Re: NAMDENERGY: function of 'cutoff' parameter when -pme option is enabled

From: Peter Freddolino (petefred_at_ks.uiuc.edu)
Date: Tue May 06 2008 - 00:37:54 CDT

Ghalib,
The function of the cutoff parameter in PME is described in the NAMD
manual at http://www.ks.uiuc.edu/Research/namd/2.6/ug/node24.html. If
you use pme, the cutoff should have very little effect on your
calculated electrostatics unless you go to the very small (<7 A or so)
or the very large (on the order of your box size), as all it represents
is the point at which you switch between short range and long range
electrostatics. Using cutoffs as long as your box can break many things
in namd, so this is not a good test.
Also, please don't cross-post to vmd-l and namd-l. This has been
repeatedly mentioned on both lists.
Best,
Peter

Ghalib Bello wrote:
> Hi all,
> I am using NAMDENERGY to calculate the non-bonding interaction energy
> between a solute and the atoms surrounding it. However, I am only
> interested in the electrostatic component of the non-bonding
> interaction energy. Therefore, I am setting the "*-pme*" parameter on,
> so that NAMDENERGY will use PME to do the electrostatic interaction
> calculation. Because PME is not a cutoff-based technique, what is the
> function of the 'cutoff' parameter in this situation ? I was initially
> under the impression that, for PME-based electrostatic energy
> calculations in NAMDENERGY, the 'cutoff' parameter has no effect. In
> an attempt to confirm this assumption, I conducted a couple tests, but
> I got some rather odd results, which I describe below.
>
> My system is quite huge (94 X 94 X 180 cubic Angstroms), and the
> solute is at the center. So I conducted the following test: I selected
> all solvent atoms between 85 and 90 Angstroms away from the solute at
> the center of the box. I then calculated the PME-based electrostatic
> interaction energy between the solute atoms (*group1*) and the
> selected solvent atoms (between 85 and 90 Ang. [*group2*]). I did this
> calculation with the NAMDENERGY "cutoff" parameter set to 12 A. Then I
> repeated the calculation, but with the NAMDENERGY "cutoff" parameter
> now set to 90 A. I got very different results:
> Average (over dcd trajectory) electrostatic energy magnitude between
> *group1* and *group2* atoms when NAMDENERGY "cutoff" parameter is set
> to 12 A = 0.67 kcal/mol
> Average (over dcd trajectory) electrostatic energy magnitude between
> *group1* and *group2* atoms when NAMDENERGY "cutoff" parameter is set
> to 90 A = 31.027 kcal/mol
>
> Could anyone please explain the observed difference ? More
> importantly, exactly how, and to what extent, does the NAMDENERGY
> "cutoff" parameter influence PME calculations ?
>
> Thanks,
> Ghalib

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:49:27 CST