Visualize.m

function [e, active, AD, Perception, ToP, unqActive, unqA, Converge] =  Visualize(Initiated_Pattern, Noise, pat, AllStates, w, scale, showHeat, ActFun, stateMag, PerceptionThreshold)
%main function that asynchronously updates network states.


%% input variables
%initiated pattern: initial state of the network
% Noise: desired SNR input
% pat: stored patterns
% AllStates: all possible energy states, generated by EnergyLandscape.m
% w: weights
% scale: scaling factor (do not use, artifact of earlier version of code)
% showHeat: show heat map for associations (extent of overlap between
%           current network state and stored patterns
%ActFun: Activation function scaling factor to change threshold of
%           activation (do not use)
% stateMag: do not use
% PerceptionThreshold: description in main

%% output variable
% e: energy trajectory of network over some time span
% active: number of active nodes over time span
% AD: association distribution, extent of overlap for each time point of
%       network with each stored pattern
% remainder were added later and used to quantify results that are plotted
% at the very end. 

%%
global n %number of stored patterns
global N %total number of nodes
global time %time over which network will run
global figurePause % toggle for realtime network evolution
global go

r = Initiated_Pattern;
e = NetEnergy(w, r, N, stateMag);
ToP = []; %time during which network is makign clear association

Vis = figure(); set(0, 'CurrentFigure', Vis)
SNR = Noise; %SNR
NoA = [];


for t=2:time;
    maxidx = [];
    r(:,t) = sign(w*awgn(r(:,t-1),SNR)-ActFun);
% r(:,t) = sign(w*imnoise(r(:,t-1),'Poisson')-ActFun);
    

    overlap = (r(:,t)'*pat/N);
    maxoverlap = max(overlap);
    maxidx = find(overlap >= PerceptionThreshold);
    
    active(:,t) = sum(1 == r(:,t));
    AD(:,t) = overlap;
    e(t) = NetEnergy(w, r(:,t), N, stateMag);
    NoA= [NoA maxidx]; % store number of associations
    
    if maxidx > 0;
        ToP(1,t) = t;
        ToP(2,t) = e(t);
    end %time during which network is makign clear association
    
    if figurePause >0 | t==max(time)
        pause(figurePause);
        set(0, 'CurrentFigure', Vis)
        subplot(2,2,[1 2]); plot (e); xlim ([0 time]); xlabel ('Iteration Number'); ylabel('Network Energy');
%         title ([ num2str(N) ' Nodes, ' num2str(n) ' Patterns, ' num2str(SNR) ' SNR, w^{', num2str(scale), '} Scaling, ActFun = ' num2str(ActFun)]);
         title ([ num2str(N) ' Nodes, ' num2str(n) ' Patterns, SNR = ' num2str(SNR) ]);
        
        if t==max(time) & ~isempty(ToP)
            hold on; plot(ToP(1,:),ToP(2,:), 'o'); hold off
        end
        
        if length(AllStates)>0           
            maxState = refline(0,max(AllStates));
            maxState.Color = 'r';
            maxState.LineStyle = '--';
            
            minState = refline(0,min(AllStates));
            minState.Color = 'r';
            minState.LineStyle = '--';
            
            if go == 1;
                set(gca, 'ylim', [(min(AllStates)-(0.1*max(AllStates))) (max(AllStates)+(0.1*max(AllStates)))]);
                legend('Energy Trajectory',['>=' num2str(PerceptionThreshold*100) '% pattern overlap'], 'Energy Boundaries');
            else
                set(gca, 'ylim', [(min(AllStates)-(0.1*max(AllStates))) (max(AllStates)+(0.1*max(AllStates)))]);
                legend('Energy Trajectory', ['>=' num2str(PerceptionThreshold*100) '% pattern overlap']);
            end            
        end
        set(0, 'CurrentFigure', Vis)        
        if showHeat ==1
            subplot(223); imagesc(overlap); set (gca, 'xtick', [1:n]); xlabel('Pattern'); set (gca, 'ytick', []);
            c = colorbar(); caxis([-1 1]); ylabel(c, 'Overlap with Stored Pattern'); colormap ('bone');
            pause(0.0001);
            for k = 1:n
                Perception(k,t) =sum(NoA == k); %proportion of perception time taken on single pattern
                Perception(k,t) = 100*Perception(k,t)/time;
                 if Perception(k,t) > 0
                Converge(k) = 1;
                else 
                    Converge(k) = 0;
                end
            end
            set(0, 'CurrentFigure', Vis)
            subplot(224); bar(Perception(:,t)); title('Associations'); ylabel('Proportion of Time Perceiving (%)'); xlabel('Pattern');
        else
            for k = 1:n
                Perception(k,t) =sum(NoA == k); %proportion of perception time taken on single pattern       
                Perception(k,t) = 100*Perception(k,t)/time;
                if Perception(k,t) > 0
                Converge(k) = 1;
                else 
                    Converge(k) = 0;
                end
               
            end
            set(0, 'CurrentFigure', Vis)
            subplot(2,2,[3 4]); bar(Perception(:,t)); title('Associations'); ylabel('Proportion of Time Perceiving (%)'); xlabel('Pattern');
            ylim([0 100]);
        end
    end    
end
    active(:,t) = sum(1 == r(:,t));
    for i = 1:N 
        if sum(1==r(i,:))>0;
    unqA(i) = 1 ; %unique nodes that were activated
        else
             unqA(i) = 0;
        end
    end
    unqActive = 100*sum(unqA)/N; %percent of entire network that is activated other entire time course (unique activations)