mm.wbugs: Run many-to-many model via WinBUGS (or JAGS)

In mixstock: functions for mixed stock analysis

Description

Sets up the many-to-many model and passes it to WinBUGS via R2WinBUGS (or R2jags)

Usage

mm.wbugs(x, sourcesize,
n.iter=20000,  n.burnin=floor(n.iter/2),
n.chains=x$R,
n.thin=max(1, floor(n.chains * (n.iter - n.burnin)/1000)),
files.only = FALSE,
inittype=c("dispersed","random"),bugs.code=c("TO","BB"),
                     returntype=c("mixstock","coda","bugs"),
pkg=c("WinBUGS","JAGS"),
mixprior=1,
which.init,
debug=FALSE,
working.directory,...)
write.TO.bugscode(fn,MIX)

Arguments

x	a mixstock data object
sourcesize	Relative sizes of sources
n.iter	Total length of each chain
n.burnin	Number of burn-in iterations
n.chains	Number of chains (default, number of sources)
n.thin	thinning rate. Must be a positive integer. Set 'n.thin' > 1 to save memory and computation time if 'n.iter' is large. Default is 'max(1, floor(n.chains * (n.iter-n.burnin) / 1000))' which will only thin if there are at least 2000 simulations.
files.only	(unimplemented) don't run WinBUGS, just produce input files
inittype	"dispersed" or "random" depending on how you want multiple chains to be initialized
bugs.code	"TO" or "BB" depending on whether you want old-style (Toshi Okuyama=TO) or more compact but possibly slower (Ben Bolker=BB) code
mixprior	Dirichlet prior for contributions to mixed stocks. Should be either (1) a a single numeric value which will be replicated the appropriate number of times; (2) a vector of length (nmix+1) [one extra for the unknown stock], identical for all sources – or (3) a matrix with (nmix+1) columns and a row for each source
which.init	for "dispersed" start with fewer chains than sources, which sources should be used as the dominant sources in the chains? (default is a random sample without replacement from the list of sources)
debug	run BUGS in debug mode? (i.e. don't exit and go back to R automatically)
...	other arguments to bugs
fn	file name to write BUGS code to
MIX	number of mixed stocks
returntype	return value as a mixstock.est object, a CODA object, or a BUGS object?
pkg	which package to use for back-end calculations
working.directory	working directory for BUGS calculations

Value

results of WinBUGS run, as a mixstock.est object by default: type varies according to returntype. write.TO.code produces a BUGS model file.

Note

For diagnostic purposes, it may be worth running the code initially with returntype="bugs" and using as.mcmc.bugs and as.mixstock.est.bugs to convert the result to either CODA format or mixstock format.

pkg="JAGS" is still experimental.

Author(s)

Ben Bolker

mixstock documentation built on May 2, 2019, 6:48 p.m.