A KD-Tree is a data structure for searching N-dimensional space in log(N) time by splitting a data set iteratively across, eg for 3 dimensions, dimensions x,y,z,x,y,z
In KDTREE-JTK, data are stored in block arrays instead of linked structures, for minimal (usually zero) consing and allocation overheads.
Balancing is done in-place with Wirth's median partition method, so it should be fast and memory efficient.
By default, all floats are double precision, but this can be changed in kdtree-jk-structs.lisp
Topic: data structure