Wrappers for other frameworks

[ogencoglu]

Thanks for the great work!

Would any of the following (mostly gradient-free) optimization frameworks would be worth to add in your opinion, similar to nevergrad wrapper?

• https://github.com/dietmarwo/fast-cma-es
• https://numericalalgorithmsgroup.github.io/pybobyqa/build/html/index.html
• https://pysot.readthedocs.io/en/latest/
• https://github.com/andresliszt/moo-rs
• https://github.com/experimental-design/bofire

[inikishev]

Thanks for the great work!

Would any of the following (mostly gradient-free) optimization frameworks would be worth to add in your opinion, similar to nevergrad wrapper?

• https://github.com/dietmarwo/fast-cma-es
• https://numericalalgorithmsgroup.github.io/pybobyqa/build/html/index.html
• https://pysot.readthedocs.io/en/latest/
• https://github.com/andresliszt/moo-rs
• https://github.com/experimental-design/bofire

Thanks for suggestions!

Currently there are wrappers for scipy, nlopt, nevergrad, directsearch, mads and fcmaes, although I haven't gotten around to adding them to docs so they are kind of hidden for now

• https://github.com/dietmarwo/fast-cma-es - the wrapper is implemented here https://github.com/inikishev/torchzero/blob/main/torchzero/optim/wrappers/fcmaes.py

• https://numericalalgorithmsgroup.github.io/pybobyqa/build/html/index.html - I definitely want to add this as it is a good solver and easy to add by copying scipy wrapper

• https://github.com/andresliszt/moo-rs - good suggestion, I didn't know this one. Should be easy to add by copying scipy wrapper too, I'll look into it

• https://pysot.readthedocs.io/en/latest/, https://github.com/experimental-design/bofire - there are a few surrogate methods in optuna and nevergrad and more on OptunaHub, and BO isn't that useful as pytorch optimizer but let me know if you have a use case

[ogencoglu]

Thanks for the swift reply. Appreciated.