[lammps-users] How to handle over 100,000 atoms?

I tried to do energy minimization with the nanowires which have over 100,000 atoms, but I got some error messages :

ERROR: Cannot use neighbor bins - box size << cutoff

So I changed neighbor cutoff style and values (bin -> nsq) and tried again but I got error message :

ERROR: Neighbor list overflow, boost neigh_modify one or page

I changed values of one and page, but I couldn’t get proper results.
How can I simulate the nanowires with over 100,000 atoms?
(I got proper results from the nanowires with under 100,000 atoms.)

Best regards.

LAMMPS (7 Jul 2009)
Reading data file …
orthogonal box = (0 -500 -500) to (28.9207 500 500)
1 by 1 by 1 processor grid
162010 atoms
Lattice spacing in x,y,z = 5.78413 5.78413 4.09
Setting up run …
Memory usage per processor = 607.734 Mbytes
Step Lx StressAl StressAl StressAl Temp TotEng PotEng KinEng Press Volume
0 28.9207 nan nan nan 0 nan nan 0 nan 4836719.2
ERROR: Cannot use neighbor bins - box size << cutoff

I tried to do energy minimization with the nanowires which have over
100,000 atoms, but I got some error messages :

100,000 atoms is "nothing" for LAMMPS. it is designed
to handle much larger systems.

ERROR: Cannot use neighbor bins - box size << cutoff

So I changed neighbor cutoff style and values (bin -> nsq) and tried
again but I got error message :

i would rather check your cutoff. how large is it?
LAMMPS assumes that you use a reasonable value. if
you don't the warnings you see may be misleading.

ERROR: Neighbor list overflow, boost neigh_modify one or page

I changed values of one and page, but I couldn't get proper results.
How can I simulate the nanowires with over 100,000 atoms?
(I got proper results from the nanowires with under 100,000 atoms.)

hard to tell without seeing your input. in general, please keep
in mind, that it is very difficult to give good advice without
knowing important details. if you check through the mailing list
archives, you'll see that people who provide (simple or smallest
possible) inputs to recreate the error or potential problem, get
helped the fastest and best. for the rest it is frequently a lot
of poking in the dark until somebody guesses right.

cheers,
   axel.

The default limit for neighbors/atom is 2000 (see the neigh_modify doc page).
If you have a very long cutoff, then you probably have a bogus model,
but if you want more than that, just use neigh_modify to reset the defaults.

Steve

2009/7/22 YooJun Hwan <[email protected]>: