Convergence Checks (for multiple chains)

Module contains convergence checks functions that require data preprocessing. Most often this means having multiple chains.

monte.convergence_checks.gelman_rubin(chains)

Calculates Gelman-Rubin statistic (R-hat) for Markov chains

Parameters:

chains (pd.DataFrame) – Markov chains used for the calculation

Returns:

Gelman-Rubin statistic (R-hat)

Return type:

float

monte.convergence_checks.multivariate_gelman_rubin(chains)

Calculates multivariate Gelman-Rubin statistic (R-hat) for Markov chains

Parameters:

chains (pd.DataFrame) – Markov chains used for the calculation with dimensions (n_chains, n_samples, n_dim), where n_chains is the number of chains, n_samples is the number of samples per chain, and n_dim is the number of dimensions.

Returns:

Multivariate Gelman-Rubin statistic (R-hat)

Return type:

float