[lammps-users] A problem when using mpirun to run python script of lammps

LAMMPS LAMMPS Mailing List Mirror
1

Dear all:
I want to loop the lammps program in a python script.
…
from lammps import lammps
for i in range(10):
lmp = lammps()
lmp.file(infile)
lmp.close()

…
When I use serial syntax, the loop can be executed. However, an error occurred when using the parallel syntax. The lammps program cannot be executed completely in the second loop.
…
Reading data file …
orthogonal box = (-16 -16 -16) to (16 16 16)
1 by 2 by 2 MPI processor grid
WARNING: Angle style in data file differs from currently defined angle style (…/read_data.cpp:559)
reading atoms …
1421 atoms
reading velocities …
1421 velocities
scanning bonds …
6 = max bonds/atom
scanning angles …
5 = max angles/atom
scanning dihedrals …
6 = max dihedrals/atom
reading bonds …
2077 bonds
reading angles …
1311 angles
reading dihedrals …
1545 dihedrals
Finding 1-2 1-3 1-4 neighbors …
special bond factors lj: 0 0 0
special bond factors coul: 0 0 0
7 = max # of 1-2 neighbors
34 = max # of 1-3 neighbors
203 = max # of 1-4 neighbors
40 = max # of special neighbors
Reading data file …
orthogonal box = (-16 -16 -16) to (16 16 16)
1 by 2 by 2 MPI processor grid
reading atoms …
5473 atoms
Finding 1-2 1-3 1-4 neighbors …
special bond factors lj: 0 0 0
special bond factors coul: 0 0 0
7 = max # of 1-2 neighbors
34 = max # of 1-3 neighbors
203 = max # of 1-4 neighbors
40 = max # of special neighbors
480 atoms in group rbc
1421 atoms in group cell
5473 atoms in group pipette
0 atoms in group water

…
The next message should be :Neighbor list information, But it has been stopped here and no longer continues to run. What could be the cause of this?

Shuai Zhang
Nankai University

2

there is no need to create a new LAMMPS instance for every iteration. what happens if you do this:

from lammps import lammps
lmp = lammps()
for i in range(10):
lmp.command(‘clear’)
lmp.file(infile)

lmp.close()

Axel

3

Thanks for your immediate answer.

After adopting your program, the result is the same as before. Only when np = 1 can be traversed completely.
In other cases, it is still stuck in the same position during the second execution.

Axel Kohlmeyer <[email protected]> 于2021年4月28日周三 下午8:23写道：

4

I cannot reproduce that behavior with the latest LAMMPS (patch) release version (8 April 2021).

i can run the following simplified script in the examples/peptide folder with mpirun -np 4 python test.py to completion. So my conclusion is that you are either using a LAMMPS version with a bug or - more likely - that there is something in your input that produces an error when run a second time in the same location.

Axel.

infile = ‘in.peptide’

from lammps import lammps

lmp = lammps()
for i in range(2):
lmp.command(‘clear’)
lmp.file(infile)
lmp.close()

for i in range(2):
lmp = lammps()
lmp.file(infile)
lmp.close()