From the AvocadoDB blog:
Cambridge mathematician Richard R. Parker presents a novel algorithm he has developed using a Hilbert curve and Polyhedrons to efficiently implement geo-indexing.
Basic premise is that points that are “near” on the line are also “near” on the Earth’s surface.
Interesting rhetoric but I think the “near” on the Earth’s surface is unnecessary.
More important to observe that a Hilbert curve when “straightened” and indexed, each point cuts across multiple dimensions of “nearness.”
Enables the isolation of “near” points in another representation, say global coordinates, quickly.
Points to consider/research:
- Basis for indexing/sharding a graph database? An particular n-dimensional Hilbert curve is used for indexing/sharding. Not all queries created equal.
- How do characteristics of the distances that compose the curve impact particular use cases?