How zvec integrates RaBitQ-Library

Zvec uses RaBitQ-Library for rotation, quantization, and distance estimation inside its own IVF and HNSW indexes. Zvec manages storage, search traversal, filtering, and result collection.

Index	How it uses RaBitQ-Library
IVF	The reformer quantizes vectors in batches and prepares query lookup tables. The search code uses FastScan estimates and extra-bit refinement for selected candidates.
HNSW	The reformer quantizes individual vectors. The query algorithm estimates distances from those codes during graph traversal.

Both integrations preserve centroids and rotator state with the index and use padded dimensions consistently for rotated vectors and queries. These are key contracts when embedding the quantizer in another system.

This overview follows zvec commit 44d8bc6f7c3d. Its bundled dependency includes local patches for lookup-table reuse and SIMD error handling; check the pinned version before adapting its code.

To integrate RaBitQ-Library, start with the quantizer and estimator guides. For a complete standalone index, see the quick start.