[gmx-users] Rigid methane molecule using virtual sites (4fdn) construction

[Ubaya Higgoda]

Dear All,

I implemented the rigid Methane molecule using virtual sites 4fdn 
construction with LINCS. Here, I constructed the 5 virtual sites (1- 
carbon and 4 hydrogen) based on 4 mass centers (each mass center having 
a mass of 1/4th of mass of the Methane molecule) arranged in a 
tetrahedron formation, each located along the C-H bond. The location of 
the mass centers were calculated such that the moment of inertia of the 
methane molecule is preserved along an axis through the C-H bond and 
another axis perpendicular to the previous. The distances between the 
mass centers were constrained.

The simulation terminates within a first few picoseconds giving the 
following errors:

"....
Step 4, time 0.004 (ps)  LINCS WARNING
relative constraint deviation after LINCS:
rms nan, max 0.533022 (between atoms 6253 and 6252)
bonds that rotated more than 30 degrees:
  atom 1 atom 2  angle  previous, current, constraint length
Wrote pdb files with previous and current coordinates
Wrote pdb files with previous and current coordinates
Wrote pdb files with previous and current coordinates
Wrote pdb files with previous and current coordinates
Wrote pdb files with previous and current coordinates
Wrote pdb files with previous and current coordinates
Wrote pdb files with previous and current coordinates
Segmentation fault (core dumped)"


However, when I constrained only 5 connections between the mass centers 
and let only one connection to be flexible (imposed a large bond 
constant), I was able to run a stable simulation with lincs-order=16 and 
lincs-iter=2, but the pressure seems to drift gradually to lower values 
(starting from 105 bar to 94 bar within 20 ns where the expected 
pressure should be 100 bar) . My concern is how can I run a stable 
simulation without imposing the additional large bond constant, which 
seems to cause the drift in pressure. I see this problem analogous to 
simulating a rigid ammonia molecule.

Thank you in advance

Best regards,
Ashandika