HiDimMaxStable (version 0.1.1)

maxstable.l.clusters: Partition-composite likelihood for multivariate max-stable distributions

Description

Computes the partition-composite likelihood for observations sampled from a multivariate max-stable distribution whose spectral random vector is Gaussian, Log-normal or has a clustered copula distribution.

Usage

maxstable.l.clusters(data,clusters=rep(1,dim(data)[1]),ln=FALSE,spatial=NULL,...)

Arguments

data

a matrix representing the data. Each column corresponds to one observation.

clusters

a vector of integers that gives the partition that is used to compute the partition-composite likelihood. Blocks of the partition should be of size smaller or equal to 7 to avoid a too long computing time. clusters=rep(1,dim(data)[1]) must be used to get the full likelihood.

ln

logical. If TRUE log-density is computed.

spatial

argument passed to the mubz.* function (where * stands for the category of the model).

...

further arguments to be passed to mubz.* function (where * stands for the category of the model). In particular, category is a character string indicating the model to be used: "normal", "lnormal" or "copula", and params gives the values of the parameters for which the likelihood is computed.

See Also

mubz.normal,mubz.lnormal, mubz.copula.

Examples

n.site<-5
xy<-matrix(runif(2 * n.site, 0, 0.5), ncol = 2)

param<-c(0.5,1.5)
n.obs<-2
library(SpatialExtremes)
data<-t(rmaxstab(n.obs, xy, "whitmat",
                 nugget = 0, range = param[1], smooth = param[2]))

d<-maxstable.l.clusters(data,clusters=c(1,1,1,2,2),
                        params=param,
                        category="normal",
                        spatial=list(sites=xy,family=spatialWhittleMatern))