help: communicate multi

LAMMPS LAMMPS Mailing List Mirror
1

My problem is the following. I would like to define 2 different bond types in
my system FENE and harmonic. The main problem is that type 1 bonded atoms
(FENE) are close to each other, whereas type 2 bonded atoms (harmonic) are very
far apart. So I would like to assign a different distance cutoff to each bond
type for inter-processor communication purposes, but I didn't succeed.

What I already tried is the following.
- I run on a single processor, but it isn't efficient.

- I also tried to use "communicate single cutoff 250.0". This choice is again
inefficient

- Then I tried to define 2 different atom types: both type atoms are involved
in bond type 1, but only type 2 atoms are involved in bond type 2. Then I
define two different groups:

group group_1 type 1
group group_2 type 2

So I would like to assign a different distance cutoff to each atom type for
inter-processor communication purposes. Let's say a low value (~2) for type 1
atoms and an high value (~250) for type 2 atoms. So I tried "communicate multi"
command but I was not able to succeed.

Does anybody know how to help me?

Thanks

Best Regards
Marco

2

So I would like to assign a different distance cutoff to each atom type for
inter-processor communication purposes.

Communicate multi will do this based on pairwise cutoffs. It's really
designed to work with non-bond interactions. But you could try
defining the pairwise 2-2 cutoff at 250 and see if it works. Unless
your box is huge, you may have other problems with bonding 2
atoms together that are 250 apart. LAMMPS may get confused
about which periodic image the atom is bonded to.

I would see if you can use fix spring between these far apart atoms.
Your model seems quite odd.

Steve