[gmx-developers] velocity constraining patch

[Berk Hess]

Rafael R. Pappalardo wrote:

>Dear Prof. Hess,
>I would like to know if the change mentioned on Oct. 16 about constraining the 
>velocities using Langrange multipliers is related with the paper:
>"A common, avoidable source of error in molecular dynamics integrators"
>J. Chem. Phys. 126, 046101 (2007).
>  
>
Yes, it is.
I implemented this new method after getting a hint from someone at the 
D.E. Shaw group.
I actually figured out the change myself (which is quite simple).
I read the paper after implementing the change and have had several 
discussions
with the authors of the paper.

The change is significant if you want very accurate energy conservation
in NVE simulations.
But the effect is completely negligible for NVT simulations.

Berk.