Re: [AMBER-Developers] default for gamma_ln?

From: Alexey Onufriev <alexey.vt.edu>
Date: Wed, 11 Mar 2015 00:57:35 -0400

Ok.

For AMBER GB, we found that the lower the better in the sense of faster
sampling. We have tested gamma=0.01 quite extensively, will add a
citation. I don't know what's best for explicit solvent, so will keep the
original recommendation with a clear distinction. best, Alexey

On Tue, Mar 10, 2015 at 7:54 PM, Jason Swails <jason.swails.gmail.com>
wrote:

> On Tue, Mar 10, 2015 at 6:54 PM, Alexey Onufriev <alexey.vt.edu> wrote:
>
> > Hi, All.
> >
> > I am wondering what was the original logic for setting the default value
> of
> > zero
> > for gamma_ln (Langevin thermostat)? It duplicates ntt=0, which seems a
> more
> > logical choice if one wants constant energy. For gamma_ln default we
> can
> > use e.g. 0.01 ps^-1, which we would recommend in practice. Thoughts?
> >
>
> ‚ÄčI have no strong feelings either way. I always think of Langevin not as a
> thermostat, but rather as an integrator for a different set of equations of
> motion (one with a stochastic element).
>
> It just so happens that when the collision frequency is 0, those EOMs
> become identical to standard Newton's equations. Perhaps that was the
> rationale? (Much like the default behavior for the standard integrator is
> to *not* have temperature control)
>
> All the best,
> Jason
>
> --
> Jason M. Swails
> BioMaPS,
> Rutgers University
> Postdoctoral Researcher
> _______________________________________________
> AMBER-Developers mailing list
> AMBER-Developers.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber-developers
>
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Tue Mar 10 2015 - 22:00:03 PDT
Custom Search