Struct SquaredExp

pub struct SquaredExp {
    pub ls: f64,
    pub ampl: f64,
}

Squared exponential kernel

Equivalent to a gaussian kernel.

k(x,y) = A exp(-||x-y||² / 2l²)

Where A is the amplitude and l the length scale.

Fields

ls: f64

The length scale of the kernel.

ampl: f64

The amplitude of the kernel.

Implementations

impl SquaredExp

pub fn new(ls: f64, ampl: f64) -> SquaredExp

Construct a new squared exponential kernel (gaussian).

Trait Implementations

impl Clone for SquaredExp

fn clone(&self) -> SquaredExp

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for SquaredExp

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

Formats the value using the given formatter.

impl Default for SquaredExp

Constructs the default Squared Exp kernel.

The defaults are:

• ls = 1
• ampl = 1

fn default() -> SquaredExp

Returns the “default value” for a type.

impl<'de> Deserialize<'de> for SquaredExp

fn deserialize<__D>(__deserializer: __D) -> Result<Self, __D::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Kernel for SquaredExp

fn kernel<S1: Storage<f64, U1, Dynamic>, S2: Storage<f64, U1, Dynamic>>( &self, x1: &SRowVector<S1>, x2: &SRowVector<S2>, ) -> f64

The squared exponential kernel function.

fn nb_parameters(&self) -> usize

Numbers of parameters (such as bandwidth and amplitude) of the kernel.

fn is_scalable(&self) -> bool

Can the kernel be rescaled (see the rescale function) ? This value is false by default.

fn gradient<S1: Storage<f64, U1, Dynamic>, S2: Storage<f64, U1, Dynamic>>( &self, x1: &SRowVector<S1>, x2: &SRowVector<S2>, ) -> Vec<f64>

Takes two equal length slices (row vector) and returns a vector containing the value of the gradient for each parameter in an arbitrary order.

fn rescale(&mut self, scale: f64)

Multiplies the amplitude of the kernel by the scale parameter such that a kernel a*K(x,y) becomes scale*a*K(x,y).

fn get_parameters(&self) -> Vec<f64>

Returns a vector containing all the parameters of the kernel in the same order as the outputs of the gradient function.

fn set_parameters(&mut self, parameters: &[f64])

Sets all the parameters of the kernel by reading them from a slice where they are in the same order as the outputs of the gradient function.

fn heuristic_fit<SM: Storage<f64, Dynamic, Dynamic>, SV: Storage<f64, Dynamic, U1>>( &mut self, training_inputs: &SMatrix<SM>, training_outputs: &SVector<SV>, )

Optional, function that fits the kernel parameters on the training data using fast heuristics. This is used as a starting point for gradient descent.

impl Serialize for SquaredExp

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

where __S: Serializer,

Serialize this value into the given Serde serializer.

impl Copy for SquaredExp