improper definition in lammps

omid_gh · May 13, 2020, 3:41pm

hi all

i have a problem with improper definition to lammps .in the picture we have a improper how can I define it to the lammps? I mean the order of atoms for definition (i, j ,k , l ) or (j, i, k , l) or ( i , k , j ,l) or something .
thanks dear friends

Andrew_Jewett · May 13, 2020, 10:41pm

Generating the correct improper interactions in LAMMPS (or any other simulation software) can be a long and complicated task. I recommend choosing a molecule builder instead of trying to do this by hand.

The answer to your question:
It completely depends on which improper_style you want to use. Unlike some MD programs, LAMMPS gives you the freedom to choose the equation used to describe forces between the 4 atoms in an improper interaction. (The “improper style”. And if none of the existing improper styles use the equation you want, then it is relatively simple to create your own custom improper style that uses the equation you want by editing the LAMMPS code.) Some of the existing improper styles are listed at the end of this web page:

https://lammps.sandia.gov/doc/improper_distance.html

If you click on the links at the end of this page, you will see that there are many different improper styles. Some of them do not even use “angles”, like the φijkl angle that you showed here. For example, some improper interactions have equations that only depend on the height of a tetrahedron formed by the 4 atoms. For example, see:
https://lammps.sandia.gov/doc/improper_distance.html

You will have to choose which improper style you want to use (which itself is usually determined by the force field you have chosen), and then read the documentation for that improper style carefully. Different improper styles in LAMMPS use different atom order. For some improper styles, the central atom is assumed to be the first atom in the list (of atoms which appear in the “Impropers” section of your data file). For other improper styles, the central atom is the second or third atom.

Suggestion:

-First you must decide what kind of molecule you want to simulate,
-then what kind of force field you want to use,
-then look up what kind of equation that force field uses to describe the improper interactions
-then look through the list of improper styles available in LAMMPS, and pick the one which matches the force field you are using
-then look at the order that the atoms appear in that improper style (ie, which one of these 4 atoms is the central atom?).
-then create a LAMMPS data file which lists the atoms in that order.

Trying to do this by hand is usually laborious. Alternatively, I think the majority users choose a molecular dynamics program (eg LAMMPS, GROMACS, NAMD, OpenMM), which comes with a decent molecule builder which uses this force field. (Trying to build a LAMMPS data file by hand is possible. But it’s complicated enough that in most cases, I would not recommend it.)

(Welcome to MD.)
Cheers

Andrew

omid_gh · May 14, 2020, 5:29am

thank you dear andrew
i dont want to make data file by myself . i make it with msi2lmp but because i should write the coeffs by hand from a paper I should recognize the improper defined by msi2lmp and check it if it is correct .
i want to use class 2 improper style (cff forcefield) and use lammps for run the code. i read the improper style class 2 and find the moving atom is the last one .
can you confirm it for me?
thanks

Andrew_Jewett · May 14, 2020, 10:42pm

thank you dear andrew
i dont want to make data file by myself . i make it with msi2lmp but because i should write the coeffs by hand from a paper I should recognize the improper defined by msi2lmp and check it if it is correct .
i want to use class 2 improper style (cff forcefield) and use lammps for run the code. i read the improper style class 2 and find the moving atom is the last one .
can you confirm it for me?

I don't know what you mean by "the moving atom".
In improper interaction is an interaction between 3 atoms connected to
a central atom.
For the "improper_style class2", that central atom is the 2nd one in
the list (not the last atom).
(Will mentioned this as well.)

References:
1) documentation for improper_style class2

(Note that atom "j" is shared by all 3 of the theta interactions.)

2) Moltemplate implementation.
The following file is used by moltemplate when generating improper
interactions for molecules using the compass force field (compass is a
class2 force field). The second atom (atom "1", since indexing begins
at 0) is the central atom:

github.com

jewettaij/moltemplate/blob/master/moltemplate/nbody_alt_symmetry/cenJsortIKL.py

try:
    from ..nbody_graph_search import Ugraph
except:
    # not installed as a module
    from nbody_graph_search import Ugraph

#    To find 4-body "improper" interactions, we would use this subgraph:
#           3
#           *                  1st bond connects atoms 1 and 0
#           |              =>  2nd bond connects atoms 1 and 2
#         _.*._                3rd bond connects atoms 1 and 3
#       *'  1  `*
#      0         2
#

bond_pattern = Ugraph([(1,0), (1,2), (1,3)])
# (Note: Ugraph atom-index counters begin at 0, not 1)


def canonical_order(match):

This file has been truncated. show original

omid_gh · May 15, 2020, 5:06am

I don’t know what you mean by “the moving atom”.
I mean the atom that moves out plane (all three atoms cant move out plane only one of them can move because they are not the same and if another atom moves out plane we should take it to account for different improper type).

I read the document and found it the moving atom is the last one and central atom is the second one (as you said).
thank you for your valuable comment.(I wrote the mail like prof.kohmeyer who is great)
omid.