I am using the scipy.optimize.**differential_evolution **function with the randtobest1binstrategy (this is the one that works the best afer testing) for optimization. After around 50 iterations, I have noticed that the RMSE increases and becomes more volatile. (as shown in the attached plots).
- What **parameters **should I adjust to improve the performance and stability of the optimization process?(The parameters I am using now are all set to their default values.)
- Why might the estimation error exhibit such behavior, and where should I start with troubleshooting and improvements?
My code:
<code>result = differential_evolution(
lambda Q: func(Q),
bounds=bound(q),
strategy=randtobest1bin,
)
</code>
<code>result = differential_evolution(
lambda Q: func(Q),
bounds=bound(q),
strategy=randtobest1bin,
)
</code>
result = differential_evolution(
lambda Q: func(Q),
bounds=bound(q),
strategy=randtobest1bin,
)
The reference for differential_evolution