Module distance 

Distance metrics for vector similarity search.

All Metric::distance functions return a value where smaller means closer, so the HNSW graph can order candidates with a single min-heap convention regardless of metric. Metric::score converts a raw distance back into a human-facing similarity where higher means more similar (used by the RAG layer and min_score cutoffs).

Enums

Metric

The distance metric used by a vector index.

Functions

dot

Dot product of two equal-length slices.

normalize

L2-normalizes a vector in place. A zero vector is left unchanged.

squared_l2

Squared Euclidean distance between two equal-length slices.