Obtaining a matrix of all the values the optimizer tried

Hi there, I have this optimizer minimizing f. The optimizer is trying various values for x variable. Would you maybe have an idea how I can obtain when the optimizer finishes a matrix with the x values the optimizer tried?

clearvars
f = @(x) x.^2;
x0=5;
x_min = -10;
x_max = 10;
TimeLimit_SA = 20;
options = optimoptions(@simulannealbnd,'TimeLimit',TimeLimit_SA);
[x_best,f_best,exitflag,output]=simulannealbnd(f,x0,x_min,x_max,options);

Answer 1

[Matt J]

You would have to write an OutputFcn to do this.

[Spyros Polychronopoulos]

Thank you Matt for your answer. I have already found that but it is rather confusing. I will try implementing it but if you have a quick example it would be very helpful. Thank you

[Matt J]

Here is an example of how to save iteration history with an OutputFcn.

https://www.mathworks.com/help/matlab/math/output-functions.html#bsgpq6q-31

The example is from the (non-Global) Optimization toolbox, but it will work very similarly for simulannealbnd

[Spyros Polychronopoulos]

Thank you very much Matt it is very clear. I didn't manage to do it for simulannealbnd yet but I am sure it wouldn't be difficult. The issue is that I have a massive function in my original code. Is it possible to add this function as an input to myproblem function? something like:

z = exp(x(1))*(4*x(1)^2+2*x(2)^2+x(1)*x(2)+2*x(2));
[x fval history] = myproblem(-1 , 1, z);

[Matt J]

Sure, just pass a handle,

[x fval history] = myproblem(-1 , 1, z, @massivefunction)

[Spyros Polychronopoulos]

Most probably I am doing something wrong here, cause I am trying what you said and I get an error importing zz. This is my main code

zz = @(x) x;
[x, fval, history] = myproblem(-1 , @zz);

and this is myproblem.m

function [x, fval, history] = myproblem(x0, n)

    history = [];
    options = optimset('OutputFcn', @myoutput);
    [x, fval] = fminsearch(@objfun, x0,options);
        function stop = myoutput(x,optimvalues,state);
        stop = false;
        if isequal(state,'iter')
          history = [history; x];
        end
    end
    function z = objfun(x);
        z=n;
    end
end

[Matt J]

I think maybe this is what you were trying to do:

zz = @(x) x.^2;
[x, fval, history] = myproblem(-1 , zz)

function [x, fval, history] = myproblem(x0, objfun)
    history = [];
    options = optimset('OutputFcn', @myoutput);
    [x, fval] = fminsearch(objfun, x0,options);
        function stop = myoutput(x,~,state)
        stop = false;
        if isequal(state,'iter')
          history = [history; x];
        end
    end

end

[Spyros Polychronopoulos]

That's it! Thank you very much. I will try to do it with simulannealbnd now.

[Spyros Polychronopoulos]

I have done the same thing for simulated annealing but the history matrix comes up empty, plus the time constrain is not working. Any ideas?

main

zz = @(x) x; %objective function
z0=5; %starting value
LB=-20; %low bound
UB=20; %upper bound
TimeLimit_SA = 1; %time constrain
[x, fval, history] = sa_store(zz, z0 ,LB,UB,TimeLimit_SA);

sa_store

function [x, fval, history] = sa_store(objfun,x0,LB,UB,t_Lim)
  history = [];
  options = optimset('MaxTime',t_Lim,'OutputFcn', @myoutput);
  [x, fval] = simulannealbnd(objfun,x0,LB,UB,options);
      function stop = myoutput(x,~,state)
      stop = false;
      if isequal(state,'iter')
        history = [history; x];
      end
  end
end