Bond atom missing at large deformation when I stretched the polymer model

LAMMPS LAMMPS Mailing List Mirror
1

Dear Lammps user,

Recently, I built a coarse-grained polymer model using harmonic potential and L-J potential. After equilibrium, I stretched the model by fixing one end and assigning a velocity at the other end. At the beginning, it ran well. However, when the deformation was large (the strain is 651%), there was an error-

“ERROR on proc 16: Bond atoms 79506 79507 missing on proc 16 at step 2924955 (…/neigh_bond.cpp:65).”

Does anyone have the same problem? How should I solve it? The version of my lammps is 10Feb15. Sincerely thank you for your reading.

Best
Chi

2

Dear Lammps user,

Recently, I built a coarse-grained polymer model using harmonic potential
and L-J potential. After equilibrium, I stretched the model by fixing one
end and assigning a velocity at the other end. At the beginning, it ran
well. However, when the deformation was large (the strain is 651%), there
was an error-

"ERROR on proc 16: Bond atoms 79506 79507 missing on proc 16 at step 2924955
(../neigh_bond.cpp:65)."

Does anyone have the same problem? How should I solve it? The version of my
lammps is 10Feb15. Sincerely thank you for your reading.

there are two factors to consider. when increasing velocities, you
usually need to decrease the time step to be able to accurately
integrate the equations of motion. besides that, the specific error
can also be made less likely by increasing the communication cutoff.
depending on your system and setup, you may need to use a combination
of both factors to have a stable and reasonably accurate simulation.

axel.

3

Ok. I will try these methods. Thank you axel!

Best
Chi

4

there are two factors to consider. when increasing velocities, you
usually need to decrease the time step to be able to accurately
integrate the equations of motion. besides that, the specific error
can also be made less likely by increasing the communication cutoff.
depending on your system and setup, you may need to use a combination
of both factors to have a stable and reasonably accurate simulation.

In addition to Axel's answer, I'll forward Steve's general comment:

"As the error section of the manual explains, this is usually
due to bad dynamics, e.g. blowing atoms out of the box,
so that a bond cannot find its partner atom. Fix your simulation
and these kinds of problems go away."
http://lammps.sandia.gov/threads/msg20249.html
http://lammps.sandia.gov/threads/msg22394.html
http://lammps.sandia.gov/doc/Section_errors.html

  For other people who run into this problem and are searching the
mailing list archives, these kinds of errors often occur early in a
simulation (due to overlapping atoms, singular dihedral angles,
etc.... For people having this problem, try visualizing the initial
configuration of your system. See:
http://lammps.sandia.gov/threads/msg47488.html)

   However that was not happening in your case. But it makes sense
that these kinds of errors crop up under harsher conditions (high
strain, high temperature, high pressure, rapid driving forces,
etc...). Axel's suggestions are the first thing to try.

  I haven't tried this but using a pair_style which supports RESPA
might help. If you have freedom to play with the force-field, then
switching from Lennard-Jones (1/r^12) to a pair_style with softer
repulsive forces can also help. You can also try use weaker harmonic
bonds, angles, etc.

Andrew

Although I don't think it will help, it can be fun to try visualizing
the trajectories of your simulation, saving often around the point
where the error occurs, and see if you can see where the explosion
occurred. (I suspect two atoms got too close together at some point.)

P.S. more pontification (suggestions for using minimization and fix nve):
http://lammps.sandia.gov/threads/msg33303.html

5

Thank you very much for the advise, Andrew. Yes, exporting the trajectories is an effective method to figure out what happened. So I plan to first use smaller timestep at the point where the error occurs and use pair_style with softer repulsive forces.

Best Regards,
Chi Ma