rxmc.model_comparison.logz_summary#

rxmc.model_comparison.logz_summary(logz, logzerr)[source]#

Replicate-aware evidence summary.

Parameters:

logz, logzerr (array_like) – log Z and its sampler-reported error for each replicate run (one value each is fine).

Returns:

(float, float, int) – (mean, err, n) with err = max(half-range across replicates, mean reported error) — the sampler’s own error is a lower bound.