Complex geometries + packing + regions

LAMMPS LAMMPS Mailing List Mirror
1

Hello everyone,

I am very new to MD. I have three lammps questions and one on MD in general:

1. In lammps, creating a region necessitates having a pre-defined lattice. But what if I am reading my data from a .pdb file and I have no lattice defined? It seems that some functionality is lost unless I am missing something.

2. How can I create some complicated geometry representing surface roughness? For instance, I have a surface of parallel layers of polymer chains (<<<I've gotten this far). Now I need to cut out some rows to create grooves (this looks straightforward). But then I need to cut in the perpendicular direction to create some pillar-like surface topology. The removed chains need to be replaced by hydrogen. Any suggestions that would make this easier? Would something like topotools help me in that problem?

3. Can lammps help when it comes to extracting PCFF parameters and potentials?

4. I am doing a basic water simulation where I start with a cube of H2O and let it morph into the natural equilibrium config., the shpere. The initial "shock" is causing the molecules to disperse too rapidly. Now I know this has to do with the initial spacing, but how do you go about this? I mean is there a specific method to calculate the best initial packing or do you learn this from experience? (I'd love to get a book on this: Chemistry books don't go beyond models. MD books don't say much besides how the simulation works "under the hood". I can't seem to find a reference with that kind of practical info)

I started doing this just this past month, so please go easy on me. Any help, whether answers or suggested readings/tools, is greatly appreciated.

Best,
MD Newbie

2

Hello everyone,

I am very new to MD. I have three lammps questions and one on MD in general:

1. In lammps, creating a region necessitates having a pre-defined
lattice. But what if I am reading my data from a .pdb file and I have
no lattice defined? It seems that some functionality is lost unless I
am missing something.

you are missing the "units box" option. the region command by
default uses "lattice units", but if you use "unitx box" you don't
need a lattice to be defined. also creating a lattice is always
possible.

2. How can I create some complicated geometry representing surface
roughness? For instance, I have a surface of parallel layers of
polymer chains (<<<I've gotten this far). Now I need to cut out some
rows to create grooves (this looks straightforward). But then I need
to cut in the perpendicular direction to create some pillar-like
surface topology. The removed chains need to be replaced by hydrogen.
Any suggestions that would make this easier? Would something like
topotools help me in that problem?

topotools by itself can only help you with part of the process.
general VMD scripting can help you to get the rest of the way.
you can pick the atoms that you want to keep with an atom selection
and save/copy them to a new "molecule". you can pick the atoms
that need to changed to hydrogens by selecting them and resetting
atom name/type, mass, and charge as needed. VMD scripting can be
extremely powerful, but requires some practice.

there are alternate approaches like writing your own program to
build the structure you need and outputting a data file directly.
that may be faster in the short run, but less flexible.

3. Can lammps help when it comes to extracting PCFF parameters and potentials?

not really.

4. I am doing a basic water simulation where I start with a cube of
H2O and let it morph into the natural equilibrium config., the shpere.
The initial "shock" is causing the molecules to disperse too rapidly.
Now I know this has to do with the initial spacing, but how do you go
about this? I mean is there a specific method to calculate the best
initial packing or do you learn this from experience? (I'd love to get
a book on this: Chemistry books don't go beyond models. MD books don't
say much besides how the simulation works "under the hood". I can't
seem to find a reference with that kind of practical info)

ideally, you want to start from a pre-equilibrated bulk water block,
then cut out your sphere (e.g. using a VMD selection) and convert
it to a LAMMPS data file. now before starting a simulation with this
right away, you first want to relax the atoms on the surface, since
their environment has changed a lot. you can use fix setforce 0.0 0.0 0.0
or fix spring/self to keep them (more or less) in place.
then you'd run a minimization to remove excess potential energy.
then you can start on those atoms (at low temperature) and
keep relaxing them and then slowly release the other atoms.

this is probably the most careful procedure and depending on your
exact boundary conditions you may be able to skip some of it or
shortcut them. there are tutorials on the web describing how to
set up a solvated protein system from a crystal structure (which
is essentially the inverse procedure) and you'll find that those
protocols are similar to my outline.

axel.