Date: Thu, 20 Jan 2000 10:31:52 -0600 (CST)
From: darden
Subject: Re: ewald and cutoff

Currently the lennard-jones interactions in amber are handled via a
cutoff. In amber6 a long-range correction to the energies and virial are
added. Using a cutoff on lennard-jones interactions gives rise to small
errors in forces but larger errors in energy and virial. The long-range
correction (discussed e.g. in allen and tildesley) is pretty accurate. I
did put in an ewald approach to lennard-jones interactions at an earlier
stage, but it proved to be expensive due to the form of the combination
rules used by amber.

So, in amber the cutoff refers to a cutoff of lennard-jones interactions,
as well as the cutoff of direct-space ewald interactions. The ewald
reciprocal interactions, approximated by pme, correct for this
latter cutoff so that the total is an accurate Ewald sum for
electrostatics. (RMS force error about 5x10^-4)

hope this helps
Tom darden
On Mon, 17 Jan 2000, Frank Yan wrote:

> does ewald summation include lennard-jones interaction in amber? and when
> doing ewald summation, what is the meaning of cut? is it associated with the
> cutoff for l-j interaction only, if it's not included in the ewald
> summation. thanks, frank