[gmx-users] energy drift in nve with conservative parameters

[Erik Lindahl]

On Tuesday, Mar 4, 2003, at 11:35 US/Pacific, Claudio J. Margulis wrote:

> Dear Erik, thanks for answering,
>
> there's no constraints in my simulation though ...
>
> >If you use PME the problem is probably the constraints combined with 
> >single precision ...
>
> Any other ideas?

Are you sure you are not using constraints? If you use SPC or TIP3P 
water, the topologies are always constrained by the SETTLE algorithm 
(since it is part of the definition of these waters, there aren't any 
real bond or angle parameters). For SPC you can force the water to be 
flexible by adding -DFLEX_SPC to the defines parameter in the mdp file, 
but since you'll have to use a smaller timestep without constraints it 
is probably better to do it in double.

Cheers,

Erik