Search over massive collections of vectors can be slow. HNSW (Hierarchical Navigable Small World) is an algorithm that significantly accelerates vector search times.

​
Basic Usage

An HNSW index can be created over any column with the vector or sparsevec type. 
CREATE INDEX ON <schema_name>.<table_name>
USING hnsw (<column_name> <distance_metric>);
​
table_name
required
The name of the table being indexed.
​
column_name
required
The name of the column being indexed. Must be of type vector.
​
distance_metric
required
The distance metric used for measuring similarity between two vectors. For the vector data type, use vector_l2_ops for L2 distance, vector_ip_ops for inner product, and vector_cosine_ops for cosine distance. For the sparsevec data type, use sparsevec_l2_ops for L2 distance, sparsevec_ip_ops for inner product, and sparsevec_cosine_ops for cosine distance.
​
schema_name
The name of the schema, or namespace, of the table. If not provided, the search path is used as a default.

​
Index Options

The following example demonstrates how to pass options when creating the HNSW index: 
CREATE INDEX ON <schema_name>.<table_name>
USING hnsw (<column_name> <distance_metric>)
WITH (m = 16, ef_construction = 64);
​
m
default:16
The maximum number of connections per layer. A higher value increases recall but also increases index size and construction time.
​
ef_construction
default:64
A higher value creates a higher quality graph, which increases recall but also construction time.