Hello,
I am using Amber 6 to run MD on carbon nanotubes, and I had the following
problem:
I could run MD on (10,10), (15,15), and (20,20) nanotubes of length 5 unit
cells, but when I tried running with nanotubes (25,25) or larger, Amber
crashes giving a segmentation fault after running for a few timesteps.
Upon examining the output, the VdW energy is just *******, which means to
me that it has some really large value. If I turn off the VdW energy by
setting the lennard jones coefficients to 0 in the input files, the
program behaves normally. Do any of you have encountered such problems?
I have looked at the source code, and it seems to read in all the bonding
information correctly. By looking at the structure that results from the
few iterations AMBER can produce before it crashes, it seems that some
C-C bonds get stretched very far (to 2.86 Angstroms). It seems that the
1-2 nearest neighbor VdW interactions don't get excluded in the energy
calculation, which makes the energy to become huge.
Stefan Andreev
------------------------------------
Stefan Andreev
12 Oxford Street
Department of Chemistry, Box 102
Harvard University
Cambridge, MA 02138
USA
Office Tel: (617) 495-8859
Office Fax: (617) 384-7920
e-mail: sandreev_at_fas.harvard.edu
Received on Sun Aug 11 2002 - 16:59:26 PDT
