Re: langevin with waterbox ?

From: Jérôme Hénin (jerome.henin_at_uhp-nancy.fr)
Date: Thu Feb 19 2004 - 13:43:04 CST

> In fact, I am quite confused too. I'd like to know how to do NVT MD
> in NAMD except using temperature rescaling method? (for example, it seems
> that there is no Nose-Hoover method?)

No, you're right, Nose-Hoover is currently not implemented. I'heard it is
among the features they are willing to implement at some point, but if you
want to know when, only Jim can answer that. But this leaves several
different temperature control methods (apart from velocity rescaling) :
* velocity reassignment, wich does not change the sampled thermodynamic
ensemble, but perturbs the dynamics very strongly. This is the farthest you
can be from Newtonian dynamics.
* Benrendsen-like temperature coupling : coupling to a heat bath
* Langevin dynamics

> Also, please correct me if I am
> wrong that if there is explicit solvent in simulation box, what would
> lagevin dynamics mean? (I think lagevin dynamics has already taken into
> accout the
> interaction of solvent with solute, so if we use explicit solvent, we do
> not use lagevin dynamics.)

Initially Langevin dynamics was meant to model the effect of a solvent on a
colloidal particle. But in NAMD, they use it in a very different context.
Every atom (including explicit solvent) is applied a Langevin-like equation of
motion, with the forces from the force field + a frictional term + a
stochastic term. The Langevin force has no direct physical meaning here,
since it doesn't represent interactions with an implicit medium.
But it's an efficient way to add or remove energy to every atom and thus
regulate the temprerature. Furthermore, the trajectories it produces sample
from the canonical ensemble.
What people usually do with NAMD is use Langevin dynamics with a rather small
friction coefficient, to keep the temperature constant without affecting
Newtonian motion too much. That way, you get a trajectory that is both
realistic on short timescales and isothermal.
Besides, the algorithm itself is robust and simple to implement.

This whole Langevin idea is a bit weird at first, but with time, you get used
to it. And when practicing it, you see that it just does the job. Well, most
of the time.

Is this what you wanted to know?

Cheers,
Jerome

```--
In principio creauit Linus Linucem.
```

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:38:25 CST