NAG Library Routine Document

f08zff (dggrqf)

1
Purpose

f08zff (dggrqf) computes a generalized RQ factorization of a real matrix pair A,B, where A is an m by n matrix and B is a p by n matrix.

2
Specification

Fortran Interface
Subroutine f08zff ( m, p, n, a, lda, taua, b, ldb, taub, work, lwork, info)
Integer, Intent (In):: m, p, n, lda, ldb, lwork
Integer, Intent (Out):: info
Real (Kind=nag_wp), Intent (Inout):: a(lda,*), b(ldb,*)
Real (Kind=nag_wp), Intent (Out):: taua(min(m,n)), taub(min(p,n)), work(max(1,lwork))
C Header Interface
#include <nagmk26.h>
void  f08zff_ (const Integer *m, const Integer *p, const Integer *n, double a[], const Integer *lda, double taua[], double b[], const Integer *ldb, double taub[], double work[], const Integer *lwork, Integer *info)
The routine may be called by its LAPACK name dggrqf.

3
Description

f08zff (dggrqf) forms the generalized RQ factorization of an m by n matrix A and a p by n matrix B 
A = RQ ,   B= ZTQ ,  
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 = n-mmm0R12() ;   if ​ mn , nm-nR11nR21() ;   if ​ m>n ,  
with R12 or R21 upper triangular,
T = nnT11p-n0() ;   if ​ pn , pn-ppT11T12() ;   if ​ p<n ,  
with T11 upper triangular.
In particular, if B is square and nonsingular, the generalized RQ factorization of A and B implicitly gives the RQ factorization of AB-1 as
AB-1= R T-1 ZT .  

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
Arguments

1:     m – IntegerInput
On entry: m, the number of rows of the matrix A.
Constraint: m0.
2:     p – IntegerInput
On entry: p, the number of rows of the matrix B.
Constraint: p0.
3:     n – IntegerInput
On entry: n, the number of columns of the matrices A and B.
Constraint: n0.
4:     alda* – Real (Kind=nag_wp) arrayInput/Output
Note: the second dimension of the array a must be at least max1,n.
On entry: the m by n matrix A.
On exit: if mn, the upper triangle of the subarray a1:mn-m+1:n contains the m by m upper triangular matrix R12.
If mn, the elements on and above the m-nth subdiagonal contain the m by n upper trapezoidal matrix R; the remaining elements, with the array taua, represent the orthogonal matrix Q as a product of minm,n 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 f08zff (dggrqf) is called.
Constraint: ldamax1,m.
6:     tauaminm,n – Real (Kind=nag_wp) arrayOutput
On exit: the scalar factors of the elementary reflectors which represent the orthogonal matrix Q.
7:     bldb* – Real (Kind=nag_wp) arrayInput/Output
Note: the second dimension of the array b must be at least max1,n.
On entry: the p by n matrix B.
On exit: the elements on and above the diagonal of the array contain the minp,n by n upper trapezoidal matrix T (T is upper triangular if pn); the elements below the diagonal, 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 f08zff (dggrqf) is called.
Constraint: ldbmax1,p.
9:     taubminp,n – Real (Kind=nag_wp) arrayOutput
On exit: the scalar factors of the elementary reflectors which represent the orthogonal matrix Z.
10:   workmax1,lwork – Real (Kind=nag_wp) arrayWorkspace
On exit: if info=0, work1 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 f08zff (dggrqf) is called.
If 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, lworkmaxn,m,p×maxnb1,nb2,nb3, where nb1 is the optimal block size for the RQ factorization of an m by n matrix by f08chf (dgerqf), nb2 is the optimal block size for the QR factorization of a p by n matrix by f08aef (dgeqrf), and nb3 is the optimal block size for a call of f08ckf (dormrq).
Constraint: lworkmax1,n,m,p or lwork=-1.
12:   info – IntegerOutput
On exit: info=0 unless the routine detects an error (see Section 6).

6
Error Indicators and Warnings

info<0
If 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 RQ factorization is the exact factorization for nearby matrices A+E and B+F, where
E2 = Oε A2   and   F2= Oε B2 ,  
and ε is the machine precision.

8
Parallelism and Performance

f08zff (dggrqf) is threaded by NAG for parallel execution in multithreaded implementations of the NAG Library.
f08zff (dggrqf) makes calls to BLAS and/or LAPACK routines, which may be threaded within the vendor library used by this implementation. Consult the documentation for the vendor library for further information.
Please consult the X06 Chapter Introduction for information on how to control and interrogate the OpenMP environment used within this routine. Please also consult the Users' Note for your implementation for any additional implementation-specific information.

9
Further Comments

The orthogonal matrices Q and Z may be formed explicitly by calls to f08cjf (dorgrq) and f08aff (dorgqr) respectively. f08ckf (dormrq) may be used to multiply Q by another matrix and f08agf (dormqr) may be used to multiply Z by another matrix.
The complex analogue of this routine is f08ztf (zggrqf).

10
Example

This example solves the least squares problem
minimize x c-Ax2   subject to   Bx=d  
where
A = -0.57 -1.28 -0.39 0.25 -1.93 1.08 -0.31 -2.14 2.30 0.24 0.40 -0.35 -1.93 0.64 -0.66 0.08 0.15 0.30 0.15 -2.13 -0.02 1.03 -1.43 0.50 ,   B= 1 0 -1 0 0 1 0 -1 ,  
c = -1.50 -2.14 1.23 -0.54 -1.68 0.82   and   d= 0 0 .  
The constraints Bx=d correspond to x1=x3 and x2=x4.
The solution is obtained by first computing a generalized RQ factorization of the matrix pair B,A. The example illustrates the general solution process.
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.

10.1
Program Text

Program Text (f08zffe.f90)

10.2
Program Data

Program Data (f08zffe.d)

10.3
Program Results

Program Results (f08zffe.r)