mcmc.chainlength.est: Estimate appropriate chain length for mixed stock analysis by...
In mixstock: functions for mixed stock analysis

Description Usage Arguments Details Value Author(s) See Also Examples

Determines an appropriate chain length for MCMC estimation of source contributions to a mixed stock by running Raftery and Lewis and Gelman & Rubin diagnostics repeatedly until convergence criteria are met

1 2	mcmc.chainlength.est(x, mult=1, inflate=sqrt(2), GR.crit=1.2, nchains=x$R, verbose=FALSE)

`x`	Mixed stock analysis data (a `mixstock.data` object or a list containing `sourcesamp` and `mixsamp`
`mult`	How many different times to run tests
`inflate`	How much to increase chain length at every failed iteration of Gelman and Rubin
`GR.crit`	Maximum value for Gelman and Rubin 97.5% quantile in order to declare convergence
`nchains`	number of separate MCMC chains to run
`verbose`	print lots of detail while running?

If mult is 1, runs Raftery and Lewis diagnostics on a chain starting from equal contributions; if mult is greater than 1, runs them on as many chains as there are sourcees, each starting from a 95% contribution from that source. Iteratively increases each chain length to that suggested by the R&L diagnostic, until all chains pass. Then runs Gelman and Rubin on a set of chains starting from each source. If mult is greater than 1, it does each step on mult different chains and takes the maximum.

The maximum chainlength needed to get convergence in all tests

Ben Bolker

gibbsC