search

ECP-WarpX / impactx

high-performance modeling of beam dynamics in particle accelerators with collective effects
https://impactx.readthedocs.io
Other
26 stars 20 forks source link

Update ABLASTR & pyAMReX #715

Open ax3l opened 5 days ago

ax3l commented 5 days ago

To latest commit on development each.

ax3l commented 4 days ago

Test that shows changes:

ax3l commented 4 days ago

Investigating if the changes in https://github.com/ECP-WarpX/WarpX/pull/5284 just need a slightly higher tolerance now or if we cut too close to the beam.

ax3l commented 4 days ago

@RemiLehe even if I increase resolution (amr.n_cell), wrapping around the beam geometry.prob_relative https://github.com/ECP-WarpX/impactx/blob/3a8ab8dc981aafccc9ab3de167e222d986f8b374/examples/cfchannel/input_cfchannel_10nC_fft.in#L43-L45 and rtol I do not et it to pass our tolerance again so far: https://github.com/ECP-WarpX/impactx/blob/3a8ab8dc981aafccc9ab3de167e222d986f8b374/examples/cfchannel/analysis_cfchannel_10nC.py#L82

We expect these values:

    [sigx, sigy, sigt, emittance_x, emittance_y, emittance_t] ==
    [
        1.2154443728379865788e-003,
        1.2154443728379865788e-003,
        4.0956844276541331005317e-004,
        1.000000000e-006,
        1.000000000e-006,
        1.000000000e-006,
    ]

and I see these:

    [
        0.00111891388596667,
        0.0011285746858011444,
        0.0003732679938919104,
        9.993789938274477e-07,
        1.0106011946996333e-06,
        1.0016124683067494e-06
    ]
EZoni commented 4 days ago

I was thinking that another possibility to check this would be to switch on/off the change of https://github.com/ECP-WarpX/WarpX/pull/5284 (a one-line change) locally and see what effect it has on a given test that currently fails, if you haven't already tried.

ax3l commented 4 days ago

Thanks @EZoni - that line causes the issue.