[7I=S#%%8ҩ.Mttb@ߔ27KzP, 5lphHb ~xBvHF"Jȩl gjN{QbJ-J2"noefk+hjm>N %MnJ_7f9h);`;,knw݋ 0$ 59!Jq^#TCe¥)*e .ΥoEf rAFmGuVo0 JR`H?&b́- ^YFi_jtʍ}ZXϷBJ;a8|I+O krNAO?fѤ5_*"-]! <"\ۦ\;\ac:ɧT4zoz7s?H2iH,cq) X8.\#M;poHw@f/➒F|T x_tnS<*!=lsDdqkZ&՗Xu6پ5~LALAÿ2^7N9WRW /+\Qp;A"̝ZBdѢ*mx,\Q{1'BtHxJiEF31 sǕ-]&'dxn(a)3 .nCf &wUg3GrgN1Msŝf*||VNR%P H'R!D!w`_pRjltA`h h]Ӽ(OkVt6kvTr0Fj{"ax;M 0Mu ˹wo uu#0koG4s'jlҧ>}åvi0[WOFAΣhohnN0 EI{o/[8}pIC~Hr9NXe%N^) 8Ɨ*; ʜ¹Pe |P?rh@O,i#Bod|>n,3ZjKq3ƹ RMhX4I1C|5wB {׭sHٯXqOXfZu55jDZ{-„bN#R8Ȕ)d1hidĻquoď~k#;bhq+6@`伀>GKB|8vRr]^%||BӤv,М;En}(6VR ybԺ^uN$]M9ŭgmFîhL<'З$?SpD)տO 7Y&(#FZ Ȭ^ H̲sV &kROYƕJjlt8!l9x0;5[oVηirxc ^޷H4f¹W`ARFa|"/xUb4 =p罸c_sDX9}a,kpCzJ+u29.݊]TAۙD;ܔCP\nּ9&Lz{´) E8ԚX:~/Jޓ`Or? %0y9z֝uމ!)ڂXpyLdeH# !VM3ְ`~2ٖe/lC[j7X{ -rv_w{Evq"I&cKմ bʚz\ZX PABV;A&0=C9Waq w wrI|{$t9; A[֭;שhha2HNJǩ d_2:dN➷TEuDppFPe%ݬ.D >'<"?z 4]@>¥jxxE^' ?cZS5Ojbms(:[rt z>,sG-|9(2 ꄠȚFsςkHWEGU5l1zӭBmT+OIMKԧqR*A#T_CH˜:h8ؘ0|W۸=x/w1A__W:5J ؾ.6|cZVf=*r9Wa)XL~/W[F(`x95"+2a&@C޾oh$M^. N'$w<(͋F6aKNvSNBxn\wb$nle]^`{̛ʳ={\Za@ݭZ5!yO ^?o5*vb]~^G0u#ZBy۔N+ v1++qOJէe*3IoMAqLBXen6gUtAE+^f2Y%nmDKϴ&zw8' _^ΣnԄ@(-Ef6"58L)/`ok2:.Xk=ide9<#ܫT<ªqϼL c"3 !I{5(葅ccW?( HU<+'`w=?GBeBbzݵ{5r$s3o1)DI:"9X~~ý=+v f!$7aO z3_-"9xzQkQ9)ȑIt24(;CGY &a0Cj$m:" Φ"KE' y.4\e 7m;A _˒CVޙTB]ظ7)&T{D=ss\;+#إ50vk@!P+h5 AGuNd27}-PZ@Y 7%H ߤXzJVGrI$uoJ1)E[0\4κ +PS 5"g?։rJtt^QH,K it3ҒR-ӂsn993!sCPҹ ^KjٵW>L5Ps秏ޫl׉.&^hC1ayv#x+>hM-EQ5N7`)P>shFJ;$]SAh /*D2,J\$]q'Jڈ$cx[Hؼ4•yiMO(Id 0w*@5A41 ( +=T W'@Mx>`rᷙ_~ ڣMMiXW-qЉ8T#ArQh>WWTF>iU*2bDLs}+'dpj<)8%8V+P6O!V[SF]A3-uOZC~RH晶vwj< g[ *dlB/^sF3/1hW}O#9{B*O¤>R@ N@p6_r3϶ṕ2p*dfx=ص#[("dܓAC._O|,a >pMQrB@zg󅎴; DMRI{sZ Z uck\lYN :H,)gŽYrtVz@!Aǧ0d"-1]V!( fݠI!"=*M5y쉁sZyOR^@EJ#OR9TLUMqm7K"(RvsN:1ȃ&vI w}:[IY=w%!v.bk} ?SG4!k ; +B3NP<[ƓqeH;`?Aѯ \ D>Qf>F8: >OD g@ iZ^tt]d)B$QlL~EbKSG9Ub4_P2W hBCP7`I95qB&o0nh&_t5_*ՑYIWSGrWV_/UoLA"z}ᛉ QWH{s1hҬZ(!lStC<&CO֡T6|R di~RʙD|r3GԈPp7 O=5^'gz"@UBuEKC&hxe,Y =FV1Bpgd`yuEA5F+"V{IAe[#I_օOYI>pQw;rq/p͠O:A 7Fwt%*}'{s/vLGw` 'GMFpԗ'_H,LϽSUjZ sI ZB`;xn3ԛ@d;ev/RڧrtTu>ҫf@gb 0C,2Z4Ttlwy]e|aSu}"4|o,W$RSҳW#Qlj5Ӛo"Fكpɸ\4]"u旻s2XMZ1т;+ 5fë10d zm:*I4O1鄺3$NUn!'\7]T@c('_sD6m]$$FeQWgDھT4KBD(ȣ|Z|mҗW p<,lP vQΊRFR W>X , lY\ Y]՘I vr|M!OXnz~ƘKVg8ޡĭnRyq  lagQ F,4@3pq=\'O^EFc|z> p ec-Ercḛ7h7쇾lCLͭByf؀{-:_.TT\j4d$8!X-w콽晗?>_pQ/NS p>^)O&^\ddAl4q4{ySI:i &WBR9nP5ܮiIgmb$3O1\` tS,uߴP?7,P:2 GB&AC%zdr|*k=C|㗽oĎ|%#54Q6-+둷;W}%bzb]O"OZEp_ *8*rVjev(f=Btky/kg_ԓ` ~(Cv n˒|nN!*#=i!u%C٠tx"@f|} QopzB=.hV.DgK`N8iO? b((=b$L/ROLQInW4emg E䏼^i/ wu> '7}mVԠֱ~ 3yoNik;ŸZ;ttdUjE /k&Ѯ&39:&)0$  N*6،@?7wC^a؂+3aŨRsk 浠Pj20>Y;6sҕ{KPAsEFdd\R GtwȩŽE$Ft6>O}ߦ~^[2! mQпn:|HC6[E&,I`s3JF S`{,Me_7eT̥+jJiE =Fq Ep2 w9kN k0 ԁԪFK8]/8 5jux=Ƶ;prBd2KRpL !9g2QT!&}Pٺom_^Uwb2}YNlw>a %[{31=03oV \t h!>Ϯ`WԆMrώrrS)_| [N% -k8]DU^R8UѿFaZ W4b$ k Y3̺\aoE6C +ua44aM1&yi*@6t٥/W3F˻]CO y&'Q$z9jJQ3ư^.te!V+s7y34q2;4egF'ڨ-Z"?f- Fihda"Q;0w%u=`\adç'F]Ja*| y8Ӣǥ;|/^4qg^~Tlay'!ݭ)K' :|X/lLZfGޡ9^#Wl;юJoNנ/I`Cv~&ƹ[bçÈ$ ACӈm*7Ou*om #t*>n I[`(kjK;o!/>I}X ~By3-~#ScRl-O#vkExFjb*ALAւy7~a$=UUj)F*d)E48)_Hfڝ(.j˗|} J`09N=}D)?O5l*k 2N;eƫ{;Ӡi(UX:yJ֝# G xF {4#1ߝ^UE@-kÈc3S8n?Z.YQb")5N6F 3lâ_gѮ$pW+6NL[!t\ HNp'=X >kP)XC. Nq+$\Kįڙ`tp7-OpsZ1l|=q+S p!i՝\1,Is;r4b pVODObm R̥"rHr Ć^gg0#GkxV@?)*:yCpR|ɱdNDx-JـWwzTrmԀ{s3d'IiE6?٪ӾRP*>e&5y+5bKgLNF[]:jӬwgwaq =@'_]rui4c.dZf1SNn1J16?bu AYT endstream endobj 400 0 obj << /Type /FontDescriptor /FontName /XQYHMN+CMBX12 /Flags 4 /FontBBox [-53 -251 1139 750] /Ascent 694 /CapHeight 686 /Descent -194 /ItalicAngle 0 /StemV 109 /XHeight 444 /CharSet (/A/B/C/D/E/F/G/I/K/L/P/T/V/a/b/c/colon/d/e/f/fi/five/four/g/h/hyphen/i/k/l/m/n/o/one/p/period/r/s/six/t/three/two/u/v/w/x/y) /FontFile 399 0 R >> endobj 401 0 obj << /Length1 1518 /Length2 7684 /Length3 0 /Length 8704 /Filter /FlateDecode >> stream xڍ4o6ԦFѢvbwmjA$Vڛڣ֪MRԬQڴx9s7쬏 vU;O$ P@ !~HA@((]_JH}cSouMO@@ *) & A HI2 h!("v%8 urF߬#Ӟ !!G:@ ڃ:`3fE{0 `BоUSFH`7?$ q AO0rt`$pcA!Owp:PCg/ S߅$n/ Azh4/; Co^`( lwG`>|/~({$F񣠰l FO/uu{i8x"PO_17&؜ hH\LH\@|읁0E@p 6pC#HOH߿P{4u'O3O|sHt#?'9a ㈁MU*E"'$@ j;f/p5!\. .[D@7_z#O濫_z`9 nP_7Dߌf7@n@oAF| ?P*wS6ڍ yGA_17Y nAhOu3r?7 nw=}"0 %9$S #ID@ߦ?o EF@ȿ `_7 @jdDސC7\/Ğhnn/R~Vͷ1*yf&7ZbMSYM6͙z:+jҩq4{^+|Ǣb8s}rk* Y Vi"d9S,yS!;ú)Tb]9Ԍ&{ ЭvF"G?= ]}v_K(ٻ|e`hpmlWw4 0~Bѝ_^(:\J?=p^?صbLsMuÜ3Q 3ś1"YpdSNQ&=YPLZS=+<24,8yue*{9F`=MXW p -g.pO Zg:ɖ-HQ zA3 a;:9=jb)0+ J /^)aVOk|VUv7+ qpIKΘxZUz+*t ^^Sc yU]b.wt ILt:c^m vS\5ʹrS uQSJPC=7tRɜ~L  N~׮rԚ5a+I{L0/9GBN 92wX,:lX6+2,_ 7gJޔݽ>؆?VE%q;z龮T=V8N)7n"(G I1+*0A~d8:&_eiCDt$/#L 7ݖw%AswزGeAuA"$isήM4jqD9m:ϢAA/tw#k}Ȭ^jߪkͽUmX@1(-50qkgLVduX)`"%xfdR/CK :78\4rmy&[!Mtx<,  jˆc ʌYG E G:lJjI.ϫS<^T7qvZ=fc!RbD$62鋜?1ʎRў{]Qz 6[dlCH.q`FWzPd?M:/v4jdӞ\XI;2c\1}0 h~qTLq/]16<E3DP|byVhE0MSd5.[ bml`LEc,GCĨ( eL|-pr9wźK> ׭y鮥G?_k1W./ٔ(<ܜz6EIf!tRGMgp);sۡCWzV -xrEb$^ Zٝ R-o] ˧S.@blOع/D,?6)n'Il^SK>Zٻ _l>!\Ve'n?W㰇#FqvTmOM!?4mWې4i{xϴ "]BG X9tf)ɩNƧ^>0UXcDT0lW!_@î1oc;#Qolq5ߧ\ƶ8Oql ]fM]ë́[k2 {dΚoO"T7S்>/v4wz#!rOCt *{;,{xDfS%I-_zY_*ɤ.[ ٚ4X?%Usq`0,4gGΪ(e H̋AS٤ځykQadz*/3`b-lnK5q. Z ݭx0uv{Wgl˱V\f°dJjAJ]:MΞr Q+Z #8d'p᷌= -}?^;E4bnɴ n6{à]ymMʄĥH4\n3[8)}/Jj^;ڝKX#ogDѿ{D9%fc$;W`;1K˖,t|1$;f %dTJm 0p 4fZS9b9.FZ Y}fBX͍ +=Bqm U)}DnTZn's8"e瞗-;QC6zKkQ胗qyiܼqj{JrR͸ |qz}E+)ρ߶RkEJzܦtZlR6+e2## 󲌥f_n}:9=(?RֲlZ<ؔSy}4Ϊdy Mb#u xT^r\AV%ٗ%^Oam .V3#- UXg@&2ظIhһCТ-=y'3D"8"آ]˯4eƂui Af|f"&AkGe EMt1s-{| WMCcԕΓ~,쓹N!ɨГ؈P-$^&{ӮBQ=H5qp/ʸ㱒-|S:Ll$ł(4lWȪz`?_3q_hPAcWWϓ!@.(7eJgkNo]!Ge:9TT~ .3r+ڝ*x*&:}`J!,V]g"#tl譌fz0%rY W-q?_609 bSUj<^-tȴps| sb-rM沷E.v28^eּ_%mU9 0zZT. S9 \$tteC%ʓC=T!KΞhˁ mHB< FJ˥Jt]D-T6q0O gf-tNZjT6@3#Liai; _3FNR ‰AA(6 ~ϷOgBl<h*"BVo]B\b"Q㶕{oSAt5=ⴖr_vǹ s,iMջKiMAؠ#.jHttwh4B^SFO lWsRn#&yV0O˅=z%lcX%+b)ٹk:>)m3.(/$#g@Vb>{mG܌DPv V}&i<~yɧ[WJttd `C㖁t<ؼzʥYNW;[`(/ʺ-jLAYܙE0GYGd/ء34Tc$'vz<;Cd4\2q{` {JlىH/|]?]y>갿=+qKo%lDMb)8hf_ SЗo^ۛ90K<-@l3[Rd=ҕ -xw*pB>cT\J:U\?Fl=|y_4N1ouJ6gjLX"GTtKa4.pc.r#qbn>AHw-\]UeNɯ_QRc_mOK[BV L)ݦT=҈Bع:ɲ!pw:="I)`kx=p*/g;@X&yG[pPTqVKT7j5ٟkB76,>nxYϯIzO z]֦_4+11rJ_tڪFNoO'xl:̝H/Ox[J] (fZDMqsB'=d?@L̛R\ei[9&y]Ş4S>Ш2; + VS-\0=?pN֗/N,b&$2:JOZ 9$/Pj&!XB崬=wd,*͵|:MT$ʠ9j͝w?g*<~3]Z|^!WD* Y\] n^y X5""NT;O=mPҢ0~ 1-7B$4s-9$ŠwˇڴefÝGفea3 Ҹޱy(2}ڣnuMD:lYAe֜}^A!ךGƏW)வxxp 1~kܠѽx`EHPc[kh #xIQ.$S=;ˏNEn$baJ;aFo,~CWX2)LCMZ:n1kBH_\ i3gp=ֽ^@WudkޱMx7Sm2!)T!u@|=E4IeNs +2*f:" kS=%@5%߾K1U&{ڠACP__hŲ3zp@@ dss˗DbOY8:7p<:'CbP9vbT0sb  B(Υ@:%ߣ{ױYɋjm,iHâV]̺Wnx8>vnV_l](\riWЧ:]$_ua9]Qsz.k۵Ƚ܁vUIk[|#)QGƁϟ٠D*|r nv]&3@VߨcO9Vɹ04!ogtw'~UJd'z8+|I!aa]z5J(a9M%@7-4-Uac&bJ(<+d94_G'GƤ_'ȅKⱅJ 5}(\P|(ᶬA1X1e5G7=B8qEP-rvRI,i6[Ft;Ƽ4vj/4Ⱥ FF>IF41e9]b- ULR*krIG$&4 Ӡcꞓ'XQK T{^?hXi9RpLcp?z{ endstream endobj 176 0 obj /ZWEPFD+CMMI10 endobj 160 0 obj << /Type /FontDescriptor /FontName /ZWEPFD+CMMI10 /Flags 4 /FontBBox [-32 -250 1048 750] /Ascent 750 /CapHeight 750 /Descent -250 /ItalicAngle -14 /StemV 72 /CharSet (/L/M/comma/e/f/kappa/n/r/y) /FontFile 401 0 R >> endobj 402 0 obj << /Length1 1385 /Length2 6067 /Length3 0 /Length 7022 /Filter /FlateDecode >> stream xڍtT.)H 1t ݩ CCH ݍ4" " )19߹w{׬ͷ~~~|@&]^y5L@@E--5Q$ p쯛h ss# >% B"N ," H7 n#0w" ngBWw@ 9OBH(G ){Eˋ·txQ};f50@ 3`hw7@ڢ n0  >` q!5x_(Av[ ɇF _@; h!y==ܡnp;;׈ʠoYatv!PSkYG װ#ll aozԔB.`(0HLTPLs~ \.[- GPnElPfG: c<~M/$?W76U3c Ho/+ `(%Ut!]E?ݢ_{%_qpYLf- APM)7"7;;!p'4i=Phh!2@7GZ0GPvh2@Bpw7F/90];׷W /#-65ah9eiKw"ćz% jl?B3ln FGP77>_oq`0(4*PvZ-OŻ6&zj*;oS`9Q2 VpVp=}yoG蘗Qu1T =4I3FI}"> ]]em $=^l*m]ř+5#jg_Gh3kHu$xU]X$ex"UW«2^%.JgLA$gbqZFN`xՐz̺GdSLO! "$.Vz#~-1 Qcjܓ{u;@ D_[F 0Z68G%o0 2 U0cl,2ټ|T$.SYATkC*.HK@\-|[1-⺸YMwmjxl.ZP6ꣻ굜zb1kᄼ'V/qǛ6~+i,r-|2؝oxIC${Ž5lW4*VEM(v%ykJZ(IWJ5eϗ`KM! IbO\(?CO 0~D9CߜPyb4w ?φ,o )/5O{ְ % ͽMS۶];#?8oXėxՀh<ۉ^#-'%eӒOMa|=V HQg;7R_E (H_Tx.ffC77‹ev:?|>1M~P聞ۥNJJj~/p3F,t@z] }sIq=L-CK؞~J{d?^꾫껳EA;(pCȓ4.z}2Ͽ3ʹ͝џ4/Q\4YRjI~b{iI&a9R?|S^K鐚3׶sY2ՐkÛU%Q‹q~F\zCTQ@+̹։Le?+% ⺖ ƶGؚEf]L$|_`ܠj,.NamoYL.{(÷f*sh3-x??gE>[mxXg֨8 >#k\?-`ϏaOCK I-__ >NՄ5zou~ȾɹkH.}gٙLAWm!IqQUP@k>=u7)"+%+ޖ8K).}71lE>:6|P12U >B!M"Ie _^lŦn"~GۭMs8 U^8`.܍V4}\t~1:#pD^($!CW<9rZU{TjH3Tՙ?}k&[C7[ oabXʛcI7.D *cŗuZj%P2vF5B4" 񞏴D ?5V $^>e;tSEH~V8F=Gvqq^ Zr@Ѣ婦5/eᨛ;SEeO:ZJcaIE05,6tUPd#r;enm5ΗkкtIm1|h؋ѻ ::YﾛXp {1s'_ònlT`ѢiИaן>3Ͻ{,E!ZٯG@ tb=錢IS֕*|02%$L)0͈ E*Oy.z)1dQͰ`m帊{ݫS߹˜%JC@FFT|- Q PLaD݋^UȒٻxoЦθFdh6lزi誺d>v*IzיQ{ΠOWgJ"Dy%5ήS]&'rӥ#t3VeJ7R7׎+4>, ,Ee-,Iqn&aV˼K|{\Jz=:-ܥƍN0v} rNdìDgx'L QW*LC;0 do>`|ͫ{悸w+ƟbO; L;9s QpE: H0"1nbrP@M"8.adCBzV*FǍ^q #qR=ڏd8-ǐ޸3[ Ɗ/OklWTJ_{^8:٫y<[e.vIL^,;F],Smqbwktcu,ͅ-*ifίB-l=#+unߕezL?;NzWԭ(c'}*M(TEґ>U |a㭅wR]Itb9fϸ]'vM!<+>A*3ßY]}G]b* 輶t {]Vա5̓)UpM)}ySJ J{5gw(8.g/ t2P~{QnՇW63BE#i:{9!!_yp˯JX4OkIQVt[v]I?y:rOT@<f Տ!} #FU!ec9AK3k.QDI5%僩/buնn) JN~ H{A3fD:0@y壤O~RUWE3h]|֛ .5Na1iN 5+چs|4Y7cKO6l=i}Za] ǰPWsr mH?L.Nm"Q^?[h|~@=0x|Uwt ucMGC9_+|h|Շ?{;#s5hjqbU_|¯X1}o3 (  ܟ"sP1s ys9.--J3JX(t]bZ[<:='j$ppƻN>mSjO֥*쩱|7t޼qR_#Dp5M~id\\mdGT#VsWOF52A^[ixGea"귘 NJA3Rt4җbb|TWoZ-fIޟPYeҬTMx2r6j|AۏųEq|#hnX#4sqړ܍ңe(-rW.֗UVj̆Μ`Ӟ^K.}a~?ƶHl`Ty[#1=LSc+Γ-ۼRȈэb_#"V %xs;1Œb~+5⓴B+jhJ.37:MJrݴ$jJ?{?0HWAv؍$/p}6.3GFmR+C=W @?>WV4$/9f#%*:kfv*Df\,wDח4 IKSHwXor4hiؤY$_*s `wi,ܣnu긞>g +bT Py3t#!!צ@ׇ=@/抶d=[vB%>ήgG%G2[phey<|k HP(d?C\Nůonsy0Sk؟^E}.Гq='-IkN_'`jDZ.,$Pf2K-2ֆd Wҕʁc>+v2eG*&)7Q{WaܙIϙiכ̃B@?-?CR]=-fK),>Um#ވEv&7S˽z3^G&[^{zJ3zȉ=x|p(Bm3椰!<%Ճ֍)IaO*ДZ[k[ĝLIo\'9w w4wBl:{-Ʒ]YhHZiŦ&cP6r=RCxsԘ k z 3X< "w?N1TYo<腌Ah6,gGwm$4 ڡ$omuHڹu]u9BYbG@ÈliQHiȳvZ9:أHp iZoŎzdmz!CbkZ. .wlqsP7 V; ħ?ّ.GLSa=$(ŧ4LdN =y9"g|VxΥe2 }԰4Ϊ~2˖`lb˭ J`hbe8MTAwyW)!ژ."ZBWGOEǟ^M?p+‰ƨlm#FW }~˹opgQɱ38 @<6J'ZyOƍAG4NZKz>w*|LԍR1w4+~'^]OЕvJxٺQȰ˩P C|Jº"*+⹭R2&{1J )S/iĪʤ\q+/ِ;^~{`z+2F+l@Eca` hb:0&!Iyqzps"/%u4ΏGPfewRh[{+<1t :g}7U __twN endstream endobj 172 0 obj /FRUNXE+CMMI7 endobj 154 0 obj << /Type /FontDescriptor /FontName /FRUNXE+CMMI7 /Flags 4 /FontBBox [-1 -250 1171 750] /Ascent 750 /CapHeight 750 /Descent -250 /ItalicAngle -14 /StemV 81 /CharSet (/n) /FontFile 402 0 R >> endobj 403 0 obj << /Length1 1347 /Length2 5981 /Length3 0 /Length 6903 /Filter /FlateDecode >> stream xڍwuTT{?Hs萘AI f`JJJBR" H("ݍJyuwݟ1׀c]X ^"V 4!` @V(</HAa1J!xN' /H9% H8%@ꇂu,zPnxB0L(*ʋv4<8 x7'!# ,0nxGsUQx7!'$@O%AF/􀥁` 6 u;Dpy v`XO/(&q(405( E` P?( u!~t5(ſP^x WYzz"0xЯQ8pKD0pF^70(o__ o+Ȃ7IJamR zHB#A|~E(pA0G'?epL` zur$ Š =eI k3W_VMMl$ĥdD "! dEU`X@z w~@%K`/7`?S鿢dt}B!OXc,a!0 AX4mx(a!40RCd$2Q> I?P䗌 ?``XE84D"A #~`Pz@&F`$#4#=\B޿ @ `OXص(ʨ/5dh:ltX#|~voLҽ3 RE`{JUHyA>FRKv1pW:.}a+b޻٫zʂuGen۔B5IQ((H_$jn*|1<րgO!ꏌ)ٹ1bl~꣕96[Bqp~cTOYjJqg X62}+0v@f #n-DF3pf/H>4Х J^qذ~lW .e-ӵlD#9vJg.xKrf3DCVeQ eF'{F:/w>q09x`g?pi) M]Q3"ac/ D{+e*>N;U}7)ڠ%u&gE3U=d GJAW<ٌȧG =PP~Kx g S\pyȌuj&kYR6Owa_$#]НtM +ܜ<ُi Tg܉VءƸʵ< msUl͚b%x۫H7$8 S. m6bnN&yY7 )p`Xz=MLrBZvF[F֦&4q;Y~4|2ո9ƱDT=x;79R;+@]dW1|`jyu1' ǡ.8e%~dv7Y?-u!Zh$HZ;)Z"+9GG3'SU3+|홫W$juSm8vd-+EB;|qM#E ;N^-S9G`.2e9I6_{@Wv0~HګDFh9iLr4,Ld޻.sD XUk~EVr-pppPr硶 33"RFhkT"++iQjt':xLD@AN\,1erd+I#U$3УcMqlʼP?B-D:s01%=6m6<,?>r&P >e?nLN3%Կ ɦ^CUMW`Ltnߖqif(S[*KK{֚& %K+ӜسrrrP}n{1Qv|!eEQf(/HR>Mgߑ߽Mdmd(0 ҩpٛ6;xD^mcQcagf1?)oRR貂&o` >Hɫ٪)!vмY]o1Q?&˺ &UA]ɭ2WO:}SYwY[KT.?a ɻVmE]Q3XޘMEkfkwQSKœV!`g9NOɪ.qQ=2tN4V 7KF.4&)G CLr80Ͷ0+F5Lf]$W?.khwdv~Cq o2B"2t^.C6%͙pVo(H[mzx/6@RJ<e{b6Yh??Tr7>Y& fh.f#? >*>gQBL71VwWBk<,{|>jCxmf]7J+Z?]qzy}OC Ok|]QVyY|AWm >ʴ22)֭> ^ꚵ(wڈ3&-rd: T3A}\ O.R8MnP}WUWSqA.̜ÚI1J$?/k^=guJ:DZj?("$IZ L ZzRZpH~&N.$ZG{|q/O>p-bL%Kx[;E46QW#cVBnqL3eM| 82Nc7mJ^ff=ƱV!X-t1;Z{"!2Al6 cz=]%0Hk`έud-M%LZ&VA+`sCgy+k3OQ/_t{8'17V&o7yN_Ҏʠtax7Oյ%#ٕ\c& mIs"_<_u&߻M;~FMs~)IS_Z!z[vdrj.Nƒ=_Grnc* OnF(o.\py&HygyB⑴x=tKg4j0 kMvҵ_ӹۈ^iRajb[g}BM:N3cd_!RhLl_Qr2S_MQ #pѺoGf͊krV|RYya&ɭ'Cq:-Ki)鹚$0^| h?,a1x%U]֗WPgIR\+<*V)l8;6'^4+}DYE"ˍ1VUjgGSEHa>o J]]:s$ϩVMY:a]k1[ϓLIdMh~ J.Q9xu$"`ޡ;~MNG*#t̼![)T?qT|ƞآYyohTOj#~ vX4U)}y^ ̊r#=Vȥ}dn&|^=qcͤt'AL ԡ~lF_xt$B?䞰Cˌ[-q𝲗su ޱEKo(q:St4L;鉝'd7rk˖;Rfa~S+&5& Ja[Ko+x-Sh2Nv򑆍> g>c,sd-uACMIk&#QR oh-*#-a{SP ꡪHkLQx+@L݋qkG2rϢ}Fb e1I-rIrfh=rwI0FJ8|(ռSF+5ARx'!Z[Ϟ+[mݕ: V@ HB Fn"r,^U4!=lv; $oApgFSI.Mo7jGC;*?>?E:)ּe%7j[[r]ZبւLm']_D7F)ڷ E96 Wt,M/+չTb̓H>KX 領U(#*pu|s*8atb.)W۞-Hw\K܌sխUzjyi'nt7;cm3{m Q@8za'P>`!, :;K~F>s" b t_5 NŸ^QQػfhM39,fGXt?Ovr\Lc/w.8A&]sSz{S,Ի.8Wӡ-o2,~u/~+To>PPBIcY kskaoI4_RL9,i/! amXfwa:c˒O^^*1{qB-.r,VIMFꛗ޼L5y@%II vaI*. =Anf3ٯx̉1n*j~-z;mUx$L%Gk/ٸO>Ūш 6':?w 7>^sMĪ;]wW C۵?``bATx%큾7,lW?$S.*SNڎ>,}|4rĤmƠ@v$1i P =k|.n7(t5%1ձsL{)O+ 6u'1>=XI&5U]V/USe&0eJa +_cCaTUR%,iI |{Lj/"m;FnwU`a[-SػѵW39IK0:>2wdjWʮ odE[n'%~ଽ~ƹT/X/(uDT1 KQC$>d{yŧս} Bif4 崝UװIod$w+zߐU_f_3IRTֱuUw9nz4:'mKfbŚskVǵQODNc/}| EiWF/F2%]u'IW_ܜ#J O$QNj#ReWY*7hs:n+?x5_u?DXvEӀp'F]bNGW}uA[umZŶ)8 e}[eAV~.&$%I}pʚEP ɾqE示>h DX9/AUBݼA_iY?nfuB1G>|adӔ):!ٔvo96z D-p2ҍ#bT|7}a{EjlC([tbrǚo@4=FqjUx rn%$EUr-q#6IJ6.o_ {y9(sA Pp:ZC(JM?] endstream endobj 180 0 obj /WRVPXU+CMMIB10 endobj 166 0 obj << /Type /FontDescriptor /FontName /WRVPXU+CMMIB10 /Flags 4 /FontBBox [-15 -250 1216 750] /Ascent 750 /CapHeight 750 /Descent -250 /ItalicAngle -14 /StemV 113 /CharSet (/alpha) /FontFile 403 0 R >> endobj 404 0 obj << /Length1 2668 /Length2 23055 /Length3 0 /Length 24571 /Filter /FlateDecode >> stream xڌPY [p=k Hp 5̼UUkY{u!'VR63J:330D33XXU-ՁNv<:A61#gP-@ `aab01q/Α fja gE=->;G 5t  olhbd P3:{=#j:9= 4j`d38rv;3g7#G d0:2\lMi9=`o+Pcpvwٚ 4vYXbnV='G {g'' _-2* ?1 G ho`fakj S{F5[ ?! o3tMWdeueog050y9Ύ.@?EpS g1- 3o Z;@ =fӯ?ejgk;2 H>;w= Kd>-dd?rm]?Z v TEdYS*o"$\rSqXX{3ہ }i偦.6+l¶;TomWuˬ-lJvN+zjtL@'$rA7Gؙۚb,#GG#8ВA @D `ds@~m(70F MF\F߈(/d0JFF߈(#6o"F .o/qQ@\#oF .ohA\#Pѿido?obnhdb̜Y}u FvL,M\l̬At5Hn =H`!omjt:ZA#1Qsr#DwC E2Ñ$b#'?@[1 A%lG YA?sq<(Tl={؛m ԅ4 ? h@l4ߕAkԬo2dAvكֶ? |H6 ]! ÿtߝZdw vP⿢btc"NtKR+Կ .@*\ fn&'5qq7"H}݁&p v&A5AwUxn;cS;_\+>9' v!/oS] -={5B6)Ng`@9РBJ$C s5K{%~sLũ G5~ ֜WQr߼{_>Q&"]2=2>Kjۆjבv@%^v\uq#"FeAs5`LdF2 {+`Qȿaep6$nahGPW3NO*Ad(xPn5 (.f6d7){EOyݹB4ʜ^k «<sVqôTO٫Qʼ5rLb"M~|6ipݏܹ j-u"=77"vna"Aٟgv8)&K Gnt4DP(xr+#_= ^ЛƩtRӦxG! y%PxMU<8ee\I]%BO7^PpG3R3xqWCHsH$.Tȼ+_ƒ' } a_ZljVWaꪪRq 80 F uگܕؘgIY|ˑ(|&"Vʹfbpׄ\w_yM͕X'V9[F߮I.A饼̽|SHv]71a0m|ꤖ,ò, u'䐬QdF!R,eH7"ӔzRCY#}ֺN4pt0GvdCqgN2|KKlrV4MJ Mbob+5\xLj2)"|R=rkm%2UmL!N@,g:Ygwͬy@_ 1SlV=m?yOA =5@;lS (,:f$DSۏ Ģ l9[vK7b Iϻ"ClmV'sYmETWr&36u$LJ6=B 3!y6L _!i S{ͮӭȕYyՅUDwJ0OJD3snӭy+;H5!m%bI`"|}~.Gj@XONmv*Ie? c[~ÿ4Z9FH(1Q|=!i;yq^6e͞?>KAZܗtT!E4D}ԩ!U,pV ޣb6b淵H|[Ҧΰ&1~Qm3S9QU508Dkgyِ |;m d+d:ባ6aP9ާ]LHS2J?kd6\s_]w>u4l BA1c݂t0;$C&.ĝjטvHk*Ɠ_VKfgu7ؔF|ܛ#X'}'T[$ϘNܽj0K8OZ|xRq]2NG鋝ThzىN|3[( öQK\ǹT͸BtlXontug0|ӗr5p섭^ N|uQ9m=qcvJm=(~d%m$Js_C( 5y 4V)XP=wɦY4jV_swH 2цF5i$PV =J>+WÆ *'tIД,[b^|Wܘ慨~Sh匪}<{㘟4u݋viGȏ Uz 7&&s||J6cgʚ:0mew ۯi:EԜ=OYc8ZKF]~kɌ oe1ajSd%֖RWl}1O *s`ԲzP]7֖ws~2%Cs~=uJ@껺^{K_|;t}nn,1 C,*Jeà]|F˄e nlI)Mt*wwhM\ =fcED5"C!?Y][aZB5rpflB?-CqHj y^g'[ۉ%37KDXY{8-_3 hv/}T˻$I8F*}inCJdT9q,Ą]Ꮋ^SEΉixtH:\qNJѾ>|[-r.e2ĕ83nZN_c$5rv,] ˬv% bz8@z$OEg`Fp.yډ|Bq-}a/ G낭62wSA[sRBf?3^?R0Rlu,X{<asa h]ϊ~+9ӯta7L[`.x|ye9:ۧjr0G1{(]dE`zBgM7PynG-ҎgEf״P,\V/&Ύ'QĮox,BvFA8/Cl6*l63] -ң*[lIa0^1Pg=ϡ?0B*]Däۜ,SlMGW[ڶ2myi/x`.wBeDa)}xF٨:cȆ#]nFxٞrFAX>CU{B;>~v4!wlC"2 y!UZ j{^E$ F4qMP >=5w&朮|t*lvn}D1>=QW@ pY7$o\3vOkuCo}+ER< [оt#7SoV2cZ%z5{Ï3ʹa>I&Mw&a9%";oavK找T=Ǘp.:Y"o0 ip_RjuW:5y!>.. ސWÞ(,sv~RA=ۓþAʶqF/ U T>5F縭`}tbn-| ,$^aSiQ9T+I-1qs XXl) XKCNhܡ_]1ZfeC+Mx#쯺H…V{cOCյi6[x˽{y)B34M^n2|x!`[%n[QpG: ͪ`+=α7M|BXTE|ð2^TI‘h K]3ٮAF)oc('ZzuȜ|<똸~*qMM)rMNlA@Kٰ@'Y.Q)[.gU e֭&]THzډB#ox݋9^ _Ӱ&n/w Krcf{K g.K"qh+Z.ޥ_Vzv$ZZq*1Zݾ~y|)t!gΗ}ZfHRcnHS^N͈XP!-ڼ)'Umc݈y 'Xs^-!fեWV<8!-oI|ė8?dpN@*OK"*5S\G :<9Q%csT$Gs)~`._!ˋi߁O`^fk:0)S,F)Q zlxRxyEC:4+Do  |0$)񗖁:t0 S"0j15S7Uۥj k+Ww~78{x5T=MJ+L-{v¥lݲN7quDr 换j^ u8r7((b}Jq]]蕠J01shso7è<;`sdRC%5m9ۡK|%&~XːtuKqhƅ1b4ޔ,f[i,i s舐a'8xpԢj)N6E2rn xxkoAs!B(ɃaQÏxp<.֒=|GbQM~ee^`^ح;Y_B4ܘz%2D**%Q2nͯRKFpشS#S Fmo] fK=8q]!c V],odۆ` !BQ1d&L? V%(5Ecy&Dp܀{R}dD% Q1JBU#RO?iHJ+C-ˏ`il=]e4yM]aæN;+|/є'7)063I\v#2N-D4"6՝m#@h?7d4s%#٭pzř:8:PNx mN-IéJUׂ 5gcword)7_ 80Q^ #rԴ8 r69t_{y"/uToMJ;VQ٤H(k((1~-<;"Z<тHQ6b;g^ Jb"c㷕OW͐Hyҥ_an/?g!/ n1A<Z@DD&dUBmuHjKJ}ᎂfl|.$ڲ}luyuϜ+)BNß>^",+Gvry6"7"ָC;y8^aC+ٹ2N,VۊdLGGVUF;5L 6i"P UO9N=AcX}GG5xNP2\3 &ESi̻y4^B܅`>/ȼ6Shj켎s<Ms@)?̱CNϊD'Żr gIbgl;-u#v`ZB;UX~;׃:Z}}.]:ЂڶDaIlJ5hxEI(o4܆qK&C?s=B@#&BUd`r+ %\_Gh}WHm|M}QX}+ıms@دe!C49z00oѓJ1ϋ ЊiD5 c7ߝ<}3gsi# ?*v8;R=/܈1]4& #A*b8 iu]D]O#8ΦS012 U pF1Il%vx񍫅NARL{Ɋ;%kF;$͖(r쫑=8ba {eic|aY«:X b4 9H{4dc/tjCRy:݃UI޲_YYk4?L&*Sh]ݱzHn9Kpɐw`:áejϗh|暙`W̯%h;[o- ̬;<_HC?*jǷsƒ >SHԚ/̹()0Ŗu4/-j(gGj0yrġ>JiQT$Ŷᬮt9w/3~n*yBZHVEI5xk)=%>x)ӶHZ~_s [,ݏt6ւeI 1]<7>=(^-f>5&!;a%uhHOt;Sx1osGF)J9ob;לMKzNtvȌ4铃/C2\?.2bfpz۰Wβ1X:Ĉ'g!*O丝wpQ !X2b4*-9T^5ŷD<қ;:NbȅȨ^kZ!$aeBʡ97+pS㚆ܧU#^QحljNƐ&r D$"_M TGJ0G|r'ŃI}igBQ<V!R藪_;ϓJ(L/=J6 c %)G8Kn0![ rk<.E]IHdISi ssp_?f"v8ۉm/ȷj%M`oӍ;&`IrMXkR5\嫶a6GtiEM w^ձJ3Gk$-uÆrH}4M7]oUxh L?,$8ccQDvQ|"L 2lطfj^C5\'n &ged$ΥkNdE=#\v"X|6 !aYb8қx{#)jh9@0oK}%֪ωR5H8'ѓsP+X?^ِ͞ȹ Y!#*( c ̄ݬ!c0 1:;G5YE=Y U%5aq's8޺RH~؅d@"vA{_u1WƑl|פ$C? +Q=.Nfd0 fNvcuH×p^]"@BQi0PHh6J\ k lt@OvIbs4+ڹ :֠H#kpNЄʇLgJ0f{@C\B1[#hOL(S+5UDzͼDnRW~YYiծ- 찈;t|PcQ;J\u5l9$LϡbX’V>K:e `dow@˓[줊*A>V̛ Y!t)t NX438L;˥mKV>bH^b >Be$ZLXDLܻR/Hng9Uu #̒~}%QV/h@QFy\^}C.եG3n읻w/rN_2{lae͜\SܕYUiXb]RN9É{Mė RZqG-oA xn~AaWT0pO8\?y,ٚ{y2p=W]m{tt Et\sthO2ޝdtU/]}^3NtK8iI?uq4M@%to';s_f]>ܺ~Ka9=Ϗg"3 tOj6x/R7$+}8S#U1K5xq)% Vq#/gRX l '˒f1,?cJw},6ךPҽ*߫#>_SWl)͔g豹ƶE(1X[. KpOW;.OВ7,)QNQo}`}1YㇿJdm#2,DĉaʀYE͍N> ' 8m5h$ A~!`m& '<]»Bl'ds A-Vg_kso!.)7nPfpr:E\aaOwu/|=!oRue_H. EnDJ? 5D{hdۣ*gkGpMى<&kYyS/-=^mߝ*O卿-­{@pNvRx`\ rK,_9uz\kS3 %.LAǎ,"6՟dPe߆YlX՞-%/ P>"֌u#' r\-V\vD͓1Gɞ$u=ht/VG!6_Oʯ9L#7Ϛhnd|Ca^|V*~gp@$Rx ؓ~͠ےA{aIJi(P.LwEp%(x+w睒9h\߀ B ȬY351vG5}An *'mlLl62J.\g^XySׯo:PNMt1+^?'OD-rt#bGPd1f=Վ%@q?qQhj%.@7KG6YpΩQfl: {B ?k|1[?p Jk8t2xsŷбݽuC Mξ+j7Z}ThN&2I|JcO>Zpj#Gr8'iet]Al΋%It{CA-+Hq:_c0<4Jf:$ܯq|:W74 [γJ23|JĈF=əSpM&M~Vuc*| V 8{fo.}MzJge+Z%P[|ʒhRB;Z-!6qv >TaHqnI@\@.HMNŤqFC41Ofn=ͥ ؂W˄}bnnO.&}݈O bځ6"Cv眀ZQ3`W: ڲ%r&mWESN_pLAeJ (:X0N?p(ĢfPGn; GűcݙSmZ .#ΙP(+(pRƮ)EuN[&'O>oҟIU ;y#qY ܆} E}2=}g$m'L JS'-z~TB 7 k2o;{~WāDu65ZPXrR5ݷ8;h!%?'nYad(kf[m[6xE4(_vh{7"IܰoxI-Rsq҄ Gpy>;iLoXY.f0@u=9QXJ2}tФ7Ta>EIh'/[>1=cⱊ;\lq@;:nFȱ2auw}hki}dl󪆋Uc4UL6O~jd>؈w0Fzo¥ wY-=q4&u}"c[4|0Įc[Mbz`)> _ݤ 2|NpEZI}m ^7@KKD't4YGq+-OU5;r)Aq8IdG4mz1]#hnbd7 Kyntʆtm͂q9>Û B$s*"k?i|ݖ1&;)3-- jj]Y:&Y9'٥*rTCMY`  jD?9 tg\#{KAAz=!Qۻm"Tbޠ1{_.-FϢhXP#,9UMeAIUbMc~n'JTa&sGQ} T.+Y0i!1 Eh1K?F[ aݽJNV4*}^=N/Jk7qg Kc-bے s8 >5m< J-iۖ@n1У `贋iz M蝐Z1Q*Oڎ* M,,M&r3 f}$zL#F FZVy 6ӗR IBIZTn0Y5%e5Y)eOìEpsIS6C jObEOlo]G(ıŢl䭄j[Ao^=? uT.G6t9õ6LBdŒuߋP&; ˥t4HAz=Tь5㲼ǐhS[K;=T.bRƥ?]jʺ,Ffr4io*i{ˤC΍glܫgZH=J̨6fEٲf:#lV/uU C-0sL`٨ĞO\z4f}N6.ˬSxv)'|r_?F,hր $8 .ʮ mfVYVg4Vg3U BߢSStZ`Xb#`IyG]]3]SZO[!t),K-UR(;jPltaz3&%'#bս1[3L[4?prvࠈ\uL_hz~?ig@EMJ \3!:$r M+G}mVߪeߖcEHyi`id9&XS -%<{=Ҟ_Sb@$͊_lvjmB/ckg=Gem:o:;+@[ɈYGҾ_8$ :٢r駭:Fc6ր5MVZ#%ӝD8߲$K N^%Y [cK$XKƚr.!Ф++=FyH;[1Z *b pVk۞|׾dgax۱=. ut 3$Y.8?: d4>uqz{kF3/}d'<];&N W /i o'XTw:I=T:F(s`Q̚QRY|ʾ.Vp1^XQF!RPioWS iE\4Z֕nF2T ܭHX&,D) tviJՉNٵxЖ˞4 S/ʈ1F$B#SM{ jGڬ*8-yi6 tݧ)&c^Av}z=MvN ) Nz.B=8An?ɇz=hA:]./nSˌcewޛk!i`L=D,ޓzҀIY.rl[hf{“l;zg|\ji$lH,BfI9a2Ki$ Y髨7X|7(YM8rbA8nQטg8&[mX28m +CעT0\O!rq䆴 xNFnA`}~%J 鏘Z<;Y#>_]~^6Emp[Duf$ZA$ PA-ޝ"+ٺ۔vg'TPw E/HUBJC%bs$p\m;=Qyp[swq*5$sXRR#vPᣧ$ rO}"]2y$T3s,_wۀXlѽ&%Wi[41eta79VzF]7$nfIB#BJTXOXCE2@&t}<E^xRɄ7xS$E=zɉ_vbuj^-%L3!׆v }GG5 3\xUP ,:. {862AIXcB'[EΧc'=MeFlgՍE &**'5wk 6x֢o0$7$hnetZa<LY7UGfZ$u(ν3|dǓG i7E&K#^rGjʗ`[.Vö cfD2Vp QTZƪJ&tKP^2%YWE3J)4Yprr>E`^}n6<uN}=2Fx^IBנL=Ȩiq*bHX^,wiڨ7/9] SDg,I)Y=?Hh/8o]2Qq4¸sr7> CM}y!E~<^L9Ƴ@Ybl{ķFNg13ٲ|C㲠 <^bGQED61 ɰ{:N(0Q 3:lጟrm-Ҿ)#R4,85JBJAaٯ\-zD'y0ZA:۵5 Gԩ`Q&piz^h6#p >2o&hɛɏ4 G+i:-^-ٖE796r8q]}pǞKV,IPgKC !GtƃQϔ|x6><HVx!~X]nc[P6Q |n7ġ +P s5p ,Ь˶_f<zu~9! /Oֈ{٩沦x{DֶF,/:UZ !*D6M>!$g[݊#䰩k/8tdݳ$0V s{w lL` üm^eeɊf4_+|(MEť:k*>%!/C9dM/qc#kmLb;躦&5pEN( d1 ,\B]uV#(5•Zgd6SW7HCp7h#gANv*dSpa&?G_4X">˳FsSB*DTNG QXs"u{p x Egh.O/:ç+iS!a.gn1ERTlt-CmdςH[*:̗3"@kX}Ԫ Ǜ7Hm89D+@[y[(6S֧NC@e)+S !#Mĭ?s8\&ʽ1U]0~̄`:XCQb"pc-y>ޞp,}?XNDjK%{tI5tPWl'{T%9<ƾq(/ {ŨWǏw=]3ݽSy>+]ugQ!vѩ fc;-av q 1Ie#F;Ȳy@𶥴@i?S]x 7*>5F4okNd*vٴl3!-3ѷ 5kO`;Yh7W:YY]Kʫik\ 1rjgz# 9ؤ7}ISL*JWyGLq2 D%mo+iGυAm.x8S~9'=7O\/w>fz Zh:T md+U$=yQ؍,5j/^"ȉm؈4d@+]uwYY1w =40#o~C{C߷aGG< G#v8A@|x;,xHPe|˝螡Xq- YֺSR@`ѻsj j߄]%+ sfU&mi`6+?j …;d¥ܰ-eÓNƆxz7:$){|bNA^,#ob`y?ݜFkANHubbkvIyYWȀ2)$iIR3ְ w/H8Y\"Œ|1`U@引dCB฀ ƶzVuBlx/I$S 7Eļj5ɜrrEe} "|5ĔM-㍢Ҟ2bKXH#vL~2FlozJ[ "iz#뿡3.u˃uªgQyzD2Mޅ1WU?*_K0mt(2`zhHnL=e&>EIz;OvqZ{}R^{RkR ՗\)teA=E$gӺBzIz)>Y96F9~HO<Ǐ;e򀾒֙HZ~ ahL01rcEN$g~8[JU1nf0;v9VcT71[->@ލ{R@K #,_+MT@z$BL,(|E @Q [+N!n۲[C Jݪlb6[Vx+p„B @O.8tenBJ˪0}V`lԯ8Y9;:kf1p'PFmy%|d $쥖v#g2!}z= 7ra|1RZs_9H~Jҫd3p<DbKmцa(۲:0#}Y˂MUsnn +Ԇu dwF/vn|љj(ci8 Fet'J{Lv{:u(_V{ʬuPte6Š@9*rX^BZc>& 490@s9&qϯh[g@DmLϷo<O.khLPwSBj ik:8>:70%b +ifw4a ]!u٧`gUtKaxMT2f~ccD$B>̑aߜLLٴ!`J 2H01Dx LoTkP ;dxٳese,Vˁeki3K\,4Y'=K8+1gzR9Իڣ/V%q #*S =<Q0-ӥէ&\TH],i R:%}Qbr-Q~SWU/'^; nwzڹwuj81_\8 8g9:Y 0&F?7{> #ǟ%1VleYpgڌKY$ ŸYkȩ8n@w&W6of<@C&Φ+/k^*!nطrV )y(O`0~J[EBYx_z10L=Aa L8JY'6 5 pF"G)wꐟ%1:j[SiBI =*Ubr 䆩7jњ> ؞ٙݖuҞ׷d"I*Le}B1hE`%|uSe&r 5Һ'B?h~<\\.@TuGOub43*ʠ .m'1\yl|8@W_\>;_C&HI謇/[A[J 㼒@,.Ǝv W>:=Ќ0??&L%N{0{=mPl"UU|&LѱkY@7/1 :bAD 5sqSRAt#B<:+yR}gEp Ծ$3"~a!(<9?m*]`ϺKwO'#۶b;',e9] l--ٺD.Ȭc>C=-F!IHC^34n"೫P|X( RUkؒ}č1r車tAsebuN*9.8 y}SٲwU&{Nw/|98TޏfpGY!*ܔ &X<A?3N̮HJޝ*M#uE/Cd;Z`vMŴ6GU/QR2F?):S/#?;B`Cht)fȲ5B6%zcPs;a<W:\^hf @aB6䲬ƙOh +|XIZQlͯ+ɞ,&XTˍYzڿ eows _[/_%RνǬ Swte l|}#*I*5/{(T)iJN@*Sv"+=8t1CL~9^\TV|& 0"8LeYIٶXP6߀?Z 0|ܱ;>}Z.oȑm``#8YDi֭kPACv&rgo| 4 ~ @Ę e .@ k(ڿ!VhaXݕ>ڟ$5Z>KCE#k o~FSS^`oBicH5reUJF&OHte}T&G`F3Lv+PN꬞8nޏ4z){|6o=!CFAEw%yT*Q]+7܆'n, ~1+C(ra>)]t0ҋLn{ډKÜ uk?Z0)wa}S م.LVS0VQuI4Le3Ԃ8fV2AesG\H}OB,n89#=n~u:+v }I \m6 _aPW)ߥ#pM\U/6dEKPٻ]c ɬsF=0g+.m 0E4BziQ!.e^WEz(>1mτtf7]|yp2$[ |)! 4ɏ\'%[!~Ed(egijH0Wo78TDy*M[ endstream endobj 174 0 obj /CADBJG+CMR10 endobj 157 0 obj << /Type /FontDescriptor /FontName /CADBJG+CMR10 /Flags 4 /FontBBox [-40 -250 1009 750] /Ascent 750 /CapHeight 750 /Descent -250 /ItalicAngle 0 /StemV 69 /CharSet (/A/B/C/D/E/F/G/H/I/J/K/L/M/N/O/P/Q/R/S/T/U/V/W/X/Y/Z/a/acute/b/bracketleft/bracketright/c/circumflex/colon/comma/d/dieresis/dotlessi/e/eight/endash/f/ff/ffi/fi/five/four/g/h/hyphen/i/j/k/l/m/n/nine/o/one/p/parenleft/parenright/period/plus/q/quoteright/r/s/semicolon/seven/six/slash/t/three/two/u/v/w/x/y/z/zero) /FontFile 404 0 R >> endobj 405 0 obj << /Length1 1987 /Length2 13553 /Length3 0 /Length 14765 /Filter /FlateDecode >> stream xڍPJ .-%Xp=ݝ %p{>-yW(* %lAN L<Q9%f+ 95?fr5-_@Cw{- l `f0s0s01X'ց fbac؂䢶vfNW15t 3t2ڼhlh P5:T|NNv< 6 ftW 's40@Og s ʶN@r|pe v@?7fO_,@'-@fS k @AB͉`2+==o /Ӟ#_-2U}A&66@#_,Zl]AS _M81,읁b y7!LLLl=flWyw;Nx{Lߛz[!x:N@o;7!03L,F@3 Ÿf?~nmc0oeb v2j+DDllzvf7 (ZGӟTi-?]sTԀ]K}h?3dyN7U+Ho7g}lor@ gv2| 3nDaǮ%m-zULL~_|8{Iq_7``>H,Oht{ [{{S[N(0!N0!n ( (ޫ(ޫ(!b0**车zA/qwd}F]}O_h/|/|k/|k/|/g`sw;OĻ__.2KW{h{ͿZ ;"˿1oӟ\m~Sp?.w5//my,:j \&gӨ=:P`6G>S -ؓWzp8ң eĔhJ)!Ia˾Xpl5rSHxSgWB}ʭ8ªӠ U2J'S D }_llF1'l6˪s?ZS!]us9ڡƼ1δw4O&ol ?iKC%Į&tn%Ə&MkemTD׭;Aά>ltEx<,S-{B+ylwƖ#%JRp^w=u}[HPf,$o'{>NЏSL! ͳ?W6/Nl#iɤvReR(8i]kT֩j[vq_Y;vkz / XnSVd@9cGns(Yj>V)\`r~yY"{nr>WJayR:> ҘoDwcR;H~e n2"m(!E9Z(lG`mw'=G?En^Z$z] g6oYllFUo( ?u]^+Ogw蕠sĹRJ8Tȓ&*t*!'ee4\wr(W `X-'<]9yץ*a9(7&Ѭ C2BqI +ѷS546kuklwV{, -04}{Bb? Mm`wX;7 (f$r8( ph6^H09Nq˛m|2m2Ed*'6 @XI((*TzTKX>dmF1:,MD=L1b7QgCUU1[vn'Ч#:!@;F"?ױ`}XԠm.o)3fίBSj(!Y& #i;HP.&f B>agjFy=uH!t&q!'ߘxw@_q}L.;++PP&SWBLM-H)}KcY&_`EY*k׿6{˩1pD13`PYbC8-tK!z B:bRZ2#ŜqPefYItR:ƋB;&q&؄F :ZgN-YO7̥g-)Gxs$saa7rD Cᅦ!}sq jCdd2芶 2YI%/d˂÷^Vq(M.N[RDLXk#t@f@Á82DG|HIcD &G͕q#Xc4֫xu|f_O KFv`Pg^Qzlyo[8fzncTn o9D5N\8socľ%%wgwBC}o_W`Qj4|2HBd0|EF0V\9PȘnǜȞUO6ֶS]rTFOTI`?ŏ'ߦ")+N8"XJ͂\hy}tn جAuSNԭ&/>fz,{SFΏډ~ J=ɁIbjKtIJUΜ ȿ,mB"2XsIV/N->8pԱC>(|^)EmA{02Ll nbR;"rRU0pgf,%Ի c.68䷏ӘA/,óVo11(N x )QL_L{#63s$^,1?y@X~VcKr‘%+J6tCLdE[\F)s24]L^nCQ~ҢN0MvYk,9C?`ELsFrh> )`ysTkyY# T ?aӤ R0*[6jgY eR(Ch{$g﹘,J*e:}:q54- +faP+ mH:ӬԸՏgΙbL5ijS >,1リe=,%qz*=9ذH2"yx'J_!`-6o1./;{u N{;֦V[MBau;o߮1)JI'j2iqFX-ӿd{@|=5=j j X'Gr3> c;DrsmӦXEpi0r!fky6Iqk"ϒ-6`<ďQMS"׀ܨ|C1/p4 #|ș6cki9fz(jorO^f>W~2Z+*ێ7Y3tkq_-A~Qv7&֘bĘ9DF,rJpÆҊWa8H=YG +&=U|f ?2/rE\1ġiFWk`ʙ)cS].B&.D8m: ;{ .^j`4ҹ9s^hF~ .V;{|lR`.ДakMǒ\*edTQR:YJ"=;tZ7s[0N"aQѳO1:H$># n=m =Ӌ<#cTn*]?2mehf{)mMԎ#Qp8~;(^<{Hnn71Ҍh-dhB<> E?U|p@C<%)~S3KDJP#Eb|=EDZZF:>~1no B%֭'aZ@O.` 3Zb;}XQ^ا#ة>o=a†Ҳ bCg;$nH 7xAJO').v&oBf}_Zm>"g0`BvID0 9k1PAS+@M?RУ.B Rk3qDvZl/wT=W h$eՂ\MA r jgDnOF>IzaSilu}A\,PԁmoחP`c\D6ͼ_^ v"6,24!ai'plwĂ])/̾j)֮NNV]azn"W\ hO :wTOP~6O9\(k.*XΟ%{cA;A";kX7k 3<%9D@Dv<?>Bɥ&h3ƾfy)5kh`ᖱMX"hX684ԐYA6GC< & -=Aimo݌R˜0;:h{pD5uR'` Bb(s z (- AW(*f4FAp4w?ml|6R)K<QA/K_4Xa-|Ksu%85{rjVaj>fn:Ņ蒈 bF Xq *1ͅ o8\GN#'vwX`_,֕ˮ(wȲ1 *". 8&+܈&?mT釁j 쫄mfCE͕d ~ƙWھ|Z9f턲6Pli'rYLc`^D'GVo@ֱ~FhDWB8 gcwani2 ibI50i)C 5tɫz>ݒ[Jy!W} G~6R-dgPͼK 1Fni{b_eUfB41ߒ WJOA[DFxN4\/FjkDf)=LR(i cghဣl-FB_heSNoZ&ӟ|=Q8fM0} dXWm=V8PN qf]qc$  6Igc΀&)GG{6ju=tLeq>v蕅k{^0ri$ωV"'jQJOP]ģx~W?8}du;WҏqR(^l1)6lA.B^T @%HeuN+zp7YqnMr QEyM[f/#Q-44ib\8Wx6dFfU'xg_l_drhxTÔ拾"ύ9Iwk̾pZJsP?,=e*NVb8ޙ<7oynUmnBSQJP>[^l_nTµ4~Ba/S^ڟf`hsƷ>T;SMv$u55%\'L/}i C#\3c]W&婕x`=ՑX9dPnT1u^8h~bp$sy^f#2!2Ҷi$4Q]{&g'tc)Sm}N“j1B$ݣ+Ä6^r.Qq@TS0]zjf~4ҏ.]^M=8W?mz{7ȋhvd 銽z zNeXB'nB޹H,%4I}Snj 4덂;u.M8 riTY݂%^Y!~]WԤ\#o9K .)},#Qs{w%O~Y}-+=2Xq{72$]eL_̏('/rhl89=7r6 %(l6?BNBWQ)AAm34 lMˁr""X$A \Ͻ LSN\Klj>^I{oMZOb0qJY`ρ ݨZY› KwqϷʄN_a HĈдOJOrJjik2^aeG旁P3w_u|b2r*V%--0 ݐ,aGNs/6E|2˿FfN}Tն`y(*̨r} ak [H҅M b+Оpe!Yq X׉Ҿ Xib4Mo7D?!J" :Bjr ײ]Nn,3Voh1+4,||MD `](CFYSTiXz TNfaa.f>gAnV ǢZk<10x^quY]0ZT[rxZ Ȥ2Ȓpg:ym'Nx=9/Hl^/p4I<+rwݭL1j1kxB)Y%' B㛾E#y`| ^N"D ׶/W!EUm>YRR*͗:fէ&,TGz6rPnI 1˚dyNܭs}Nƹf*زQdQ|gmTG9DwgIm֥D"\6 @|j^I=Yn|FnɐJ [7q:kj@;11裾]waMRd3:'+tZz n.EySD/`TU Y$'1Ѫ2C |64q`cSڃA5ckC: EȒaq—j&bRkJR}\q  NKpi Smŗ1FJǴrm8F`JJlEu2DB& h^(_!\FUk{U%*?oKh)\w-߶Wm)(EzB}U'->=41Cv02<Ĥe{%FroIJD.px~x,+./kN|ˬ QجY8e'x7Z'!g 6G0}V陨2q /ss1EPf I%P1Өs דT213JȎc6zEL<o ^R˧lBoˉgTo-qI%4g׋nX!2/lPb#/v b9;V]CS84N*IO-$V8`eИSb} T MTYCGSp:/г\A$L/_21s$1l&Rea(0饳-)T&E VgPY[L&K N3bKnO\G {oKGUTl F h>>Ln[^VSO¡ǼbuoۑNHzϣ>.m49 FA#-2#ە kktUp\FM:Dm }SsCIa6I(<3 .а;~<4$Rf"8f0MHOMWjJ:Q? Q'ZnGnH0^ h;dVE~"<gG *ig[$;a nHkWŤJPQZubq{c]kWR ek%ݞ3d'Qíir(k\}eџ8 dns=cyxzgw90kZ߉[brGݘ3+}oZΚ(5:>a;Gިڶ/.٦ Q8\k1?KW/ '-ZwZ^Ww[x17HO@kA`ְR~cht666NOy6;P"Oӧ%6RʲC|y 2q5 ]Wo4Პ?._W2"5B]}s-t?yCu3iDI~ZEIT?%YA /%j=SR$ܷOd6 >o.7CFP2{F%56 ztvr|_(~ݷdKaeEJC;n~, "B>bneOK*p&-5":-EgakejpUĢ+|JcqߪՐɔ#f lDET޾\uO1sC17}Z(u0' P$ɖC2"KkHKщG%`9hW3r(kЊ͘pq}UimiVV@x:zZ<_`1cO6kzag(e?⡹jE*CQx"nl4H^>3̽oLӎb{>'j~cqb8(_@̞XK&,؎[~ ]/Vk}imu 7-':qwh˝<9d~\5굕`[/kgY0o!g9Znl=GM|0Y:,u&7dׄr&6X%X=$*-"=WYOmzsRKkc6sx w-j*\ά>BԏÅ<R%׿\bGx$j͑p%UUG}ZbΝI󧽣]8&G9gÓt񏖩tCJI?Pi4/͞P68;_uw cEoMiTfw Dt;?LHyuPڷU4@&w+Cic&m5܄ցpBw%NF+`ѯF!X)E#҉IGReMP7!pr6%(᭝ycB|$:2%y򌓏Ux5D3jcW6ntx0/Yw1^:( Z~|Oi[!e@'{߉UA _y8)xS[INJ0_E"@Ϩd馄/ d(C)J\9oUhTF+oo [^-b3h7q T+ 6`򟓤v4h62Xt~ $-J3FwX;bցGՐ LɯN\c9#l8vkAoCWADz*[,SPaXOVo/=҈~z(FU"vz%S|4NӇp7ÿ&٤^"汢Y[̓(c'HItuh5G)L)F==NKTJRH ad׉_CX~3*j%y)tKC{DŽ#3daoo>(T LI&%&T,Рw /eet6m}u_ +ܛQ:+7A4j%LɌ}MIk+RθU eٸWe= uSVbУoje vx~ZvYz CMP6_.AVf4M /M7m J9A~G72 2m>(N[4؉)4l BR?[QwSʅR2߳VxZ?jֵMB%?:^TAxqtH`WIg H d;ƤY/|,!wrz8Y2뺕K/jz'R$0NS=jR€xPRoCL:Z+mk^ Kf;m^'3|8c=.l:a4Ug"Ҿ;W.4;@rqLҴE^~=dfD 8s9 nW*^ ^j&0P:&]rǝʾ+h6 )8Vp]IvzoDC[TQTv.،oy'DRJ!Z5Jzc1mYbQK{V!s8!.1JĢKeyľnȻ`G|W5K6^mC4zFi.i "4"vao38$tVА>˃|#*}AU+CHanՂ [eNPըn9'NKT ԃ=8Dye-*tvi&-(k웙n-A#5+ʮ $6gZ6J3I)1O KT>}bn2N5fs:\EV]7)p6G>lV;_T݈72W+1`/!=Wx[QFO]aT{H-^@Gr|Wm~Pl1Im-! +p)}GWEXk*8mn!,8=B}yiHGr&L HC1 YI$H]+])6-Q!]*OeVYN=ՠ?ct endstream endobj 406 0 obj << /Type /FontDescriptor /FontName /XCNSQF+CMR12 /Flags 4 /FontBBox [-34 -251 988 750] /Ascent 694 /CapHeight 683 /Descent -194 /ItalicAngle 0 /StemV 65 /XHeight 431 /CharSet (/A/B/C/D/E/F/G/J/K/O/P/R/S/T/U/V/X/a/b/c/comma/d/e/f/g/h/hyphen/i/l/m/n/o/one/p/r/s/seven/t/two/u/v/y/zero) /FontFile 405 0 R >> endobj 407 0 obj << /Length1 1685 /Length2 10100 /Length3 0 /Length 11175 /Filter /FlateDecode >> stream xڍP{2; 2 3k %@xdwUT19ݧP`!`WVv:/FjKB tvA2p$\_ `= FWC@d PbC@ 3%~- 30@]+[WWG66V3V03jP݁ vƊBд)׀Xz9/{:^4*@Ɗ0:+hV { @EZӕ`mhfy7s7ٛQ@ZL `_X8]]X]@[d唥 w} g˱{yv`/-~7aȦ9$2y#xyy@'†wxM/GJ|!&~ +; ` pA`V;</`xYB^qlʊb22LvN\ apyj \Vž v R -@ό[?\wیoAnf { ^fe /[_S;9Ve#E TZ9,ʵ~/= T~*vѽl2/)`l2H<<Uz16V0Ҟ F~$>`^,&>n `3^C6wlB_ `rج/%/y/ADK"ȿ%=K^%_7c6;/!=(K Pۚж*1bq=%v*37o>~\ݑ]&9lChh>׊870(VOL¢)W;S泓j.ΝGgmhڷ7 ,qZAE49s,Hg73 L(~Gq\>59] HaGh}Ŀ_).lb}4:BY7 WF<ٌ| C@:nC'-FIj{7v,?U2|5ޱs2. ~V{?f7O3\(^zq=,= 2˹1WTSy;0һ% Ƙ ]b ;T~.(V٢]җiO8ЬJsϔ?zpS&y'|v|vGg; twD='WÉo$񿻧 N}y#c} {LEj)I60?n tĆ3kΑ"kZ!EoDNJ-Cc/P.uLf'ޞ9f2lbA!{VJH Y.NӇT]a+2N=4SrߋZ5+ӷ z= (6qjmZ_psaYB\&b.K+`/ P >$D Gc/}ŏ=/ڄcj4f$OA0NlLUu'rwٵiDLt A:Gܤ @2Rh3Q|W__K{y|^ }PH69^R(\j⪟ T̸D, #4%Lu8N-aO`#?˩(Z"Яvp6W<`Rrݰ\|Vld3r*%.1$3_ !ESF.o3!r4n\l_y7YnͺBY"/sD8l KYOtrQԠ=wQ|Q<:e00&v!)RF%;?]S슏0iC17bIAEL2N6kQ9%>/{}Ddt Ps[7wE;ZS!5ej[XdCK}*4@{uv?5-:`QK)[Z[28p3#Cr*Nz!*VMPѧ2Ry ޘp* *PN.б-CeWC\@Lp-o:Q{"Յz&yJ 8woc tʹ}Hf5gbh dh_xpT)baI3a%2m?Rᭈ2.Rd~*, BȴetQfB%ǿoU¯@6&қepD0-rhx#_#O 8&roE%:ImgCDkwfJ W"TWzu`Zi!4mr.le$8Fb;XQQ`Wg!\.ӭFlYvhr%Y<ԡTKβl\),f!F jLF/1 )F(u?W.FIi_,첚oJkGɽ5`HVHWv90Z˲CƓtlᣠnʵ✎{o%H%710˃sbr37߱Qx)lMmfu1 h0{ukm-$P.=UBe٤@dW)Z49L$I{=#|1YZ2 يC%ҫO"_,DHM` _[xh yێF6eOB^7~Ǵ DZR*DAI--*[WSgMȖ*vı/+J}3N ,V{Se@O}wB )R}(fTCЉÂd}ruytUcJ`>f'{%Y&k~ 줢$ / WHbu}s(sC%4Z[Cѿ&M%Qⶵh*Ign,HVmHRM.+P᩶wDBw{*270"$мA6vHVn 3mh3yyf|s-+ac `xju:v_9q9Fx躞 (p aze!1h X+B7UP?GԾ\۟6s?H"D}ނ#aRws ]fңQD"L)ЌrMU$>rFH}?ZiL۷J [߯}Py5݆u--vW^6& VH|óKUkfgWҎxDR!¤_JrBx9 o )7T_M\B ԽSBK<jQ!A_2S]E+oE&՟GuM,ӈs=.ޡ `ô$ {5߁`71J0BĔz>~{(:r#~ x쪝6l~6Kџ,Wn/5pxmGXөE!`NKKT Ns^4p#w"y#="'yc.kz#A9$s- xgL沔o, SGƵ N;G6u);%QTQs}Dz#t#FP6TORʣߦsrv'^ҫO.xtD&p9!$+H܈vBFx`b04&$8ƴ42~}s@NHvjoiqȚ-&ٻ!ӛeoHf}]x&=IL$BOmٛ5 r/!3~IVx9onhEXkμޚFf2t}2hK/"g)m藵b[w!=3uX=A ݣ.1ljY3' 0&\3QA^{1"TSȁzy|Y ~L=vc,iMIݟ2aoLk"L݅SwoPkAՂ@-ũucS(hHޭrXSSD !T_^2ZtJ>DxT)U;7Ob>Skku1 \ף.j8QA6'ְDeeN{EO?0i_Wqbdtu([9acy%^z{EՔ\"^p#ֆS-9Ј2Y}i/'=ɲyNFf&nC j;|Ons/<+¡ecJ' קN|]nk^cCԣ`֒GYoXt+0>/UԎ6-nyN@7Zi!JFK者$vP$q[moߌIK7:*'aFi@Ԫ`GUO' xa~XCo|fj@u0X 3Fy nTdE!+z"?9&wI][νԬ۵kr_(ex#= WW+l*xuzPi8|N[ <%#Ey_X𽯪\@Zr1ui26kDW ~8Gc>,T췈@ېm>a5w̶T!2t[uYگߔ F?k`מu ~r9}nϦY F9tޝFtmx`"+}`'I颤RA'† 2ݵoti"4*bCכ 69dtzYí[Lpd,͏-# C=ލD‹ <_0@_oA rUOTPީMD؍&P i.͙+o\5q |~N[*b^nٔcy?jTYs-h-|Pd X~[Hzb[O1sÀ_7;YJKN'Hi@$(e@d+6۬b~~8$I)vnh""G/gؕ{8֐Q`X_gxێ$i[yrY>h#C!i(*<}@MP-*S3bkJ}qc1Rʫik UoҤmFs;*]5(sgVF0_WIV.P HU :x<0 udrY OYixe%U>H~jPZ]*DϤ_JXO*PkD5G,Ƭ@$"@LAQۘʏO7]|qmz{v&*'Y>`}HEXg>9 =Ӎf +W9̖RXw߫s̡xv.3&xő![[U]MLI=(E~V힚aucߟiCz*!tnTY,/_Ėu^mվquRtrCڶ\Mr 90Z,(߂/c򱔞?QgoaH{y?;XW z9,,ǥ *1p'\ 臃]WOJ,"Uc .S)\Mj腪wv5DC֏*5;CDu21$r.{|󒁮*yl_;9wɳbz7Hy P;rP:f%awVH E\w3PWF~&"`7 uAu?˝)S|]Mƥ4,35P6o1o_dóT5UF,aFzgt5RMR|`"j܃}'|]/W1$A_4QK|;UfIkJ:3T_jVa}Y| VLri).P,Z ƍ(!-|OӺ^}~®ȈaqZ1|JUz2!@$c_ұcq8 p&h\鏉d/\ĕP k{[RJ+,(FO]}9@djѫð{\̦&!Ιh}͞c2=۰8P0#ی8+|ۺ/»DLWw#=K;B0C;%) 5V8>h-"ࠤٖ^/5њ41¥B1fpQ$zQc \JD3ŕ"7z8 1kJG.Z.T K_^Vr7$uHI|y5;?NbqaIJvR9?pUfqPb.ߦvYL2믓!)eأQ QVT?BQ3wb#ncy^$Vs@0B_ M9#}\&.cUσco0¶< WIDV/Xܯ,g)|$:H0GI+(ZXՌ 16h<̂c)O?W'YP&:uyg0|m~mϨ,@g\)J'ˀ+@ W%ih . ׷yJ;Fspݲ;ֆ ͮ>S(K]%}3?Ap䉭6tC,_jޅ'_: n5re2RCamLlȵhڿwGa&6 `$O46fAvי0kO$zJX-r|yTzD_-^WLbq =7!+b#hӶTL9dGVubL&%X&,> uZupЄ,}<tOp ΗruWbSS!1/:k#O5kBYزg2vLpՌ"Ԫ%rTPH*WX ~vKĿjH5B#Aof%Y'mbdf#Y^ԖLQWy[1iNHVxӌ-zj l*蒫 qm:ZY\hpؼu2Hx%9G7a!nȑ"y25@jR>=m eh@҅._ ^CI4d:{<|L9,yj?4٨Mayj}_4*6ڧ5VKSuB6KYJ_ֆW\e'wX4긻Quz?@Bo1qJW_ߡ>I)ppY蔓n4%N,O&; ~0瘶8-|Efzvn4^LU8(M|80mSp~s -{@Vc̸2 ;sLWJzULC}`cx MZN4šK>wN?cɐY|1wkܔla;iX{l>XI`~%HX:RjBCі6$nbKZvqy1 LHdmQcbckKPQ,V{iJJ@7NNWN5teSYXF UC1eͿP&N4iW* I-iTԃ0/>+ܚ,NF9Kj;Ȏ6x=?񞙑G_:[ QNZ2>uG,slg2oe5Ҧǥlx1]49j,&l~mH8Ĭa{BdWS<}'oP~(rvd2}lOۚǠd8՚n:&[W#(d[XOG!&M5*YYԁH&iKw^D}*PF~ 2reZ: w)g맴!W^$q+eS:-?I:Re 셌γ\Ći~I$JdmOBz+>IL1 +.%tiYX&)V"3 z(Ho:4gIBe#(^!7ou_aX{ ;kȯarw(K B 7j{|pXH.{y|6$+͋jJ#4q0%4߬. ;L Q_U؅DW{E_z3 ) P&%?Nhƨ*Ƒ}5;`Ƹ" ;v5F KaTyIp2JOkF*!9̎ 4_jOtPR6]qhp;ɻc}MHʠCh;ŵpǜHD&!~Qc[T:6Qz#J 5 E@zJ!W&YzƵx,h>؀FHsauoK_mK6J֡9Ux=ERɪ`$=>g帖ͼQ7{( {7*Ս5]&4;%8ob)nje3k9o,3ЭZs2lAK3M}:%U`I=nhDdG>a/:I6c^Gmzuguný\볢>)+dN>-u endstream endobj 408 0 obj << /Type /FontDescriptor /FontName /NLANGG+CMR17 /Flags 4 /FontBBox [-33 -250 945 749] /Ascent 694 /CapHeight 683 /Descent -195 /ItalicAngle 0 /StemV 53 /XHeight 430 /CharSet (/A/D/F/K/T/a/b/c/d/e/f/g/i/l/m/n/o/p/r/t/u/v/x) /FontFile 407 0 R >> endobj 409 0 obj << /Length1 1357 /Length2 5945 /Length3 0 /Length 6875 /Filter /FlateDecode >> stream xڍWTlS@$T1:6QBc66CB@B 4P@N}w9=Ͽ{LHD"B`a,@T @@9#ZP^p$BP0'SqfH@ %eR @!% P]$ETCz`Qpg4.˿^(,##%Cv2B!n3$CcW{ #QΊ|_8` |`_ !ߍ S.p?b3p789x#a(.7LG`C1c k40 Bp8`#M}a4- Bܼ8q.T1@p՝@{ {~u(+ nG5; U:⦎VWG8:jCeQ-s HJZ0P_ͱJ/1@ ,P{A|`4QG8 p9GljaNͣ <0[Ho1UWUH _H`1(_UA8!2M_}޿g,C$0I/]7t?vsZ; Wo4HijCW#?:h*g h["3Fz=(!0:G *4L"KTBA X ܊q7 ?GCG7"$pB(~S\ {~ )Bh{0 A)f&Pmǵ*7|ƀVgPOIjQU^_{4˟gɦ#/)6UE~S\y%32ԕC-Mm\HwV ]쇨eI=ʳQxq`C Z= kyCl[ bK'~^\L}W]K׽>x!c'T߲1j]n,/3!N\bv{){ xޜ"FTQ Mr}ŁŕA EjRi۾b#ȴ-|Z8w6rS"gɱ/$u=٬J]*R쏦YX\ژ}ВJTEO^Nkn'UH~Y?]-V+H?<5+*ɐEɖ I+j"y X~%]y6Amo^^gHxF`^UH!5}R@AUxvV#kR%+HDᆩ(L۾ZHې_X^=Qy5\gBSV{x:zQU+^4Fa VV1a%?OiMv\u3@`%J;x\ZNu^W? KJTwBZ\Π]=s B)E;K3VR&,7S*EW׭<_!\ȥ'{12{{Nq~U;{/ˈ هI1u}.T. ]5͔H9vz/[h"/I:]H";mvX=L^hf\ .\p:V+zmLR1N9Ujp;[),x,w5!stjh\󺘹|mktI:5J;o^"-cJ5yZmߨn>$IOxLRz\G"nr$4B0$o5 !,*h2U\;"p:v17H`duu%K`Pv "sޛ] 6|g @"tQ:糙X  CgL` *n1L)& Iv%[w"֮`vrxL=1#J}fFG3oK4ǠkZyw}|mؘHhcwNE=Ԗq9aA}e#l-}DH1t{fE1SZN6_f9} dzD.HGKͰseQˢ"6f|.˨Kҕ*cWwؗV:y {) jZ\ʎ3]uMA}*`0X=xْ3k۫=ݲ+A_{:29^8l\3fffrzAW]&6*$ V>g]p_tO/*uu0V~jeʷVן!+%e NλpfSI#;TTp6p#uRaގ NbUjx @O?eHiƶܤWM<)(ِJ!|rї#U\}Bfuy#i!X}yv7ъntHtu:_Z__'_ȖfoF%o`I'sS}wJ0&[ 6m--5^?XLJiar@^ R-4,2f;Ӿ;@ʩ8,}Uڒux\n T:>Oms< m.wlrlH)biK k U{^t{jec%QAu+ f0Dտ,&K 0XR滏L߱l,8ڜbpVgWrS8T m I35UX6?*'QPByΔ;[E:W1(Tԗr]^@/ݰ6qV'tk'LYY>+ۓ$ ԕK]vccg J9 Ɋeh&W'4To㍩a}Eeq7P: 3,9Z-RUr&x348uxTk\+ife#^6Y0FVr%" ;aýqם"15KR*o#m1T(d Rtv~ȉPu1WC)͊7LLdFh.b:mwuj$".㒉ğvM(GeŔggcIޡ6o^9|H)M GCS+Ff*dUvV@Ny셼2GKEѮwl '_|"2')*W6XNWÚd1nzn CXK~)vWܸTlOH9{ g'/ۏ)ƅXӍp iNh wWxž{Ծ$bv[3zۤSݟTxE?,z&?hepD촎2ҋu%'}L> T'T98c7SQ`+ħn0vy}7'4!,} S4%=IũS} _j@{p㐋fvkpCِȠ8o[/ Ր=N0Y9;q=MJ$|.{١>5^}l G7c.ۓ}A Cjb6a9}Ssi* KToOŎJp*Vː{kG>(1[cgws7#ԷK/\XbXއZ oj֎ r6(TTT!+s\rzv<pZiRrг|P<:ewm [/`#(=/Ŗo=4cx۞z :dLdNqNqp85zSM!R-Ya\6Ƚ)KM7fA8p(3zgKk޹d̷g5">& rlZ @SZn۲=Mo/Vzcps s{wdL,g.֤G6x~-ܹ&zXav|ՒQ+{C9HD*a{VDMA#CO"-%&X6oLWWPGJf3*Tk@JpN5ϲt;_T,'J5T>|ޛԞn矇 ~#S~ۀv-ѼnۛwD6a:oU So¶D4Ts&}B+<%و1x@Pl@gF!tⶳEk5Q_4+)>#9SLyGXf$q¯ګ,h06#u5JXCf4Dtf[\G&Nsw`7;Mʫ33S!V #7P[u</-LҞuU\Aji_!6td_p:aܞmxV59/yi cLVUMx]ndk1&7A {3aOy`@`7EB\Qܓ9|5jړU)X"W4Ͱ/C9 endstream endobj 410 0 obj << /Type /FontDescriptor /FontName /WIRDCL+CMR6 /Flags 4 /FontBBox [-20 -250 1193 750] /Ascent 694 /CapHeight 683 /Descent -194 /ItalicAngle 0 /StemV 83 /XHeight 431 /CharSet (/one) /FontFile 409 0 R >> endobj 411 0 obj << /Length1 1390 /Length2 6198 /Length3 0 /Length 7148 /Filter /FlateDecode >> stream xڍvTl7%1eJo#;0b(%H )"HHHIHHK(;~L𲉹+B!` %7 h`(cPA QH7"*@`$,GFG %z($ TG0w@!0"//'; `P$* @s + H@}(0p?k\{0 6G8FH,! @c4B$ +W"w0C  $p81 EP7ԅq(PK%tXa%_~!\&UGⰀ_i 0pփ@D nۯ\ K$AGe`<a_-F/5P<F" q?x(- +H? jy "xL@+ , tEIM ċK%e@DR(G8; Wbun(f ھ_;X8Pۃe0CoGUB}A xo"0߮?t5"|۪8tX-D y/y#pACX"<X&r.IY +&H2@<@CWxoAH$ tCa)-_? PUa0FȿIa -^Vr * XOa(2DcT3:>.k L7^R66:4&/zZC-nrbE@ܪ'wΤ0K;b]Ċ3Y}ӊIc.q<"LcLyCz$b݂ݣ'6;v}w#WjzlesEc~%Rۈ&n!R=W%c)W5xL2 M"v gOT0սk=},~l|eNE鼻;JӼ{Uc@Zt/ú@>D wr.#2J4ETibz8v kllEՊyyx7$"Ͼ8%4Au\pnZ^6A$Մ $;A{Lԙ"O+ԳqηU/\<ÖjSzFia#f#LhW.N^Ξ;㻟۲k5+-Q_jw=ՌϴGY$N7: |fxJ3)uה K-Of=,cqq~ϼD$E 9Q%Apՙa Ltk ֭҆:7U^0zD<]VP|Gʟroc[x ; yi}a&ԂݜnDu/;!T\5‚l0-Uv)e}C)7҄˚AVml-$4Aq$y.)]{oh' y|?/=u\?Z[ϽeJ7P˼J2#d{; F?LvO HE;m*T/of)ըgv0>J䦬//  t1PD z~(/j$q|Vѭ4wU1?U w']:ΗP᥎5O=6HJ뇉D6*$޻x}?e~ |=K2 MB?:qimLH'? aM;l1f5zmÂO]Z<^'twNw<x"Z+j>M  pT}θCGa[g)}bcy$n]C,CQ'*Toi:ȿxX@B(N޶]rGB AnN+m6;J*yPjw+cCd:JS?vWŌfrn.|q.u^ JY;_}4+tn[LOXw'/*1MwќlfuVe%#R/*N5OHb)GEFe1z;j{Nl,qg`iɚeYӦF^eBYw[vEU5[W$y)1ac$5*B-6Gm ංf8Q[JGb|OSֳWm*%2q#5G"ܺ׬TUyYA}m[[l\Ǭ5n㬙_uS!hJ:/awëַȀFru⭈FF ī4ZT|$$b@YhoZRV3~C$@t5>avd¶Wn*jG(3b roRũl ̈*VWXu1[ /H} 9!G<5Xʭ}hie!Qk~%3^`_myx\70wiؘ]!oO{7h8'؜#y"ϴe x~{ygY;omΎ]}YrS~h: xʍ=K=&iAkƇ%3{hJ'!g!vOrUW+=;cEqd^XL#ö%*Z2B*/1ϧzu@._=%:p' ,e6@ַܵT˳1S5.j6U}_y?ӆ<14Ȋww4#ND'y2:Q7 6늧bGh(#H%]X:[yvjYvv:n68y(Dz\bLuQq[eS_k{oY&g ^X4&w^Ӆ({NEs_3(nDa?k9jͦdy47=}?}RiPޙ\rbgָcQ%ECʓ)Kt1ر4!g8%b{̊˂ %_Y5?ݎ.FTD?kJ{1w38'}ꥳ~ ζJCDyef[1Nwg._/eR՝,Y,*5bJc#Lv._S؅K~‹g(En2wr_ s#+ݓ eV:E X%~d(Cz݅O8` IfBn/i KD)l)Xn VcY~i}epB;øT^ r2*aO` s689h `egp2XRN<'`cb[Í5oR^k*ya;r-mj .|@_Ok*7>3(tzQ6j&7h j2e#6w" .yHE3cS 9n!eܾww.)/gw _ oL ? wrҨ8{vE*bt Pñ$֔ګ~ QO$NsiuJlxI[m ,Z%}HZnQԂ ?XOM@6e(<7Rf[8Y>JH5|I{ճEmb!ʊ}AkF,Hꆥ>/\GkcjGb[gbE]FuX\ԇxf"U)ǮN؁(K~#xuWB#Iki1“,u[m- {YYV-݌a Ÿ Eqk):Dlw٢x&O\mznFko\kyTͭ)Ըah|&{߃ ؏)3E Z H Qz nj6+vF)j_j˪)v᎗Y;oeM6딴o[2ti8j̨Bt_Ġ尙 {O =$OdV }7,`8NqPw [J,s6lru w#LFu;9]ZtJӛ%]8QVWue ]*75X+ 5rָG˺d#k訒tD*n^+T!}y->@xĺsp=B( bJ=(x*vK?yУtZvb'lߎU 8nmtE gtHq[;BG߻_CԊaRHh pg*hL.3gɲE&R*Obt](ӉӤpJZ\2q)kx܇ˡ!+vFM=?n\柧5}L^ h[Eq%, C=!9q[)&).@n>#bdS7ytϵq`^vw/ %ִ] Drꭱ JSݘ+`vJhr54lDb>/lR,!OK}ws Ȕz$Þ,I׬-=]+v|-,H4˂BNɰY-X*/gڞ(P7r˛~S(9Z}J%ҟ;}XN(#Ʃ"PaZ)pMhv5|WA2c7_FS eV5ot́QjQrپOJKXWU*64N^ dL 6NR"5ֺ=#?LLGTyωڸߏ)e)&n`UX nEOۿ6?Ԉip|1e[ZEe绷)qpocR3kn4S $ӹƑjVtglDz>=>gB,l^YA(k25.SwUn wc͛fHAd$Ae@>^);#-dϊp?ڱE_ e >"RMn{q*nF t<ή|ݵLeO0K+XL Gx%KSCswZt+)AseXx0~LsuW>a RޕclSI$4/_I|oЕ07|`o CJY~zq9Zma [!RyPC{[V PDIwCu%Q^wBe" ^[k6naP iDx$m DqzFa;WysȫCR4id0 0zwYOS9𪸳q*Z׷zb5Ibt3pj9۩ 4-|V?"WM^advO4gGS|=f魗wH֋lϯ KzRODSxKXG>NnySHٲI5k"2T 3A]G )n1h?njgᦼ+ Z@ ]s~Z,Pz00%yjF9mіS\r3̣BϾ괐5>6n6 4}Zҕ)V!S/I0d:3%Ph9afVw̨t{ ?U9Z)xDkQsa GW,F endstream endobj 412 0 obj << /Type /FontDescriptor /FontName /TMNHJG+CMR7 /Flags 4 /FontBBox [-27 -250 1122 750] /Ascent 694 /CapHeight 683 /Descent -194 /ItalicAngle 0 /StemV 79 /XHeight 431 /CharSet (/one/plus/two) /FontFile 411 0 R >> endobj 413 0 obj << /Length1 1511 /Length2 7485 /Length3 0 /Length 8503 /Filter /FlateDecode >> stream xڍTk6Lw Ctw4 1-]ҍ ! 7yyk}ߚ, :ܲPk5D||<||8,,`3b@Vp`GiB!5gPc<Mrap-9m8@11Y lchZ@.mP0p8/;f ;@ '\ wa<8,Xjl#< q7@_U 1ci@ lecuq!;3s ݡ+O+5wV%Y]s]<`_ xdE<O  ^݇w[ P/ߟbk[W'HU/ fDĀm++7? C ہ8~V  oſ% ` A`?0< 0Cd-j)iip.o- D"CXʂ_U 'Y+'aϿ`pK `, >gvc(o>Jο /=p5 SПqق=\W B̀,G+A:`0g0uZ(n SeX>VC+!6P_/$ |p-FHB? b mA޿ @'?%!? ?zDDvb^ɈyA7"@l[x0!/?1E _%x`MI{G]@ o +,}mnhImHқn8I?QU@f0 S%C@ҫWٻby4|~[wFXb;I쳐Y|9JF 87#coٳjI8q~&?|W ݻL(PHFXv_=X++^Qmw}.a{?YY&٩(yQTcbZT]J1s?yivNk;XOg K}IIohx7puzE # WatXHI%Ge,->ë3tzIc`,NRzp!0(_5BFMX[Z=I%Ą(/)>quspH|vA*,Zql'KXq&=i])D0<=}Fx-#$t|:MK2e㎅\%QPЉDXn3U6[v"[%?ƨ  w_{dNY(IDp^/[5ѮiyIi7jP_lL{4y% mv:Qh UD}CudpوVIꮰϒX(.khqe_^&&VhLɾG_c)19XM۸؜*Q4Im0*P(z`3%I. mRa a0)ilX?"; 7H &8io]s}ЪLxzhP\v?tLPGa7)ijٳ#Go"g9Z%d4IG ֪;P|TΛEF N_cL='oSI'MG7|8b >|~fXl3!ބWA V򤅨u.Af1یpD*"#=I-85T3{,֙,m(={ m/Ghr'8CGNSJw;-͇| ox8Ş&^,/m66ߍ͔S?5p/\7G}I,b%<^cVi˼Zl]Ɯ%Zp92/(r5=9tvFgp;#kFҶk5Fmz5Y~[rˑE]9n TR.Yl`|]{ zmѺ(pӑ^yX oVTX;9(f_KqIZ18̞O{i1l@LYp˩*^ͱ30h%ko@fSSK_&y*df?̢^"2Ә|2R &)S]z̟5殮̓↾@XΒ 0 Ls%syEmo+G%TFCs1t1K{/>{`DŽ>ږ(I/Yo 9V_Nq:~{axc}_M"bpl[H)Ȱg֒;ExΚp KѤpLPV)a_vn ;_5iDa{8oWZknvDcj%|IY |X?r-85=.LA\{e8Wa7lG} Kk~wwhz/I E= 123 ̏Ϲ c}˚IِpŒsx)XQ].ʆKJ$ 8śCF8]Y3ɷϵۀo \=z{1cO3r9ny''QtHmRȯ4i"|k|fcFa%#I[,<̯ؤ?a:(.w1JjI`IYUcˆ,n s3v@6TXE,rK)+Knv>7g=䛮? B Qß,l%O' UNtB ';ijGd#\ˎf%G'+#~ibװ ۢ]Bn{N  IJ@x"2&Oz;xemZM"v?Кk?̿A:Ҩ^b0 !o\Q=D}Ħw%K,1WB6< ^$RJO'uk܋h"eh,G1~AK0q1sњ>M=YXbj&SXrPTփsv7݁H随OVv L6^S`Xr*21NsS<|8+jˁ*V_5QYC϶qZךq/8A0F\Bՙ?ŞsEIݢ*>}'NM-x5br x:tLp+p-Mpo9 Q\)áH"$@ߖRK8Mn~s}m}d퉬o@9ӑڨ)Y l*+K'bYhϲv>I.PEů nTMH^^H D'%-qOM 0L&: ݊XqTaa7@JG@(@v^˒ qdDFEMn `I. ?MJOE 9ՓfRn\ TEӲb)찜&#-;XX[l?Xƨ$ jwF?ܩz;TֵƤ=K/zmHF,ZE:[xvjQj`KB9~hKzF1ZpdLB0nҔzkB:3efۃDlD +5ū*0)`/./Jc]cI^QR\̓焋L ߡ kV+!5}g#їW^V63CpJ܍f؛0Ǐّl@z܄q<$04lj']^M?F<6>?48:J*Cn6ff1a&%n1uwB tRk=}]11V{2H}̯GK}w=RFh1%O:bEqhW"^16=nV-M<yʀ<>SErOQIs; X VpMˮgXc])\8VKdiX3PYS1i6ٯVh#FGgwedZ=GYӎ/Ge$bYïa(6O:.Ҧ`zI7ۍBcsOPNPˀiSo3+IwNkdA gfa`-62aǯ_/4B5x}Ǔ-EGT%D;j4lSo~OGnZiX'bP<(tD%sjt,sJEiR_xٝb^<޸[ ]RUaKucvTP֖C~^&d`*f)k&Cظoe*#f%miFLm!=|.^ՑL~oiN~(X-EWCKbױocRSI YbD-6z+ MפH{fdoҥ#^?oZK.8fy/ r0mC&acB)Pڋ6l{r*m^"UN G'>YjoD"~/p'iR0!ٟt${ OcwԽ",WpO/8MhU=Z^ܞK9'/܂ZגLv,w4|:Oݛ8'1O+X7v{`nۯjsB:vl _Pmc^ AvfE bz>o#.*8Væydm姜h>!kᨒ+!0h7UZbBefNG ZO) ؎z<r&16L?Y]QjF_.,!IyTUL|+fCH >RGKri?>mb3I$$5E9yݴ7~܅!Qy_hKwYb1WRgl4/GȚ7N,TwtF>ჺhT΍6ulf=7x&$kTOJg^oɝ [gOЖCqi=,'>c8XwT2c ZpؼK)Zㄥ3Nc W! J=}0^zNA?|r{>6U\6U=ɏh,4'm!&#)}hf%#n[yn^;jn(Ty7+Xq!ś26x%LO.=~2A&IkHn=]& (^ Aug *w߾v،X_Z֟dE3?-U.\ \앝kRiun8e*m:ų2Cg#h Ջd1m Zvay5[A|L)gbkoFgI>)~r_c*V˖ŪtdbC0ű /6%:Dވ;ہ_ڐgӶ5P;!uex)Cu($nq){C"[rXgzD┆ M4oTqu*c\3PokK3ԁC\g7;@t`L@_޺c]/o!AY.CQf|(K|6?d<P i٨T($/k'{jm~(UJ[cu)N19 FvH&nE-h/ J(-jek:>%OW$ПYDGT<$!lu5CaL%.([t}Q}?[) w/Ǜ[F~i̕<)b֙k$zpL}Ќ'QS (r kAUvswj|+E^8AjO}|/ t@őŗdrw]x455ӇLˉ0Xj@;dXsO· Uۖ7՗;ѓ2 _>kxz4ZT@\bǏ}3t[~"9RƑRCUdxEpAqi gi}jecn˒ =&ò;IL~LrMJ7I@[C 2vzstdVqḯ:g'jE*w)&-9Ѕ{hO"f /S Jm(C\GEmJ;\csH=Ϥ, &p'‹ ~p&SiQwv8d@Rן/O.=ڟ9M~'-Q̴+`:RaI?f4Ƭ[P]h\g4ILiKxݍTd(E&S:QCJ\[F`pㆷ}fFF+M1N8drPH91^twk.1P><;~YO:HﶖtOSw%{Ǭ OQLϾ]e^:":NPbU+QѻSHGglbW-UFIIy*Bi4&x+sykG^= a75a]o׃C;p< x{[HucF=,MK v ՕW .OB}`/ |sXeZ4D#E@tZk鹢pwfda>pAҎ'MRe[Ikc&Z$XkŪemOSGhW ;Sij_9|Q a Lj a1IJ|F_ruqT=+߸Ś6lfL'0$9fD'&Z7DTɥɜ9'VkCJj䆀ZI7~IrA&huhr_|yzӵ6: endstream endobj 414 0 obj << /Type /FontDescriptor /FontName /NFNMIT+CMR8 /Flags 4 /FontBBox [-36 -250 1070 750] /Ascent 694 /CapHeight 683 /Descent -194 /ItalicAngle 0 /StemV 76 /XHeight 431 /CharSet (/eight/five/four/nine/one/seven/six/three/two/zero) /FontFile 413 0 R >> endobj 415 0 obj << /Length1 1489 /Length2 6988 /Length3 0 /Length 8001 /Filter /FlateDecode >> stream xڍxPk/AtHoBH RC@HB*JGTM(M@;I}I&Lv[ (j A|P$ E%B bDl(x"A~~7 'xCY>nPG$ꜿ`N(ow lԉ`B> !DJ^^^|6.|7NЇC7`#P66P- ;uU_ݠHw>w(W_aPmV)"\\ p;n0> Gxp;_ey'u0(?: /."",<@_B~Q5"\2 P{@y@ Fl!P8?Qj577E?,P Ca>gW1*( ~^q~@TT86п_p{@O>_kA8@17ovW';#췝۸@a>!P@@ CM VW bpo: py*Po. vÚ?z_!w_6ԒQ;M\e8akE6nn6>Y$aj+ ޿ H Uck/I$&G@T|ןmQM]p0+j;m/jn,2&ȿ7 /|@%Ü^u\y~^[N2F=sQ~`mz$Sw\v_t<?J.H2]fL,a$ ̳< Ս&nc q4|6WV gN4zkD^Eѐ>dIH{Ʀ |XLSa-ch=y蠭|WhG=Bt~}\o@BꖔhGL;/5 XA/&%kVkQZvi4d=I}e, h. BAD:box- e%UfrZyE?Ҕ1~c'b:8ɴ3o?Y]=1Ncmř8_ޥ*HN|~9D\hH:gg&.4CBDzJdtHtXӥ;Jz$uwzaY&ٺq-5 9Zj?D %Aw̦E?(4 "+{>Gם/Q`iӁV/F; $z3yɞvXUb9Vٔ(ܦ2@md>`qނ3jt@RRh+rQ돊' J{WZAl~QƁJ_L6~97od5)Yb[bH뿫__!k|8 (Q Z4y#` RV+vZYD iYhক`Wt?|*i}vwmFzӒliY0F[ǁ *'%&#aWܰPn <#K}0jWU.[$a:C Lw^ϼCUrD2{$kCAklz%:a U3ӬDxa/ ;;󋜴$Ur\>*PHB=F`LA,N+=fFԵ4JFeϊ.B9vj}6( E&3!s,lqN-H`[laaoI9Ll׫>/L֯Sw0J&_{XD f(H.y— 7q_e'Mt0O]Yx f2c@ܔm]CK 3g*Cо8;έk Te5>mScq+;C~tj,f9$DyXb֋',1t\/&]\v,=9 LC@٨*I44 Hu9ț8ZɵUUef {1eďl7JvLP~0[l^ zM7LE2^蚶HHlBrm}dL~p= 7:|VZN$n\0}\>\79RXb_҅Ip&)[eO?3X}^َ@wݭaN|pov:͓Qa/orb2R\ !%gE`"蹎2&fc N5FGdNm|xTWʲuY_tf/I1CjM%v vHpjۣS􊧰٤8IrQGqN.ʧ8jpj/7.<:_ă[?kؑFڛ9F鞥ƼX]r` Qw\߄짌b\^OK~b爟2$l}J§vpy+T)(L8"u KFMc}@=JPQR?o-H79d냋燑ں[}@P^Scq6$'iX Y@3?/+baegzg ٻ[e;8!f\{:%Ԉ2i@={3"X,5O^yiG  nlИs s :V!LThxDIDq`O9b7+xf=x}Ǒwb!tPCy o)^CҮi!w:r"ӕW86kY*HJ:lkޭ; Un_׾#t4`Zc\쾝Ib ڟ?ʥv6{8_Rz֐k'MM )-Iw8ܝ~~8͕<+"κYōw x5 _*}i5i>r2Kth͏}(6Nѕ.sxDweǴYDL#jGG #:oR̺gIYmI\5fjj~-^fS94DW4u-< &=.2)"-nz^4=$CΧTÂ8EkU7 #>Q86ܧ+[b5Z8k“"QŸE u|2m _܋Wx]TYm)۸?Cv:^ `$> CbX'{I([A#!eO1\\aU`悙u@ZO#cѻSh3>1q; `@nnY!(6i$O]eGB6m`3 rTt1/X#*lݹ}Ĵ=gg*>']bW̹oRbō/~N8TK)O}R c4X`C~Zgbyf+N+U ʗ;8?]xq "f}j_Pjj,cy|".arZ!{ȒyOЈǴ@ͩS-ca~qngO&2ɫs2Щ"ןas!s{aQgXqRTy׸{KhnnEc9m}$2I³:hk^"Cja鲭e/Om`,W:k[a^V1LfOS篳Wte9luVu?U9ތvr`}o5%˯eL]J1CFU̅ry=K)EJ,/]%*-XlOöSe"$1 <^1.?`[| 0f:ůUb#=g%<)=X6cNxa˽V'bV]tS"IH\9/ q*,AUx:5x~dز}aE\_YL@Ւp5G_4LqԖV︿RUeK(؊Øؼێ#I+\3͜xy؟If !lu5l6_Nbk·$e= vpqϽ&7:n ns*`uY>ގG9k4a@<gz}G@WՅ7ԂYʐ+ZG촸u_>F齦=?5+p$xҠmSEW!cA!?n۝3^z w\Njg0nr#$3>(c T2*sJzi'EĬe AhG,9.xv]q,i7;yw$6V8xE y,=_Y# NNc:yy="%ff֮IJkVZ> R^ aѹy^@F~h9-ږeM.,OFHSJۀ'Vvh/-uhPzיl/23e-3Ԗ(!qC['B߇@ %n~>o!["̟iy)dxx]dUnʋ  8+2*UU0Q)GL~#?u} [o9~]%R S1RW[Xvi912b\TdZ 1&:?z_Pu"9; jh$#>u6^zLv{/WDҀ6|OfDzf~%x5@[.qBApMHo#;{^̱< BB[=,DُdoG :2^ɖǟ9} a[6n@h5  }}s5ϐ(8Vo  zѼ3ڥ -bC2::(Dpg6EN'ٷR3w!=WLX~@کK%-ݏڢoFggR7;(9rs1uYYP' dLDQ- )ͩMo2qsP؍o)PC\co.c֣tӔaŞ |-H;^y'*B: 8g k] _WkL.BB\= NJ,ɋ}:ҢwBj+JTo+@XԄX7'OLbn&3\3 e*^RwB*N>A?ˬ)t' j|/[5Bnp1C{6d@͓a$*gv3NWgDbIq endstream endobj 178 0 obj /QUBPBM+CMSY10 endobj 163 0 obj << /Type /FontDescriptor /FontName /QUBPBM+CMSY10 /Flags 4 /FontBBox [-29 -960 1116 775] /Ascent 775 /CapHeight 775 /Descent -960 /ItalicAngle -14 /StemV 40 /CharSet (/H/O/X/arrowright/bullet/minus/periodcentered) /FontFile 415 0 R >> endobj 416 0 obj << /Length1 2139 /Length2 17083 /Length3 0 /Length 18378 /Filter /FlateDecode >> stream xڌPZրݵCpw`ơq ,̙߫j[^k7{u&Q3{33+@\QCʎHErGH ta!4v~I;*\ll6n>|vVV;$]AfEfH%nt~ZS:/DmN Sc;%- @t|,,nnnƶ`f{' !:F\fZ(3"@[nof l@@;𛋋 .Pvm#߇`cfo{d/gcSS{[c;d(K)0;;324ۿlM U1@JT`:`_=%mmv`Ŀ9M݃ߗkmgf2ٙՆG; PV6o"?2 3tM-YJ/[>^6> s?D/+ `2u-@v;XƏ M]1G_; `cc|xqTAv޿};7]ߔuoER.66m7xmr]߶@mU\lVmD,&o9,rM-7P ybe?%3~{Eo/m7_ 0vr2@|7xm_ `aw~s0wBb,?`CV!N( -zˠ_yˠ2 -Kom6oR?3yH89M, ;OAԿI,o-d{KK?,d[mo'!?l~~+hGo+[?]כ96)½=,ΖN?Ž?v [ܸM<X-ThvDr<&g<11H$+zpz{UQO5e|J91Yj><;Eq wK1#*9U-[9ݕp&ۄ(ҤG?|ɚ!'!\z(J&)Uӷퟩf.CaHs¢ccRjkET_,_>1} 'ӇԮpm HqV,Q8 Mݶ(Rwv}[Cǫ=/nNn*Yɲ4h)Dtw!$|?F60U M׬qڥk՛9^$hAS1"/#<|Usgl(Ց7))3?bt˞/7zaTqvZ\.Skf9Z-I0mdא&?a4Q!k'?)_,E٧7z>KWoS&OZ4 Fk~YfI@A 1> "mIM#{GǨ"1&Kr=I;'9j^wi <+^$eu?st "c9R -,T-~9+m$7R体?/jI*CJ{yKSub9e2%5ȁVCw`xp;[\~)f6`3Əkja[.n>}^QG,a댁&a m.٤%,eA;3,.Q Br2"؟Eb`D s')1<̓i^vf#aY5I˕~GRfwvNr|@624BרUK6qŀ\XOD?dM;$~cI|>譼,Ͱڋ69LRc n(~j_Z @nCi3M1dOƃr 2_]vp*'#Yth,踰*-2t❤['4zx+g-ߙZx A.P۫R9ײXmԗBǑ]j\0M..Y#+55Sϥ_[g9g{j$jF2=8|Vxl|3pV@KLCݐ09r̎Pm߻E,%PL\dԩKaJvc4Ks߂ ,2*:5!UEpd8)ÒLd(DW/z%jx8fBuǢxCU`X4gͪ7v1@kPȞ^;^`Y7t3jTtc]V`|.bލrgiaQpP }e+|jP3TI_^r|x5´pC)^>CjՄyOm)ڿI#@wɿ> =~^jx!HeF1fSfqɔRG1Y!jUIJ>Sd]$셥,;E1_kP]MH8uSef\*taYYR%vMCxNl˻мm10bV߭IeW@شCZI،Lf{qnKē1%G8PzXȚ~ }o@}P9yti_d{ᔇ>3=6 0;uoBSՙ})[Cx9ܪD.jP d0sGÅyb3vGTlz,À6;/JMx~:6Sn3oD6N gTq9E"GlKaYh¸2!Xc~YÅh,rf2#v=!3v:3$|O;CIb1zC7քV*P{՘w*N|ã[8Qy)31!QŃˣ5N"uN\am6^fM!Q$y|BKYoMwV:doBj<@#\ju3NjW(E5MN;z37Y(^Wk]QOeeCc-UĪApuP!A$8ȭ[i6GjL1f(-UbiwK <NJ Z9t z 49>ڼ=Kx0[|VvqUa^jh}c& jB< ;] 5'a&%'%5}E yc[s||55$d&3ݖwFQ0p˩'BgUTk;?m՜ ׷*DCh Q6_ X#F1Of,Īg nO%o zGTG{:mGP# Jj<g]dAr_-;vyP#if]g!Y)#I^ ѓeګ$X\[m΁ê#eNjS֦c7'^ijM Б]E-)h TT LR=E= ܇`Z>GØ  լ!nj+fqy:@*ڞ&ւ^ZB*@ Y7B20ayK.8yX!&2y9o S=p# k΢?WbMߡ\LAX lT+pCADl`suߺw?2}Ҩɢu;6hVMaCfѴ[GʝmRS> A|xx;iy*|8:Zl8,CdOεt~:i)-Yj;ۆ~:s/H~?>It^ӵt^4Y8Vc}6h t-QtKN1ANεV,6[K ;aNi. W$K\deN^an K&#{].|XiO񥘈kJ0:B+NpGwbA@+IA0t{2l&KO^amTl˿`.91zvsᎂ +ԦkS< GC8=-?bUlo 5-c1J]z>+z|aY #}un=e9x -==CjY;b  9r;ug\|Nګ:E9cZUppD frZU/ 3 W^V [oUKhBDgk'pya>uښG)|9覂1Qt+nR(SIcmSߡq 6ja|PCm|<pULH z^H9南/O7Sn2+ډ5rtV J%0~2M9&wƭqyQ{Ƙzƥ1.ud $b>n^y+hTǬXKaJh|C@rճ(A;7A"+/YW+fR'ST=/ctjzZ+})g(fALA1K:>]V}g"[N"Izbnzx|6RK3JFg|(v#>ѹ[읿/y[YPf),cjA_XHxM|~q< }FN hS!osYFiŊ@fp5+9>vd>sV44ܭ}y^'I)sh.d?lb{c$T3)NI/6ݹ(Rȿ!~b">%*7פ{4އe?:ŭw^[ZGYFMLr0&N;,.(i]J2iXTHD0Ds#coan >R<1F)@\"H/<1U0X 5hL??i|ܥp}a@=#uNPb".vCM*= [=>SBc}*fvK'⪥ 4 >>Rw6SfFsGfr`?@8c /J[ o[S 8)gë&#.e!a7ߴ +tg1-u1U(zJڏͭSl\Uf='Pc=A/<-k鋀,ў,xv<* P' `]Tx'Qru'SVa j}#QAGbZ-tlN8{GUwPSkWtߺұ=mH!hM=XMb#1`"iތ2pHB2MΙ1?쑅^]{Ay P)Pfy7?$l.hUChګUD2z~ɲozPG i>Ufm;(T]t1I6}C{qLL||nGFvYX;1YYUr(R}r\+R+uƋ`F./T $/t3N:2lmiz~FY]CıSdP6߁ hV2wB:}T ;w\{;LqpSfLM~c:jJV@2 66i7 ogaQ0px̋;3 v >zj*v/ʸ3uJ$㫖sήkQ?e!=Bp3X,L:x2:?򐸿N.7{wj7żh?Ez"&IJ{i4'UPOS_ M)hh P'TvR&ƅj!2Y-{Cf}>؃ (J8cڣ{[RUVT5. SHs2IӁN5$UV +U t& SpWأ8'RbIL 5#܇%SH2G5}YzJ 7ʧ+-Zh} hnO}?i|60ncw: X^נ(pg'O!X^ ȥ8Jx2ҚɯLEE|[!mep&2Xy/<uPۤQKU(ڎy[g7w/fFc_{" C)Ѣ(ឬrřmG(VwT~!m"pq__`WT)PZO[q"ɮhgYѷ1AZq]@3ܬܞC[z0-V+'(iЖVpP^|T0ٽ%.@ʲA_RM7bK }pbn >EGeܣ̎وᤋQ۬k3¸rYDnK;f`¦p^X `Zuc N8`:XAAqn D| 1yn dlIoBXB^%;eԬZ0J%g֟㕶x0.ͽ2˅Y P0RUv1{WQLTj P_U7ZWJvVtSS5= gE,ᛄŅ5ݍ FmW)-C5u@}s~elϾ?b(JrVOkˍ#)<)_5T|OㄢKOY:t7;殣.$W1zk<_IzP"ƺ+D *;T_Tcț׎-3H|7mmCG _Ssbb,lNgO9 riS"\sm: A<- Ek÷ڇ>YθcMIR3h, ozlMr0:z>n.yd5gBx(6[t6$ l'ZUbA1Ixq i QX?FSٙj!eK0GdSL7Of>qR3MUPjOY }RzGe N4%&LH5"3p=&6iԇűѢx!irG@gSay:`>XSq+!B!Vx*qm#7\@`;;*]-g{Qn|ʐȖį9B@K;RWAF Ү" KPGs{nIJŴ꧄~W25oht ЇliDSIb X$dhj e~rjRÖl/C+bH-#6N#4wrӣ TF[&πEK't 'A LҩZ?[S|]etc~^-+w _C;T ڹ}$A^]h;KC?lygNr9lͅ \6]L$Rxcu&#.#q򚍫qn)\T'Z^Xn!x8v֜U脞N{ř_>FZ*$Øe6 8PNKs'ִk; N"ގ}t-)LxĄe'Hn3׶Y`M~rVRmc|aR>^VVlFbXwxBfP/N4[U \x ^(Kmtu VI2+#G@ܘ,=k6-(TpG0:}Bc҈wa1IZDwG31c}#E-u$1#@i'@ wt%cy64zc i-QjI`HdҢyOW- 4teM"ׄy'I\-B6-_KK#GqgOX= o? j9U0`-C m[p*KrdS3G0PȒ 6aB=kc=tZʉD_u@5Eڮ:izJ?92P+М+4 zT~J"}7Urg]S7O0#v٩.q;sGA~]ŕX>}-j^4 vWwܐ}x7R56{ kD_=E&{t RȔIQ5ǒ,%>#"{d5 &~2:Pzf7M:PGLC$];F}Q%'f٣P̱m?[OF KPŽ*v\eӽK\BsRNOj8̥Ҋ%5%2*Luj)*Wh Z:YdLc~ک |T:b N$\XS(`]nɼS<|Jm+c/K# @,meC ɤ36dA"l sqճ\Q3*py}FL;>ұ$:ae֚̋ Y!.iID/>Py!(RÃK5$ q6Cb8rU=+CDr_b< =;@n~`шId0oo:>>$JɾWJ;Śt%񑼪!9w$j 'yqđҤ W-;偰"KV쯗3DfU7AZxAG@0On_e5m w>"o{ ۴`35bJt$>,83QfRvg@[-w?R^ɚmZg-Jh/f&gVҐEl.Yh7(ږ8(Q^Jy1[gO^RCj¨*HyT}/0==,fM_Wlk w\m63BK"~cSR٤pg 񽙕*z嚅Dbx!E1a=)4>#%ԖsPٯ+㿈}y^6] s[K= ͛(Z*lFʥl\02b(dCg4|ڂ56J2ya%^Ջ?/BuD^1I0^ ׿ K%z8X~XŻ U"L.Bd[^/6WtVp dv;](V<QAH 9z֚ =exI F=,`\^U{d%kUxaJpMr9K[:emK :k׀og$Y<ʻ#ctaʸvg/d'8W5-l)4>NdN3go"kّRܛ~@̀(jZd, ^:(A|/V.A1')qJ`t3z?k :BO@#hNXVxvבg2AtdH(i [%c9;+'<δI’[#O%I pFS}]zR*$@!%w_2_EksqiE(rqX%XqQ}8R+PbK KWםBZQKr+5W\qv}^5}pҖ-x>"Jǽq2cB`mgmizU2EQT!5Z!#M>)i5YD: Ft KC:E7[Lɚ:QU>H = 9R/lo;C.d87fr0J%>p|NvPfps}?tg*G-FM6c$䵙 R ^,ggqW`kyo=d: "T|la/o%Jlkc?eNj`R zΔ"#7 ǫ=}Vٌ%`l|U;?T$*n؊' >P;`vVRvɀZt.8 n++bUYL GL!&̌sS`֢}J9V0Ƭ&xzմV~ 0.uje3xi>%&+SŪ!SWܧiC$0cop]6X.z"Š0[d_'K [hK=agߡMcJA.M:FjWQNTۘX|[]M>oc6Cs.KK]Dǃ\aM̹@;qY}tfju7 5Ywl 덶sRMT}]t],%.yzL#J¾[޲OLWK]fy(p=Z;-k@)_2َ?& ]n"d$Yh_xbl9ѐ*qqydf4Py[$R˫y­ J ;ܾ4/mZGC,R̝a^3VMGɯ 8{C\ImsU!lpsr?{OǞBG=*` FtDtUW:0;!w!6etI\Vƙ_i 5=Wر^d5߈`@O!VqgI$n"p2JTURQpJWRhl=l3EVAAUمfhmiC%+5vtE_0eç_6-˙MsE}r endstream endobj 417 0 obj << /Type /FontDescriptor /FontName /NTHGWU+CMTI10 /Flags 4 /FontBBox [-35 -250 1124 750] /Ascent 694 /CapHeight 683 /Descent -194 /ItalicAngle -14 /StemV 68 /XHeight 431 /CharSet (/A/C/D/E/F/I/J/L/M/N/O/P/S/T/W/a/ampersand/b/c/colon/comma/d/e/f/five/g/h/hyphen/i/k/l/m/n/o/one/p/parenleft/parenright/r/s/seven/six/t/three/two/u/v/w/y/zero) /FontFile 416 0 R >> endobj 418 0 obj << /Length1 1612 /Length2 16558 /Length3 0 /Length 17391 /Filter /FlateDecode >> stream xڬce]%ve%+uҶmgVfqvm۶m۶mw=۷o8ckƌއ@NFXځ^ZF#௑Dh`fc-` P56 p$!oS*Z N{ҿNƖ6V!7*L&f!Y9u_br_1ckc%@ mfhlmoL0,XS=_,{>6cCc\[cg=7PolfֆFk7![`r6@3[߬r¢`On{nH#CJ/_=\#3{[K}׿E2,;Y'^_m897o3k8E@oj?3C񗄾+ߔ;iDoEyo !0ǖhI2#Pq hdo=@beIu ץx%yϺA{1uকPܐܱ52ӻ{ @46QNƤm^X<.VӰѕM1OW!N%@ G÷M `\¦@cJINHNjoki/&E5K|+?˜1)HfM&-QdžP(8XJ;m^zN~B0 1d躌[]:c\Ǝ=-VNb;k~3V~:;^C/kƘxeS€1Isl.;&{Cd~)N@ZU CnpXW/+[6kP~L 'ѼW^@zY튴,o #?!)|b߮cGi)I[WGJZH2^1r~TN&uM$#mz-YL lDT`ϼALj( zÊȣٱgmL10BLb{ 2:k&߀_^X;F\mWU=DHitU06NKP\;h _rlI\iß%-nBʧѥȎ}m1x=8 NF(˂=0Rlnph>=~ަ ! pmϨ 2Vgiuܴ.Nۡ2 n_ȇf<# v+.:y5k~1VPSu^b&wxNѸK9,sؐXrB=%&3$;Qϡj&䈩&-3GᘨWS,!IA+ˍ'˹~l=[la6p0"↪A<4m],Ng. j^6O1>.q_ FUd\9dx3^/4O bϯ,:XYYQ AF$$,HMV6"$IJah@| Sy. I;8SQ#)>Aܼ9tp;plJh(&*Pq ɄwƱ9l&|uȾ K:EKT!Z/і]%{lMr}՞|pH z5e 8zJ|9x"{^^ >ɱ۳NlQ7g(S}AJ߈4@(C8{L 7`Q4Fi7FczDגZ8+ʑ0 'PBե\}Fo7;@*9]}kJq"rسh^Rfơ=d1KlUj^w7%ࢰ ߃bAsӰ{lv^xn#.q ỀE> ZmglN+R1(f8|EzߤW)o2qÀ>stJUOD3kY"u+0"ἿBӆgwDg*/P F@)ђFCab"лrM?Wyș*.(CCQ0q& -~s&8Eq\,Kx}O)\RZ1I`KŅE/U\Daɛ={a QcFEپ4/?cax_;xrnj%Xiώ(Ma+LRvpٳv*-}B4(HD}Pyh9ҌBȫ@ G>|#>QgD,w\x okdˍFseEF\z _E?1rϸR`(X,(:zC=Yej3ΈYl;פ %>'YVuvg`q[B㹜.<|է;h^F^N<'5!{Ff>T'Ho$\os$Vq!zC•20/|eZX-'WXg#C\xAm\TҴep,&:|f8"J`+<]z,S#3w P !3GB"ƒ-V~|oJ,\#95I[1aH$(V8x'7-nǃcm063֑A66 H9.@=įN"i,]YS} ~%4+x3t \9\Dx\Ưȇo aq O%jymьzyܸڟLPa _ToBfkpiW~Œ#Xd!=P@mĠ,j&RG½ww-se~{ y_8*W7 h8e`ʧ\)#h,-z3X-;7lLf*Hm!ujO`φF.II:nw`}~"mOko+;2$p}:^lߝRg&D?%Ъ7y ;6J4Ų60єo8D{;!_fx󝂇˞Qi$(#ra)]Cz lA6bm-'ȈL&Z#:.q~zzkzle}+hwǶ"F>ݔ\Zݞh^D7uE߲9!k rm~5lx.kgW?S:_ҷC[}syiuJ4[ݲBE>s-ud֡?9,tT\'lK?WM6~Cwӫ4g#p&MO@/:vH|B8G:J.(cRuioפXfZ2ڰlY621wy(Q#"j)p/n/}xԧ 3n t|\3S#S[7JO/&E>s,cM?-mnl>/)ˬ4x{eV=]~T|23PO!D?K($/ P"x˷p d&>gb8\4O;@y@SBiZWd=Ldhn9J G cI2ŚhoԺ8rp $x_/{ƴ%5TeLHl'܊7U<1~Kk~7[=cczJ2-h*P!rK >jT!:'MTt ,ſ :N>5OӦ:Xd&QQO;Es1=E0ɩۍF9ʢs2Ii|i9fuBΆ![YGf B8UT'TݲX'(HD'#+ѫ\qCy@Ge3"͈IhКG!IiB9~pCČp!R`Aj;"H>p ,#K}i '2VOHϚ"m5 ;rUM `]Gc=cutdl-fиӣͨ=bH]sPb5wGbp_ Ws^GL cpG673vr=eH jM-DMy{l*7L9}1RV26ѣ]ދ{$ tw8F'dg I"c#^[E_մgG3t3_އʎ eH]vbQ7u YvhATHЋrs"Cl$aTY0lMϔfz1NaC-U)aKL{reXEpYzH)1hQѓ~ZScwEtNm#{wukynjo.e#gx6+zzUF- .G{,%]C6BoԌ^YC[Ib/4 $bdzߔm5O!'?l.<Ŵh)c]vZ~]E M %uyY&re[pAOLwJN+:O6 âׂ)&p@77=C$Mߑ?SȎ7=I]rbSR}aʘ<Lqv@wqpcǵNim <VJ#j@eT4Sޱ_qpt2?ߎA Mq3k30S]H;ԥO.3[ke9Լ͚hFsvo=N Q#ZRN9?=,f(-]ԟ v>uvIȓ4m%Wۨ6dF|i} I@j}:X/ EۋV? _h*eNٚTD48i5Lo+fǑ5a@.vOBȍQadֻSV`QN%w3߾߯WKu`Agb d`m'ɲv2 94yX  ˗WqtV=wAH}„EލAWN}b+luBsFlt=4^(9J`;ȕaS;#R6R,6;p[:es2[~#/ mf %G(:44~2EF!pDU=l,q54;->-ؑpq i3c~&/ geҘvOl6% l4ü c}N6#8cJ7-Ֆ؞~kYy 60|\6=)d1n?b GWxg jo'KD֮afCxz'sWؘZ7lSثq1CYtV>T9 s\Ij Ý$^3)GDR?4(f5W%r=F*Z!y#YXDm&O8ʦkglϘ0N b C buQa_PFur䐭%b͑@:FkwVz99WY#t2FDe6חv>0~/`#S-JU6=r567ʱaɁ7)~<>q)~;BHJ(Qɂ5nN>Hў.SQz=22 X4N6hW3DK@AHՍu6&u% 2K@*ye0;rQHoC½#^r<*q,ʽ}*2ʭػ[΅(M8a!F'/ Pr}Y[ױ*S&L)"0D\JI4?)3 a@_PYb(Υ*1paݩ`1̀\xV9?K.7HG)8s¸!'2&'̛%7wdm?QZ#rLF&LQOn,5h'np<"Qzc X$%nOG!dl |+WEv&/_E]5Kl~ˀ76]Gp}_TY8v|jAct{tnd8'td^/d8)I*O U(b}3ӽ/̸|- X|Eϼ0h sDNY?Lٻ"򦳨ZWB#[ ƭL ݑ7>BWI) @ϙ}ɩqb,c3yïD0[@kIlEhwp~7 1u*Brq#FzrAKDZc.X CPJ|'!ƶدv1!NJ]XXG2||ﺚHbZFǫSiL>ž:5l7MhlU-Uan"Wg=fTCx#+TEvI,N]/rB}^њa@+>mv\noW/76sW"I&^~16õ)CS6If顨-Α>+~) /g/ :ܬ/uW(hwYHTmՖX^ԯgr񖰍rQ`f zZ;)S~h ̽1/m+d3;{)T@ݶbz.44;wPģto2:tL[ wh*@HvZ9> mԞD/pcq$g"\HUyxD+ `<JkeC'gj{8D_*ꠛێ)P䘔l3;^y*@eޒ&>;!vedzl)w-(9JTɆgެF,yK륀ͺg!p2Dm9E !# 8ʦD"=%2F=F_8):|&y_m5tըT9ď ?hW; Mo-d(*Q]1j[\ ]W *iK6)̒(Rlg% :<뜠璲O{lΩ8jf<樒,ds+Ѷo@hf>X_XAU*'OAH΅@ٲp<c14F4ToH%D/y7S\AR 4MZYqQQK hTuSxQR.nߏyfF OBzdhJ lLxf)[C^G/^p׈b7?;L&ܦ2įQ rPw@`O%p-{)g_~xlT-nh'N"rc(-PW{t}=%{{v,65521m1(!+w〩\R9hGvh\5uk5WۮɟqJG"cRu( N܄בRu \Sy"T`VXdAWfјRz {o=VtW#4ۀI6v5K( 16uAOw%kۙw17>VQUD['`nqc$tLa7envB.My5H*iMr%.avxgpD DDdL\> ٨|N~]!!w+IsHgix1dC©1q 3thq?,8ɒ_o#!݄Vy C#BLr"=k M=K;fۍºMScD5.1bjm4dM"@9~j>Рي=$4 T7YOnܟV3Dy4|\j;AYB'$KeEa{X::f-%qQ~L" 'WJ;y~TtX?(Kƒc4->ɬگA]re2q>hj361 ~?\ݝI n~j}^yH&Y¾]MP d9y|2##.CN8xe8>w"D̃T2j=sqo7p`U]ޕq^5Q`ؙ+{4`DeRp}1r& BÄQJ^Kb$G ׏g٣oEp/&>J鬳fdNbX\i`$>4c7tInI g~3̧aFLLIRЫ,xfuQkJ%~x Cp ? ީB,>Fm0Aa3ea âVC0/etv4y fەT X!vj0gwjyg-To?(Jmp]# | &+K#bdOZ\}O ̘U{ƩV/UBU;.3pkaQȴv@Bn" QMxeZ4c0Vчڎ<3j_6N4ރT㟓͔-:{+'̊,(lxAa>XR)kOIͶߗт~KN9.= xv28J.xeYoBxߩR V Zmj NX ~oJdUF|[qvoa9R075v+gtIKMO"6_77muE#=!&oĘg9@2`t`{žD<+IkлjFD@7Owt% ݫy%2ҁ_>nK#:UǠ;(٤29Nov>6FEZ 4Bt<$̴_H 䝑J6yCYMxlb&*q7:Y<oib:V_/OmIK?G]PEӒL9ͬ[ ? K!@Bڔ}Z^⇴s|3rKK/f@2V&[GJ$B<س(ݵ2BAJ&vMI nz˥pY6vsbGL: %Vl[ħ4 3HǝMƝ_iC_x&lk ~V{a^+Zzun|8ĩ_p$ӭv^hɖlLۋ~h3+DfG-O[Vq4ىEuΦyjT*•% `;.q}-! ֯2?Zh)?˃oY*Tw mOXϻÆNZ˟;uz7jyzhsoBf?`%%4{~1i;6O gZƖ]$GS%ԎfMxGGC!cRu R; Bk1Cѩ~Lv SC]G= ?2 3wA!fmGNL0[Qf=1ȃ2!uF1,ʯ׎``y{wn2e?l]!WrLpθǵwS0!㈸S6˽$JMx@B"B":"@AAE'l;{2[cȼ~WӋ`@) #<(a.\ A2#HZ\f'dHl͑E JhW)P U@zE>T5Y7 qY1p D'Fc)U Lg!ìCNײiIx֔!*o:w4~wD*O1Jђl\ݚ:ȋ8jrgk B>l3n#Kjd2ϰ{{j=|pe-< tE+W {6#}L ۤ׼όA1"ۙe9 z[ f\]V 6sD*/g  2Qd=FNHvlD$YBBNs N*/&dž,g*@bX\wz8Q XMoK ڄ{LWBI>Utv͸{\C3Fy Iva+ȇJқLTB,M|EU3܉R!Ǖ V;1w#q$a)0<&ic A<#z}8dEibb"B*"3:iMAbd>1Js5Eq/AzTz"[/T>S{mlI__WS Jphۼ85. k?'x]ɸdoD塔%0`1ȬէXV %, &eQ/|pRȗh͚)fe17<6@&' X*r.*wD/хRбe/WA-:mHG %a':scA2. $uO_`/vTpwr%D]K0]_ž9*!:t̻`CE>U:T36~6_U-nuxh7$B^Z͂ vجW,)L\@\8sU \>I 8*HDR/?( nU9 *~r"H\ЃU3;ׄ?wwnjV{%!Е.:?bb;'h^ 3GS:i@qYVhdyWV9z<ɿI=(W+DT%Ӭ$Q"L݆_C? w/{-U; KrӈQ;NU>g/u\vC v| eEaQ Nc qG^,o)o 56MKs㑳l DyxcԃZ?z$w~mWY&4!QڑiXI#\y,M# YPS l&'S)c2T44ے4I$t*$ｷN{VS\?}-D>EZNAVd=$>f(%ܰ>հu훍YKK3OeoLr !ʨaڻoF3 P4n.2 lB423Ea9@fĨΤC>o%?0;$eM8R( iO6!ԢGD*V,DIX3,~[/O*#<,nŔ< veۑ ϱ$n4ZxBg艹1n#n#q^ԀQc( {YG??'Dzpg_C*B g1H>{HR|/-Zج1j7~٭cۃ_6~zK\]!7EWүWiKЅ5nQese8$u?zƇW2lP^0Q NzpzO.Z{5mȱN:T!ݭN;V=j:X q'B-ND?pw)Kt$3.]@'klOc*RP7}v%w$@.={=_O4:#o[v_\IqS#_la5 z>dwr5YSxSPFﻛ?>&eGOoI=Qc\? w,b^` ݯ:9UoMG6JVgI'] G Gx :XZWe6s!My^JcnALnAOq}ɽ}W2mWs_x SRhMuډz?[ $W/!\tSu}QdjEC>a} !SPww^a9ߜ5/ŝrZ uByPNI$W8+-|aԚF摋 9w,!ev*N~ A<׉C;э%ty!K1~aV7]UE:9ơ'F$2eb*۶I, J)>W4?L`*1H8U/8/VJap+ٛg%'aF\gTc#5;{ 9 3!_k.';@YGzbAE6r Tpٛ^ΈҚn7y0& ȞO׼ johlu7߂x̔¡ { rP_o g]βh-_tۮ*NQ5dh vCe{ ǽD*$E!7KUw#6VX5==4s_oy^Ȑf.f^k銣ᷜ>rZ@edx޺QTuE[ AI9EZc}corϭ;ut0' ]Nww".YQ}/SxŻV=PbBA>imzfح?98P-a0Xk}A2O zk!:sJBKd?28 @IΚh4 @714E/ ;dHvΔ?+~ 3'5w[:~w8qZ~wc+_סVn{S>4AC) Ԣ}U`ćg\. â'IKsi{ݡڴUPxJYs]7rRÐ_{x(8Bw3J^o7[eA!8y S*N5u#c]hV^lN*j%[-+78|à={F KsV6#Oc<SLnyaMU{K ܲ>tV CtD d[cei腞21RЩK2p@?ϝ'e[S=5ǛYG˜UUr+%[SZqGݻcnU_9бdZ<7۱ dgLۋ+TV90jy%#i{$pIh YfBVkHM{/T2Jpa ^&̫YN1muwƙX7-L;e$ZtnP$N/8C`R WsHSt$?!R+%`ok=~X&tthi=.Ͱ=e(Ts0=Ab\ޫ̪^6cW`m5׀ .bZVcl#_ւ{D"Wbhe* 5:%Ii7NvX*. ]ֻZt_~NL% vU3852qI+Nr ]W8ѝ :eBZ?#4TO endstream endobj 419 0 obj << /Type /FontDescriptor /FontName /GGZGOL+NimbusMonL-Regu /Flags 4 /FontBBox [-12 -237 650 811] /Ascent 625 /CapHeight 557 /Descent -147 /ItalicAngle 0 /StemV 41 /XHeight 426 /CharSet (/A/B/D/E/F/G/H/I/K/L/M/N/O/P/R/S/T/U/X/Y/a/asterisk/at/b/backslash/braceleft/braceright/bracketleft/bracketright/c/circumflex/colon/comma/d/e/eight/equal/f/five/four/g/greater/h/hyphen/i/j/k/l/less/m/n/o/one/p/parenleft/parenright/percent/period/quoteright/r/s/semicolon/six/slash/t/three/tilde/two/u/underscore/v/w/x/y/z/zero) /FontFile 418 0 R >> endobj 391 0 obj << /Type /Encoding /Differences [37/percent 39/quoteright/parenleft/parenright/asterisk 44/comma/hyphen/period/slash/zero/one/two/three/four/five/six 56/eight 58/colon/semicolon/less/equal/greater 64/at/A/B 68/D/E/F/G/H/I 75/K/L/M/N/O/P 82/R/S/T/U 88/X/Y 91/bracketleft/backslash/bracketright 95/underscore 97/a/b/c/d/e/f/g/h/i/j/k/l/m/n/o/p 114/r/s/t/u/v/w/x/y/z/braceleft 125/braceright 136/circumflex 152/tilde] >> endobj 85 0 obj << /Type /Font /Subtype /Type1 /BaseFont /VOGEUD+CMBX10 /FontDescriptor 151 0 R /FirstChar 12 /LastChar 123 /Widths 395 0 R >> endobj 107 0 obj << /Type /Font /Subtype /Type1 /BaseFont /XQYHMN+CMBX12 /FontDescriptor 400 0 R /FirstChar 12 /LastChar 121 /Widths 393 0 R >> endobj 134 0 obj << /Type /Font /Subtype /Type1 /BaseFont /ZWEPFD+CMMI10 /FontDescriptor 160 0 R /FirstChar 59 /LastChar 121 /Widths 389 0 R >> endobj 136 0 obj << /Type /Font /Subtype /Type1 /BaseFont /FRUNXE+CMMI7 /FontDescriptor 154 0 R /FirstChar 110 /LastChar 110 /Widths 388 0 R >> endobj 86 0 obj << /Type /Font /Subtype /Type1 /BaseFont /CADBJG+CMR10 /FontDescriptor 157 0 R /FirstChar 11 /LastChar 127 /Widths 394 0 R >> endobj 77 0 obj << /Type /Font /Subtype /Type1 /BaseFont /XCNSQF+CMR12 /FontDescriptor 406 0 R /FirstChar 44 /LastChar 121 /Widths 396 0 R >> endobj 76 0 obj << /Type /Font /Subtype /Type1 /BaseFont /NLANGG+CMR17 /FontDescriptor 408 0 R /FirstChar 65 /LastChar 120 /Widths 397 0 R >> endobj 367 0 obj << /Type /Font /Subtype /Type1 /BaseFont /WIRDCL+CMR6 /FontDescriptor 410 0 R /FirstChar 49 /LastChar 49 /Widths 385 0 R >> endobj 148 0 obj << /Type /Font /Subtype /Type1 /BaseFont /TMNHJG+CMR7 /FontDescriptor 412 0 R /FirstChar 43 /LastChar 50 /Widths 387 0 R >> endobj 191 0 obj << /Type /Font /Subtype /Type1 /BaseFont /NFNMIT+CMR8 /FontDescriptor 414 0 R /FirstChar 48 /LastChar 57 /Widths 386 0 R >> endobj 123 0 obj << /Type /Font /Subtype /Type1 /BaseFont /QUBPBM+CMSY10 /FontDescriptor 163 0 R /FirstChar 1 /LastChar 79 /Widths 390 0 R >> endobj 375 0 obj << /Type /Font /Subtype /Type1 /BaseFont /NTHGWU+CMTI10 /FontDescriptor 417 0 R /FirstChar 38 /LastChar 121 /Widths 384 0 R >> endobj 121 0 obj << /Type /Font /Subtype /Type1 /BaseFont /GGZGOL+NimbusMonL-Regu /FontDescriptor 419 0 R /FirstChar 37 /LastChar 152 /Widths 392 0 R /Encoding 391 0 R >> endobj 79 0 obj << /Type /Pages /Count 6 /Parent 420 0 R /Kids [69 0 R 83 0 R 104 0 R 110 0 R 117 0 R 131 0 R] >> endobj 149 0 obj << /Type /Pages /Count 6 /Parent 420 0 R /Kids [145 0 R 184 0 R 220 0 R 270 0 R 301 0 R 322 0 R] >> endobj 348 0 obj << /Type /Pages /Count 5 /Parent 420 0 R /Kids [345 0 R 364 0 R 371 0 R 377 0 R 381 0 R] >> endobj 420 0 obj << /Type /Pages /Count 17 /Kids [79 0 R 149 0 R 348 0 R] >> endobj 421 0 obj << /Type /Outlines /First 6 0 R /Last 66 0 R /Count 6 >> endobj 66 0 obj << /Title 67 0 R /A 64 0 R /Parent 421 0 R /Prev 62 0 R >> endobj 62 0 obj << /Title 63 0 R /A 60 0 R /Parent 421 0 R /Prev 58 0 R /Next 66 0 R >> endobj 58 0 obj << /Title 59 0 R /A 56 0 R /Parent 421 0 R /Prev 30 0 R /Next 62 0 R >> endobj 54 0 obj << /Title 55 0 R /A 52 0 R /Parent 30 0 R /Prev 50 0 R >> endobj 50 0 obj << /Title 51 0 R /A 48 0 R /Parent 30 0 R /Prev 46 0 R /Next 54 0 R >> endobj 46 0 obj << /Title 47 0 R /A 44 0 R /Parent 30 0 R /Prev 42 0 R /Next 50 0 R >> endobj 42 0 obj << /Title 43 0 R /A 40 0 R /Parent 30 0 R /Prev 38 0 R /Next 46 0 R >> endobj 38 0 obj << /Title 39 0 R /A 36 0 R /Parent 30 0 R /Prev 34 0 R /Next 42 0 R >> endobj 34 0 obj << /Title 35 0 R /A 32 0 R /Parent 30 0 R /Next 38 0 R >> endobj 30 0 obj << /Title 31 0 R /A 28 0 R /Parent 421 0 R /Prev 14 0 R /Next 58 0 R /First 34 0 R /Last 54 0 R /Count -6 >> endobj 26 0 obj << /Title 27 0 R /A 24 0 R /Parent 14 0 R /Prev 22 0 R >> endobj 22 0 obj << /Title 23 0 R /A 20 0 R /Parent 14 0 R /Prev 18 0 R /Next 26 0 R >> endobj 18 0 obj << /Title 19 0 R /A 16 0 R /Parent 14 0 R /Next 22 0 R >> endobj 14 0 obj << /Title 15 0 R /A 12 0 R /Parent 421 0 R /Prev 6 0 R /Next 30 0 R /First 18 0 R /Last 26 0 R /Count -3 >> endobj 10 0 obj << /Title 11 0 R /A 8 0 R /Parent 6 0 R >> endobj 6 0 obj << /Title 7 0 R /A 4 0 R /Parent 421 0 R /Next 14 0 R /First 10 0 R /Last 10 0 R /Count -1 >> endobj 422 0 obj << /Names [(Doc-Start) 75 0 R (Hfootnote.1) 368 0 R (Item.1) 113 0 R (Item.10) 188 0 R (Item.2) 114 0 R (Item.3) 120 0 R] /Limits [(Doc-Start) (Item.3)] >> endobj 423 0 obj << /Names [(Item.4) 122 0 R (Item.5) 135 0 R (Item.6) 137 0 R (Item.7) 138 0 R (Item.8) 139 0 R (Item.9) 187 0 R] /Limits [(Item.4) (Item.9)] >> endobj 424 0 obj << /Names [(chapter*.1) 106 0 R (chapter*.2) 374 0 R (chapter.1) 5 0 R (chapter.2) 13 0 R (chapter.3) 29 0 R (chapter.4) 57 0 R] /Limits [(chapter*.1) (chapter.4)] >> endobj 425 0 obj << /Names [(chapter.5) 61 0 R (chapter.6) 65 0 R (cite.blum1998line) 140 0 R (cite.chen2012quantized) 330 0 R (cite.engel2004kernel) 325 0 R (cite.fernandez2015probabilistic) 354 0 R] /Limits [(chapter.5) (cite.fernandez2015probabilistic)] >> endobj 426 0 obj << /Names [(cite.gao2013kernel) 351 0 R (cite.kalman1960new) 355 0 R (cite.kivinen2004online) 142 0 R (cite.liu2008kernel) 327 0 R (cite.liu2009extended) 332 0 R (cite.parreira2012stochastic) 181 0 R] /Limits [(cite.gao2013kernel) (cite.parreira2012stochastic)] >> endobj 427 0 obj << /Names [(cite.pokharel2013mixture) 352 0 R (cite.richard2009online) 334 0 R (cite.rzepka2012fixed) 349 0 R (cite.sayed2003fundamentals) 335 0 R (cite.singh2012random) 331 0 R (cite.slavakis2008online) 350 0 R] /Limits [(cite.pokharel2013mixture) (cite.slavakis2008online)] >> endobj 428 0 obj << /Names [(cite.takizawa2013efficient) 337 0 R (cite.vanvaerenbergh2006sliding) 326 0 R (cite.vanvaerenbergh2010fixed) 328 0 R (cite.vanvaerenbergh2012estimation) 333 0 R (cite.vanvaerenbergh2012kernel) 329 0 R (cite.vanvaerenbergh2013comparative) 369 0 R] /Limits [(cite.takizawa2013efficient) (cite.vanvaerenbergh2013comparative)] >> endobj 429 0 obj << /Names [(cite.wilson2014best) 218 0 R (cite.yukawa2012multikernel) 336 0 R (cite.zhao2013fixed) 353 0 R (figure.3.1) 141 0 R (lstlisting.3.-1) 189 0 R (lstlisting.3.-2) 263 0 R] /Limits [(cite.wilson2014best) (lstlisting.3.-2)] >> endobj 430 0 obj << /Names [(lstlisting.3.-3) 265 0 R (lstlisting.3.-4) 267 0 R (lstlisting.3.-5) 273 0 R (lstlisting.3.-6) 275 0 R (lstnumber.-1.1) 190 0 R (lstnumber.-1.10) 200 0 R] /Limits [(lstlisting.3.-3) (lstnumber.-1.10)] >> endobj 431 0 obj << /Names [(lstnumber.-1.11) 201 0 R (lstnumber.-1.12) 202 0 R (lstnumber.-1.13) 203 0 R (lstnumber.-1.14) 204 0 R (lstnumber.-1.15) 205 0 R (lstnumber.-1.16) 206 0 R] /Limits [(lstnumber.-1.11) (lstnumber.-1.16)] >> endobj 432 0 obj << /Names [(lstnumber.-1.17) 207 0 R (lstnumber.-1.18) 208 0 R (lstnumber.-1.19) 209 0 R (lstnumber.-1.2) 192 0 R (lstnumber.-1.20) 210 0 R (lstnumber.-1.21) 211 0 R] /Limits [(lstnumber.-1.17) (lstnumber.-1.21)] >> endobj 433 0 obj << /Names [(lstnumber.-1.22) 212 0 R (lstnumber.-1.23) 213 0 R (lstnumber.-1.24) 214 0 R (lstnumber.-1.25) 215 0 R (lstnumber.-1.26) 216 0 R (lstnumber.-1.27) 217 0 R] /Limits [(lstnumber.-1.22) (lstnumber.-1.27)] >> endobj 434 0 obj << /Names [(lstnumber.-1.28) 223 0 R (lstnumber.-1.29) 224 0 R (lstnumber.-1.3) 193 0 R (lstnumber.-1.30) 225 0 R (lstnumber.-1.31) 226 0 R (lstnumber.-1.32) 227 0 R] /Limits [(lstnumber.-1.28) (lstnumber.-1.32)] >> endobj 435 0 obj << /Names [(lstnumber.-1.33) 228 0 R (lstnumber.-1.34) 229 0 R (lstnumber.-1.35) 230 0 R (lstnumber.-1.36) 231 0 R (lstnumber.-1.37) 232 0 R (lstnumber.-1.38) 233 0 R] /Limits [(lstnumber.-1.33) (lstnumber.-1.38)] >> endobj 436 0 obj << /Names [(lstnumber.-1.39) 234 0 R (lstnumber.-1.4) 194 0 R (lstnumber.-1.40) 235 0 R (lstnumber.-1.41) 236 0 R (lstnumber.-1.42) 237 0 R (lstnumber.-1.43) 238 0 R] /Limits [(lstnumber.-1.39) (lstnumber.-1.43)] >> endobj 437 0 obj << /Names [(lstnumber.-1.44) 239 0 R (lstnumber.-1.45) 240 0 R (lstnumber.-1.46) 241 0 R (lstnumber.-1.47) 242 0 R (lstnumber.-1.48) 243 0 R (lstnumber.-1.49) 244 0 R] /Limits [(lstnumber.-1.44) (lstnumber.-1.49)] >> endobj 438 0 obj << /Names [(lstnumber.-1.5) 195 0 R (lstnumber.-1.50) 245 0 R (lstnumber.-1.51) 246 0 R (lstnumber.-1.52) 247 0 R (lstnumber.-1.53) 248 0 R (lstnumber.-1.54) 249 0 R] /Limits [(lstnumber.-1.5) (lstnumber.-1.54)] >> endobj 439 0 obj << /Names [(lstnumber.-1.55) 250 0 R (lstnumber.-1.56) 251 0 R (lstnumber.-1.57) 252 0 R (lstnumber.-1.58) 253 0 R (lstnumber.-1.59) 254 0 R (lstnumber.-1.6) 196 0 R] /Limits [(lstnumber.-1.55) (lstnumber.-1.6)] >> endobj 440 0 obj << /Names [(lstnumber.-1.60) 255 0 R (lstnumber.-1.61) 256 0 R (lstnumber.-1.62) 257 0 R (lstnumber.-1.63) 258 0 R (lstnumber.-1.64) 259 0 R (lstnumber.-1.65) 260 0 R] /Limits [(lstnumber.-1.60) (lstnumber.-1.65)] >> endobj 441 0 obj << /Names [(lstnumber.-1.66) 261 0 R (lstnumber.-1.67) 262 0 R (lstnumber.-1.7) 197 0 R (lstnumber.-1.8) 198 0 R (lstnumber.-1.9) 199 0 R (lstnumber.-2.1) 264 0 R] /Limits [(lstnumber.-1.66) (lstnumber.-2.1)] >> endobj 442 0 obj << /Names [(lstnumber.-3.1) 266 0 R (lstnumber.-4.1) 268 0 R (lstnumber.-5.1) 274 0 R (lstnumber.-6.1) 276 0 R (lstnumber.-6.10) 285 0 R (lstnumber.-6.11) 286 0 R] /Limits [(lstnumber.-3.1) (lstnumber.-6.11)] >> endobj 443 0 obj << /Names [(lstnumber.-6.12) 287 0 R (lstnumber.-6.13) 288 0 R (lstnumber.-6.14) 289 0 R (lstnumber.-6.15) 290 0 R (lstnumber.-6.16) 291 0 R (lstnumber.-6.17) 292 0 R] /Limits [(lstnumber.-6.12) (lstnumber.-6.17)] >> endobj 444 0 obj << /Names [(lstnumber.-6.18) 293 0 R (lstnumber.-6.19) 294 0 R (lstnumber.-6.2) 277 0 R (lstnumber.-6.20) 295 0 R (lstnumber.-6.21) 296 0 R (lstnumber.-6.22) 297 0 R] /Limits [(lstnumber.-6.18) (lstnumber.-6.22)] >> endobj 445 0 obj << /Names [(lstnumber.-6.23) 298 0 R (lstnumber.-6.24) 299 0 R (lstnumber.-6.3) 278 0 R (lstnumber.-6.4) 279 0 R (lstnumber.-6.5) 280 0 R (lstnumber.-6.6) 281 0 R] /Limits [(lstnumber.-6.23) (lstnumber.-6.6)] >> endobj 446 0 obj << /Names [(lstnumber.-6.7) 282 0 R (lstnumber.-6.8) 283 0 R (lstnumber.-6.9) 284 0 R (page.1) 74 0 R (page.10) 324 0 R (page.11) 347 0 R] /Limits [(lstnumber.-6.7) (page.11)] >> endobj 447 0 obj << /Names [(page.12) 366 0 R (page.13) 373 0 R (page.14) 379 0 R (page.15) 383 0 R (page.2) 112 0 R (page.3) 119 0 R] /Limits [(page.12) (page.3)] >> endobj 448 0 obj << /Names [(page.4) 133 0 R (page.5) 147 0 R (page.6) 186 0 R (page.7) 222 0 R (page.8) 272 0 R (page.9) 303 0 R] /Limits [(page.4) (page.9)] >> endobj 449 0 obj << /Names [(section.1.1) 9 0 R (section.2.1) 17 0 R (section.2.2) 21 0 R (section.2.3) 25 0 R (section.3.1) 33 0 R (section.3.2) 37 0 R] /Limits [(section.1.1) (section.3.2)] >> endobj 450 0 obj << /Names [(section.3.3) 41 0 R (section.3.4) 45 0 R (section.3.5) 49 0 R (section.3.6) 53 0 R] /Limits [(section.3.3) (section.3.6)] >> endobj 451 0 obj << /Kids [422 0 R 423 0 R 424 0 R 425 0 R 426 0 R 427 0 R] /Limits [(Doc-Start) (cite.slavakis2008online)] >> endobj 452 0 obj << /Kids [428 0 R 429 0 R 430 0 R 431 0 R 432 0 R 433 0 R] /Limits [(cite.takizawa2013efficient) (lstnumber.-1.27)] >> endobj 453 0 obj << /Kids [434 0 R 435 0 R 436 0 R 437 0 R 438 0 R 439 0 R] /Limits [(lstnumber.-1.28) (lstnumber.-1.6)] >> endobj 454 0 obj << /Kids [440 0 R 441 0 R 442 0 R 443 0 R 444 0 R 445 0 R] /Limits [(lstnumber.-1.60) (lstnumber.-6.6)] >> endobj 455 0 obj << /Kids [446 0 R 447 0 R 448 0 R 449 0 R 450 0 R] /Limits [(lstnumber.-6.7) (section.3.6)] >> endobj 456 0 obj << /Kids [451 0 R 452 0 R 453 0 R 454 0 R 455 0 R] /Limits [(Doc-Start) (section.3.6)] >> endobj 457 0 obj << /Dests 456 0 R >> endobj 458 0 obj << /Type /Catalog /Pages 420 0 R /Outlines 421 0 R /Names 457 0 R /PageMode/UseOutlines/PageLabels<>1<>2<>]>> /OpenAction 68 0 R >> endobj 459 0 obj << /Author()/Title()/Subject()/Creator(LaTeX with hyperref package)/Producer(pdfTeX-1.40.17)/Keywords() /CreationDate (D:20170116122559+01'00') /ModDate (D:20170116122559+01'00') /Trapped /False /PTEX.Fullbanner (This is MiKTeX-pdfTeX 2.9.6050 (1.40.17)) >> endobj xref 0 460 0000000127 65535 f 0000059247 00000 n 0000059267 00000 n 0000059287 00000 n 0000000015 00000 n 0000012277 00000 n 0000245575 00000 n 0000000060 00000 n 0000000090 00000 n 0000012337 00000 n 0000245516 00000 n 0000000137 00000 n 0000000176 00000 n 0000014076 00000 n 0000245392 00000 n 0000000222 00000 n 0000000250 00000 n 0000014137 00000 n 0000245318 00000 n 0000000298 00000 n 0000000329 00000 n 0000014322 00000 n 0000245231 00000 n 0000000377 00000 n 0000000431 00000 n 0000014383 00000 n 0000245157 00000 n 0000000479 00000 n 0000000524 00000 n 0000017188 00000 n 0000245032 00000 n 0000000570 00000 n 0000000614 00000 n 0000017249 00000 n 0000244958 00000 n 0000000662 00000 n 0000000716 00000 n 0000017558 00000 n 0000244871 00000 n 0000000764 00000 n 0000000819 00000 n 0000025275 00000 n 0000244784 00000 n 0000000867 00000 n 0000000910 00000 n 0000028413 00000 n 0000244697 00000 n 0000000958 00000 n 0000001026 00000 n 0000035069 00000 n 0000244610 00000 n 0000001074 00000 n 0000001119 00000 n 0000037710 00000 n 0000244536 00000 n 0000001167 00000 n 0000001233 00000 n 0000040139 00000 n 0000244448 00000 n 0000001279 00000 n 0000001316 00000 n 0000044316 00000 n 0000244360 00000 n 0000001362 00000 n 0000001408 00000 n 0000050140 00000 n 0000244285 00000 n 0000001454 00000 n 0000001478 00000 n 0000002116 00000 n 0000002238 00000 n 0000003071 00000 n 0000006666 00000 n 0000001526 00000 n 0000006547 00000 n 0000006606 00000 n 0000242772 00000 n 0000242630 00000 n 0000002996 00000 n 0000243794 00000 n 0000002867 00000 n 0000006418 00000 n 0000007537 00000 n 0000007429 00000 n 0000006839 00000 n 0000241913 00000 n 0000242488 00000 n 0000008705 00000 n 0000008855 00000 n 0000009008 00000 n 0000009159 00000 n 0000009312 00000 n 0000009465 00000 n 0000009618 00000 n 0000009768 00000 n 0000009921 00000 n 0000010074 00000 n 0000010227 00000 n 0000010380 00000 n 0000010531 00000 n 0000010684 00000 n 0000010836 00000 n 0000010988 00000 n 0000011201 00000 n 0000008467 00000 n 0000007670 00000 n 0000011139 00000 n 0000242056 00000 n 0000012065 00000 n 0000012521 00000 n 0000011934 00000 n 0000011348 00000 n 0000012216 00000 n 0000012397 00000 n 0000012459 00000 n 0000013821 00000 n 0000014444 00000 n 0000013690 00000 n 0000012656 00000 n 0000014015 00000 n 0000014198 00000 n 0000243623 00000 n 0000014260 00000 n 0000243337 00000 n 0000016651 00000 n 0000016810 00000 n 0000019351 00000 n 0000000000 00000 f 0000024834 00000 n 0000016963 00000 n 0000017619 00000 n 0000016504 00000 n 0000014605 00000 n 0000017127 00000 n 0000242200 00000 n 0000017310 00000 n 0000242344 00000 n 0000017372 00000 n 0000017434 00000 n 0000017496 00000 n 0000052671 00000 n 0000025214 00000 n 0000053042 00000 n 0000025002 00000 n 0000025336 00000 n 0000019211 00000 n 0000017805 00000 n 0000025153 00000 n 0000243055 00000 n 0000243908 00000 n 0000021434 00000 n 0000080915 00000 n 0000024774 00000 n 0000021722 00000 n 0000113050 00000 n 0000024714 00000 n 0000021878 00000 n 0000145251 00000 n 0000024545 00000 n 0000022034 00000 n 0000105644 00000 n 0000024453 00000 n 0000022189 00000 n 0000204285 00000 n 0000024352 00000 n 0000022344 00000 n 0000120313 00000 n 0000024289 00000 n 0000022497 00000 n 0000022651 00000 n 0000080883 00000 n 0000022677 00000 n 0000113019 00000 n 0000022703 00000 n 0000145220 00000 n 0000023166 00000 n 0000105612 00000 n 0000023748 00000 n 0000204253 00000 n 0000024263 00000 n 0000120280 00000 n 0000056078 00000 n 0000028066 00000 n 0000030206 00000 n 0000027934 00000 n 0000025550 00000 n 0000028228 00000 n 0000028289 00000 n 0000028351 00000 n 0000028474 00000 n 0000028535 00000 n 0000243196 00000 n 0000028596 00000 n 0000028658 00000 n 0000028720 00000 n 0000028782 00000 n 0000028844 00000 n 0000028906 00000 n 0000028968 00000 n 0000029030 00000 n 0000029092 00000 n 0000029153 00000 n 0000029215 00000 n 0000029277 00000 n 0000029339 00000 n 0000029401 00000 n 0000029463 00000 n 0000029525 00000 n 0000029587 00000 n 0000029649 00000 n 0000029711 00000 n 0000029773 00000 n 0000029834 00000 n 0000029896 00000 n 0000029958 00000 n 0000030020 00000 n 0000030082 00000 n 0000030144 00000 n 0000058914 00000 n 0000035502 00000 n 0000032422 00000 n 0000030392 00000 n 0000032534 00000 n 0000032595 00000 n 0000032655 00000 n 0000032717 00000 n 0000032779 00000 n 0000032841 00000 n 0000032903 00000 n 0000032965 00000 n 0000033027 00000 n 0000033089 00000 n 0000033151 00000 n 0000033213 00000 n 0000033275 00000 n 0000033336 00000 n 0000033398 00000 n 0000033460 00000 n 0000033522 00000 n 0000033584 00000 n 0000033646 00000 n 0000033708 00000 n 0000033770 00000 n 0000033832 00000 n 0000033893 00000 n 0000033955 00000 n 0000034017 00000 n 0000034079 00000 n 0000034141 00000 n 0000034203 00000 n 0000034265 00000 n 0000034327 00000 n 0000034389 00000 n 0000034451 00000 n 0000034513 00000 n 0000034573 00000 n 0000034635 00000 n 0000034697 00000 n 0000034759 00000 n 0000034821 00000 n 0000034883 00000 n 0000034945 00000 n 0000035007 00000 n 0000035130 00000 n 0000035192 00000 n 0000035254 00000 n 0000035316 00000 n 0000035378 00000 n 0000035440 00000 n 0000039318 00000 n 0000037413 00000 n 0000035663 00000 n 0000037525 00000 n 0000037586 00000 n 0000037648 00000 n 0000037771 00000 n 0000037833 00000 n 0000037895 00000 n 0000037957 00000 n 0000038019 00000 n 0000038080 00000 n 0000038142 00000 n 0000038204 00000 n 0000038266 00000 n 0000038328 00000 n 0000038390 00000 n 0000038452 00000 n 0000038514 00000 n 0000038576 00000 n 0000038638 00000 n 0000038700 00000 n 0000038760 00000 n 0000038822 00000 n 0000038884 00000 n 0000038946 00000 n 0000039008 00000 n 0000039070 00000 n 0000039132 00000 n 0000039194 00000 n 0000039256 00000 n 0000040200 00000 n 0000039966 00000 n 0000039479 00000 n 0000040078 00000 n 0000041756 00000 n 0000041919 00000 n 0000042090 00000 n 0000042255 00000 n 0000042416 00000 n 0000042587 00000 n 0000042758 00000 n 0000042922 00000 n 0000043085 00000 n 0000043246 00000 n 0000043421 00000 n 0000043584 00000 n 0000043749 00000 n 0000043918 00000 n 0000044087 00000 n 0000045526 00000 n 0000045689 00000 n 0000044377 00000 n 0000041512 00000 n 0000040348 00000 n 0000044255 00000 n 0000052794 00000 n 0000058852 00000 n 0000053104 00000 n 0000058790 00000 n 0000056574 00000 n 0000052733 00000 n 0000056326 00000 n 0000056016 00000 n 0000058729 00000 n 0000056202 00000 n 0000056388 00000 n 0000058976 00000 n 0000056512 00000 n 0000045850 00000 n 0000046015 00000 n 0000046176 00000 n 0000046343 00000 n 0000046504 00000 n 0000046678 00000 n 0000046900 00000 n 0000045338 00000 n 0000044525 00000 n 0000046839 00000 n 0000244025 00000 n 0000056264 00000 n 0000056450 00000 n 0000052918 00000 n 0000056140 00000 n 0000059038 00000 n 0000052856 00000 n 0000052980 00000 n 0000048836 00000 n 0000049018 00000 n 0000049196 00000 n 0000049380 00000 n 0000049557 00000 n 0000049890 00000 n 0000049711 00000 n 0000050263 00000 n 0000048656 00000 n 0000047047 00000 n 0000050079 00000 n 0000242914 00000 n 0000050201 00000 n 0000056636 00000 n 0000053166 00000 n 0000052436 00000 n 0000050461 00000 n 0000052548 00000 n 0000052609 00000 n 0000243479 00000 n 0000056698 00000 n 0000055843 00000 n 0000053326 00000 n 0000055955 00000 n 0000059100 00000 n 0000058556 00000 n 0000056833 00000 n 0000058668 00000 n 0000059340 00000 n 0000059845 00000 n 0000059870 00000 n 0000059949 00000 n 0000060014 00000 n 0000060039 00000 n 0000060422 00000 n 0000241478 00000 n 0000060876 00000 n 0000061341 00000 n 0000061978 00000 n 0000062622 00000 n 0000063259 00000 n 0000063689 00000 n 0000064040 00000 n 0000081239 00000 n 0000096446 00000 n 0000096788 00000 n 0000105877 00000 n 0000113257 00000 n 0000120528 00000 n 0000145765 00000 n 0000160651 00000 n 0000160973 00000 n 0000172269 00000 n 0000172531 00000 n 0000179526 00000 n 0000179746 00000 n 0000187014 00000 n 0000187243 00000 n 0000195866 00000 n 0000196132 00000 n 0000204537 00000 n 0000223036 00000 n 0000223414 00000 n 0000240926 00000 n 0000244134 00000 n 0000244211 00000 n 0000245684 00000 n 0000245857 00000 n 0000246019 00000 n 0000246203 00000 n 0000246463 00000 n 0000246745 00000 n 0000247041 00000 n 0000247395 00000 n 0000247646 00000 n 0000247879 00000 n 0000248113 00000 n 0000248346 00000 n 0000248580 00000 n 0000248813 00000 n 0000249047 00000 n 0000249280 00000 n 0000249514 00000 n 0000249746 00000 n 0000249978 00000 n 0000250212 00000 n 0000250441 00000 n 0000250670 00000 n 0000250904 00000 n 0000251137 00000 n 0000251366 00000 n 0000251562 00000 n 0000251729 00000 n 0000251891 00000 n 0000252086 00000 n 0000252240 00000 n 0000252367 00000 n 0000252503 00000 n 0000252627 00000 n 0000252751 00000 n 0000252863 00000 n 0000252970 00000 n 0000253008 00000 n 0000253184 00000 n trailer << /Size 460 /Root 458 0 R /Info 459 0 R /ID [<1ABBFC7F36E6F69E2E5965BEFA2E2338> <1ABBFC7F36E6F69E2E5965BEFA2E2338>] >> startxref 253459 %%EOF kafbox-2.0/doc/kafbox_doc.tex000066400000000000000000000375661303712747200162420ustar00rootroot00000000000000\documentclass[]{report} \usepackage{fixltx2e,fix-cm} \usepackage{amssymb} \usepackage{amsmath} \usepackage{algorithm} \usepackage{algorithmic} \usepackage{courier} \usepackage{color} \usepackage{listings} \usepackage{tikz} % for diagram in first Fig. \usetikzlibrary{shapes,arrows} \tikzstyle{block} = [draw, rectangle, fill=white, minimum height=2em, minimum width=4em] \tikzstyle{sum} = [draw, circle, node distance=2em] \tikzstyle{input} = [] \tikzstyle{node} = [coordinate, node distance=2em] \usetikzlibrary{external} \tikzexternalize % activate \definecolor{matgreen}{rgb}{0,0.6,0} \definecolor{matpurple}{rgb}{0.625,0.125,0.9375} \definecolor{gray}{rgb}{0.5,0.5,0.5} \lstset{language=Matlab, keywords={break,case,catch,classdef,continue,else,elseif,end,for,function,global,if,otherwise,persistent,properties,return,switch,try,while}, basicstyle=\ttfamily\footnotesize, keywordstyle=\color{blue}, commentstyle=\color{matgreen}, stringstyle=\color{matpurple}, numbers=left, numberstyle=\color{gray}, stepnumber=1, numbersep=10pt, backgroundcolor=\tiny\color{white}, tabsize=4, showspaces=false, showstringspaces=false} \lstnewenvironment{shortlisting}{ \lstset{language=Matlab, keywords={break,case,catch,classdef,continue,else,elseif,end,for,function,global,if,otherwise,persistent,properties,return,switch,try,while}, basicstyle=\ttfamily, keywordstyle=\color{blue}, commentstyle=\color{matgreen}, stringstyle=\color{matpurple}, numbers=none, numberstyle=\tiny\color{gray}, stepnumber=1, numbersep=10pt, backgroundcolor=\color{white}, tabsize=4, showspaces=false, showstringspaces=false} }{} \def\d{{\mathbf d}} \def\x{{\mathbf x}} \def\K{{\mathbf K}} %\usepackage{url} \usepackage{hyperref} \begin{document} \title{Documentation for \\Kernel Adaptive Filtering Toolbox} \author{Steven Van Vaerenbergh} \date{~\\[1em]\includegraphics[width=4cm]{fig/kafbox_logo}\\[1em] {\Large Technical Report} \\[1em] TR-KAFBOX-02 \\[1em] January 2017\\[5em] \begin{minipage}[t]{\textwidth} \centering \begin{minipage}[t]{0.2\textwidth} ~\\ \includegraphics[width=1.65cm]{fig/unican_logo} \end{minipage} % \begin{minipage}[t]{0.7\textwidth} ~\\ Advanced Signal Processing Group\\ Department of Communications Engineering\\ University of Cantabria, Spain \\ \end{minipage} \end{minipage} } \maketitle \begin{abstract} The Kernel Adaptive Filtering Toolbox is a benchmarking toolbox to evaluate and compare kernel adaptive filtering algorithms in Matlab. Kernel adaptive filtering algorithms are online machine learning techniques suitable for nonlinear filtering, prediction, tracking and regression. This toolbox contains implementations of all major kernel adaptive filtering algorithms, and tools to measure and compare the performance, memory usage, speed and other properties. \end{abstract} %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \tableofcontents \chapter{Introduction} %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \section{Goals of the toolbox} The goals of this toolbox are twofold: \begin{enumerate} \item To provide a repository for Matlab implementations of kernel adaptive filtering algorithms; \item To provide tools that allow to compare all aspects of the different algorithms. \end{enumerate} A list of the included algorithms and tools can be found in Chapter \ref{chap:algos}. %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \chapter{First use} \section{Installation} \begin{enumerate} \item Download the toolbox and run the \verb"installation.m" file in the root folder. \item Type \verb"savepath" to save the changes to the path. \end{enumerate} \section{Directories included in the toolbox} \begin{itemize} \item \verb"data/" contains data sets. \item \verb"demo/" contains demos and test files. \item \verb"doc/" contains the documentation for the toolbox. \item \verb"lib/" contains the algorithm libraries and utilities. \end{itemize} \section{Versions and compatibility} KAFBOX uses the \verb"classdef" command which is not supported in Matlab pre-2008a and not yet in Octave. The older 0.x versions of this toolbox do not use \verb"classdef" and can therefore be used with all versions of Matlab and Octave. \url{https://github.com/steven2358/kafbox/releases/} The current version of KAFBOX is v2. %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \chapter{Kernel adaptive filtering} \section{Framework for online kernel methods} % summary: growing functional representation Online learning methods update their solution iteratively. In the standard online learning framework, each iteration consists of several trials \cite{blum1998line}. During the $n$-th iteration: \begin{enumerate} \item The algorithm first receives an input datum, $\x_n$; \item Then, it calculates the estimated output $\hat y_n$ corresponding to this datum. \item The true outcome $y_n$ is made available shortly thereafter, which enables the algorithm to calculate the loss $L(\cdot)$ incurred on the data pair $(\x_n,y_n)$; \item Finally, the solution is updated. \end{enumerate} A typical setup for online system identification with a kernel-based method is depicted in Fig. \ref{ork:fig:diagram}. It represents an unknown nonlinear system, whose input data $\x_n$ and response $y_n$ (including additive noise $r_n$) can be measured at different time steps, and an adaptive kernel-based algorithm, which is used to identify the system's response. \begin{figure}[h] \centering \tikzsetnextfilename{kaf_diagram} \begin{tikzpicture}[auto, node distance=2cm,>=latex] \node [input, name=input] {$\mathbf{x}_n$}; \node [node, right of=input] (node1) {}; \draw (input) -- (node1); \node [block, right of=node1] (system) {Nonlinear system}; \draw [->] (node1) -- (system); \node [sum, right of=system, xshift=7em] (plus) {}; \draw [->] (system) -- node[pos=0.8, below] {$+$} node {$f(\mathbf{x}_n)$} (plus); \node [input, name=noise, above of=plus, node distance=3em] {$r_n$}; \draw [->] (noise) -- node[pos=0.8, right] {$+$}(plus); \node [block, below of=system, yshift=1em] (transformation) {$\mathcal{X}\rightarrow \mathcal{H}$}; \draw [->] (node1) |- (transformation); \node [node, right of=transformation, xshift=3em] (node2) {}; \draw (transformation) -- (node2); \node [block, right of=node2] (filter) {$\boldsymbol{\mathbf{\alpha}}_n$}; \draw [->] (transformation) -- node {$\kappa(\mathbf{x}_n,\cdot)$} (filter); \node [sum, right of=filter, xshift=4em] (minus) {}; \draw [->] (filter) -- node[pos=0.9, below] {$-$} node {$\hat y_n$} (minus); \draw [->] (plus) -| node[pos=0.9, right] {$+$} node {$y_n$} (minus); \node [input, right of=minus, xshift=-2em] (output) {$e_n$}; \draw [->] (minus) -- (output); \node [block, below of=filter, yshift=1em] (algorithm) {Adaptive algorithm}; \draw [->] (node2) |- (algorithm); \draw [->] (minus) |- node {$e_n$} (algorithm); \node [node, above of=algorithm, yshift=1em] (node3) {}; \draw (algorithm) -- (node3); \node [node, above of=node3, xshift=1.5em, yshift=1.5em] (node4) {}; \draw [->] (node3) -- (node4); \node [block, right of=node2] (filter2) {$\boldsymbol{\mathbf{\alpha}}_n$}; % repeat to fix z-index \end{tikzpicture} %\centerline{\includegraphics[height=150pt]{chapters/chapter_ork/fig/diagram.eps}} \caption{Kernel-based adaptive system identification. Figure adapted from \cite{parreira2012stochastic}.} \label{ork:fig:diagram} \end{figure} Online algorithms should be capable of operating during extended periods of time, processing large amounts of data. Kernel methods rely on a functional representation that grows as the amount of observations increases. A na\"ive implementation of an online kernel method will therefore require growing computational resources during operation, leading to performance issues once the available memory is insufficient to store the training data or once the computations for one update take more time than the interval between incoming data \cite{kivinen2004online}. %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \section{Kernel adaptive filtering algorithms} Kernel adaptive filtering algorithms are online machine learning algorithms that build the solution to a regression or classification problem. While early kernel adaptive filtering algorithms had a clear connection to classical adaptive filters, such as LMS, RLS and KAPA filters, later implementations explored Gaussian process, projections-based and multi-kernel algorithms. The different kernel adaptive filtering algorithms proposed in the literature vary in several aspects: \begin{itemize} \item The type of filter they relate to: typically LMS, RLS or KAPA; \item The computational complexity w.r.t. the amount of elements in memory, $M$: typically $\mathcal{O}(M)$ or $\mathcal{O}(M^2)$; \item The approach to building a sparse dictionary and to possible prune data from it afterwards; \item The convergence speed; \item The ability to employ multiple kernels and learn their respective weights. \end{itemize} The algorithm profiler described in Chapter \ref{sec:profiler} allows to compare these different aspects empirically. \section{Algorithm code structure} Since all algorithms share the same structure, they are coded in a common framework. In particular: \begin{itemize} \item Each algorithm is contained in a single file in the \verb"/lib" folder. \item Each algorithm is implemented as an object in matlab, using the \verb"classdef" syntax. \item Each algorithm has two basic public methods: \verb"evaluate" and \verb"train". The object code contains only one iteration of the algorithm. The for-loop over the time index that governs the online operation should go in an external script. \end{itemize} Furthermore, efforts are made to make the code \begin{enumerate} \item As human-readable as possible, in the first place; \item Short and structured, in the second place. \end{enumerate} As far as possible, algorithm structure should follow the pseudocode from the corresponding publication. Finally, the following best practices for scientific computing are taken into account \cite{wilson2014best}: \begin{itemize} \item Variable naming should correspond to the nomenclature used in the corresponding publication whenever possible; \item Comments should be used sparingly. Document the design and purpose of the code rather than its mechanics. \end{itemize} \section{Template for kernel adaptive filtering algorithms} \lstinputlisting{../lib/kafbox_template.m} \section{Performing online learning} Each kernel adaptive filtering algorithm is implemented as a Matlab class. To perform online learning with an algorithm, first define its options: \begin{shortlisting} options = struct('nu',1E-4,'kerneltype','gauss','kernelpar',32); \end{shortlisting} Next, create an instance of the filter. E.g., for an instance of the KRLS algorithm that uses the ALD criterion run: \begin{shortlisting} kaf = aldkrls(options); \end{shortlisting} One iteration of training is performed by feeding one input-output data pair to the filter: \begin{shortlisting} kaf.train(x,y); \end{shortlisting} The outputs for one or more test inputs are evaluated as follows: \begin{shortlisting} Y_test = kaf.evaluate(X_test); \end{shortlisting} \section{Example learning script: time-series prediction} \lstinputlisting{../demo/demo_prediction.m} %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \chapter{Algorithm profiler} \label{sec:profiler} The algorithm profiler is a tool that allows to compare the trade-off between cost and performance between several algorithm configurations and/or several algorithms. A demo configuration script is given in the file\\ \verb"demo/demo_profiler_prediction_lorenz.m". %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \chapter{List of included algorithms} \label{chap:algos} Matlab implementations of the following algorithms are included in the toolbox: \begin{itemize} \item Approximate Linear Dependency Kernel Recursive Least-Squares (ALD-KRLS) \cite{engel2004kernel}. \item Sliding-Window Kernel Recursive Least-Squares (SW-KRLS) \cite{vanvaerenbergh2006sliding}. \item Naive Online Regularized Risk Minimization Algorithm (NORMA) \cite{kivinen2004online}. \item Kernel Least-Mean-Square (KLMS) \cite{liu2008kernel}. \item Fixed-Budget Kernel Recursive Least-Squares (FB-KRLS) \cite{vanvaerenbergh2010fixed}. \item Kernel Recursive Least-Squares Tracker (KRLS-T) \cite{vanvaerenbergh2012kernel}. \item Quantized Kernel Least Mean Squares (QKLMS) \cite{chen2012quantized}. \item Random Fourier Feature Kernel Least Mean Square (RFF-KLMS) algorithm \cite{singh2012random}. \item Extended Kernel Recursive Least Squares (EX-KRLS) \cite{liu2009extended}. \item Gaussian-Process based estimation of the parameters of KRLS-T \cite{vanvaerenbergh2012estimation}. \item Kernel Affine Projection (KAP) algorithm with Coherence Criterion \cite{richard2009online}. \item Kernel Normalized Least-Mean-Square (KNLMS) algorithm with Coherence Criterion \cite{richard2009online}. \item Recursive Least-Squares algorithm with exponential weighting (RLS) \cite{sayed2003fundamentals}. \item Multikernel Normalized Least Mean Square algorithm with Coherence-based Sparsification (MKNLMS-CS) \cite{yukawa2012multikernel}. \item Parallel HYperslab Projection along Affine SubSpace (PHYPASS) algorithm \cite{takizawa2013efficient}. \item Fixed-budget kernel least mean squares (FB-KLMS) algorithm \cite{rzepka2012fixed}. \item Leaky Kernel Affine Projection Algorithm (LKAPA, including KAPA-1 and KAPA-3) and Normalized Leaky Kernel Affine Projection Algorithm (NLKAPA, including KAPA-2 and KAPA-4) \cite{liu2008kernel}. \item Kernel Affine Projection Subgradient Method (KAPSM) \cite{slavakis2008online}. \item Kernel Least Mean Squares algorithm with Coherence-Sparsification criterion and L1-norm regularization (KLMS-CSL1) and with active L1-norm regularization (KLMS-CSAL1) \cite{gao2013kernel}. \item Mixture Kernel Least Mean Square (MxKLMS) algorithm \cite{pokharel2013mixture}. \item Fixed budget quantized kernel least-mean-square algorithm (QKLMS-FB) \cite{zhao2013fixed} \item Probabilistic Least-Mean-Squares (ProbLMS) algorithm \cite{fernandez2015probabilistic}. \item Discrete-time Kalman filter \cite{kalman1960new}. \end{itemize} %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \chapter{About} \begin{itemize} \item {\bf Name:} Kernel Adaptive Filtering Toolbox \item {\bf Abbreviation:} KAFBOX \item {\bf Current version:} v2 \item {\bf Punchline:} a Matlab benchmarking toolbox for kernel adaptive filtering \item {\bf Author:} Steven Van Vaerenbergh \begin{itemize} \item Homepage: \url{http://gtas.unican.es/people/steven/} \item Email: \url{steven.vanvaerenbergh@unican.es} \item Affiliation: Advanced Signal Processing Group, Department of Communications Engineering, University of Cantabria, Spain. \end{itemize} \item {\bf Collaborators:} Miguel Lazaro-Gredilla, Sohan Seth, Masahiro Yukawa, Masa-aki Takizawa, Osamu Toda, Dominik Rzepka, Pantelis Bouboulis, Stephan Rhode \item {\bf URL:} \url{https://github.com/steven2358/kafbox/} \item {\bf Citing:} Published reports of research using this code (or a modified version) should cite \cite{vanvaerenbergh2013comparative}. \item {\bf Environment:} Matlab\footnote{\url{http://www.mathworks.com/products/matlab/}}. \item {\bf Origin:} KAFBOX is a fork of Kernel Methods Toolbox (KMBOX) v0.6 (\url{http://sourceforge.net/p/kmbox/}) \item {\bf Extending the code:} Template files are provided to encourage external authors to include their own code into the toolbox. Contributions can be made through Github's fork and pull system. \item {\bf License:} FreeBSD \end{itemize} %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \bibliographystyle{alpha} \bibliography{kaf,extra} \end{document} kafbox-2.0/install.m000066400000000000000000000006331303712747200144610ustar00rootroot00000000000000% Installation file. Adds local folders to path. fprintf('Adding KAFBOX folders to Matlab path... ') addpath(genpath(fullfile(pwd,'data'))); % add data folder with subfolders addpath(genpath(fullfile(pwd,'lib'))); % add lib folder with subfolders % addpath(fullfile(pwd,'demo')); % add demo folder without subfolders fprintf('done.\n') disp('Type "savepath" if you wish to store the changes.') % savepath; kafbox-2.0/lib/000077500000000000000000000000001303712747200134015ustar00rootroot00000000000000kafbox-2.0/lib/exkrls.m000066400000000000000000000060571303712747200150770ustar00rootroot00000000000000% Extended Kernel Recursive Least Squares algorithm % % W. Liu, I.M. Park. Y. Wang, and J.C. Principe, "Extended Kernel Recursive % Least Squares Algorithm," IEEE Transactions on Signal Processing, vol. % 57, no. 10, pp. 3801-3814, Oct. 2009, % http://dx.doi.org/10.1109/TSP.2009.2022007 % % Remark: implementation of the tracking model, includes a maximum % dictionary size M % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ classdef exkrls < handle properties (GetAccess = 'public', SetAccess = 'private') alphaf = .999; % state forgetting factor, "alpha" in publication beta = .995; % data forgetting factor lambda = 1E-2; % regularization q = 1E-3; % trade-off between modeling variation and measurement disturbance M = 500; % maximum dictionary size kerneltype = 'gauss'; % kernel type kernelpar = 1; % kernel parameter end properties (GetAccess = 'public', SetAccess = 'private') dict = []; % dictionary rho = []; Q = []; i = 0; % iteration number; alpha = []; % expansion coefficients, "a" in publication end methods function kaf = exkrls(parameters) % constructor allpars = {'alphaf','lambda','beta','q','kerneltype','kernelpar','M'}; if (nargin > 0) for j=1:length(allpars), p = allpars{j}; if isfield(parameters,p), kaf.(p) = parameters.(p); end end end end function y_est = evaluate(kaf,x) % evaluate the algorithm if size(kaf.dict,1)>0 k = kernel(kaf.dict,x,kaf.kerneltype,kaf.kernelpar); y_est = k'*kaf.alpha; else y_est = zeros(size(x,1),1); end end function train(kaf,x,y) % train the algorithm kaf.i = kaf.i + 1; k = kernel([kaf.dict; x],x,kaf.kerneltype,kaf.kernelpar); kt = k(1:end-1); ktt = k(end); if numel(kt)==0 % initialize kaf.alpha = kaf.alphaf*y/(kaf.lambda*kaf.beta+ktt); kaf.rho = kaf.lambda*kaf.beta/(kaf.alphaf^2*kaf.beta + kaf.lambda*kaf.q); kaf.Q = kaf.alphaf^2/((kaf.beta*kaf.lambda+ktt)*(kaf.alphaf^2+kaf.beta*kaf.lambda*kaf.q)); kaf.dict = x; else if (size(kaf.dict,1) 0) % copy valid parameters for fn = fieldnames(parameters)', if ismember(fn,fieldnames(kaf)), kaf.(fn{1}) = parameters.(fn{1}); end end end end function y_est = evaluate(kaf,x) % evaluate the algorithm if size(kaf.dict,1)>0 k = kernel(kaf.dict,x,kaf.kerneltype,kaf.kernelpar); y_est = k'*kaf.alpha; else y_est = zeros(size(x,1),1); end end function train(kaf,x,y) % train the algorithm if size(kaf.dict,2)==0 % initialize kaf.dict = x; k = kernel(kaf.dict,x,kaf.kerneltype,kaf.kernelpar); kaf.diagkdict(1) = k; kaf.alpha = kaf.eta*y*k/(k'*k); else kt = kernel([kaf.dict;x],x,kaf.kerneltype,kaf.kernelpar); k = kt(1:end-1); y_est = k'*kaf.alpha; e = y - y_est; kaf.alpha = kaf.alpha + kaf.eta*e*k/(k'*k); % growth criterion kx = kt(end); dependency = kx - k./kaf.diagkdict; if min(dependency) >= kaf.nu, % expand dictionary kaf.dict = [kaf.dict; x]; kaf.diagkdict = [kaf.diagkdict; kx]; kaf.alpha = [kaf.alpha; kaf.eta*e/(k'*k)]; if length(kaf.alpha) > kaf.M, % prune dictionary [~, id] = min(abs(kaf.alpha)); kaf.dict(id,:) = []; kaf.diagkdict(id) = []; kaf.alpha(id) = []; end end end end end end kafbox-2.0/lib/fbkrls.m000066400000000000000000000063671303712747200150560ustar00rootroot00000000000000% Fixed-Budget Kernel Recursive Least Squares algorithm % % S. Van Vaerenbergh, I. Santamaria, W. Liu, and J.C. Principe, "Fixed- % budget kernel recursive least-squares," 2010 IEEE International % Conference on Acoustics Speech and Signal Processing (ICASSP), pp. 1882- % 1885, 14-19 March 2010, http://dx.doi.org/10.1109/ICASSP.2010.5495350 % % Remark: label update is not implemented (mu=0) % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ classdef fbkrls < handle properties (GetAccess = 'public', SetAccess = 'private') M = 100; % dictionary size lambda = 1E-2; % regularization parameter kerneltype = 'gauss'; % kernel type kernelpar = 1; % kernel parameter end properties (GetAccess = 'public', SetAccess = 'private') dict = []; % dictionary dicty = []; % output dictionary alpha = []; % expansion coefficients Kinv = []; % inverse kernel matrix end methods function kaf = fbkrls(parameters) % constructor if (nargin > 0) % copy valid parameters for fn = fieldnames(parameters)', if ismember(fn,fieldnames(kaf)), kaf.(fn{1}) = parameters.(fn{1}); end end end end function y_est = evaluate(kaf,x) % evaluate the algorithm if size(kaf.dict,1)>0 k = kernel(kaf.dict,x,kaf.kerneltype,kaf.kernelpar); y_est = k'*kaf.alpha; else y_est = zeros(size(x,1),1); end end function train(kaf,x,y) % train the algorithm kaf.dict = [kaf.dict; x]; % grow kaf.dicty = [kaf.dicty; y]; % grow k = kernel(kaf.dict,x,kaf.kerneltype,kaf.kernelpar); kaf.Kinv = kaf.grow_kernel_matrix(kaf.Kinv,k,kaf.lambda);% grow kaf.alpha = kaf.Kinv*kaf.dicty; if (size(kaf.dict,1) > kaf.M) ape = abs(kaf.alpha)./diag(kaf.Kinv); % a posteriori error [~,ind] = min(ape); kaf.dict(ind,:) = []; % prune kaf.dicty(ind) = []; % prune kaf.Kinv = kaf.prune_kernel_matrix(kaf.Kinv,ind); % prune end kaf.alpha = kaf.Kinv*kaf.dicty; end end methods (Static = true) function Kinv = grow_kernel_matrix(Kinv,k,c) % calculate inverse of expanded matrix K = [K_inv b;b' d] b = k(1:end-1); d = k(end) + c; % add regularization if numel(b)>0 g_inv = d - b'*Kinv*b; g = 1/g_inv; f = -Kinv*b*g; E = Kinv - Kinv*b*f'; Kinv = [E f;f' g]; else Kinv = 1/d; end end function Kinv = prune_kernel_matrix(Kinv,ind) % calculate inverse of pruned kernel matrix m = size(Kinv,1); noind = 1:m; noind(ind) = []; G = Kinv(noind,noind); f = Kinv(noind,ind); e = Kinv(ind,ind); Kinv = G - f*f'/e; end end end kafbox-2.0/lib/kafbox_template.m000066400000000000000000000040431303712747200167250ustar00rootroot00000000000000% This is the template used for kernel adaptive filtering algorithms in % the kernel adaptive filtering toolbox. [Delete this line.] % % [The name of the algorithm goes here] % % [A reference to the original publication of the algorithm goes here, % including a link to its DOI url: http://dx.doi.org/xxx] % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ classdef kafbox_template < handle properties (GetAccess = 'public', SetAccess = 'private') % parameters param1 = 1; param2 = 2; kerneltype = 'gauss'; % kernel type kernelpar = 1; % kernel parameter end properties (GetAccess = 'public', SetAccess = 'private') % variables dict = []; % dictionary alpha = []; % expansion coefficients end methods function kaf = kafbox_template(parameters) % constructor if (nargin > 0) % copy valid parameters for fn = fieldnames(parameters)', if ismember(fn,fieldnames(kaf)), kaf.(fn{1}) = parameters.(fn{1}); end end end end function y_est = evaluate(kaf,x) % evaluate the algorithm if size(kaf.dict,1)>0 k = kernel(kaf.dict,x,kaf.kerneltype,kaf.kernelpar); y_est = k'*kaf.alpha; else y_est = zeros(size(x,1),1); end end function train(kaf,x,y) % train the algorithm if size(kaf.dict,2)==0 % initialize kaf.dict = x; kaf.alpha = 0; else % [main algorithm training goes here] % [example of a helper function] z = kaf.helper1(x,y); end end end methods (Static = true) % [helper functions go here] function z = helper1(x,y) z = x*y; % operations end end end kafbox-2.0/lib/kalman.m000066400000000000000000000057531303712747200150340ustar00rootroot00000000000000% Discrete-time Kalman filter. % % R. E. Kalman, "A New Approach to Linear Filtering and Prediction % Problems", ASME Journal of Basic Engineering, 1960, 82(1), pp. 35-45. % http://dx.doi.org/10.1115/1.3662552 % % Notation is taken from: R. Faragher, "Understanding the Basis of the % Kalman Filter Via a Simple and Intuitive Derivation [Lecture Notes]," in % IEEE Signal Processing Magazine, vol. 29, no. 5, pp. 128-132, Sept. 2012. % http://dx.doi.org10.1109/MSP.2012.2203621 % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ classdef kalman < handle properties (GetAccess = 'public', SetAccess = 'private') % parameters F; % state transition matrix B; % control input matrix Q; % state noise covariance matrix % H; % state-to-measurements transition matrix R; % measurement noise covariance matrix end properties (GetAccess = 'public', SetAccess = 'private') % variables x_mean; % state mean P; % state covariance end methods function obj = kalman(parameters) % constructor if (nargin > 0) % copy valid parameters for fn = fieldnames(parameters)', if ismember(fn,fieldnames(obj)), obj.(fn{1}) = parameters.(fn{1}); end end end end function [mean_test, var_test] = evaluate(obj,H) % evaluate if ~isempty(obj.x_mean) mean_test = H*obj.x_mean; var_test = H*obj.P*H'; % predictive variance else mean_test = zeros(size(H,1),1); var_test = nan(size(H,1),1); % signal scale is unknown end end function train(obj,H,z,u) % train the algorithm if isempty(obj.x_mean) % initialize [n,m] = size(H); % measurement and state dimensions obj.x_mean = zeros(m,1); % filter coefficients obj.P = eye(m); if isempty(obj.F) % default state transition: random walk obj.F = eye(m); obj.Q = 0.1*eye(m); % default control input obj.B = eye(m); % default measurement noise covariance obj.R = 1E-2*eye(n); end end if nargin<4 u = zeros(size(H,2),1); % no control input end % state prediction equations obj.x_mean = obj.F*obj.x_mean + obj.B*u; obj.P = obj.F*obj.P*obj.F + obj.Q; % calculate the kalman gain K = obj.P*H'/(H*obj.P*H' + obj.R); % measurement update equations obj.x_mean = obj.x_mean + K*(z - H*obj.x_mean); % mean obj.P = obj.P - K*H*obj.P; % covariance end end end kafbox-2.0/lib/kap.m000066400000000000000000000060041303712747200143320ustar00rootroot00000000000000% Kernel Affine Projection (KAP) algorithm with coherence criterion % % C. Richard, J.C.M. Bermudez, and P. Honeine, "Online Prediction of Time % Series Data With Kernels," IEEE Transactions on Signal Processing, % vol. 57, no. 3, pp. 1058-1067, March 2009, % http://dx.doi.org/10.1109/TSP.2008.2009895 % % Remark: memories are initialized empty in this implementation % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ classdef kap < handle properties (GetAccess = 'public', SetAccess = 'private') mu0 = .95; % coherence criterion threshold eta = .5; % step size eps = 1E-2; % regularization p = 20; % memory length kerneltype = 'gauss'; % kernel type kernelpar = 1; % kernel parameter end properties (GetAccess = 'public', SetAccess = 'private') memx = []; % input memory memy = []; % output memory dict = []; % dictionary modict = []; % modulus of the dictionary elements alpha = []; % expansion coefficients end methods function kaf = kap(parameters) % constructor if (nargin > 0) % copy valid parameters for fn = fieldnames(parameters)', if ismember(fn,fieldnames(kaf)), kaf.(fn{1}) = parameters.(fn{1}); end end end end function y_est = evaluate(kaf,x) % evaluate the algorithm if size(kaf.dict,1)>0 k = kernel(kaf.dict,x,kaf.kerneltype,kaf.kernelpar); y_est = k'*kaf.alpha; else y_est = zeros(size(x,1),1); end end function train(kaf,x,y) % train the algorithm if (length(kaf.memy) < kaf.p) kaf.memx = [kaf.memx; x]; % grow the memory kaf.memy = [kaf.memy; y]; % grow the memory else kaf.memx = [kaf.memx(2:end,:); x]; % sliding memory kaf.memy = [kaf.memy(2:end); y]; % sliding memory end if size(kaf.dict,2)==0 % initialize k = kernel(x,x,kaf.kerneltype,kaf.kernelpar); kaf.dict = x; kaf.modict = sqrt(k); kaf.alpha = 0; else k = kernel(kaf.dict,x,kaf.kerneltype,kaf.kernelpar); kx = kernel(x,x,kaf.kerneltype,kaf.kernelpar); C = k./(sqrt(kx)*kaf.modict); % coherence if (max(C) <= kaf.mu0), % coherence criterion kaf.dict = [kaf.dict; x]; % order increase kaf.alpha = [kaf.alpha; 0]; % order increase end end H = kernel(kaf.memx,kaf.dict,kaf.kerneltype,kaf.kernelpar); kaf.alpha = kaf.alpha + ... kaf.eta*H'/... (kaf.eps*eye(size(H,1)) + H*H')*... (kaf.memy - H*kaf.alpha); end end end kafbox-2.0/lib/kapsm.m000066400000000000000000000131151303712747200146730ustar00rootroot00000000000000% Kernel Affine Projection Subgradient Method % % K. Slavakis, S. Theodoridis, and I. Yamada, "Online kernel-based % classification using adaptive projection algorithms," IEEE Transactions % on Signal Processing, Vol. 56, No. 7, pp. 2781-2796, 2008. % http://dx.doi.org/10.1109/TSP.2008.917376 % % Remark: implemented with L2-ball forgetting. Code contributed by % Pantelis Bouboulis. % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ classdef kapsm < handle properties (GetAccess = 'public', SetAccess = 'private') % parameters M = 200; % dictionary size Delta = 10; % L2-ball forgetting radius mu = 1.8, % learning rate Q = 10; % number of subgradients loss = 'l2'; % loss function type loss_param = 2; % Huber loss parameter kerneltype = 'gauss'; % kernel type kernelpar = .5; % kernel parameter end properties (GetAccess = 'public', SetAccess = 'private') % variables dict = []; % dictionary alpha = []; % expansion coefficients norm_f = []; % function norm for L2-ball forgetting xmem = []; % input memory ymem = []; % output memory end methods function kaf = kapsm(parameters) % constructor if (nargin > 0) % copy valid parameters for fn = fieldnames(parameters)', if ismember(fn,fieldnames(kaf)), kaf.(fn{1}) = parameters.(fn{1}); end end end end function y_est = evaluate(kaf,x) % evaluate the algorithm if size(kaf.dict,1)>0 k = kernel(kaf.dict,x,kaf.kerneltype,kaf.kernelpar); y_est = k'*kaf.alpha; else y_est = zeros(size(x,1),1); end end function train(kaf,x,y) % train the algorithm if size(kaf.xmem,1) kaf.Delta kaf.alpha = kaf.Delta/kaf.norm_f * kaf.alpha; kaf.norm_f = kaf.Delta; end % sliding-window pruning if length(kaf.alpha)>kaf.M, k = kernel(kaf.dict,kaf.dict(1,:),... kaf.kerneltype,kaf.kernelpar); kaf.norm_f = sqrt(kaf.norm_f.^2 + kaf.alpha(1)^2*k(1) - ... 2*kaf.alpha(1)*k'*kaf.alpha); kaf.dict(1,:) = []; kaf.alpha(1) = []; end end end methods (Static = true) function loss = loss_fun(ksi,loss_type,loss_param) switch loss_type case 'l2' loss = ksi.^2; case 'l1' loss = abs(ksi); case 'huber' sigma = loss_param; loss = zeros(length(ksi),1); for i=1:length(ksi), if abs(ksi(i)) <= sigma loss(i) = ksi(i)^2/(2*sigma); else loss(i) = abs(ksi(i)) - sigma/2; end end end end function subgrad_coef = ... compute_subgrad_coef(e,loss_type,loss_param) subgrad_coef = zeros(length(e),1); for i=1:length(e) switch loss_type case 'l2' subgrad_coef(i) = -2*e(i); case 'l1' subgrad_coef(i) = -sign(e(i)); case 'huber' sigma = loss_param; if abs(e(i)) <= sigma subgrad_coef(i) = -e(i)/sigma; else subgrad_coef(i) = -sign(e(i)); end end end end end end kafbox-2.0/lib/klms.m000066400000000000000000000036341303712747200145330ustar00rootroot00000000000000% Kernel Least-Mean-Square algorithm % % W. Liu, P.P. Pokharel, and J.C. Principe, "The Kernel Least-Mean-Square % Algorithm," IEEE Transactions on Signal Processing, vol. 56, no. 2, pp. % 543-554, Feb. 2008, http://dx.doi.org/10.1109/TSP.2007.907881 % % Remark: implementation includes a maximum dictionary size M % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ classdef klms < handle properties (GetAccess = 'public', SetAccess = 'private') eta = .5; % learning rate M = 10000; % maximum dictionary size kerneltype = 'gauss'; % kernel type kernelpar = 1; % kernel parameter end properties (GetAccess = 'public', SetAccess = 'private') dict = []; % dictionary alpha = []; % expansion coefficients end methods function kaf = klms(parameters) % constructor if (nargin > 0) % copy valid parameters for fn = fieldnames(parameters)', if ismember(fn,fieldnames(kaf)), kaf.(fn{1}) = parameters.(fn{1}); end end end end function y_est = evaluate(kaf,x) % evaluate the algorithm if size(kaf.dict,1)>0 k = kernel(kaf.dict,x,kaf.kerneltype,kaf.kernelpar); y_est = k'*kaf.alpha; else y_est = zeros(size(x,1),1); % zeros if not initialized end end function train(kaf,x,y) % train the algorithm if (size(kaf.dict,1) 0) % copy valid parameters for fn = fieldnames(parameters)', if ismember(fn,fieldnames(kaf)), kaf.(fn{1}) = parameters.(fn{1}); end end end end function y_est = evaluate(kaf,x) % evaluate the algorithm if size(kaf.dict,1)>0 k = kernel(kaf.dict,x,kaf.kerneltype,kaf.kernelpar); y_est = k'*kaf.alpha; else y_est = zeros(size(x,1),1); end end function train(kaf,x,y) % train the algorithm if size(kaf.dict,2)==0 % initialize kd = kernel(x,x,kaf.kerneltype,kaf.kernelpar); kaf.dict = x; kaf.modict = sqrt(kd); kaf.alpha = y/kd; w = 1./abs(kaf.alpha + kaf.eps_alpha); % for sparsification else w = 1./abs(kaf.alpha + kaf.eps_alpha); % for sparsification kd = kernel(kaf.dict,x,kaf.kerneltype,kaf.kernelpar); kx = kernel(x,x,kaf.kerneltype,kaf.kernelpar); C = kd./(sqrt(kx)*kaf.modict); % coherence if (max(C) <= kaf.mu0), % coherence criterion kaf.dict = [kaf.dict; x]; % order increase kaf.modict = [kaf.modict; sqrt(kx)]; kaf.alpha = [kaf.alpha; 0]; % order increase kx = kernel(x,x,kaf.kerneltype,kaf.kernelpar); kd = [kd; kx]; w = [w; 1]; end end y_est = kaf.evaluate(x); e = y - y_est; kaf.alpha = kaf.alpha + kaf.eta * kd * e; kaf.alpha = kaf.prox(kaf.lambda,kaf.alpha,w); % prune idx = find(kaf.alpha == 0); if ~isempty(idx) kaf.dict(idx,:) = []; kaf.modict(idx,:) = []; kaf.alpha(idx,:) = []; end end end methods (Static = true) % proximal operator for l1 norm function alphap = prox(lambda,alpha,w) alphap = sign(alpha).*max(abs(alpha) - lambda*w, 0); end end end kafbox-2.0/lib/klms_csl1.m000066400000000000000000000065231303712747200154550ustar00rootroot00000000000000% Kernel Least Mean Squares algorithm with Coherence-Sparsification % criterion and L1-norm regularization % % Wei Gao, Jie Chen, Cedric Richard, Jianguo Huang, and Remi Flamary, % "Kernel LMS algorithm with forward-backward splitting for dictionary % learning,", 2013 IEEE International Conference on Acoustics, Speech, and % Signal Processing (ICASSP 2013), Vancouver, Canada, March 2013. % http://dx.doi.org/10.1109/ICASSP.2013.6638763 % % Remark: Code contributed by Wei Gao and Cedric Richard. % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ classdef klms_csl1 < handle properties (GetAccess = 'public', SetAccess = 'private') % parameters eta = .1; % step-size mu0 = .95; % threshold for coherence criterion lambda = 5E-4; % sparsification threshold kerneltype = 'gauss'; % kernel type kernelpar = .5; % kernel parameter end properties (GetAccess = 'public', SetAccess = 'private') % variables dict = []; % dictionary modict = []; % modulus of the dictionary elements alpha = []; % expansion coefficients end methods function kaf = klms_csl1(parameters) % constructor if (nargin > 0) % copy valid parameters for fn = fieldnames(parameters)', if ismember(fn,fieldnames(kaf)), kaf.(fn{1}) = parameters.(fn{1}); end end end end function y_est = evaluate(kaf,x) % evaluate the algorithm if size(kaf.dict,1)>0 k = kernel(kaf.dict,x,kaf.kerneltype,kaf.kernelpar); y_est = k'*kaf.alpha; else y_est = zeros(size(x,1),1); end end function train(kaf,x,y) % train the algorithm if size(kaf.dict,2)==0 % initialize kd = kernel(x,x,kaf.kerneltype,kaf.kernelpar); kaf.dict = x; kaf.modict = sqrt(kd); kaf.alpha = 0; else kd = kernel(kaf.dict,x,kaf.kerneltype,kaf.kernelpar); kx = kernel(x,x,kaf.kerneltype,kaf.kernelpar); C = kd./(sqrt(kx)*kaf.modict); % coherence if (max(C) <= kaf.mu0), % coherence criterion kaf.dict = [kaf.dict; x]; % order increase kaf.modict = [kaf.modict; sqrt(kx)]; kaf.alpha = [kaf.alpha; 0]; % order increase kx = kernel(x,x,kaf.kerneltype,kaf.kernelpar); kd = [kd; kx]; end end y_est = kaf.evaluate(x); e = y - y_est; kaf.alpha = kaf.alpha + kaf.eta * kd * e; kaf.alpha = kaf.prox(kaf.lambda,kaf.alpha); % prune idx = find(kaf.alpha == 0); if ~isempty(idx) kaf.dict(idx,:) = []; kaf.modict(idx,:) = []; kaf.alpha(idx,:) = []; end end end methods (Static = true) % proximal operator for l1 norm function alphap = prox(lambda,alpha) alphap = sign(alpha).*max(abs(alpha) - lambda, 0); end end end kafbox-2.0/lib/knlms.m000066400000000000000000000051721303712747200147100ustar00rootroot00000000000000% Kernel Normalized Least-Mean-Square algorithm with Coherence Criterion % % C. Richard, J.C.M. Bermudez, and P. Honeine, "Online Prediction of Time % Series Data With Kernels," IEEE Transactions on Signal Processing, % vol. 57, no. 3, pp. 1058-1067, March 2009, % http://dx.doi.org/10.1109/TSP.2008.2009895 % % Remark: memories are initialized empty in this implementation % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ classdef knlms < handle properties (GetAccess = 'public', SetAccess = 'private') % parameters eta = .5; % step size mu0 = .95; % coherence criterion threshold eps = 1E-2; % regularization kerneltype = 'gauss'; % kernel type kernelpar = 1; % kernel parameter end properties (GetAccess = 'public', SetAccess = 'private') % variables dict = []; % dictionary modict = []; % modulus of the dictionary elements alpha = []; % expansion coefficients end methods function kaf = knlms(parameters) % constructor if (nargin > 0) % copy valid parameters for fn = fieldnames(parameters)', if ismember(fn,fieldnames(kaf)), kaf.(fn{1}) = parameters.(fn{1}); end end end end function y_est = evaluate(kaf,x) % evaluate the algorithm if size(kaf.dict,1)>0 k = kernel(kaf.dict,x,kaf.kerneltype,kaf.kernelpar); y_est = k'*kaf.alpha; else y_est = zeros(size(x,1),1); end end function train(kaf,x,y) % train the algorithm if size(kaf.dict,2)==0 % initialize k = kernel(x,x,kaf.kerneltype,kaf.kernelpar); kaf.dict = x; kaf.modict = sqrt(k); kaf.alpha = 0; else k = kernel(kaf.dict,x,kaf.kerneltype,kaf.kernelpar); kx = kernel(x,x,kaf.kerneltype,kaf.kernelpar); C = k./(sqrt(kx)*kaf.modict); % coherence if (max(C) <= kaf.mu0), % coherence criterion kaf.dict = [kaf.dict; x]; % order increase kaf.modict = [kaf.modict; sqrt(kx)]; kaf.alpha = [kaf.alpha; 0]; % reserve spot end end k = kernel(kaf.dict,x,kaf.kerneltype,kaf.kernelpar); kaf.alpha = kaf.alpha + ... % update coefficients kaf.eta / (kaf.eps + k'*k) * (y - k'*kaf.alpha) * k; end end end kafbox-2.0/lib/krls.m000066400000000000000000000057231303712747200145410ustar00rootroot00000000000000% Kernel Recursive Least Squares algorithm with Approximate Linear % Dependency criterion % % Y. Engel, S. Mannor, and R. Meir, "The kernel recursive least-squares % algorithm," IEEE Transactions on Signal Processing, vol. 52, no. 8, pp. % 2275-2285, Aug. 2004, http://dx.doi.org/10.1109/TSP.2004.830985 % % Remark: implementation includes a maximum dictionary size M % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ classdef krls < handle properties (GetAccess = 'public', SetAccess = 'private') nu = 1E-4; % ALD threshold M = 1000; % maximum dictionary size kerneltype = 'gauss'; % kernel type kernelpar = 1; % kernel parameter end properties (GetAccess = 'public', SetAccess = 'private') dict = []; % dictionary alpha = []; % expansion coefficients P = []; % inverse A'*A Kinv = []; % inverse kernel matrix end methods function kaf = krls(parameters) % constructor if (nargin > 0) % copy valid parameters for fn = fieldnames(parameters)', if ismember(fn,fieldnames(kaf)), kaf.(fn{1}) = parameters.(fn{1}); end end end end function y_est = evaluate(kaf,x) % evaluate the algorithm if size(kaf.dict,1)>0 k = kernel(kaf.dict,x,kaf.kerneltype,kaf.kernelpar); y_est = k'*kaf.alpha; else y_est = zeros(size(x,1),1); end end function train(kaf,x,y) % train the algorithm k = kernel([kaf.dict; x],x,kaf.kerneltype,kaf.kernelpar); kt = k(1:end-1); ktt = k(end); if numel(kt)==0, % initialize kaf.Kinv = 1/ktt; kaf.alpha = y/ktt; kaf.P = 1; kaf.dict = x; else at = kaf.Kinv*kt; % coefficients of best linear combination delta = ktt - kt'*at; % residual of linear approximation if (delta>kaf.nu && size(kaf.dict,1) 0) % copy valid parameters for fn = fieldnames(parameters)', if ismember(fn,fieldnames(kaf)), kaf.(fn{1}) = parameters.(fn{1}); end end end end function [mean_test, var_test] = evaluate(kaf,x) % evaluate if size(kaf.dict,1)>0 k = kernel(kaf.dict,x,kaf.kerneltype,kaf.kernelpar); q = kaf.Q*k; mean_test = q'*kaf.mu; % predictive mean if nargout>1 ktt = kernel(x,x,[kaf.kerneltype '-diag'],kaf.kernelpar); sf2 = ktt + kaf.jitter + sum(k.*((kaf.Q*kaf.Sigma*kaf.Q-kaf.Q)*k),1)'; sf2(sf2<0) = 0; var_test = kaf.s02*(kaf.sn2 + sf2); % predictive variance end else mean_test = zeros(size(x,1),1); % prior var_test = nan(size(x,1),1); % signal scale is unknown end end function train(kaf,x,y) % train the algorithm m = size(kaf.Sigma,1); if m<1 % initialize k = kernel(x,x,kaf.kerneltype,kaf.kernelpar); k = k + kaf.jitter; kaf.Q = 1/k; kaf.mu = y*k/(k+kaf.sn2); kaf.Sigma = k - k^2/(k+kaf.sn2); kaf.dict = x; % dictionary bases kaf.nums02ML = y^2/(k+kaf.sn2); kaf.dens02ML = 1; kaf.s02 = kaf.nums02ML / kaf.dens02ML; else % forget K = kernel(kaf.dict,kaf.dict,kaf.kerneltype,kaf.kernelpar) + kaf.jitter*eye(m); kaf.Sigma = kaf.lambda*kaf.Sigma + (1-kaf.lambda)*K; % forget kaf.mu = sqrt(kaf.lambda)*kaf.mu; % forget % predict k = kernel([kaf.dict; x],x,kaf.kerneltype,kaf.kernelpar); kt = k(1:end-1); ktt = k(end) + kaf.jitter; q = kaf.Q*kt; y_mean = q'*kaf.mu; % predictive mean gamma2 = ktt - kt'*q; gamma2(gamma2<0)=0; % projection uncertainty h = kaf.Sigma*q; sf2 = gamma2 + q'*h; sf2(sf2<0)=0; % noiseless prediction variance sy2 = kaf.sn2 + sf2; % y_var = s02*sy2; % predictive variance % include a new sample and add a basis Qold = kaf.Q; % old inverse kernel matrix p = [q; -1]; kaf.Q = [kaf.Q zeros(m,1);zeros(1,m) 0] + 1/gamma2*(p*p'); p = [h; sf2]; kaf.mu = [kaf.mu; y_mean] + (y - y_mean)/sy2*p; % posterior mean kaf.Sigma = [kaf.Sigma h; h' sf2] - 1/sy2*(p*p'); % posterior covariance m = m + 1; kaf.dict = [kaf.dict; x]; % estimate s02 via ML kaf.nums02ML = kaf.nums02ML + kaf.lambda*(y - y_mean)^2/sy2; kaf.dens02ML = kaf.dens02ML + kaf.lambda; kaf.s02 = kaf.nums02ML/kaf.dens02ML; kaf.prune = false; % delete a basis if necessary if (m>kaf.M || gamma2 end criterion = [ones(1,m-1) 0]; else % MSE pruning criterion errors = (kaf.Q*kaf.mu)./diag(kaf.Q); criterion = abs(errors); end [~, r] = min(criterion); % remove element r, which incurs in the minimum error smaller = 1:m; smaller(r) = []; if r == m, % if we must remove the element we just added, perform reduced update instead kaf.Q = Qold; kaf.reduced = true; else Qs = kaf.Q(smaller, r); qs = kaf.Q(r,r); kaf.Q = kaf.Q(smaller, smaller); kaf.Q = kaf.Q - (Qs*Qs')/qs; kaf.reduced = false; end kaf.mu = kaf.mu(smaller); kaf.Sigma = kaf.Sigma(smaller, smaller); kaf.dict = kaf.dict(smaller,:); kaf.prune = true; end end end end end kafbox-2.0/lib/lkapa.m000066400000000000000000000061131303712747200146500ustar00rootroot00000000000000% Leaky Kernel Affine Projection Algorithm % % W. Liu and J.C. Principe, "Kernel Affine Projection Algorithms", EURASIP % Journal on Advances in Signal Processing, Volume 2008, Article ID 784292, % 12 pages. http://dx.doi.org/10.1155/2008/784292 % % Remark: This implementation includes a maximum dictionary size M. With % M=Inf this algorithm is equivalent to KAPA-3 from the publication. With % M=Inf and lambda=0 it is equivalent to KAPA-1. % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ classdef lkapa < handle properties (GetAccess = 'public', SetAccess = 'private') % parameters eta = .05; % learning rate lambda = 1E-2; % Tikhonov regularization M = 1000; % maximum dictionary size P = 20; % number of regressors kerneltype = 'gauss'; % kernel type kernelpar = 1; % kernel parameter end properties (GetAccess = 'public', SetAccess = 'private') % variables xmem = []; % input memory ymem = []; % output memory dict = []; % dictionary alpha = []; % expansion coefficients end methods function kaf = lkapa(parameters) % constructor if (nargin > 0) % copy valid parameters for fn = fieldnames(parameters)', if ismember(fn,fieldnames(kaf)), kaf.(fn{1}) = parameters.(fn{1}); end end end end function y_est = evaluate(kaf,x) % evaluate the algorithm if size(kaf.dict,1)>0 k = kernel(kaf.dict,x,kaf.kerneltype,kaf.kernelpar); y_est = k'*kaf.alpha; else y_est = zeros(size(x,1),1); end end function train(kaf,x,y) % train the algorithm if size(kaf.dict,2)==0 % initialize kaf.dict = x; kaf.alpha = kaf.eta*y; kaf.xmem = x; kaf.ymem = y; else if size(kaf.dict,1) < kaf.M, if size(kaf.xmem,1) 0) % copy valid parameters for fn = fieldnames(parameters)', if ismember(fn,fieldnames(obj)), obj.(fn{1}) = parameters.(fn{1}); end end end end function y_est = evaluate(obj,x) % evaluate the algorithm if numel(obj.w)>0 y_est = x*obj.w; else y_est = zeros(size(x,1),1); end end function train(obj,x,y) % train the algorithm if numel(obj.w)==0, % initialize obj.w = zeros(length(x),1); end % Algorithm 5.2.1 in reference err = y - x*obj.w; % instantaneous error obj.w = obj.w + obj.mu*x'*err; % update filter coefficients end end end kafbox-2.0/lib/memory_cell.m000066400000000000000000000014321303712747200160660ustar00rootroot00000000000000% Dummy prediction method that always returns the last seen output. % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ classdef memory_cell < handle properties (GetAccess = 'public', SetAccess = 'private') % variables y_mem; end methods % dummy constructor function obj = memory_cell(~) end % evaluate the algorithm function y_est = evaluate(obj,x) if ~isempty(obj.y_mem) y_est = repmat(obj.y_mem,size(x,1),1); else y_est = zeros(size(x,1),1); end end % train the algorithm function train(obj,~,y) obj.y_mem = y; end end end kafbox-2.0/lib/mknlms_cs.m000066400000000000000000000066151303712747200155550ustar00rootroot00000000000000% Multikernel Normalized Least Mean Square algorithm With Coherence-Based % Sparsification (MKNLMS-CS) % % M. Yukawa, "Multikernel Adaptive Filtering," IEEE Transactions on Signal % Processing, vol.60, no.9, pp.4672,4682, Sept. 2012, % http://dx.doi.org/10.1109/TSP.2012.2200889 % % Remark: version01, August 2013 % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ classdef mknlms_cs < handle properties (GetAccess = 'public', SetAccess = 'private') % parameters delta = .95; % coherence criterion threshold eta = .5; % step size rho = 1E-2; % regularization kerneltype = 'gauss'; % kernel type kernelpars = .5:.5:2; % kernel parameters end properties (GetAccess = 'public', SetAccess = 'private') % variables dict = []; % dictionary alpha = []; % expansion coefficients, H in the original article end methods function kaf = mknlms_cs(parameters) % constructor if (nargin > 0) % copy valid parameters for fn = fieldnames(parameters)', if ismember(fn,fieldnames(kaf)), kaf.(fn{1}) = parameters.(fn{1}); end end end end function y_est = evaluate(kaf,x) % evaluate the algorithm N = size(x,1); if size(kaf.dict,1)>0 K = multikernel_dict(kaf,x); y_est = reshape(K,N,[])*kaf.alpha(:); else y_est = zeros(size(x,1),1); end end function train(kaf,x,y) % train the algorithm M = length(kaf.kernelpars); % number of distinct kernels if size(kaf.dict,2)==0 % initialize kaf.dict = x; kaf.alpha = zeros(1,M); % row with coefficients for all kernels else K = multikernel_dict(kaf,x); if (max(K(:)) <= kaf.delta), % coherence criterion kaf.dict = [kaf.dict; x]; % order increase kaf.alpha = [kaf.alpha; zeros(1,M)]; % order increase end end K = multikernel_dict(kaf,x); kaf.alpha = kaf.alpha + kaf.eta /... (kaf.rho + K(:)'*K(:)) * (y - K(:)'*kaf.alpha(:)) * ... reshape(K,size(kaf.dict,1),M); end function K = multikernel_dict(kaf,X) % multikernel for dictionary M = length(kaf.kernelpars); % number of distinct kernels N = size(X,1); D = size(kaf.dict,1); K = zeros(N,D,M); switch kaf.kerneltype case 'gauss' % RBF kernel norms1 = sum(X.^2,2); norms2 = sum(kaf.dict.^2,2); mat1 = repmat(norms1,1,D); mat2 = repmat(norms2',N,1); d2 = mat1 + mat2 - 2*X*kaf.dict'; % distance matrix Kalpha = -1./(2*kaf.kernelpars.^2); for m=1:M, K(:,:,m) = exp(d2*Kalpha(m)); end otherwise % default case for m=1:M, k = kernel(X,kaf.dict,kaf.kerneltype,kaf.kernelpars(m)); K(:,:,m) = k; end end end end end kafbox-2.0/lib/mxklms.m000066400000000000000000000100711303712747200150710ustar00rootroot00000000000000% Mixture Kernel Least Mean Square algorithm % % Mixture Kernel Least Mean Square (MXKLMS) algorithm, as proposed in R. % Pokharel, S. Seth, and J.C. Principe, "Mixture kernel least mean square," % The 2013 International Joint Conference on Neural Networks (IJCNN), % pp.1-7, 4-9 Aug. 2013. http://dx.doi.org/10.1109/IJCNN.2013.6706867 % % Remark: implementation includes a maximum dictionary size M % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ classdef mxklms < handle properties (GetAccess = 'public', SetAccess = 'private') % parameters mu = 2; % learning rate for gradient descent update of v eta = .8; % learning rate for updating alpha M = 1000; % maximum dictionary size kerneltype = 'gauss'; % kernel type kernelpars = .5:.5:2; % kernel parameters end properties (GetAccess = 'public', SetAccess = 'private') % variables dict = []; % dictionary alpha = []; % expansion coefficients matrix. rows: dict, cols: kern psi = []; % kernel weights v = []; % gate parameters used to calculate psi end methods function kaf = mxklms(parameters) % constructor if (nargin > 0) % copy valid parameters for fn = fieldnames(parameters)', if ismember(fn,fieldnames(kaf)), kaf.(fn{1}) = parameters.(fn{1}); end end end end function y_est = evaluate(kaf,x) % evaluate the algorithm if size(kaf.dict,1)>0 N = size(x,1); K = multikernel_dict(kaf,x); a = kaf.alpha*diag(kaf.psi); y_est = reshape(K,N,[])*a(:); else y_est = zeros(size(x,1),1); end end function train(kaf,x,y) % train the algorithm if (size(kaf.dict,1) 0) % copy valid parameters for fn = fieldnames(parameters)', if ismember(fn,fieldnames(kaf)), kaf.(fn{1}) = parameters.(fn{1}); end end end end function y_est = evaluate(kaf,x) % evaluate the algorithm if size(kaf.dict,1)>0 k = kernel(kaf.dict,x,kaf.kerneltype,kaf.kernelpar); y_est = k'*kaf.alpha; else y_est = zeros(size(x,1),1); end end function train(kaf,x,y) % train the algorithm if size(kaf.dict,2)==0 % initialize kaf.dict = x; kaf.alpha = kaf.eta*y; kaf.xmem = x; kaf.ymem = y; else if size(kaf.dict,1) < kaf.M, if size(kaf.xmem,1) % prefer inv to \ to avoid instability end end end end end kafbox-2.0/lib/nlms.m000066400000000000000000000026271303712747200145370ustar00rootroot00000000000000% Normalized Least Mean Squares algorithm % % From A. H. Sayed, "Fundamentals of adaptive filtering}", Wiley-IEEE % Press, 2003, Chapter 5. % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ classdef nlms < handle properties (GetAccess = 'public', SetAccess = 'private') mu = .9; % step size eps = 1E-6; % regularization end properties (GetAccess = 'public', SetAccess = 'private') w = []; end methods function obj = nlms(parameters) % constructor if (nargin > 0) % copy valid parameters for fn = fieldnames(parameters)', if ismember(fn,fieldnames(obj)), obj.(fn{1}) = parameters.(fn{1}); end end end end function y_est = evaluate(obj,x) % evaluate the algorithm if numel(obj.w)>0 y_est = x*obj.w; else y_est = zeros(size(x,1),1); end end function train(obj,x,y) % train the algorithm if numel(obj.w)==0, % initialize obj.w = zeros(length(x),1); end % Algorithm 5.6.1 in reference err = (y-x*obj.w); obj.w = obj.w + obj.mu/(obj.eps + x*x')*x'*err; end end end kafbox-2.0/lib/norma.m000066400000000000000000000041201303712747200146700ustar00rootroot00000000000000% Naive Online regularized Risk Minimization Algorithm % % J. Kivinen, A.J. Smola, and R.C. Williamson, "Online learning with % kernels," IEEE Transactions on Signal Processing, vol. 52, no. 8, % pp. 2165-2176, Aug. 2004, http://dx.doi.org/10.1109/TSP.2004.830991 % % Remark: using squared loss function % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ classdef norma < handle properties (GetAccess = 'public', SetAccess = 'private') tau = 500; % memory size (terms retained in truncation) lambda = 1E-2; % regularization parameter eta = .5; % learning rate tcoeff = 0; % learning rate coefficient: eta_t = eta * t^tcoeff kerneltype = 'gauss'; % kernel type kernelpar = 1; % kernel parameter end properties (GetAccess = 'public', SetAccess = 'private') dict = []; % dictionary alpha = []; % expansion coefficients end methods function kaf = norma(parameters) % constructor if (nargin > 0) % copy valid parameters for fn = fieldnames(parameters)', if ismember(fn,fieldnames(kaf)), kaf.(fn{1}) = parameters.(fn{1}); end end end end function y_est = evaluate(kaf,x) % evaluate the algorithm if size(kaf.dict,1)>0 k = kernel(kaf.dict,x,kaf.kerneltype,kaf.kernelpar); y_est = k'*kaf.alpha; else y_est = zeros(size(x,1),1); end end function train(kaf,x,y) % train the algorithm kaf.alpha = (1-kaf.lambda*kaf.eta)*kaf.alpha; y_est = kaf.evaluate(x); err = y - y_est; kaf.alpha = [kaf.alpha; kaf.eta*err]; % grow kaf.dict = [kaf.dict; x]; % grow if length(kaf.alpha)>kaf.tau kaf.alpha(1) = []; % prune kaf.dict(1,:) = []; % prune end end end end kafbox-2.0/lib/phypass.m000066400000000000000000000120271303712747200152500ustar00rootroot00000000000000% Parallel HYperslab Projection along Affine SubSpace (PHYPASS) algorithm % % M. Takizawa and M. Yukawa, "An Efficient Data-Reusing Kernel Adaptive % Filtering Algorithm Based on Parallel Hyperslab Projection Along Affine % Subspace," in Proc. ICASSP, May. 2013, pp.3557-3561. % % Remark: version01, August 2013 % Contributor for this code: Masa-aki Takizawa % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ classdef phypass < handle properties (GetAccess = 'public', SetAccess = 'private') mu = 0.5; % step size s = 1; % number of update coefficients (dictionary size) sigma = 0.95; % threshold for dictionary p = 8; % number of hyper slabs kerneltype = 'gauss'; % kernel type kernelpar = 1; % kernel parameter omega = 1/8; % weight end properties (GetAccess = 'public', SetAccess = 'private') mem = []; % memory dict = []; % dictionary alpha = []; % expansion coefficients d = []; % output window end methods function kaf = phypass(parameters) % constructor if (nargin > 0) % copy valid parameters for fn = fieldnames(parameters)', if ismember(fn,fieldnames(kaf)), kaf.(fn{1}) = parameters.(fn{1}); end end end end function y_est = evaluate(kaf,x) % evaluate the algorithm if size(kaf.dict,1)>0 k = kernel(kaf.dict,x,kaf.kerneltype,kaf.kernelpar); y_est = k'*kaf.alpha; else y_est = zeros(size(x,1),1); end end function train(kaf,x,y) % train the algorithm if (length(kaf.d) < kaf.p) % grow the memory kaf.mem = [kaf.mem; x]; kaf.d = [kaf.d; y]; else % sliding window kaf.mem = [kaf.mem(2:end,:); x]; kaf.d = [kaf.d(2:end); y]; end if size(kaf.dict,1) == 0 % initialize kaf.dict = x; kaf.alpha = 0; else k = kernel(x,kaf.dict,kaf.kerneltype,kaf.kernelpar); if max(k) < kaf.sigma % coherence criterion kaf.dict = [kaf.dict ; x]; % add the current input x kaf.alpha = [kaf.alpha ; 0]; end end num = length(kaf.d); % length of memory num_d = min(size(kaf.dict,1),kaf.s); % number of dictionary elements to update dict_id = zeros(num_d,num); % memory of dictionary index Kmemdict = kernel(kaf.mem,kaf.dict,kaf.kerneltype,kaf.kernelpar); Kdict = kernel(kaf.dict,kaf.dict,kaf.kerneltype,kaf.kernelpar); for k=1:num, d_check = Kmemdict(k,:); % kernel between k'th memory element and full dictionary [mm,ii] = sort(d_check,'descend'); %#ok dict_id(:,k) = ii(1:num_d); % memory indexes end k_n = zeros(num_d,num_d,num); y_n = zeros(num_d,num); alpha_new = zeros(num_d,num); for k=1:num k_n(:,:,k) = Kdict(dict_id(:,k),dict_id(:,k)); y_n(:,k) = Kmemdict(k,dict_id(:,k))'; alpha_new(:,k) = k_n(:,:,k)\y_n(:,k); % compute projections onto dictionary subspaces end ln_nume = 0; numerator = zeros(num,1); denominator = zeros(num,1); beta = zeros(num,1); for k=1:num, numerator(k) = Kmemdict(k,:)*kaf.alpha; denominator(k) = Kmemdict(k,dict_id(:,k))*alpha_new(:,k); beta(k) = (kaf.d(k) - numerator(k))/(denominator(k)); % progress of the projection onto hyperplanes ln_nume = ln_nume + kaf.omega * beta(k)^2 *... alpha_new(:,k)' * k_n(:,:,k) * alpha_new(:,k); % numerator of the extrapolation coefficcient end G = zeros(num,num); for k = 1:num, Grow = alpha_new(:,k)'*Kdict(dict_id(:,k),dict_id); sti = 1; ndi = num_d; for l = 1:num, G(k,l) = Grow(1,sti:ndi)*alpha_new(:,l); sti = ndi + 1; ndi = ndi + num_d; end end ln_deno = kaf.omega^2 * beta' * G * beta; % denominator of the extrapolation coefficient ln = ln_nume/(ln_deno); % extrapolation coefficient alpha_new2 = zeros(length(kaf.alpha),num); for k=1:num alpha_new2(dict_id(:,k),k) = alpha_new(:,k); end for k=1:num kaf.alpha = kaf.alpha + kaf.mu * ln *... kaf.omega * beta(k) * alpha_new2(:,k); % update expansion coefficients end end end end kafbox-2.0/lib/problms.m000066400000000000000000000036521303712747200152430ustar00rootroot00000000000000% Probabilistic Least-Mean-Squares Filter % % Jesus Fernandez-Bes, Victor Elvira, and Steven Van Vaerenbergh, "A % probabilistic least-mean-squares filter," 2015 IEEE International % Conference on Acoustics, Speech and Signal Processing (ICASSP), 2015. % http://dx.doi.org/10.1109/ICASSP.2015.7178361 % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ classdef problms < handle properties (GetAccess = 'public', SetAccess = 'private') sigma2_n = 1E-6; % variance of observation noise sigma2_d = 1E-4; % variance of filter weight diffusion lambda = 1; % forgetting factor end properties (GetAccess = 'public', SetAccess = 'private') w = []; sigma2_k = 1E-2; end methods function obj = problms(parameters) % constructor if (nargin > 0) % copy valid parameters for fn = fieldnames(parameters)', if ismember(fn,fieldnames(obj)), obj.(fn{1}) = parameters.(fn{1}); end end end end function y_est = evaluate(obj,x) % evaluate the algorithm if numel(obj.w)>0 y_est = x*obj.w; else y_est = zeros(size(x,1),1); end end function train(obj,x,y) % train the algorithm m = length(x); if numel(obj.w)==0, % initialize obj.w = zeros(m,1); end x_norm2 = x*x'; K_k = (obj.sigma2_k + obj.sigma2_d) / ... ((obj.sigma2_k + obj.sigma2_d) * x_norm2 + obj.sigma2_n); err = y - obj.lambda*x*obj.w; obj.w = obj.lambda*obj.w + K_k*x'*err; obj.sigma2_k = (1 - K_k*x_norm2/m)*... (obj.sigma2_k + obj.sigma2_d); end end end kafbox-2.0/lib/profiler/000077500000000000000000000000001303712747200152235ustar00rootroot00000000000000kafbox-2.0/lib/profiler/exkrls_profiler.m000066400000000000000000000053601303712747200206170ustar00rootroot00000000000000% Profiler extension for Extended Kernel Recursive Least Squares algorithm % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ classdef exkrls_profiler < exkrls properties (GetAccess = 'public', SetAccess = 'private') elapsed = 0; % elapsed time prev_dict_size = 0; % previous dictionary size for growth check end methods function kaf = exkrls_profiler(parameters) % constructor if nargin<1, parameters = struct(); end kaf = kaf@exkrls(parameters); end function flops = lastflops(kaf) % flops for last iteration m = size(kaf.dict,1); if kaf.prev_dict_size < m, % growing m1 = m; m2 = m - 1; floptions = struct(... 'sum', 1 + m2 - 1 + m2 + 1 + m2 + 2*m2 - 1 + 1 + m2^2 + m2 - 1, ... 'mult', m2 + m2 + 2 + m2 + m2 + 2 + 4 + 2*m2^2 + m2 + 2, ... 'div', 1 + 1, ... sprintf('%s_kernel',kaf.kerneltype), [m1, 1, size(kaf.dict,2)]); else floptions = struct('sum', 1); % no kernel calculation end flops = kflops(floptions); end %% flops breakdown % k = kernel([kaf.dict; x],x,kaf.kerneltype,kaf.kernelpar); % kernel: m1 % kaf.i = kaf.i + 1; % sum: 1 % z = kaf.Q*kt; % sum: m2 - 1 % mult: m2 % r = kaf.beta^kaf.i*kaf.rho + ktt - kt'*z; % sum: m2 + 1 % mult: m2 + 2 % assumes previous kaf.beta^kaf.i is stored % err = y - kt'*kaf.alpha; % sum: m2 % mult: m2 % kaf.alpha = kaf.alphaf*[kaf.alpha - z*err/r; err/r]; % sum: 2*m2 - 1 % mult: m2 + 2 % div: 1 % dummy = kaf.alphaf^2 + kaf.beta^kaf.i*kaf.q*kaf.rho; % sum: 1 % mult: 4 % kaf.rho = kaf.rho/dummy; % div: 1 % kaf.Q = kaf.alphaf^2/(r*dummy)*[kaf.Q*r + z*z', -z; -z', 1]; % sum: m2^2 + m2 - 1 % mult: 2*m2^2 + m2 + 2 % div: 1 %% function train_profiled(kaf,x,y) kaf.prev_dict_size = size(kaf.dict,1); t1 = tic; kaf.train(x,y); t2 = toc(t1); kaf.elapsed = kaf.elapsed + t2; end function bytes = lastbytes(kaf) % bytes used in last iteration m = size(kaf.dict,1); bytes = 8*(m^2 + m + 1 + m*size(kaf.dict,2)) + 4; % 8 bytes for double precision, 4 for uint32 % Q, alpha, rho, dict, i end end end kafbox-2.0/lib/profiler/fbkrls_profiler.m000066400000000000000000000062501303712747200205710ustar00rootroot00000000000000% Profiler extension for Fixed-Budget Kernel Recursive Least Squares % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ classdef fbkrls_profiler < fbkrls properties (GetAccess = 'public', SetAccess = 'private') elapsed = 0; % elapsed time prev_dict_size = 0; % previous dictionary size for growth check end methods function kaf = fbkrls_profiler(parameters) % constructor if nargin<1, parameters = struct(); end kaf = kaf@fbkrls(parameters); end function flops = lastflops(kaf) % flops for last iteration m = size(kaf.dict,1); if kaf.prev_dict_size < m, % growing m1 = m; m2 = m - 1; m4 = m; floptions = struct(... 'sum', m2^2 + m2^2 - m2 + m2^2 + m^2 - m + m4^2 - m4, ... 'mult', m2^2 + m2 + m2^2 + m^2 + m4^2, ... 'div', 1, ... sprintf('%s_kernel',kaf.kerneltype), [m1,1,size(kaf.dict,2)]); else m1 = m + 1; m2 = m; m3 = m; m4 = m + 1; m5 = m + 1; floptions = struct(... 'sum', m2^2 + m2^2 - m2 + m2^2 + m3^2 + m^2 - m + m4^2 - m4, ... 'mult', m2^2 + m2 + m2^2 + m3^2 + m3 + m^2 + m4^2, ... 'div', 1 + 1 + m5, ... sprintf('%s_kernel',kaf.kerneltype), [m1,1,size(kaf.dict,2)]); end flops = kflops(floptions); end %% flops breakdown % k = kernel(kaf.dict,x,kaf.kerneltype,kaf.kernelpar); % grow Kinv % kernel: m1 % d = k(end) + kaf.c; % grow Kinv % sum: 1 % g_inv = d - b'*kaf.Kinv*b; % grow Kinv % sum: m2^2 % mult: m2^2 + m2 % g = 1/g_inv; % grow Kinv % div: 1 % f = -kaf.Kinv*b*g; % grow Kinv % sum: m2^2 - m2 % mult: m2^2 + m2 % E = kaf.Kinv - kaf.Kinv*b*f'; % grow Kinv % sum: m2^2 % mult: m2^2 % kaf.alpha = kaf.Kinv*kaf.dicty; % before prune check % sum: m4^2 - m4 % prod: m4^2 % err_ap = abs(kaf.alpha)./diag(kaf.Kinv); % finding index to prune % div: m5 % kaf.Kinv = G - f*f'/e; % prune Kinv % sum: m3^2 % mult: m3^2 + m3 % div: 1 % kaf.alpha = kaf.Kinv*kaf.dicty; % end of training % sum: m^2 - m % prod: m^2 %% function train_profiled(kaf,x,y) kaf.prev_dict_size = size(kaf.dict,1); t1 = tic; kaf.train(x,y); t2 = toc(t1); kaf.elapsed = kaf.elapsed + t2; end function bytes = lastbytes(kaf) % bytes used in last iteration m = size(kaf.dict,1); bytes = 8*(m^2 + m + m + m*size(kaf.dict,2)); % 8 bytes for double precision % Kinv, alpha, dicty, dict end end end kafbox-2.0/lib/profiler/kafbox_profiler.m000066400000000000000000000035571303712747200205670ustar00rootroot00000000000000% KAFBOX_PROFILER Profiler program. Runs through different algorithms, each % with a number of configurations, each for a number of simulations. % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ function [data,algorithms,results] = ... kafbox_profiler(data,sim_opts,algorithms,outputdir) num_algo = length(algorithms); numsim = 1; if isfield(sim_opts,'numsim'), numsim = sim_opts.numsim; end % pre-allocate structure results = cell(num_algo,1); for algo_ind = 1:num_algo, num_config = length(algorithms{algo_ind}.options.sweep_val); algo_results = cell(num_config,1); for config_ind = 1:num_config, algo_results{config_ind} = cell(numsim,1); end results{algo_ind} = algo_results; end for sim_ind = 1:numsim, % process simulations fprintf('SIM %2d\n',sim_ind); for algo_ind = 1:num_algo % process algorithms algo = algorithms{algo_ind}; sweep_par = algo.options.sweep_par; sweep_val = algo.options.sweep_val; num_config = length(sweep_val); % number of iterations in sweep for config_ind = 1:num_config, % process configurations algo_config = algo; fprintf('%9s %10s: ',algo_config.name,... struct2str(struct(sweep_par,sweep_val(config_ind)))); % set sweep parameter value eval(sprintf('algo_config.options.%s = sweep_val(config_ind);',sweep_par)); % perform one simulation for this configuration simresults = kafbox_profiler_simulation(data,sim_opts,... algo_config,sim_ind,outputdir); results{algo_ind}{config_ind}{sim_ind} = simresults; fprintf(' %.2fs',simresults.elapsed); fprintf('\n'); end end fprintf('\n'); end kafbox-2.0/lib/profiler/kafbox_profiler_convergence_analysis.m000066400000000000000000000045611303712747200250440ustar00rootroot00000000000000% Convergence analysis. Returns steady state MSE, numbers of iterations to % reach specified MSE value and MSE reached after specified number of % iterations. % % Input: % - MSE_curve: array containing MSE curve % - target_mse: target MSE value % - target_it: target number of iterations % % Output: % - ss: steady-state MSE level calculated over the last 1000 samples % - it_reached: number of iterations it takes to reach the specified MSE % - mse_reached: mse reached after the specified number of iterations % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ function [ss,it_reached,mse_reached] = ... kafbox_profiler_convergence_analysis(MSE_curve, target_mse, target_it) xs = find(~isnan(MSE_curve)); % calculate steady state as MSE over last 1000 points ss = 10*log10(mean(MSE_curve(xs(xs>length(MSE_curve)-1000)))); if nargin > 1 % check when error_measure reaches certain MSE xsi = find(10*log10(MSE_curve(xs)) < target_mse); if numel(xsi)>0 if xsi(1)==1 index = xsi(1); else % linear interpolation to find value sti = xsi(1)-1; ndi = xsi(1); stv = 10*log10(MSE_curve(xs(sti))); ndv = 10*log10(MSE_curve(xs(ndi))); index = xs(sti)+(xs(ndi) - xs(sti))*(stv-target_mse)/(stv-ndv); % plot(index,target_MSE,'ro') % plot([0 index],[target_MSE target_MSE],'r','LineWidth',2) % ylim = get(gca,'YLim'); % plot([index index],[ylim(1) target_MSE],'r') end else % fprintf('No convergence to %d dB.\n',target_MSE); index = nan; end it_reached = index; end if nargin > 2 % check what error is reached after certain number of iterations xsi = find((xs) >= target_it); if xsi(1)==1 value = 10*log10(MSE_curve(xsi(1))); else % linear interpolation to find value sti = xsi(1)-1; ndi = xsi(1); stv = 10*log10(MSE_curve(xs(sti))); ndv = 10*log10(MSE_curve(xs(ndi))); value = stv + (ndv - stv)*(xs(sti)-target_iter)/(xs(sti)-xs(ndi)); % plot(target_iter,value,'ko') % plot([0 target_iter],[value value],'k') % ylim = get(gca,'YLim'); % plot([target_iter target_iter],[ylim(1) value],'k','LineWidth',2) end mse_reached = value; end kafbox-2.0/lib/profiler/kafbox_profiler_msecurves.m000066400000000000000000000031731303712747200226550ustar00rootroot00000000000000% Proces MSE results into MSE curves: average out and/or limit to a % subset of temporal indices. % % Input: % - results: a configresults cell produced by kafbox_profiler. % - inds: array of indices for which to calculate output % % Output: % - MSE_avg_setups: cell containing averaged out MSE curves, structure % corresponds to "results" cell structure. % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ function MSE_avg_setups = kafbox_profiler_msecurves(results,inds) num_setups = length(results); MSE_avg_setups = cell(num_setups,1); for setup_ind = 1:num_setups, setup_results = results{setup_ind}; num_configs = length(setup_results); MSE_avg_configs = cell(num_configs,1); for config_ind = 1:num_configs, config_results = setup_results{config_ind}; if isfield(config_results{1},'NMSE') N = length(config_results{1}.NMSE); % temporary else N = length(config_results{1}.MSE); end MSE = zeros(N,1); num_sim = length(config_results); for sim_ind = 1:num_sim, simresults = config_results{sim_ind}; if isfield(simresults,'NMSE') % temporary simresults.MSE = simresults.NMSE; end if nargin<2 inds = 1:length(simresults.MSE); end mm = min(length(simresults.MSE(inds)),N); MSE = MSE(1:mm) + simresults.MSE(inds(1:mm))/num_sim; end MSE_avg_configs{config_ind} = MSE; end MSE_avg_setups{setup_ind} = MSE_avg_configs; end kafbox-2.0/lib/profiler/kafbox_profiler_plotconvergence.m000066400000000000000000000017311303712747200240340ustar00rootroot00000000000000function [f,h] = kafbox_profiler_plotconvergence(algorithms,... mse_curves,resinds) figure; hold all set(gcf,'Position',[200, 200, 500 300]) titles = cell(length(algorithms),1); for i=1:size(resinds,1) algo = algorithms{resinds(i,1)}; ls = '-'; % line style if isfield(algo.figstyle,'line') ls = algo.figstyle.line; end lw = 1; % line width if isfield(algo.figstyle,'linewidth') lw = algo.figstyle.linewidth; end curve = mse_curves{resinds(i,1)}{resinds(i,2)}; xs= ~isnan(curve); inds = 1:length(mse_curves{resinds(i,1)}{resinds(i,2)}); plot(inds(xs),10*log10(curve(xs)),'color',algo.figstyle.color,... 'LineWidth',lw,'LineStyle',ls) titles{i} = algo.name; end h = legend(titles); grid on; box on xlabel('iteration'); ylabel('(N)MSE'); % Expand figure axes f = gcf; style = hgexport('factorystyle'); style.Bounds = 'tight'; hgexport(f,'-clipboard',style,'applystyle', true); drawnow; kafbox-2.0/lib/profiler/kafbox_profiler_plotresults.m000066400000000000000000000133611303712747200232410ustar00rootroot00000000000000% Produces a plot showing the performance trade-off for two chosen % performance measures for different algorithms. Includes a data cursor % showing extra data. % % Input: % - algorithms: cell containing algorithm setups % - mse_curves: cell containing MSE curves (possibly averaged out) % - results: cell containing complete results % - pmeasures: 2x1 cell containing two performance measures % - opt: optional parameters / options for performance measures % % Output: % - f: figure axis handle % - h: legend axis handle % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ function [f,h] = kafbox_profiler_plotresults(algorithms,mse_curves,... results,pmeasures,opt) logaxis = [0 0]; labels = cell(2); for i=1:2 switch pmeasures{i} case 'bytes' labels{i} = 'max bytes'; logaxis(i) = 1; case 'flops' labels{i} = 'max flops'; logaxis(i) = 1; case 'ssmse' labels{i} = 'steady-state MSE'; case 'ssplus' labels{i} = sprintf('time to ss+%d dB',opt); logaxis(i) = 1; case 'timeto' labels{i} = sprintf('time to %d dB',opt); logaxis(i) = 1; case 'timetopct' % time to fall to percentage of steady state labels{i} = sprintf('time to %d%% of steady-state',opt); logaxis(i) = 1; otherwise error('undefined case') end end f = figure; hold all %#ok set(gcf,'Position',[200, 200, 500 300]) starts = zeros(length(algorithms),2); titles = cell(length(algorithms),1); for algo_ind = 1:length(algorithms) algo = algorithms{algo_ind}; algo_results = results{algo_ind}; xdata = zeros(length(algo_results),1); ydata = zeros(length(algo_results),1); num_algo = length(algo_results); for config_ind = 1:num_algo config_results = algo_results{config_ind}; numsim = length(config_results); flops = 0; bytes = 0; for sim_ind = 1:numsim, sim_results = config_results{sim_ind}; bytes = max(bytes,sim_results.bytes); flops = max(flops,sim_results.flops); end mse_curve = mse_curves{algo_ind}{config_ind}; ss = kafbox_profiler_convergence_analysis(mse_curve); r.flops = flops; r.bytes = bytes; r.ssmse = ss; clear target_mse for i=1:2, switch pmeasures{i} case 'ssplus' target_mse = ss + opt; case 'timeto' target_mse = opt; case 'timetopct' target_mse = ss * opt/100; end if exist('target_mse','var') [ss,tt] = kafbox_profiler_convergence_analysis(... mse_curve,target_mse); r.(pmeasures{i}) = tt; end end xdata(config_ind) = r.(pmeasures{1}); ydata(config_ind) = r.(pmeasures{2}); end titles{algo_ind} = algo.name; % tooltip data uservalues = algo.options.sweep_val; userstrings.method = algo.name; userstrings.param = algo.options.sweep_par; userstrings.x = xdata; userstrings.y = ydata; userstrings.xaxis = labels{1}; userstrings.yaxis = labels{2}; xxi = xdata; if logaxis(1), xxi = log10(xdata); end yyi = ydata; if logaxis(2), yyi = log10(ydata); end starts(algo_ind,:) = [xxi(1),yyi(1)]; hLine = line(xxi, yyi,... 'Marker', algo.figstyle.marker,... 'MarkerSize', 8,... 'Color', algo.figstyle.color,... 'UserData', {uservalues, userstrings},... 'LineStyle','-',... 'LineWidth',2); %#ok cursorMode = datacursormode(gcf); set(cursorMode, 'enable','on', 'UpdateFcn',@setDataTipTxt); if verLessThan('matlab', '8.4') set(cursorMode, 'NewDataCursorOnClick',false); end end h = legend(titles); grid on; box on xlabel(labels{1}); ylabel(labels{2}); if logaxis(1), xtl = get(gca,'XTickLabel'); set(gca,'XTickLabel',str2log10(xtl)); end if logaxis(2), ytl = get(gca,'YTickLabel'); set(gca,'YTickLabel',str2log10(ytl)); end % Expand figure axes f = gcf; style = hgexport('factorystyle'); style.Bounds = 'tight'; hgexport(f,'-clipboard',style,'applystyle', true); drawnow; % plot start points for algo_ind=1:length(algorithms) xi = starts(algo_ind,1); yi = starts(algo_ind,2); yyi = yi; hLine = line(xi(1), yyi(1),... 'Marker', 'o', 'MarkerSize', 6,... 'Color', 'black', 'MarkerFaceColor','black'); %#ok end function output_txt = setDataTipTxt(~,event_obj) % ~ Currently not used (empty) % event_obj Object containing event data structure % output_txt Data cursor text (string or cell array of strings) p = event_obj.Position; t = event_obj.Target; ud = get(t,'UserData'); % get index in data series by looking up closest point p_all = [get(t,'XData')', get(t,'YData')']; N = size(p_all,1); d = p_all-repmat(p,N,1); d2 = sum(d.^2,2); [mm,ii] = min(d2); %#ok uservalue = ud{1}(ii); userstrings = ud{2}; % fill text tooltip output_txt{1} = userstrings.method; output_txt{2} = print_formatted_value(userstrings.param,uservalue); output_txt{3} = print_formatted_value(userstrings.xaxis,userstrings.x(ii)); output_txt{4} = print_formatted_value(userstrings.yaxis,userstrings.y(ii)); function str = print_formatted_value(s,v) if (v-round(v))^2<1E-8, str = sprintf('%s = %d',s,v); else str = sprintf('%s = %.2f',s,v); end % add log10 to string numbers function b = str2log10(a) b = cell(length(a),1); for i=1:length(a), if isa(a,'cell') el = a{i}; else el = a(i,:); end b{i} = ['1E' strtrim(el)]; end kafbox-2.0/lib/profiler/kafbox_profiler_simulation.m000066400000000000000000000054751303712747200230340ustar00rootroot00000000000000% Perform a single profiler simulation. % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ function simresults = kafbox_profiler_simulation(data,sim_opts,... algo_config,sim_ind,output_dir) simdata = data; if isfield(data,'offset') simdata.offset = (sim_ind-1)*data.offset; elseif isfield(data,'permutation') simdata.permutation = (sim_ind-1)*data.permutation; % start with 0 = no permutation else % fix random seed per simulation rs = sim_ind; rng('default'); rng(sim_ind); simdata.rs = rs; end % check if results for this simulation have been stored before simresults = kafbox_profiler_storet(simdata,algo_config,output_dir); if isempty(simresults), % perform simulation [X,Y,X_test,Y_test] = kafbox_data(simdata); % load data every = 1; if isfield(data,'test_every') every = data.test_every; end N = size(X,1); all_fl = zeros(N,1); % flops per iteration all_bytes = zeros(N,1); % flops per iteration erm = sim_opts.error_measure; % error measure eval(sprintf('%s = nan*zeros(N,1);',erm)); var_test = var(Y_test); % for NMSE kaf = feval(sprintf('%s_profiler',algo_config.class),algo_config.options); results2store = {'elapsed','flops','bytes',erm}; y_test_ind = ones(N,1); if isfield(data,'N_switch') y_test_ind(data.N_switch:end) = 2; end for i=1:N, Y_est = kaf.evaluate(X(i,:)); % predict kaf.train_profiled(X(i,:),Y(i)); % train all_fl(i) = kaf.lastflops(); all_bytes(i) = kaf.lastbytes(); if isfield(data,'test_every_conv') de = data.test_every_conv; if i<10*data.test_every_conv % start from 1 at i=1 and go exponentially to every at % i=10*every every = max(1,round(10^(log10(de)/10/de*i))); else every = data.test_every_conv; end end if mod(i,every) == 0 if isempty(X_test) Y_test = Y(i); % test prediction else Y_est = kaf.evaluate(X_test); % test on multiple data end erm_val = mean((Y_test(:,y_test_ind(i)) - Y_est).^2); if strcmp(erm,'NMSE'), erm_val = erm_val/var_test; %#ok end eval(sprintf('%s(i) = erm_val;',erm)) end end elapsed = kaf.elapsed; %#ok flops = max(all_fl); %#ok bytes = max(all_bytes); %#ok for j=1:length(results2store) eval(sprintf('simresults.%s = %s;',results2store{j},results2store{j})) end kafbox_profiler_storet(simdata,algo_config,output_dir,simresults); fprintf('calculated'); else fprintf('retrieved '); end kafbox-2.0/lib/profiler/kafbox_profiler_storet.m000066400000000000000000000046741303712747200221700ustar00rootroot00000000000000% Handles storage and retrieval of profiler results. % % Checks if the results for a given configuration were stored before, and % saves new results if necessary. % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ function results = kafbox_profiler_storet(data,config,output_dir,results) if nargin<4 option = 'check'; else option = 'save'; end dataset = lower(data.name); if isfield(data,'class') dataset = sprintf('%s_%s',lower(data.class),dataset); end dataset_algo = [dataset '_' config.class]; results_path = fullfile(output_dir,dataset_algo); index_path = fullfile(results_path,'index.mat'); if ~exist(results_path,'file') mkdir(results_path); end % all relevant values in one array my_config = config.options; fn = fieldnames(data); for i = 1:length(fn), if ~strcmp(fn{i},'name') my_config.(sprintf('data_%s',fn{i})) = data.(fn{i}); end end skiplist = {'sweep_par','sweep_val','data_numsim'}; for i = 1:length(skiplist), if isfield(my_config,skiplist{i}) my_config = rmfield(my_config,skiplist{i}); end end switch option case 'check' if exist(index_path,'file')==2 load(index_path); % loads "configs" results = find_results(my_config,configs,results_path); %#ok else results = []; end case 'save' if exist(index_path,'file')==2 load(index_path); % loads "configs" ind = length(configs)+1; %#ok else ind = 1; end id = datestr(now,30); configs{ind}.id = id; configs{ind}.cstr = struct2str(my_config); save(index_path,'configs'); save_results(results,results_path,id) otherwise error('unknown option') end % check if config was already processed and return corresponding results function r = find_results(my_config,configs,results_path) str = struct2str(my_config); for i=1:length(configs), stri = configs{i}.cstr; if strcmp(str,stri), id = configs{i}.id; r = load_results(results_path,id); return end end r = []; function save_results(results,results_path,id) %#ok fname = sprintf('%s/%s',results_path,id); save(fname,'results'); function results = load_results(results_path,id) fname = sprintf('%s/%s.mat',results_path,id); if exist(fname,'file')==2 r = load(fname); results = r.results; else results = []; end kafbox-2.0/lib/profiler/kafbox_template_profiler.m000066400000000000000000000031341303712747200224510ustar00rootroot00000000000000% Profiler extension template. % % Profiler extension for [algorithm] % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ classdef kafbox_template_profiler < kafbox_template properties (GetAccess = 'public', SetAccess = 'private') elapsed = 0; % elapsed time end methods function kaf = kafbox_template_profiler(parameters) % constructor if nargin<1, parameters = struct(); end kaf = kaf@kafbox_template(parameters); end function flops = lastflops(kaf) % flops for last iteration % numbers of operations m1 = 1; m2 = 1; m3 = 1; m4 = 1; floptions = struct(... 'sum', m1, ... 'mult', m2, ... 'div', m3, ... sprintf('%s_kernel',kaf.kerneltype), [m4,1,size(kaf.dict,2)]); flops = kflops(floptions); end %% flops breakdown % [space for remarks on number of operations used above] %% function train_profiled(kaf,x,y) t1 = tic; kaf.train(x,y); t2 = toc(t1); kaf.elapsed = kaf.elapsed + t2; end function bytes = lastbytes(kaf) % bytes used in last iteration m = size(kaf.dict,1); m2 = 1; bytes = (m2 + m*size(kaf.dict,2)); % 8 bytes for double precision % [list variables] end end end kafbox-2.0/lib/profiler/kap_profiler.m000066400000000000000000000052221303712747200200570ustar00rootroot00000000000000% Profiler extension for Kernel Affine Projection algorithm with Coherence % Criterion % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ classdef kap_profiler < kap properties (GetAccess = 'public', SetAccess = 'private') elapsed = 0; % elapsed time prev_dict_size = 0; % previous dictionary size for growth check end methods function kaf = kap_profiler(parameters) % constructor if nargin<1, parameters = struct(); end kaf = kaf@kap(parameters); end function flops = lastflops(kaf) % flops for last iteration m = size(kaf.dict,1); if kaf.prev_dict_size < m, % growing m1 = m - 1; m2 = m - 1; else m1 = m; m2 = m; end m3 = size(kaf.memx,1); p2 = size(kaf.memx,1); floptions = struct(... 'sum', m2 + p2^2*m2 - p2^2 + m2*(p2-1) + p2*m2 + (p2-1)*p2*(2*p2-1)/6 - (p2-1)*p2/2, ... 'mult', m2*p2 + p2^2*m2 + p2*p2 + p2*m2 + (p2-1)*p2*(2*p2-1)/6 - (p2-1)*p2/2, ... 'div', p2 - 1, ... sprintf('%s_kernel',kaf.kerneltype), [m1+m3*m2, 1, size(kaf.dict,2)]); flops = kflops(floptions); end %% flops breakdown % k = kernel(x,kaf.dict,kaf.kerneltype,kaf.kernelpar); % kernel: m1 % kernel(kaf.memx,kaf.dict,kaf.kerneltype,kaf.kernelpar); % kernel: m3*m2 % kaf.alpha = kaf.alpha + ... % kaf.eta*H'/... % (kaf.eps*eye(size(H,1)) + H*H')*... % (kaf.memd - H*kaf.alpha); % sum: m2 + p2^2*m2 - p2^2 + m2*(p2-1) + p2*m2 % without division % mult: m2*p2 + p2^2*m2 + p2*p2 + p2*m2 % without division % matrix division in previous operation, assuming Gaussian elimination % sum: (p2-1)*p2*(2*p2-1)/6 - (p2-1)*p2/2 % mult: (p2-1)*p2*(2*p2-1)/6 - (p2-1)*p2/2 % div: p2 - 1 %% function train_profiled(kaf,x,y) kaf.prev_dict_size = size(kaf.dict,1); t1 = tic; kaf.train(x,y); t2 = toc(t1); kaf.elapsed = kaf.elapsed + t2; end function bytes = lastbytes(kaf) % bytes used in last iteration m = size(kaf.dict,1); bytes = 8*(m + m*size(kaf.dict,2) + kaf.p + kaf.p*size(kaf.memx,2)); % 8 bytes for double precision % alpha, dict, memd, memx end end end kafbox-2.0/lib/profiler/kflops.m000066400000000000000000000026371303712747200167070ustar00rootroot00000000000000function fl = kflops(operations) % Calculate the number of FLOPS needed to perform the specified operations. % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ % values for x86 processor, change as desired flops_sum = 1; flops_mult = 1; flops_div = 8; flops_sqrt = 8; flops_exp = 20; fl = 0; f = fields(operations); for op = f' num = operations.(op{1}); switch op{1} case 'sum' fl = fl + num*flops_sum; case 'mult' fl = fl + num*flops_mult; case 'div' fl = fl + num*flops_div; case 'exp' fl = fl + num*flops_exp; case 'sqrt' fl = fl + num*flops_sqrt; case 'gauss_kernel' n1 = operations.(op{1})(1); n2 = operations.(op{1})(2); m = operations.(op{1})(3); fl = fl + kflops(struct('sum',n1*n2*(2*m-1),... 'mult',n1*n2*(m+1),'exp',n1*n2)); % breakdown: % fl = kaf_flops('sum',N1*N2*M) + ... % all X1(i,:)-X(2(j,:) % kaf_flops('mult',N1*N2*M) + ... % square all elements % kaf_flops('sum',N1*N2*(M-1)) + ... % sum M-1 elements per entry % kaf_flops('mult',N1*N2) + ... % multiply by 1/(2*sgm^2) % kaf_flops('exp',N1*N2); otherwise error('unknown option') end end kafbox-2.0/lib/profiler/klms_profiler.m000066400000000000000000000036321303712747200202550ustar00rootroot00000000000000% Profiler extension for Kernel Least-Mean-Square algorithm % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ classdef klms_profiler < klms properties (GetAccess = 'public', SetAccess = 'private') elapsed = 0; % elapsed time prev_dict_size = 0; % previous dictionary size for growth check end methods function kaf = klms_profiler(parameters) % constructor if nargin<1, parameters = struct(); end kaf = kaf@klms(parameters); end function flops = lastflops(kaf) % flops for last iteration m = size(kaf.dict,1); if kaf.prev_dict_size < m, % growing m1 = m - 1; floptions = struct(... 'sum', m1 - 1, ... 'mult', m1 + 1, ... sprintf('%s_kernel',kaf.kerneltype), [m1, 1, size(kaf.dict,2)]); flops = kflops(floptions); else flops = 0; end end %% flops breakdown % k = kernel(kaf.dict,x,kaf.kerneltype,kaf.kernelpar); % kernel: m1 % y_est = k'*kaf.alpha; % sum: m1 - 1 % mult: m1 % err = y - y_est; % sum: 1 % kaf.alpha = [kaf.alpha; kaf.mu*err]; % mult: 1 %% function train_profiled(kaf,x,y) kaf.prev_dict_size = size(kaf.dict,1); t1 = tic; kaf.train(x,y); t2 = toc(t1); kaf.elapsed = kaf.elapsed + t2; end function bytes = lastbytes(kaf) % bytes used in last iteration m = size(kaf.dict,1); bytes = 8*(m + m*size(kaf.dict,2)); % 8 bytes for double precision % alpha, dict end end end kafbox-2.0/lib/profiler/knlms_profiler.m000066400000000000000000000040771303712747200204370ustar00rootroot00000000000000% Profiler extension for Kernel Normalized Least-Mean-Square algorithm with % Coherence Criterion % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ classdef knlms_profiler < knlms properties (GetAccess = 'public', SetAccess = 'private') elapsed = 0; % elapsed time prev_dict_size = 0; % previous dictionary size for growth check end methods function kaf = knlms_profiler(parameters) % constructor if nargin<1, parameters = struct(); end kaf = kaf@knlms(parameters); end function flops = lastflops(kaf) % flops for last iteration m = size(kaf.dict,1); if kaf.prev_dict_size < m, % growing m1 = m - 1; n2 = 1; else m1 = m; n2 = 0; end m3 = m; floptions = struct(... 'sum', 3*m3, ... 'mult', 3*m3 + 1, ... sprintf('%s_kernel',kaf.kerneltype), [m1+n2, 1, size(kaf.dict,2)]); flops = kflops(floptions); end %% flops breakdown % k = kernel(x,kaf.dict,kaf.kerneltype,kaf.kernelpar); % kernel: m1 % h = kernel(x,kaf.dict,kaf.kerneltype,kaf.kernelpar); % kernel: n2 % 1 element when growing, 0 when not % kaf.alpha = kaf.alpha + kaf.eta / (kaf.eps + h*h') * (y - h*kaf.alpha) * h'; % sum: 3*m3 % mult: 3*m3 + 1 % div: 1 %% function train_profiled(kaf,x,y) kaf.prev_dict_size = size(kaf.dict,1); t1 = tic; kaf.train(x,y); t2 = toc(t1); kaf.elapsed = kaf.elapsed + t2; end function bytes = lastbytes(kaf) % bytes used in last iteration m = size(kaf.dict,1); bytes = 8*(m + m*size(kaf.dict,2)); % 8 bytes for double precision % alpha, dict end end end kafbox-2.0/lib/profiler/krls_profiler.m000066400000000000000000000065051303712747200202640ustar00rootroot00000000000000% Profiler extension for Kernel Recursive Least Squares with Approximate % Linear Dependency % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ classdef krls_profiler < krls properties (GetAccess = 'public', SetAccess = 'private') elapsed = 0; % elapsed time prev_dict_size = 0; % previous dictionary size for growth check end methods function kaf = krls_profiler(parameters) % constructor if nargin<1, parameters = struct(); end kaf = kaf@krls(parameters); end function flops = lastflops(kaf) % flops for last iteration m = size(kaf.dict,1); if kaf.prev_dict_size < m, % growing m1 = m; m2 = m - 1; m3 = m - 1; floptions = struct(... 'sum', m2^2 - m2 + m2 + m3^2 + m3 + m3, ... 'mult', m2^2 + m2 + m3^2 + 2*m3 + m3 + 1 + m3, ... 'div', 1, ... sprintf('%s_kernel',kaf.kerneltype), [m1,1,size(kaf.dict,2)]); else m1 = m + 1; m2 = m; m4 = m; floptions = struct(... 'sum', m2^2 - m2 + m2 + m4^2 + m4^2 + m4^2 + m4, ... 'mult', m2^2 + m2 + m4^2 + m4 + m4^2 + m4^2 + 2*m4, ... 'div', 1, ... sprintf('%s_kernel',kaf.kerneltype), [m1,1,size(kaf.dict,2)]); end % note: update with dictionary changed has 2*m^2 operations, % while update with dictionary unchanged has 8*m^2 operations. flops = kflops(floptions); end %% flops breakdown % k = kernel([kaf.dict; x],x,kaf.kerneltype,kaf.kernelpar); % kernel: m1 % at = kaf.Kinv*kt; % check linear dependency % sum: m2^2 - m2 % mult: m2^2 % delta = ktt - kt'*at; % check linear dependency % sum: m2 % mult: m2 % kaf.Kinv = 1/delta*[delta*kaf.Kinv + at*at', -at; -at', 1]; % grow Kinv % sum: m3^2 % mult: m3^2 + 2*m3 + 1 % div: 1 % ode = 1/delta*(y-kt'*kaf.alpha); % grow Kinv % sum: m3 % mult: m3 + 1 % kaf.alpha = [kaf.alpha - at*ode; ode]; % grow Kinv % sum: m3 % mult: m3 % q = kaf.P*at/(1+at'*kaf.P*at); % only update alpha % sum: m4^2 % mult: m4^2 + m4 % div: 1 % kaf.P = kaf.P - q*(at'*kaf.P); % only update alpha % sum: m4^2 % mult: m4^2 % kaf.alpha = kaf.alpha + kaf.Kinv*q*(y-kt'*kaf.alpha); % only update alpha % sum: m4^2 + m4 % mult: m4^2 + 2*m4 %% function train_profiled(kaf,x,y) kaf.prev_dict_size = size(kaf.dict,1); t1 = tic; kaf.train(x,y); t2 = toc(t1); kaf.elapsed = kaf.elapsed + t2; end function bytes = lastbytes(kaf) % bytes used in last iteration m = size(kaf.dict,1); bytes = 8*(m^2 + m^2 + m + m*size(kaf.dict,2)); % 8 bytes for double precision % Kinv, P, alpha, dict end end end kafbox-2.0/lib/profiler/krlst_profiler.m000066400000000000000000000075671303712747200204610ustar00rootroot00000000000000% Profiler extension for Kernel Recursive Least-Squares Tracker algorithm % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ classdef krlst_profiler < krlst properties (GetAccess = 'public', SetAccess = 'private') elapsed = 0; % elapsed time end methods function kaf = krlst_profiler(parameters) % constructor if nargin<1, parameters = struct(); end kaf = kaf@krlst(parameters); end function flops = lastflops(kaf) % flops for last iteration m = size(kaf.dict,1); if kaf.prune m1 = m; else m1 = m - 1; end if kaf.reduced m3 = 0; else m3 = m; end m2 = m1 + 1; floptions = struct(... 'sum', m1^2 + m1 + 1 + m1^2 - m1 + m1 - 1 + m1 + m1 - 1 + m1 + 1 + m2^2 + m2 + 1 + m2^2 + 2 + 1 + m2^2 - m2 + m3^2, ... 'mult', 2*m1^2 + m1^2 + m1 + m1 + m1 + m1 + m2^2 + 1 + 1 + m2^2 + 3 + m2^2 + m3^2, ... 'div', 1 + 1 + 2 + m2 + m3, ... sprintf('%s_kernel',kaf.kerneltype), [m1^2 + 1,1,size(kaf.dict,2)]); flops = kflops(floptions); end %% flops breakdown % K = kernel(kaf.dict,kaf.dict,kaf.kerneltype,kaf.kernelpar); % kernel: m1^2 % kaf.Sigma = kaf.lambda*kaf.Sigma + (1-kaf.lambda)*K + kaf.jitter*eye(m); % forget % sum: m1^2 + m1 + 1 % mult: 2*m1^2 % kaf.mu = sqrt(kaf.lambda)*kaf.mu; % square can be pre-calculated % mult: m1 % k = kernel([kaf.dict; x],x,kaf.kerneltype,kaf.kernelpar); % kernel: m1 % q = kaf.Q*kt; % sum: m1^2 - m1 % mult: m1^2 % y_mean = q'*kaf.mu; % predictive mean % sum: m1 - 1 % mult: m1 % gamma2 = ktt - kt'*q; gamma2(gamma2<0)=0; % projection uncertainty % sum: m1 % mult: m1 % h = kaf.Sigma*q; % sum: m1 - 1 % mult: m1 % sf2 = gamma2 + q'*h; sf2(sf2<0)=0; % noiseless prediction variance % sum: m1 % mult: m1 % sy2 = kaf.sn2 + sf2; % sum: 1 % kaf.Q = [kaf.Q zeros(m,1);zeros(1,m) 0] + 1/gamma2*(p*p'); with p = [q; -1]; % sum: m2^2 % mult: m2^2 + 1 % div: 1 % kaf.mu = [kaf.mu; y_mean] + (y - y_mean)/sy2*p; % with p = [h; sf2]; % sum: m2 + 1 % mult: m2 + 1 % div: 1 % kaf.Sigma = [kaf.Sigma h; h' sf2] - 1/sy2*(p*p'); % posterior covariance % sum: m2^2 % mult: m2^2 % kaf.nums02ML = kaf.nums02ML + kaf.lambda*(y - y_mean)^2/sy2; % sum: 2 % mult: 3 % kaf.dens02ML = kaf.dens02ML + kaf.lambda; % sum: 1 % kaf.s02 = kaf.nums02ML/kaf.dens02ML; % div: 1 % errors = (kaf.Q*kaf.mu)./diag(kaf.Q); % MSE pruning criterion % sum: m2^2 - m2 % mult: m2^2 % div: m2 % kaf.Q = kaf.Q - (Qs*Qs')/qs; % if removed element is not the last % sum: m3^2 % mult: m3^2 % div: m3 %% function train_profiled(kaf,x,y) % kaf.prev_dict_size = size(kaf.dict,1); t1 = tic; kaf.train(x,y); t2 = toc(t1); kaf.elapsed = kaf.elapsed + t2; end function bytes = lastbytes(kaf) % bytes used in last iteration m = size(kaf.dict,1); bytes = 8*(m^2 + m^2 + m + 2 + m*size(kaf.dict,2)); % 8 bytes for double precision % Q, Sigma, mu, nums02ML, dens02ML, dict end end end kafbox-2.0/lib/profiler/lkapa_profiler.m000066400000000000000000000031141303712747200203720ustar00rootroot00000000000000% Profiler extension for Leaky Kernel Affine Projection Algorithm % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ classdef lkapa_profiler < lkapa properties (GetAccess = 'public', SetAccess = 'private') elapsed = 0; % elapsed time end methods function kaf = lkapa_profiler(parameters) % constructor if nargin<1, parameters = struct(); end kaf = kaf@lkapa(parameters); end function flops = lastflops(kaf) % flops for last iteration % numbers of operations m1 = 1; m2 = 1; m3 = 1; m4 = 1; floptions = struct(... 'sum', m1, ... 'mult', m2, ... 'div', m3, ... sprintf('%s_kernel',kaf.kerneltype), [m4,1,size(kaf.dict,2)]); flops = nan; % dummy. kflops(floptions); end %% flops breakdown % [space for remarks on number of operations used above] %% function train_profiled(kaf,x,y) t1 = tic; kaf.train(x,y); t2 = toc(t1); kaf.elapsed = kaf.elapsed + t2; end function bytes = lastbytes(kaf) % bytes used in last iteration m = size(kaf.dict,1); m2 = 1; bytes = nan; % dummy. (m2 + m*size(kaf.dict,2)); % 8 bytes for double precision % [list variables] end end end kafbox-2.0/lib/profiler/nlkapa_profiler.m000066400000000000000000000031351303712747200205530ustar00rootroot00000000000000% Profiler extension for Normalized Leaky Kernel Affine Projection % Algorithm % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ classdef nlkapa_profiler < nlkapa properties (GetAccess = 'public', SetAccess = 'private') elapsed = 0; % elapsed time end methods function kaf = nlkapa_profiler(parameters) % constructor if nargin<1, parameters = struct(); end kaf = kaf@nlkapa(parameters); end function flops = lastflops(kaf) % flops for last iteration % numbers of operations m1 = 1; m2 = 1; m3 = 1; m4 = 1; floptions = struct(... 'sum', m1, ... 'mult', m2, ... 'div', m3, ... sprintf('%s_kernel',kaf.kerneltype), [m4,1,size(kaf.dict,2)]); flops = nan; % dummy. kflops(floptions); end %% flops breakdown % [space for remarks on number of operations used above] %% function train_profiled(kaf,x,y) t1 = tic; kaf.train(x,y); t2 = toc(t1); kaf.elapsed = kaf.elapsed + t2; end function bytes = lastbytes(kaf) % bytes used in last iteration m = size(kaf.dict,1); m2 = 1; bytes = nan; % dummy. (m2 + m*size(kaf.dict,2)); % 8 bytes for double precision % [list variables] end end end kafbox-2.0/lib/profiler/nlms_profiler.m000066400000000000000000000025501303712747200202560ustar00rootroot00000000000000% Profiler extension for Normalized Least Mean Squares algorithm % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ classdef nlms_profiler < nlms properties (GetAccess = 'public', SetAccess = 'private') elapsed = 0; % elapsed time end methods function obj = nlms_profiler(parameters) % constructor if nargin<1, parameters = struct(); end obj = obj@nlms(parameters); end function flops = lastflops(obj) % flops for last iteration m = size(obj.w,1); floptions = struct(... 'sum', 3*m, ... 'mult', 3*m, ... 'div', 1); flops = kflops(floptions); end %% flops breakdown % obj.w = obj.w + obj.mu/(obj.eps + x*x')*x*(y-x*obj.w'); % sum: 3*m % mult: 3*m % div: 1 %% function train_profiled(obj,x,y) t1 = tic; obj.train(x,y); t2 = toc(t1); obj.elapsed = obj.elapsed + t2; end function bytes = lastbytes(obj) % bytes used in last iteration m = size(obj.w,1); bytes = 8*m; % 8 bytes for double precision % w end end end kafbox-2.0/lib/profiler/norma_profiler.m000066400000000000000000000042251303712747200204220ustar00rootroot00000000000000% Profiler extension for Naive Online regularized Risk Minimization % Algorithm % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ classdef norma_profiler < norma properties (GetAccess = 'public', SetAccess = 'private') elapsed = 0; % elapsed time prev_dict_size = 0; % previous memory size for prune check end methods function kaf = norma_profiler(parameters) % constructor if nargin<1, parameters = struct(); end kaf = kaf@norma(parameters); end function flops = lastflops(kaf) % flops for last iteration m = size(kaf.dict,1); if kaf.prev_dict_size < m, % growing (no pruning) m1 = m-1; else m1 = m; end floptions = struct(... 'sum', m1 - 1 + 1 + 1, ... 'mult', 2*m1 + 3 + 1, ... 'exp', 1, ... sprintf('%s_kernel',kaf.kerneltype), [m1,1,size(kaf.dict,2)]); flops = kflops(floptions); end %% flops breakdown % k = kernel(kaf.dict,x,kaf.kerneltype,kaf.kernelpar); % kernel: m1 % y_est = k'*(kaf.alpha.*kaf.beta(length(kaf.alpha):-1:1)); % sum: m1 - 1 % mult: 2*m1 % err = y - y_est; % sum: 1 % kaf.alpha = (1-kaf.lambda*kaf.eta*kaf.t^kaf.tcoeff)*kaf.alpha; % sum: 1 % mult: 3 % exp: 1 % kaf.alpha = [kaf.alpha; kaf.mu*err]; % mult: 1 %% function train_profiled(kaf,x,y) kaf.prev_dict_size = size(kaf.dict,1); t1 = tic; kaf.train(x,y); t2 = toc(t1); kaf.elapsed = kaf.elapsed + t2; end function bytes = lastbytes(kaf) % bytes used in last iteration m = size(kaf.dict,1); bytes = 4 + 8*(m + m*size(kaf.dict,2)); % 8 bytes for double precision, 4 for uint32 % t, alpha, dict end end end kafbox-2.0/lib/profiler/qklms_profiler.m000066400000000000000000000047461303712747200204450ustar00rootroot00000000000000% Profiler extension for Kernel Least-Mean-Square algorithm % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ classdef qklms_profiler < qklms properties (GetAccess = 'public', SetAccess = 'private') elapsed = 0; % elapsed time prev_dict_size = 0; % previous dictionary size for growth check end methods function kaf = qklms_profiler(parameters) % constructor if nargin<1, parameters = struct(); end kaf = kaf@qklms(parameters); end function flops = lastflops(kaf) % flops for last iteration m = size(kaf.dict,1); if kaf.prev_dict_size < m, % growing m1 = m-1; n4 = 1; n5 = 1; else m1 = m; n4 = 1; n5 = 0; end if strcmp('kaf.kernel','gauss') m2 = 0; m3 = 0; % quantization criterion can use values calculated by kernel else m2 = m1; m3 = size(kaf.dict,2); end floptions = struct(... 'sum', m1 - 1 + 1 + (2*m2-1)*m3 + n4 + n5, ... 'mult', m1 + m2*m3 + n4 + n5, ... sprintf('%s_kernel',kaf.kerneltype), [m1,1,size(kaf.dict,2)]); flops = kflops(floptions); end %% flops breakdown % k = kernel(kaf.dict,x,kaf.kerneltype,kaf.kernelpar); % kernel: m1 % y_est = k'*kaf.alpha; % sum: m1 - 1 % mult: m1 % err = y - y_est; % sum: 1 % [d2,j] = min(sum((kaf.dict - repmat(x,m,1)).^2,2)); % sum: (2*m2-1)*m3 % mult: m2*m3 % kaf.alpha(j) = kaf.alpha(j) + kaf.eta*err; % sum: n4 % mult: n5 % kaf.alpha = [kaf.alpha; kaf.mu*err]; % mult: n4 %% function train_profiled(kaf,x,y) kaf.prev_dict_size = size(kaf.dict,1); t1 = tic; kaf.train(x,y); t2 = toc(t1); kaf.elapsed = kaf.elapsed + t2; end function bytes = lastbytes(kaf) % bytes used in last iteration m = size(kaf.dict,1); bytes = 8*(m + m*size(kaf.dict,2)); % 8 bytes for double precision % alpha, dict end end end kafbox-2.0/lib/profiler/rls_profiler.m000066400000000000000000000033041303712747200201030ustar00rootroot00000000000000% Profiler extension for Recursive Least-Squares algorithm % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ classdef rls_profiler < rls properties (GetAccess = 'public', SetAccess = 'private') elapsed = 0; % elapsed time end methods function obj = rls_profiler(parameters) % constructor if nargin<1, parameters = struct(); end obj = obj@rls(parameters); end function flops = lastflops(obj) % flops for last iteration m = size(obj.w,1); floptions = struct(... 'sum', 2*m^2 - m + m + 2*m - 1 + 2*m^2 - m, ... 'mult', m + m + m + 3*m^2, ... 'div', 1 + 1); flops = kflops(floptions); end %% flops breakdown % k = obj.P*x'/(obj.lambda+x*obj.P*x'); % sum: 2*m^2 - m % mult: m % div: 1 % z = y - x*obj.w; % sum: m % mult: m % obj.w = obj.w + k*z; % sum: 2*m - 1 % mult: m % obj.P = obj.lambda\(obj.P - k*x*obj.P); % sum: 2*m^2 - m % mult: 3*m^2 % div: 1 %% function train_profiled(obj,x,y) % measures elapsed time of training t1 = tic; obj.train(x,y); t2 = toc(t1); obj.elapsed = obj.elapsed + t2; end function bytes = lastbytes(obj) % bytes used in last iteration m = size(obj.w,1); bytes = 8*(m + m^2); % 8 bytes for double precision % w, P end end end kafbox-2.0/lib/profiler/struct2str.m000066400000000000000000000014421303712747200175410ustar00rootroot00000000000000% convert a structure to a string function str = struct2str(my_struct) fields = sortrows(fieldnames(my_struct)); % avoid permutations str = ''; for i=1:length(fields) fn = fields{i}; fv = my_struct.(fn); switch class(fv) case 'char' str = sprintf('%s %s=%s',str,fn,fv); case 'double' if (round(fv)==fv) str = sprintf('%s %s=%d',str,fn,fv); else found = 0; for j=0:10, if (round(10^j*fv)==10^j*fv) && ~found str = sprintf(sprintf('%%s %%s=%%.%df',j),str,fn,fv); found = 1; end end end otherwise error('unknown field class'); end end str = str(2:end); kafbox-2.0/lib/profiler/swkrls_profiler.m000066400000000000000000000055221303712747200206340ustar00rootroot00000000000000% Profiler extension for Sliding-Window Kernel Recursive Least Squares % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ classdef swkrls_profiler < swkrls properties (GetAccess = 'public', SetAccess = 'private') elapsed = 0; % elapsed time prev_dict_size = 0; % previous dictionary size for growth check end methods function kaf = swkrls_profiler(parameters) % constructor if nargin<1, parameters = struct(); end kaf = kaf@swkrls(parameters); end function flops = lastflops(kaf) % flops for last iteration m = size(kaf.dict,1); if kaf.prev_dict_size < m, % growing m1 = m; m2 = m - 1; floptions = struct(... 'sum', m2^2 + m2^2 - m2 + m2^2 + m^2 - m, ... 'mult', m2^2 + m2 + m2^2 + m^2, ... 'div', 1, ... sprintf('%s_kernel',kaf.kerneltype), [m1,1,size(kaf.dict,2)]); else m1 = m + 1; m2 = m; m3 = m; floptions = struct(... 'sum', m2^2 + m2^2 - m2 + m2^2 + m3^2 + m^2 - m, ... 'mult', m2^2 + m2 + m2^2 + m3^2 + m3 + m^2, ... 'div', 1 + 1, ... sprintf('%s_kernel',kaf.kerneltype), [m1,1,size(kaf.dict,2)]); end flops = kflops(floptions); end %% flops breakdown % k = kernel(kaf.dict,x,kaf.kerneltype,kaf.kernelpar); % grow Kinv % kernel: m1 % d = k(end) + kaf.c; % grow Kinv % sum: 1 % g_inv = d - b'*kaf.Kinv*b; % grow Kinv % sum: m2^2 % mult: m2^2 + m2 % g = 1/g_inv; % grow Kinv % div: 1 % f = -kaf.Kinv*b*g; % grow Kinv % sum: m2^2 - m2 % mult: m2^2 + m2 % E = kaf.Kinv - kaf.Kinv*b*f'; % grow Kinv % sum: m2^2 % mult: m2^2 % kaf.Kinv = G - f*f'/e; % prune Kinv % sum: m3^2 % mult: m3^2 + m3 % div: 1 % kaf.alpha = kaf.Kinv*kaf.dicty; % end of training % sum: m^2 - m % prod: m^2 %% function train_profiled(kaf,x,y) kaf.prev_dict_size = size(kaf.dict,1); t1 = tic; kaf.train(x,y); t2 = toc(t1); kaf.elapsed = kaf.elapsed + t2; end function bytes = lastbytes(kaf) % bytes used in last iteration m = size(kaf.dict,1); bytes = 8*(m^2 + m + m + m*size(kaf.dict,2)); % 8 bytes for double precision % Kinv, alpha, dicty, dict end end end kafbox-2.0/lib/qklms.m000066400000000000000000000043471303712747200147160ustar00rootroot00000000000000% Quantized Kernel Least Mean Square algorithm % % B. Chen, S. Zhao, P. Zhu, and J.C. Principe, "Quantized Kernel Least Mean % Square Algorithm," IEEE Transactions on Neural Networks and Learning % Systems, vol. 23, no. 1, pp. 22-32, Jan. 2012, % http://dx.doi.org/10.1109/TNNLS.2011.2178446 % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ classdef qklms < handle properties (GetAccess = 'public', SetAccess = 'private') eta = .9; % learning rate epsu = .1; kerneltype = 'gauss'; % kernel type kernelpar = 1; % kernel parameter end properties (GetAccess = 'public', SetAccess = 'private') dict = []; % codebook (dictionary) alpha = []; % expansion coefficients end methods function kaf = qklms(parameters) % constructor if (nargin > 0) % copy valid parameters for fn = fieldnames(parameters)', if ismember(fn,fieldnames(kaf)), kaf.(fn{1}) = parameters.(fn{1}); end end end end function y_est = evaluate(kaf,x) % evaluate the algorithm if size(kaf.dict,1)>0 k = kernel(kaf.dict,x,kaf.kerneltype,kaf.kernelpar); y_est = k'*kaf.alpha; else y_est = zeros(size(x,1),1); % zeros if not initialized end end function train(kaf,x,y) % train the algorithm y_est = kaf.evaluate(x); % evaluate function output err = y - y_est; % instantaneous error m = size(kaf.dict,1); if m==0 d2 = kaf.epsu^2 + 1; % force addition of initial base else % find distance to closest dictionary element [d2,j] = min(sum((kaf.dict - repmat(x,m,1)).^2,2)); end if d2 <= kaf.epsu^2, % reduced coefficient update kaf.alpha(j) = kaf.alpha(j) + kaf.eta*err; else kaf.dict = [kaf.dict; x]; % add base to dictionary kaf.alpha = [kaf.alpha; kaf.eta*err]; % add new coefficient end end end end kafbox-2.0/lib/qklms_fb.m000066400000000000000000000124011303712747200153530ustar00rootroot00000000000000% Fixed budget quantized kernel least-mean-square algorithm % % S. Zhao, B. Chen, P. Zhu, J. C. Principe, "Fixed budget quantized kernel % least-mean-square algorithm", Signal Processing, Volume 93, Issue 9, % September 2013, Pages 2759-2770, % http://dx.doi.org/10.1016/j.sigpro.2013.02.012. % % Remark: significance calculation only implemented for Gaussian kernel. % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ classdef qklms_fb < handle properties (GetAccess = 'public', SetAccess = 'private') eta = .9; % learning rate epsu = .1; % quantization threshold beta = .95; % forgetting factor for influence M = 500; % dictionary size (K in publication) kerneltype = 'gauss'; % kernel type kernelpar = 1; % kernel parameter end properties (GetAccess = 'public', SetAccess = 'private') dict = []; % codebook alpha = []; % expansion coefficients E = []; % significance vector (~ importance) lambda = []; % influence vector (~ how many samples are quantized % to each centre) end methods function kaf = qklms_fb(parameters) % constructor if (nargin > 0) % copy valid parameters for fn = fieldnames(parameters)', if ismember(fn,fieldnames(kaf)), kaf.(fn{1}) = parameters.(fn{1}); end end end end function y_est = evaluate(kaf,x) % evaluate the algorithm if size(kaf.dict,1)>0 k = kernel(kaf.dict,x,kaf.kerneltype,kaf.kernelpar); y_est = k'*kaf.alpha; else y_est = zeros(size(x,1),1); end end function train(kaf,x,y) % train the algorithm y_est = kaf.evaluate(x); err = y - y_est; m = size(kaf.dict,1); if m==0 d2 = kaf.epsu^2 + 1; % force dictionary growth else [d2,j] = min(sum((kaf.dict - repmat(x,m,1)).^2,2)); end c1 = pi*kaf.kernelpar^2/2; % pre-calculate if d2 <= kaf.epsu^2, % new basis under quantization threshold kaf.alpha(j) = kaf.alpha(j) + kaf.eta*err; % update significance for quantizing y to j-th centre (16) if m>1 inds = 1:m; inds(j) = []; kaf.E(inds) = kaf.beta*kaf.E(inds) + c1*abs(kaf.alpha(inds)).* ... kernel(kaf.dict(inds,:),kaf.dict(j,:),kaf.kerneltype,kaf.kernelpar); end kaf.E(j) = abs(1 + kaf.eta*err/kaf.alpha(j)) * ... kaf.beta*kaf.E(j) + ... abs(kaf.alpha(j) + kaf.eta*err) * ... c1*kernel(kaf.dict(j,:),kaf.dict(j,:),kaf.kerneltype,kaf.kernelpar); % update influence kaf.lambda = kaf.beta*kaf.lambda; kaf.lambda(j) = kaf.lambda(j) + 1; else % new basis not under quantization threshold if m < kaf.M % still room for extra centres kaf.dict = [kaf.dict; x]; % add to codebook kaf.alpha = [kaf.alpha; kaf.eta*err]; % update significance for addition of m+1-th centre (15) kaf.E(m+1,1) = 0; kaf.E = kaf.beta*kaf.E + abs(kaf.alpha(m+1)) * ... kernel(kaf.dict,kaf.dict(m+1,:),kaf.kerneltype,kaf.kernelpar); % update influence kaf.lambda = kaf.beta*kaf.lambda; % initial influence kaf.lambda(m+1) = 1; else % no room for extra centres [~,L] = min(kaf.E); % centre with lowest significance % update significance for removing L-th centre (17) kaf.E = kaf.E - kaf.lambda(L) * c1*abs(kaf.alpha) .* ... kernel(kaf.dict,kaf.dict(L,:),kaf.kerneltype,kaf.kernelpar); % prune kaf.dict(L,:) = []; kaf.alpha(L) = []; kaf.E(L) = []; kaf.lambda(L) = []; % re-calculate error y_est = kaf.evaluate(x); err = y - y_est; % grow kaf.dict = [kaf.dict; x]; kaf.alpha = [kaf.alpha; kaf.eta*err]; % update significance for addition of m-th centre (15) kaf.E(m,1) = 0; kaf.E = kaf.beta*kaf.E + abs(kaf.alpha(m)) * ... kernel(kaf.dict,kaf.dict(m,:),kaf.kerneltype,kaf.kernelpar); % update influence kaf.lambda = kaf.beta*kaf.lambda; % initial influence kaf.lambda(m) = 1; end end end end end kafbox-2.0/lib/readme.md000066400000000000000000000042431303712747200151630ustar00rootroot00000000000000Algorithm code structure ------------------------ - Each algorithm is contained in a single file in the `/lib` folder. - Each algorithm is implemented as an object in matlab, using the `classdef` syntax. - Each algorithm has two basic public methods: `evaluate` and `train`. - The object code contains only one iteration of the algorithm. The for-loop over the time index that governs the online operation should go in an external script. Quick steps to code an algorithm in the toolbox' format ------------------------------------------------------- 1. Make a copy of `lib/kafbox_template.m` and name it `MY_ALGORITHM.m`. 2. Open the file and replace all mentions of "kafbox_template" to MY_ALGORITHM. 3. Clean up the file header by replacing all text between square brackets. 4. In the first properties section, fill in the parameter values used by the algorithm with their corresponding default values. Default values should be chosen such that the algorithm performs well on a whitened input signal (see `demo/demo_sinc_all.m`). 5. In the second properties section, fill in the variables that will be calculated by the algorithm. 6. [optional] Adjust the `evaluate` method. 7. Add the code for training the algorithm to the `train` method. It may be useful to rely on "helper" functions to keep the code modular. For coding style, see below. 8. Check if the code is formatted correctly by running the script `unit_test('MY_ALGORITHM')`. How to contribute an algorithm to the toolbox --------------------------------------------- Option 1: email it to me (steven.vanvaerenbergh@unican.es). Option 2: fork the toolbox on GitHub (https://github.com/steven2358/kafbox), push your change to a named branch, then send me a pull request. Coding style ------------ **Code** should be 1. as human-readable as possible, in the first place; 2. short and structured, in the second place. **Algorithm structure** should follow the pseudocode from the corresponding publication. **Variable naming** should correspond to the nomenclature used in the corresponding publication whenever possible. **Comments** should be used sparingly. Document the design and purpose of the code rather than its mechanics. kafbox-2.0/lib/rffklms.m000066400000000000000000000040731303712747200152270ustar00rootroot00000000000000% Random Fourier Feature Kernel Least Mean Square algorithm. % % Abhishek Singh, Narendra Ahuja and Pierre Moulin, "Online learning with % kernels: Overcoming the growing sum problem," 2012 IEEE International % Workshop on Machine Learning for Signal Processing (MLSP), Sept. 2012. % http://dx.doi.org/10.1109/MLSP.2012.6349811 % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ classdef rffklms < handle properties (GetAccess = 'public', SetAccess = 'private') % parameters seed = 1; mu = .9; % step size D = 1000; % RFF dimension kerneltype = 'gauss'; % kernel type kernelpar = 1; % kernel parameter end properties (GetAccess = 'public', SetAccess = 'private') % variables omega = []; % Omega = []; % weight vector b = 0; % end methods function kaf = rffklms(parameters) % constructor if (nargin > 0) % copy valid parameters for fn = fieldnames(parameters)', if ismember(fn,fieldnames(kaf)), kaf.(fn{1}) = parameters.(fn{1}); end end end end function y_est = evaluate(kaf,x) % evaluate the algorithm if numel(kaf.omega) Psi = cos(kaf.omega*x' + repmat(kaf.b,1,size(x,1))); y_est = Psi'*kaf.Omega/kaf.D; else y_est = zeros(size(x,1),1); end end function train(kaf,x,y) % train the algorithm if ~numel(kaf.omega) rng('default'); rng(kaf.seed); kaf.omega = 1/kaf.kernelpar*randn(kaf.D,size(x,2)); kaf.b = 2*pi*rand(kaf.D,1); kaf.Omega = zeros(kaf.D,1); end Psi = cos(kaf.omega*x' + kaf.b); y_est = kaf.Omega'*Psi/kaf.D; err = y - y_est; kaf.Omega = kaf.Omega + kaf.mu*err*Psi; end end end kafbox-2.0/lib/rls.m000066400000000000000000000032741303712747200143650ustar00rootroot00000000000000% Recursive Least-Squares Algorithm with exponential weighting % % From S. Haykin, "Adaptive Filtering Theory (3rd Ed.)", Prentice Hall, % Chapter 13. % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ classdef rls < handle properties (GetAccess = 'public', SetAccess = 'private') lambda = .99; % forgetting factor c = 1E-4; % regularization end properties (GetAccess = 'public', SetAccess = 'private') P = []; % inverse autocorrelation matrix w = []; % filter coefficients end methods function obj = rls(parameters) % constructor if (nargin > 0) % copy valid parameters for fn = fieldnames(parameters)', if ismember(fn,fieldnames(obj)), obj.(fn{1}) = parameters.(fn{1}); end end end end function y_est = evaluate(obj,x) % evaluate the algorithm if numel(obj.w)>0 y_est = x*obj.w; else y_est = zeros(size(x,1),1); end end function train(obj,x,y) % train the algorithm if numel(obj.w)==0, % initialize m = length(x); obj.w = zeros(m,1); obj.P = obj.c\eye(m); end g = obj.P*x'/(obj.lambda+x*obj.P*x'); % gain vector err = y - x*obj.w; % instantaneous error obj.w = obj.w + g*err; % update filter coefficients obj.P = obj.lambda\(obj.P - g*x*obj.P); % update inv. autocorr. end end end kafbox-2.0/lib/swkrls.m000066400000000000000000000061351303712747200151110ustar00rootroot00000000000000% Sliding-Window Kernel Recursive Least Squares algorithm % % S. Van Vaerenbergh, J. Via, and I. Santamaria, "A Sliding-Window Kernel % RLS Algorithm and Its Application to Nonlinear Channel Identification," % 2006 IEEE International Conference on Acoustics, Speech and Signal % Processing (ICASSP), May 2006, % http://dx.doi.org/10.1109/ICASSP.2006.1661394 % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ classdef swkrls < handle properties (GetAccess = 'public', SetAccess = 'private') % parameters M = 250; % dictionary size c = 1E-2; % regularization parameter kerneltype = 'gauss'; % kernel type kernelpar = 1; % kernel parameter end properties (GetAccess = 'public', SetAccess = 'private') % variables dict = []; % dictionary dicty = []; % output dictionary alpha = []; % expansion coefficients Kinv = []; % inverse kernel matrix end methods function kaf = swkrls(parameters) % constructor if (nargin > 0) % copy valid parameters for fn = fieldnames(parameters)', if ismember(fn,fieldnames(kaf)), kaf.(fn{1}) = parameters.(fn{1}); end end end end function y_est = evaluate(kaf,x) % evaluate the algorithm if size(kaf.dict,1)>0 k = kernel(kaf.dict,x,kaf.kerneltype,kaf.kernelpar); y_est = k'*kaf.alpha; else y_est = zeros(size(x,1),1); end end function train(kaf,x,y) % train the algorithm kaf.dict = [kaf.dict; x]; % add base to dictionary kaf.dicty = [kaf.dicty; y]; % add y to output dictionary k = kernel(kaf.dict,x,kaf.kerneltype,kaf.kernelpar); kaf.Kinv = kaf.grow_kernel_matrix(kaf.Kinv,k,kaf.c); % grow if (size(kaf.dict,1) > kaf.M) % prune kaf.dict(1,:) = []; % remove oldest base from dictionary kaf.dicty(1) = []; % prune output dictionary kaf.Kinv = kaf.prune_kernel_matrix(kaf.Kinv); % prune end kaf.alpha = kaf.Kinv*kaf.dicty; end end methods (Static = true) function Kinv = grow_kernel_matrix(Kinv,k,c) % calculate inverse of expanded matrix K = [K_inv b;b' d] b = k(1:end-1); d = k(end) + c; % add regularization if numel(b)>0 g_inv = d - b'*Kinv*b; g = 1/g_inv; f = -Kinv*b*g; E = Kinv - Kinv*b*f'; Kinv = [E f;f' g]; else Kinv = 1/d; end end function Kinv = prune_kernel_matrix(Kinv) % calculate inverse of pruned kernel matrix Kp, K = [a b';b Kp] m = size(Kinv,1); G = Kinv(2:m,2:m); f = Kinv(2:m,1); e = Kinv(1,1); Kinv = G - f*f'/e; end end end kafbox-2.0/lib/test/000077500000000000000000000000001303712747200143605ustar00rootroot00000000000000kafbox-2.0/lib/test/unit_test.m000066400000000000000000000112501303712747200165530ustar00rootroot00000000000000% UNIT_TEST Test function for kernel adaptive filtering algorithms. % % The input arguments contains the algorithms to be tested, as separate % strings. If no argument is provided, or a single 'all' argument, all % algorithms from the 'lib' folder are tested. % USAGE: unit_test('klms') % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ function kaf = unit_test(varargin) % get list of algorithms in 'lib' folder pathstr = fileparts(which('unit_test.m')); fdir = [pathstr '/..']; % lib folder files = dir(fullfile(fdir,'*.m')); [~,allfiles] = cellfun(@fileparts, {files.name}, 'UniformOutput',0); % remove files that do not represent classes allfiles = filter_classes(allfiles); % get list of algorithms to run if isempty(varargin) || (length(varargin)==1 && strcmp(varargin{1},'all')) algorithms = allfiles; else algorithms = lower(varargin); fdir = ''; % local folder files = dir(fullfile(fdir,'*.m')); [~,localfiles] = cellfun(@fileparts, {files.name}, 'UniformOutput',0); % remove files that do not contain classes localfiles = filter_classes(localfiles); allfiles = horzcat(allfiles,localfiles); end % check for invalid names for a=algorithms, unexisting = isempty(find(ismember(allfiles,a{1}),1)); if unexisting error('Algorithm not found: %s.',a{1}); end end % perform test for each specified algorithm fprintf('\n') for ii=1:length(algorithms) t1 = tic; algorithm = algorithms{ii}; fprintf('%d. %s:\n',ii,upper(algorithm)); fprintf('Constructing object............\n') try feval(algorithm); catch err error('Error: %s object cannot be constructed: %s',... upper(algorithm), err.message); end fprintf('Testing arguments..............\n') % constructor test 1 opts = struct(); try feval(algorithm, opts); catch err me = err.stack(1); error(['Error in %s constructor: ',... '',... '%s at %d\n',... '%s'],... upper(algorithm), me.file, me.line, me.name, me.line,... err.message); end % constructor test 2: test if bogus arguments are processed opts = struct('shmaloombah',5,'ker',4); try kaf = feval(algorithm, opts); catch err me = err.stack(1); error(['Constructor of %s object is prone ',... 'to bogus arguments: ',... '',... '%s at %d\n',... '%s'],... upper(algorithm), me.file, me.line, me.name, me.line,... err.message); end % generate some data c = 10; N = 10000; x = rand(N,2)*c; y = sin(3*x(:,1)).*cos(x(:,1)+x(:,2)); fprintf('Evaluating before training.....') y_est = kaf.evaluate(x(1:2,:)); if length(y_est)<2 error(['Warning: '... '%s',... ' does not allow evaluation on multiple data',... 'before training'],algorithm, algorithm); end fprintf('\n') fprintf('1 training step................\n') for i=1, try kaf.train(x(i,:),y(i)); catch err me = err.stack(1); error(['Error in %s training: ',... '',... '%s at %d\n',... '%s'],... upper(algorithm), me.file, me.line, me.name, me.line,... err.message); end end fprintf('Training with repeated data....') for i=1:10, if ~mod(i,floor(N/10)), fprintf('.'); end kaf.train(x(1,:),y(1)); % y_test = kaf.evaluate(x(i+1,:)); end fprintf('\n') fprintf('Long training........') for i=1:N, if ~mod(i,floor(N/10)), fprintf('.'); end kaf.train(x(i,:),y(i)); % y_test = kaf.evaluate(x(i+1,:)); end fprintf('\n') fprintf('Testing evaluation.............\n') [x1,x2] = meshgrid(0:.2:c, 0:.2:c); yt = kaf.evaluate([x1(:) x2(:)]); z = reshape(yt,size(x1,1),size(x2,1)); close; figure; surf(x1,x2,z); title(sprintf('%s',upper(strrep(algorithm,'_','\_')))); colormap('spring'); view(20,70); drawnow; t2 = toc(t1); fprintf('OK. %.2fs\n\n',t2); end if length(algorithms)>1 fprintf('All OK.\n\n') end % remove files that do not represent classes function files = filter_classes(files) inds = []; for i=1:length(files) if ~exist(files{i},'class') inds = [inds i]; %#ok end end files(inds) = []; kafbox-2.0/lib/test/unit_test_profiler.m000066400000000000000000000106311303712747200204570ustar00rootroot00000000000000% UNIT_TEST_PROFILER Test function for profilers of kernel adaptive % filtering algorithms. % % The input arguments contains the algorithms to be tested, as separate % strings. If no argument is provided, or a single 'all' argument, all % algorithms from the 'lib' folder are tested. % USAGE: unit_test_profiler('klms') % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ function unit_test_profiler(varargin) % get list of algorithms in 'lib' folder pathstr = fileparts(which('unit_test_profiler.m')); fdir = [pathstr '/../profiler']; % profiler folder files = dir(fullfile(fdir,'*.m')); [~,allfiles] = cellfun(@fileparts, {files.name}, 'UniformOutput',0); % remove files that do not represent classes allfiles = filter_classes(allfiles); % get list of algorithms to run if isempty(varargin) || (length(varargin)==1 && strcmp(varargin{1},'all')) algorithms = allfiles; else algorithms = lower(varargin); for i=1:length(algorithms) algorithms{i} = [algorithms{i} '_profiler']; end fdir = ''; % local folder files = dir(fullfile(fdir,'*.m')); [~,localfiles] = cellfun(@fileparts, {files.name}, 'UniformOutput',0); % remove files that do not contain classes localfiles = filter_classes(localfiles); allfiles = horzcat(allfiles,localfiles); end % check for invalid names for a=algorithms, unexisting = isempty(find(ismember(allfiles,a{1}),1)); if unexisting error('Algorithm not found: %s.',a{1}); end end % perform test for each specified algorithm fprintf('\n') for ii=1:length(algorithms) t1 = tic; algorithm = algorithms{ii}; fprintf('%d. %s:\n',ii,upper(algorithm)); fprintf('Constructing object............\n') try feval(algorithm); catch err error('Error: %s object cannot be constructed: %s',... upper(algorithm), err.message); end fprintf('Testing arguments..............\n') % constructor test 1 opts = struct(); try feval(algorithm, opts); catch err me = err.stack(1); error(['Error in '... '%s',... ' constructor: ',... '',... '%s at %d\n',... '%s'],... algorithm, algorithm, me.file, me.line, me.name, me.line,... err.message); end % constructor test 2: test if bogus arguments are processed opts = struct('shmaloombah',5,'ker',4); try kaf = feval(algorithm, opts); catch err me = err.stack(1); error(['Constructor of %s object is prone ',... 'to bogus arguments: ',... '',... '%s at %d\n',... '%s'],... upper(algorithm), me.file, me.line, me.name, me.line,... err.message); end % generate some data c = 10; N = 1000; x = rand(N,2)*c; y = sin(3*x(:,1)).*cos(x(:,1)+x(:,2)); fprintf('1 training step................\n') for i=1, try kaf.train_profiled(x(i,:),y(i)); catch err me = err.stack(1); % keyboard error(['Error in '... '%s',... ' training: ',... '',... '%s at %d\n',... '%s'],... algorithm, algorithm, me.file, me.line, me.name, me.line,... err.message); end end fprintf('Profiled training....') flops_all = zeros(N,1); bytes_all = zeros(N,1); for i=1:N, if ~mod(i,floor(N/10)), fprintf('.'); end kaf.train_profiled(x(i,:),y(i)); flops_all(i) = kaf.lastflops(); bytes_all(i) = kaf.lastbytes(); end fprintf('\n') close all; figure; plot(bytes_all); ylabel('bytes') figure; plot(flops_all); ylabel('flops') drawnow; t2 = toc(t1); fprintf('OK. %.2fs\n\n',t2); end if length(algorithms)>1 fprintf('All OK.\n\n') end % remove files that do not represent classes function files = filter_classes(files) inds = []; for i=1:length(files) if ~exist(files{i},'class') inds = [inds i]; %#ok end end files(inds) = []; kafbox-2.0/lib/util/000077500000000000000000000000001303712747200143565ustar00rootroot00000000000000kafbox-2.0/lib/util/gpml/000077500000000000000000000000001303712747200153155ustar00rootroot00000000000000kafbox-2.0/lib/util/gpml/Copyright000066400000000000000000000014101303712747200172040ustar00rootroot00000000000000 Software that implements GAUSSIAN PROCESS REGRESSION AND CLASSIFICATION Copyright (c) 2005 - 2007 by Carl Edward Rasmussen and Chris Williams Permission is granted for anyone to copy, use, or modify these programs for purposes of research or education, provided this copyright notice is retained, and note is made of any changes that have been made. These programs are distributed without any warranty, express or implied. As these programs were written for research purposes only, they have not been tested to the degree that would be advisable in any important application. All use of these programs is entirely at the user's own risk. The code and associated documentation are avaiable from http://www.GaussianProcess.org/gpml/code kafbox-2.0/lib/util/gpml/kafbox_covLambda.m000066400000000000000000000030311303712747200207120ustar00rootroot00000000000000function [A, B] = kafbox_covLambda(logtheta, xt, z) % KAFBOX_COVLAMBDA covariance function for an AR(1) process with fixed % power, parameterized as k(x^(t),x^(t+n)) = lambda^|n|. Uses the % covariance functions constructors from GPML toolbox version 2.0. % % INPUT: - logtheta = [ log(lambda/(1-lambda))] % hyperparameter % - xt: input data, each row is a data point. First column % contains temporal indices, other columns are "spatial" data. % - z: test set % OUTPUT: - A, B: depend on the number of input arguments % USAGE: [A, B] = kaf_covLambda(logtheta, x, z) % % Author: Miguel Lazaro Gredilla, 2012 % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ if nargin == 0, A = '1'; return; end % report number of parameters t = xt(:,1); % calculate covariance function only on temporal indices lambda = exp(logtheta(1))/(1+exp(logtheta)); sf2 = 1; n = size(t,1); if nargin == 2 A = sf2*lambda.^abs(repmat(t,1,n)-repmat(t',n,1)); elseif nargout == 2 % compute test set covariances z = z(:,1); ntst = size(z,1); A = sf2*ones(ntst,1); B = sf2*lambda.^abs(repmat(t,1,ntst)-repmat(z',n,1)); else % compute derivative matrix if z == 1 % wrt lambda absn = abs(repmat(t,1,n)-repmat(t',n,1)); A = sf2*absn.*lambda.^absn.*(1-lambda); elseif z == 2 % wrt sf2 display('It is fixed, you should not be trying to compute this.') end end kafbox-2.0/lib/util/gpml/kafbox_covNoise.m000066400000000000000000000020601303712747200206100ustar00rootroot00000000000000function [A, B] = kafbox_covNoise(logtheta, x, z); % Independent covariance function, ie "white noise", with specified variance. % The covariance function is specified as: % % k(x^p,x^q) = s2 * \delta(p,q) % % where s2 is the noise variance and \delta(p,q) is a Kronecker delta function % which is 1 iff p=q and zero otherwise. The hyperparameter is % % logtheta = [ log(sqrt(s2)) ] % % For more help on design of covariance functions, try "help covFunctions". % % (C) Copyright 2006 by Carl Edward Rasmussen, 2006-03-24. if nargin == 0, A = '1'; return; end % report number of parameters s2 = exp(2*logtheta); % noise variance if nargin == 2 % compute covariance matrix A = s2*eye(size(x,1)); elseif nargout == 2 % compute test set covariances A = s2; B = 0; % zeros cross covariance by independence else % compute derivative matrix A = 2*s2*eye(size(x,1)); end kafbox-2.0/lib/util/gpml/kafbox_covProd.m000066400000000000000000000050351303712747200204440ustar00rootroot00000000000000function [A, B] = kafbox_covProd(covfunc, logtheta, x, z); % covProd - compose a covariance function as the product of other covariance % functions. This function doesn't actually compute very much on its own, it % merely does some bookkeeping, and calls other covariance functions to do the % actual work. % % For more help on design of covariance functions, try "help covFunctions". % % (C) Copyright 2006 by Carl Edward Rasmussen, 2006-04-06. for i = 1:length(covfunc) % iterate over covariance functions f = covfunc(i); if iscell(f{:}), f = f{:}; end % dereference cell array if necessary j(i) = cellstr(feval(f{:})); end if nargin == 1, % report number of parameters A = char(j(1)); for i=2:length(covfunc), A = [A, '+', char(j(i))]; end return end [n, D] = size(x); v = []; % v vector indicates to which covariance parameters belong for i = 1:length(covfunc), v = [v repmat(i, 1, eval(char(j(i))))]; end switch nargin case 3 % compute covariance matrix A = ones(n, n); % allocate space for covariance matrix for i = 1:length(covfunc) % iteration over factor functions f = covfunc(i); if iscell(f{:}), f = f{:}; end % dereference cell array if necessary A = A .* feval(f{:}, logtheta(v==i), x); % multiply covariances end case 4 % compute derivative matrix or test set covariances if nargout == 2 % compute test set cavariances A = ones(size(z,1),1); B = ones(size(x,1),size(z,1)); % allocate space for i = 1:length(covfunc) f = covfunc(i); if iscell(f{:}), f = f{:}; end % dereference cell array if necessary [AA BB] = feval(f{:}, logtheta(v==i), x, z); % compute test covariances A = A .* AA; B = B .* BB; % and accumulate end else % compute derivative matrices A = ones(n, n); ii = v(z); % which covariance function j = sum(v(1:z)==ii); % which parameter in that covariance for i = 1:length(covfunc) f = covfunc(i); if iscell(f{:}), f = f{:}; end % dereference cell array if necessary if i == ii A = A .* feval(f{:}, logtheta(v==i), x, j); % multiply derivative else A = A .* feval(f{:}, logtheta(v==i), x); % multiply covariance end end end end kafbox-2.0/lib/util/gpml/kafbox_covSEiso.m000066400000000000000000000031441303712747200205610ustar00rootroot00000000000000function [A, B] = kafbox_covSEiso(loghyper, x, z) % Squared Exponential covariance function with isotropic distance measure. The % covariance function is parameterized as: % % k(x^p,x^q) = sf2 * exp(-(x^p - x^q)'*inv(P)*(x^p - x^q)/2) % % where the P matrix is ell^2 times the unit matrix and sf2 is the signal % variance. The hyperparameters are: % % loghyper = [ log(ell) % log(sqrt(sf2)) ] % % For more help on design of covariance functions, try "help covFunctions". % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ % % Original covSEiso.m (C) Copyright 2006 by Carl Edward Rasmussen as part % of the GPML toolbox version 2.0. The code and associated documentation % are avaiable from http://www.GaussianProcess.org/gpml/code if nargin == 0, A = '2'; return; end % report number of parameters ell = exp(loghyper(1)); % characteristic length scale sf2 = exp(2*loghyper(2)); % signal variance if nargin == 2 A = sf2*exp(-kafbox_sq_dist(x'/ell)/2); elseif nargout == 2 % compute test set covariances A = sf2*ones(size(z,1),1); B = sf2*exp(-kafbox_sq_dist(x'/ell,z'/ell)/2); else % compute derivative matrix if z == 1 % first parameter A = sf2*exp(-kafbox_sq_dist(x'/ell)/2).*kafbox_sq_dist(x'/ell); else % second parameter A = 2*sf2*exp(-kafbox_sq_dist(x'/ell)/2); end end kafbox-2.0/lib/util/gpml/kafbox_covSEiso2.m000066400000000000000000000035621303712747200206470ustar00rootroot00000000000000function [A, B] = kafbox_covSEiso2(loghyper, xt, z) % KAFBOX_COVSEISO2 Squared Exponential covariance function with isotropic % distance measure. The covariance function is parameterized as: % k(x^p,x^q) = sf2 * exp(-(x^p - x^q)'*inv(P)*(x^p - x^q)/2), where the P % matrix is ell^2 times the unit matrix and sf2 is the signal variance. % % INPUT: - loghyper = [log(ell); log(sqrt(sf2))] % hyperparameters % - xt: input data, each row is a data point. First column % contains temporal indices, other columns are "spatial" data. % - z: test set % OUTPUT: - A, B: depend on the number of input arguments % USAGE: [A, B] = kaf_covSEiso2(loghyper, xt, z) % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ % % Original covSEiso.m (C) Copyright 2006 by Carl Edward Rasmussen as part % of the GPML toolbox version 2.0. The code and associated documentation % are avaiable from http://www.GaussianProcess.org/gpml/code if nargin == 0, A = '2'; return; end % report number of parameters x = xt(:,2:end); % calculate covariance function only on spatial data ell = exp(loghyper(1)); % characteristic length scale sf2 = exp(2*loghyper(2)); % signal variance if nargin == 2 A = sf2*exp(-kafbox_sq_dist(x'/ell)/2); elseif nargout == 2 % compute test set covariances z = z(:,2:end); A = sf2*ones(size(z,1),1); B = sf2*exp(-kafbox_sq_dist(x'/ell,z'/ell)/2); else % compute derivative matrix if z == 1 % first parameter A = sf2*exp(-kafbox_sq_dist(x'/ell)/2).*kafbox_sq_dist(x'/ell); else % second parameter A = 2*sf2*exp(-kafbox_sq_dist(x'/ell)/2); end end kafbox-2.0/lib/util/gpml/kafbox_covSum.m000066400000000000000000000045461303712747200203120ustar00rootroot00000000000000function [A, B] = kafbox_covSum(covfunc, logtheta, x, z); % covSum - compose a covariance function as the sum of other covariance % functions. This function doesn't actually compute very much on its own, it % merely does some bookkeeping, and calls other covariance functions to do the % actual work. % % For more help on design of covariance functions, try "help covFunctions". % % (C) Copyright 2006 by Carl Edward Rasmussen, 2006-03-20. for i = 1:length(covfunc) % iterate over covariance functions f = covfunc(i); if iscell(f{:}), f = f{:}; end % dereference cell array if necessary j(i) = cellstr(feval(f{:})); end if nargin == 1, % report number of parameters A = char(j(1)); for i=2:length(covfunc), A = [A, '+', char(j(i))]; end return end [n, D] = size(x); v = []; % v vector indicates to which covariance parameters belong for i = 1:length(covfunc), v = [v repmat(i, 1, eval(char(j(i))))]; end switch nargin case 3 % compute covariance matrix A = zeros(n, n); % allocate space for covariance matrix for i = 1:length(covfunc) % iteration over summand functions f = covfunc(i); if iscell(f{:}), f = f{:}; end % dereference cell array if necessary A = A + feval(f{:}, logtheta(v==i), x); % accumulate covariances end case 4 % compute derivative matrix or test set covariances if nargout == 2 % compute test set cavariances A = zeros(size(z,1),1); B = zeros(size(x,1),size(z,1)); % allocate space for i = 1:length(covfunc) f = covfunc(i); if iscell(f{:}), f = f{:}; end % dereference cell array if necessary [AA BB] = feval(f{:}, logtheta(v==i), x, z); % compute test covariances A = A + AA; B = B + BB; % and accumulate end else % compute derivative matrices i = v(z); % which covariance function j = sum(v(1:z)==i); % which parameter in that covariance f = covfunc(i); if iscell(f{:}), f = f{:}; end % dereference cell array if necessary A = feval(f{:}, logtheta(v==i), x, j); % compute derivative end end kafbox-2.0/lib/util/gpml/kafbox_gpr.m000066400000000000000000000054241303712747200176220ustar00rootroot00000000000000function [out1, out2] = kafbox_gpr(logtheta, covfunc, x, y, xstar) % gpr - Gaussian process regression, with a named covariance function. Two % modes are possible: training and prediction: if no test data are given, the % function returns minus the log likelihood and its partial derivatives with % respect to the hyperparameters; this mode is used to fit the hyperparameters. % If test data are given, then (marginal) Gaussian predictions are computed, % whose mean and variance are returned. Note that in cases where the covariance % function has noise contributions, the variance returned in S2 is for noisy % test targets; if you want the variance of the noise-free latent function, you % must substract the noise variance. % % usage: [nlml dnlml] = gpr(logtheta, covfunc, x, y) % or: [mu S2] = gpr(logtheta, covfunc, x, y, xstar) % % where: % % logtheta is a (column) vector of log hyperparameters % covfunc is the covariance function % x is a n by D matrix of training inputs % y is a (column) vector (of size n) of targets % xstar is a nn by D matrix of test inputs % nlml is the returned value of the negative log marginal likelihood % dnlml is a (column) vector of partial derivatives of the negative % log marginal likelihood wrt each log hyperparameter % mu is a (column) vector (of size nn) of prediced means % S2 is a (column) vector (of size nn) of predicted variances % % For more help on covariance functions, see "help covFunctions". % % (C) copyright 2006 by Carl Edward Rasmussen (2006-03-20). if ischar(covfunc), covfunc = cellstr(covfunc); end % convert to cell if needed [n, D] = size(x); %#ok if eval(feval(covfunc{:})) ~= size(logtheta, 1) error('Error: Number of parameters do not agree with covariance function') end K = feval(covfunc{:}, logtheta, x); % compute training set covariance matrix L = chol(K)'; % cholesky factorization of the covariance alpha = kafbox_solve_chol(L',y); if nargin == 4 % if no test cases, compute the negative log marginal likelihood out1 = 0.5*y'*alpha + sum(log(diag(L))) + 0.5*n*log(2*pi); if nargout == 2 % ... and if requested, its partial derivatives out2 = zeros(size(logtheta)); % set the size of the derivative vector W = L'\(L\eye(n))-alpha*alpha'; % precompute for convenience for i = 1:length(out2) out2(i) = sum(sum(W.*feval(covfunc{:}, logtheta, x, i)))/2; end end else % ... otherwise compute (marginal) test predictions ... [Kss, Kstar] = feval(covfunc{:}, logtheta, x, xstar); % test covariances out1 = Kstar' * alpha; % predicted means if nargout == 2 v = L\Kstar; out2 = Kss - sum(v.*v)'; end end kafbox-2.0/lib/util/gpml/kafbox_minimize.m000066400000000000000000000214521303712747200206520ustar00rootroot00000000000000function [X, fX, i] = kafbox_minimize(X, f, length, varargin) % Minimize a differentiable multivariate function. % % Usage: [X, fX, i] = minimize(X, f, length, P1, P2, P3, ... ) % % where the starting point is given by "X" (D by 1), and the function named in % the string "f", must return a function value and a vector of partial % derivatives of f wrt X, the "length" gives the length of the run: if it is % positive, it gives the maximum number of line searches, if negative its % absolute gives the maximum allowed number of function evaluations. You can % (optionally) give "length" a second component, which will indicate the % reduction in function value to be expected in the first line-search (defaults % to 1.0). The parameters P1, P2, P3, ... are passed on to the function f. % % The function returns when either its length is up, or if no further progress % can be made (ie, we are at a (local) minimum, or so close that due to % numerical problems, we cannot get any closer). NOTE: If the function % terminates within a few iterations, it could be an indication that the % function values and derivatives are not consistent (ie, there may be a bug in % the implementation of your "f" function). The function returns the found % solution "X", a vector of function values "fX" indicating the progress made % and "i" the number of iterations (line searches or function evaluations, % depending on the sign of "length") used. % % The Polack-Ribiere flavour of conjugate gradients is used to compute search % directions, and a line search using quadratic and cubic polynomial % approximations and the Wolfe-Powell stopping criteria is used together with % the slope ratio method for guessing initial step sizes. Additionally a bunch % of checks are made to make sure that exploration is taking place and that % extrapolation will not be unboundedly large. % % See also: checkgrad % % Copyright (C) 2001 - 2006 by Carl Edward Rasmussen (2006-09-08). INT = 0.1; % don't reevaluate within 0.1 of the limit of the current bracket EXT = 3.0; % extrapolate maximum 3 times the current step-size MAX = 20; % max 20 function evaluations per line search RATIO = 10; % maximum allowed slope ratio SIG = 0.1; RHO = SIG/2; % SIG and RHO are the constants controlling the Wolfe- % Powell conditions. SIG is the maximum allowed absolute ratio between % previous and new slopes (derivatives in the search direction), thus setting % SIG to low (positive) values forces higher precision in the line-searches. % RHO is the minimum allowed fraction of the expected (from the slope at the % initial point in the linesearch). Constants must satisfy 0 < RHO < SIG < 1. % Tuning of SIG (depending on the nature of the function to be optimized) may % speed up the minimization; it is probably not worth playing much with RHO. % The code falls naturally into 3 parts, after the initial line search is % started in the direction of steepest descent. 1) we first enter a while loop % which uses point 1 (p1) and (p2) to compute an extrapolation (p3), until we % have extrapolated far enough (Wolfe-Powell conditions). 2) if necessary, we % enter the second loop which takes p2, p3 and p4 chooses the subinterval % containing a (local) minimum, and interpolates it, unil an acceptable point % is found (Wolfe-Powell conditions). Note, that points are always maintained % in order p0 <= p1 <= p2 < p3 < p4. 3) compute a new search direction using % conjugate gradients (Polack-Ribiere flavour), or revert to steepest if there % was a problem in the previous line-search. Return the best value so far, if % two consecutive line-searches fail, or whenever we run out of function % evaluations or line-searches. During extrapolation, the "f" function may fail % either with an error or returning Nan or Inf, and minimize should handle this % gracefully. if max(size(length)) == 2, red=length(2); length=length(1); else red=1; end if length>0, S='Linesearch'; else S='Function evaluation'; end i = 0; % zero the run length counter ls_failed = 0; % no previous line search has failed [f0 df0] = feval(f, X, varargin{:}); % get function value and gradient fX = f0; i = i + (length<0); % count epochs?! s = -df0; d0 = -s'*s; % initial search direction (steepest) and slope x3 = red/(1-d0); % initial step is red/(|s|+1) while i < abs(length) % while not finished i = i + (length>0); % count iterations?! X0 = X; F0 = f0; dF0 = df0; % make a copy of current values if length>0, M = MAX; else M = min(MAX, -length-i); end while 1 % keep extrapolating as long as necessary x2 = 0; f2 = f0; d2 = d0; f3 = f0; df3 = df0; success = 0; while ~success && M > 0 try M = M - 1; i = i + (length<0); % count epochs?! [f3 df3] = feval(f, X+x3*s, varargin{:}); if isnan(f3) || isinf(f3) || any(isnan(df3)+isinf(df3)), error(''), end success = 1; catch % catch any error which occured in f x3 = (x2+x3)/2; % bisect and try again end end if f3 < F0, X0 = X+x3*s; F0 = f3; dF0 = df3; end % keep best values d3 = df3'*s; % new slope if d3 > SIG*d0 || f3 > f0+x3*RHO*d0 || M == 0 % are we done extrapolating? break end x1 = x2; f1 = f2; d1 = d2; % move point 2 to point 1 x2 = x3; f2 = f3; d2 = d3; % move point 3 to point 2 A = 6*(f1-f2)+3*(d2+d1)*(x2-x1); % make cubic extrapolation B = 3*(f2-f1)-(2*d1+d2)*(x2-x1); x3 = x1-d1*(x2-x1)^2/(B+sqrt(B*B-A*d1*(x2-x1))); % num. error possible, ok! if ~isreal(x3) || isnan(x3) || isinf(x3) || x3 < 0 % num prob | wrong sign? x3 = x2*EXT; % extrapolate maximum amount elseif x3 > x2*EXT % new point beyond extrapolation limit? x3 = x2*EXT; % extrapolate maximum amount elseif x3 < x2+INT*(x2-x1) % new point too close to previous point? x3 = x2+INT*(x2-x1); end end % end extrapolation while (abs(d3) > -SIG*d0 || f3 > f0+x3*RHO*d0) && M > 0 % keep interpolating if d3 > 0 || f3 > f0+x3*RHO*d0 % choose subinterval x4 = x3; f4 = f3; d4 = d3; % move point 3 to point 4 else x2 = x3; f2 = f3; d2 = d3; % move point 3 to point 2 end if f4 > f0 x3 = x2-(0.5*d2*(x4-x2)^2)/(f4-f2-d2*(x4-x2)); % quadratic interpolation else A = 6*(f2-f4)/(x4-x2)+3*(d4+d2); % cubic interpolation B = 3*(f4-f2)-(2*d2+d4)*(x4-x2); x3 = x2+(sqrt(B*B-A*d2*(x4-x2)^2)-B)/A; % num. error possible, ok! end if isnan(x3) || isinf(x3) x3 = (x2+x4)/2; % if we had a numerical problem then bisect end x3 = max(min(x3, x4-INT*(x4-x2)),x2+INT*(x4-x2)); % don't accept too close [f3 df3] = feval(f, X+x3*s, varargin{:}); if f3 < F0, X0 = X+x3*s; F0 = f3; dF0 = df3; end % keep best values M = M - 1; i = i + (length<0); % count epochs?! d3 = df3'*s; % new slope end % end interpolation if abs(d3) < -SIG*d0 && f3 < f0+x3*RHO*d0 % if line search succeeded X = X+x3*s; f0 = f3; fX = [fX' f0]'; % update variables fprintf('%s %6i; Value %4.6e\r', S, i, f0); s = (df3'*df3-df0'*df3)/(df0'*df0)*s - df3; % Polack-Ribiere CG direction df0 = df3; % swap derivatives d3 = d0; d0 = df0'*s; if d0 > 0 % new slope must be negative s = -df0; d0 = -s'*s; % otherwise use steepest direction end x3 = x3 * min(RATIO, d3/(d0-realmin)); % slope ratio but max RATIO ls_failed = 0; % this line search did not fail else X = X0; f0 = F0; df0 = dF0; % restore best point so far if ls_failed || i > abs(length) % line search failed twice in a row break; % or we ran out of time, so we give up end s = -df0; d0 = -s'*s; % try steepest x3 = 1/(1-d0); ls_failed = 1; % this line search failed end end fprintf('\n'); kafbox-2.0/lib/util/gpml/kafbox_solve_chol.m000066400000000000000000000017461303712747200211720ustar00rootroot00000000000000% solve_chol - solve linear equations from the Cholesky factorization. % Solve A*X = B for X, where A is square, symmetric, positive definite. The % input to the function is R the Cholesky decomposition of A and the matrix B. % Example: X = solve_chol(chol(A),B); % % NOTE: The program code is written in the C language for efficiency and is % contained in the file solve_chol.c, and should be compiled using matlabs mex % facility. However, this file also contains a (less efficient) matlab % implementation, supplied only as a help to people unfamiliar with mex. If % the C code has been properly compiled and is avaiable, it automatically % takes precendence over the matlab code in this file. % % Copyright (c) 2004, 2005, 2006 by Carl Edward Rasmussen. 2006-02-08. function x = kafbox_solve_chol(A, B); if nargin ~= 2 | nargout > 1 error('Wrong number of arguments.'); end if size(A,1) ~= size(A,2) | size(A,1) ~= size(B,1) error('Wrong sizes of matrix arguments.'); end x = A\(A'\B); kafbox-2.0/lib/util/gpml/kafbox_sq_dist.c000066400000000000000000000036131303712747200204640ustar00rootroot00000000000000/* sq_dist - a mex function to compute a matrix of all pairwise squared distances between two sets of vectors, stored in the columns of the two matrices that are arguments to the function. The length of the vectors must agree. If only a single argument is given, the missing argument is taken to be identical to the first. If an optional third matrix argument Q is given, it must be of the same size as the output, but in this case a vector of the traces of the product of Q and the coordinatewise squared distances is returned. Copyright (c) 2003, 2004 Carl Edward Rasmussen. 2003-04-22. */ #include "mex.h" #include void mexFunction(int nlhs, mxArray *plhs[], int nrhs, const mxArray *prhs[]) { double *a, *b, *C, *Q, z, t; int D, n, m, i, j, k; if (nrhs < 1 || nrhs > 3 || nlhs > 1) mexErrMsgTxt("Usage: C = sq_dist(a,b)\n or: C = sq_dist(a)\n or: c = sq_dist(a,b,Q)\nwhere the b matrix may be empty."); a = mxGetPr(prhs[0]); m = mxGetN(prhs[0]); D = mxGetM(prhs[0]); if (nrhs == 1 || mxIsEmpty(prhs[1])) { b = a; n = m; } else { b = mxGetPr(prhs[1]); n = mxGetN(prhs[1]); if (D != mxGetM(prhs[1])) mexErrMsgTxt("Error: column lengths must agree"); } if (nrhs < 3) { plhs[0] = mxCreateDoubleMatrix(m, n, mxREAL); C = mxGetPr(plhs[0]); for (i=0; i 3 | nargout > 1 error('Wrong number of arguments.'); end if nargin == 1 | isempty(b) % input arguments are taken to be b = a; % identical if b is missing or empty end [D, n] = size(a); [d, m] = size(b); if d ~= D error('Error: column lengths must agree.'); end if nargin < 3 C = zeros(n,m); for d = 1:D C = C + (repmat(b(d,:), n, 1) - repmat(a(d,:)', 1, m)).^2; end % C = repmat(sum(a.*a)',1,m)+repmat(sum(b.*b),n,1)-2*a'*b could be used to % replace the 3 lines above; it would be faster, but numerically less stable. else if [n m] == size(Q) C = zeros(D,1); for d = 1:D C(d) = sum(sum((repmat(b(d,:), n, 1) - repmat(a(d,:)', 1, m)).^2.*Q)); end else error('Third argument has wrong size.'); end end kafbox-2.0/lib/util/gpml/readme.md000066400000000000000000000010131303712747200170670ustar00rootroot00000000000000GPML in KAFBOX ============== This folder contains excerpts from the GPML toolbox v2.0. The full toolbox is available at http://gaussianprocess.org/gpml/code/matlab/release/oldcode.html Prefixes have been added to the files in order to avoid interference with newer versions of GPML located on the same system. The files included here are used for the parameter estimation of kernel adaptive filters. Demos can be found in `demo/demo_parameter_estimation_lorenz.m` and `demo/demo_parameter_estimation_nonstationary.m`. kafbox-2.0/lib/util/kafbox_parameter_estimation.m000066400000000000000000000043361303712747200223100ustar00rootroot00000000000000function [sigma,reg,lambda] = ... kafbox_parameter_estimation(X,Y,time_ind,epochs) % KAFBOX_PARAMETER_ESTIMATION Gaussian-process based estimation of the % parameters of the KRLS-T algorithm. % % Steven Van Vaerenbergh, Ignacio Santamaria, and Miguel Lazaro-Gredilla, % "Estimation of the forgetting factor in kernel recursive least squares." % 2012 IEEE International Workshop on Machine Learning for Signal % Processing (MLSP), 2012. http://dx.doi.org/10.1109/MLSP.2012.6349749 % % INPUT: - X: input data, each row is a data point % - Y: output data, one column. % - time_ind: temporal indices of points % - epochs: number of epochs for minimization % OUTPUT: - sigma: estimated kernel length scale (width) % - reg: estimated regularization % - lambda: estimated forgetting factor. % USAGE: [sigma,reg,lambda] = % kafbox_parameter_estimation(X,Y,time_ind,epochs) % % Author: Steven Van Vaerenbergh, 2013. % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ N = size(X,1); if nargin<4 epochs = 15; end if nargin<3 time_ind = (0:N-1)'; % assume that samples are taken on synchronous time instants end Xt = [time_ind X]; % initialization values noisevar0 = mean(Y)^2/4; lambda0 = 0.99; ell0 = 1; sf20 = 1; % The standard covariance function in kernel adaptive filtering is a % Gaussian kernel with additive noise. Forgetting is introduced through an % AR(1) process with forgetting factor lambda. See publication for details. covfunc = {'kafbox_covSum',{{'kafbox_covProd',... {'kafbox_covLambda', 'kafbox_covSEiso2'}},'kafbox_covNoise'}}; loghyper0 = [log(lambda0/(1-lambda0)); log(ell0); log(sqrt(sf20)); 0.5*log(noisevar0)]; % Minimize negative log maximal likelihood using GPML toolbox loghyper1 = kafbox_minimize(loghyper0,'kafbox_gpr',epochs,covfunc,Xt,Y); lambda_est = exp(loghyper1(1))/(1+exp(loghyper1(1))); lambda_est = lambda_est^2; % correspondance between st-gp and krls-t ell_est = exp(loghyper1(2)); sf2_est = exp(2*loghyper1(3)); noisevar_est = exp(2*loghyper1(4)); sigma = ell_est; % length-scale reg = noisevar_est/sf2_est; % noise-to-signal parameter (regularization) lambda = lambda_est; % forgetting factor kafbox-2.0/lib/util/kafbox_predictionloop.m000066400000000000000000000012311303712747200211150ustar00rootroot00000000000000% Wrapper program for time series prediction. Performs progression of time % and calls adaptive filter during each iteration. % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ function [e,kaf] = kafbox_predictionloop(kaf,X,y,vb) if nargin<4, vb = 1; end N = size(X,1); e = zeros(N,1); for n=1:N, if ~mod(n,floor(N/10)) && vb, fprintf('.'); % progress indicator (10 dots) end y_est = kaf.evaluate(X(n,:)); % predict the next output e(n) = y(n)-y_est; % store error kaf.train(X(n,:),y(n)); % train with one input-output pair end if vb, fprintf('\n'); end; kafbox-2.0/lib/util/kafbox_quickrun.m000066400000000000000000000017011303712747200177260ustar00rootroot00000000000000% Quick-run an algorithm on a dataset. % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % http://sourceforge.net/projects/kafbox/ function kafbox_quickrun(kafname,datasetname,kafopt,dataopt) rng(1) t1 = tic; if nargin > 2, kaf = feval(kafname,kafopt); else kaf = feval(kafname); end % get data opt = struct('N_test',50); [X,y,y_ref,X_test,y_test] = generate_channel_switch(opt); %#ok N = size(X,1); N_switch = 500; MSE = zeros(N,1); for n=1:N, if ~mod(n,floor(N/10)), fprintf('.'); % progress indicator (10 dots) end y_est = kaf.evaluate(X_test); if n<=N_switch MSE(n) = mean((y_test(:,1)-y_est).^2); else MSE(n) = mean((y_test(:,2)-y_est).^2); end kaf.train(X(n,:),y(n)); % train with one input-output pair end fprintf(' %.2fs. Final MSE=%3.2fdB.\n',toc(t1),10*log10(mean(MSE(N-500:N,1)))); %% OUTPUT figure; plot(10*log10(MSE(:))) legend(kafname) kafbox-2.0/lib/util/kernel.m000066400000000000000000000033111303712747200160120ustar00rootroot00000000000000% Calculate the kernel matrix for two data sets. % Author: Steven Van Vaerenbergh, 2013 % % This file is part of the Kernel Adaptive Filtering Toolbox for Matlab. % https://github.com/steven2358/kafbox/ function K = kernel(X1,X2,ktype,kpar) N1 = size(X1,1); N2 = size(X2,1); switch ktype case 'gauss' % RBF kernel norms1 = sum(X1.^2,2); norms2 = sum(X2.^2,2); mat1 = repmat(norms1,1,N2); mat2 = repmat(norms2',N1,1); dist2 = mat1 + mat2 - 2*X1*X2'; % full distance matrix K = exp(-dist2/(2*kpar^2)); case 'laplace' % Laplace kernel norms1 = sum(X1.^2,2); norms2 = sum(X2.^2,2); mat1 = repmat(norms1,1,N2); mat2 = repmat(norms2',N1,1); dist2 = mat1 + mat2 - 2*X1*X2'; % full distance matrix K = exp(-sqrt(dist2)/(2*kpar^2)); case 'gauss-diag' % diagonal of RBF kernel K = exp(-sum((X1-X2).^2,2)/(2*kpar^2)); case 'gauss-aniso' % anisotropic RBF kernel D = kpar; % anisotropic scaling X1 = X1*D; X2 = X2*D; K = kernel(X1,X2,'gauss',1); case 'poly' % polynomial kernel p = kpar(1); % polynome order c = kpar(2); % additive constant K = (X1*X2' + c).^p; case 'linear' % linear kernel K = X1*X2'; case 'sum', a = kpar.a; ktype1 = kpar.ktype1; kpar1 = kpar.kpar1; b = kpar.b; ktype2 = kpar.ktype2; kpar2 = kpar.kpar2; K = a*kernel(X1,X2,ktype1,kpar1) + b*kernel(X1,X2,ktype2,kpar2); otherwise % default case error ('unknown kernel type') end