%% Correlation
function [correlation] = correl (n_samples, mean_value, standard_deviation, n_connections)
%a = 0:0.1:0.9;
a=0;
%correlation = zeros(1, size(a,2));
%for j = 1:size(a,2)
corr = zeros(1, n_samples);
for i =1:n_samples
x_unp = normrnd(mean_value, standard_deviation, 1, n_connections) .* (rand(1,n_connections) > a); %Also: mean_value + (standard_deviation).*randn(1,N);
y_unp = normrnd(mean_value, standard_deviation, 1, n_connections) .* (rand(1,n_connections) > a); %Also: mean_value + (standard_deviation).*randn(1,N);
x = x_unp((x_unp+y_unp)~=0); %uses logical index to cut away the zero elements from the distributions when they have the same index in x and y (w_{ij}=w_{ji}=0)
y = y_unp((x_unp+y_unp)~=0);
distr = [ x+y; abs(x-y)];
mean_val = [mean(x+y), mean(abs(x-y))];
variance = [ var(x+y), var(abs(x-y))];
covariance = sum((distr(1,:)-mean_val(1)) .* (distr(2,:)-mean_val(2))) / n_connections;
corr(i) = covariance / (sqrt(variance(1))*sqrt(variance(2)));
end
correlation = mean(corr);
% correlation(j) = mean(corr);
% display ('Iteration done')
%end