ConvergenceTest._assess#

classmethod ConvergenceTest._assess(draws, *, names, rhat_tol=_RHAT_TOL, min_ess=_MIN_ESS_PER_CHAIN, source='draws')[source]#

Compute every diagnostic from (C, N, d) draws.

Parameters:

  • draws (NDArray[double]) – C chains of N kept draws of d scalar quantities.

  • names (tuple[str, ...]) – d labels.

  • rhat_tol (float) – Flagging threshold for R-hat.

  • min_ess (float) – Flagging floor for effective draws, per chain.

  • source (str) – Label for the summary title.

Raises:

  • DimensionError – If draws is not three-dimensional or names does not match its last axis.

  • SpecificationError – If the thresholds are not positive or a chain is shorter than the minimum.

Return type:

Self