Benchmarking MHD-drift-kinetic EP simulations

[spossann]

Scope

The aim of this project is to benchmark the model LinearMHDDriftkineticCC with other codes like HYMAGIK and ORB5. As a representative test case we will use, among others, the resonant TAE from Byung-Kyu 2025, section 5.2. This case is a well-established benchmark from Koenies et al. 2018.

1. Preparation

• Install Struphy from source (devel branch) on your laptop and get familiar with tutorials, Userguide etc.
• Check the new Claude simulation skill Added setup-simulation claude skill #304.
• Install Struphy from source (devel branch) on the clusters Viper and Raven and the TOK cluster and test it using the batch system (SLURM). Simple test files can be found here: https://github.com/struphy-hub/struphy/tree/devel/profiling/examples. An example scaling test can be run by executing python submit_poisson_strong_scaling.py where the submit scripts are here: https://github.com/struphy-hub/struphy/tree/devel/profiling.
• Report back the results of the Poisson strong scaling test on the three clusters.

2. Weak and strong scaling tests for TAE mode

At this stage we turn off the energetic particles and just run the TAE mode from the above papers with the model LinearMHD.

• Set up and test the TAE mode on your laptop, for only a few time steps.
• With the Poisson scaling test as an example, set up strong and weak scaling tests for the TAE in the folder https://github.com/struphy-hub/struphy/tree/devel/profiling/examples. The corresponding submit_*.py script goes to https://github.com/struphy-hub/struphy/tree/devel/profiling. Much like the Poisson test, the TAE test must check in post-processing whether the obtained results are correct and save some pictures and possibly data (like errors).
• Report the results of the TAE scaling tests on the three clusters.
• Make a PR into devel to commit the new scaling tests to the repo.

3. TAE with energetic particles

The model LinearMHDDriftkineticCC is currently being re-factored in #275. We need to wait for this PR to be merged before we can continue. @spossann is on it - it is almost finished.

@xinwang-ipp @Sinkhole09 @max-models

[xinwang-ipp]

Hi Stefan, ITPA TAE is an adhoc equilibrium. Can STRUPHY read realistic equilibrium input?

[spossann]

Hi @xinwang-ipp, yes it can read EQDSK, GVEC and DESC equilibria:

https://struphy-hub.github.io/struphy/sections/fluid-equils.html

[xinwang-ipp]

Then Shrut can run my current case which i used for ORB5, GENE. What do you think?

[spossann]

Alfven mode, energetic particles, in Tokamaks, linear physics, using TAE as an example. Teach me everything.

[Sinkhole09]

Mode structure, real frequency, growth rate, electrostatic potential,