Interpolation in absolute units

[ddahlbom]

When retrieving intensities from classical structure factors, rounding to nearest available wave vector currently takes place in "RLU space." This is in some sense incorrect, and the rounding should instead be done in the global coordinate system.

Probably the most appropriate way to "round" would be to perform a Voronoi decomposition of the sample points as they are distributed in the global frame. Since we are dealing with a regular lattice, this decomposition can probably be precomputed in some way (rather than performing a local search to make a local Voronoi decomposition).

The issue is then: perform this "Voronoi rounding" in a fast and general way or implement an appropriate heuristic that approximates this type of rounding.

EDIT: Additionally, even with the Voronoi decomposition, we'll frequently encounter edge cases which should be dealt with in a satisfactory way. For example, in the attached image, the choice of wave vector is ambiguous when the path (red line) runs directly on top of the cell boundary for some finite length.