bridgestat.diag: Iterative truncation procedure based on a bridge statistic.

In codadiags: Markov chain Monte Carlo burn-in based on "bridge" statistics

Description

Iterative truncation procedure based on a bridge statistic.

Usage

  bridgestat.diag(x, bridge = "student", stat = "E",
    param = "asymptotic", trunc = 1, eps = 0.1,
    pvalue = 0.3)

Arguments

x	coda::mcmc sequence (will be cast to if necessary) to truncate transient
bridge	bridge type to use: "brownian", student" or "loglik"
stat	statistic to use for testing bridge: - if student bridge, "E","var","autocov","loglik_mean","loglik_extremum","extremum","ratio_extremum","ratio_loglik_extremum" - if brownian bridge, "E","var","autocov","loglik_mean","loglik_extremum","extremum","ratio_extremum","ratio_loglik_extremum" - if loglik bridge, "E","var","autocov","extremum","ratio_extremum"
param	if "asymptotic" use asymptotic statistics, else if a list of 'N' and 'rho' use these parameters, if NULL estimate N and rho
trunc	number of mcmc iterations to delete: if >=1, it is a constant number, if <1, a percentage of remaining batches
eps	Target value for ratio of halfwidth to sample mean (for compatibility with heidel.diag)
pvalue	significance level to use in iterative test

References

Heidelberger P and Welch PD. Simulation run length control in the presence of an initial transient. Opns Res., 31, 1109-44 (1983)

See Also

coda::heidel.diag

Examples

require(codadiags)
set.seed(123)
x = AR1()
print(bridgestat.diag(x))
y = add.transient(x)
print(bridgestat.diag(y,trunc=10))

codadiags documentation built on May 1, 2019, 9:26 p.m.