JohannesBuchner / PyMultiNest

Pythonic Bayesian inference and visualization for the MultiNest Nested Sampling Algorithm and PyCuba's cubature algorithms.
http://johannesbuchner.github.io/PyMultiNest/
Other
191 stars 87 forks source link

Use Scikit-build for easier install #210

Open Nicholaswogan opened 2 years ago

Nicholaswogan commented 2 years ago

Would be more convenient if you used scikit-build for an easier install. The benefit is that the MultiNest library could just be built and contained within PyMultiNest package.

I've used it before for Fortran. For example: https://github.com/Nicholaswogan/NumbaMinpack

We can use CMake Package Manager to make MultiNest a dependency.

Anyway, if you are interested, I can put together a pull request. Let me know!

JohannesBuchner commented 2 years ago

Interesting. scikit-build says it is about CPython extensions, so I guess we would have to modify quite a bit? I'd rather keep MultiNest (which is not my code) identical to upstream, because otherwise I become in charge of maintaining it and resolving bugs ...

I am not sure whether there would be a potential licensing issue with redistributing MultiNest, since it is not FOSS.

JohannesBuchner commented 2 years ago

For a easier-to-install, free and open source software package, there is UltraNest.

Nicholaswogan commented 2 years ago

No modifications of MultiNest would be necessary.

I am not sure of the licensing.

I should look closer at UltraNest. Thanks!