Boltzmann interface dev

[uvilla]

Developing the interface for the Boltzmann solver interface.

Key changes:

• Implemented a Tps2Boltzmann class (both c++ and python wrappers) to handle the sharing of information between the TPS and Boltzmann solver. A new section [boltzmannInterface] in the input file has been added to pass parameters to the interface.
• Added functionalities in M2ulPhysS to push thermodynamic state variables to Tps2Boltzmann.
• Modified QuasiMagnetostaticSolver* to optionally compute/store the electric field.
• Interpolate the electric field to the fluid mesh in CycleAvgJouleCoupling.
• Add a demo of the new interface (including a BoltzmannMockSolver) in tps-time-loos.py.
• Resolved many memory leaks in QuasiMagnetostaticSolver* and M2ulPhysS.

Next milestones:

• Create a pytps python package
• Improve the interface of Tps2Boltzmann to include variable name dictionaries and units
• Extend Tps2Boltzmann to handle GPU data directly. Dependents on #228 that allow EM solver to run on the build for GPUs
• Implement methods to read back Boltzmann fields in TPS

[uvilla]

@trevilo :

There is a built in that fails in the VPATH Testing that I can not reproduce locally. Other than that this first chuck is ready to be reviewed and merged.

Cheers,

Umberto

[trevilo]

@trevilo :

There is a built in that fails in the VPATH Testing that I can not reproduce locally. Other than that this first chuck is ready to be reviewed and merged.

Cheers,

Umberto

Thanks @uvilla. The failing test is non-deterministic (by design unfortunately) so it fails occasionally. I'd like to change that, but... that is a battle for another day. I've triggered it to run again, and I'm guessing it will pass, especially since you couldn't reproduce locally. If it passes, I'll look this over and get it merged asap. If not, I'll take a closer look.