[lammps-users] WARNING: Dihedral problem

LAMMPS LAMMPS Mailing List Mirror
1

Dear LAMMPS users:

     I am simulating a triblock copolymer using COMPASS parameters, solvated in
water using TIP3P and SHAKE. I have been running nvt simulations that have run
for about 10,000 timesteps and then ended in errors with "WARNING: Dihedral
problem" and "Shake determinant = 0.0." I have checked all of my polymer's
parameters.

    I changed the command for the neigh_modify list to "every 1," to see if it's
a neighbor list problem, but I still got dihedral errors after 12,000 timesteps.

    Decreasing the timestep from 0.5 to 0.05 (real units) and running for an
equivalent period of time did not solve the error problem.

    I also increased the max iter. number for SHAKE from 10 to 20, but I still
got the same errors.

    Additionally, simulating the polymer for about 100,000 timesteps (0.5 fs)
without water did not give any dihedral problems.

    Any help in this matter would be appreciated. If needed, I can attach my
input and data files.

Thanks

Gaurav Pranami

Graduate Student
Chemical Engineering Dept.
Iowa State University

2

The dihedral warning occurs when the orientation of the 4 atoms
in the dihedral is going bad. It prints the values, so you should
be able to examine them and see it if looks bad, according
to your ff params.

Are you just shaking water or H bonds in the polymer as well?
SHAKE should not give that error unless the unconstrained
positions are way off. So again this indicates something
is going badly wrong with your simulation. I would probably
try to viz things and see what is happening.

Steve

3

This was a couple of 1-character typos in the angle class2 file that
were introduced before the 22Jan08 release. See the 21 Apr 08
patch. It should work as before now.

Running this for a while I noticed you are not getting good coupling
between the polymer and water in your thermostat. This is a common
issue with NVT. You will see this if you plot the T of polymer and water
separately. You may want to thermostat them independently or by
another means. See the howto section on thermostatting for more details.

Steve