Module builder 

Index builder for DiskANN

Implements the greedy best-first algorithm for constructing Vamana graphs. The builder incrementally adds vectors to the graph, maintaining connectivity and using robust pruning to select high-quality neighbors.

Build Algorithm

1. Add vectors incrementally
2. For each vector, search for nearest neighbors using beam search
3. Prune neighbors using robust pruning strategy
4. Update reverse edges (make graph bidirectional)
5. Select entry points (medoids)

References

• DiskANN: Fast Accurate Billion-point Nearest Neighbor Search on a Single Node (Jayaram Subramanya et al., NeurIPS 2019)

Structs

DiskAnnBuildStats

Index builder statistics

DiskAnnBuilder

Index builder for DiskANN