Hybrid/Overlay Pair Style with AIREBO Potential

Dear LAMMPS-users

I am trying to use the hybrid/overlay pair style with AIREBO and lj/cut/coul/cut potentials as follows:
(i have 20 types of atoms so I will not typed them all)

pair_style hybrid/overlay airebo 2.5 1 1 lj/cut/coul/long 12.000000

pair_coeff * * airebo CH.airebo C H C NULL NULL NULL NULL NULL NULL … … …
pair_coeff 1 4 lj/cut/coul/long 0.005827 3.179795 #CN-o
pair_coeff 1 5 lj/cut/coul/long 0.005243 3.324835 #CN-n3
pair_coeff 1 6 lj/cut/coul/long 0.001593 2.234375 #CN-hn

pair_coeff 2 4 lj/cut/coul/long 0.002433 2.779780 #ha-o
pair_coeff 2 5 lj/cut/coul/long 0.002189 2.924820 #ha-n3
pair_coeff 2 6 lj/cut/coul/long 0.000665 1.834360 #ha-hn

pair_coeff 3 4 lj/cut/coul/long 0.009105 3.179795 #c1-o
pair_coeff 3 5 lj/cut/coul/long 0.008192 3.324835 #c1-n3
pair_coeff 3 6 lj/cut/coul/long 0.002490 2.234375 #c1-hn

pair_coeff 4 4 lj/cut/coul/long 0.009105 2.959920 #o-o
pair_coeff 4 5 lj/cut/coul/long 0.008192 3.104960 #o-n3
pair_coeff 4 6 lj/cut/coul/long 0.002490 2.014500 #o-hn


However afther about 0.3 ns the hydrogen moves away from the carbon atom and I begin to receive ‘nan’ in output. I use NPT at 400K. Do you have any idea what can be wrong?

Thanks in advance!

Pawel

Dear LAMMPS-users

I am trying to use the hybrid/overlay pair style with AIREBO and
lj/cut/coul/cut potentials as follows:
(i have 20 types of atoms so I will not typed them all)

pair_style hybrid/overlay airebo 2.5 1 1 lj/cut/coul/long 12.000000

pair_coeff * * airebo CH.airebo C H C NULL NULL NULL NULL NULL NULL .. .. ..
pair_coeff 1 4 lj/cut/coul/long 0.005827 3.179795 #CN-o
pair_coeff 1 5 lj/cut/coul/long 0.005243 3.324835 #CN-n3
pair_coeff 1 6 lj/cut/coul/long 0.001593 2.234375 #CN-hn
..
..
pair_coeff 2 4 lj/cut/coul/long 0.002433 2.779780 #ha-o
pair_coeff 2 5 lj/cut/coul/long 0.002189 2.924820 #ha-n3
pair_coeff 2 6 lj/cut/coul/long 0.000665 1.834360 #ha-hn
..
..
pair_coeff 3 4 lj/cut/coul/long 0.009105 3.179795 #c1-o
pair_coeff 3 5 lj/cut/coul/long 0.008192 3.324835 #c1-n3
pair_coeff 3 6 lj/cut/coul/long 0.002490 2.234375 #c1-hn
..
..
pair_coeff 4 4 lj/cut/coul/long 0.009105 2.959920 #o-o
pair_coeff 4 5 lj/cut/coul/long 0.008192 3.104960 #o-n3
pair_coeff 4 6 lj/cut/coul/long 0.002490 2.014500 #o-hn
..
..
..
..

However afther about 0.3 ns the hydrogen moves away from the carbon atom and
I begin to receive 'nan' in output. I use NPT at 400K. Do you have any idea
what can be wrong?

lots of things. impossible to say with so little information.

axel.

My system is carbon nanotubes (C type 1) witch attached to the wall of CNT 24 hydrogen atoms (H type 2) and 24 PEG chain (C from chain typ3). Always only one hydrogen atom moves away from CNT surface and for each time with different id. If I run program from restart_data saved just before the hydrogen atom moves away, program works a little longer but anyway the hydrogen moves away and i get ‘nan’.
what additional information will be helpful?

Pawel

My system is carbon nanotubes (C type 1) witch attached to the wall of CNT
24 hydrogen atoms (H type 2) and 24 PEG chain (C from chain typ3). Always
only one hydrogen atom moves away from CNT surface and for each time with
different id. If I run program from restart_data saved just before the
hydrogen atom moves away, program works a little longer but anyway the
hydrogen moves away and i get 'nan'.
what additional information will be helpful?

first and foremost. you have to understand, that if there is a problem
on the science side, it will be difficult to provide any help short of
doing the project in your stead. so first you need to figure out,
whether you are providing consistent data and meaningful input. nobody
can tell this from a few lines of pair_coeff statements. there are
many things that can go wrong:
- incorrect choice or implementation of the force field or potential
parameters, e.g. it is highly suspicious that you are using
hybrid/overlay for your system. why not hybrid?
- incorrect units or unit conversion
- bad choice of simulation parameters (time step, neighbor list
settings, cutoffs etc.)
- incorrect input structure/topology
- incorrect physics of your model
- unphysical choice of system manipulation (time integration and other fixes).

for most of those, it can be difficult to spot problems from just
looking at the input deck unless they are blatantly obvious issues.
for everything else, you need to simplify and cut down your input to
the smallest possible size that still reproduces the problem at hand
(it doesn't have to make sense scientifically). if this is small
enough, people can experiment with it and narrow down potential
causes.

apart from that, there is also the chance of missing bugfixes to the
code, i.e. you should always state what version of LAMMPS are you
using and on what platform and if it is not the current one, you
should test with the very latest patchlevel or the svn/git version.

please also see here for mailing list posting guidelines:

axel.

okay, thanks anyway for answering