Struct opensrdk_probability::nonparametric::elliptical_process::BaseEllipticalProcessParams

pub struct BaseEllipticalProcessParams<K, T> where
    K: PositiveDefiniteKernel<T>,
    T: RandomVariable,  { /* private fields */ }

You can use these methods:

• exact
• sparse
• kiss_love

Implementations

impl<K, T> BaseEllipticalProcessParams<K, T> where
    K: PositiveDefiniteKernel<T>,
    T: RandomVariable, 

pub fn exact(
    self,
    y: &[f64]
) -> Result<ExactEllipticalProcessParams<K, T>, DistributionError>

Elliptical Process without approximation for scalability.

• Pre-computation time: O(n^3)
• Pre-computation storage: O(n^2)
• Prediction time: O(n^2)

impl<K, T> BaseEllipticalProcessParams<Convolutional<K>, T> where
    K: PositiveDefiniteKernel<Vec<f64>>,
    T: RandomVariable + Convolutable, 

pub fn kiss_love(
    self,
    y: &[f64]
) -> Result<KissLoveEllipticalProcessParams<K, T>, DistributionError>

Lanczos Variance Estimate Kernel Interpolation for Scalable Structured Gaussian Process

• Pre-computation time: O(k(n + m log m))
• Pre-computation storage: O(km)
• Prediction time: O(k) with k = 100

impl<K, T> BaseEllipticalProcessParams<K, T> where
    K: PositiveDefiniteKernel<T>,
    T: RandomVariable, 

pub fn sparse(
    self,
    y: &[f64],
    u: Vec<T>
) -> Result<SparseEllipticalProcessParams<K, T>, DistributionError>

Elliptical Process with approximation called the Fully Independent Training Conditional (FITC) for scalability.

• Pre-computation time: O(nm^2)
• Pre-computation storage: O(m^2)
• Prediction time: O(m^2)

impl<K, T> BaseEllipticalProcessParams<K, T> where
    K: PositiveDefiniteKernel<T>,
    T: RandomVariable, 

pub fn new(
    kernel: K,
    x: Vec<T>,
    theta: Vec<f64>,
    sigma: f64
) -> Result<Self, DistributionError>

• kernel: Kernel function
• x: Input value
• theta: Params of kernel function
• sigma: White noise variance for a regularization likes Ridge regression

Trait Implementations

impl<K: Clone, T: Clone> Clone for BaseEllipticalProcessParams<K, T> where
    K: PositiveDefiniteKernel<T>,
    T: RandomVariable, 

fn clone(&self) -> BaseEllipticalProcessParams<K, T>

Returns a copy of the value.

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

Performs copy-assignment from source.

impl<K: Debug, T: Debug> Debug for BaseEllipticalProcessParams<K, T> where
    K: PositiveDefiniteKernel<T>,
    T: RandomVariable, 

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

Formats the value using the given formatter.