[gmx-users] Re: gmx-users Digest, Vol 66, Issue 85

[Mark Abraham]

Pan Wu wrote:
> Thanks Dr. Dallas Warren. I have plotted the picture of pressure with 
> respect to simulation time, it is not fluctuation, it 
> is monotonically decrease from about 500 to 3 bar, this is why I am 
> confused.

Actual monotonicity would be wildly unlikely unless nstenergy is fairly 
large. In an equilibrated NPT system it would be unremarkable to have 
the sign of dP/dt change at every step - but nobody would care anyway.

> If it really should take some time to converge the required pressure, my 
> MD simulation is 4 ns, during the whole simulation time, the pressure 
> keeps on decrease. (Will this pressure convergence take that long time?)

That does seem rather too long for normal equilibration - but if you 
really are changing volume dramatically (and somehow stably!) then it 
could take a while!

> The initial box size is 12 ns *12 ns *12 ns, after the simulation, it 
> turned out to be 52 * 52 *52 .

If you mean 12nm, then that contradicts the below:

>        Here is how I set up the system,
> 
>        1. I solvate the proteini use spc216.gro, the log file shows:
> 
>                     Output configuration contains 40990 atoms in 12386
>     residues
> 
>                     Volume: 412.593 (nm ^3)
> 
>                     Density: 1015.79 g/L
>                     Number of SOL molecules: 13410

If your initial setup was at all reasonable, you won't see a four-fold 
linear expansion of the box dimensions. It seems rather more likely that 
you'd have seen such a simulation blow up. I expect you've observed 
something wrongly. Perhaps you should copy and paste some output text 
that you are using so you/we can work out your error.

Mark