[gmx-users] Changing tau_p for parrinello-rahman made my simulation stable

[Hermann, Johannes]

Dear all,

I had a hard time making my simulation stable. Perhaps someone has an 
explanation, perhaps it (just) saves someone a lot of time. However, I 
want to share my experience:

I have a dense simulation box (80% water, 20% protein). Free simulation, 
no restrains, constrains on the H-bonds. After some time (around 15ns) I 
always get LINCS warnings, which lead the simulation to stop either 
because of a "Segmentation fault (11)" or "1 particles communicated to 
PME rank 0 are more than 2/3 times the cut-off..". I looked at the 
trajectories but could not identify any strange behavior.

I tried a lot of things, but what made my simulations eventually stable 
was to increase tau_p for parrinello-rahman from 1.0 to 2.0. Can it be 
guessed whether this value is okay, or will it produce un-physical behavior?

I know I did not provide a lot of specifics about my setup and run 
parameters, I can happily provide that if someone wants to dig deeper that.

All the best

Johannes