Module vector 

Vector indexing data structures

This module provides a modular architecture for vector search:

Module Structure

• ivf - Core IVF (Inverted File Index) infrastructure

  • CoarseCentroids - k-means clustering for coarse quantization
  • ClusterData / ClusterStorage - generic cluster storage
  • SoarConfig / MultiAssignment - SOAR geometry-aware assignment

• quantization - Vector quantization methods

  • RaBitQCodebook - RaBitQ binary quantization (32x compression)
  • PQCodebook - Product Quantization with OPQ (ScaNN-style)
  • Quantizer trait - common interface for quantizers

• index - Ready-to-use IVF indexes

  • IVFRaBitQIndex - IVF + RaBitQ
  • IVFPQIndex - IVF + PQ

SOAR (Spilling with Orthogonality-Amplified Residuals)

The IVF module includes Google’s SOAR algorithm for improved recall:

• Assigns vectors to multiple clusters (primary + secondary)
• Secondary clusters chosen to have orthogonal residuals
• Improves recall by 5-15% with ~1.3-2x storage overhead

Re-exports

pub use ivf::ClusterData;

pub use ivf::ClusterStorage;

pub use ivf::CoarseCentroids;

pub use ivf::CoarseConfig;

pub use ivf::MultiAssignment;

pub use ivf::QuantizedCode;

pub use ivf::SoarConfig;

pub use quantization::DistanceTable;

pub use quantization::PQCodebook;

pub use quantization::PQConfig;

pub use quantization::PQVector;

pub use quantization::QuantizedQuery;

pub use quantization::QuantizedVector;

pub use quantization::Quantizer;

pub use quantization::RaBitQCodebook;

pub use quantization::RaBitQConfig;

pub use index::IVFPQConfig;

pub use index::IVFPQIndex;

pub use index::IVFRaBitQConfig;

pub use index::IVFRaBitQIndex;

pub use index::RaBitQIndex;

Modules

index

Vector index implementations

ivf

IVF (Inverted File Index) module for vector search

quantization

Vector quantization methods for IVF indexes