[gmx-users] Dielectric constant calculation with Gromacs-3.3.1

[David van der Spoel]

On 2012-03-09 19:47, Liang Shi wrote:
> Hi~
>
> I am going to calculate the dielectric constant of rigid water model,
> but I am not quite sure about the use of boundary condition. I know
> there is an option "epsilon_surface", and its default number is 0. I am
> using PME to calculate electrostatic interaction. From the literature
> and the mailing list, it seems there are 2 following options:
>
> 1. set epsilon_surface to be its default value 0, and use g_dipoles to
> calculate the dielectric constant.
>
0 means inifinite, also in e.g. the epsilon_RF in your mdp file.
This is what you should use if your simulation was done with PME.

> 2. set epsilon_surface to be inf (someone calls this conducting boundary
> condition), and use g_dipoles to calculate the dielectric constant. Here
> I am not sure which option I should use. epsilon_RF seems just to work
> with reaction field method, not PME. But I don't see any option for PME.
>
> And I am not sure what formula is used for the dielectric constant in
> Gromacs:
> Let a = (4*pi*<M^2>) / (3*k*T*V) in Gaussian unit assuming <M>=0
>
> For case 1: it seems to be: epsilon = 1/(1-a)
>
> For case 2: it seems to be: epsilon = 1+a
>
> And finally I think these two cases (epsilon_surface in MD and
> corresponding formula) should give very close answer. But I am not sure
> which way is better. Could any one please help me sort this out? And I
> am using Gromacs-3.3.1.
>
> Thanks,
> Liang
>
>


-- 
David van der Spoel, Ph.D., Professor of Biology
Dept. of Cell & Molec. Biol., Uppsala University.
Box 596, 75124 Uppsala, Sweden. Phone:	+46184714205.
spoel at xray.bmc.uu.se    http://folding.bmc.uu.se