neighbor command

LAMMPS LAMMPS Mailing List Mirror
1

Dear Lammps users

I am trying to understand how neighbor command works in my case. My modelling system is circular polymeric chain. Each monomers of the chain has bonds with neighbors, Bonds are determined by harmonic potential. All system exists in a huge box. I run this system in one proseccor.

As I understood the harmonic potential hasn’t cut off distance and cut off distance is equal to the skin option in the neigthbor command. Am I right? But I haven’t pair interaction in my programm only bonds but without neighbor command it doesn’t work. Initially I have used neighbor 2 bin command, but I recieved a mistake neighbor bins-box size<< cut off ->ERROR. I have changed it to the neighbor 2 nsq. It is working now. But what it is really happen?

As I understand the way how lammps works it needs neighbor list for long-range interaction (for instance LJ). But I don’t have this interaction in my system I have only bonds. Could you explain how it really works or give me a link to the Lammps documentation.

Kind regards!
Tatyana Pichugina

My input script

2

Dear Lammps users

dear tatyana,

I am trying to understand how neighbor command works in my case. My
modelling system is circular polymeric chain. Each monomers of the chain has
bonds with neighbors, Bonds are determined by harmonic potential. All system
exists in a huge box. I run this system in one proseccor.

As I understood the harmonic potential hasn't cut off distance and cut off
distance is equal to the skin option in the neigthbor command. Am I right?

no. harmonic bonds have no cutoff. period.

But I haven't pair interaction in my programm only bonds but without
neighbor command it doesn't work. Initially I have used neighbor 2 bin

that is a different story. the problem you are having is
only indirectly related to the neighbor command, the
real culprit is the lack of a non-bonded potential, or
rather the lack of a non-bonded cutoff associated with
it. this cutoff is instrumental in the generation of
neighbor lists, which includes the lists of bonds.

command, but I recieved a mistake neighbor bins-box size<< cut off ->ERROR.
I have changed it to the neighbor 2 nsq. It is working now. But what it is
really happen?

the sum of the "skin" value of 2.0 and the largest non-bonded
cutoff is usually the base length used for the construction of
cells that lead to the assembly of neighbor lists. the lack of
a pair style turns this into a problem, which you can circumvent
by making the "skin" value very large.

As I understand the way how lammps works it needs neighbor list for
long-range interaction (for instance LJ). But I don't have this interaction
in my system I have only bonds. Could you explain how it really works or

the fact, that you don't have any non-bonded interactions
makes your system _very_ unusual, and it doesn't match
the usual assumptions of lammps.

i don't think that there is any explanation in the lammps
documentation, but i faintly remember, that a similar
problem has occurred earlier and thus should be mentioned
in the mailing list archives. i suggest you search them
for an authoritative answer on what is the best way to
deal with this situation.

axel.

3

Axel's description is correct. Instead of
using the neighbor skin setting to induce
communication of ghost atoms from far enough
away to find the harmonic bond neighbors,
you can just use the communicate cutoff
command. Then you will not use/need neighbor
lists at all, since your script doesn't require
them.

Steve