[gmx-users] Unable to swith from NVT to NPT

[Mark Abraham]

Stéphane Téletchéa wrote:
> I'm probalby doing something wrong but since neither the manual nor the 
> mailing list has driven me to the correct answer, i'm asking you the 
> problem. Thanks in advance for you answers and time.
> 
> I want to simulate my system with short NVT (for temperature 
> equilibration) and then NPT since i've observed that full NPT leads to a 
> delay in the temperature scaling (about 100ps is lost during this stage, 
> not that big, but i think this is because i'm not doing things correctly).
> 
> To set up the NVT calculation, i've just set pcoupl to no (as i think is 
> required), but *let* other pressure related parameters the same (i 
> assumed if pcoupl is off, pressure parameters are not taken into account).
> 
> I did 20 ps with my *supposed* NVT :
> 
> #####
> ; Pression, semiisotropic recommended for membrane simulations
> Pcoupl                   = no
> Pcoupltype               = semiisotropic
> tau_p                    = 10 10        ; see below for recommandations
> ref_p                    = 1.0 1.0
> compressibility          = 4.5e-5 0
> #####
> 
> I did the next NPT step for 10ps with parameters :
> #####
> ; Pression, semiisotropic recommended for membrane simulations
> Pcoupl                   = Berendsen
> Pcoupltype               = semiisotropic
> tau_p                    = 10 10        ; see below for recommandations
> ref_p                    = 1.0 1.0
> compressibility          = 4.5e-5 0
> #####
> 
> 
> Each run was done with gromp for generating the tpr + mdrun (for 4 nodes).
> 
> grompp->mdrun (4nodes) -> NVT, then
> grompp (with NPT)
> 
> The problem is that running the second grompp returns me :
> 
> ####
> Opened ../pr1/fullmd_ions.edr as single precision energy file
> Reading frame     40 time   20.000
> -------------------------------------------------------
> Program grompp_lam, VERSION 3.3.1
> Source code file: enxio.c, line: 364
> 
> Fatal error:
> Could not find energy term named 'Pcoupl-Mu-XX'
> -------------------------------------------------------

Using grompp here and supplying an energy file leads it to expect 
pressure coupling terms, since the only purpose of supplying energy 
files is for preservation of pressure coupling continuity (or 
Nose-Hoover temperature coupling, IIRC). Since you wouldn't care about 
preserving a temperature coupling parameter when you're changing 
ensembles, grompp assumes you knew what you were doing when you chose to 
give it an energy file, and that you gave it an erroneous file. Just 
don't give it an energy file and you're fine.

Thanks for the complete problem description. You assumed correctly about 
pcoupl=no, by the way.

Mark