Effective sample size by the initial monotone sequence estimator

Geyer's (1992) initial positive sequence with the monotone step, on the autocovariances computed by FFT, as posterior::ess_basic(x, split = FALSE): no rank normalization, no split. With a matrix (one column per chain) the chains share one autocorrelation estimate, so chains that disagree lower the result. iid draws give about the number of draws. For the bulk and tail ESS of Vehtari et al. (2021), see mcmc_diagnostics().

Usage

ess(x)

Arguments

x

numeric vector (one chain) or matrix of draws, one column per chain

Value

a single number, the effective sample size.

References

Geyer, C.J. (1992). Practical Markov chain Monte Carlo. Statistical Science 7:473-483.