Dear all,
For crystals (or graphene monolayer) with bonds, angles, and dihedrals explicitly defined: as LAMMPS does not automatically create the connectivities between atoms that straddle periodic boundaries, I don’t think I can I use periodic boundary conditions in this case, right? For example, can I model a infinite large graphene single layer (in x,y plane) with explicit bonds, angles and dihedrals by using periodic boundary conditions in x and y directions?
I also notice in previous mailing list from Axel say that “potential with implicit bonds for graphene”. Does that mean bonds interacted with pair (and/or three-body) interaction potentials, in which the bonds connectivities are not explicitly defined?
thanks,
Jibao
Dear all,
For crystals (or graphene monolayer) with bonds, angles, and dihedrals
explicitly defined: as LAMMPS does not automatically create the
connectivities between atoms that straddle periodic boundaries, I don't
think I can I use periodic boundary conditions in this case, right? For
example, can I model a infinite large graphene single layer (in x,y plane)
with explicit bonds, angles and dihedrals by using periodic boundary
conditions in x and y directions?
yes, you can use PBC. you will get a warning about inconsistent image
flags, but LAMMPS will select the proper closest images in the respective
bond, angle and dihedral lists. i suggest you build a some simple test
cases, e.g. a linear chain of atoms connected with bonds only to verify. in
addition you could put those atoms at equal distances, but alternate the
bond type, so they have different r0 values. during minimization, the atom
positions should adjust accordingly. or have different distances, but only
one bond type and then all atoms should become equidistant.
I also notice in previous mailing list from Axel say that "potential with
implicit bonds for graphene". Does that mean bonds interacted with pair
(and/or three-body) interaction potentials, in which the bonds
connectivities are not explicitly defined?
yes. pair styles like AIREBO or ReaxFF (both of which are regularly
applied to graphene) determine bonds (and bond orders) on the fly from
within their respective pair styles and then include forces due to them.
this is why the require much more computational effort than a simple
classical (class 1) force field.
axel.
Dear Axel,
Thank you! I will do the test.
Best,
Jibao
Dear Axel,
Before the test, I would like just to confirm one point: regarding to “yes, you can use PBC. you will get a warning about inconsistent image flags, but LAMMPS will select the proper closest images in the respective bond, angle and dihedral lists.” I think you mean that: if I have a linear chain of atoms connected with bonds (specified by bond list) with the PBC in the direction of the chain, the chain will be as with infinite length because all the atoms in the chain (including the atoms in the box and the ones in the periodic images) interact with each other with the same bond interaction.
However, how would I do if I just want to simulate a box of alkyl chains with a fixed length using PBC?
Thanks,
Jibao
Dear Axel,
Before the test, I would like just to confirm one point: regarding to "*yes,
you can use PBC. you will get a warning about inconsistent image flags, but
LAMMPS will select the proper closest images in the respective bond, angle
and dihedral lists.*" I think you mean that: if I have a linear chain of
atoms connected with bonds (specified by bond list) with the PBC in the
direction of the chain, the chain will be as with infinite length because
all the atoms in the chain (including the atoms in the box and the ones in
the periodic images) interact with each other with the same bond
interaction.
i don't understand what you are saying here. simplified you have the
following scenario assuming you have 4 atoms arranged in a chain like this:
1--2--3--4 |
to make it an "infinite chain" across PBC, you need to add a bond between 1
and 4.
-1--2--3--4-|
the "inconsistent image flag" warning will come from the fact, that for
the 1--2 bond, you need the periodic image of 1 that is in the same box as
2. however, for the 1-4 bond, you need the bond where one of the two is a
periodic image from a neighboring box. when checking the data file for
consistency, LAMMPS cannot tell, that this is intentional and thus prints
the warning (as it is most other cases a sign of a bad input geometry).
However, how would I do if I just want to simulate a box of alkyl chains
with a fixed length using PBC?
it is the same principle. that is why i suggested to make those tests. do
them and you shall see. it is a waste of time to discuss hypothetical
outcomes before you have understood the basic principle and this is best
done here with a small test.
axel.
Hi Axel,
Thank you. I appreciate your help!
Best,
Jibao