function [number_edges_store,InitialMatrix,FinalMatrix] = Organization_Control_An_Evolving_Interdependent_Population_Code(N,n_l,p,b,num_reproduce,numGen,tau,num_selected,beta,initMatrix,storeInitialMatrix,storeFinalMatrix)
% N = number of cells
% n_l = initial number of links
% p = number of mutations is approximately p*N^2
% b = benefit to fitness for taking vs. giving
% num_reproduce = number that reproduce (by replacement)
% numGen = number of generations
% initMatrix = given initial. Input of [] will generate a random one
% tau = when to apply dose
% num_selected = number of individuals selected in medicine step
% beta = probabilistic number of links affected

% initialize
if isempty(initMatrix)
    connectivitymatrix = spalloc(N,N,N+n_l);
    one_idx = randperm(N^2,n_l); % choose n_l random nodes (allowing diagonal entries to be included)
    connectivitymatrix(one_idx) = 1; % make n_l random links
else
    connectivitymatrix = sparse(initMatrix);
end

if tau == 1
    affected_cells = randperm(N,num_selected);
    for j = 1:length(affected_cells)
        choicevector = (rand(1,N) > beta);
        connectivitymatrix(affected_cells(j),:) = connectivitymatrix(affected_cells(j),:).*choicevector; % turn beta percent of 1s to 0 in the row
    end
end

if storeInitialMatrix == 'y'
    InitialMatrix = connectivitymatrix;
else
    InitialMatrix = []; % save blank
end

number_edges_store = zeros(1,numGen); % number of edges at a given timestep
number_edges_store(1) = nnz(connectivitymatrix);

% note: we treat diagonal elements (self-loops) as 1 INCOMING arrow and 1 OUTGOING arrow (0 benefit, increase in complexity)

for i = 2:numGen
    
    if i == tau
        affected_cells = randperm(N,num_selected);
        for j = 1:length(affected_cells)
            choicevector = (rand(1,N) > beta);
            connectivitymatrix(affected_cells(j),:) = connectivitymatrix(affected_cells(j),:).*choicevector; % turn beta percent of 1s to 0 in the row (affect giving)
        end
    end
    
    % Mutation
    connectivitymatrix = abs(connectivitymatrix-logical(sprand(N,N,p)));
    % Note: p should generally be relatively small compared to 1/N: if p*N^2 is large, there may be significantly fewer zeros generated by this function than expected

    
    % Fitness calculation
    connectivitymatrix = connectivitymatrix - diag(diag(connectivitymatrix)); % remove diagonals
    fitness_vector = b*sum(connectivitymatrix,1)-(sum(connectivitymatrix,2))';

    % Updating
    [~,I] = sort(fitness_vector);
    min_n = I(1,(1:num_reproduce));
    max_n = I(1,(end-num_reproduce+1:end));
    connectivitymatrix(min_n,:) = 0;
    connectivitymatrix(:,min_n) = 0;

    connectivitymatrix(min_n,:) = connectivitymatrix(max_n,:);
    connectivitymatrix(:,min_n) = connectivitymatrix(:,max_n);
    
    % record number of edges
    number_nonzero_elements = nnz(connectivitymatrix);
    number_edges_store(i) = number_nonzero_elements;
    
end

if storeFinalMatrix == 'y'
    FinalMatrix = connectivitymatrix;
else
    FinalMatrix = [];
end