There is a known bug in the FIELDLINES (and possbily the BEAMS3D) codes where if one attempts to run the code with fewer field lines than processors the code will crash on writing to the HDF5 file. The fix is to add some code to the fieldlines_write_parhdf5.f90 code which creates a subcommunicator of just those processors which actually did work for the paralle HDF5 part. For the serial HDF5 part we probably just need to skip those threads whose mystart is greater than myend.
There is a known bug in the FIELDLINES (and possbily the BEAMS3D) codes where if one attempts to run the code with fewer field lines than processors the code will crash on writing to the HDF5 file. The fix is to add some code to the
fieldlines_write_parhdf5.f90code which creates a subcommunicator of just those processors which actually did work for the paralle HDF5 part. For the serial HDF5 part we probably just need to skip those threads whosemystartis greater thanmyend.