[gmx-users] Pressure coupling and tau_p values

[chris.neale at utoronto.ca]

> For example, a 3ns simulation using a tau_p of 0.5 I get
> a surface tension of 30 dynes/cm and using a tau_p of 5.0 I get a surface
> tension of 25 dynes/cm.

First thing that comes to mind is that your area per lipid is not at  
equilibrium. Assuming that your bilayer normal lies in the Z  
dimension, plot your xy area vs. simulation time. My bet is that it is  
not converged and that the run using a tau_p of 0.5 is approaching  
convergence faster than the run using a tau_p of 5.0. Once it is  
converged, I would recommend something more like 20ns for a production  
run.

> Is this difference negligible?

Test that out yourself. Do some repeats, or perhaps take the 3ns  
snapshot from the simulation at tau_p of 0.5 and continue it using a  
tau_p of 5.0. Also take the 3ns snapshot from the simulation at tau_p  
of 5.0 and continue it using a tau_p of 0.5.

> Therefore, is it more reliable when simulating a lipid
> bilayer system to use a short tau_p ~0.5ps or a longer tau_p ~ 5.0ps?

Sorry, I don't know the answer to that one.

Chris.