Struct HNSWConfig 

pub struct HNSWConfig {
    pub m: usize,
    pub m0: usize,
    pub ef_construction: usize,
    pub ef_search: usize,
    pub ml: f64,
    pub sparsity_threshold: f32,
    pub max_nodes: usize,
    pub distance_metric: HNSWDistanceMetric,
}

Configuration for HNSW index.

Fields

m: usize

Maximum number of connections per node per layer (default: 16)

m0: usize

Maximum connections at layer 0 (default: 2*m = 32)

ef_construction: usize

Candidates to consider during construction (default: 200)

ef_search: usize

Candidates to consider during search (default: 50)

ml: f64

Level multiplier for layer selection (default: 1/ln(m))

sparsity_threshold: f32

Sparsity threshold for auto-sparse storage (default: 0.5 = 50% zeros)

When using insert_auto(), vectors with sparsity above this threshold will be stored as sparse vectors for memory savings.

max_nodes: usize

Maximum number of nodes allowed in the index (0 = unlimited). Default: 10,000,000 (10M nodes) to prevent memory exhaustion.

distance_metric: HNSWDistanceMetric

Distance metric for similarity computation.

Determines how distances are calculated during index construction and search. Must match the embedding model’s training objective:

• Use Cosine for normalized embeddings (most common)
• Use Euclidean for spatial/geometric embeddings
• Use DotProduct for MIPS (maximum inner product search)

Default: HNSWDistanceMetric::Cosine

Implementations

impl HNSWConfig

pub fn high_recall() -> HNSWConfig

Create a config optimized for high recall (slower but more accurate)

pub fn high_speed() -> HNSWConfig

Create a config optimized for speed (faster but lower recall)

pub const fn with_distance_metric( self, metric: HNSWDistanceMetric, ) -> HNSWConfig

Set the distance metric for this configuration.

Example

use tensor_store::{HNSWConfig, HNSWDistanceMetric};

let config = HNSWConfig::default()
    .with_distance_metric(HNSWDistanceMetric::Euclidean);

Trait Implementations

impl Clone for HNSWConfig

fn clone(&self) -> HNSWConfig

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for HNSWConfig

fn fmt(&self, f: &mut Formatter<'_>) -> Result<(), Error>

Formats the value using the given formatter.

impl Default for HNSWConfig

fn default() -> HNSWConfig

Returns the “default value” for a type.

impl<'de> Deserialize<'de> for HNSWConfig

fn deserialize<__D>( __deserializer: __D, ) -> Result<HNSWConfig, <__D as Deserializer<'de>>::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Serialize for HNSWConfig

fn serialize<__S>( &self, __serializer: __S, ) -> Result<<__S as Serializer>::Ok, <__S as Serializer>::Error>

where __S: Serializer,

Serialize this value into the given Serde serializer.