[gmx-users] What becomes of kinetic energy in local minima?

[Mark Abraham]

yutong.zhao at utoronto.ca wrote:
> Hi Gmx,
> 
> 
> I often hear the notion that "simulations" get stuck in local minima 
> (for potential energy I assume), how come? What happens to the "kinetic 
> energy" that is supposed to kick the system out of local minima? (ie. 
> when I roll a marble down a dip it tends to come out so long as there's 
> no friction. If the barrier is too high then wouldn't the system would 
> sort of "oscillate" in the well? 

Yes, that's exactly what can happen. Dynamical simulations can get 
trapped in a well if its depth is a suitable multiple of kT.

> How does GMX treat KE loss via 
> micromolecular friction?

There isn't any. MDS integrate Newton's laws. Under constant-energy 
conditions (NVE), energy can only convert from form to form. Under 
constant-temperature conditions (NVT, NPT), velocities are manipulated 
to keep the average T near a value.

Mark