Avoiding too many figures from loop in GA Toolbox

I am creating a function which looks like this:

function F = fitnessfun(x)
...
  score = x1+x2+x3
...  

figure('Color',[1 1 1],'position', [500, 250, 640, 480]);
imagesc(score);
title(['Score: ', num2str(score)],'Fontsize',24);

Now, when I run this function in Genetic Algorithm toolbox, i.e. I call @fitnessfun from the Optimtoolbox, everything runs smoothly, except the script making a new figure window for each GA generation. Tried to prevent this by using classical trick:

figure('Color',[1 1 1],'position', [500, 250, 640, 480]);
hold on
imagesc(score);
title(['Score: ', num2str(score)],'Fontsize',24);
hold off

But that does NOT help, new figure windows are created until termination (e.g. 500th generation) and causing my PC to crash.

Any solution to close the figure window automatically everytime a new figure is opened?

Thanks in advance.

Answer 1

[OCDER]

One way to do this is to create a template figure, and then pass this to your function to prevent creating/closing a new figure every time (which is slow, especially when doing 500+ iterations).

function F = fitnessfun(x, varargin)
...
  score = x1+x2+x3
...  

%If you already have a figure, just modify what you need
if ~isempty(varargin) && isa(varargin{1}, 'matlab.ui.Figure')
    Gx = varargin{1};                  % figure handle
    Im = findobj(Gx, 'type', 'image'); % image handle
    Ax = findobj(Gx, 'type', 'axes');  % axes handle
    if ~isempty(Im) && ~isempty(Ax)
        Im.CData = score;
        title(Ax, ['Score: ', num2str(score)], 'Fontsize', 24);
        return
    end
end

%Otherwise, create a whole new figure
figure('Color',[1 1 1],'position', [500, 250, 640, 480]);
imagesc(score);
title(['Score: ', num2str(score)],'Fontsize',24);

Summon your function as this

for j = 1:1000
   if j == 1
       F = fitnessfun(x(j))  %Start a template figure the first time,
   else
       F = fitnessfun(x(j), gcf)  %Pass on current figure handle to your function
   end
end

[cniv_we]

Thanks!

[OCDER]

You're welcome!