[gmx-users] Simulating Infinite Nanotube

[Patrick Lafond]

Hello, I was wondering if perhaps someone could help me out with running a
simulation of an infinite CNT.  I have searched through the archives before
and have tried all of the suggestions I was able to find but still can't
seem to figure out what the problem is so I decided it best to try and ask
for direct help.  Here is what I have done:

     1) Taken the PDB file which is exactly the structure for the tube and
converted it to a gro file with the proper dimensions for periodic height
and ample room on the sides.
     2) Used x2top to convert the gro file to a topology file using the
"-pbc" option.  In the topology file I see that it is infact specifying
bonds between the carbons at the top of the tube and carbons at the bottom
of the tube.
     3) I have tried running the simulation using different options for the
pbc in the mdp file.  I at first tried pbc = xyz but then changed that to
full.  I also tried fiddling with
non-bonded exclusion trying 3, 4, and 5.  I also tried changing the time
steps to shorter steps.
     4) When none of the above mentioned attempts seemed to work I tried
making a new slightly shorter tube using the same method mentioned above.

     Regardless of what I seem to do every time the simulation yields the
same results which is that it won't run.  LINCS provides a warning for just
about every carbon in the tube a handful of times and then reports that
there were errors in the simulation.  Upon reviewing what it had
accomplished the ends of the tube, and 2 rows in the very middle of the tube
crumple together and them fly apart presumably from how close together the
atoms are forced there.  I have a very strong suspicion that the simulation
isn't recognizing that the atoms are actually bonded to atoms in the image
(which would explain the crumpling from how close the atoms are to the
periodic image) but cannot figure out how to actually achieve this
recognition in the simulation.  I read just recently that LINCS has problems
with molecules containing a lot of bonds but then read the problem is more
with the constraints applied.  It doesn't seem like the error is in LINCS
rather it is a lack of recognition of bonding between the actual molecule
and the image.  Thanks for your time, I hope I can get some help on this.

(By the way, I am running the simulation on version 3.1.1 if that might
account for any of the problems)

Sincerely,
Patrick
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20071101/256613d4/attachment.html>