superkmeans-rs 0.1.0

A Rust port of SuperKMeans: fast k-means clustering for high-dimensional vector embeddings
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50

//! Rust port of SuperKMeans — a fast k-means clustering library for
//! high-dimensional vector embeddings using BLAS+ADSampling pruning.
//!
//! The C++ original lives in `SuperKMeans/`; this crate re-implements the
//! C++ public API (no Python surface) in pure Rust with no FFI dependencies.

// The numeric kernels favor explicit index arithmetic and wide, positional
// argument lists that mirror the C++ original, so a handful of Clippy style
// lints are relaxed crate-wide rather than obscuring the ports.
#![allow(
    clippy::needless_range_loop,
    clippy::too_many_arguments,
    clippy::manual_memcpy,
    clippy::field_reassign_with_default,
    clippy::doc_lazy_continuation
)]

// OpenBLAS is linked by `build.rs` (via pkg-config or OPENBLAS_LIB_DIR);
// Accelerate is linked by the framework attribute in `gemm.rs`.
#[cfg(all(
    feature = "blas",
    not(any(feature = "accelerate", feature = "openblas"))
))]
compile_error!(
    "the `blas` feature is a marker — enable a concrete backend instead: \
     `accelerate` (macOS) or `openblas` (Linux/Windows/macOS)"
);

pub mod adsampling;
pub mod batch;
pub mod common;
pub mod distance;
pub mod gemm;
pub mod hierarchical;
pub mod layout;
pub mod pdxearch;
pub mod superkmeans;
pub mod utils;

pub use common::{DistanceFunction, KnnCandidate};
pub use hierarchical::{
    HierarchicalSuperKMeans, HierarchicalSuperKMeansConfig, HierarchicalSuperKMeansIterationStats,
};
pub use superkmeans::{
    ClusterBalanceStats, SuperKMeans, SuperKMeansConfig, SuperKMeansIterationStats,
};
pub use utils::{
    TicToc, compute_l2_squared, compute_norms_row_major, find_nearest_neighbor_brute_force,
    generate_random_vectors, make_blobs,
};