[gmx-users] RE: Temperature Fluctuations

[Mark Abraham]

On 18/11/2010 3:28 AM, Nimesh Jain wrote:
> Hi,
>
> I am doing Replica Exchange (RE) simulations with 8 replicas 10 degree 
> apart. My system has implicit solvent and there is only one molecule 
> in the system (around 270 atoms). I am using Nose-Hover thermostat and 
> tabulated potentials.
> When I do simulations with exchange frequency of 1 ns, the system 
> blows up and the temperature goes to a few 10,000K. When I remove RE, 
> the temperature fluctuations are +/- 20 degrees and they increase with 
> time.
> Does anyone know what the problem is ? Let me know if you need any 
> other info about the system I am simulating.

If one of your simulations is not stable on its own, then you've not 
equilibrated under your thermostat properly. What was your preparation 
protocol and final .mdp?

Mark