Integer, Intent (In)	::	m, n, l, nb, lda, ldb, ldt
Integer, Intent (Out)	::	info
Complex (Kind=nag_wp), Intent (Inout)	::	a(lda,), b(ldb,), t(ldt,), work()

C Header Interface

#include nagmk26.h

void	f08bpf_ ( const Integer m, const Integer n, const Integer l, const Integer nb, Complex a[], const Integer lda, Complex b[], const Integer ldb, Complex t[], const Integer ldt, Complex work[], Integer info)

The routine may be called by its LAPACK name ztpqrt.

3

Description

f08bpf (ztpqrt) forms the

Q R

factorization of a complex

(m + n)

n

triangular-pentagonal matrix

C

C = (\begin{matrix} A \\ B \end{matrix})

where

A

is an upper triangular

n

n

matrix and

B

is an

m

n

pentagonal matrix consisting of an

(m - l)

n

rectangular matrix

B_{1}

on top of an

l

n

upper trapezoidal matrix

B_{2}

B = (\begin{matrix} B_{1} \\ B_{2} \end{matrix}) .

The upper trapezoidal matrix

B_{2}

consists of the first

l

rows of an

n

n

upper triangular matrix, where

0 \leq l \leq \min (m, n)

. If

l = 0

B

m

n

rectangular; if

l = n

and

m = n

B

is upper triangular.

A recursive, explicitly blocked,

Q R

factorization (see f08apf (zgeqrt)) is performed on the matrix

C

. The upper triangular matrix

R

, details of the unitary matrix

Q

, and further details (the block reflector factors) of

Q

are returned.

Typically the matrix

A

B_{2}

contains the matrix

R

from the

Q R

factorization of a subproblem and f08bpf (ztpqrt) performs the

Q R

update operation from the inclusion of matrix

B_{1}

For example, consider the

Q R

factorization of an

l

n

matrix

\hat{B}

with

l < n

\hat{B} = \hat{Q} \hat{R}

\hat{R} = (\begin{matrix} \hat{R_{1}} & \hat{R_{2}} \end{matrix})

, where

\hat{R_{1}}

l

l

upper triangular and

\hat{R_{2}}

(n - l)

n

rectangular (this can be performed by f08apf (zgeqrt)). Given an initial least squares problem

\hat{B} \hat{X} = \hat{Y}

where

X

and

Y

are

l

nrhs

matrices, we have

\hat{R} \hat{X} = {\hat{Q}}^{H} \hat{Y}

Now, adding an additional

m - l

rows to the original system gives the augmented least squares problem

B X = Y

where

B

is an

m

n

matrix formed by adding

m - l

rows on top of

\hat{R}

and

Y

is an

m

nrhs

matrix formed by adding

m - l

rows on top of

{\hat{Q}}^{H} \hat{Y}

f08bpf (ztpqrt) can then be used to perform the

Q R

factorization of the pentagonal matrix

B

; the

n

n

matrix

A

will be zero on input and contain

R

on output.

In the case where

\hat{B}

r

n

r \geq n

\hat{R}

n

n

upper triangular (forming

A

) on top of

r - n

rows of zeros (forming first

r - n

rows of

B

). Augmentation is then performed by adding rows to the bottom of

B

with

l = 0

4

References

Elmroth E and Gustavson F (2000) Applying Recursion to Serial and Parallel

Q R

Factorization Leads to Better Performance IBM Journal of Research and Development. (Volume 44) 4 605–624

Golub G H and Van Loan C F (2012) Matrix Computations (4th Edition) Johns Hopkins University Press, Baltimore

5

Arguments

1: $m$ – IntegerInput

On entry:

m

, the number of rows of the matrix

B

Constraint:

m \geq 0

2: $n$ – IntegerInput

On entry:

n

, the number of columns of the matrix

B

and the order of the upper triangular matrix

A

Constraint:

n \geq 0

3: $l$ – IntegerInput

On entry:

l

, the number of rows of the trapezoidal part of

B

(i.e.,

B_{2}

Constraint:

0 \leq l \leq \min (m, n)

4: $nb$ – IntegerInput

On entry: the explicitly chosen block-size to be used in the algorithm for computing the

Q R

factorization. See Section 9 for details.

Constraints:

$nb \geq 1$ ;
if $n > 0$ , $nb \leq n$ .

5: $a (lda *)$ – Complex (Kind=nag_wp) arrayInput/Output

Note: the second dimension of the array a must be at least

\max (1, n)

On entry: the

n

n

upper triangular matrix

A

On exit: the upper triangle is overwritten by the corresponding elements of the

n

n

upper triangular matrix

R

6: $lda$ – IntegerInput

On entry: the first dimension of the array a as declared in the (sub)program from which f08bpf (ztpqrt) is called.

Constraint:

lda \geq \max (1, n)

7: $b (ldb *)$ – Complex (Kind=nag_wp) arrayInput/Output

Note: the second dimension of the array b must be at least

\max (1, n)

On entry: the

m

n

pentagonal matrix

B

composed of an

(m - l)

n

rectangular matrix

B_{1}

above an

l

n

upper trapezoidal matrix

B_{2}

On exit: details of the unitary matrix

Q

8: $ldb$ – IntegerInput

On entry: the first dimension of the array b as declared in the (sub)program from which f08bpf (ztpqrt) is called.

Constraint:

ldb \geq \max (1, m)

9: $t (ldt *)$ – Complex (Kind=nag_wp) arrayOutput

Note: the second dimension of the array t must be at least

n

On exit: further details of the unitary matrix

Q

. The number of blocks is

b = ⌈\frac{k}{nb}⌉

, where

k = \min (m, n)

and each block is of order nb except for the last block, which is of order

k - (b - 1) \times nb

. For each of the blocks, an upper triangular block reflector factor is computed:

T_{1}, T_{2}, \dots, T_{b}

. These are stored in the

nb

n

matrix

T

T = [T_{1} | T_{2} | \dots | T_{b}]

10: $ldt$ – IntegerInput

On entry: the first dimension of the array t as declared in the (sub)program from which f08bpf (ztpqrt) is called.

Constraint:

ldt \geq nb

11: $work (*)$ – Complex (Kind=nag_wp) arrayWorkspace

Note: the dimension of the array work must be at least

nb \times n

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 factorization is the exact factorization of a nearby matrix

(A + E)

, where

{‖E‖}_{2} = O (ε) {‖A‖}_{2},

and

ε

is the machine precision.

8

Parallelism and Performance

f08bpf (ztpqrt) 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 total number of floating-point operations is approximately

\frac{2}{3} n^{2} (3 m - n)

m \geq n

\frac{2}{3} m^{2} (3 n - m)

m < n

The block size, nb, used by f08bpf (ztpqrt) is supplied explicitly through the interface. For moderate and large sizes of matrix, the block size can have a marked effect on the efficiency of the algorithm with the optimal value being dependent on problem size and platform. A value of

nb = 64 ≪ \min (m, n)

is likely to achieve good efficiency and it is unlikely that an optimal value would exceed

340

To apply

Q

to an arbitrary complex rectangular matrix

C

, f08bpf (ztpqrt) may be followed by a call to f08bqf (ztpmqrt). For example,

Call ZTPMQRT('Left','Transpose',m,p,n,l,nb,b,ldb, &
              t,ldt,c,ldc,c(n+1,1),ldc,work,info)

forms

C = Q^{H} C

, where

C

(m + n)

p

To form the unitary matrix

Q

explicitly set

p = m + n

, initialize

C

to the identity matrix and make a call to f08bqf (ztpmqrt) as above.

10

Example

This example finds the basic solutions for the linear least squares problems

minimize {‖A x_{i} - b_{i}‖}_{2}, i = 1, 2

where

b_{1}

and

b_{2}

are the columns of the matrix

B

A = (\begin{array}{r} 0.96 - 0.81 i & - 0.03 + 0.96 i & - 0.91 + 2.06 i & - 0.05 + 0.41 i \\ - 0.98 + 1.98 i & - 1.20 + 0.19 i & - 0.66 + 0.42 i & - 0.81 + 0.56 i \\ 0.62 - 0.46 i & 1.01 + 0.02 i & 0.63 - 0.17 i & - 1.11 + 0.60 i \\ - 0.37 + 0.38 i & 0.19 - 0.54 i & - 0.98 - 0.36 i & 0.22 - 0.20 i \\ 0.83 + 0.51 i & 0.20 + 0.01 i & - 0.17 - 0.46 i & 1.47 + 1.59 i \\ 1.08 - 0.28 i & 0.20 - 0.12 i & - 0.07 + 1.23 i & 0.26 + 0.26 i \end{array}) and

B = (\begin{array}{r} - 2.09 + 1.93 i & 3.26 - 2.70 i \\ 3.34 - 3.53 i & - 6.22 + 1.16 i \\ - 4.94 - 2.04 i & 7.94 - 3.13 i \\ 0.17 + 4.23 i & 1.04 - 4.26 i \\ - 5.19 + 3.63 i & - 2.31 - 2.12 i \\ 0.98 + 2.53 i & - 1.39 - 4.05 i \end{array}) .

Q R

factorization is performed on the first

4

rows of

A

using f08apf (zgeqrt) after which the first

4

rows of

B

are updated by applying

Q^{T}

using f08aqf (zgemqrt). The remaining row is added by performing a

Q R

update using f08bpf (ztpqrt);

B

is updated by applying the new

Q^{T}

using f08bqf (ztpmqrt); the solution is finally obtained by triangular solve using

R

from the updated

Q R

NAG Library Routine Document

f08bpf (ztpqrt)

▸▿ Contents

1 Purpose

2 Specification

3 Description

4 References

5 Arguments

6 Error Indicators and Warnings

7 Accuracy

8 Parallelism and Performance

9 Further Comments

10 Example

10.1 Program Text

10.2 Program Data

10.3 Program Results

1

Purpose

2

Specification

3

Description

4

References

5

Arguments

6

Error Indicators and Warnings

7

Accuracy

8

Parallelism and Performance

9

Further Comments

10

Example

10.1

Program Text

10.2

Program Data

10.3

Program Results