Fix GWO

[k-onoue]

Contributor Agreements

Please read the contributor agreements and if you agree, please click the checkbox below.

• I agree to the contributor agreements.

Tip

Please follow the Quick TODO list to smoothly merge your PR.

Motivation

Description of the changes

TODO List towards PR Merge

Please remove this section if this PR is not an addition of a new package.
Otherwise, please check the following TODO list:

• Copy ./template/ to create your package
• Replace <COPYRIGHT HOLDER> in LICENSE of your package with your name
• Fill out README.md in your package
• Add import statements of your function or class names to be used in __init__.py
• Apply the formatter based on the tips in README.md
• Check whether your module works as intended based on the tips in README.md

[k-onoue]

Apologies for the delay in making modifications despite receiving clear and helpful comments on the previous implementation.

I’ve addressed all the points mentioned in this PR, but I still have a few uncertainties.

One major point of concern is how to update the value of a, which serves as a decaying factor to balance exploration and exploitation. Initially, I treated a as linearly decreasing from 2 to 0 based on the number of trials, as follows:

# Linearly decrease from 2 to 0
a = 2 * (1 - len(study.trials) / (self.max_iter * self.population_size))

It was suggested that a should be updated on an iteration-by-iteration basis. However, after reviewing both the paper and the author-provided MATLAB implementation, I believe that the "Max number of iterations" in them corresponds to n_trials in Optuna. That said, I’m still not entirely confident, so any corrections or further suggestions would be appreciated.

Additionally, I’m unsure of the exact types of input variables that are allowed for this sampler, so I’ve currently omitted support for categorical distributions in _lazy_init method.

[HideakiImamura]

@eukaryo Could you review this PR?

[eukaryo]

LGTM! Sorry for my late reply.