Does genetic algorithm take a long time to find a solution?

Question

Wendy 2014-6-20

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/136676-does-genetic-algorithm-take-a-long-time-to-find-a-solution

回答： Prateekshya 2024-7-22

Hi all,

I am using the genetic algorithm to find the solution for 28 kinetic rates. The system consists of 14 ODEs and 9 time course data points. I use dde23 to solve the equations. It seems that GA is really slow. I set the display option as 'iter'. I did not get output after running my program overnight. This is the first time that I use GA optimization so that I am not sure if my case is usual. Any comment is appreciated. Thank you.

Wendy

1 个评论
显示 -1更早的评论隐藏 -1更早的评论

Star Strider 2014-6-20

Did you change your code so the sum squared error is the value you are minimising, and not its inverse?

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Prateekshya 2024-7-22

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/136676-does-genetic-algorithm-take-a-long-time-to-find-a-solution#answer_1488881

在 MATLAB Online 中打开

Hi Wendy,

I understand that Genetic Algorithm is taking more time than expected in your case. Here are a few troubleshooting steps:

Set appropriate tolerances for the ODE solver to balance accuracy and speed.If possible, vectorize the calculations within the ODE function to speed up evaluations.
MATLAB's Global Optimization Toolbox supports parallel computing, which can significantly speed up GA evaluations by distributing them across multiple cores or workers.

options = optimoptions('ga', 'UseParallel', true, 'Display', 'iter');

You may reduce the Population Size and Generations for the first iteration and increase it later if needed. You may also try using custom initial population.

I hope this helps!

Thank you.