NAG Library Routine Document

f08pbf  (dgeesx)

 Contents

    1  Purpose
    7  Accuracy

1
Purpose

f08pbf (dgeesx) computes the eigenvalues, the real Schur form T, and, optionally, the matrix of Schur vectors Z for an n by n real nonsymmetric matrix A.

2
Specification

Fortran Interface
Subroutine f08pbf ( jobvs, sort, select, sense, n, a, lda, sdim, wr, wi, vs, ldvs, rconde, rcondv, work, lwork, iwork, liwork, bwork, info)
Integer, Intent (In):: n, lda, ldvs, lwork, liwork
Integer, Intent (Out):: sdim, iwork(max(1,liwork)), info
Real (Kind=nag_wp), Intent (Inout):: a(lda,*), wr(*), wi(*), vs(ldvs,*)
Real (Kind=nag_wp), Intent (Out):: rconde, rcondv, work(max(1,lwork))
Logical, External:: select
Logical, Intent (Inout):: bwork(*)
Character (1), Intent (In):: jobvs, sort, sense
C Header Interface
#include nagmk26.h
void  f08pbf_ ( const char *jobvs, const char *sort,
logical (NAG_CALL *sel)( const double *wr, const double *wi),
const char *sense, const Integer *n, double a[], const Integer *lda, Integer *sdim, double wr[], double wi[], double vs[], const Integer *ldvs, double *rconde, double *rcondv, double work[], const Integer *lwork, Integer iwork[], const Integer *liwork, logical bwork[], Integer *info, const Charlen length_jobvs, const Charlen length_sort, const Charlen length_sense)
The routine may be called by its LAPACK name dgeesx.

3
Description

The real Schur factorization of A is given by
A = Z T ZT ,  
where Z, the matrix of Schur vectors, is orthogonal and T is the real Schur form. A matrix is in real Schur form if it is upper quasi-triangular with 1 by 1 and 2 by 2 blocks. 2 by 2 blocks will be standardized in the form
a b c a  
where bc<0. The eigenvalues of such a block are a±bc.
Optionally, f08pbf (dgeesx) also orders the eigenvalues on the diagonal of the real Schur form so that selected eigenvalues are at the top left; computes a reciprocal condition number for the average of the selected eigenvalues (rconde); and computes a reciprocal condition number for the right invariant subspace corresponding to the selected eigenvalues (rcondv). The leading columns of Z form an orthonormal basis for this invariant subspace.
For further explanation of the reciprocal condition numbers rconde and rcondv, see Section 4.8 of Anderson et al. (1999) (where these quantities are called s and sep respectively).

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
Golub G H and Van Loan C F (1996) Matrix Computations (3rd Edition) Johns Hopkins University Press, Baltimore

5
Arguments

1:     jobvs – Character(1)Input
On entry: if jobvs='N', Schur vectors are not computed.
If jobvs='V', Schur vectors are computed.
Constraint: jobvs='N' or 'V'.
2:     sort – Character(1)Input
On entry: specifies whether or not to order the eigenvalues on the diagonal of the Schur form.
sort='N'
Eigenvalues are not ordered.
sort='S'
Eigenvalues are ordered (see select).
Constraint: sort='N' or 'S'.
3:     select – Logical Function, supplied by the user.External Procedure
If sort='S', select is used to select eigenvalues to sort to the top left of the Schur form.
If sort='N', select is not referenced and f08pbf (dgeesx) may be called with the dummy function f08paz.
An eigenvalue wrj+-1×wij is selected if selectwrj,wij is .TRUE.. If either one of a complex conjugate pair of eigenvalues is selected, then both are. Note that a selected complex eigenvalue may no longer satisfy selectwrj,wij=.TRUE. after ordering, since ordering may change the value of complex eigenvalues (especially if the eigenvalue is ill-conditioned); in this case info is set to n+2 (see info below).
The specification of select is:
Fortran Interface
Function select ( wr, wi)
Logical:: select
Real (Kind=nag_wp), Intent (In):: wr, wi
C Header Interface
#include nagmk26.h
Nag_Boolean  select ( const double *wr, const double *wi)
1:     wr – Real (Kind=nag_wp)Input
2:     wi – Real (Kind=nag_wp)Input
On entry: the real and imaginary parts of the eigenvalue.
select must either be a module subprogram USEd by, or declared as EXTERNAL in, the (sub)program from which f08pbf (dgeesx) is called. Arguments denoted as Input must not be changed by this procedure.
4:     sense – Character(1)Input
On entry: determines which reciprocal condition numbers are computed.
sense='N'
None are computed.
sense='E'
Computed for average of selected eigenvalues only.
sense='V'
Computed for selected right invariant subspace only.
sense='B'
Computed for both.
If sense='E', 'V' or 'B', sort='S'.
Constraint: sense='N', 'E', 'V' or 'B'.
5:     n – IntegerInput
On entry: n, the order of the matrix A.
Constraint: n0.
6:     alda* – Real (Kind=nag_wp) arrayInput/Output
Note: the second dimension of the array a must be at least max1,n.
On entry: the n by n matrix A.
On exit: a is overwritten by its real Schur form T.
7:     lda – IntegerInput
On entry: the first dimension of the array a as declared in the (sub)program from which f08pbf (dgeesx) is called.
Constraint: ldamax1,n.
8:     sdim – IntegerOutput
On exit: if sort='N', sdim=0.
If sort='S', sdim= number of eigenvalues (after sorting) for which select is .TRUE.. (Complex conjugate pairs for which select is .TRUE. for either eigenvalue count as 2.)
9:     wr* – Real (Kind=nag_wp) arrayOutput
Note: the dimension of the array wr must be at least max1,n.
On exit: see the description of wi.
10:   wi* – Real (Kind=nag_wp) arrayOutput
Note: the dimension of the array wi must be at least max1,n.
On exit: wr and wi contain the real and imaginary parts, respectively, of the computed eigenvalues in the same order that they appear on the diagonal of the output Schur form T. Complex conjugate pairs of eigenvalues will appear consecutively with the eigenvalue having the positive imaginary part first.
11:   vsldvs* – Real (Kind=nag_wp) arrayOutput
Note: the second dimension of the array vs must be at least max1,n if jobvs='V', and at least 1 otherwise.
On exit: if jobvs='V', vs contains the orthogonal matrix Z of Schur vectors.
If jobvs='N', vs is not referenced.
12:   ldvs – IntegerInput
On entry: the first dimension of the array vs as declared in the (sub)program from which f08pbf (dgeesx) is called.
Constraints:
  • if jobvs='V', ldvs max1,n ;
  • otherwise ldvs1.
13:   rconde – Real (Kind=nag_wp)Output
On exit: if sense='E' or 'B', contains the reciprocal condition number for the average of the selected eigenvalues.
If sense='N' or 'V', rconde is not referenced.
14:   rcondv – Real (Kind=nag_wp)Output
On exit: if sense='V' or 'B', rcondv contains the reciprocal condition number for the selected right invariant subspace.
If sense='N' or 'E', rcondv is not referenced.
15:   workmax1,lwork – Real (Kind=nag_wp) arrayWorkspace
On exit: if info=0, work1 contains the minimum value of lwork required for optimal performance.
16:   lwork – IntegerInput
On entry: the dimension of the array work as declared in the (sub)program from which f08pbf (dgeesx) is called.
If lwork=-1, a workspace query is assumed; the routine only calculates upper bounds on the optimal sizes of the work and iwork arrays, returns these values as the first entries of the work and iwork arrays, and no error messages related to lwork or liwork is issued.
If sense='E', 'V' or 'B', lworkn+2×sdim×n-sdim, where sdim is the number of selected eigenvalues computed by this routine. Note that n+2×sdim×n-sdimn+n×n/2.
Note also that an error is only returned if lwork<max1,3×n, but if sense='E', 'V' or 'B' this may not be large enough.
Suggested value: for optimal performance, lwork must generally be larger than the minimum; increase the workspace by, say, nb×n , where nb  is the optimal block size for f08nef (dgehrd).
Constraint: lworkmax1,3×n.
17:   iworkmax1,liwork – Integer arrayWorkspace
On exit: if info=0, iwork1 returns the optimal liwork.
18:   liwork – IntegerInput
On entry: the dimension of the array iwork as declared in the (sub)program from which f08pbf (dgeesx) is called.
If liwork=-1, a workspace query is assumed; the routine only calculates upper bounds on the optimal sizes of the work and iwork arrays, returns these values as the first entries of the work and iwork arrays, and no error messages related to lwork or liwork is issued.
Constraints:
  • if sense='V' or 'B', liworksdim×n-sdim;
  • otherwise liwork1.
Note: sdim×n-sdimn×n/4. Note also that an error is only returned if liwork<1, but if sense='V' or 'B' this may not be large enough.
19:   bwork* – Logical arrayWorkspace
Note: the dimension of the array bwork must be at least max1,n if sort'N', and at least 1 otherwise.
If sort='N', bwork is not referenced.
20:   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=1ton
If info=i and in, the QR algorithm failed to compute all the eigenvalues.
info=n+1
The eigenvalues could not be reordered because some eigenvalues were too close to separate (the problem is very ill-conditioned).
info=n+2
After reordering, roundoff changed values of some complex eigenvalues so that leading eigenvalues in the Schur form no longer satisfy select=.TRUE.. This could also be caused by underflow due to scaling.

7
Accuracy

The computed Schur factorization satisfies
A+E = ZTZT ,  
where
E2 = Oε A2 ,  
and ε is the machine precision. See Section 4.8 of Anderson et al. (1999) for further details.

8
Parallelism and Performance

f08pbf (dgeesx) is threaded by NAG for parallel execution in multithreaded implementations of the NAG Library.
f08pbf (dgeesx) 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 proportional to n3.
The complex analogue of this routine is f08ppf (zgeesx).

10
Example

This example finds the Schur factorization of the matrix
A = 0.35 0.45 -0.14 -0.17 0.09 0.07 -0.54 0.35 -0.44 -0.33 -0.03 0.17 0.25 -0.32 -0.13 0.11 ,  
such that the real positive eigenvalues of A are the top left diagonal elements of the Schur form, T. Estimates of the condition numbers for the selected eigenvalue cluster and corresponding invariant subspace are also returned.
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 (f08pbfe.f90)

10.2
Program Data

Program Data (f08pbfe.d)

10.3
Program Results

Program Results (f08pbfe.r)

© The Numerical Algorithms Group Ltd, Oxford, UK. 2017