lammps-users Digest, Vol 88, Issue 54

Dear all,

Firstly, many thanks to Oscar G. and Adian.

I’m looking for a way to join some atomic layers, say for example 3 or 4 fcc atomic layers, as a single plane. I would like lammps keep this plane unchanged until the end of simulation.
I should add that I have defined them firstly as a specific region and then as a group but still they are deformed and I can’t prevent the deformation of these layers. Does anyone know a command to define some atomic layers as a plane?

Another question would be ‘is there any command in lammps to encode atoms?’ I mean, to define specific atoms with color coding or indices?

Thank you all in advanced.



Have you tried „fix rigid“

color coding:

this is a visualization problem and therefore not really a lammps problem. There are several ways. Read the documentation:

  • You could use atom type molecule and give each group a specific mol number

  • You can dump cfg and use “dump_modify 1 element C…” Here you can give an Element for each atom type

  • You can dump each group in a separate file and load them all in e.g. vmd

I am pretty new to lammps too, so this list may not be complete or the best way. But these were the first ideas that came to my mind. I personally would go for the second or third method.



Or just don’t time integrate the atoms in the 4-layer




please don't quote entire digests, but edit it down to the relevant parts.

this just clogs up everybody's inputs, particularly those that read
the digested version, since sending out a digest is not just dependent
on the time since the last, but also on the amount of content that has
accumulated (and that includes all quoted text).

also, keep in mind that there are people reading e-mails on mobile devices.


Dear Wolfgang,

Thank you for your response.

Actually I have tried ‘fix rigid command’ but I don’t know why when stretching the component the rigid body doesn’t move, which cause bursting in stress inside the component but not the strain. Do you have any idea about the reason?

About coding atoms, I search for a way to reach to any layers of atoms.

By the way, Dear Prof. Steve Plimpton thank you for your response and what do you mean by “don’t time integrate the atoms in the 4-layer group”? Can you please say with what command?

Thanks for your time.

You have e.g. fix npt which follows “fix ID group-ID style_name keyword value”. you can include in the group-id all atoms except those you don’t want to move.

Actually, I want to move one group of atoms at one end while the other one is fix. Because this is a stretching test. Here not deforming the fix layer and the moving one at the two ends is desirable. You can imagine that one side of a baulk element is fixed to a point and we stretch the element from the other side. Here, we don’t expect atoms in any side change their place and deform.

I also tried what you mentioned but still there is no change in the result.

Could fix_deform be suitable for you?

Could fix_deform be suitable for you?

LAMMPS Molecular Dynamics Simulator

fix move is likely the better choice. please see the description of a
simulation protocol that i just gave in a different discussion thread.
fix move just moves atoms according to the prescribed parameters and
does not care in any way about the forces on them.

fix deform only changes the box unless you tell it to update the
coordinates accordingly, which makes the most sense when applied to a
bulk system.


Dear Wolfgang,

I have used ‘fix deform erate … remap x’ with exempting atoms at two ends but there was no change in the result.



Dear Axel,

Thanks really much for your suggestion.

I used ‘fix move’ command. Now the stress figures grow to negative large numbers!!

Why is it so?

Thanks in advanced.


Dear Axel,

Thanks really much for your suggestion.
I used 'fix move' command. Now the stress figures grow to negative large
Why is it so?

because that is what LAMMPS computes based on the data you feed it.


Dear Steve,

Thanks so much for your offer. It was a great choice.

I ran the program with exempting one (or more) atomic layer at two ends from time integration. The result was good, the two ends were fixed and didn’t deform. Just a little problem arises: the atomic layers distance from the body of the wire. This seems natural but what should I do to figure out this problem?

Thanks indeed in advanced.

Best regards,


What you are asking is not clear to me.


Dear Steve,

I put the part of the program related to the applied ensembles and the regions here.

region left_side block -10 -9 -5 5 -5 5
region right_side block 9 10 -5 5 -5 5
region middle block -9 9 -5 5 -5 5
group edge1 region left_side
group edge2 region right_side
group body region middle

fix 1 all npt temp 500.0 500.0 0.1 x 0 0 0.05
run 15000

fix 2 body nve
run 5000
unfix 1

reset_timestep 0

fix 4 all deform 1 x erate 0.05

stretching computation commands

Thanks indeed in advanced.

Best regards,


I wonder how many times people like you who go back to Steve every time their simulation is not working add him to the acknowledgements of their thesis. Steve must be a guy with infinite patience.


You have 2 integration fixes (npt, nve) both integrating
some of the same atoms, which is bad, Did LAMMPS

not give you a warning about this?