jaykalinani/asterx

AsterX is a GPU-accelerated GRMHD code for dynamical spacetimes

asterx's Issues

[Improvement] Implement unit tests for reconstruction operators

It would be nice to have unit tests implemented for the reconstruction testing mathematical properties of these operators.

These should be in the spirit of test.cxx.

[Improvement] Schedule all IF condition parameter checks only once at the beginning

Currently, there are certain IF conditions within scheduled routine that check which reconstruction schemes, flux solvers, C2Ps and EOSs are used. One optimization could be to schedule these checks/settings only once in the beginning, as suggested by Michal Chabanov.

[Improvement] Extend flux solver to higher order

The implemented HLLE flux solver is only 3rd order accurate. We should extend it to higher order, as mentioned, for instance, in the Appendix A.2 of the second Spritz paper: https://arxiv.org/pdf/2012.10174.pdf

[Improvement] Reconstruct pressure instead of epsilon

Reconstructing epsilon, as done in fluxes.cxx, may lead to inaccurate long-term evolution of the vector potential, as reported in Espino et al 2022 (https://arxiv.org/abs/2210.13481).
It is suggested to reconstruct pressure instead of epsilon.

[Improvement] Include symmetry preserving summations

Implement symmetry preserving summations which take care of the associative property. See this paper for details: https://www.sciencedirect.com/science/article/pii/S0045793018308399

[BugFix] Implement higher order correction to fluxes

Implement high order correction to the fluxes, to extend to 4th or 6th order accuracy. See (Del Zanna et al. 2007, Equation A.21), and Appendix A.2 of Cipolletta+2021 (https://arxiv.org/pdf/2012.10174.pdf)

[Improvement] Add option to use ErrorEstimator thorn and disable local error estimator

Currently, only the local error estimator updates the regrid_error grid function. An option needs to be added to allow usage of the ErrorEstimator thorn, and in that case, disable the local error estimator. Schedule needs to be updated accordingly.

`AsterX` dependency in `AsterSeeds`

Currently, AsterSeeds can only set initial data for AsterX. This isn't ideal, as it would be nice to have a general-purpose thorn for ID instead of having multiple. This issue is connected to a related CarpetX issue.

As stated in that issue, one solution would be to move Avec to HydroBaseX. This requires the ability to set the rhs tag for these variables at runtime, which may not be currently supported.

Alternatively, AsterSeeds could use the method of VolumeIntegrals_vacuum to change which variable it is accessing via a CCTK_STRING parameter. However, I am not sure how this method will interact with the new tiling/parallelism in CarpetX.

[Improvement] Add more unit tests to Con2PrimFactory module

Need to add more unit tests for different con2prim schemes going over a range of values for velocities and magnetic fields.
https://github.com/jaykalinani/AsterX/blob/main/Con2PrimFactory/src/test.cxx