Nicely said Saeed. Just adding that you can also integrate the velocity
autocorrelation function (only in unthermostatted systems) to obtain the
diffusion coefficient (Green-Kubo relationship). Both methods (Einstein
and Green-Kubo) are exact (you still will have sampling error) and provide
a good check that the calculation is being done correctly.
the derivation of the Green-Kubo relationship for diffusion also, for me,
provides the most intuitive explanation as to why the diffusion coefficient
is affected by thermostats.
Best,
Tom
On Wed, Dec 19, 2018 at 3:52 PM Saeed Izadi <izadi.vt.edu> wrote:
> Hi Lucas,
>
> What type of ensemble are you using for the calculation? .. note that
> unlike many other properties of water which are calculated in NPT ensemble,
> self-diffusion coefficient is safer to be estimated with no thermostat
> (i.e., NVE ensemble). The use of a thermostat affects how the water
> molecules move, usually leading to under estimation of self-diffusion
> value. To make sure the system’s temperature does not deviate from the
> target temperature, it is suggested to run a large number of short (NVE)
> simulations starting from well- equilibrated evenly spaced NPT-sampled
> coordinates and velocities as initial conditions. The Einstein relation can
> then be used to infer the diffusion constant from the time dependence of
> the mean-squared displacement, and then final value of the self-diffusion
> coefficient can be obtained by averaging over all the NVE trajectories. You
> can also take a look at the following paper for more information:
> https://onlinelibrary.wiley.com/doi/full/10.1002/wcms.1347
>
> best,
> Saeed
>
> On Wed, Dec 19, 2018 at 9:14 AM Lucas Bandeira <bandeiralucas97.gmail.com>
> wrote:
>
> > Hello,
> >
> > I'm trying to calculate the self coefficient diffusion of water using the
> > FF TIP3P. I know that to TIP3P the value of the diffusion coefficient of
> > water is different than the obtained experimentally, 5.19 * 10⁻⁹ m²*s⁻¹
> > instead of 2.3 * 10⁻⁹ m² * s⁻¹, but when I try to calculated the
> diffusion
> > coefficient I get the value 4.21 * 10⁻⁹ m² * s⁻¹. I'm using the formula
> on
> > AMBER Manual:
> >
> > [image: image.png]
> >
> > Using n = 3. The MSD value I compute using the out_r.xmgr generated when
> I
> > use cpptraj.
> > My input looks like:
> >
> > diffusion :WAT.O 0.2 out
> >
> > I'm using Amber14 so I can't calculate diffusion coefficient directly.
> >
> > Thank you in advance,
> >
> > Lucas
> > _______________________________________________
> > AMBER mailing list
> > AMBER.ambermd.org
> > http://lists.ambermd.org/mailman/listinfo/amber
> >
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
--
************************************************
Tom Kurtzman, Ph.D.
Associate Professor
Department of Chemistry
Lehman College, CUNY
250 Bedford Park Blvd. West
Bronx, New York 10468
718-960-8832
http://www.lehman.edu/faculty/tkurtzman/
<http://www.lehman.edu/faculty/tkurtzman/index.html>
************************************************
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Dec 20 2018 - 05:30:02 PST
