[gmx-users] simulation crash

[herbst at fhi-berlin.mpg.de]

Hi all,

I am trying to simulate an alpha-helix with different sidechains. I do
simulations with two different temperatures (200K and 600K). When I use a
pure alanine helix, both temperatures run fine. When I exchange a few of
the sidechains with other amino acids, only the 600 K simulations work.
The 200K simulations crash after a short time, because the system is
flying apart.

Step 715282, time 2145.85 (ps)  LINCS WARNING
relative constraint deviation after LINCS:
max 222372.468750 (between atoms 16 and 17) rms 43464.628906
bonds that rotated more than 30 degrees:
 atom 1 atom 2  angle  previous, current, constraint length
     ...
      4      5   90.0    0.1000 3301.5032      0.1000
     10     11   90.0    0.1000 2889.1333      0.1000
     16     17   90.0    0.1189 22237.3457      0.1000
     22     23   90.0    0.1000 2224.6282      0.1000
     73     74   90.0    2.6739   2.6304      0.1059
     73     75   90.0    3.5212   2.3823      0.1059
     73     76   90.0    2.4126   2.8698      0.1059
     77     78   90.0    3.0696   4.4763      0.1059
     77     79   90.0    3.7020   3.3081      0.1059
    ...

Back Off! I just backed up step715281.pdb to ./#step715281.pdb.1#
Wrote pdb files with previous and current coordinates
Fatal error: ci = -2147483648 should be in 0 .. 575 [FILE nsgrid.c, LINE 218]

I use constraints on the c-alpha atoms, so the helix should be stable.
I of course optimized the starting geometry, I tried to use the output
geometry of a 600 K run as input file, and I tried the simulation with a
larger box, but none of this worked. What else can I try ? What could be
going wrong ?

Thank you for your help,
Anna