LAMMPS WWW Site - LAMMPS Documentation - LAMMPS Mailing List Archives
Re: [lammps-users] dihedral_style multi/harmonic
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [lammps-users] dihedral_style multi/harmonic

From: Andrew Jewett <jewett@...1937...>
Date: Wed, 18 Oct 2017 14:01:33 -0700

It sounds as though you are asking a more general question than the
one you posted.

You have a simulation containing molecules, and you would like to know
how to choose all of the force-field parameters for the interactions
between these atoms, (including the dihedral interactions).  You may
also be seeking a way to automatically generate lists of these angle,
dihedral, and improper interactions for your molecule(s), and also the
partial charge of each atom.

This is a non-trivial task.  Most people don't manually create the
files that LAMMPS reads (input, data).  It is the most complicated
part of preparing any simulation.  Most people use 3rd party tools:

Check out:

I have some knowledge of topotools, moltemplate, EMC, and ATB, and
I'll make a stab at trying to compare their strengths and weaknesses.
(I wrote moltemplate, so I am naturally biased toward it.)  All of
these tools eliminate the tedious task of replicating molecules and
generating lists of angles, dihedrals, and impropers automatically.
Some tools also generate force field parameters and atom charges

The ATB is a large database of molecules currently based on the GROMOS
force field which are completely parameterized.  For each of these
molecules, the angles, dihedrals, impropers, force field parameters,
atom types and partial charges (basically everything) have chosen for
you automatically.  This is probably the safest way to get started
running accurate molecular dynamics simulations in LAMMPS without
making any serious mistakes.  It is a big database, but there are some
molecules which nobody got around to adding.  Other tools enable you
to build your own molecules of arbitrary size.

Moltemplate and EMC generate lists of angles, dihedrals and impropers
for arbitrary molecules automatically, and someone took the time to
build databases of force field parameters for several popular force
fields including OPLS(moltemplate,EMC), COMPASS(EMC, moltemplate),
AMBER(moltemplate) PCFF(EMC, moltemplate*) and CHARMM(EMC,
moltemplate*), and Born(EMC).  (*available soon).  Each of these force
fields (OPLS, AMBER, COMPASS,...) is a set of force field parameters
that work well together.  Different force fields work well for
different types of molecules.   You must search online and ask experts
around you to decide what kind of force field works well for your
molecules, and then choose the tool which supports that particular
force field.  However with moltemplate, at least, you must be careful
choosing these atom types.  (As an example how difficult this can be,
there are many different types of carbon atoms with different
hybridization states.  You must choose which one applies to the carbon
atoms in your molecules.  Moltemplate has a couple examples to help
you choose, but it needs more.)

I have not tried EMC, but my impression is that it might be able to
help automatically choose atom types for you (for certain force fields
at least), which can reduce the chance of making a mistake.  EMC also
supports other ways of inputing molecules (SMILES strings) and other
force fields (Born), and it has been around longer and it has probably
undergone more testing (so the force field parameters for COMPASS and
PCFF are more likely to be correct!).

The advantage of topotools and moltemplate is that they give you a
high degree of autonomy in building your system.  They are not "black
box" programs.  Topotools and moltemplate give you complete autonomy
to build your system and customize interactions for certain atoms and
molecules and manually add and delete bonds.  All they really do is
eliminate the most redundant tasks when building LAMMPS input & data
files (such as replicating molecules and generating interactions.)

--- partial charges ---

For most force fields, both moltemplate and EMC can generate partial
charges for your molecule acording to atom type.  However some force
fields (OPLS3 and AMBER) require the user to calculate the charge of
each atom using quantum chemistry software.  (This must be done using
an external program.  Afterwards, you can apply these charges to the
atoms in your LAMMPS system using a series "set atom N charge Q"
commands, or by manually editing the DATA file.)

If you download molecules from ATB, however, both atom types and
partial charges have been determined for you, so there is less change
of making catastrophic mistakes.

hope this helps

P.S.  I should mention there are also some old tools msi2lmp, which rely on 3rd party tools (eg, DiscoveryStudio)
and convert the output into LAMMPS format.  My impression is that
these tools are aging and the original author has moved on and is not
willing to fix bugs.  And they do have bugs (serious bugs in the case
of msi2lmp).  I am somewhat biased against these tools for this

On Wed, Oct 18, 2017 at 4:24 AM, Ali Golbaf via lammps-users
<> wrote:
> Dear all,
> My simulation includes molecules with different atoms.
> how I can  get the "dihedral_style multi/harmonic" coefficients between
> different atoms?
> ------------------------------------------------------------------------------
> Check out the vibrant tech community on one of the world's most
> engaging tech sites,!
> _______________________________________________
> lammps-users mailing list