tip4p

LAMMPS LAMMPS Mailing List Mirror
1

Hi LAMMPS users,

I run the two systems with same data file, paramter , input file, with only one difference of pair_style,

  1. pair_style hybrid lj/cut/coul/long 12.0 lj/cut/coul/long/tip4p 1 2 1 1 0.1577 12.0

  2. pair_style lj/cut/coul/long/tip4p 1 2 1 1 0.1577 12.0

With system 1, the volume converged to lesser value, leading to more in density.
but system 2, volume converged to a exact value giving correct density.

can anyone try to explain me , where I went wrong.

thanks in advance

karthik kumar

2

If you are running NPT and the volumes end up different
then the pressures should be different. If you run 0
steps of a system with the 2 pair commands below,
do you get different pressures or different
thermo output?

Steve

3

Hi steve sir,

at each step, I can see the difference in pressure, but I am looking at the convergence of the volume ( which in turn relates the density of the system).

with same force field, same datafile & input file and same parameter, except for hybrid and non-hybrid style of tip4p
Density should be same right, how can the averaged volume be different in two pair_styles…

As Axel sir previously mentioned both the hybrid and non hybrid will do the same, except for computational cost.

thanks,

karthik kumar

4

Hi steve sir,

at each step, I can see the difference in pressure, but I am looking at
the convergence of the volume ( which in turn relates the density of the
system).

what do you supposed is responsible for
the volume to change, if not the pressure?

with same force field, same datafile & input file and same parameter,
except for hybrid and non-hybrid style of tip4p
Density should be same right, how can the averaged volume be different in
two pair_styles..

you can do averaging only after your system
has reached equilibrium. the last input you
posted has been *very* far from equilibrium.

it is near impossible to discuss such matters
on this general level.

if you believe that LAMMPS is at fault,
please post a minimal test input, that
allows to easily and quickly reproduce
and investigate it.

As Axel sir previously mentioned both the hybrid and non hybrid will do the
same, except for computational cost.

as steve said, they should give the same result
at step 0. but since you sum up forces differently,
trajectories will eventually diverge (same as when
using a different number of processors) due to
floating point truncation errors (floating point
math does not commute, so the result can depend
on the order of operations due to the limited accuracy)

axel

5

Please post a simple example where you run for 0 steps and
get a different pressure with the 2 forms of the pair style.

Steve