box/relax for nanolaminates

I am trying to obtain the relaxed structure of bimetallic nanolaminates, in this case Cu Ni. I use the fix box/relax aniso 0 in conjunction with cg minimzation. Before too long after running the minimization, the simulation box awkwardly starts expanding and the separation between atoms become too large. Even if vmax is made an order less, the same happens. That is obviously not physical but not sure why it is happening. (i am attaching the few lines of the input script below)

I would appreciate your advice
thanks
Firas

--------------- INITIALIZATION ------------------

clear

units metal

dimension 3

boundary p p p

atom_style atomic

lattice fcc 3.52

region upper block -19. 19. -19. 19. 0 7

lattice fcc 3.615

region lower block -18.5 18.5 -18.5 18.5 7 14

region whole union 2 upper lower

create_box 2 whole

lattice fcc 3.52 orient x 1 0 0 orient y 0 1 0 orient z 0 0 1

create_atoms 1 region upper

lattice fcc 3.615 origin 0 0 .247 orient x 1 0 0 orient y 0 1 0 orient z 0 0 1

create_atoms 2 region lower

pair_style eam/alloy

pair_coeff * * FeCuNi.eam.alloy Cu Ni

compute csym all centro/atom fcc

compute eng all pe/atom

compute eatoms all reduce sum c_eng

compute strs all stress/atom

reset_timestep 0

thermo 10

thermo_style custom step pe lx ly lz press pxx pyy pzz c_eatoms

dump 1 all custom 50 dump.comp.* id type xs ys zs c_csym c_eng c_strs[1] c_strs[2] c_strs[3]

fix 1 all box/relax aniso 0 vmax 0.001

min_style cg

minimize 1e-15 1e-15 10000 10000

Aidan may be able to comment.

Steve

Well, since the target pressure is zero, there is a physical solution at
infinite volume. It's just not the solution you care about, but the
minimizer does not know that.

There are lots of ways you can encourage the minimizer to converge to a
finite volume solution. For example, start out by setting the target
stress in fix box/relax to be the stress in the initial configuration, and
see if you can get that to relax. Or just relax the atom positions at
fixed volume. You can also use commands like change_box and fix deform to
manually search out low-stress box dimensions. In all of these cases, you
can then use fix box/relax to fully converge the stress to zero.

Aidan

Firas,
The minimizer could be converging to some sort of gas phase where the
atoms are very far away from one another and which will correspond to
a local energy minima. If you visualize your initial structure with
VMD, atomeye or any other code and measure the initial distances
between the atoms, how much different are they from the eq lengths of
your potential? Make sure that you are placing the atoms properly when
invoking the lattice and create_atoms commands.
In addition, think twice about the physical meaning of lines like the
one you have used below:

minimize 1e-15 1e-15 10000 10000

The energy and force tolerance values above are ridiculous, especially
for a test run. Usually it is better to use something like

minimize 0.0 \{force\_tol\} {Steps} ${Steps}

where the first 0.0 makes the minimizer to use force convergence
criteria instead of the energy one. Next, you set your force_tol
variable to a reasonable starting guess (not an ultra low value) so
that minimization doesn't take forever. Make also sure the output form
the minimizer shows that indeed it converged!
If needed you can always lower force_tol as you go to improve the
convergence of your system.
Carlos

Dear Steve, All

I would appreciate your advice on this issue with change_box command. My input file contains a change_box command and runs fine on my desktop PC which has a recent version (2012) of LAMMPS. When I transfer the same file to run it on an HPC cluster an error occurs like “illegal change_box command”. The LAMMPS version on the cluster dates back to 2010 and my guess is that the problem is version related. Since the cluster is at a different location and I have no control over it, I was wondering how can I find out what the proper syntax of the command is for that version. I looked at LAMMPS website but could not find the answer. I would deeply appreciate some direction.
ON a different note, is there any identity which allows general users who do not have local access to HPC to run LAMMPS. I am at a US university and have NSF grant (if that helps)

thank you in advance,
Regards
Firas

That’s an error that indicates the syntax of your change_box command
is not compatible with the current version. See the doc page
for the version you are running (in the doc dir). If you’re not
running the current version, please upgrade. If you’re still
confused after that, post the command (or entire script)
you are using.

Steve

That's an error that indicates the syntax of your change_box command
is not compatible with the current version. See the doc page
for the version you are running (in the doc dir). If you're not
running the current version, please upgrade. If you're still
confused after that, post the command (or entire script)
you are using.

Steve

Dear Steve, All

I would appreciate your advice on this issue with change_box command. My
input file contains a change_box command and runs fine on my desktop PC
which has a recent version (2012) of LAMMPS. When I transfer the same file
to run it on an HPC cluster an error occurs like "illegal change_box
command". The LAMMPS version on the cluster dates back to 2010 and my guess
is that the problem is version related. Since the cluster is at a different
location and I have no control over it, I was wondering how can I find out

LAMMPS is an entirely user-level program with a standalone executable
and thus can be compiled and installed by *any* user. many LAMMPS
users do this all the time at all kinds of locations.

what the proper syntax of the command is for that version. I looked at
LAMMPS website but could not find the answer. I would deeply appreciate some
direction.
ON a different note, is there any identity which allows general users who
do not have local access to HPC to run LAMMPS. I am at a US university and
have NSF grant (if that helps)

www.xsede.org.

Dear Steve, All

I would appreciate your advice on this issue with change_box command. My
input file contains a change_box command and runs fine on my desktop PC
which has a recent version (2012) of LAMMPS. When I transfer the same file

well, given the pace of development in LAMMPS, i would not call a
version from 2012 "recent".

I am bound to use LAMMPS (1 May 2010) because that is what the institution that kindly gave me free access to their cluster has. I do not have the least authority to ask them to upgrade. What I am looking for is to find out what is the proper syntax for change_box command in that version. Is there a repository of all LAMMPS manuals by version. That is what I looked for but could not find.
(as I mentioned before the command runs just find on LAMMPS 1July 2012)
I would deeply appreciate your advice.

Thanks
Firas

I am bound to use LAMMPS (1 May 2010) because that is what the institution
that kindly gave me free access to their cluster has. I do not have the
least authority to ask them to upgrade. What I am looking for is to find out

you are not paying attention to what i wrote. you do not have to ask
anybody. just download the source and compile it yourself. probably
the majority of LAMMPS users does that.

what is the proper syntax for change_box command in that version. Is there a
repository of all LAMMPS manuals by version. That is what I looked for but
could not find.
(as I mentioned before the command runs just find on LAMMPS 1July 2012)
I would deeply appreciate your advice.

again, you *do* have to pay attention or else people will in the
future disregard you. steve already told you that each LAMMPS version
ships with its own documentation that matches that specific version.
all you have to do is to look up *that* documentation and determined
the syntax.

but then again, the benefits of upgrading to the current version in my
personal opinion by far outweigh the effort required. there have been
a ton of improvements and lots of bugfixes. the average code quality
has improve a lot over the last couple of years due to several
concerted efforts to do more thorough testing and validation and
regularly building and packaging LAMMPS packages for RPM-based
distributions and windows.

axe;/

the version-specific documentation is in the doc dir
of the tarball that is downloaded.

Steve