Fitness function increases rather than decreasing with fmincon using hybrid Genetic Algorithm

Question

Ankur Shah 2020-7-28

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/571762-fitness-function-increases-rather-than-decreasing-with-fmincon-using-hybrid-genetic-algorithm

评论： Matt J 2020-7-29

While running a hybrid GA with fmincon for minimising the objective function to 0. Once GA stops and fmincon starts. The fitness value in iteration as shown in figure instead of decreasing increases. GA achieves perfect solutions, but with fmincon usage approximate solution is obtained. I am using interior algorithm with central difference in fmincon. Note this code is related to my topic. For standard bencmark problems the process works just fine .

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

Matt J 2020-7-28

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/571762-fitness-function-increases-rather-than-decreasing-with-fmincon-using-hybrid-genetic-algorithm#answer_472024

This might happen if your objective has discontinuities or non-differentiabilities. Such functions are outside the scope of fmincon.

2 个评论
显示无隐藏无

Ankur Shah 2020-7-29

编辑：Ankur Shah 2020-7-29

Hey Matt,

It might be an issue. When using GA with fminunc and even fminsearch, both the functions can't decrease the fitness function any further than that by GA. Is there anything which can be done to minimize further?