NEB fix first, last replica

LAMMPS LAMMPS Mailing List Mirror
1

Hi,

I want to run neb simulations on lammps 7th Aug, 19 version. I realized that the first and last replica for neb are not fixed in lammps. I wish to restrict the movement of my first and last replica. Has anyone done this before or could guide me how to do this in lammps neb code?

Thanks,
Srijan

2

If both end points are (nearly) at energy minimums, then they will not
move (much). If they are not at energy minimums, you typically
want them to relax to the minimum. Is there some reason you
want to prevent that?

Steve

3

Hi Steve,

Thanks for your reply.
I ran NEB for my system with FIRE minimizer. In the screen output files generated for the images, I observe that with FIRE minimizer, the potential energy of the system increases to a very large number during initial steps of the run, and then comes down again as minimization proceeds further. This might allow my first and replicas to settle in some other minima, which I do not want, as I am interested in finding a Minimum energy path between 2 given stable conformations. Therefore, I was thinking fixing the first and the last image would be a good idea. But still, for the intermediate replicas also, energy should not increase during minimization ever! I am attaching my NEB input script and a screenshot of my screen output file here.

Please let me know your thoughts about this.

Thanks,
Srijan Singh

neb.inp (936 Bytes)

screen.png

4

The min_style doc page indicates you can use a smaller timestep with the FIRE
minimizer which may help. It should definitely not take a system out of
a (nearly converged) minimum it is already in. You can expt with this
issue outside of NEB, by simply loading your first or last state into
LAMMPS and running minimize with FIRE. The first or last replicas
should behave no differently when they are part of NEB.

Steve