The R Project SVN R

\name{qr}
\title{The QR Decomposition of a Matrix}
\usage{
qr(x, tol=1e-07)
qr.coef(qr, y)
qr.qy(qr, y)
qr.qty(qr, y)
qr.resid(qr, y)
qr.fitted(qr, y, k = qr$rank)
qr.solve(a, b, tol = 1e-7)

is.qr(x)
as.qr(x)
}
\alias{qr}
\alias{qr.coef}
\alias{qr.qy}
\alias{qr.qty}
\alias{qr.resid}
\alias{qr.fitted}
\alias{qr.solve}
\alias{is.qr}
\alias{as.qr}
\arguments{
  \item{x}{a matrix whose QR decomposition is to be computed.}
  \item{tol}{the tolerance for detecting linear dependencies in the
    columns of \code{x}.}
  \item{qr}{a QR decomposition of the type computed by \code{qr}.}
  \item{y, b}{a vector or matrix of right-hand sides of equations.}
  \item{a}{A matrix or QR decomposition.}
}
\description{
  \code{qr} computes the QR decomposition of a matrix.  It provides an
  interface to the techniques used in the LINPACK routine DQRDC.  
}
\details{
  The QR decomposition plays an important role in many
  statistical techniques.  In particular it can be used to solve the
  equation \eqn{\bold{Ax} = \bold{b}} for given matrix \eqn{\bold{A}},
  and vector \eqn{\bold{b}}.  It is useful for computing regression
  coefficients and in applying the Newton-Raphson algorithm.

  The functions \code{qr.coef}, \code{qr.resid}, and \code{qr.fitted}
  return the coefficients, residuals and fitted values obtained when
  fitting \code{y} to the matrix with QR decomposition \code{qr}.
  \code{qr.qy} and \code{qr.qty} return \code{Q \%*\% y} and
  \code{t(Q) \%*\% y}, where \code{Q} is the \eqn{\bold{Q}} matrix.

  \code{qr.solve} solves systems of equations via the QR decomposition.

  \code{is.qr} returns \code{TRUE} if \code{x} is a list with a
  component named \code{qr} and \code{FALSE} otherwise.

  It is not possible to coerce objects to mode \code{"qr"}.  Objects
  either are QR decompositions or they are not.
}
\value{
  The QR decomposition of the matrix as computed by LINPACK.
  The components in the returned value correspond directly
  to the values returned by DQRDC.
  \item{qr}{a matrix with the same dimensions as \code{x}.
    The upper triangle contains the \eqn{\bold{R}} of the decomposition
    and the lower triangle contains information on the \eqn{\bold{Q}} of
    the decomposition (stored in compact form).}
  \item{qraux}{a vector of length \code{ncol(x)} which contains
    additional information on \eqn{\bold{Q}}.}
  \item{rank}{the rank of \code{x} as computed by the decomposition.}
  \item{pivot}{information on the pivoting strategy used during
    the decomposition.}
}
\note{To compute the determinant of a matrix (do you \emph{really} need it?),
    the QR decomposition is much more efficient than using Eigen values
    (\code{\link{eigen}}).  See \code{det2} in the examples below.
}
\references{
  Dongarra, J. J., J. R. Bunch, C. B. Moler and G. W. Stewart (1978).
  \emph{LINPACK Users Guide}.
  Philadelphia, PA:  SIAM Publications.
}
\seealso{
  \code{\link{qr.Q}},  \code{\link{qr.R}},  \code{\link{qr.X}} for
  reconstruction of the matrices.
  \code{\link{solve.qr}},  \code{\link{lsfit}},
  \code{\link{eigen}}, \code{\link{svd}}.}
\examples{
## The determinant of a matrix  -- if you really must have it
det2 <- function(x) prod(diag(qr(x)$qr))*(-1)^(ncol(x)-1)
det2(print(cbind(1,1:3,c(2,0,1))))

hilbert <- function(n) { i <- 1:n; 1 / outer(i - 1, i, "+") }
h9 <- hilbert(9); h9
qr(h9)$rank           #--> only 7
qrh9 <- qr(h9, tol = 1e-10)
qrh9$rank             #--> 9
##-- Solve linear equation system  H \%*\% x = y :
y <- 1:9/10
x <- qr.solve(h9, y, tol = 1e-10) # or equivalently :
x <- qr.coef(qrh9, y) #-- is == but much better than
                      #-- solve(h9) \%*\% y
h9 \%*\% x              # = y
}
\keyword{algebra}
\keyword{array}
\keyword{regression}