Restart error when using a coarsened field diagnostic

[mtrocadomoreira]

Description

Simulation restart from a checkpoint fails when there is a coarsened field diagnostic (see error message below).

Expected behavior

The simulation runs successfully on the first go, and saves checkpoint files. I expected the simulation to restart from the saved checkpoint files successfully.

How to reproduce

I have included a minimal working example (MWE) to reproduce the error. Please run input_restart first, which will save some diagnostics and a checkpoint. Then run input_checkpoint to attempt the restart and reproduce the error.

The error only happens if the simulation runs long enough (restarting from iteration 500 in the MWE does not lead to the error), which I suspect has something to do with the time at which the plasma electrons are introduced.

System information

Please check all relevant boxes and provide details.

• Operating system (name and version):
  • Linux: e.g., Ubuntu 22.04 LTS
  • macOS: e.g., macOS Monterey 12.4
  • Windows: e.g., Windows 11 Pro
• Version of WarpX: 26.04 (26.04-64-g6b32fecc73d3)
• Installation method:
  • Conda
  • Spack
  • PyPI
  • Brew
  • From source with CMake
  • Module system on an HPC cluster
• Other dependencies: no
• Computational resources:
  • MPI: 1 MPI processes
  • OpenMP: 128 OpenMP threads
  • CPU: 128 CPUs
  • GPU:

The HPC cluster this was run on (ARCHER2) is not yet supported in the WarpX documentation.

Steps taken so far

• I have reduced the original input file to a MWE that can reproduce the problem
• I have tried to set the diag_lo and diag_hi parameters explicitly for the coarsened diagnostic
• I have tried to switch the order in which the checkpoint and coarsened diagnostics are defined in the input file

I have not tried debugging the code.

Additional information

Error message:

0::Assertion `m_crse_ratio[idim]==1' failed, file "/work/e985/e985/moreira/software/src/warpx/Source/Diagnostics/FullDiagnostics.cpp", line 750, Msg: 
### ERROR   : coarsening ratio in reduced dimension must be 1.
 !!!
SIGABRT
See Backtrace.0.0 file for details
MPICH ERROR [Rank 0] [job id ] [Fri Aug 28 17:51:03 2026] [ln01] - Abort(6) (rank 0 in comm 496): application called MPI_Abort(comm=0x84000000, 6) - process 0

I have included the log and the Backtrace file in the attachment.

Backtrace0.0.txt
input_checkpoint.txt
input_restart.txt
log.txt

[ax3l]

Hi @mtrocadomoreira,

It's great to read you again 🎉 Welcome to WarpX!

Thanks for reporting this bug, I'll try to reproduce it and give an update here.

[ax3l]

@atmyers this is a variant of the bug in #6392 #6944

@mtrocadomoreira it is possible the patch in #6399 (alt: #6400) might already help you. @atmyers is working on a clean version of this as well.

[mtrocadomoreira]

Hi @ax3l !
Thank you very much for the quick response. The #6399 patch solves the problem for me! I've subscribed to the PR anyway to keep up with the final implementation.
Thanks again!