**problem with calculation of forces** |
- Date: 2014/03/28 14:57
- Name:
**Andrzej Lusakowski**
- Dear Professor Ozaki,
I use OpenMX since many years. Recently I started to make calculations on larger clusters (60 or more CPUs). I meet following problem. Everything goes very well up to the moment of force calculations. The program practically stops and for the next 15 - 30 minutes I obtain only the following statements: <MD=23> Force calculation Failure of saving ./gemnte_rst/gemnte.crst_check Failure of saving ./gemnte_rst/gemnte.crst_check Failure of saving ./gemnte_rst/gemnte.crst_check Failure of saving ./gemnte_rst/gemnte.crst_check Failure of saving ./gemnte_rst/gemnte.crst_check Failure of saving ./gemnte_rst/gemnte.crst_check Failure of saving ./gemnte_rst/gemnte.crst_check Failure of saving ./gemnte_rst/gemnte.crst_check Failure of saving ./gemnte_rst/gemnte.crst_check Failure of saving ./gemnte_rst/gemnte.crst_check
The next line Failure of saving ./gemnte_rst/gemnte.crst_check appears after a few minutes and so on.
At a certain moment the program "overcomes" difficulties with saving and goes to the next iteration of geometry optimization. The same problem I have with calculations without geometry optimization, but in this case the this problem appears, of course, only once. Unfortunately the cluster does not belong to me (I have no superuser privileges) and I cannot even say how the makefile looks like. The problem appears only when I calculate larger systems (64 atoms or more), for two atom bulk system I have not noticed anything wrong. Have you met with such a problem and have you any suggestions? With best regards Andrzej Lusakowski P. S. Trying to understand what is going on I have noticed that such communicates may appear due to reading or writing *_rst/*_check (RestartFileDFT.c and Input_std.c), but this does not help me. The file ./gemnte_rst/gemnte.crst_check always exists after calculations. Best regards Andrzej Lusakowski
| |