Module batch_distance 

Batch distance computation for HNSW neighbor selection.

Instead of computing distances one-at-a-time in a loop, collects candidate vectors and computes distances in bulk. This improves cache utilization and enables SIMD-friendly memory access patterns.

Used by select_neighbors_heuristic in build.rs to accelerate the diversity check during HNSW graph construction.

Functions

batch_distances

Compute distances from a query vector to multiple candidate vectors.

is_diverse_batched

Precompute all pairwise distances between selected neighbors and a candidate.