Re: [AMBER] about dielectric constant..

From: Robert Duke <>
Date: Wed, 28 Jul 2010 09:09:59 -0400

Hello Wookyung,

About the message:
"ERROR : Non-default values of intdiel are not supported."

This is pmemd-specific. The story here, as I can best recollect, is that
when support for modifying intdiel was added to sander, the same thing was
not done in pmemd because it was new functionality that was missed. In
addition, upon reflection, we decided that the ability to change intdiel was
mostly there for folks researching GB methods, not for production use by
users. Also, introducing this capability into pmemd was likely to have some
performance impacts, perhaps not huge, but some. So we decided to just
block this capability in pmemd. I don't know if the differing functionality
is what you find really confusing, but that's the explanation.
Regards - Bob Duke

----- Original Message -----
From: "case" <>
To: "AMBER Mailing List" <>
Sent: Wednesday, July 28, 2010 8:32 AM
Subject: Re: [AMBER] about dielectric constant..

> On Wed, Jul 28, 2010, Wookyung Yu wrote:
>> 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. "
> It means more or less exactly what it says: use of non-default values of
> intdiel is a research topic, and you should expect to have to study the
> options and results carefully.
>> As I know, in real protein, dielectric constant is not 1 .
> But many of the motions that contribute to the "real" dielectric constant
> are being explcitly modeled in the GB/MD simulation. Furthermore, the
> charge model was designed for dielectric 1, and things like H-bond
> energies can be wrong if you use charges designed for epsilon=1 in some
> other model.
> Of course, people are encouraged to run test caclulations and learn new
> things
> about force fields and solvation models. Just be aware that you would be
> breaking some new ground here.
>> 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."
> Means what it says; I'm not sure how else to word it to make it more
> clear.
> ...dac
> _______________________________________________
> AMBER mailing list

AMBER mailing list
Received on Wed Jul 28 2010 - 06:30:04 PDT
Custom Search