refrence of reax potential parameter set

LAMMPS LAMMPS Mailing List Mirror
1

Dear Lammps Users and Developers

I tried to find the refrence of “ffield.reax.Al_Al0_AlN” Reax parameter set that exist in potential folder. Please some one help me.

Thanks

2

That file isn’t part of the LAMMPS distro, so not sure
what you are asking.

Steve

3

The refrence of all reax potential set is written in README.reax file exept ffield.reax.Al_Al0_AlN.

I found it usefull for special work and I want to know the refrence of this patameter set.

I think it is a reasonable question to ask from LAMMPS Developers the refrence of a potential parameter set that is exist in the LAMMPS tar file.

Thanks

4

The refrence of all reax potential set is written in README.reax file exept
ffield.reax.Al_Al0_AlN.
I found it usefull for special work and I want to know the refrence of this
patameter set.
I think it is a reasonable question to ask from LAMMPS Developers the
refrence of a potential parameter set that is exist in the LAMMPS tar file.

what steve is saying is, that it does *not* exist in the (current)
LAMMPS distribution. if it had ever had been distributed in the past,
than it must have been in error and thus it is recommended against
using it. please see below. it is not there.

axel.

[[email protected]... lammps]$ find . -name ffield.reax\* -print
./potentials/ffield.reax.cho
./potentials/ffield.reax.FC
./potentials/ffield.reax.rdx
./potentials/ffield.reax.lg
./potentials/ffield.reax.AB
./potentials/ffield.reax.mattsson
./potentials/ffield.reax.ZnOH
./potentials/ffield.reax.AuO
./potentials/ffield.reax.budzien
./potentials/ffield.reax.V_O_C_H
./potentials/ffield.reax.Fe_O_C_H
./bench/POTENTIALS/ffield.reax
./examples/reax/AB/ffield.reax.AB
./examples/reax/Zn_O_H/ffield.reax.ZnOH
./examples/reax/V_O_H/ffield.reax.V_O_C_H
./examples/reax/ffield.reax
./examples/reax/Fluorographene/ffield.reax.FC
./examples/reax/Au_O/ffield.reax.AuO
./examples/reax/RDX/ffield.reax.rdx
./examples/reax/CHO/ffield.reax.cho
./examples/reax/Fe_O_H/ffield.reax.Fe_O_C_H

5

I seem to recall we removed a ReaxFF potential file
or two, for exactly these reasons. They were
not documented with a paper from the literature.
People are welcome to use such files if
they find or create them, but caveat emptor.

Steve

6

Dear Lammps users and developers,

I have a question concerning nudged elastic band (NEB) calculations.
In the "neb" command in LAMMPS, "etol" and "ftol" are two convergence criteria, and "Nevery" is the timestep for printing information:

neb etol ftol N1 N2 Nevery filename

It is my understanding that "Nevery" should not affect the calculation.
However what I observed is that changing Nevery impacts the number of steps required to complete the NEB calculation.

The attached example shows how to reproduce this behaviour. It aims at computing the energy barrier for Mg migration in MgSiO3 perovskite. Using Nevery=1000 the NEB calculation completes in less than 3000 steps (stopping criterion: energy convergence), while when using Nevery=1 the NEB calculation is still not complete after 100,000 steps (stopping criterion: max iterations).

So my question is, why does the parameter Nevery affect the calculation so much? Is there a specific problem when using a too small value of Nevery? What reasonable value of Nevery should I use?

PS: I am running LAMMPS 6Apr13 with the following command:
mpirun -np 4 lmp_openmpi -partition 4x1 -in input_diff.in

Best regards,
Antoine

example.tar.gz (4.76 KB)

7

Which minimizer are you using? It could be that
FIRE or QuickMin is re-initializing something
each time the minimizer starts up (which happens
in chunks of Nevery steps), so maybe that
is the problem.

Steve

8

In the attached example I'm using the FIRE minimizer.
But I observed the same behaviour with Quickmin.

Antoine

9

Hi,

you can also try to ask Prof. van Duin, who is, I think, the main expert in ReaxFF parametrizations. Keep in mind that CHON part is the same as in ffield.reax.budzien. You may ask about Reax potentials that are based on force field that Budzien used, and that also include S, Si and Al.

Please share the reference if you find any.

Regards,
Oleg.

03.07.2013, 08:24, "mohammad ashajer" <[email protected]...>:

10

Dear Oleg

is it possibel for you to send me the force field that Budzien used, and also include and Al.

Thanks

Mohammad

11

The budzien potential is, I believe, modified from Chenoweth’s combustion potential by folks at CalTech. The budzien potential has not been published as a potential development paper, but as an energetic material application paper. No details of its parameterization can be found.

The Al part is just as mysterious - we don’t know where the S, Si and Al that was added to the budzien potential came from. We think it is best to have reaxff force field files in LAMMPS distro that can be clearly cited and traced, and this is the reason the Al part was taken away.

Ray

12

Dear Mohammad,

it is the Al_Al0_AlN potential - it's budzien+Al. But again, I don't know the citation for it, and it doesn't seem anyone here knows.

Oleg.

06.07.13, 18:07, "mohammad ashajer" <[email protected]...>":

13

Please try the 8Aug patch and see if that fixes
your problem … another user complained about
convergence issues with FIRE and I think
it was related to your problem as well.

Steve