calibrate.rjmcmc: initialize proposal width

Description

estimating a reasonable proposal width to initiate sampling for Markov sampling

Usage

calibrate.rjmcmc(phy, dat, nstep = 10000, widths = 2^(-3:3), type=c("bm", 
    "rbm", "jump-bm", "jump-rbm"), ...)

Arguments

phy: a phylogenetic tree of class 'phylo'
dat: a named vector of continuous trait values, associated with each species in phy
nstep: number of proposal steps over which to assess proposal widths
widths: if unspecified, a series of proposal widths from 1/8 to 8 will be considered
type: a model type available in rjmcmc.bm
...: arguments to be passed to make.gbm and rjmcmc.bm

Author

JM Eastman

Details

This function may be useful for constraining subsequent runs after an adequate proposal width has been approximated. MCMC samples from this calibration are not stored and do not become available to the user. This function is solely used to give the user a sense of acceptance rates that can be expected for different proposal widths. The narrower the width, the more easily the chain can become stuck. With a wider width, the chain will more quickly explore a broader parameter space, yet acceptance rates may become unacceptably low.

Examples

Run this code

# \donttest{
n=40
phy=rcoal(n=n)
dat=rTraitCont(phy=phy, model="BM", sigma=sqrt(0.1))
r=paste(sample(letters,9,replace=TRUE),collapse="")

## calibrate proposal width
calibrate.rjmcmc(phy=phy, dat=dat, nstep=500, widths=2^(-3:0), type="rbm")	
# }