[gmx-users] replica exchange

[X.Periole]

On Thu, 12 Jan 2006 16:17:51 -0700
  "Andrea C. Vaiana" <vaiana at lanl.gov> wrote:
> Hi everyone,
> Mine are still test runs so I still don't have an 
>optimal temperature  distribution... so I still have to 
>work a bit to rule out problems  due to this.
> 
> 
> 
> 
> 
>>> this is due to the different density, you will see that 
>>>at high  
>>> temperature fewer inteeractions are calculated.
>> should this increase the speed ??
> 
> Actually I think it should... instead, my 2 replica run 
>(~300 and 310  K) is faster than my 20 replica one (~300 
>to ~510 K) by 20%. So I  think it is not due to the fewer 
>interactions... Could it be that at  higher temperatures 
>(lower density) the ewald summation gets slower  because 
>of a suboptimal distribution of particles in real and 
> reciprocal space?

you could run 2 replicas at hight temperatures to see if
it a temperature effect ...

XAvier