[gmx-developers] NVE gromacs

[Berk Hess]

Hi,

This is really a question for gmx-users, not developers.
We should add a point on the wiki for this. But short answer:
Decrease verlet-buffer-tolerance to a vaue you find acceptable for the 
energy drift (you should be using cutoff-scheme=verlet)
Increase the constraint accuracy, assuming you are using constraints and 
LINCS (and double precision), increase lincs-iter until you have 
sufficiently good energy conservation.
Other settings should not matter, if not chosen strangly.

Cheers,

Berk

On 11/29/2016 02:24 AM, Michael K. Gilson wrote:
> Hello,
>
> My lab is trying to get good energy conservation with GROMACS in NVE, 
> but not doing very well at it.  We are running with double precision, 
> long cutoffs, short time-steps, etc.
>
> I was wondering if you have (or could direct us to) a sample input 
> file which you'd expect to give good results.  We'd also be happy to 
> give you more info on our system and what we've tried.
>
> All the best,
> Mike
>