ring statistics #284
Description
While testing another library for graphs, I used matscipy.rings.ring_statistics to get some reference values to compare to. When I then used a C60 Buckminster fullerene as test structure, I noticed that matscipy gives some very large (incorrect) rings in addition to the correct ones:
matscipy.rings.ring_statistics(at, 1.85, 30)
gives
[ 0 0 0 0 0 12 20 0 0 0 0 0 0 0 0 0 0 0 10]
If one is just interested in the total numbers of small rings, that won't matter. However, if one computed the fractions of ring sizes, those would change by a lot in this case! (E.g., 5-rings would go from 37.5% to 28.6%)
(I just tested this on a few C60 isomers - they show the same behavior, but the exact size and number of phantom rings depends on the structure.)