Problem reading WFK
Posted: Thu Mar 30, 2017 11:39 am
Hi,
I am studying a big system and, after more than 300 steps, I decided to stop the calculation (putting the abinit.exit file in the work directory) because we have limited time in our cluster.
Convergence was 5.0E-06, still far from the desired value (1.0E-8), so I restarted the job with the WFK file (writing in the in file: irdwfk 1) but I obtained an error:
--- !ERROR
src_file: m_wffile.F90
src_line: 2180
mpi_rank: 0
message: |
Error when reading record markers of file acasurface_i_WFK
...
I do not know if the problem is due to the fact that I am running this job in parallel (120 cpus, 4 nodes) or if it is related with the band occupation (since is a metal surface, the occupation of the input WFK in some bands is not 0 or 2)... I only know that, when I tested the "irdwfk 1" in an isolated molecule (1 node, band occupation 0 or 2), it worked.
Any advice?
Thank you!
Fernando
I am studying a big system and, after more than 300 steps, I decided to stop the calculation (putting the abinit.exit file in the work directory) because we have limited time in our cluster.
Convergence was 5.0E-06, still far from the desired value (1.0E-8), so I restarted the job with the WFK file (writing in the in file: irdwfk 1) but I obtained an error:
--- !ERROR
src_file: m_wffile.F90
src_line: 2180
mpi_rank: 0
message: |
Error when reading record markers of file acasurface_i_WFK
...
I do not know if the problem is due to the fact that I am running this job in parallel (120 cpus, 4 nodes) or if it is related with the band occupation (since is a metal surface, the occupation of the input WFK in some bands is not 0 or 2)... I only know that, when I tested the "irdwfk 1" in an isolated molecule (1 node, band occupation 0 or 2), it worked.
Any advice?
Thank you!
Fernando