Module diskann 

DiskANN / Vamana SSD-Backed Approximate Nearest Neighbor Index

Implements the Vamana graph index from the DiskANN paper (Subramanya et al., 2019). Each node connects to R neighbors chosen via alpha-RNG pruning – a relaxed Relative Neighborhood Graph balancing proximity and angular diversity.

Why DiskANN achieves 95%+ recall at sub-10ms

• Vamana graph: alpha > 1.0 retains long-range shortcuts for O(log n) hops.
• SSD layout: node vector + neighbors packed in aligned pages; one read per hop.
• Page cache: LRU cache keeps hot pages in memory (80-95% hit rates typical).
• Filtered traversal: predicates evaluated during search, not post-filter.

Alpha-RNG Pruning

A candidate c is kept only if for every already-selected neighbor n, dist(p, c) <= alpha * dist(n, c), ensuring angular diversity.

Structs

DiskIndex

Simulated SSD-backed index with page-aligned reads and LRU cache.

DiskNode

A node stored in SSD-backed layout: id + neighbors + vector in one page.

IOStats

IO statistics for disk-based search.

MedoidFinder

Finds the geometric medoid (point minimising sum of distances to all others).

PageCache

LRU page cache tracking access recency via a clock counter.

VamanaConfig

Configuration for the Vamana graph index.

VamanaGraph

In-memory Vamana graph for building and searching.