[lammps-users] How to create a 'Huge' box

LAMMPS LAMMPS Mailing List Mirror
1

By huge , I mean, for example(using following commands),

units metal
lattice fcc 60
region box block 0 1000 0 1000 0 2000
create_box 1 box

But it seems that the code has a limit in setting this. When I submit the input script, it hangs there without any output.
I checked the code, but did not find out where is the limit indicated.Thanks.

2

This code works fine for me, using LAMMPS from 04/17/09:

lutz:~ cat test\.in units metal lattice fcc 60 region box block 0 1000 0 1000 0 2000 create\_box 1 box lutz:\~ lmp -in test.in
LAMMPS (9 Jan 2009)
Lattice spacing in x,y,z = 60 60 60
Created orthogonal box = (0 0 0) to (60000 60000 120000)
  1 by 1 by 1 processor grid
lutz:~ $

What version are you using?

  Lutz

3

Hi, Lutz,thanks.

I am using the latest version (May,2009).

Yes, the code is hanging there with the information of :

LAMMPS (9 Jan 2009)
Lattice spacing in x,y,z = 60 60 60
Created orthogonal box = (0 0 0) to (60000 60000 120000)
1 by 1 by 1 processor grid
…

4

Maybe I am misunderstanding, but what makes you think that LAMMPS is
getting stuck on the create_box command? It seems to complete it just
fine. Try this:

units metal
lattice fcc 60
region box block 0 1000 0 1000 0 2000
create_box 1 box
print "Finished."

Does LAMMPS reach the print statement in your version?

Hope this helps,

  Lutz

5

Your output indicates LAMMPS created the box just fine.
It's the next operation(s) that are hanging. If that includes
icreating atoms, as would be the usual case, then
you are creating 8 billion atoms on a single processsor
so it might take a while. And a bit of memory.

Steve

6

Thanks, Steve

My full commands are as follows:

units metal
boundary f f f
atom_style atomic
lattice fcc 60
region box block 0 1000 0 1000 0 2000
create_box 1 box
region particle sphere 30000 30000 30000 200 units box
create_atoms 1 region particle
mass 1 60221.0
pair_style lj/cut 111.60725
pair_coeff * * 42.7103 35.6359
thermo 100
timestep 0.004
run 0

7

ok - this makes more sense - I think
what's happening is that create_atoms
loops over all 8 billion (4x1Kx1Kx2K) lattice
sites in the big box, checking each to
see if it's in the region - I'm not looking
at the source code at the moment.

So that's very inefficient for your problem.
Instead I suggest this: create the atoms in
a small box first. The use displace_box
to change the box size to big, using the
remap none option so the atoms don't move.

BTW, you can always run LAMMPS with
the -echo screen option to see commands
processed one at a time and figure out
which one is giving you problems.

Steve