Isaac

First time seeing these categories/categorize in action, so I am bit lost. Just let me mention that for the hexagons example I made earlier, one should be able to have...

Just to elaborate a bit more on the ideas we briefly discussed yesterday in Discord. If fragment categories were possible, one could easily do the atom classification that @pfebrer was...

[hNPG_str.fdf.txt](https://github.com/zerothi/sisl/files/15060638/hNPG_str.fdf.txt) This certainly looks like a potential approach. However, I've found some issues when testing it with a *hybrid-NPG* structure, like the one I am attaching (these issues may also...

I find some issues though... First, as you may see with the NPG example, one also gets points of "rings" that are not really aryl rings - the nanopore in...

> And it would actually separate the layers by using my method described above. So this might be an even simpler method to separate layers. E.g. a tilted bi-layer, a...

> In any case, handling periodicity is not trivial, no? Would you expand to the auxiliary cell to run the algorithm? This is actually a very good point which, by...

Well, the GNR separation in NPGs would be one usage for `Rings` - but there are many other cases where having a list of `Rings` becomes quite handy, like working...

> I think handling periodicity in the graph case is also not trivial. But maybe it is simpler, I don't know :) But inter-cell bonds will be present in the...

From what I see in the example by Pol, I understand that the implementation you are thinking about would require each atom to be uniquely associated to one category, right?

Just one question for me to follow... What is the difference between classifier and categorize?