GenQR: Generalized QR decomposition

In RegSDC: Information Preserving Regression-Based Tools for Statistical Disclosure Control

Generalized QR decomposition

Description

Matrix X decomposed as Q and R (X=QR) where columns of Q are orthonormal. Ordinary QR or SVD may be used.

Usage

GenQR(x, doSVD = FALSE, findR = TRUE, makeunique = findR, tol = 1e-07)

Arguments

x	Matrix to be decomposed
doSVD	When TRUE SVD instead of QR
findR	When FALSE only Q returned
makeunique	When TRUE force uniqueness by positive diagonal elements (QR) or by column sums (SVD)
tol	As input to qr or, in the case of svd(), similar as input to MASS::ginv().

Details

To handle dependency a usual decomposition of X is PX=QR where P is a permutation matrix. This function returns RP^T as R. When SVD, Q=U and R=SV^T.

Value

List with Q and R or just Q

Author(s)

Øyvind Langsrud

Examples

   GenQR(matrix(rnorm(15),5,3))
   GenQR(matrix(rnorm(15),5,3)[,c(1,2,1,3)])
   GenQR(matrix(rnorm(15),5,3)[,c(1,2,1,3)],TRUE)

RegSDC documentation built on Aug. 19, 2022, 9:08 a.m.