AMBER Archive (2006)

Subject: Re: AMBER: How to set Bondi Radii using xleap

From: Carlos Simmerling (carlos_at_csb.sunysb.edu)
Date: Tue Aug 01 2006 - 12:43:04 CDT

I think that this is more a code issue than a GB one.
the calculation of effective radii has a step that divides by the
intrinsic radius, so as Dave suggests this code definitely won't work
unless you rewrite it. This will be a problem even if the effective
radius would end up non-zero (which might avoid the problem
mentioned by Don Bashford where you use 1/reff).

David A. Case wrote:

>On Tue, Aug 01, 2006, Ilyas Yildirim wrote:
>
>
>>Is there any way to set the bondii radii to zero or any value for a
>>specific atom type using xleap?
>>
>>
>
>No.
>
>
>
>>Another question is the following: When the bondii radii for a specific
>>type is set to zero, and igb=2 is used, I am getting NaN results in the
>>output file (when bondii radii is not zero - even when it is very close
>>to zero - the result is ok). I was wondering why this is creating a
>>singularity in the system when implicit solvent is used.
>>
>>
>
>You can't use a zero radius for any atom with a non-zero partial charge: that
>would create real problems with the electrostatic energies. Even for atoms
>with zero partial charge, using a zero dielectric radius is probably dangerous
>(as you have found), since the code was certainly never carefully checked for
>this possibility.
>
>...dac
>
>-----------------------------------------------------------------------
>The AMBER Mail Reflector
>To post, send mail to amber_at_scripps.edu
>To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu
>
>

-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu