Wrapping solvers from Siconos?

[SteffenPL]

I wanted to share this link: Which contains a list (and implementations) of many related solvers:
https://nonsmooth.gricad-pages.univ-grenoble-alpes.fr/siconos/users_guide/problems_and_solvers/lcp.html#

Maybe at some point, it could be useful to wrap their solvers for comparison. I could imagine that some of the people behind Siconos would also be interested in the possibilities Julia opens.

(However, it is probably too early at this stage.)

[avik-pal]

We could wrap Siconos in a separate Siconos.jl repository and integrate it with ComplementaritySolve.jl. That way, we could also differentiate through their solvers which would be interesting.

[ChrisRackauckas]

Yes, and we definitely need to since that would be important for benchmarks in any paper.

[avik-pal]

For CP we only need to wrap the numerics part of siconos which happens to be in C fortunately https://github.com/siconos/siconos/tree/master/numerics

[SteffenPL]

Sounds good, obviously, I'm quite motivated for Siconos as well for the differential inclusions part ;)
Also for CPs they have a lot of tests which would be nice here as well.

@avik-pal shall I try to get Siconos.jl to become a thing, or do you want to do that?

[avik-pal]

@SteffenPL, you can try to get it to work. That will be very useful.

[SteffenPL]

@avik-pal, I'm on it. I just wanted to let you know that I'm away for the next two weeks. I will be more active in August.