[gmx-users] Lincs warnings

[Andreas Kukol]

Dear all, 

When trying to energy minimise a lipid bilayer of DMPC, I continuously get LINCS warnings and a subsequent MD simulation crashes.

step 0, time 0 (ps)  LINCS WARNING
relative constraint deviation after LINCS:
max 2.427052 (between atoms 5487 and 5488) rms 0.068787
bonds that rotated more than 30 degrees:
 atom 1 atom 2  angle  previous, current, constraint length
    650    651   39.1    0.1437   0.1434      0.1430
    661    662   35.9    0.2929   0.1525      0.1530
     ...        ...       ...

The atoms concerned (5487, 5488) refer to a DMPC molecule, which is located at the corner of the cubic simulation box in the starting structure (obtained from Peter Tielemans website). In fact, some atoms of this molecule are distributed over three adjacent corners of the simulation box.
Also other LINCS warnings refer to molecules which are broken up in similar ways. However, this should not be a problem due to periodic boundary conditions.
I do not understand these warnings, which eventually cause the simulation to crash.
Any idea how to solve this problem ?

Many thanks
Andreas