Hi,
I run a very simple code in lammps on a single processor ( there are tqo run commands used in it), every thing goes well to “All Done!”. But as I run the same code on multiprocessors, the first run command in the code is completed successfully, but as it starts setting up second run, it multiplies the number of atoms by the number of processors and give me lammps Error: Lost atoms.
I can’t figure it out how to avoid this. I have run many other lammps code and this repeatedly occurs i.e. once the second run is setting up, it multiplies atoms by # processors and stops simulation.
Please tell me how I can avoid this problem and make the simulation (with several runs) complete successfully.
The code and the log file are attached.
BTW, I used “mpirun -np 4 lmp<model3.txt”
Thanks,
Masoumeh
log.txt (2.97 KB)
Al99.eam.alloy (791 KB)
model3.txt (1.08 KB)
Hi,
I run a very simple code in lammps on a single processor ( there are tqo run
commands used in it), every thing goes well to "All Done!". But as I run the
same code on multiprocessors, the first run command in the code is completed
successfully, but as it starts setting up second run, it multiplies the
number of atoms by the number of processors and give me lammps Error: Lost
atoms.
I can't figure it out how to avoid this. I have run many other lammps code
and this repeatedly occurs i.e. once the second run is setting up, it
multiplies atoms by # processors and stops simulation.
Please tell me how I can avoid this problem and make the simulation (with
several runs) complete successfully.
try a current version of the code. it works for me.
in general, you should *always* verify with the current version of the
code, if a problem persists. nobody wants to fix bugs, that have
already been fixed.
thanks,
axel.