openPMD viewer: visualize AMR simulation

[kli-jfp]

Hi! I recently got into running WarpX with mesh refinement with multiple levels. I can view the individual levels separately, but not together. Is there a current solution for viewing them together in the same plot? What is the recommended way to visualize this?

Also, is there a way to create complex mesh refinement shapes or can you only make boxes as of right now? For example, I've tried creating a refined torus shape in 3D, but the result is a box enclosing the torus.

Thanks!

[RemiLehe]

Thanks for the question.

Regarding visualizing the fields with MR, there are two options:

• Having WarpX write the data in plotfile format, and then using yt (see here), which does have extensive support for MR visualization.
• Having WarpX write the data in openPMD format, and then using openPMD-viewer. In that particular case, you can use the fact that, for refined levels, the fields outside the MR patch are seen as NaN (MR Reads: Zero->NaN Padding openPMD/openPMD-viewer#334)
  So, if you call get_field with plot=True, and call the fields in the order going from low-level to high level, you will automatically see the right thing. (since, in matplotlib's imshow, NaNs are just transparent):

ts.get_field('E', 'x', plot=True)  # This only plots level 0
ts.get_field('E_lvl1', 'x', plot=True)
ts.get_field('E_lvl2', 'x', plot=True)

[kli-jfp]

Thanks Remi, that works great!

[RemiLehe]

Regarding refined refinement: thanks for reporting this. Did you use this feature, in order to create the patch:
#4299

One thing to keep in mind is that, by construction in WarpX/AMReX, the MR refinement patch consists of a collection of patches of size amr.blocking_factor. In other words, the granularity of the MR patch shape is amr.blocking_factor. In the above-mentioned torus case, maybe reducing amr.blocking_factor` could help?

[kli-jfp]

Yes, I used ref_patch_function(x,y,z). Thanks for the tip, I'll try different settings of blocking_factor and see if it works.