[ase-users] NeighborList radii

Peterson, Andrew andrew_peterson at brown.edu
Wed May 7 18:04:46 CEST 2014


On Wed, May 7, 2014 at 5:29 AM, Jens Jørgen Mortensen <jensj at fysik.dtu.dk>wrote:

> You give each atom a radius that will define a sphere for each atom.  Then
> neighbors of an atom will be those spheres that overlap with the sphere of
> that atom.  Maybe we should rename the "cutoffs" argument to "radii" -
> would that help?
>

I guess there are alternate ways to define it:

1. Each atom has a cutoff sphere, and if any other atom is within that
sphere it is counted. (This is the behavior I've seen in neural network
potentials, and what I was expecting.)

2. Each atom has a cutoff sphere, and if any part of the cutoff sphere of
another atom is within the first's cutoff sphere, they are counted as
neighbors. I can qualitatively guess why this would be desired for
electronic structure calculations (GPAW and EMT).

Probably the simplest is to just update the documentation to clarify this.
I'll take care of that at some point.

thanks!
Andy
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.fysik.dtu.dk/pipermail/ase-users/attachments/20140507/595a169e/attachment.html>


More information about the ase-users mailing list