[lammps-user] New looking for initial help.

Hey everyone,
I am a totally rookie when it comes to everything lammps, I just started a few weeks ago. I’m a junior Chem major looking to use bead-spring modeling to predict how different groups on a polymer interact with each other based on the force fields I give them. I went through the documentation on the site and did some tutorials, but I was wondering if any of you have any tips/good sites to help me get going more. I need to learn more on the syntax of the scripts for polymers. Thanks in advanced, and I’m sorry if this is in the wrong place, I’ve been searching the Internet for awhile and its a little over my head!
-Daniel Darcy

Hi,

The best thing to do is to learn Lammps expert (?) around you.

What is a possible approach for learning Lammps or running something with Lammps?

Think, think and think. First you need a correct input files (data file and input script).

Most important thing is how a data file and input script are formatted. There are several examples

in Lammps/example directory. You can take carefully look at those examples and then read

lammps manual. You might think that "how can I create a data file including system coordinates,

bond, angle, and dihedral topologies etc?". Go to lammps archive and find similar questions and

read answers. When you success creating a data file, then make your input script to test your

system. You might have error messages and have a headache to find solutions. You might also

look at lammps archive for similar questions before you send an email to lammps-users list.

I guess that this is a normal procedure to learn lammps.

I recommend you "try to find similar examples to what you want to simulate in the

lammps/examples directory and understand how input and data files consist of.

I hope this will help you learning lammps.

Thanks.

Best regards,
Changwoon Jang

Hey everyone,

hey daniel,

I am a totally rookie when it comes to everything lammps, I just started a
few weeks ago. I'm a junior Chem major looking to use bead-spring modeling
to predict how different groups on a polymer interact with each other based
on the force fields I give them. I went through the documentation on the
site and did some tutorials, but I was wondering if any of you have any
tips/good sites to help me get going more. I need to learn more on the

the internet is not a good place to obtain the information you are
looking for. you are best off to look for somebody local with more
experience. in addition, there are two kinds of problems that you may
run into: 1) problems with regard to general understanding of doing
molecular simulations and 2) problems specific to understanding the
syntax of lammps. as a beginner, you are likely to mix those two up.
again, talking to somebody local with more experience in molecular
simulations (it need not be with lammps) can get you get over a lot of
problems quickly and leave you with syntax issues, for which you can
then consult the manual (sometimes it needs two or three reads to
fully understand what it is supposed to say) and *then* you should
post your question here.

the lammps webpage also links to a rather detailed tutorial at
mississippi state, that may be helpful as well. but you should
definitely start by talking to somebody else and perhaps work you way
though some of the examples shipped with lammps (they are supposed to
work) and discuss them with somebody. in general, it is always a good
idea to build up experience and confidence slowly and not to rush into
a project or assignment too quickly.

good luck,
     axel.

I agree with everything Axel and Changwoon said, especially regarding
obtaining local guidance (how to select a research topic, how to
select the software you need, how to install it, how to use it,
etc...)

  I would add that there are several very general examples of how to
build and customize your own "toy" molecules from scratch, including
polymers, and example input files containing how to run them are at:
http://moltemplate.org
(Warning: Also, in order to try these examples, you will need to have
access to a computer with proper bourne-shell, such as linux or mac,
or minGW, and know how to set your PATH environment variable.)

If you need to simulate entangled, polymer melts, this is harder.
Take a look at the discussions here:

http://lammps.sandia.gov/threads/msg32450.html
http://lammps.sandia.gov/threads/msg32472.html
http://sourceforge.net/mailarchive/message.php?msg_id=30427755
  some relevant-looking papers:
http://jcp.aip.org/resource/1/jcpsa6/v77/i3/p1554_s1
http://labs.cas.usf.edu/softmattertheory/papers/PhysRevE_72_061802.pdf
http://labs.cas.usf.edu/softmattertheory/papers/PhysRevE_80_031803.pdf

If you want to run a simulation of real all-atom molecules using
realistic force-fields, then most of the examples currently on the
"moltemplate" web page will probably not be very helpful. Instead
try:

http://lammps.sandia.gov/tutorials.html
http://lammps.sandia.gov/scripts.html
http://lammps.sandia.gov/howto.html
http://lammps.sandia.gov/prepost.html

(Did I leave anything out?)

Because LAMMPS gives you so much control over so many simulation
details. this means that there is a steep learning curve.

In fact, LAMMPS might not be the best program to use if you are
simulating an all-atom realistic biopolymer (proteins/nucleic-acids).
The problem is that LAMMPS does not come with the tools to build the
(rather large) input files which describe the details of the (complex)
molecule you are simulating. For that application, it would be easier
to use Gromacs, NAMD (or Amber, if you have access to it) and the
tools which come with them.

(Also consider ESPResSo. It was designed for running coarse-grained
simulations and it looks good too. There is also a new program called
OpenMM/Zephyr which comes with a basic GUI. I have not tried it yet.
These are both new. Older programs like DLpoly and Tinker are also
respectable.) If you need anything QMM, that is a different story.

http://en.wikipedia.org/wiki/Molecular_design_software
http://en.wikipedia.org/wiki/Category:Molecular_dynamics_software

So figure out what kind of question you want to investigate, and then
what kind of molecule(s) you are trying to simulate (is it a
coarse-grained, all-atom, organic, inorganic, charged, solvent?, ion
properties, electronic states, if applicable, etc...). Then worry
about which software simulates that kind of system, and look for
molecule-builder scripts (preprocessors) can be used to create input
files for those simulation programs. This may take you a lot of time.
I will add that I think it would be difficult to start and finish a
project for a typical 10-week undergraduate course, unless you have
help, or are willing to tinker with something else which has already
been done.

Good luck
Andrew

A key piece of advice that I wish I had heard not once but repeatedly when I started working in this area:

Start small… check for correctness… scale up.

The last thing you want is to waste many hours obtaining simulation data to later learn that there was a catastrophic typo in your input script that resulted in a wrong interaction or some other problem (not that this ever happened to me :wink:

A corollary to this is: Just do it.

Don’t be afraid to try many different simulations, force fields, calculation, analyses, and such. IN A SMALL SCALE. Experiment extensively with small systems and when you find something that works, solves your problem, and seems interesting. Proceed with scale up.

Finally, as you work through learning the practical aspects of MD, work to have a good understanding of the underlying statistical mechanics. There are numerous books that can help you with this. One of my favorites is McQuarrie’s Statistical Mechanics. For MD in particular, Rapaport’s The Art of Molecular Dynamics Simulation is a great text.

Salo