Extend the methodology to periodic systems

[tfrederiksen]

Currently only molecules can be analyzed. Extension to systems with periodicity in 1D or 2D could be very powerful.

[brandimarte]

Need to include k-point sampling for the Huckel bond order analysis.

[brandimarte]

Should we consider this issue as done with c5410c4?

[tfrederiksen]

Can we do graphene? If not I think we shoud try to fix this first before closing.

[tfrederiksen]

Added some graphene xyz-files (varying real-space repetitions) to examples/geometries but the results obtained with get_all_kekule_periodic.py look incorrect: the bond orders vary over the structure, but all bonds should be strictly identical.

[tfrederiksen]

I think the remaining issue is related to the fact that the code currently only samples over the coupling to 4 neighboring cells indexed by (-1, 0), (1, 0), (0, -1), (0, 1) and not all 8 adjacent cells in 2D. In other words, the coupling to cells (-1, -1), (-1, 1), (1, -1) (1, 1) are missing.

[brandimarte]

Yep, I know, the implementation was just to have it working for 1D systems periodic in x or y directions, but it should be easily generalized. I'll try to make it!

[brandimarte]

I've include the summation over all neighboring cells in 4e929e7, but there is something with the weights in the reciprocal space. For the graphene_rep1x2.xyz example it calculates BOs higher than 2 (even > 4).

[brandimarte]

At de699d6 I've tried to fix the mistakes @tfrederiksen pointed on comment, however the HBO is still wrong for graphene_rep1x2 example.