About NEB calculation with intermediate replica

LAMMPS LAMMPS General Discussion
1

Nice to meet you.
I want to do do a NEB calculation with one intermediate replica specified in Lammps (i.e. I want to specify one replica at reaction coordinate 0.5), but the calculation diverges. Also, I want to do the calculation with 31 replicas.
I am using the file-style of “each” in the neb command, am I using it incorrectly?
The working directory has relaxed_initial.pos, relaxed_final.pos.15, relaxed_final.pos.30, FeP_mm.eam.fs.
The execution commands and input file are as follows.

-----------------------------input file start----------------------------------
units metal
atom_style atomic

atom_modify map array
boundary p p p

neighbor 2.0 bin
neigh_modify delay 1 every 1

read_data relaxed_initial.pos
mass 1 55.845

pair_style eam/fs
pair_coeff * * FeP_mm.eam.fs Fe

compute ene all pe/atom pair
variable u uloop 1000
dump 1 all custom 500 NEB.dump.$u id type x y z c_ene
dump_modify 1 format line “%d %d %g %g %g %20.15g”

fix 1 all neb 0.1 parallel ideal

thermo 1
thermo_style custom step temp etotal ke pe press

timestep 0.002
min_style quickmin

variable i index 15 30
neb 0.0 0.001 10000 10000 10 each relaxed_final.pos.$i
------------------------------------input file end------------------------------------------

execution command :
OMPI_MCA_rmaps_base_oversubscribe=1 mpirun -np 31 /home/lammps-29Oct20/src/lmp_mpi -partition 31x1 -in in.neb

2

yes. “each” means one file per replica. In your input all will read the relaxed_final.pos.15 file. See the “neb” command documentation.

1 Like
3

Thanks for the reply.

I looked at the neb command, but I don’t know what command would accomplish what I want to do.
I want to do a neb calculation with 3 files (initial replica, one intermediate replica and an end replica).
Which command can I use to accomplish this calculation?

I apologize for the inconvenience and thank you in advance for your help.

4

According to the documentation, that is not supported. You would either have to forego providing the midpoint, or create all intermediate positions (e.g. from linear interpolation) and provide them.

A feature to assign explicit coordinates to selected replica and let LAMMPS do the interpolation would need some C++ programming.

1 Like
5

Thanks to you, I understand that to specify an intermediate replica, I need a replica of all the reaction coordinates.
I will try to solve this problem by extrapolating a program that does linear interpolation.

Thanks for your support!