Hi Lammps users:
I have a restart file on one machine and want to restart the calculation on another cluster. I used the restart2data tool to transfer the restart file to a data file. The problem is that when I launched a NPT MD run with this data file, the pressure was not computed correctly. The initial pressure was very high thus the system got blew away in just a few steps. But when I restarted the MD run using the restart file on the same machine on which the file was generated, there was no such problem. What might be the reason for this problem? Thanks.
Shaorui
I'm guessing the issue is not related to the 2 machines. Try
this test on the same machine:
a) convert the restart file to a data file and do
a 2nd simulation reading the data file
b) run a 2nd simulation using the restart file
directly via read_restart
If (a) blows up and (b) does not, then it means you are not specifying
something in the input script that is needed to
restart correctly from the reduced information
in a data file. E.g. a fix has not been respecified
correctly.
Steve
Hmm. If you still are having trouble, try this:
Make a copy of your old data file, and use a text editor to open it
and replace the "Atoms" and "Velocities" sections with the new
coordinates from the data file created by restart2data. That should
be safe.
restart2data scrambled the angle coeffs in my water system. I
don't trust it. (My apologies for being too lazy to post a bug
report. I can still dig out these input files.)
Andrew
Hmm. If you still are having trouble, try this:
Make a copy of your old data file, and use a text editor to open it
and replace the "Atoms" and "Velocities" sections with the new
coordinates from the data file created by restart2data. That should
be safe.
This may be self evident, but if you are running an NPT simulation,
then you also have to copy the boundary box dimensions "LAMMPS
Description" section from the top of the new data file. I forgot
that.
Hi Steve:
I did the test you suggested. Problem exists using the same machine. But I don’t think the reason is that something is not specified correctly in the new input script. I used exactly the same fix. Besides, it is at STEP 0 that big difference of pressure is observed. So guess it’s actually not related any fix. The new data file may contain some mistakes.
Shaorui