multscatter: Function to Compute Several Scatter Matrices for the Same...

In JADE: Blind Source Separation Methods Based on Joint Diagonalization and Some BSS Performance Criteria

Function to Compute Several Scatter Matrices for the Same Data

Description

The function can be used to compute several scatter matrices for the same data.

Usage

multscatter(scatterlist, X, toshape = TRUE)

Arguments

scatterlist	a vector with the names of the scatter matrices to be computed. Note that each of these functions should only return a matrix of size p times p.
X	the n times p data matrix for which the scatter should be computed.
toshape	logical, whether scatter matrices should be converted to shape matrices. If TRUE, all matrices will have determinant 1.

Details

It is important that the functions do not need any additional imput and that they return only the p times p scatter matrix. Hence it might be sometimes necessary to write wrappers for some of the functions. See examples.

Value

An array of dimension c(p,p,k) where k is the number of scatter matrices.

Author(s)

Klaus Nordhausen

Examples

# example requires the packages ICS and ICSNP
library(ICSNP)
X <- cbind(rexp(1000), rt(1000,6), runif(1000)) 

my.tM1 <- function(X,df=1) tM(X,)$V
my.tM2 <- function(X,df=2) tM(X,)$V

multscatter(c("cov","cov4","HP1.shape","my.tM1", "my.tM2"), X)
multscatter(c("cov","cov4","HP1.shape","my.tM1", "my.tM2"), X, toshape=FALSE)

JADE documentation built on Sept. 18, 2023, 1:06 a.m.