rxmc.evidence.Evidence#

class rxmc.evidence.Evidence(constraints: list[Constraint] | None = None, weights: ndarray = None)[source]#

Bases: object

A collection of independent constraints sharing a common physical model.

Parameters:

  • constraints (list of Constraint) – All constraints. Those with n_params > 0 are exposed (in order) as parametric_constraints.

  • weights (np.ndarray, optional) – 1-D array of per-constraint weights. Defaults to all ones.

Raises:

ValueError – If constraints is empty, if any constraint uses different physical-model parameters than the first, if the model is under-constrained, or if weights does not match the constraint count.

constraints#

All constraints.

Type:

list of Constraint

parametric_constraints#

The subset with n_params > 0, in the order they appear.

Type:

list of Constraint

parametric_indices#

Global index into constraints for each entry of parametric_constraints (e.g. to look up its weights entry).

Type:

list of int

__init__(constraints: list[Constraint] | None = None, weights: ndarray = None)[source]#

Methods

__init__([constraints, weights])

log_likelihood(model_params[, cov_params])

Weighted sum of log likelihoods over all constraints.

weighted_marginal_log_likelihood(lm_index, ...)

Weighted marginal log likelihood of one parametric constraint.
log_likelihood(model_params, cov_params: list | None = None)[source]#

Weighted sum of log likelihoods over all constraints.

Parameters:

  • model_params (tuple) – Physical-model parameters.

  • cov_params (list of tuple, optional) – One tuple of constraint parameters per entry in parametric_constraints (in that order). Defaults to [].

Returns:

float – Total weighted log likelihood.

weighted_marginal_log_likelihood(lm_index, ym, *cov_params)[source]#

Weighted marginal log likelihood of one parametric constraint.

Applies the same weights entry that log_likelihood() uses for this constraint, so Gibbs-style conditional updates target the same joint distribution as the model block.

Parameters:

  • lm_index (int) – Index into parametric_constraints.

  • ym (list of np.ndarray) – One prediction array per observation of that constraint.

  • *cov_params (float) – The constraint’s parameters, in its params order.