problems with running lammps on the multiprocessor cluster

Hello,
We have some problems with running lammps on the multiprocessor cluster.

After we installed the version of 28Feb12 we got two problems:
1) "read_data" command has stopped working. The log message is "ERROR: Incorrect atom format in data file (atom.cpp:833)"
2) if we use more than 9 nodes of 8 processors (>72processors) we get the "lost atoms" error.

We have no problems on desktop system with 4 processors and with the earlier versions of lammps (data file is the same).

Please suggest us a solution regarding this issue.
Thanks for any information.

Eugeny.

An earlier version of LAMMPS works with the same data? Maybe if you provide us the input file and the file with data of the atoms we could help, without these files we could just guess what is wrong…

Lost atoms error is usually associated with a bad dynamics of your system, on a serial run you will hardly obtain an error like this one because the atoms are always in the same processor (they are not communicated to other processors, and that is when you lost atoms, you atom could for instance walk a BIG distance in one step and fall out of your box…). Monitor the kinetic energy is a good start, but again, without the input files there is not so much thing that we can do to help you.

Hello,
We have some problems with running lammps on the
multiprocessor cluster.

After we installed the version of 28Feb12 we got two
problems:
1) "read_data" command has stopped working. The log
message is "ERROR: Incorrect atom format in data file
(atom.cpp:833)"

which means, that you may be affected by some changes
in the read_data command. search this link for remarks
about changes that are not fully backward compatible.

http://lammps.sandia.gov/bug.html

2) if we use more than 9 nodes of 8 processors
(>72processors) we get the "lost atoms" error.

We have no problems on desktop system with 4 processors
and with the earlier versions of lammps (data file is the
same).

Please suggest us a solution regarding this issue.
Thanks for any information.

please check your data file with the current documentation.
i'd also recommend to update to the current version, since
it includes even more bugfixes than the 28Feb version.

axel.