Integer, Intent (In)	::	ijob, m, n, lda, ldb, ldc, ldd, lde, ldf, lwork
Integer, Intent (Out)	::	iwork(m+n+2), info
Real (Kind=nag_wp), Intent (Out)	::	scale, dif
Complex (Kind=nag_wp), Intent (In)	::	a(lda,), b(ldb,), d(ldd,), e(lde,)
Complex (Kind=nag_wp), Intent (Inout)	::	c(ldc,), f(ldf,)
Complex (Kind=nag_wp), Intent (Out)	::	work(max(1,lwork))
Character (1), Intent (In)	::	trans

C Header Interface

#include nagmk26.h

void

f08yvf_ ( const char *trans, const Integer *ijob, const Integer *m, const Integer *n, const Complex a[], const Integer *lda, const Complex b[], const Integer *ldb, Complex c[], const Integer *ldc, const Complex d[], const Integer *ldd, const Complex e[], const Integer *lde, Complex f[], const Integer *ldf, double *scal, double *dif, Complex work[], const Integer *lwork, Integer iwork[], Integer *info, const Charlen length_trans)

The routine may be called by its LAPACK name ztgsyl.

3

Description

f08yvf (ztgsyl) solves either the generalized complex Sylvester equations

\begin{matrix} A R - L B & = α C \\ D R - L E & = α F, \end{matrix}

(1)

or the equations

\begin{matrix} A^{H} R + D^{H} L & = α C \\ R B^{H} + L E^{H} & = - α F, \end{matrix}

(2)

where the pair

(A, D)

are given

m

m

matrices in generalized Schur form,

(B, E)

are given

n

n

matrices in generalized Schur form and

(C, F)

are given

m

n

matrices. The pair

(R, L)

are the

m

n

solution matrices, and

α

is an output scaling factor determined by the routine to avoid overflow in computing

(R, L)

Equations (1) are equivalent to equations of the form

Z x = α b,

where

Z = (\begin{matrix} I \otimes A - B^{H} \otimes I \\ I \otimes D - E^{H} \otimes I \end{matrix})

and

\otimes

is the Kronecker product. Equations (2) are then equivalent to

Z^{H} y = α b .

The pair

(S, T)

are in generalized Schur form if

S

and

T

are upper triangular as returned, for example, by f08xnf (zgges), or f08xsf (zhgeqz) with

job ='S'

Optionally, the routine estimates

Dif [(A, D), (B, E)]

, the separation between the matrix pairs

(A, D)

and

(B, E)

, which is the smallest singular value of

Z

. The estimate can be based on either the Frobenius norm, or the

1

-norm. The

1

-norm estimate can be three to ten times more expensive than the Frobenius norm estimate, but makes the condition estimation uniform with the nonsymmetric eigenproblem. The Frobenius norm estimate provides a low cost, but equally reliable estimate. For more information see Sections 2.4.8.3 and 4.11.1.3 of Anderson et al. (1999) and Kågström and Poromaa (1996).

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

Kågström B (1994) A perturbation analysis of the generalized Sylvester equation

(A R - L B, D R - L E) = (c, F)

SIAM J. Matrix Anal. Appl. 15 1045–1060

Kågström B and Poromaa P (1996) LAPACK-style algorithms and software for solving the generalized Sylvester equation and estimating the separation between regular matrix pairs ACM Trans. Math. Software 22 78–103

5

Arguments

1: $trans$ – Character(1)Input

On entry: if

trans ='N'

, solve the generalized Sylvester equation (1).

trans ='C'

, solve the ‘conjugate transposed’ system (2).

Constraint:

trans ='N'

'C'

2: $ijob$ – IntegerInput

On entry: specifies what kind of functionality is to be performed when

trans ='N'

$ijob = 0$: Solve (1) only.
$ijob = 1$: The functionality of $ijob = 0$ and $3$ .
$ijob = 2$: The functionality of $ijob = 0$ and $4$ .
$ijob = 3$: Only an estimate of $Dif [(A, D), (B, E)]$ is computed based on the Frobenius norm.
$ijob = 4$: Only an estimate of $Dif [(A, D), (B, E)]$ is computed based on the $1$ -norm.

trans ='C'

, ijob is not referenced.

Constraint: if

trans ='N'

0 \leq ijob \leq 4

3: $m$ – IntegerInput

On entry:

m

, the order of the matrices

A

and

D

, and the row dimension of the matrices

C

F

R

and

L

Constraint:

m > 0

4: $n$ – IntegerInput

On entry:

n

, the order of the matrices

B

and

E

, and the column dimension of the matrices

C

F

R

and

L

Constraint:

n > 0

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

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

\max (1, m)

On entry: the upper triangular matrix

A

6: $lda$ – IntegerInput

On entry: the first dimension of the array a as declared in the (sub)program from which f08yvf (ztgsyl) is called.

Constraint:

lda \geq \max (1, m)

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

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

\max (1, n)

On entry: the upper triangular matrix

B

8: $ldb$ – IntegerInput

On entry: the first dimension of the array b as declared in the (sub)program from which f08yvf (ztgsyl) is called.

Constraint:

ldb \geq \max (1, n)

9: $c (ldc *)$ – Complex (Kind=nag_wp) arrayInput/Output

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

\max (1, n)

On entry: contains the right-hand-side matrix

C

On exit: if

ijob = 0

1

2

, c is overwritten by the solution matrix

R

trans ='N'

and

ijob = 3

4

, c holds

R

, the solution achieved during the computation of the Dif estimate.

10: $ldc$ – IntegerInput

On entry: the first dimension of the array c as declared in the (sub)program from which f08yvf (ztgsyl) is called.

Constraint:

ldc \geq \max (1, m)

11: $d (ldd *)$ – Complex (Kind=nag_wp) arrayInput

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

\max (1, m)

On entry: the upper triangular matrix

D

12: $ldd$ – IntegerInput

On entry: the first dimension of the array d as declared in the (sub)program from which f08yvf (ztgsyl) is called.

Constraint:

ldd \geq \max (1, m)

13: $e (lde *)$ – Complex (Kind=nag_wp) arrayInput

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

\max (1, n)

On entry: the upper triangular matrix

E

14: $lde$ – IntegerInput

On entry: the first dimension of the array e as declared in the (sub)program from which f08yvf (ztgsyl) is called.

Constraint:

lde \geq \max (1, n)

15: $f (ldf *)$ – Complex (Kind=nag_wp) arrayInput/Output

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

\max (1, n)

On entry: contains the right-hand side matrix

F

On exit: if

ijob = 0

1

2

, f is overwritten by the solution matrix

L

trans ='N'

and

ijob = 3

4

, f holds

L

, the solution achieved during the computation of the Dif estimate.

16: $ldf$ – IntegerInput

On entry: the first dimension of the array f as declared in the (sub)program from which f08yvf (ztgsyl) is called.

Constraint:

ldf \geq \max (1, m)

17: $scale$ – Real (Kind=nag_wp)Output

On exit:

α

, the scaling factor in (1) or (2).

0 < scale < 1

, c and f hold the solutions

R

and

L

, respectively, to a slightly perturbed system but the input arrays a, b, d and e have not been changed.

scale = 0

, c and f hold the solutions

R

and

L

, respectively, to the homogeneous system with

C = F = 0

. In this case dif is not referenced.

Normally,

scale = 1

18: $dif$ – Real (Kind=nag_wp)Output

On exit: the estimate of

Dif

. If

ijob = 0

, dif is not referenced.

19: $work (\max (1, lwork))$ – Complex (Kind=nag_wp) arrayWorkspace

On exit: if

info = 0

, the real part of

work (1)

contains the minimum value of lwork required for optimal performance.

20: $lwork$ – IntegerInput

On entry: the dimension of the array work as declared in the (sub)program from which f08yvf (ztgsyl) is called.

lwork = - 1

, a workspace query is assumed; the routine only calculates the minimum 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.

Constraints:

lwork \neq - 1

if $trans ='N'$ and $ijob = 1$ or $2$ , $lwork \geq 2 \times m \times n$ ;
otherwise $lwork \geq 1$ .

21: $iwork (m + n + 2)$ – Integer arrayWorkspace

22: $info$ – 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.

$info > 0$: $(A, D)$ and $(B, E)$ have common or close eigenvalues and so no solution could be computed.

7

Accuracy

See Kågström (1994) for a perturbation analysis of the generalized Sylvester equation.

8

Parallelism and Performance

f08yvf (ztgsyl) 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 needed to solve the generalized Sylvester equations is approximately

8 m n (n + m)

. The Frobenius norm estimate of

Dif

does not require additional significant computation, but the

1

-norm estimate is typically five times more expensive.

The real analogue of this routine is f08yhf (dtgsyl).

10

Example

This example solves the generalized Sylvester equations

\begin{matrix} A R - L B & = α C \\ D R - L E & = α F, \end{matrix}

where

A = (\begin{array}{r} 4.0 + 4.0 i & 1.0 + 1.0 i & 1.0 + 1.0 i & 2.0 - 1.0 i \\ 0 & 2.0 + 1.0 i & 1.0 + 1.0 i & 1.0 + 1.0 i \\ 0 & 0 & 2.0 - 1.0 i & 1.0 + 1.0 i \\ 0 & 0 & 0 & 6.0 - 2.0 i \end{array}),

B = (\begin{array}{r} 2.0 & 1.0 + 1.0 i & 1.0 + 1.0 i & 3.0 - 1.0 i \\ 0 & 1.0 & 2.0 + 1.0 i & 1.0 + 1.0 i \\ 0 & 0 & 1.0 & 1.0 + 1.0 i \\ 0 & 0 & 0 & 2.0 \end{array}),

D = (\begin{array}{r} 1.0 + 1.0 i & 1.0 - 1.0 i & 1.0 + 1.0 i & 1.0 - 1.0 i \\ 0 & 6.0 - 4.0 i & 1.0 - 1.0 i & 1.0 + 1.0 i \\ 0 & 0 & 2.0 + 4.0 i & 1.0 - 1.0 i \\ 0 & 0 & 0 & 2.0 + 3.0 i \end{array}),

E = (\begin{array}{r} 1.0 & 1.0 + 1.0 i & 1.0 - 1.0 i & 1.0 + 1.0 i \\ 0 & 2.0 & 1.0 + 1.0 i & 1.0 - 1.0 i \\ 0 & 0 & 2.0 & 1.0 + 1.0 i \\ 0 & 0 & 0 & 1.0 \end{array}),

C = (\begin{array}{r} - 13.0 + 9.0 i & 2.0 + 8.0 i & - 2.0 + 8.0 i & - 2.0 + 5.0 i \\ - 9.0 - 1.0 i & 0.0 + 5.0 i & - 7.0 - 3.0 i & - 6.0 - 0.0 i \\ - 1.0 + 1.0 i & 4.0 + 2.0 i & 4.0 - 5.0 i & 9.0 - 5.0 i \\ - 6.0 + 6.0 i & 9.0 + 1.0 i & - 2.0 + 4.0 i & 22.0 - 8.0 i \end{array})

and

F = (\begin{array}{r} - 6.0 + 5.0 i & 4.0 - 4.0 i & - 3.0 + 11.0 i & 3.0 - 7.0 i \\ - 5.0 + 11.0 i & 12.0 - 4.0 i & - 2.0 + 2.0 i & 0.0 + 14.0 i \\ - 5.0 - 1.0 i & 0.0 + 4.0 i & - 2.0 + 10.0 i & 3.0 - 1.0 i \\ - 6.0 - 2.0 i & 1.0 + 1.0 i & - 7.0 - 3.0 i & 4.0 + 7.0 i \end{array}) .

NAG Library Routine Document

f08yvf (ztgsyl)

▸▿ 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