clustering::index::RangeIndex Concept Reference

Contract for spatial indexes that can surface the radius-neighborhood adjacency over a borrowed point cloud in one call. More...

#include <clustering/index/range_query.h>

Concept definition

template<class Q, class T>
concept RangeIndex = std::constructible_from<Q, const NDArray<T, 2> &> &&
                     requires(const Q &q, T radius, std::size_t minPts, math::Pool pool) {
                       { q.query(radius, minPts, pool) } -> std::same_as<CoreAdjacency>;
                     }

Detailed Description

Contract for spatial indexes that can surface the radius-neighborhood adjacency over a borrowed point cloud in one call.

DBSCAN reduces core-point detection and cluster expansion to neighbor lookups on the radius graph; any satisfying backend is free to build that graph however its geometry favours (tree walk for low dim, blocked pairwise for high dim). Carrying the core threshold into the query lets a backend skip materializing edge halves no consumer reads.

Template Parameters

Q	Candidate index type.
T	Element type of the point cloud.

Definition at line 53 of file range_query.h.