NAME

Logic::Probabilistic::Inference::MCMC - Markov Chain Monte Carlo (Metropolis-Hastings) inference engine.

SYNOPSIS

use Logic::Probabilistic::Inference::MCMC;

my $mcmc = Logic::Probabilistic::Inference::MCMC->new(
    program => $prob_prog,
    chains  => 4,
    samples => 5000,
    burn_in => 500,
    lag     => 5,
);
my $result = $mcmc->query(query => 'burglary', evidence => ['alarm']);
say "R-hat: " . $result->r_hat;
say "ESS:   " . $result->ess;

DESCRIPTION

Samples from conditional distributions using the Metropolis-Hastings algorithm with multi-chain execution, Gelman-Rubin convergence diagnostics (R-hat), and Effective Sample Size (ESS) estimation.