Module distance 

Distance metrics for vector similarity search.

Re-exports shared scalar implementations from nodedb-types. No SIMD intrinsics — works on all targets (native, WASM, iOS, Android).

Enums

DistanceMetric

Distance metric selection.

Functions

chebyshev

Chebyshev (L-infinity) distance: max absolute difference.

cosine_distance

Cosine distance: 1.0 - cosine_similarity(a, b).

distance

Compute distance using the specified metric (scalar dispatch).

hamming_f32

Hamming distance for f32 vectors (values > 0.5 treated as 1).

jaccard

Jaccard distance for f32 vectors (values > 0.5 treated as set membership).

l2_squared

Euclidean (L2) squared distance.

manhattan

Manhattan (L1) distance: sum of absolute differences.

neg_inner_product

Negative inner product (for max-inner-product search via min-heap).

pearson

Pearson distance: 1 - Pearson correlation coefficient.