F08 Chapter Contents
F08 Chapter Introduction
NAG Library Manual

# NAG Library Routine DocumentF08ZEF (DGGQRF)

Note:  before using this routine, please read the Users' Note for your implementation to check the interpretation of bold italicised terms and other implementation-dependent details.

## 1  Purpose

F08ZEF (DGGQRF) computes a generalized $QR$ factorization of a real matrix pair $\left(A,B\right)$, where $A$ is an $n$ by $m$ matrix and $B$ is an $n$ by $p$ matrix.

## 2  Specification

 SUBROUTINE F08ZEF ( N, M, P, A, LDA, TAUA, B, LDB, TAUB, WORK, LWORK, INFO)
 INTEGER N, M, P, LDA, LDB, LWORK, INFO REAL (KIND=nag_wp) A(LDA,*), TAUA(min(N,M)), B(LDB,*), TAUB(min(N,P)), WORK(max(1,LWORK))
The routine may be called by its LAPACK name dggqrf.

## 3  Description

F08ZEF (DGGQRF) forms the generalized $QR$ factorization of an $n$ by $m$ matrix $A$ and an $n$ by $p$ matrix $B$
 $A =QR , B=QTZ ,$
where $Q$ is an $n$ by $n$ orthogonal matrix, $Z$ is a $p$ by $p$ orthogonal matrix and $R$ and $T$ are of the form
 $R = mm(R11) n-m 0 , if ​n≥m; nm-nn(R11R12) , if ​n
with ${R}_{11}$ upper triangular,
 $T = p-nnn(0T12) , if ​n≤p, pn-p(T11) p T21 , if ​n>p,$
with ${T}_{12}$ or ${T}_{21}$ upper triangular.
In particular, if $B$ is square and nonsingular, the generalized $QR$ factorization of $A$ and $B$ implicitly gives the $QR$ factorization of ${B}^{-1}A$ as
 $B-1A= ZT T-1 R .$

## 4  References

Anderson E, Bai Z, Bischof C, Blackford S, Demmel J, Dongarra J J, Du Croz J J, Greenbaum A, Hammarling S, McKenney A and Sorensen D (1999) LAPACK Users' Guide (3rd Edition) SIAM, Philadelphia http://www.netlib.org/lapack/lug
Anderson E, Bai Z and Dongarra J (1992) Generalized QR factorization and its applications Linear Algebra Appl. (Volume 162–164) 243–271
Hammarling S (1987) The numerical solution of the general Gauss-Markov linear model Mathematics in Signal Processing (eds T S Durrani, J B Abbiss, J E Hudson, R N Madan, J G McWhirter and T A Moore) 441–456 Oxford University Press
Paige C C (1990) Some aspects of generalized $QR$ factorizations . In Reliable Numerical Computation (eds M G Cox and S Hammarling) 73–91 Oxford University Press

## 5  Parameters

1:     N – INTEGERInput
On entry: $n$, the number of rows of the matrices $A$ and $B$.
Constraint: ${\mathbf{N}}\ge 0$.
2:     M – INTEGERInput
On entry: $m$, the number of columns of the matrix $A$.
Constraint: ${\mathbf{M}}\ge 0$.
3:     P – INTEGERInput
On entry: $p$, the number of columns of the matrix $B$.
Constraint: ${\mathbf{P}}\ge 0$.
4:     A(LDA,$*$) – REAL (KIND=nag_wp) arrayInput/Output
Note: the second dimension of the array A must be at least $\mathrm{max}\phantom{\rule{0.125em}{0ex}}\left(1,{\mathbf{M}}\right)$.
On entry: the $n$ by $m$ matrix $A$.
On exit: the elements on and above the diagonal of the array contain the $\mathrm{min}\phantom{\rule{0.125em}{0ex}}\left(n,m\right)$ by $m$ upper trapezoidal matrix $R$ ($R$ is upper triangular if $n\ge m$); the elements below the diagonal, with the array TAUA, represent the orthogonal matrix $Q$ as a product of $\mathrm{min}\phantom{\rule{0.125em}{0ex}}\left(n,m\right)$ elementary reflectors (see Section 3.3.6 in the F08 Chapter Introduction).
5:     LDA – INTEGERInput
On entry: the first dimension of the array A as declared in the (sub)program from which F08ZEF (DGGQRF) is called.
Constraint: ${\mathbf{LDA}}\ge \mathrm{max}\phantom{\rule{0.125em}{0ex}}\left(1,{\mathbf{N}}\right)$.
6:     TAUA($\mathrm{min}\phantom{\rule{0.125em}{0ex}}\left({\mathbf{N}},{\mathbf{M}}\right)$) – REAL (KIND=nag_wp) arrayOutput
On exit: the scalar factors of the elementary reflectors which represent the orthogonal matrix $Q$.
7:     B(LDB,$*$) – REAL (KIND=nag_wp) arrayInput/Output
Note: the second dimension of the array B must be at least $\mathrm{max}\phantom{\rule{0.125em}{0ex}}\left(1,{\mathbf{P}}\right)$.
On entry: the $n$ by $p$ matrix $B$.
On exit: if $n\le p$, the upper triangle of the subarray ${\mathbf{B}}\left(1:n,p-n+1:p\right)$ contains the $n$ by $n$ upper triangular matrix ${T}_{12}$.
If $n>p$, the elements on and above the $\left(n-p\right)$th subdiagonal contain the $n$ by $p$ upper trapezoidal matrix $T$; the remaining elements, with the array TAUB, represent the orthogonal matrix $Z$ as a product of elementary reflectors (see Section 3.3.6 in the F08 Chapter Introduction).
8:     LDB – INTEGERInput
On entry: the first dimension of the array B as declared in the (sub)program from which F08ZEF (DGGQRF) is called.
Constraint: ${\mathbf{LDB}}\ge \mathrm{max}\phantom{\rule{0.125em}{0ex}}\left(1,{\mathbf{N}}\right)$.
9:     TAUB($\mathrm{min}\phantom{\rule{0.125em}{0ex}}\left({\mathbf{N}},{\mathbf{P}}\right)$) – REAL (KIND=nag_wp) arrayOutput
On exit: the scalar factors of the elementary reflectors which represent the orthogonal matrix $Z$.
10:   WORK($\mathrm{max}\phantom{\rule{0.125em}{0ex}}\left(1,{\mathbf{LWORK}}\right)$) – REAL (KIND=nag_wp) arrayWorkspace
On exit: if ${\mathbf{INFO}}={\mathbf{0}}$, ${\mathbf{WORK}}\left(1\right)$ contains the minimum value of LWORK required for optimal performance.
11:   LWORK – INTEGERInput
On entry: the dimension of the array WORK as declared in the (sub)program from which F08ZEF (DGGQRF) is called.
If ${\mathbf{LWORK}}=-1$, a workspace query is assumed; the routine only calculates the optimal size of the WORK array, returns this value as the first entry of the WORK array, and no error message related to LWORK is issued.
Suggested value: for optimal performance, ${\mathbf{LWORK}}\ge \mathrm{max}\phantom{\rule{0.125em}{0ex}}\left({\mathbf{N}},{\mathbf{M}},{\mathbf{P}}\right)×\mathrm{max}\phantom{\rule{0.125em}{0ex}}\left(\mathit{nb1},\mathit{nb2},\mathit{nb3}\right)$, where $\mathit{nb1}$ is the optimal block size for the $QR$ factorization of an $n$ by $m$ matrix, $\mathit{nb2}$ is the optimal block size for the $RQ$ factorization of an $n$ by $p$ matrix, and $\mathit{nb3}$ is the optimal block size for a call of F08AGF (DORMQR).
Constraint: ${\mathbf{LWORK}}\ge \mathrm{max}\phantom{\rule{0.125em}{0ex}}\left(1,{\mathbf{N}},{\mathbf{M}},{\mathbf{P}}\right)$ or ${\mathbf{LWORK}}=-1$.
12:   INFO – INTEGEROutput
On exit: ${\mathbf{INFO}}=0$ unless the routine detects an error (see Section 6).

## 6  Error Indicators and Warnings

Errors or warnings detected by the routine:
${\mathbf{INFO}}<0$
If ${\mathbf{INFO}}=-i$, argument $i$ had an illegal value. An explanatory message is output, and execution of the program is terminated.

## 7  Accuracy

The computed generalized $QR$ factorization is the exact factorization for nearby matrices $\left(A+E\right)$ and $\left(B+F\right)$, where
 $E2 = O⁡ε A2 and F2= O⁡ε B2 ,$
and $\epsilon$ is the machine precision.

The orthogonal matrices $Q$ and $Z$ may be formed explicitly by calls to F08AFF (DORGQR) and F08CJF (DORGRQ) respectively. F08AGF (DORMQR) may be used to multiply $Q$ by another matrix and F08CKF (DORMRQ) may be used to multiply $Z$ by another matrix.
The complex analogue of this routine is F08ZSF (ZGGQRF).

## 9  Example

This example solves the general Gauss–Markov linear model problem
 $minx y2 subject to d=Ax+By$
where
 $A = -0.57 -1.28 -0.39 -1.93 1.08 -0.31 2.30 0.24 -0.40 -0.02 1.03 -1.43 , B= 0.5 0.0 0.0 0.0 0.0 1.0 0.0 0.0 0.0 0.0 2.0 0.0 0.0 0.0 0.0 5.0 and d= 1.32 -4.00 5.52 3.24 .$
The solution is obtained by first computing a generalized $QR$ factorization of the matrix pair $\left(A,B\right)$. The example illustrates the general solution process, although the above data corresponds to a simple weighted least squares problem.
Note that the block size (NB) of $64$ assumed in this example is not realistic for such a small problem, but should be suitable for large problems.

### 9.1  Program Text

Program Text (f08zefe.f90)

### 9.2  Program Data

Program Data (f08zefe.d)

### 9.3  Program Results

Program Results (f08zefe.r)