1DTGSEN(1)               LAPACK routine (version 3.1.1)               DTGSEN(1)
2
3
4

NAME

6       DTGSEN - the generalized real Schur decomposition of a real matrix pair
7       (A, B) (in terms of an orthonormal equivalence trans-  formation  Q'  *
8       (A,  B)  * Z), so that a selected cluster of eigenvalues appears in the
9       leading diagonal blocks of the upper quasi-triangular matrix A and  the
10       upper triangular B
11

SYNOPSIS

13       SUBROUTINE DTGSEN( IJOB,  WANTQ,  WANTZ,  SELECT,  N,  A,  LDA, B, LDB,
14                          ALPHAR, ALPHAI, BETA, Q, LDQ, Z,  LDZ,  M,  PL,  PR,
15                          DIF, WORK, LWORK, IWORK, LIWORK, INFO )
16
17           LOGICAL        WANTQ, WANTZ
18
19           INTEGER        IJOB, INFO, LDA, LDB, LDQ, LDZ, LIWORK, LWORK, M, N
20
21           DOUBLE         PRECISION PL, PR
22
23           LOGICAL        SELECT( * )
24
25           INTEGER        IWORK( * )
26
27           DOUBLE         PRECISION  A( LDA, * ), ALPHAI( * ), ALPHAR( * ), B(
28                          LDB, * ), BETA( * ), DIF( * ), Q( LDQ, * ), WORK(  *
29                          ), Z( LDZ, * )
30

PURPOSE

32       DTGSEN  reorders  the  generalized  real  Schur decomposition of a real
33       matrix pair (A, B) (in terms of an orthonormal equivalence trans-  for‐
34       mation  Q'  *  (A,  B)  * Z), so that a selected cluster of eigenvalues
35       appears in the leading diagonal blocks of  the  upper  quasi-triangular
36       matrix  A  and  the  upper triangular B. The leading columns of Q and Z
37       form orthonormal bases of the corresponding left and right eigen-  spa‐
38       ces  (deflating  subspaces).  (A,  B) must be in generalized real Schur
39       canonical form (as returned by DGGES), i.e. A is block upper triangular
40       with 1-by-1 and 2-by-2 diagonal blocks. B is upper triangular.
41
42       DTGSEN also computes the generalized eigenvalues
43
44                   w(j) = (ALPHAR(j) + i*ALPHAI(j))/BETA(j)
45
46       of the reordered matrix pair (A, B).
47
48       Optionally,  DTGSEN computes the estimates of reciprocal condition num‐
49       bers  for  eigenvalues  and  eigenspaces.  These  are   Difu[(A11,B11),
50       (A22,B22)]  and  Difl[(A11,B11),  (A22,B22)],  i.e.  the  separation(s)
51       between the matrix pairs (A11, B11) and (A22,B22)  that  correspond  to
52       the  selected  cluster  and the eigenvalues outside the cluster, resp.,
53       and norms of "projections" onto left and right eigenspaces w.r.t.   the
54       selected cluster in the (1,1)-block.
55
56

ARGUMENTS

58       IJOB    (input) INTEGER
59               Specifies  whether condition numbers are required for the clus‐
60               ter of eigenvalues (PL and PR) or the deflating subspaces (Difu
61               and Difl):
62               =0: Only reorder w.r.t. SELECT. No extras.
63               =1:  Reciprocal  of  norms of "projections" onto left and right
64               eigenspaces w.r.t. the selected cluster (PL and PR).  =2: Upper
65               bounds on Difu and Difl. F-norm-based estimate
66               (DIF(1:2)).
67               =3: Estimate of Difu and Difl. 1-norm-based estimate
68               (DIF(1:2)).   About 5 times as expensive as IJOB = 2.  =4: Com‐
69               pute PL, PR and DIF (i.e. 0, 1 and 2 above):  Economic  version
70               to  get  it  all.   =5: Compute PL, PR and DIF (i.e. 0, 1 and 3
71               above)
72
73       WANTQ   (input) LOGICAL
74
75       WANTZ   (input) LOGICAL
76
77       SELECT  (input) LOGICAL array, dimension (N)
78               SELECT specifies the eigenvalues in the selected  cluster.   To
79               select  a  real  eigenvalue w(j), SELECT(j) must be set to w(j)
80               and w(j+1), corresponding to a 2-by-2  diagonal  block,  either
81               SELECT(j)  or  SELECT(j+1)  or  both must be set to either both
82               included in the cluster or both excluded.
83
84       N       (input) INTEGER
85               The order of the matrices A and B. N >= 0.
86
87       A       (input/output) DOUBLE PRECISION array, dimension(LDA,N)
88               On entry, the upper quasi-triangular matrix A, with (A,  B)  in
89               generalized real Schur canonical form.  On exit, A is overwrit‐
90               ten by the reordered matrix A.
91
92       LDA     (input) INTEGER
93               The leading dimension of the array A. LDA >= max(1,N).
94
95       B       (input/output) DOUBLE PRECISION array, dimension(LDB,N)
96               On entry, the upper triangular matrix B, with (A, B) in  gener‐
97               alized real Schur canonical form.  On exit, B is overwritten by
98               the reordered matrix B.
99
100       LDB     (input) INTEGER
101               The leading dimension of the array B. LDB >= max(1,N).
102
103       ALPHAR  (output) DOUBLE PRECISION array, dimension (N)
104               ALPHAI  (output) DOUBLE PRECISION  array,  dimension  (N)  BETA
105               (output)   DOUBLE  PRECISION  array,  dimension  (N)  On  exit,
106               (ALPHAR(j) + ALPHAI(j)*i)/BETA(j), j=1,...,N, will be the  gen‐
107               eralized    eigenvalues.     ALPHAR(j)    +   ALPHAI(j)*i   and
108               BETA(j),j=1,...,N  are the diagonals of the complex Schur  form
109               (S,T)  that  would  result if the 2-by-2 diagonal blocks of the
110               real generalized Schur form of (A,B) were  further  reduced  to
111               triangular  form  using  complex  unitary  transformations.  If
112               ALPHAI(j) is zero, then the j-th eigenvalue is real;  if  posi‐
113               tive, then the j-th and (j+1)-st eigenvalues are a complex con‐
114               jugate pair, with ALPHAI(j+1) negative.
115
116       Q       (input/output) DOUBLE PRECISION array, dimension (LDQ,N)
117               On entry, if WANTQ = .TRUE., Q is an N-by-N matrix.  On exit, Q
118               has  been  postmultiplied by the left orthogonal transformation
119               matrix which reorder (A, B); The leading M columns  of  Q  form
120               orthonormal  bases  for  the specified pair of left eigenspaces
121               (deflating subspaces).  If WANTQ = .FALSE.,  Q  is  not  refer‐
122               enced.
123
124       LDQ     (input) INTEGER
125               The leading dimension of the array Q.  LDQ >= 1; and if WANTQ =
126               .TRUE., LDQ >= N.
127
128       Z       (input/output) DOUBLE PRECISION array, dimension (LDZ,N)
129               On entry, if WANTZ = .TRUE., Z is an N-by-N matrix.  On exit, Z
130               has  been  postmultiplied by the left orthogonal transformation
131               matrix which reorder (A, B); The leading M columns  of  Z  form
132               orthonormal  bases  for  the specified pair of left eigenspaces
133               (deflating subspaces).  If WANTZ = .FALSE.,  Z  is  not  refer‐
134               enced.
135
136       LDZ     (input) INTEGER
137               The  leading  dimension  of  the  array Z. LDZ >= 1; If WANTZ =
138               .TRUE., LDZ >= N.
139
140       M       (output) INTEGER
141               The dimension of the specified pair of left  and  right  eigen-
142               spaces (deflating subspaces). 0 <= M <= N.
143
144       PL      (output) DOUBLE PRECISION
145               PR       (output)  DOUBLE PRECISION If IJOB = 1, 4 or 5, PL, PR
146               are lower bounds on the reciprocal of the norm of "projections"
147               onto  left  and  right eigenspaces with respect to the selected
148               cluster.  0 < PL, PR <= 1.  If M = 0 or M = N, PL =  PR   =  1.
149               If IJOB = 0, 2 or 3, PL and PR are not referenced.
150
151       DIF     (output) DOUBLE PRECISION array, dimension (2).
152               If IJOB >= 2, DIF(1:2) store the estimates of Difu and Difl.
153               If IJOB = 2 or 4, DIF(1:2) are F-norm-based upper bounds on
154               Difu  and  Difl.  If  IJOB  = 3 or 5, DIF(1:2) are 1-norm-based
155               estimates of Difu and Difl.  If M =  0  or  N,  DIF(1:2)  =  F-
156               norm([A, B]).  If IJOB = 0 or 1, DIF is not referenced.
157
158       WORK    (workspace/output) DOUBLE PRECISION array,
159               dimension  (MAX(1,LWORK)) On exit, if INFO = 0, WORK(1) returns
160               the optimal LWORK.
161
162       LWORK   (input) INTEGER
163               The dimension of the array WORK. LWORK >=  4*N+16.  If  IJOB  =
164               1,  2 or 4, LWORK >= MAX(4*N+16, 2*M*(N-M)).  If IJOB = 3 or 5,
165               LWORK >= MAX(4*N+16, 4*M*(N-M)).
166
167               If LWORK = -1, then a workspace query is assumed;  the  routine
168               only  calculates  the  optimal  size of the WORK array, returns
169               this value as the first entry of the WORK array, and  no  error
170               message related to LWORK is issued by XERBLA.
171
172       IWORK   (workspace/output) INTEGER array, dimension (MAX(1,LIWORK))
173               IF  IJOB  = 0, IWORK is not referenced.  Otherwise, on exit, if
174               INFO = 0, IWORK(1) returns the optimal LIWORK.
175
176       LIWORK  (input) INTEGER
177               The dimension of the array IWORK. LIWORK >= 1.  If IJOB = 1,  2
178               or  4, LIWORK >=  N+6.  If IJOB = 3 or 5, LIWORK >= MAX(2*M*(N-
179               M), N+6).
180
181               If LIWORK = -1, then a workspace query is assumed; the  routine
182               only  calculates  the  optimal size of the IWORK array, returns
183               this value as the first entry of the IWORK array, and no  error
184               message related to LIWORK is issued by XERBLA.
185
186       INFO    (output) INTEGER
187               =0: Successful exit.
188               <0: If INFO = -i, the i-th argument had an illegal value.
189               =1:  Reordering of (A, B) failed because the transformed matrix
190               pair (A, B) would be too far from generalized Schur  form;  the
191               problem  is  very  ill-conditioned.   (A, B) may have been par‐
192               tially reordered.  If requested, 0 is returned  in  DIF(*),  PL
193               and PR.
194

FURTHER DETAILS

196       DTGSEN  first collects the selected eigenvalues by computing orthogonal
197       U and W that move them to the top left corner  of  (A,  B).   In  other
198       words, the selected eigenvalues are the eigenvalues of (A11, B11) in:
199
200                     U'*(A, B)*W = (A11 A12) (B11 B12) n1
201                                   ( 0  A22),( 0  B22) n2
202                                     n1  n2    n1  n2
203
204       where  N  = n1+n2 and U' means the transpose of U. The first n1 columns
205       of U and W span the  specified  pair  of  left  and  right  eigenspaces
206       (deflating subspaces) of (A, B).
207
208       If  (A, B) has been obtained from the generalized real Schur decomposi‐
209       tion of a matrix pair (C, D) = Q*(A, B)*Z', then the reordered general‐
210       ized real Schur form of (C, D) is given by
211
212                (C, D) = (Q*U)*(U'*(A, B)*W)*(Z*W)',
213
214       and  the first n1 columns of Q*U and Z*W span the corresponding deflat‐
215       ing subspaces of (C, D) (Q and Z store Q*U and Z*W, resp.).
216
217       Note that if the selected eigenvalue is  sufficiently  ill-conditioned,
218       then  its value may differ significantly from its value before reorder‐
219       ing.
220
221       The reciprocal condition numbers of  the  left  and  right  eigenspaces
222       spanned  by  the  first  n1  columns of U and W (or Q*U and Z*W) may be
223       returned in DIF(1:2), corresponding to Difu and Difl, resp.
224
225       The Difu and Difl are defined as:
226
227            Difu[(A11, B11), (A22, B22)] = sigma-min( Zu )
228       and
229            Difl[(A11, B11), (A22, B22)] = Difu[(A22, B22), (A11, B11)],
230
231       where  sigma-min(Zu)  is   the   smallest   singular   value   of   the
232       (2*n1*n2)-by-(2*n1*n2) matrix
233
234            Zu = [ kron(In2, A11)  -kron(A22', In1) ]
235                 [ kron(In2, B11)  -kron(B22', In1) ].
236
237       Here,  Inx  is the identity matrix of size nx and A22' is the transpose
238       of A22. kron(X, Y) is the Kronecker product between the matrices X  and
239       Y.
240
241       When  DIF(2)  is small, small changes in (A, B) can cause large changes
242       in the deflating subspace. An approximate  (asymptotic)  bound  on  the
243       maximum angular error in the computed deflating subspaces is
244
245            EPS * norm((A, B)) / DIF(2),
246
247       where EPS is the machine precision.
248
249       The reciprocal norm of the projectors on the left and right eigenspaces
250       associated with (A11, B11) may be returned in PL and PR.  They are com‐
251       puted  as follows. First we compute L and R so that P*(A, B)*Q is block
252       diagonal, where
253
254            P = ( I -L ) n1           Q = ( I R ) n1
255                ( 0  I ) n2    and        ( 0 I ) n2
256                  n1 n2                    n1 n2
257
258       and (L, R) is the solution to the generalized Sylvester equation
259
260            A11*R - L*A22 = -A12
261            B11*R - L*B22 = -B12
262
263       Then PL = (F-norm(L)**2+1)**(-1/2) and PR  =  (F-norm(R)**2+1)**(-1/2).
264       An  approximate (asymptotic) bound on the average absolute error of the
265       selected eigenvalues is
266
267            EPS * norm((A, B)) / PL.
268
269       There are also global error bounds which valid for perturbations up  to
270       a  certain  restriction:  A lower bound (x) on the smallest F-norm(E,F)
271       for which an eigenvalue of (A11, B11) may move and coalesce with an ei‐
272       genvalue  of (A22, B22) under perturbation (E,F), (i.e. (A + E, B + F),
273       is
274
275        x = min(Difu,Difl)/((1/(PL*PL)+1/(PR*PR))**(1/2)+2*max(1/PL,1/PR)).
276
277       An approximate bound on x can be computed from DIF(1:2), PL and PR.
278
279       If y = ( F-norm(E,F) / x) <= 1, the angles between the  perturbed  (L',
280       R')  and  unperturbed (L, R) left and right deflating subspaces associ‐
281       ated with the selected cluster in the (1,1)-blocks can be bounded as
282
283        max-angle(L, L') <= arctan( y * PL / (1 - y * (1 - PL * PL)**(1/2))
284        max-angle(R, R') <= arctan( y * PR / (1 - y * (1 - PR * PR)**(1/2))
285
286       See LAPACK User's Guide section 4.11 or the  following  references  for
287       more information.
288
289       Note that if the default method for computing the Frobenius-norm- based
290       estimate DIF is not wanted (see DLATDF), then the parameter IDIFJB (see
291       below)  should be changed from 3 to 4 (routine DLATDF (IJOB = 2 will be
292       used)). See DTGSYL for more details.
293
294       Based on contributions by
295          Bo Kagstrom and Peter Poromaa, Department of Computing Science,
296          Umea University, S-901 87 Umea, Sweden.
297
298       References
299       ==========
300
301       [1] B. Kagstrom; A Direct Method for Reordering Eigenvalues in the
302           Generalized Real Schur Form of a Regular Matrix Pair (A, B), in
303           M.S. Moonen et al (eds), Linear Algebra for Large Scale and
304           Real-Time Applications, Kluwer Academic Publ. 1993, pp 195-218.
305
306       [2] B. Kagstrom and P. Poromaa; Computing Eigenspaces with Specified
307           Eigenvalues of a Regular Matrix Pair (A, B) and Condition
308           Estimation: Theory, Algorithms and Software,
309           Report UMINF - 94.04, Department of Computing Science, Umea
310           University, S-901 87 Umea, Sweden, 1994. Also as LAPACK Working
311           Note 87. To appear in Numerical Algorithms, 1996.
312
313       [3] B. Kagstrom and P. Poromaa, LAPACK-Style Algorithms and Software
314           for Solving the Generalized Sylvester Equation and Estimating the
315           Separation between Regular Matrix Pairs, Report UMINF - 93.23,
316           Department of Computing Science, Umea University, S-901 87 Umea,
317           Sweden, December 1993, Revised April 1994, Also as LAPACK Working
318           Note 75. To appear in ACM Trans. on Math. Software, Vol 22, No 1,
319           1996.
320
321
322
323
324 LAPACK routine (version 3.1.1)  February 2007                       DTGSEN(1)
Impressum