Differences in evolution of simulation using Group and not using Group Command

LAMMPS LAMMPS Mailing List Mirror
1

Dear LAMMPS users,

I post to the LAMMPS forum in order to ask you a couple of things reffering to the group command.

I want to simulate a system consisted of 49 identical molecules. What I was doing till now

is that I was grouping all the atoms of the system with the group command as follows:

group mobile union all.

After that in the fix command where I give the simulation details (such are temperature, pressure etc) i write:

fix 1 mobile npt temp 1000.00 1000.00 10.0 iso 1.000000 1.000000 350.0

and I execute the simulation for 10000 steps (1fs timestep) writing at the log file every 100 steps.

In an effort to speed up my simulation, and taking into account that I don’t have different types of molecules so as to

use the group command, (thinking that this might accelerate the simulation) I deleted the group commad and in the fix command I made the following change:

fix 1 all npt temp 1000.00 1000.00 10.0 iso 1.000000 1.000000 350.0

What I observed is that the simulation does not get faster but the evolution of the simulation is not EXACTLY THE SAME with the case that

I used the group command.

Actually at the first ~400fs of both simulations the Energies are IDENTICAL but as the time evolves there is significant difference between both cases.

In order this to be more clear to you I attach both log files (log_group.lammps and log_no_group.lammps) and .in file in order you to have a better view of the situation.

Is there any explanation why the evolution of both simulations is not the same?

Thank you in advance for your help and time.

Best,

Emmanuel N. Skountzos.

log_group.lammps (72.7 KB)

log_no_group.lammps (72.7 KB)

PT_50m_49ch.in (1.23 KB)

2

Dear LAMMPS users,

I post to the LAMMPS forum in order to ask you a couple of things reffering
to the group command.

I want to simulate a system consisted of 49 identical molecules. What I was
doing till now

is that I was grouping all the atoms of the system with the group command as
follows:

group mobile union all.

After that in the fix command where I give the simulation details (such are
temperature, pressure etc) i write:

fix 1 mobile npt temp 1000.00 1000.00 10.0 iso 1.000000 1.000000
350.0

and I execute the simulation for 10000 steps (1fs timestep) writing at the
log file every 100 steps.

In an effort to speed up my simulation, and taking into account that I don't
have different types of molecules so as to

use the group command, (thinking that this might accelerate the simulation)

no. it doesn't. groups are implemented into LAMMPS in a way that after
atoms have been assigned to a group, there is no additional cost.

I deleted the group commad and in the fix command I made the following
change:

fix 1 all npt temp 1000.00 1000.00 10.0 iso 1.000000 1.000000
350.0

What I observed is that the simulation does not get faster but the evolution
of the simulation is not EXACTLY THE SAME with the case that

I used the group command.

Actually at the first ~400fs of both simulations the Energies are IDENTICAL
but as the time evolves there is significant difference between both cases.

In order this to be more clear to you I attach both log files
(log_group.lammps and log_no_group.lammps) and .in file in order you to have
a better view of the situation.

Is there any explanation why the evolution of both simulations is not the
same?

yes, it only has been explained on this list about two hundred times
and also, there is an explanation in the manual.

axel.