How to create a .gif animation visualizing the progress of an optimization process?

I am using a genetic algorithm to optimize the shape of a bezier curve for a specific application. The curve is generated using the x- and y-coordinates of its four control points. The two end control points are kept fixed during the optimization process, while the x- and y-coordinates of the two other (middle) control points are considered the design variables.

I am asking about the possibility of creating an animation to visualize the progress of the optimization process (i.e., plot the bezier curve at each iteration and create a .gif animation from those plots).

Note: some irrelevant parts of the code are removed for simplicity.

[x,fval,exitflag,output] = ga(@objectiveFcn,4,A,b,Aeq,beq,lb,ub,@nonlcon,[],options)

function f = objectiveFcn(x)
    xc = [0  x(1)  x(2)  1];          % x-coordinates of the 4 control points of the bezier curve
    yc = [0  x(3)  x(4)  0];          % y-coordinates of the 4 control points of the bezier curve
    
    % generating the bezier curve given the control points
    t = 0:0.01:1;
    for i = 1:length(t)               
        xb(i) = xc(1)*(1-t(i))^3 + 3*xc(2)*t(i)*(1-t(i))^2 + 3*xc(3)*(1-t(i))*t(i)^2 + xc(4)*t(i)^3;
        yb(i) = yc(1)*(1-t(i))^3 + 3*yc(2)*t(i)*(1-t(i))^2 + 3*yc(3)*(1-t(i))*t(i)^2 + yc(4)*t(i)^3;
    end
    % plot the bezier curve
    plot(xb,yb)
    .
    .
    .
    
    % calculation of the objective function in ANSYS
    system('C:\"Program Files"\"ANSYS Inc"\v221\Framework\bin\Win64\RunWB2.exe -B -R finaljournal.wbjn');

    % reading the value of the objective function
    f = xlsread('export data1.csv','export data1','J8')
end

[Voss]

For MATLAB versions older than R2022a:

https://www.mathworks.com/help/matlab/ref/imwrite.html#mw_00bce5fe-d730-4bcb-8104-bcebe7d5262a

For MATLAB R2022a and newer:

https://www.mathworks.com/matlabcentral/answers/94495-how-can-i-create-animated-gif-images-in-matlab

[Usama]

Thanks so much! This is really helpful.

Any idea how to update the value of idx inside the objective function in:

im{idx} = frame2im(frame);

Because I want to save a seperate image (im) each time the objective function is evaluated.

Answer 1

[chicken vector]

The simplest way is to add a few lines of command that saves your current objective function data.

You can add something like this to objectiveFcn after the xlsread command:

for j = 1 : 10
    f = rand(10,1);
    try
        load('optimisationData.mat');
        iterationData(length(iterationData)+1).objectiveFunction = f;
        save('optimisationData.mat','iterationData');
    catch
        iterationData = struct;
        iterationData(1).objectiveFunction = f;
        save('optimisationData.mat','iterationData');
    end
end

You can later access the data from the .mat file to create the .gif:

load('optimisationData.mat');
xLimits = [0 length(iterationData)] +.5;
yLimits = [.9*min([iterationData(:).objectiveFunction]) 1.1*max([iterationData(:).objectiveFunction])];
figure('WindowState','maximized')
grid on;
for frame = 1 : length(iterationData)
    b = bar([iterationData(1:frame).objectiveFunction]);
    xlim(xLimits);
    ylim(yLimits);
    exportgraphics(gcf,'optimisation.gif','Append',true);
    delete b
end

[Usama]

I don't understand why you added the line

f = rand(10,1);

Please note that the value of the objective function is stored in f as well, and now you are updating it's value by the vector rand(10,1).

Also, what is the pupose of the for loop:

for j = 1 : 10

[chicken vector]

I tried to setup an example to show you the working principle but I should have explained it better.

The part of code you are interested in is the following:

try
    load('optimisationData.mat');
    iterationData(length(iterationData)+1).objectiveFunction = f;
    save('optimisationData.mat','iterationData');
catch
    iterationData = struct;
    iterationData(1).objectiveFunction = f;
    save('optimisationData.mat','iterationData');
end