cholPermute: Sparse Cholesky Factorisation with fill-in reducing...

In andrewzm/linalg: Linalg functions

Description

This function is similar to chol(A,pivot=T) when A is a sparse matrix. The fill-in reduction permutation is the approximate minimum degree permutation of Davis' SuiteSparse package configured to be slightly more aggressive than that in the Matrix package. If the Cholesky factor fails, the matrix is coerced to be symmetric.

Usage

cholPermute(Q, method = "amd", matlab_server = NULL)

Arguments

Q	matrix (sparse or dense), the Cholesky factor of which needs to be found
method	If "amd", Timothy Davis SuiteSparse algorithm is used, if not that in the R Matrix package is employed
matlab_server	A matlab server initiated using the R.matlab package by Henrik Bengtsson. Sparse Cholesky factorisation in MATLAB is generally much faster than in R.

Value

A list with two elements, Qpermchol (the permuted Cholesky factor) and P (the pivoting order matrix)

References

Havard Rue and Leonhard Held (2005). Gaussian Markov Random Fields: Theory and Applications. Chapman & Hall/CRC Press

Examples

require(Matrix)
cholPermute(sparseMatrix(i=c(1,1,2,2),j=c(1,2,1,2),x=c(0.1,0.2,0.2,1)))

andrewzm/linalg documentation built on May 10, 2019, 11:15 a.m.