Trait linfa_nn::distance::Distance

pub trait Distance<F: Float>: Clone + Send + Sync + Unpin {
    // Required method
    fn distance<D: Dimension>(
        &self,
        a: ArrayView<'_, F, D>,
        b: ArrayView<'_, F, D>
    ) -> F;

    // Provided methods
    fn rdistance<D: Dimension>(
        &self,
        a: ArrayView<'_, F, D>,
        b: ArrayView<'_, F, D>
    ) -> F { ... }
    fn rdist_to_dist(&self, rdist: F) -> F { ... }
    fn dist_to_rdist(&self, dist: F) -> F { ... }
}

A distance function that can be used in spatial algorithms such as nearest neighbour.

Required Methods

fn distance<D: Dimension>( &self, a: ArrayView<'_, F, D>, b: ArrayView<'_, F, D> ) -> F

Computes the distance between two points. For most spatial algorithms to work correctly, this metric must satisfy the Triangle Inequality.

Panics if the points have different dimensions.

Provided Methods

fn rdistance<D: Dimension>( &self, a: ArrayView<'_, F, D>, b: ArrayView<'_, F, D> ) -> F

A faster version of the distance metric that keeps the order of the distance function. That is, dist(a, b) > dist(c, d) implies rdist(a, b) > rdist(c, d). For most algorithms this is the same as distance. Unlike distance, this function does not need to satisfy the Triangle Inequality.

fn rdist_to_dist(&self, rdist: F) -> F

Converts the result of rdistance to distance

fn dist_to_rdist(&self, dist: F) -> F

Converts the result of distance to rdistance

Object Safety

This trait is not object safe.

Implementors

impl<F: Float> Distance<F> for L1Dist

impl<F: Float> Distance<F> for L2Dist

impl<F: Float> Distance<F> for LInfDist

impl<F: Float> Distance<F> for LpDist<F>