triplot: Plot density of a 3D hyperdirichlet distribution

In hyperdirichlet: A Generalization of the Dirichlet Distribution

Description

Plot (log) density of a 3D hyperdirichlet distribution

Usage

triplot(HD, l = 100, do_image=TRUE, do_contour=TRUE, discard = 0.05, labels=NULL, ...)

Arguments

HD	Either a function of a three-element vector, or a hyperdirichlet object
l	Size of plot; larger values look much better but take longer
do_image,do_contour	Boolean, indicating whether to plot the image and contour respectively
discard	Numeric; default 0.05 means to disregard densities less than the fifth percentile (i.e. quantile 0.05). This makes the contour plot intervals prettier
labels	Length three character vector for corner labels with default NULL meaning to use pnames(HD)
...	Further arguments passed to image() and contour()

Details

If argument HD is not a function it is interpreted as a hyperdirichlet object. If HD is a function, triplot() will pass it a three-element vector.

If argument HD is a hyperdirichlet object, then triplot() plots contours of the log of the density (ie support). If you want likelihood instead, see examples section for how to do this.

Labels will be printed at the appropriate corners; use par(xpd=TRUE) to suppress clipping.

Value

The function returns (invisibly) an l-by-l matrix holding the support as a function of the two independent variables, with NA outside the domain.

Author(s)

Robin K. S. Hankin

Examples

a <- dirichlet(1:3)
triplot(a,l=20)

old.par <- par(no.readonly=TRUE);par(xpd=TRUE)   # (needed for labels)


pnames(a) <- letters[1:3]
triplot(a,l=33)

f <- function(p){1-p[1]+p[2]+8*p[2]^2*p[3]}
triplot(f,labels=LETTERS[1:3])

g <- function(p){dhyperdirichlet(p, a, log=FALSE)}
triplot(g,l=20)



par(old.par)

hyperdirichlet documentation built on May 31, 2017, 5:18 a.m.