https://github.com/afd-illinois/kharma - KHARMA 2024.9

This release adds "excised" transmitting boundary conditions allowing fluxes across the pole, and makes EMHD problems much faster on AMD GPUs. It also updates and cleans up all of the included parameter files, updates the tests, and fixes a few obscure bugs.

There have also been a bunch of updates to the wiki, which should now be more useful for users and developers alike!

Full Changelog: https://github.com/AFD-Illinois/kharma/compare/v2024.5.1...v2024.9

- C++
Published by bprather almost 2 years ago

https://github.com/afd-illinois/kharma - KHARMA 2024.5.1

This release is all the bugfixes since the May release, without the features. * Most notable is that "transmitting" boundary conditions work without mitigations. They are not default because of the lingering requirement that one one meshblock be used in phi/X3, but they are recommended. * Also important is a fix to generating initial conditions with face-centered B fields. * Reductions and history files have been fixed as well, but history files are still pretty sparse by default. To re-enable all history outputs set parthenon/outputX/variables=all_reductions, see the PR. * Some minor fixes to restarting simulations. * Limiters reflecting physical limits on the sound speed and overall signal speed were added.

- C++
Published by bprather almost 2 years ago

https://github.com/afd-illinois/kharma - 2024.5

A few stability features. * Forced reconnection at the pole for transmitting boundary conditions * Some more flexible boundary conditions generally * Ideal GRMHD guesses in EGRMHD * More flexible floor prescriptions and FOFC application * Eliminated some compile warnings

- C++
Published by bprather about 2 years ago

https://github.com/afd-illinois/kharma - KHARMA 2024.4

Lots of features which were waiting for the last release (e-, multizone, reconstructions, CT schemes, etc), plus a ton of stability features to try to make the large-scale INCITE runs reliable with fewer of the traditional GRMHD crutches. * First-order flux corrections (FOFC) * Kastaun robust primitive variable recovery * Multizone improvements * Implicit/EMHD bugfixes * electron temperature tracking fixes and tests * Upwinded B fluxes * New reconstruction schemes * Split/simplified floor injection * Reconstruction fallback * "Transmitting" polar boundary conditions: correct ghosts and EMFs on polar edges, but no excision

- C++
Published by bprather about 2 years ago

https://github.com/afd-illinois/kharma - KHARMA 2023.12

Highlights: * SMR/AMR with new split face-centered B fields. New initialization for magnetic fields. * Basic support for "multi-zone" chained overlapping runs (see dev for more recent progress there, though) * EMHD with pivoted QR decomposition for stability * HIP support * Reorganized package structure with callbacks, simplified drivers (and new, simpler driver for testing) * Most features are now split into packages, reducing weight of GRMHD package and non-package components * Much faster and more extensible machinery for reductions, incl. counting failure/floor flags, etc. * New package for handling boundaries allowing user boundary choices * Simplify problems, e.g. single Bondi problem * Simplified range management with KDomain, eliminate some of the more confusing lambda conventions for loops * Move to nearly main-line Parthenon, simplify other dependencies

- C++
Published by bprather over 2 years ago

https://github.com/afd-illinois/kharma - KHARMA 2022.12

As used for up-scaled "focused" runs of the ngGRMHD library.

Support for interpolation and magnetic field cleanup, better tests & CI features. Significant code cleanup and several new convenience features when compiling. Faster file I/O and MPI Reductions thanks to Parthenon improvements.

- C++
Published by bprather over 3 years ago

https://github.com/afd-illinois/kharma - KHARMA 2022.3

KHARMA as used to begin "base" production runs of ngGRMHD INCITE project.

Fast, reliable, flexible & feature-complete core GRMHD evolution, plus core implementations of semi-implicit stepping and electron thermodynamics.

- C++
Published by bprather about 4 years ago