[lammps-users] hydrogen bonding potential in lammps?

Hi Chetan (and others interested in the DREIDING ff) -

I reworked and released Tod's DREIDING files today as
a patch. So please try them out and see if they work for your problem.

Steve

Thanks, Steve. I am in the process of working on exactly the same thing. I will send you feedback in a week or two.

Hi Steve

First of all, i would like to thank both Steve and Tod.
So i tried Dreiding hydrogen bonding potential in lammps. My first few comments are:

  1. Total potential energy gets more and more negative during energy minimization. I am mentioning this, since with earlier code that was shared between us and Tod, potential energy, although reducing with minimization, still stayed substantially positive for the same system.

  2. Also, i would like to suggest adding ehbond thermo keyword to print out that energy. Earlier i did use to print that, but it was always very small and close to zero.

  3. Minor thing: Syntax line on http://lammps.sandia.gov/doc/pair_hbond_dreiding.html needs to have N before distance cut-off.

  4. I am glad that there is no restriction such as newton off with the earlier files. That should save substantial computational costs.

  5. My runs for this are going on and i will get back to you in few days after my runs are over and i analyze results.

cheers
Chetan

I'll ask Tod to look at your example of the problem
with the energy and see if he thinks the current
behavior is bad, as compared to his original
version.

Re: the ehbond energy, you can get that out
via the compute pair command. See the
pair_dreiding doc page for details.

I'll change the doc page.

Thanks,
Steve

Thanks, Steve. I would just like to make one point: It seems that in current behavior, the reference points of hydrogen bond energy are compatible with others, so total potential energy is negative. While negative or positive depends on reference points, it certainly seems more tractable to follow current behavior than previous.

Hi Steve and Tod

So i tested both LJ and Morse form of DREIDING hb potential and here are my observations:

  1. Both the forms can produce strong (polymeric) acid-base bonding. However, diffusion of solvents through such blend membrane is better observed with Morse form. It produces experimentally observed trends.

  2. I printed out hydrogen bond energy and but it is always zero or something like exponential (-308). I do not know why it is happening. However, i can confidently say that hydrogen bonding potential is working to produce better bonding than the case where it would be absent.

Thanks
Chetan

Can you post a simple input script (with supporting files)
that illustrates how you are printing the hbond energy
and the near zero values?

Steve

Hi Steve

I have attached files herewith.

Thanks
Chetan

restartf2neg.75000 (2.16 MB)

in.bl_dmfc2 (2.5 KB)

Also attached sample output file herewith.

sample.output (622 Bytes)

Chetan - Tod sent me an upgraded set of Dreiding pair files which
I just posted as a 30May11 patch today. Hopefully they
fix all the issues, including in the doc pages, that you mentioned
in your earlier emails.

Steve

Hey Steve

Thanks. Are these new developments available through svn repository as well?

Thanks
Chetan

Hey Steve
Thanks. Are these new developments available through svn repository as well?

everything that steve commits to the svn is usually available in the public
repository within 24 hours. for small changes, the svn may even be more
up-to-date than the patches.

you can check before updating through browsing the git repository here.
http://git.icms.temple.edu/git/?p=lammps-ro.git;a=summary

this repository is updated from the svn repository about 1 hour after
the svn repository is updated.

cheers,
    axel.

Hi Steve and Tod

Is this patch of 30th May 2011 to lammps to fix any serious issue for DREIDING hydrogen bonding potential originally released? This is since i was going to use results and lammps executable generated before this patch, but am not sure yet to what extent i should trust these older results and executables. They are producing experimentally desired trend, as i noted earlier, but just wanted to know the extent of error that may be associated if 30th May patch is not used.

Thanks
Chetan

Hi Steve and Tod
Is this patch of 30th May 2011 to lammps to fix any serious issue for
DREIDING hydrogen bonding potential originally released? This is since i was
going to use results and lammps executable generated before this patch, but
am not sure yet to what extent i should trust these older results and
executables. They are producing experimentally desired trend, as i noted
earlier, but just wanted to know the extent of error that may be associated
if 30th May patch is not used.

from reading through the patch, the changes seem to be quite significant.

also, if i am reading the source code right, there is still one small bug
left in accumulating the the total hydrogen bond energy for use with
compute pair. this is one of the few potential bugs that i came across
while adding multi-threading to the code. i don't have a good example
to validate it yet, and haven't had the time to read up on how to set up
a DREIDING based simulation to generate one myself.

axel.