AMBER Archive (1999)

Subject: Re: ANAL Questions!

From: Yongxing Liu (yliu_at_mail.wesleyan.edu)
Date: Thu Jul 15 1999 - 09:17:27 CDT

Hi,

Some time ago, for my specific system(a DNA decamer) in a box(~45x45x60 A),
I did an experiment to compare the Amber PME energy and ANAL energy with
cutoff. I added some code in sander to print energy components and then
calculate the energy components using ANAL with cutoff changing from 5A to
40A. I found that at 18-22 A level, these two energies become almost the
same. Then I use 30A cutoff in my later ANAL calculations.

Yongxing Liu
 

At 07:20 PM 7/14/99 -0700, you wrote:
> 1. ANAL does not have PME feature incorporated into it, so how is
> long-range electrostatic interactions handled? Will it be ignored if the
> distance between 2 charges greater than the cutoff distanceb (using the
> pairlist)?
>
>Yes.
>
> If so, How big is the cutoff distance sufficient in order to
> achieve acceptable accuracy (e.g. Rcut = 15.0 (A)?)
>
>The cutoff should maybe be less than the shortest box dimension,
>and perhaps for solute systems that remain integral within the
>box, the box should be ignored and iftres=0 (all solute-solute
>interactions used).
>
>What does "acceptable accuracy" mean? If you are trying to analyze
>PME potentials with a pairwise-only method, you have to consider
>that you are comparing a local pairwise interaction (non-PME) to
>an infinite lattice of pairwise interaction. The good news is that,
>unless you are really interested in a crystalline situation, the
>pairwise analysis may be more like reality. The effect of say 8A
>vs 10A cutoff can be determined in this situation by doing both
>calcs and comparing energies. In general, I believe that the
>consensus is that the contribution beyond 9-10A is small
>energetically but important for getting a trajectory right on a
>very long time scale.
>
> 2. ANAL contains bug, as shown on AMBER webpage, that is not consistant
> with SANDER in computing energy. Can anyone tell how serious it is? I'd
> like to use ANAL to compare energies of different sub-systems. Can
> anyone give me your opinions on this?
>
>I'm looking forward to opinions here myself. :-)
>
>Bill Ross
>
>