[gmx-users] Re: RE: heat appears

[Nguyen Hoang Phuong]

> Spiwok Vojtech wrote:
> >>I use the twin-range cutoff 0.8/1.2 and the temperature of the system
> >>consisting of 500 H2O increased up to 310K while the reference temperature
> >>is 300K.
> >
> >
> > Use a single cut-off of 1.2 (or more) and the fever will be cured !
>
> That doesn't seem like a good solution. If you simulate only water (nothing
> in the water, like protein, ions or anything), you should be very close to
> your ref_t.
>
> What time constant do you have for temp-coupling (tau_t)?
> What do you have for nstlist (neighborlist update frequency)?

tau_t=0.1,rlist=10

>
> Instead of going to single cut-off I'd suggest using PME for the long-range
> electrostatics. Check the manual and/or mail archives how to do it.

that is true, but just would like to understand the problem.

>
> > It was just guess. I don't know any sophisticated explanation.
>
> The potential is discontinuous at the cut-off. This generates artificial
> forces that accelerate your atoms, hence an increase in temperature.

but the artificial force can slow down the atoms --->decrease in
temperature.

Phuong