KLSH tightens neighborhoods
Config B (RBF + KLSH) clusters top-8 neighbors above 0.94 similarity, versus a sparser cosine LSH decay in Config A.
Benchmarks
Simulation results
Curated top-k similarity measurements from the Concentric Vector Chord prototype paper. Configurations A/B/C use seed 42 on Apple Silicon.
Run setup
- Nodes
- 16
- Vectors
- 256
- Dimension
- 32
- Metric
- Cosine LSH
- KLSH
- off
- Query key
- 25921
- Route hops
- 0
- Init + query
- < 100 ms
Top-8 similarity decay (self-query)
Concentric clustering needs overflow
Without Chord successor overflow, bucket saturation failed near 300 vectors on 24 nodes. Overflow stabilized insertion to 512 vectors.
Sub-100 ms simulation queries
In-process init and k-NN query for 256–512 vectors completed in under 100 ms on Apple Silicon (seed 42).
Vulkan validates placement
Interactive visualization confirmed shell guides, ring topology, vector projections, and routing hop paths during development.