amrex

amrex copied to clipboard

Summary

Further updates of #2853 where @atmyers and @ax3l can also push to.

Additional background

Checklist

The proposed changes:

• [ ] fix a bug or incorrect behavior in AMReX
• [x] add new capabilities to AMReX
• [ ] changes answers in the test suite to more than roundoff level
• [ ] are likely to significantly affect the results of downstream AMReX users
• [ ] include documentation in the code and/or rst files, if appropriate

@Thierry992 @atmyers let us make sure this PR is passing this week, because @Thierry992 and @np-eazy will use this in ImpactX for the hackathon next week :)

I think amrex::ParticleContainer::make_alike, e.g., in

// create a host-side particle buffer
auto tmp = pc.make_alike<amrex::PinnedArenaAllocator>();

is not yet working.

@Thierry992 please implement make_alike & ContainerLike for the new SoA container for ImpactX/WarpX: https://github.com/AMReX-Codes/amrex/blob/826cd378f8ba0d844c64e1029f7914c3b066debd/Src/Particle/AMReX_ParticleContainer.H#L1248-L1271

@Thierry992 I think (Const)ParticleTile methods:

• int numParticles () const { return m_aos_tile.numParticles(); }
• int numRealParticles () const { return m_aos_tile.numRealParticles(); }
• int numNeighborParticles () const { return m_aos_tile.numNeighborParticles(); }
• int numTotalParticles () const { return m_aos_tile.numTotalParticles() ; }

Do not yet return the number of particles for pure SoA particle containers, right?

@atmyers I updated this branch following AMReX' C++17 transition. Do you potentially have time to look further into the MPI debugging?

I fixed a couple of issues, the ImpactX examples seems to work on multiple ranks now.

Awesome - I merged development in and we can now fix the warnings and get this ready :rocket: :tada:

Looking right now in pyAMReX for changes from this PR to low-level data access: https://github.com/AMReX-Codes/pyamrex/pull/119

Currently, I cannot find these properties (this PR but old layout)

• ParticleTileType::size
• ParticleTileType::numParticles
• ParticleTileType::numRealParticles
• ParticleTileType::numNeighborParticles
• ParticleTileType::numTotalParticles
• ParticleTileDataType::getSuperParticle

update: Ah, maybe need to be explicit to resolve the sfinae in the methods in pybind code - all good.

@atmyers @WeiqunZhang shall we merge this now? :) That way, we have still two weeks to update downstream codes and fix potential regressions before tagging a new release.