If RH is interrupted (ie, system problem), the convergence matrix in output_indata.hdf5 (mpi.convergence) has a fill value of 9.96921e+36 for columns that were never run. In a re-run, these should be run anew, but after the change to HDF5 this no longer works. This PR should fix this bug.
If RH is interrupted (ie, system problem), the convergence matrix in
output_indata.hdf5
(mpi.convergence
) has a fill value of 9.96921e+36 for columns that were never run. In a re-run, these should be run anew, but after the change to HDF5 this no longer works. This PR should fix this bug.