[lammps-users] parallel simulation problem

Dear all,

I have a problem with my parallel lammps. Everything looked good in the beginning -- it
compiled, and my small example worked on 6 processors without problems.
However, when I tried to simulate ~26000 atoms in parallel, I got the
following message:

Deleting bonds ...
ERROR on proc 0: Bond atom missing in delete_bonds
application called MPI_Abort(MPI_COMM_WORLD, 1) - process 0
ERROR on proc 1: Bond atom missing in delete_bonds
application called MPI_Abort(MPI_COMM_WORLD, 1) - process 1
rank 0 in job 9 s0793217-VPCF12Z1E_43784 caused collective abort of
all ranks
   exit status of rank 0: return code 1

Again, when I run this in parallel mode on 1 processor it works. Could
anybody possibly help me with th is? Thank You very much in advance!

Sincerely Yours,
Anna

dear anna,

this looks like a bug in lammps. can you produce some
simple input that reproduces it.

i am wondering, why you insist on doing this bond deletion
within LAMMPS, it would be _so_ much easier to do this
outside of LAMMPS when you build your input.

that is probably the reason, why the code is failing
on you (as it has not been used a lot).

cheers,
    axel.

Dear Dr Axel,

Thank You very much for Your reply! Unfortunately, because of the initial configuration that I have (20 chains, each of which forms a specific structure and these are separate from each other), LAMMPS does not allow me not to delete bonds -- everything flies apart if I do not do so! ( I have a chain of 2000 atoms, and by translation, this configuration is repeated 20 times, which means that the last atom on one chain and the next atom of the next chain can be far away from each other...). Perhaps I should try to think of another way to solve this...

I will look at this carefully and will try to reproduce this behaviour for a simple output soon!

Thank You very much again!

Sincerely Yours,
Anna

No - this error message means LAMMPS could not find
the 2nd atom in the bond - likely b/c it is outside the
cutoff. Did you define a cutoff and is it long enough
to encompass any bond in the system?

Steve