I've found that azimuthal average can crash in parallel when the M_max<L_max. The issue seems to be to do with masks, the error ends in
in operate
comp_out[m0_out] = comp_in[m0_in]
~~~~~~~~^^^^^^^^
ValueError: NumPy boolean array indexing assignment cannot assign 96 input values to the 64 output values where the mask is true
Attached is mwe_az_av.py which shows the issue. Run as
mpiexec -n 8 python3 mwe_az_av.py
I've found that azimuthal average can crash in parallel when the M_max<L_max. The issue seems to be to do with masks, the error ends in
Attached is mwe_az_av.py which shows the issue. Run as