Questions about the bond atom missing when I stretch the model

LAMMPS LAMMPS Mailing List Mirror
1

Dear Lammps user,

Recently I am doing a simulation about stretching a polymer bar. The potential is LJ and FENE bond. However, when the model is stretched too long, there is an error called “bond atom missing.” I think it is caused by the long distance between these two atoms.

Has anyone met such a problem? How can we solve it? Does it work that I use command “fix bond/break?”

Best
Chi

2

Dear Lammps user,

Recently I am doing a simulation about stretching a polymer bar. The
potential is LJ and FENE bond. However, when the model is stretched too
long, there is an error called "bond atom missing." I think it is caused by
the long distance between these two atoms.

Has anyone met such a problem? How can we solve it? Does it work that I use
command "fix bond/break?"

LAMMPS uses domain decomposition with ghost atoms for parallelization
and periodic replication. the distance as to how far away from a
sub-domain the coordinates of these ghost atoms are created depends on
the "communication" cutoff. atoms in a bonded structure have to be
within this and also atoms within the cutoff (plus skin distance).
however, if you stretch a bond that is near the domain boundary beyond
this cutoff, you will "lose" the ghost atom.

the question is, how realistic is your model? do you really *need*
those long bonds? fix bond/break would be a change in the model, but
that may be what you desire. otherwise you can work around the losing
of atoms by increasing the communication cutoff with the comm_modify
command. of course that will slow down your calculation the longer the
cutoff. if you have a very limited number of such long bonds, you may
have a look at the pair style list, which can compute forces between
arbitrary pairs of atoms in the system, but has other drawbacks... so
in the end, you will have to make a choice what is going to represent
the correct physics of your model and then choose the right method to
implement this.

axel

3

Thanks Axel. I will try them and appreciate your advice.

Best
Chi

4

You can also use comm_modify cutoff to set a communication

distance further than the neighbor cutoff. So that the code

can still “find” far-away atoms in a bond. However, as Axel said,

the real problem is likely that you have a model giving bad dynamics.

Steve

5

Hi Steve

Yes, it is. Hence I plan to use command “fix bond/break” to delete the extreme long bonds and I am testing it.

Best
Chi

6

Thank you Steve.

Best
Chi