Re: [AMBER] about dielectric constant..

From: Jason Swails <>
Date: Wed, 28 Jul 2010 07:59:16 -0400


My comments are below:

On Tue, Jul 27, 2010 at 11:18 PM, Wookyung Yu <> wrote:

> Hello!
> I want to do MD simulation on differenect dielectric constant. with GB and
> pmemd.cuda.
> But Manual said
> intdiel
> Sets the interior dielectric constant of the molecule of interest.
> Default is 1.0. Othervalues have not been extensively tested.
> What's the meaning of " Othervalues have not been extensively tested. "

The model has been parameterized to reproduce experimental data for
solvation energies, etc. See the original papers cited in the Amber manual
for a more detailed description of the development of the methods.

> If I do MD simulation using another dielectric constant,
> then It's incorrect result?

It's not necessarily incorrect, but it will have to be validated in the same
way that the original parameters were validated.

> As I know, in real protein, dielectric constant is not 1 .

I've seen values of 2 to 4 to account for electronic polarizability.
However, GB was parameterized with indiel=1 and extdiel=80 (or 78.5, I don't
remember which), so unless you plan to redo the validation using different
values, you should probably use the defaults. Changing intdiel is probably
more appropriate for PB calculations since that's a more formally exact
theory without the parameterization inherent in GB.

> And another question is
> If I set "intdiel" parameter in pmemd.cuda.
> the error message is occurred.
> "ERROR : Non-default values of intdiel are not supported."
> If I use pmemd.cuda , It's impossible to use another dielectric constant?

That's what the error message says, so yes.

Hope this helps,

> Thanks in advance.
> _______________________________________________
> AMBER mailing list

Jason M. Swails
Quantum Theory Project,
University of Florida
Ph.D. Graduate Student
AMBER mailing list
Received on Wed Jul 28 2010 - 05:30:03 PDT
Custom Search