How can I enable adaptive mesh refinement with pyAMReX?

[adamdarx]

As titled, I didn't find any relevant codes in the manual about adaptive mesh refinement. It seems that both two example codes (HeatEquation.py and MultiFab.py) do not enable this function at all. I wonder whether there is a python code involving adaptive mesh refinement for reference. Thx!

[WeiqunZhang]

Not everything in amrex is ported to python. For an AMR code, you will have to write in C++ first. Then you can add some wrappers that allow you to drive your code from python.

[ax3l]

I agree with @WeiqunZhang for the current state.

One can certainly set up a lot for a MR simulation already from Python, e.g., setting the amr.max_level via amrex.ParmParse.add and likewise for amr.n_cell.

A couple of helper functions, e.g., from amrex::AmrCore to mark the cells for refinement are a bit hard to expose to Python right now due to their virtual nature, but once one has derived a C++ simulation class from them once could go back to Python and implement details of AmrCore::ErrorEst and AmrCore::MakeNewLevelFromScratch to mark and allocate cells for refinement from Python.

Update: for the last paragraph, we could actually expose AmrCore in pyAMReX (like this) and let users implement their simulation class fully in Python.