[gmx-users] TUBE - Lincs or bonds?

[Steven Neumann]

Dear Gmx Users,

I have a tube made of atoms connected in lines:

Picture here: http://speedy.sh/BTRHx/tube.png

The edges of the tube are on the box edges with pbc - tube is infinite in
lenght then. I wish this tube to move only in XY directions so I froze the
Z coordinate using freezedim and freezgrps. I do not have any pressure in
my simulations - NVT ensemble.

I wish that tube to be rigid and move together with all the atoms in XY
directions. I used LINCS between neighboring atoms as well as oppositely
placed within each circle (diameter lenght).

I get lots of LINCS errors because of deviations e.g.:

Step 147790, time 1182.32 (ps)  LINCS WARNING
relative constraint deviation after LINCS:
rms 0.438139, max 0.889697 (between atoms 48 and 49)
bonds that rotated more than 30 degrees:
 atom 1 atom 2  angle  previous, current, constraint length
     40     41   79.3    0.1432   0.1715      0.3450

And the simulation crashes, although the trajectory looks ok.

Would you suggest something else please or how to overcome those errors?

Please, advise.

Steven