Struct BallTree

pub struct BallTree<T: Float + Send + Sync, D: Distance<T>> { /* private fields */ }

Ball tree for efficient nearest neighbor searches

The ball tree partitions data into a set of nested hyperspheres (balls), which allows for efficient nearest neighbor searches, especially in high-dimensional spaces. Each node in the tree represents a ball containing a subset of the points.

Type Parameters

• T: Floating point type (f32 or f64)
• D: Distance metric that implements the Distance trait

Implementations

impl<T: Float + Send + Sync + 'static, D: Distance<T> + Send + Sync + 'static> BallTree<T, D>

pub fn new( data: &ArrayView2<'_, T>, leaf_size: usize, distance: D, ) -> SpatialResult<BallTree<T, D>>

Create a new ball tree from the given data points

Arguments

• data - 2D array of data points (n_samples x n_features)
• leaf_size - Maximum number of points in leaf nodes
• distance - Distance metric to use

Returns

• SpatialResult<BallTree<T, D>> - A new ball tree

pub fn query( &self, point: &[T], k: usize, return_distance: bool, ) -> SpatialResult<(Vec<usize>, Option<Vec<T>>)>

Query the k nearest neighbors to the given point

Arguments

• point - Query point
• k - Number of neighbors to find
• return_distance - Whether to return distances

Returns

• SpatialResult<(Vec<usize>, Option<Vec<T>>)> - Indices and optionally distances of the k nearest neighbors

pub fn query_radius( &self, point: &[T], radius: T, return_distance: bool, ) -> SpatialResult<(Vec<usize>, Option<Vec<T>>)>

Find all points within a given radius of the query point

Arguments

• point - Query point
• radius - Radius to search within
• return_distance - Whether to return distances

Returns

• SpatialResult<(Vec<usize>, Option<Vec<T>>)> - Indices and optionally distances of points within radius

pub fn query_radius_tree( &self, other: &BallTree<T, D>, radius: T, ) -> SpatialResult<Vec<(usize, usize)>>

Find all pairs of points from two trees that are within a given radius

Arguments

• other - Another ball tree
• radius - Radius to search within

Returns

• SpatialResult<Vec<(usize, usize)>> - Pairs of indices (self_idx, other_idx) within radius

pub fn get_data(&self) -> &Array2<T>

Get the original data points

pub fn get_n_samples(&self) -> usize

Get the number of data points

pub fn get_n_features(&self) -> usize

Get the dimension of data points

pub fn get_leaf_size(&self) -> usize

Get the leaf size

impl<T: Float + Send + Sync + 'static> BallTree<T, EuclideanDistance<T>>

pub fn with_euclidean_distance( data: &ArrayView2<'_, T>, leaf_size: usize, ) -> SpatialResult<BallTree<T, EuclideanDistance<T>>>

Create a new ball tree with default Euclidean distance metric

Arguments

• data - 2D array of data points (n_samples x n_features)
• leaf_size - Maximum number of points in leaf nodes

Returns

• SpatialResult<BallTree<T, EuclideanDistance<T>>> - A new ball tree

Trait Implementations

impl<T: Clone + Float + Send + Sync, D: Clone + Distance<T>> Clone for BallTree<T, D>

fn clone(&self) -> BallTree<T, D>

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl<T: Debug + Float + Send + Sync, D: Debug + Distance<T>> Debug for BallTree<T, D>

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

Formats the value using the given formatter.