Edoardo Zoni

Small update: I re-ran the test with the latest version of WarpX and can confirm that I still see the issue reported here. Note that the RZ test runs quite...

@RemiLehe One difference between the 3D input and the RZ input is about the plasma lenses. In the 3D input we set: ``` particles.B_ext_particle_init_style = repeated_plasma_lens particles.repeated_plasma_lens_period = 0.35 particles.repeated_plasma_lens_starts...

This is an updated version of the original input file (with only small changes that do not change the results significantly), sorted so that it can be compared more easily...

As a result of the latest debugging efforts, these are the most prominent conclusions: 1. Current correction to guarantee charge conservation plays a fundamental role and needs to be activated...

Here's the result for a setup very similar to the original one, [inputs_issue_2462.txt](https://github.com/ECP-WarpX/WarpX/files/8672138/inputs_issue_2462.txt), run on 24 GPUs on Summit, with AMReX (22.04-24-gf2d7f67848de), WarpX (21.10-585-g4a7c47678d42-dirty) and PICSAR (a621a5a41fb8), with the only...

The following videos show the electric field Ey and Ez (mirrored), the electron beam, and the first and second plasma stages, with a combination of fixes from #3092, #3101, and...

CI tests currently failing: - `VayDeposition2D` (checksums) - `bilinear_filter` (error v. tolerance): fixed by 9127c44 - `VayDeposition3D` (checksums) - `LaserAcceleration_single_precision_comms` (checksums)

Setting this as "[WIP]" while investigating with @oshapoval a potential bug that shows up only in DEBUG mode.

@oshapoval @RemiLehe The bug briefly mentioned in the last comment above is due to the fact that the real-space box `realspace_ba` used to construct the object `spectral_solver_fp` is grown by...

@RemiLehe The checksum changes, all related to CI tests that use mesh refinement, are due to the fact that, by modifying `ng_alloc_J` directly in the guard cells manager (so that...