calculate_gelmanRubin: Calculate the potential scale reduction factor from parallel...

In hillarykoch/CLIMB: Composite LIkelihood eMpirical Bayes

Calculate the potential scale reduction factor from parallel MCMCs

Description

The Gelman-Rubin convergence diagnostic calculates the "potential scale reduction factor" (PSRF) for a series of MCMC chains. If the collection of parallel MCMCs have converged for a given parameter, the PSRF will be close to 1.

Usage

calculate_gelmanRubin(chain_list, burnin_list)

Arguments

chain_list	A list of MCMC runs output from extract_chains.
burnin_list	A vector or list of vectors specifying the iterations to discard as burn-in for each chain.

Details

NAs are returns for parameters associated with a 0 label, since these are not sampled during MCMC.

Value

PSRF_mu	Calculated PSRF for cluster mean estimates.
PSRF_Sigma	Calculated PSRF for cluster covariance estimates.
PSRF_prop	Calculated PSRF for cluster proportion estimates.

Author(s)

[e-mail]

References

Gelman, A., and D. B. Rubin. 1992. Inference from Iterative Simulation Using Multiple Sequences. Statistical Science 7: 457–511.

hillarykoch/CLIMB documentation built on Oct. 24, 2022, 4:27 a.m.