API Reference

Configuration

High-level configuration objects for assembling a calibration problem and handing it to an external sampler (emcee, dynesty, etc.).

rxmc.config.CalibrationConfig	End-to-end configuration for Bayesian calibration.
rxmc.config.ParameterConfig	Configuration for a single sector of parameters.

Priors

Prior distribution classes that satisfy the generic prior protocol required by ParameterConfig. Any user-defined class with logpdf, rvs, and (optionally) prior_transform methods can be used directly.

rxmc.priors.IndependentPrior	Independent prior built from an arbitrary list of scipy.stats distributions.
rxmc.priors.TruncatedNormalPrior	Independent truncated-normal prior for a vector of parameters.

Core building blocks

rxmc.constraint.Constraint	Pair observations with a physical model and a stacked covariance.
rxmc.evidence.Evidence	A collection of independent constraints sharing a common physical model.
rxmc.observation.Observation	Experimental data: x, y, and the statistical error on y.
rxmc.params.Parameter	A single scalar model parameter.
rxmc.physical_model.PhysicalModel	Abstract base class for parametric physical models.
rxmc.physical_model.Polynomial	Polynomial model of fixed order.

Transforms

One low-level, numpy-style transform type shared by observations (the comparison space, e.g. transform=log), models (parametric transforms such as a latent normalisation) and covariance terms (coordinate transforms).

rxmc.transforms.Transform	A numpy-style transform with optional parameters.
rxmc.transforms.as_transform	Coerce None (identity), a callable, or a Transform.
rxmc.transforms.identity	A numpy-style transform with optional parameters.
rxmc.transforms.log	Natural log; -inf (no warnings) where the argument is not positive.
rxmc.transforms.exp	A numpy-style transform with optional parameters.
rxmc.transforms.scale	A latent multiplicative normalisation \(\rho\, y\).
rxmc.transforms.per_observation_scaling	One latent normalisation \(\rho_i\) per dataset, routed by identity.

Covariance terms

The stacked covariance of a Constraint is assembled additively from Term objects — a single generic type: a numpy-style callable of a TermContext (the term’s local x/y/ym) and its parameters, plus a kind ("diag"/"mode"/"matrix"). The factory helpers build the common terms in one line; anything else is a direct Term(fn, params, kind=...). A StackContext bundles the stacked x/y/ym that a ConstraintCovariance is evaluated on.

rxmc.covariance.Term	One additive contribution to the stacked covariance.
rxmc.covariance.TermContext	A term's local view of the stack on its own support.
rxmc.covariance.statistical_term	Always-on, genuinely uncorrelated statistical diagonal diag(stat_err**2).
rxmc.covariance.normalization_term	A correlated normalisation systematic outer(eta * ym, eta * ym).
rxmc.covariance.offset_term	A correlated absolute-offset systematic outer(omega, omega).
rxmc.covariance.noise_term	Unknown statistical noise diag((epsilon * basis)**2).
rxmc.covariance.noise_fraction_term	Unknown fractional noise diag((epsilon * ym)**2).
rxmc.covariance.model_error_term	Unknown uncorrelated model error diag((gamma * z)**2).
rxmc.covariance.systematic_term	A correlated mode outer(s * u, s * u) with a user basis u = basis(c, ...).
rxmc.covariance.kernel_term	A Gaussian-process kernel a a^T * K(x, x; theta) over the support.
rxmc.covariance.ones	Constant unit basis.
rxmc.covariance.ym	The model prediction — the prediction-scaled basis.
rxmc.covariance.averaging	0.5 * (y + ym) — the averaging model-error basis.
rxmc.covariance.x_basis	Basis x / scale (e.g. x_basis(np.pi) for theta/180 on radians).
rxmc.covariance.exp_growth	Parametric basis base(c) * exp(slope * x / scale).
rxmc.covariance.constant_amplitude	Kernel amplitude exp(log_amplitude), constant over the support.
rxmc.covariance.exp_growth_amplitude	Kernel amplitude exp(log_amplitude) * exp(slope * x / scale) (two params).
rxmc.covariance.stacked_supports	One contiguous index array per observation, in stacking order.
rxmc.covariance.ConstraintCovariance	The stacked covariance of a constraint, assembled from Term s.
rxmc.covariance.StackContext	Bundle of stacked arrays passed to every Term.add_to().

Likelihood functionals

Thin functionals of the pre-computed Mahalanobis statistics (d2, logdet, n); all covariance modeling lives on the ConstraintCovariance.

rxmc.likelihood_model.Likelihood	A functional of the pre-computed Mahalanobis statistics (d2, logdet, n).
rxmc.likelihood_model.GaussianLikelihood	Multivariate-normal likelihood over the stacked residual.
rxmc.likelihood_model.StudentT	Multivariate Student-t likelihood with a degrees-of-freedom parameter.
rxmc.likelihood_model.Chi2	Generalised chi-squared functional — drops the log-det normalisation.
rxmc.likelihood_model.mahalanobis_distance_sqr_cholesky	Squared Mahalanobis distance and log-determinant via Cholesky factorisation.
rxmc.likelihood_model.log_likelihood	Multivariate-normal log likelihood from pre-computed statistics.

Predictive utilities

Posterior-predictive helpers, including Gaussian-process discrepancy propagation.

rxmc.predictive.predictive_band	Percentile band over a matrix of predictive draws.
rxmc.predictive.gp_posterior_predictive	Posterior mean and covariance of a GP discrepancy at X_pred.
rxmc.predictive.total_predictive_band	Data-space predictive band that propagates total uncertainty.

Model comparison

Sampler-agnostic posterior-predictive checks, held-out scoring, and nested-sampling evidence bookkeeping.

rxmc.model_comparison.predictive_draws	Posterior-predictive draws on the constraint's active points.
rxmc.model_comparison.coverage_curve	Empirical coverage of central predictive intervals at each nominal level.
rxmc.model_comparison.coverage_error	max |coverage(level) - level| — a single calibration score.
rxmc.model_comparison.sharpness	Per-point width of a central predictive interval.
rxmc.model_comparison.heldout_log_predictive	log p(y_held | theta_i) for each posterior row.
rxmc.model_comparison.log_posterior_predictive	log E_post[p(y_held | theta)]: the joint log posterior predictive density of the held-out block.
rxmc.model_comparison.logz_summary	Replicate-aware evidence summary.
rxmc.model_comparison.compare_logz	Delta log Z = a - b with a conservative tie verdict.
rxmc.model_comparison.log_jacobian	Comparison-space log-Jacobian of a constraint (sum over active points).
rxmc.model_comparison.split_samples	Split flat sampler rows into (model_samples, [cov_samples, ...]).

Sampling

rxmc.walker.Walker	Gibbs-style MCMC coordinator for a Bayesian calibration problem.
rxmc.param_sampling.Sampler	Base class wrapping a sampling algorithm with chain recording.
rxmc.param_sampling.MetropolisHastingsSampler	Metropolis-Hastings sampler with a fixed proposal distribution.
rxmc.param_sampling.AdaptiveMetropolisSampler	Metropolis sampler that adapts the proposal covariance with a sliding window.
rxmc.param_sampling.BatchedAdaptiveMetropolisSampler	Metropolis sampler that updates the proposal covariance after each batch.
rxmc.proposal.ProposalDistribution	Abstract base class for Metropolis-Hastings proposal distributions.
rxmc.proposal.NormalProposalDistribution	Multivariate-normal proposal centred at the current state.
rxmc.proposal.HalfNormalProposalDistribution	Half-normal proposal, useful for strictly non-negative parameters.
rxmc.proposal.LogspaceNormalProposalDistribution	Normal proposal operating in log space, for strictly positive parameters.

Sampling algorithms

Low-level sampling functions used internally by the sampler classes.

rxmc.metropolis_hastings.metropolis_hastings	Metropolis-Hastings MCMC sampling.
rxmc.adaptive_metropolis.adaptive_metropolis	Adaptive Metropolis algorithm with sliding-window covariance adaptation.

Domain-specific models

Reaction-physics observation and model classes for elastic differential cross sections and isobaric-analog (p,n) cross sections.

rxmc.elastic_diffxs_observation.ElasticDifferentialXSObservation	Observation for elastic differential cross sections.
rxmc.elastic_diffxs_observation.momentum_transfer	Momentum transfer \(q = 2k\sin(\theta/2)\) (fm^-1) on the data angles.
rxmc.elastic_diffxs_model.ElasticDifferentialXSModel	A model that predicts the elastic differential xs for a given reaction.
rxmc.ias_pn_observation.IsobaricAnalogPNObservation	Observation for (p,n) isobaric analog state (IAS) reactions.
rxmc.ias_pn_model.IsobaricAnalogPNXSModel	A model that predicts the (p,n) IAS differential xs for a given reaction.