Lammps - angle missing problem

LAMMPS LAMMPS Mailing List Mirror
1

Dear lammps users,

I am equilibrating a system with 400 molecules of an ionic fluid, starting from a low-density lattice.

I pre-equilibrate the system at 0K and then I raise the temperature in two stages, from 0 to 250 K, and then from 250K to 350K, with a very small time step. I am deforming the box to a target volume at the same time temperature goes from 250K to 350K. Everything seems ok in VMD.

Problem:

If I run it in 1 proc., it works, if I run it in N processors it crashes after more than 50000 steps saying the following:

ERROR on proc N: Angle atoms XXXX XXXX XXXX missing on proc. N at step 55307 (neigh_bond.cpp:176)

This seems to me as a problem of neighbour lists and for that I am rebuilding neighbour lists very often. My cut-off is 12 A for a box length of

Lx: -20.25 20.25

Ly: -20.25 20.25

Lz: -41.5 41.5

Herein I print you my input file:

#INITIALIZATION

units metal

atom_style full

newton off

dimension 3

boundary p p p

pair_style lj/cut/coul/cut 14.0

bond_style harmonic

angle_style harmonic

dihedral_style opls

read_data data.IL

kspace_style none

special_bonds lj 0.0 0.0 0.5

special_bonds coul 0.0 0.0 0.5

time step and run_style

timestep 0.00001

run_style verlet

Rebuild neighbor list

neighbor 2.0 bin

neigh_modify every 1 delay 0 check yes page 1000000 one 10000

ensamble

fix microcanonic all nve

thermostat

fix rescaletemp all temp/rescale 1 1.0 1.0 1.00 0.90

shake

fix cluster ct-hc shake 0.0001 20 1000 b 3

run 2000

unfix rescaletemp

timestep 0.00001

fix rescaletemp all temp/rescale 1 1.0 250.0 1.00 0.90

run 50000

timestep 0.00005

unfix rescaletemp

fix rescaletemp all temp/rescale 1 250.0 350.0 1.00 0.90

#Deform box

fix lowV all deform 500 x final -17.35 17.35 y final -17.35 17.35 z final -35.6 35.6 units box

run 50000

timestep 0.00005

unfix rescaletemp

fix rescaletemp all temp/rescale 1 350.0 350.0 1.00 0.90

run 50000

clear

Can you please help me? Thanks a lot!

Ana Catarina Mendonça

PhD Student

2

Dear lammps users,

I am equilibrating a system with 400 molecules of an ionic fluid, starting
from a low-density lattice.

I pre-equilibrate the system at 0K and then I raise the temperature in two
stages, from 0 to 250 K, and then from 250K to 350K, with a very small time
step. I am deforming the box to a target volume at the same time temperature
goes from 250K to 350K. Everything seems ok in VMD.

Problem:

If I run it in 1 proc., it works, if I run it in N processors it crashes
after more than 50000 steps saying the following:

ERROR on proc N: Angle atoms XXXX XXXX XXXX missing on proc. N at step 55307
(neigh_bond.cpp:176)

This seems to me as a problem of neighbour lists and for that I am
rebuilding neighbour lists very often. My cut-off is 12 A for a box length
of

no. this is usually a consequence of bad dynamics.
e.g. when inserting too much kinetic energy in a
meta stable system with high potential energy.
you need to quench your system (i.e. run a minimiation)
for a bit or and/use an integrator like fix nve/limit to remove
that excess potential energy.

...and *please* don't call temp/rescale a thermostat.
it is a brutal (and not very helpful in case of "hot spots")
method to inserting/removing kinetic energy into a system.
use something like fix langevin instead and later switch
to fix nvt. you'll do yourself a big favor.

axel.