Struct MarkovChainMonteCarlo 

pub struct MarkovChainMonteCarlo {
    pub step_size: f64,
    pub n_warmup: usize,
}

Markov Chain Monte Carlo samplers.

Implements Metropolis-Hastings (random-walk), Gibbs sampling helpers, and a simplified dual-averaging NUTS-like step-size adaptation.

Fields

step_size: f64

Step size (proposal std for MH).

n_warmup: usize

Number of warm-up (burn-in) steps.

Implementations

impl MarkovChainMonteCarlo

pub fn new(step_size: f64, n_warmup: usize) -> Self

Creates a new MCMC sampler with the given step size and warm-up count.

pub fn metropolis_hastings<F>( &self, log_target: F, init: f64, n_samples: usize, seed: u64, ) -> Vec<f64>

where F: Fn(f64) -> f64,

Runs a Metropolis-Hastings random-walk chain.

Arguments

• log_target - Closure returning log π(θ) (up to normalisation).
• init - Initial parameter value.
• n_samples - Number of post-warmup samples to return.
• seed - RNG seed.

pub fn metropolis_hastings_vec<F>( &self, log_target: F, init: Vec<f64>, n_samples: usize, seed: u64, ) -> Vec<Vec<f64>>

where F: Fn(&[f64]) -> f64,

Metropolis-Hastings for a vector-valued parameter.

pub fn gibbs_bivariate_gaussian( mu1: f64, mu2: f64, sigma1: f64, sigma2: f64, rho: f64, n_samples: usize, seed: u64, ) -> Vec<[f64; 2]>

Gibbs sampler for a bivariate Gaussian with known full conditionals.

Samples from N([μ1, μ2], [[σ1², ρσ1σ2],[ρσ1σ2, σ2²]]).

pub fn nuts_step<F>( &self, log_target: F, grad_log_target: impl Fn(f64) -> f64, init: f64, seed: u64, ) -> f64

where F: Fn(f64) -> f64,

Simplified NUTS (No-U-Turn Sampler) step using leapfrog integration.

Returns a single sample given gradient function grad_log_target.

pub fn effective_sample_size(chain: &[f64]) -> f64

Returns the effective sample size (ESS) from a chain using autocorrelation.

Trait Implementations

impl Clone for MarkovChainMonteCarlo

fn clone(&self) -> MarkovChainMonteCarlo

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for MarkovChainMonteCarlo

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

Formats the value using the given formatter.