Re: [AMBER-Developers] Please read: we need to look at pmemd.cuda pair list fidelity

From: Daniel Roe <>
Date: Tue, 16 Jan 2018 10:57:33 -0500


On Mon, Jan 15, 2018 at 10:11 PM, David Cerutti <> wrote:
> What's happening is that, for non-orthogonal unit cells (and the
> only case the tleap really knows how to construct is the truncated
> octahedron, translating it into a parallelepiped with all angles at
> 109.47), sander and pmemd do not know how to properly gauge the distance
> between the box faces, and therefore the number of hash cells to create
> when building their pair lists.

I don't think you are right about this. Sander (and cpptraj and I
think pmemd as well) determines the number of grid cells in each
dimension using the length of the *reciprocal* cell vectors, which is
the inverse of the fractional coordinate system. Because this
determination is done in reciprocal (fractional) space, it doesn't
matter if the unit cell is orthorhombic or not. All of the "gridding"
during the pair list build is also done in fractional space, so I
think the pair list builds are fine. Maybe I'm not understanding what
you're saying the problem is though. Do you have a specific test case
where the pair list is not properly set up (i.e. an atom that is
within the cutoff of another atom is skipped)?


Daniel R. Roe
Laboratory of Computational Biology
National Institutes of Health, NHLBI
5635 Fishers Ln, Rm T900
Rockville MD, 20852
AMBER-Developers mailing list
Received on Tue Jan 16 2018 - 08:00:02 PST
Custom Search