M1 Support about fast-cma-es

[dietmarwo]

Don't think that it is possible to use the C++ algorithms on M1 at the moment without rosetta (the arm64 emulator). My plan is to port fcmaes to Rust, then it should be easy to compile it on any platform. Will take some time. But you may use the pure Python algorithms.

[dietmarwo]

Update for fcmaes version 1.5.8 recently released (pip install fcmaes --upgrade):

Pure Python versions of the algorithms are now usable also for parallel retry. Pure Python features: Algorithms: CMA-ES, CR-FM-NES, DE, MODE (multiple objective), Map-Elites+Diversifier (quality diversity). All Python algorithms support an ask-tell interface and parallel function evaluation. Additionally parallel retry / advanced retry (smart boundary management) are supported for these algorithms.

You may have to delete
/opt/homebrew/Caskroom/miniforge/base/envs/petri/lib/python3.9/site-packages/fcmaes/lib/libacmalib.dll
to avoid the error you get.

According to
https://stackoverflow.com/questions/65970469/what-does-platform-system-and-platform-architecture-return-on-apple-m1-silic

I would have expected that

elif 'mac' in sys.platform or 'darwin' in sys.platform:
libcmalib = ct.cdll.LoadLibrary(basepath + '/lib/libacmalib.dylib')

should have been loaded. Since I have no mac this shared library file is slightly outdated and would anyway only work under rosetta (the arm64 emulator on the M1). It is probably better to avoid the C++ algorithms on M1.

[iyzg]

That seems to have done the trick! Thanks so much for the help :)