setInits function

Set up and dump initial values given the model and prior

Given a model of class modelSegratioMM and priors of class priorsSegratioMM, initial values are computed using approximate expected values by setInits and then written to file by dumpInits

setInits(model, priors, seed = 1)

dumpInits(inits, stem = "test", inits.file = paste(stem, "-inits.R",
  sep = ""))

Arguments

• model: Object of class modelSegratioMM providing model attributes like the number of components and ploidy level
• priors: Object of class priorsSegratioMM
• seed: Seed to be used for JAGS runs. If a number of chains are to be run a vector of starting values may be specified. However, see note below.
• inits: A list of initial values usually produced by setInits
• stem: File name stem for inits file (default test )
• inits.file: Inits file name which is automatically generated from stem if not specified

Returns

Returns a list with the following initial values: - mu: Mean of dosage classes on logit scale: usually c(0,NA,NA, ,NA)

• P: Initial value for proportion in each dosage class

• tau: Precision of means which depends on whether priors are strong or weak

• theta: Differences in means (for parameterisation employed for better convergence)

• seed: Sets seed for each MCMC chain (Default:1)

• taub: If the model contains a random effect then sets initial value of precision of random effect b which is normally distributed with mean 0 and precision taub

Author(s)

Peter Baker p.baker1@uq.edu.au

Note

Warning: If a number of chains are to be produced then several seeds may be specified. Currently, this is largely untested and so it is highly unlikely that this will actually work for all functions in this package.

See Also

setModel setPriors

setControl dumpInits

Examples

## simulate small autooctaploid data set
a1 <- sim.autoMarkers(8,c(0.7,0.2,0.1),n.markers=100,n.individuals=50)

## set up model, priors, inits etc and write files for JAGS
x <- setModel(3,8)
x2 <- setPriors(x)
inits <- setInits(x,x2)
dumpInits(inits)