Dear all
While I was trying to model a polymeric network, I had to define 3 types of dihedrals. I have to set all the pair-wise interactions (1-2, 1-3 and 1-4) as zeros in 2 types and have to allow a non-zero pairwise interaction between atoms separated by 3 bonds in third type.
What is the way to implement this in LAMMPS?
Can I use special_bonds specifically for each dihedral types?
I would be grateful to you for your valuable help.
Thank you
Jeeno Jose
Dear all
While I was trying to model a polymeric network, I had to define 3 types of
dihedrals. I have to set all the pair-wise interactions (1-2, 1-3 and 1-4)
as zeros in 2 types and have to allow a non-zero pairwise interaction
between atoms separated by 3 bonds in third type.What is the way to implement this in LAMMPS?
there is no support for this in the current official LAMMPS version.
in general, what you describe would be the result of mixing parameters
from different force fields and that is as such *strongly*
discouraged, since each of these force field has different
parameterization strategies and balances for the non-bonded
interactions and thus mixing such parameter sets would result in
incorrect mixing behavior. there are some special cases, where this
still makes sense, e.g. when you have a two separate components with
particular adapted parameters for the interface interaction.
to support such systems, i have experimental support for this kind of
setup in my LAMMPS-ICMS branch git repository. this is implemented as
options to using pair style hybrid where it is possible to assign
special bonds settings to each substyle which are then used to
temporarily change the global special bonds settings. this is more
complex as it may sound, since the special bonds settings also affect
the neighbor list algorithm and thus certain constraints have to be
enforced to require that the global settings are compatible with the
substyle settings. preliminary documentation is currently here.
http://git.lammps.org/git/?p=lammps-icms.git;a=blob;f=doc/pair_hybrid.txt;hb=HEAD#l311
this feature is currently submitted for inclusion into the upstream
LAMMPS versions, but this has not yet happened (as it is not a highly
important change).
Can I use special_bonds specifically for each dihedral types?
no. that would be pointless, since special_bonds factors affect pair styles.
axel.
Dear all
Let me put my problem in a broad perspective, than what discussed in previous submission.
While I was trying to model a polymeric network, I had to set zero pair-wise interaction between all possible 1-2, 1-3 and 1-4 occurrences, excluding one dihedral type, where I have to have 1-4 interaction to be non-zero (but just a fraction of original interaction between those two atom types if they were not in 1-4 positions).
What is the way to implement this in LAMMPS?
Is it like creating some number of dihedrals, in an exhaustive manner to contain all atoms, and then using “special_bonds” command? But then, can I use “special_bonds” specifically for each dihedral types, so that I could include the variable weighting coefficients (as discussed in second paragraph of this mail)?
I would be grateful to you for your valuable help.
Thank you
Jeeno Jose
Dear all
Let me put my problem in a broad perspective, than what discussed in
previous submission.While I was trying to model a polymeric network, I had to set zero pair-wise
interaction between all possible 1-2, 1-3 and 1-4 occurrences, excluding one
dihedral type, where I have to have 1-4 interaction to be non-zero (but just
a fraction of original interaction between those two atom types if they were
not in 1-4 positions).
please note that exclusions are *only* applied to non-bonded
interactions that are connected via bonds. it is not clear from your
description whether you are aware of that. if you were not, you have
some serious additional thinking to do.
What is the way to implement this in LAMMPS?
Is it like creating some number of dihedrals, in an exhaustive manner to
contain all atoms, and then using "special_bonds" command? But then, can I
use "special_bonds" specifically for each dihedral types, so that I could
include the variable weighting coefficients (as discussed in second
paragraph of this mail)?
your description is describing the same model as you did in the
previous e-mail and thus my explanation and comments are equally
applicable. whether having multiple exclusions settings is a valid
model or not, is ultimately up to you to decide. i was just remarking
that the vast majority of cases where people ask to use it, are not a
good model. i also explained that with the current LAMMPS code this is
not possible. exclusions are a global property. if i understand
correctly what you describe, this would be best implemented by
programming a new dihedral style that includes the scaled 1-4
interaction as additional components of the dihedral style. globally,
you can keep those interactions excluded.
What I meant as pair-wise interactions, were non-bonded interactions itself.
What I meant as pair-wise interactions, were non-bonded interactions itself.
this is as confusing a statement as some of the other descriptions you
have given. thus i still have the suspicion that you are not fully
realizing how exclusions work overall and thus may be trying to solve
the wrong problem. none of your statements so far have indicated that
you do understand what LAMMPS is doing. and you also have not provided
a specific enough description, let alone a particular example with
proper comments and verification that what you think you are doing is
actually what LAMMPS is going to do.
axel.