Struct GaussianProcess 

pub struct GaussianProcess { /* private fields */ }

Gaussian Process (GP) surrogate model for function approximation.

Provides posterior mean and variance predictions at query points using Cholesky-based computations. Supports hyperparameter optimization via NLML.

Implementations

impl GaussianProcess

pub fn new(kernel_type: KernelType, noise_variance: f64) -> Self

Create a new unfitted Gaussian Process with the given kernel and noise variance.

pub fn fit(&mut self, x: &[Vec<f64>], y: &[f64]) -> Result<()>

Fit the GP model to training data (normalizes, computes kernel, Cholesky, alpha).

pub fn predict(&self, x_new: &[f64]) -> Result<(f64, f64)>

Predict the posterior mean and variance at a new point (de-normalized).

pub fn predict_batch(&self, x_new: &[Vec<f64>]) -> Result<(Vec<f64>, Vec<f64>)>

Predict mean and variance for multiple points.

pub fn negative_log_marginal_likelihood(&self) -> Result<f64>

Compute the negative log marginal likelihood (for hyperparameter optimization).

pub fn optimize_hyperparameters( &mut self, x: &[Vec<f64>], y: &[f64], n_restarts: usize, ) -> Result<()>

Optimize kernel hyperparameters via multi-start coordinate descent on NLML.

pub fn sample_posterior( &self, x_points: &[Vec<f64>], rng: &mut impl Rng, ) -> Result<Vec<f64>>

Sample from the GP posterior at the given points (for Thompson Sampling).

pub fn kernel_type(&self) -> &KernelType

Get the current kernel type.

pub fn length_scale(&self) -> f64

Get the current length scale.

pub fn signal_variance(&self) -> f64

Get the current signal variance.

pub fn n_train(&self) -> usize

Get the number of training points.

Trait Implementations

impl Clone for GaussianProcess

fn clone(&self) -> GaussianProcess

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for GaussianProcess

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

Formats the value using the given formatter.