Particle fields in RZ Geometry with openPMD

[PhysicsDan]

Description

Particle Fields do not work when using the openpmd backend in RZ geometry
(Only tested using the precompiled executable from conda)

Minimal Example

When adding the snippet

diag1.particle_fields_to_plot = z
diag1.particle_fields_species = electrons beam
diag1.particle_fields.z(x,y,z,ux,uy,uz) = z

to
inputs_test_rz_laser_acceleration_opmd
The code instantly fails with the error

Initializing AMReX (25.02)...
OMP initialized with 8 OMP threads
AMReX (25.02) initialized
PICSAR (25.01)
WarpX (Unknown)

    __        __             __  __
    \ \      / /_ _ _ __ _ __\ \/ /
     \ \ /\ / / _` | '__| '_ \\  /
      \ V  V / (_| | |  | |_) /  \
       \_/\_/ \__,_|_|  | .__/_/\_\
                        |_|

Level 0: dt = 1.72851157e-16 ; dx = 1.171875e-07 ; dz = 6.640625e-08
0::Assertion `ncomp_from_src == m_varnames.size()' failed, file "/Users/runner/miniforge3/conda-bld/warpx_1738873300730/work/Source/Diagnostics/FullDiagnostics.cpp", line 540 !!!
SIGABRT
See Backtrace.0.0 file for details

Backtrace

 0: amrex::BLBackTrace::print_backtrace_info(__sFILE*) (in libamrex_2d.dylib) + 64

 1: amrex::BLBackTrace::handler(int) (in libamrex_2d.dylib) + 640

 2: amrex::Assert_host(char const*, char const*, int, char const*) (in libamrex_2d.dylib) + 268

 3: FullDiagnostics::InitializeFieldFunctorsRZopenPMD(int) (in warpx.rz) + 7260

 4: Diagnostics::InitData() (in warpx.rz) + 152

 5: MultiDiagnostics::InitData() (in warpx.rz) + 32

 6: WarpX::InitData() (in warpx.rz) + 776

 7: main (in warpx.rz) + 76

 8: start (in dyld) + 2840


===== TinyProfilers ======
main()
WarpX::InitData()

However, when the snippet is added to this file (i.e. the same simulation but using the plotfile output type), there us no error
inputs_test_rz_laser_acceleration

Note, this is also not an issue in cartesian geometry as WarpX/Examples/Tests/particle_fields_diags also runs without any issue.

System information

MacOS 15.3
WarpX 25.02 via Conda

[ax3l]

Hi @PhysicsDan , thanks for the report!

Did you test if this still worked with 25.01? If yes, it's likely a regression to #5481

[ax3l]

Thanks for the report!

I wonder if creating fields from particles for diagnostics is at all supported in RZ yet...

Here is what I see debugging it:
RZ default case, the values of the assert are

ncomp_from_src=29
m_varnames.size()=29

The moment that we add a single particle_fields_to_plot as in the block you posted above:

ncomp_from_src=31
m_varnames.size()=35
0::Assertion `ncomp_from_src == m_varnames.size()' failed, file "/home/axel/src/warpx/Source/Diagnostics/FullDiagnostics.cpp", line 542 !!!

The problem indeed disappears if I remove

diag1.format = openpmd

, which is funny because this is frontend logic throwing the assert, before we go into the I/O backend.

The m_varnames are:

v=Er_0_real
v=Er_1_real
v=Er_1_imag
v=Bt_0_real
v=Bt_1_real
v=Bt_1_imag
v=Bz_0_real
v=Bz_1_real
v=Bz_1_imag
v=jr_0_real
v=jr_1_real
v=jr_1_imag
v=jt_0_real
v=jt_1_real
v=jt_1_imag
v=jz_0_real
v=jz_1_real
v=jz_1_imag
v=rho_0_real
v=rho_1_real
v=rho_1_imag
v=part_per_cell
v=part_per_grid
v=rho_beam_0_real
v=rho_beam_1_real
v=rho_beam_1_imag
v=rho_electrons_0_real
v=rho_electrons_1_real
v=rho_electrons_1_imag
v=z_electrons_0_real
v=z_electrons_1_real
v=z_electrons_1_imag
v=z_beam_0_real
v=z_beam_1_real
v=z_beam_1_imag

[PhysicsDan]

I wonder if creating fields from particles for diagnostics is at all supported in RZ yet...

Just to confirm, I have been able to use particle_fields with the plotfile outputs. The particle_fields outputs look to be ~correct

[ax3l]

Discussed today in the developer meeting:

• This works well for plotfiles and openPMD in RZ if we only dump mode zero for particle fields.
• We do not have support for multiple modes for particle-to-field yet and the output for plotfiles might not crash but is very likely wrong.
  • Add a clean error message.
  • Document in an issue.

[chiehjen]

I wonder if creating fields from particles for diagnostics is at all supported in RZ yet...

Just to confirm, I have been able to use particle_fields with the plotfile outputs. The particle_fields outputs look to be ~correct

Hi, I encountered similar problem, i.e., particle_fields_to_plot in RZ.

Although it runs until finish with plotfile format, the output of whatever within "particle_fields_to_plot" is always zero (other parts are correctly executed though).

The input file is "inputs_base_rz", but with the following diagnosis :

`
diag1.electrons.variables = x z w ux uy uz orig_x orig_z
diag1.beam.variables = x z w ux uy uz

diag1.particle_fields_to_plot = z
diag1.particle_fields_species = electrons
diag1.particle_fields.z(x,y,z,ux,uy,uz) = 1 + z

diag1.file_prefix = diags/plotfiles/plt
`

Then, I check the plotfiles with amrvis2d, the "z_electrons" is always zero.

Did I missed something here?