[gmx-users] LINCS warning involving hydrogen

[Senthil Kandasamy]

Hi all, 

I have a glycerol headgroup for a lipid and I am running into LINCS
warnings during a short md run which eventually crashes.

structure:

H-O-CH2-CH-O-H
         |
         CH2-......lipid

as united atoms:

H-O-C-C-O-H
      |
      C-......lipid

this is the warning I get :
_____________________________________________
Step 400, time 0.8 (ps)  LINCS WARNING
relative constraint deviation after LINCS:
max 0.120018 (between atoms 334 and 335) rms 0.002054
bonds that rotated more than 30 degrees:
 atom 1 atom 2  angle  previous, current, constraint length
    333    335   33.8    0.2081   0.2040      0.2000
    334    335   72.2    0.1297   0.1120      0.1000
   4311   4313   36.0    0.2086   0.1984      0.2000
   4312   4313   78.1    0.1089   0.1081      0.1000
   6147   6149   37.1    0.2184   0.1979      0.2000
   6148   6149   75.7    0.1187   0.1077      0.1000
____________________________________________________

335, 4313 and 6149 are hydrogen atoms attached to the -CH-O- moiety.
333  4311 and 6147 are the carbon atoms of the -CH-O-H and 
334  4312 and 6148 are the oxygen atoms of the -CH-O-H.

What would prevent this hydrogen atom from moving around so much? How
can I constrain it? Some earlier posts, mention something about the
pairs directive. Should I remove all the pairs involving that hydrogen?
I am a bit confused. 

(I had created the topology file using prodrg for this molecule and then
edited it )

Senthil