1
       2
       3
       4
       5
       6
       7
       8
       9
      10
      11
      12
      13
      14
      15
      16
      17
      18
      19
      20
      21
      22
      23
      24
      25
      26
      27
      28
      29
      30
      31
      32
      33
      34
      35
      36
      37
      38
      39
      40
      41
      42
      43
      44
      45
      46
      47
      48
      49
      50
      51
      52
      53
      54
      55
      56
      57
      58
      59
      60
      61
      62
      63
      64
      65
      66
      67
      68
      69
      70
      71
      72
      73
      74
      75
      76
      77
      78
      79
      80
      81
      82
      83
      84
      85
      86
      87
      88
      89
      90
      91
      92
      93
      94
      95
      96
      97
      98
      99
     100
     101
     102
     103
     104
     105
     106
     107
     108
     109
     110
     111
     112
     113
     114
     115
     116
     117
     118
     119
     120
     121
     122
     123
     124
     125
     126
     127
     128
     129
     130
     131
     132
     133
     134
     135
     136
     137
     138
     139
     140
     141
     142
     143
     144
     145
     146
     147
     148
     149
     150
     151
     152
     153
     154
     155
     156
     157
     158
     159
     160
     161
     162
     163
     164
     165
     166
     167
     168
     169
     170
     171
     172
     173
     174
     175
     176
     177
     178
     179
     180
     181
     182
     183
     184
     185
     186
     187
     188
     189
     190
     191
     192
     193
     194
     195
     196
     197
     198
     199
     200
     201
     202
     203
     204
     205
     206
     207
     208
     209
     210
     211
     212
     213
     214
     215
     216
     217
     218
     219
     220
     221
     222
     223
     224
     225
     226
     227
     228
     229
     230
     231
     232
     233
     234
     235
     236
     237
     238
     239
     240
     241
     242
     243
     244
     245
     246
     247
     248
     249
     250
     251
     252
     253
     254
     255
     256
     257
     258
     259
     260
     261
     262
     263
     264
     265
     266
     267
     268
     269
     270
     271
     272
     273
     274
     275
     276
     277
     278
     279
     280
     281
     282
     283
     284
     285
     286
     287
     288
     289
     290
     291
     292
     293
     294
     295
     296
     297
     298
     299
     300
     301
     302
     303
     304
     305
     306
     307
     308
     309
     310
     311
     312
     313
     314
     315
     316
     317
     318
     319
     320
     321
     322
     323
     324
     325
     326
     327
     328
     329
     330
     331
     332
     333
     334
     335
     336
     337
     338
     339
     340
     341
     342
     343
     344
     345
     346
     347
     348
     349
     350
     351
     352
     353
     354
     355
     356
     357
     358
     359
     360
     361
     362
      SUBROUTINE ZTGSY2( TRANS, IJOB, M, N, A, LDA, B, LDB, C, LDC, D,
     $                   LDD, E, LDE, F, LDF, SCALE, RDSUM, RDSCAL,
     $                   INFO )
*
*  -- LAPACK auxiliary routine (version 3.3.1) --
*  -- LAPACK is a software package provided by Univ. of Tennessee,    --
*  -- Univ. of California Berkeley, Univ. of Colorado Denver and NAG Ltd..--
*  -- April 2011                                                      --
*
*     .. Scalar Arguments ..
      CHARACTER          TRANS
      INTEGER            IJOB, INFO, LDA, LDB, LDC, LDD, LDE, LDF, M, N
      DOUBLE PRECISION   RDSCAL, RDSUM, SCALE
*     ..
*     .. Array Arguments ..
      COMPLEX*16         A( LDA, * ), B( LDB, * ), C( LDC, * ),
     $                   D( LDD, * ), E( LDE, * ), F( LDF, * )
*     ..
*
*  Purpose
*  =======
*
*  ZTGSY2 solves the generalized Sylvester equation
*
*              A * R - L * B = scale * C               (1)
*              D * R - L * E = scale * F
*
*  using Level 1 and 2 BLAS, where R and L are unknown M-by-N matrices,
*  (A, D), (B, E) and (C, F) are given matrix pairs of size M-by-M,
*  N-by-N and M-by-N, respectively. A, B, D and E are upper triangular
*  (i.e., (A,D) and (B,E) in generalized Schur form).
*
*  The solution (R, L) overwrites (C, F). 0 <= SCALE <= 1 is an output
*  scaling factor chosen to avoid overflow.
*
*  In matrix notation solving equation (1) corresponds to solve
*  Zx = scale * b, where Z is defined as
*
*         Z = [ kron(In, A)  -kron(B**H, Im) ]             (2)
*             [ kron(In, D)  -kron(E**H, Im) ],
*
*  Ik is the identity matrix of size k and X**H is the conjuguate transpose of X.
*  kron(X, Y) is the Kronecker product between the matrices X and Y.
*
*  If TRANS = 'C', y in the conjugate transposed system Z**H*y = scale*b
*  is solved for, which is equivalent to solve for R and L in
*
*              A**H * R  + D**H * L   = scale *  C           (3)
*              R  * B**H + L  * E**H  = scale * -F
*
*  This case is used to compute an estimate of Dif[(A, D), (B, E)] =
*  = sigma_min(Z) using reverse communicaton with ZLACON.
*
*  ZTGSY2 also (IJOB >= 1) contributes to the computation in ZTGSYL
*  of an upper bound on the separation between to matrix pairs. Then
*  the input (A, D), (B, E) are sub-pencils of two matrix pairs in
*  ZTGSYL.
*
*  Arguments
*  =========
*
*  TRANS   (input) CHARACTER*1
*          = 'N', solve the generalized Sylvester equation (1).
*          = 'T': solve the 'transposed' system (3).
*
*  IJOB    (input) INTEGER
*          Specifies what kind of functionality to be performed.
*          =0: solve (1) only.
*          =1: A contribution from this subsystem to a Frobenius
*              norm-based estimate of the separation between two matrix
*              pairs is computed. (look ahead strategy is used).
*          =2: A contribution from this subsystem to a Frobenius
*              norm-based estimate of the separation between two matrix
*              pairs is computed. (DGECON on sub-systems is used.)
*          Not referenced if TRANS = 'T'.
*
*  M       (input) INTEGER
*          On entry, M specifies the order of A and D, and the row
*          dimension of C, F, R and L.
*
*  N       (input) INTEGER
*          On entry, N specifies the order of B and E, and the column
*          dimension of C, F, R and L.
*
*  A       (input) COMPLEX*16 array, dimension (LDA, M)
*          On entry, A contains an upper triangular matrix.
*
*  LDA     (input) INTEGER
*          The leading dimension of the matrix A. LDA >= max(1, M).
*
*  B       (input) COMPLEX*16 array, dimension (LDB, N)
*          On entry, B contains an upper triangular matrix.
*
*  LDB     (input) INTEGER
*          The leading dimension of the matrix B. LDB >= max(1, N).
*
*  C       (input/output) COMPLEX*16 array, dimension (LDC, N)
*          On entry, C contains the right-hand-side of the first matrix
*          equation in (1).
*          On exit, if IJOB = 0, C has been overwritten by the solution
*          R.
*
*  LDC     (input) INTEGER
*          The leading dimension of the matrix C. LDC >= max(1, M).
*
*  D       (input) COMPLEX*16 array, dimension (LDD, M)
*          On entry, D contains an upper triangular matrix.
*
*  LDD     (input) INTEGER
*          The leading dimension of the matrix D. LDD >= max(1, M).
*
*  E       (input) COMPLEX*16 array, dimension (LDE, N)
*          On entry, E contains an upper triangular matrix.
*
*  LDE     (input) INTEGER
*          The leading dimension of the matrix E. LDE >= max(1, N).
*
*  F       (input/output) COMPLEX*16 array, dimension (LDF, N)
*          On entry, F contains the right-hand-side of the second matrix
*          equation in (1).
*          On exit, if IJOB = 0, F has been overwritten by the solution
*          L.
*
*  LDF     (input) INTEGER
*          The leading dimension of the matrix F. LDF >= max(1, M).
*
*  SCALE   (output) DOUBLE PRECISION
*          On exit, 0 <= SCALE <= 1. If 0 < SCALE < 1, the solutions
*          R and L (C and F on entry) will hold the solutions to a
*          slightly perturbed system but the input matrices A, B, D and
*          E have not been changed. If SCALE = 0, R and L will hold the
*          solutions to the homogeneous system with C = F = 0.
*          Normally, SCALE = 1.
*
*  RDSUM   (input/output) DOUBLE PRECISION
*          On entry, the sum of squares of computed contributions to
*          the Dif-estimate under computation by ZTGSYL, where the
*          scaling factor RDSCAL (see below) has been factored out.
*          On exit, the corresponding sum of squares updated with the
*          contributions from the current sub-system.
*          If TRANS = 'T' RDSUM is not touched.
*          NOTE: RDSUM only makes sense when ZTGSY2 is called by
*          ZTGSYL.
*
*  RDSCAL  (input/output) DOUBLE PRECISION
*          On entry, scaling factor used to prevent overflow in RDSUM.
*          On exit, RDSCAL is updated w.r.t. the current contributions
*          in RDSUM.
*          If TRANS = 'T', RDSCAL is not touched.
*          NOTE: RDSCAL only makes sense when ZTGSY2 is called by
*          ZTGSYL.
*
*  INFO    (output) INTEGER
*          On exit, if INFO is set to
*            =0: Successful exit
*            <0: If INFO = -i, input argument number i is illegal.
*            >0: The matrix pairs (A, D) and (B, E) have common or very
*                close eigenvalues.
*
*  Further Details
*  ===============
*
*  Based on contributions by
*     Bo Kagstrom and Peter Poromaa, Department of Computing Science,
*     Umea University, S-901 87 Umea, Sweden.
*
*  =====================================================================
*
*     .. Parameters ..
      DOUBLE PRECISION   ZERO, ONE
      INTEGER            LDZ
      PARAMETER          ( ZERO = 0.0D+0, ONE = 1.0D+0, LDZ = 2 )
*     ..
*     .. Local Scalars ..
      LOGICAL            NOTRAN
      INTEGER            I, IERR, J, K
      DOUBLE PRECISION   SCALOC
      COMPLEX*16         ALPHA
*     ..
*     .. Local Arrays ..
      INTEGER            IPIV( LDZ ), JPIV( LDZ )
      COMPLEX*16         RHS( LDZ ), Z( LDZ, LDZ )
*     ..
*     .. External Functions ..
      LOGICAL            LSAME
      EXTERNAL           LSAME
*     ..
*     .. External Subroutines ..
      EXTERNAL           XERBLA, ZAXPY, ZGESC2, ZGETC2, ZLATDF, ZSCAL
*     ..
*     .. Intrinsic Functions ..
      INTRINSIC          DCMPLXDCONJGMAX
*     ..
*     .. Executable Statements ..
*
*     Decode and test input parameters
*
      INFO = 0
      IERR = 0
      NOTRAN = LSAME( TRANS, 'N' )
      IF.NOT.NOTRAN .AND. .NOT.LSAME( TRANS, 'C' ) ) THEN
         INFO = -1
      ELSE IF( NOTRAN ) THEN
         IF( ( IJOB.LT.0 ) .OR. ( IJOB.GT.2 ) ) THEN
            INFO = -2
         END IF
      END IF
      IF( INFO.EQ.0 ) THEN
         IF( M.LE.0 ) THEN
            INFO = -3
         ELSE IF( N.LE.0 ) THEN
            INFO = -4
         ELSE IF( LDA.LT.MAX1, M ) ) THEN
            INFO = -5
         ELSE IF( LDB.LT.MAX1, N ) ) THEN
            INFO = -8
         ELSE IF( LDC.LT.MAX1, M ) ) THEN
            INFO = -10
         ELSE IF( LDD.LT.MAX1, M ) ) THEN
            INFO = -12
         ELSE IF( LDE.LT.MAX1, N ) ) THEN
            INFO = -14
         ELSE IF( LDF.LT.MAX1, M ) ) THEN
            INFO = -16
         END IF
      END IF
      IF( INFO.NE.0 ) THEN
         CALL XERBLA( 'ZTGSY2'-INFO )
         RETURN
      END IF
*
      IF( NOTRAN ) THEN
*
*        Solve (I, J) - system
*           A(I, I) * R(I, J) - L(I, J) * B(J, J) = C(I, J)
*           D(I, I) * R(I, J) - L(I, J) * E(J, J) = F(I, J)
*        for I = M, M - 1, ..., 1; J = 1, 2, ..., N
*
         SCALE = ONE
         SCALOC = ONE
         DO 30 J = 1, N
            DO 20 I = M, 1-1
*
*              Build 2 by 2 system
*
               Z( 11 ) = A( I, I )
               Z( 21 ) = D( I, I )
               Z( 12 ) = -B( J, J )
               Z( 22 ) = -E( J, J )
*
*              Set up right hand side(s)
*
               RHS( 1 ) = C( I, J )
               RHS( 2 ) = F( I, J )
*
*              Solve Z * x = RHS
*
               CALL ZGETC2( LDZ, Z, LDZ, IPIV, JPIV, IERR )
               IF( IERR.GT.0 )
     $            INFO = IERR
               IF( IJOB.EQ.0 ) THEN
                  CALL ZGESC2( LDZ, Z, LDZ, RHS, IPIV, JPIV, SCALOC )
                  IF( SCALOC.NE.ONE ) THEN
                     DO 10 K = 1, N
                        CALL ZSCAL( M, DCMPLX( SCALOC, ZERO ),
     $                              C( 1, K ), 1 )
                        CALL ZSCAL( M, DCMPLX( SCALOC, ZERO ),
     $                              F( 1, K ), 1 )
   10                CONTINUE
                     SCALE = SCALE*SCALOC
                  END IF
               ELSE
                  CALL ZLATDF( IJOB, LDZ, Z, LDZ, RHS, RDSUM, RDSCAL,
     $                         IPIV, JPIV )
               END IF
*
*              Unpack solution vector(s)
*
               C( I, J ) = RHS( 1 )
               F( I, J ) = RHS( 2 )
*
*              Substitute R(I, J) and L(I, J) into remaining equation.
*
               IF( I.GT.1 ) THEN
                  ALPHA = -RHS( 1 )
                  CALL ZAXPY( I-1, ALPHA, A( 1, I ), 1, C( 1, J ), 1 )
                  CALL ZAXPY( I-1, ALPHA, D( 1, I ), 1, F( 1, J ), 1 )
               END IF
               IF( J.LT.N ) THEN
                  CALL ZAXPY( N-J, RHS( 2 ), B( J, J+1 ), LDB,
     $                        C( I, J+1 ), LDC )
                  CALL ZAXPY( N-J, RHS( 2 ), E( J, J+1 ), LDE,
     $                        F( I, J+1 ), LDF )
               END IF
*
   20       CONTINUE
   30    CONTINUE
      ELSE
*
*        Solve transposed (I, J) - system:
*           A(I, I)**H * R(I, J) + D(I, I)**H * L(J, J) = C(I, J)
*           R(I, I) * B(J, J) + L(I, J) * E(J, J)   = -F(I, J)
*        for I = 1, 2, ..., M, J = N, N - 1, ..., 1
*
         SCALE = ONE
         SCALOC = ONE
         DO 80 I = 1, M
            DO 70 J = N, 1-1
*
*              Build 2 by 2 system Z**H
*
               Z( 11 ) = DCONJG( A( I, I ) )
               Z( 21 ) = -DCONJG( B( J, J ) )
               Z( 12 ) = DCONJG( D( I, I ) )
               Z( 22 ) = -DCONJG( E( J, J ) )
*
*
*              Set up right hand side(s)
*
               RHS( 1 ) = C( I, J )
               RHS( 2 ) = F( I, J )
*
*              Solve Z**H * x = RHS
*
               CALL ZGETC2( LDZ, Z, LDZ, IPIV, JPIV, IERR )
               IF( IERR.GT.0 )
     $            INFO = IERR
               CALL ZGESC2( LDZ, Z, LDZ, RHS, IPIV, JPIV, SCALOC )
               IF( SCALOC.NE.ONE ) THEN
                  DO 40 K = 1, N
                     CALL ZSCAL( M, DCMPLX( SCALOC, ZERO ), C( 1, K ),
     $                           1 )
                     CALL ZSCAL( M, DCMPLX( SCALOC, ZERO ), F( 1, K ),
     $                           1 )
   40             CONTINUE
                  SCALE = SCALE*SCALOC
               END IF
*
*              Unpack solution vector(s)
*
               C( I, J ) = RHS( 1 )
               F( I, J ) = RHS( 2 )
*
*              Substitute R(I, J) and L(I, J) into remaining equation.
*
               DO 50 K = 1, J - 1
                  F( I, K ) = F( I, K ) + RHS( 1 )*DCONJG( B( K, J ) ) +
     $                        RHS( 2 )*DCONJG( E( K, J ) )
   50          CONTINUE
               DO 60 K = I + 1, M
                  C( K, J ) = C( K, J ) - DCONJG( A( I, K ) )*RHS( 1 ) -
     $                        DCONJG( D( I, K ) )*RHS( 2 )
   60          CONTINUE
*
   70       CONTINUE
   80    CONTINUE
      END IF
      RETURN
*
*     End of ZTGSY2
*
      END