[gmx-users] NVT and NVE

[Mark Abraham]

JMandumpal wrote:
> Dear gromacs users,
> 
> 
> I ran 100ps equilibraion, NVT, prior to running 60ps NVE to check the 
> energy conservation with different timesteps: 1.0fs, 1.5fs and 2.0fs, 
> but  these NVE simulations seem not to converge even after. Following is 
> the command I used for running 60ps NVE :
> ----------------------------------------------------------------------
> grompp_d -np $PBS_NCPUS -f mdout_NVE.mdp -p topol.top -c conf_100.gro -o 
> binary.tpr
> mpirun mdrun_mpi_d -s binary.tpr -o pr.trr -c output_1.gro -g topol.log
> -------------------------------------------------------------------------
> conf_100.gro is the equilibrated box, from my 100ps NVT, which I used 
> for the NVE's. Is this the right way of doing it? or Am I miss something 
> else in those commands.
> My system is 500 TIP5P box.

Well the pertinent parameters are all in the .mdp file, so we can't 
tell. Certainly you are not doing an exact restart 
(http://wiki.gromacs.org/index.php/Doing_Restarts), and so must be 
re-generating velocities, which removes some of the value of the NVT 
equilibration. Changing ensemble requires a re-equilibration period, but 
you should attempt to minimize the severity of the transition by doing 
the restart from the .trr file with velocities, and not a .gro file that 
(presumably) lacks velocities and has low-resolution coordinates.

Your second query about using tpbconv will not work - the tpbconv 
restarts preserve the ensemble.

Mark