function [sign_eigenval]=Sign_subset_1el
%used to compute Figures 5 and 6 in DBS paper
%uses PCA analysis to find the # of significant eigenvalues among those
%computed via PCAgenerator.m
%we use 80% threshold to choose the significant subset of eigenvalues
load PCA_eigenval_w0.3.mat
S = 0;
M = 0;
sign_eigenval = zeros(38,10,5);
for i = 1:38
for j = 1:10
for k = 1:5
if size(pca_eigenval{i,j,k}) ~= 0
n = 10;
S = sum(pca_eigenval{i,j,k});
M = pca_eigenval{i,j,k}(n);
while n >= 1
n = n - 1;
if M > 0.8*S
sign_eigenval(i,j,k) = 10 - n;
n = 0;
else M = M + pca_eigenval{i,j,k}(n);
end
end
else
sign_eigenval(i,j,k) = 0;
end
end
end
end
save('Sign_eig_w0.3.mat','sign_eigenval')
x1_ix = 1;
for iapp = 5:5
x1_ix=x1_ix+1;
x2_ix = 0;
for gsyn = .5:.1:1.4
x2_ix=x2_ix+1;
x3_ix = 0;
for Kn = 0:2:74
x3_ix=x3_ix+1;
%use computed PCA with Cn=0 as a baseline
ii = sign_eigenval(x3_ix,x2_ix,x1_ix)-sign_eigenval(1,x2_ix,x1_ix);
if ii <= -3
color='k';
elseif ii == -2
color='b';
elseif ii == -1
color='b';
elseif ii == 0
color='w';
elseif ii ==1
color='y';
elseif ii ==2
color='y';
elseif ii >= 3
color='r';
end
scatter(gsyn,Kn,25,color,'filled')
hold on
end
end
xlabel('g_{syn}');ylabel('K_n');
fOut2 = sprintf('diffgsynKnplane_w0.3_%s%1.1f.fig','Iapp',iapp);
hgsave(fOut2);
end
end