Utility function to change the default options for the runjags package. Options will be used for all runjags function calls until the runjags package is unloaded. For a permanent solution, create a named list called '.runjags.options' containing the desired options in an R profile file - on loading, runjags will check to see if this object exists in the global environment and set the options automatically.
runjags.options(...)runjags.getOption(name)
runJAGS.getOption(name)
The current value of all available runjags options (after applying any changes specified) is returned invisibly as a named list.
named option(s) to change - for a list of available options, see details below.
the name of the option to get the current value of - for a list of available options, see details below.
The following default options can be specified:
jagspath - the path to JAGS to use unless over-ridden in a function call (uses the findjags() function by default).
method - the runjags method to use unless over-ridden in a function call (default is 'rjags' if the rjags package is installed, or 'interruptible' otherwise).
tempdir - default to temporary directory unless over-ridden in a function call (default TRUE).
plot.layout - the layout for plots unless over-ridden in a function call. Must be a numeric vector of length 2.
new.windows - use multiple windows for plots unless over-ridden in a function call (default is platform dependent).
modules - the modules to load unless over-ridden in a function call (default none).
factories - the factories to load unless over-ridden in a function call (default none).
bg.alert - an optional command to run once background JAGS processes have completed. Note that this command is run on the command line via system(), so will be system dependent. The default attempts to make an alert sound using a system appropriate method, which may not work on all platforms.
linenumbers - display line numbers when printing runjags model, data and inits class objects unless over-ridden in a function call (default none).
inits.warning - display warning messages about initial values being not specified or re-used.
rng.warning - display warning messages relating to pseudo-random number generation for parallel chains.
summary.warning - display a warning message if summary statistics are requested for a small number of samples (and a few other similar situations).
blockcombine.warning - display a warning message if multiple data or inits blocks are combined in a model file.
blockignore.warning - display a warning message if ignoring monitors, data or inits in the model file because a character argument was given for the same parameters to the run.jags function.
tempdir.warning - display a warning message if tempdir=TRUE is requested with a background method.
nodata.warning - display a warning message if the model has been run without any data.
adapt.incomplete - all models are checked to make sure that the adaptive phase has completed - this option controls the behaviour of runjags if this adaptation is not complete before MCMC sampling. If adapt.incomplete='silent' no action is taken, if 'warning' then the model run is continued but a warning is given after the simulation is finished, and if 'error' an error will be returned. Note that for most methods the error is returned immediately following the adapt/burnin phases (so the sample iterations are not run), but for the simple and snow methods the full model will be run before the error is given.
repeatable.methods - option to ensure that the MCMC object produced by the rjags and rjparallel methods are identical to those produced by other methods (given the same starting values). This is primarily for extending compiled models, where additional burnin iterations will be done to replace unnecessary adpative steps for consistency with other methods, and following dic sampling, where the rjags model will be reset to the state it was in before dic sampling. Note that the precision of the numbers returned may differ between methods on some platforms.
silent.jags - suppress output of JAGS (or rjags) when updating models.
silent.runjags - suppress feedback provided by the runjags functions.
predraw.plots - automatically pre-calculate convergence diagnostic plots (this will save time when displaying plots at the cost of increased storage requirement for the runjags object).
force.summary - override the default behaviour to omit calculation of summary statistics for greater than 50 variables, and attempt to calculate summary statistics anyway (this may cause long delays in the processing of results).
mode.continuous - calculate the mode of continuous variables for summary statistics (requires the "modeest" package to be installed).
timeout.import - the maximum number of seconds for runjags to wait for coda files to finish being written before giving up. If a large number of monitored variables are being written, either the timeout can be increased or results.jags() can be used once the files have been written.
partial.import - force runjags to read in successful simulations even when parallel simulations crashed. If this option is set to TRUE, it is not guaranteed that a model result will contain the requested number of chains!
keep.crashed.files - allows folders containing crashed simulations to be preserved even if keep.jags.files = FALSE. Any folders kept will be deleted when runjags is unlaoded or when R quits.
full.cleanup - when unloading the runjags package, should all simulation folders preserved using keep.jags.files=TRUE be deleted? This option may not work as expected on all systems when quitting R, but should always work for unloadNamespace('runjags'). Note also that folders for any failed JAGS runs are always deleted on exit - if you want to keep these, they will have to be copied manually.
debug - display internal debugging output.
run.jags, findjags, runjags-class
if (FALSE) {
# Create a list of options in the global environment (perhaps in an
# R startup profile file) BEFORE load()ing runjags:
.runjags.options <- list(inits.warning=FALSE, rng.warning=FALSE)
# Or if it is run in a different environment:
# .runjags.options <<- list(inits.warning=FALSE, rng.warning=FALSE)
# Then load runjags and verify that the options have been set:
library('runjags')
print(runjags.options())
# Change the default option to remove all feedback provided by
# runjags and JAGS/rjags (only errors will be printed to screen):
runjags.options(silent.jags=TRUE, silent.runjags=TRUE)
}
Run the code above in your browser using DataLab