When computing neighbor lists of ideal crystals with all periodic boundaries, the average coordination number is occasionally slightly less than the expected value.
Example: 3x3x3 supercell of the HCP conventional cell gives mean coord as 11.9629... rather than 12.
The missing neighbor pairings seem to be related to binning near the c and a box edges.
Computing dmag of the trouble atoms versus all other atoms shows the correct number of neighbors below the cutoff.
Applying an arbitrary shift of ~1 Angstrom to atomic positions in either the x or z direction also results in the correct coordination.
When computing neighbor lists of ideal crystals with all periodic boundaries, the average coordination number is occasionally slightly less than the expected value.
Example: 3x3x3 supercell of the HCP conventional cell gives mean coord as 11.9629... rather than 12.
The missing neighbor pairings seem to be related to binning near the c and a box edges.