[lammps-users] Fire and quickmin minimization styles

LAMMPS LAMMPS Mailing List Mirror
1

Hello all,

I have a problem when I minimize with FIRE style.

When I write commands like this:

     min_style fire
     minimize 1.0e-7 1.0e-5 9000 90000

I always get equal energies on iterations 0 and 1:

    min_style fire
    minimize 1.0e-7 1.0e-5 9000 90000
    WARNING: Resetting reneighboring criteria during minimization
    Memory usage per processor = 1.96328 Mbytes
    Step energy
       0 -1789.6471239502
       1 -1789.6471239502
   Loop time of 0.00388503 on 1 procs for 1 steps with 24 atoms
  Minimization stats:
    Stopping criterion = energy tolerance
    Energy initial, next-to-last, final =
        -1789.64712395 -1789.64712395 -1789.64712395
    Force two-norm initial, final = 171.902 171.902
    Force max component initial, final = 45.2298 45.2298
    Final line search alpha, max atom move = 0 0
    Iterations, force evaluations = 1 1

As you can see energies of the 0 and 1 iteration are equal, so NO
minimization at all takes place!

This happens when I minimize energy of different molecules using
ReaxFF potential.

Can I fix it somehow?

Thanks

2

Try running with an energy tolerance of 0.0 and
see if it works better. If not, please post a simple
input script/data fille that has the problem.

The quickmin and fire minimizers
have intermittant start-up iterations where the energy
does not decrease, so it's better to think about them
as operating to reduce forces rather than energy.
We need to add some logic to these minimizers to
skip the energy checks during the start-up phase,
but haven't done it yet ...

Steve