Output function for particleswarm command

Hi I'm trying to extract all the best values during particle swarm iterations using built in particle swarm function of matlab

if true
  % code
options = optimoptions('particleswarm','SwarmSize',100,'display','iter','PlotFcn',@pswplotbestf,'OutputFcn',@outfun)
x=particleswarm(fun,nvars,lb,ub,options)
end

So I've to use an output function to do that. Unfortunately, the descriptions in Matlab help didn't help. Until now, I've been able to show them on command window through this code:

if true
% code
function stop = outfun(optimValues,state)
stop = false;
switch state
   case 'iter'
       fprintf('%g %0.30f\n',[optimValues.iteration;optimValues.bestfval]')
%     ya(optimValues.iteration,1)=optimValues.bestfval;
%     assignin('base', 'var', ya)
end
 end

end

As you can see in the last two lines, I've tried to record the best values but it didn't work. It gives me a vector of zeros to the length of iterations (optimValues.iteration) and records the last best value only in the last cell as shown below:

if true
  % code

 var
var =

   1.0e-04 *

         0
         0
         0
         0
         0
         0
         0
         0
         0
         0
         0
         0
         0
         0
         0
         0
         0
         0
         0
    0.2915
  end

Can anyone help me write the right code to record the best values (optimValues.bestfval) in all iterations? P.S.: I'm using Matlab R2015a

Best Regards

[Ahmed Hussein]

This could simply be done using a global variable. However, I do not think this is the optimum solution. In my opinion the particleswarm implementation is still missing a lot of functionality, compared to ga implementation. In ga the best value in all iterations is already available in the state variable along with other useful information. For reference here is the Global Variable method:

function stop = outfun(optimValues,state)
global BstFval
stop = false;
switch state
    case 'iter'
        BstFval(optimValues.iteration) = optimValues.bestfval;
end

The variable is then accessible from anywhere. I would also appreciate if anyone has figured a better method to do this.

Answer 1

[Alan Weiss]

Try this function:

function stop = pswoutfun(optimValues,state)

persistent hist
stop = false;
switch state
    case 'init'
        hist = [0,optimValues.bestfval];
    case 'iter'
        nextline = [optimValues.iteration,optimValues.bestfval];
        disp(nextline)
        hist = [hist;nextline];
    case 'done'
        assignin('base','hist',hist);
end

Do you think that this is this worth giving as an example in the documentation?

Alan Weiss

MATLAB mathematical toolbox documentation

[Farshina Nazrul Shimim]

Hello Alan,

This is indeed very helpful and would be great if it were in the documentation. In my opinion, the documentation for PSO should have more explanations, specially with optimoptions. I keep getting errors for a lot of parameter tuning attempts.

Best,

Farshina