On Sun, May 9, 2010 at 10:10 AM, case <case.biomaps.rutgers.edu> wrote:
> On Sat, May 08, 2010, Jason Swails wrote:
> >
> > DV/DL is just calculated as a difference between the energies computed
> with
> > both hamiltonians...
>
> This is correct.
>
> >.... (so the closer together the lambdas are, the more the
> > slope approaches a true derivative).
>
> This is not correct: the DV/DL values are always true derivatives, and have
> nothing to do with how closely spaced the lambdas are. The code doesn't
> know
> or care if there are (in fact) and "other lambdas".
>
Ah yes, so it is. Page 96 of amber11 manual: "This capability in sander is
implemented by calling the force() routine twice, once for Vo, once for
V1..." The lambdas in both mdins are the same (as are the energy terms
posted in each mdout file and the coordinates printed to each restart/mdcrd
in my experience with TI). Having many lambdas closely spaced improves the
quality of the numerical integration, not the derivatives (a significant
difference). Sorry for the mistake.
> Concerning other parts of this thread: the constant pH implementation uses
> the "dummy atom" approach for H atoms that disappear, not the "softcore"
> approach with different numbers of atoms at the two endpoints. I'm not
> sure
> which approach the original poster is using, and which terms are present or
> absent depends on that choice.
>
As I recall, it was about the pKa tutorial on the amber website which uses
the dummy-atom approach, not softcore, and why is it OK to include the
bond/angle/dihedral terms involving the "disappearing" hydrogen.
All the best,
Jason
--
Jason M. Swails
Quantum Theory Project,
University of Florida
Ph.D. Graduate Student
352-392-4032
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Sun May 09 2010 - 22:00:03 PDT
