Induviduals dissapearing after some iterations using GA algorithm and parsim.

Question

Jesper Ramsberg 2023-3-29

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1937354-induviduals-dissapearing-after-some-iterations-using-ga-algorithm-and-parsim

回答： Sugandhi 2023-8-23

Hi!

When using the built in GA algorithm optimising on a simulink model using parsim the number of simulations ran each iterations seem to reduce by two after one or two iterations. What can the explaination be for this.

options2 = optimoptions('ga',...

'Display','diagnose',...

'MaxTime',5000,...

'MaxStallGenerations',10,...

'PopulationSize',population,...

'MaxGenerations',10,...

'FunctionTolerance',0.001,...

'UseVectorized',true,...

'PlotFcn',{@gaplotbestf,@gaplotstopping, @gaplotbestindiv});

x = ga(@CostFunction3,nvars,A,B,[],[],lb,ub,[],options2);

In function

results = parsim(SimIn,'TransferBaseWorkspaceVariables','on','ShowProgress','on','UseFastRestart','on','StopOnError','on');

I thought as long as I dont use non linear constraint i would get as many simulation runs as population size. In this case the population was 40, but already on the second iteration I got only 38 simulation runs.

Thanks!

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Sugandhi 2023-8-23

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1937354-induviduals-dissapearing-after-some-iterations-using-ga-algorithm-and-parsim#answer_1291717

Hi Jesper Ramsberg,

I understand that when you are using the built in GA algorithm optimising on a Simulink model using ‘parsim’, individuals disappearing after some iterations.

Few of the possible workarounds could be:

Check the Simulink model and ensure that it is configured correctly, without any errors or issues that could cause simulations to fail.
Review the output and error messages from the ‘parsim’ function to identify any specific simulation failures or errors that may be occurring.
Verify the constraints in your optimization problem and ensure that they are correctly defined and feasible for the given parameter bounds.
Monitor the parallel computing environment and check if there are any limitations or resource constraints that could affect the number of available workers.