question about how to run a stimulation of TIP4P

Dear. Mr. Plimpton

Recently, my tutor and me tried using lammps to stimulate a NPT-ensemble of water by TIP4P model. Unfortunately a problem harassed us nearly a week.

Our scripts are sent with this email in the attachment. Everytime we use the command–“mpirun −np 12 lmp_g++ -echo both <in.main”–to execute the program, it will be terminated and following messages will be displayed

PPPM initialization …
G vector = 0.0308819
grid = 3 3 3
stencil order = 5
RMS precision = 2.45677e-06
WARNING: Reducing PPPM order b/c stencil extends beyond neighbor processor
G vector = 0.0290852
grid = 3 3 3
stencil order = 4
RMS precision = 7.2159e-06
WARNING: Reducing PPPM order b/c stencil extends beyond neighbor processor
G vector = 0.0274372
grid = 4 3 3
stencil order = 3
RMS precision = 1.83155e-05
WARNING: Reducing PPPM order b/c stencil extends beyond neighbor processor
G vector = 0.0252743
grid = 8 6 6
stencil order = 2
RMS precision = 5.72749e-05
brick FFT buffer size/proc = 180 32 216
Setting up run …
APPLICATION TERMINATED WITH THE EXIT STRING: Hangup (signal 1)

When we checked the log file, no any information were recorded. We did not catch up with the same problems in the past, stimulating ensemble like Ni, Argon and other matters. So that we hope to get some suggestion from you and express deep appreciation for any possible help.

Best regard!

data.tip4p (96.8 KB)

in.main (1.12 KB)

in.NPT (1016 Bytes)

in.TIP4P (822 Bytes)

2011/10/18 曲原 <[email protected]>:

From: 曲原 <[email protected]>
Date: 2011/10/14
Subject: question about how to run a stimulation of TIP4P
To: [email protected]

Dear. Mr. Plimpton

Recently, my tutor and me tried using lammps to stimulate a NPT-ensemble of
water by TIP4P model. Unfortunately a problem harassed us nearly a week.

Our scripts are sent with this email in the attachment. Everytime we use the
command--"mpirun -np 12 lmp_g++ -echo both <in.main"--to execute the
program, it will be terminated and following messages will be displayed

PPPM initialization ...
  G vector = 0.0308819
  grid = 3 3 3
  stencil order = 5
  RMS precision = 2.45677e-06
WARNING: Reducing PPPM order b/c stencil extends beyond neighbor processor
  G vector = 0.0290852
  grid = 3 3 3
  stencil order = 4
  RMS precision = 7.2159e-06
WARNING: Reducing PPPM order b/c stencil extends beyond neighbor processor
  G vector = 0.0274372
  grid = 4 3 3
  stencil order = 3
  RMS precision = 1.83155e-05
WARNING: Reducing PPPM order b/c stencil extends beyond neighbor processor
  G vector = 0.0252743
  grid = 8 6 6
  stencil order = 2
  RMS precision = 5.72749e-05
  brick FFT buffer size/proc = 180 32 216
Setting up run ...
APPLICATION TERMINATED WITH THE EXIT STRING: Hangup (signal 1)

When we checked the log file, no any information were recorded. We did not
catch up with the same problems in the past, stimulating ensemble like Ni,
Argon and other matters. So that we hope to get some suggestion from you and
express deep appreciation for any possible help.

you are using a 100 A coulomb cutoff.

that is ridiculously large and will cause abysmal performance.
as a consequence your PPPM grid is so small, that it cannot
be parallelized across the 12 CPU (core)s that you are using.
hence the refusal of LAMMPS to continue.

you should try with a coulomb cutoff that is identical
to the LJ cutoff and then adjust it so that you get the
best overall performance.

if you haven't done so, you should upgrade to the latest
version of LAMMPS, which has an optimized version
of the pair style that you are using.

axel.

oh, and your relaxations times for fix NPT are similarly crazy.
check out the documentation for what is reasonable. yours
are by many orders of magnitude too small.

axel.