1STGEX2(1)           LAPACK auxiliary routine (version 3.1)           STGEX2(1)
2
3
4

NAME

6       STGEX2  -  adjacent  diagonal  blocks (A11, B11) and (A22, B22) of size
7       1-by-1 or 2-by-2 in an upper (quasi) triangular matrix pair (A,  B)  by
8       an orthogonal equivalence transformation
9

SYNOPSIS

11       SUBROUTINE STGEX2( WANTQ, WANTZ, N, A, LDA, B, LDB, Q, LDQ, Z, LDZ, J1,
12                          N1, N2, WORK, LWORK, INFO )
13
14           LOGICAL        WANTQ, WANTZ
15
16           INTEGER        INFO, J1, LDA, LDB, LDQ, LDZ, LWORK, N, N1, N2
17
18           REAL           A( LDA, * ), B( LDB, * ), Q( LDQ, * ), WORK( * ), Z(
19                          LDZ, * )
20

PURPOSE

22       STGEX2 swaps adjacent diagonal blocks (A11, B11) and (A22, B22) of size
23       1-by-1 or 2-by-2 in an upper (quasi) triangular matrix pair (A,  B)  by
24       an orthogonal equivalence transformation.
25
26       (A, B) must be in generalized real Schur canonical form (as returned by
27       SGGES), i.e. A is block upper triangular with 1-by-1 and 2-by-2  diago‐
28       nal blocks. B is upper triangular.
29
30       Optionally,  the  matrices  Q  and  Z  of generalized Schur vectors are
31       updated.
32
33              Q(in) * A(in) * Z(in)' = Q(out) * A(out) * Z(out)'
34              Q(in) * B(in) * Z(in)' = Q(out) * B(out) * Z(out)'
35
36
37

ARGUMENTS

39       WANTQ   (input) LOGICAL
40
41       WANTZ   (input) LOGICAL
42
43       N       (input) INTEGER
44               The order of the matrices A and B. N >= 0.
45
46       A      (input/output) REAL arrays, dimensions (LDA,N)
47              On entry, the matrix A in the pair (A, B).  On exit, the updated
48              matrix A.
49
50       LDA     (input)  INTEGER
51               The leading dimension of the array A. LDA >= max(1,N).
52
53       B      (input/output) REAL arrays, dimensions (LDB,N)
54              On entry, the matrix B in the pair (A, B).  On exit, the updated
55              matrix B.
56
57       LDB     (input)  INTEGER
58               The leading dimension of the array B. LDB >= max(1,N).
59
60       Q       (input/output) REAL array, dimension (LDZ,N)
61               On entry, if WANTQ = .TRUE., the orthogonal matrix Q.  On exit,
62               the updated matrix Q.  Not referenced if WANTQ = .FALSE..
63
64       LDQ     (input) INTEGER
65               The  leading  dimension  of  the array Q. LDQ >= 1.  If WANTQ =
66               .TRUE., LDQ >= N.
67
68       Z       (input/output) REAL array, dimension (LDZ,N)
69               On entry, if WANTZ =.TRUE., the orthogonal matrix Z.  On  exit,
70               the updated matrix Z.  Not referenced if WANTZ = .FALSE..
71
72       LDZ     (input) INTEGER
73               The  leading  dimension  of  the array Z. LDZ >= 1.  If WANTZ =
74               .TRUE., LDZ >= N.
75
76       J1      (input) INTEGER
77               The index to the first block (A11, B11). 1 <= J1 <= N.
78
79       N1      (input) INTEGER
80               The order of the first block (A11, B11). N1 = 0, 1 or 2.
81
82       N2      (input) INTEGER
83               The order of the second block (A22, B22). N2 = 0, 1 or 2.
84
85       WORK    (workspace) REAL array, dimension (MAX(1,LWORK)).
86
87       LWORK   (input) INTEGER
88               The dimension of the array WORK.   LWORK  >=   MAX(  N*(N2+N1),
89               (N2+N1)*(N2+N1)*2 )
90
91       INFO    (output) INTEGER
92               =0: Successful exit
93               >0: If INFO = 1, the transformed matrix (A, B) would be too far
94               from generalized Schur form; the blocks are not swapped and (A,
95               B)  and  (Q,  Z) are unchanged.  The problem of swapping is too
96               ill-conditioned.  <0: If INFO = -16: LWORK is too small. Appro‐
97               priate value for LWORK is returned in WORK(1).
98

FURTHER DETAILS

100       Based on contributions by
101          Bo Kagstrom and Peter Poromaa, Department of Computing Science,
102          Umea University, S-901 87 Umea, Sweden.
103
104       In the current code both weak and strong stability tests are performed.
105       The user can omit the strong stability test by  changing  the  internal
106       logical parameter WANDS to .FALSE.. See ref. [2] for details.
107
108       [1] B. Kagstrom; A Direct Method for Reordering Eigenvalues in the
109           Generalized Real Schur Form of a Regular Matrix Pair (A, B), in
110           M.S. Moonen et al (eds), Linear Algebra for Large Scale and
111           Real-Time Applications, Kluwer Academic Publ. 1993, pp 195-218.
112
113       [2] B. Kagstrom and P. Poromaa; Computing Eigenspaces with Specified
114           Eigenvalues of a Regular Matrix Pair (A, B) and Condition
115           Estimation: Theory, Algorithms and Software,
116           Report UMINF - 94.04, Department of Computing Science, Umea
117           University, S-901 87 Umea, Sweden, 1994. Also as LAPACK Working
118           Note 87. To appear in Numerical Algorithms, 1996.
119
120
121
122
123 LAPACK auxiliary routine (versionNo3v.e1m)ber 2006                       STGEX2(1)
Impressum