Re: AMBER: end point contribution in MCTI

From: David A. Case <case.scripps.edu>
Date: Fri, 5 Nov 2004 08:36:31 -0800

On Fri, Nov 05, 2004, Cai, Yufeng wrote:

> I read a paper of doing MCTI using charmm package. In which they run
> MD on the separate lambda value: 0.02 0.1 0.2...0.9 0.98, and use
> trapezoidal rule to calculate the energy between lambda 0.02 and 0.98.
> Then calculate the end point contribution separately.
> In sander MCTI, a table of lambda value and weight is provided in
> the manual. The weights add up to 1. And I don't understand how people
> get these lambda values and weights. Are we assuming that the end point
> energy will be smooth in sander?

The points and weights in the Amber manual are just for Gaussian quadrature
of a (smooth) function. Check your math books, or google for "Gaussian
quadrature" for information.

If you expect a non-smooth integrand, then you should either think about
setting klambda in sander and/or using a different quadrature scheme. It
never hurts to investigage end-point effects carefully.

....dac

-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber.scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
Received on Fri Nov 05 2004 - 16:53:00 PST
Custom Search