coord2Sig: Coordinate descent algorithm for the MLE.

In pzwiernik/mtp2: Working with Gaussian totally positive distributions

Description

This function implements a simple coordinate descent algorithm to find the maximum likelihood estimator over Gaussian MTP2 distributions. The algorithm performs local updates of the covariance matrix $Sigma$. For details see Lauritzen, Uhler, Zwiernik (2017); look for Algorithm 4.

Usage

coord2Sig(S, n = 1, Sig0 = S, tol = 1e-10, graph = FALSE)

Arguments

S	the sample covariance matrix
n	the sample size (default 1)
Sig0	starting point of the algorithm
tol	convergence criterion sum(abs(Sig-Sig0))<tol
graph	a graph to which the procedure restricts. If FALSE, the complete graph is taken.

Examples

library(gRbase)
data("carcass")
S <- cov(carcass)[1:6,1:6]
coord2Sig(S)

pzwiernik/mtp2 documentation built on Aug. 9, 2020, 12:34 p.m.