search

IMTEK-Simulation / NuMPI

Utilities for MPI-parallel numerical calculations with Python
MIT License
2 stars 1 forks source link

Feat/20 parallel bugnicourt arbitrary bounds #41

Closed sannant closed 3 years ago

sannant commented 3 years ago

I have done two things:

I parallelized Bugnicourt.

I introduced the possibility to provide bounds different from 0 in Bugnicourt. This is the simplest way to handle simulations with padding regions in ContactMechanics (FreeFFTElasticHalfspace).

@sindhu2993 please have a look and check I didn't break anything.

We would then merge this into the branch 20_optimizer.

Last step is to clean up the code for bugnicourt and polonsky and keer, and merge this into the main branch.

I assume that we now stick to keeping the two algorithms implemented separately.

sindhu2993 commented 3 years ago

So I have tested it for mean value and non-adhesive problem. It works good.

For the adhesive problem, I will use the bugnicourt benchmark test, so i will run this probably in the night.

sindhu2993 commented 3 years ago

I have tested against the bugnicourt benchmark as well. Everything works perfect!

sannant commented 3 years ago

I last minute verified that the case with padding area works as well in Adhesion. You can merge it.