File:  [local] / rpl / lapack / lapack / zgges.f
Revision 1.12: download - view: text, annotated - select for diffs - revision graph
Fri Dec 14 14:22:46 2012 UTC (11 years, 5 months ago) by bertrand
Branches: MAIN
CVS tags: rpl-4_1_16, rpl-4_1_15, rpl-4_1_14, rpl-4_1_13, rpl-4_1_12, rpl-4_1_11, HEAD
Mise à jour de lapack.

    1: *> \brief <b> ZGGES computes the eigenvalues, the Schur form, and, optionally, the matrix of Schur vectors for GE matrices</b>
    2: *
    3: *  =========== DOCUMENTATION ===========
    4: *
    5: * Online html documentation available at 
    6: *            http://www.netlib.org/lapack/explore-html/ 
    7: *
    8: *> \htmlonly
    9: *> Download ZGGES + dependencies 
   10: *> <a href="http://www.netlib.org/cgi-bin/netlibfiles.tgz?format=tgz&filename=/lapack/lapack_routine/zgges.f"> 
   11: *> [TGZ]</a> 
   12: *> <a href="http://www.netlib.org/cgi-bin/netlibfiles.zip?format=zip&filename=/lapack/lapack_routine/zgges.f"> 
   13: *> [ZIP]</a> 
   14: *> <a href="http://www.netlib.org/cgi-bin/netlibfiles.txt?format=txt&filename=/lapack/lapack_routine/zgges.f"> 
   15: *> [TXT]</a>
   16: *> \endhtmlonly 
   17: *
   18: *  Definition:
   19: *  ===========
   20: *
   21: *       SUBROUTINE ZGGES( JOBVSL, JOBVSR, SORT, SELCTG, N, A, LDA, B, LDB,
   22: *                         SDIM, ALPHA, BETA, VSL, LDVSL, VSR, LDVSR, WORK,
   23: *                         LWORK, RWORK, BWORK, INFO )
   24:    25: *       .. Scalar Arguments ..
   26: *       CHARACTER          JOBVSL, JOBVSR, SORT
   27: *       INTEGER            INFO, LDA, LDB, LDVSL, LDVSR, LWORK, N, SDIM
   28: *       ..
   29: *       .. Array Arguments ..
   30: *       LOGICAL            BWORK( * )
   31: *       DOUBLE PRECISION   RWORK( * )
   32: *       COMPLEX*16         A( LDA, * ), ALPHA( * ), B( LDB, * ),
   33: *      $                   BETA( * ), VSL( LDVSL, * ), VSR( LDVSR, * ),
   34: *      $                   WORK( * )
   35: *       ..
   36: *       .. Function Arguments ..
   37: *       LOGICAL            SELCTG
   38: *       EXTERNAL           SELCTG
   39: *       ..
   40: *  
   41: *
   42: *> \par Purpose:
   43: *  =============
   44: *>
   45: *> \verbatim
   46: *>
   47: *> ZGGES computes for a pair of N-by-N complex nonsymmetric matrices
   48: *> (A,B), the generalized eigenvalues, the generalized complex Schur
   49: *> form (S, T), and optionally left and/or right Schur vectors (VSL
   50: *> and VSR). This gives the generalized Schur factorization
   51: *>
   52: *>         (A,B) = ( (VSL)*S*(VSR)**H, (VSL)*T*(VSR)**H )
   53: *>
   54: *> where (VSR)**H is the conjugate-transpose of VSR.
   55: *>
   56: *> Optionally, it also orders the eigenvalues so that a selected cluster
   57: *> of eigenvalues appears in the leading diagonal blocks of the upper
   58: *> triangular matrix S and the upper triangular matrix T. The leading
   59: *> columns of VSL and VSR then form an unitary basis for the
   60: *> corresponding left and right eigenspaces (deflating subspaces).
   61: *>
   62: *> (If only the generalized eigenvalues are needed, use the driver
   63: *> ZGGEV instead, which is faster.)
   64: *>
   65: *> A generalized eigenvalue for a pair of matrices (A,B) is a scalar w
   66: *> or a ratio alpha/beta = w, such that  A - w*B is singular.  It is
   67: *> usually represented as the pair (alpha,beta), as there is a
   68: *> reasonable interpretation for beta=0, and even for both being zero.
   69: *>
   70: *> A pair of matrices (S,T) is in generalized complex Schur form if S
   71: *> and T are upper triangular and, in addition, the diagonal elements
   72: *> of T are non-negative real numbers.
   73: *> \endverbatim
   74: *
   75: *  Arguments:
   76: *  ==========
   77: *
   78: *> \param[in] JOBVSL
   79: *> \verbatim
   80: *>          JOBVSL is CHARACTER*1
   81: *>          = 'N':  do not compute the left Schur vectors;
   82: *>          = 'V':  compute the left Schur vectors.
   83: *> \endverbatim
   84: *>
   85: *> \param[in] JOBVSR
   86: *> \verbatim
   87: *>          JOBVSR is CHARACTER*1
   88: *>          = 'N':  do not compute the right Schur vectors;
   89: *>          = 'V':  compute the right Schur vectors.
   90: *> \endverbatim
   91: *>
   92: *> \param[in] SORT
   93: *> \verbatim
   94: *>          SORT is CHARACTER*1
   95: *>          Specifies whether or not to order the eigenvalues on the
   96: *>          diagonal of the generalized Schur form.
   97: *>          = 'N':  Eigenvalues are not ordered;
   98: *>          = 'S':  Eigenvalues are ordered (see SELCTG).
   99: *> \endverbatim
  100: *>
  101: *> \param[in] SELCTG
  102: *> \verbatim
  103: *>          SELCTG is a LOGICAL FUNCTION of two COMPLEX*16 arguments
  104: *>          SELCTG must be declared EXTERNAL in the calling subroutine.
  105: *>          If SORT = 'N', SELCTG is not referenced.
  106: *>          If SORT = 'S', SELCTG is used to select eigenvalues to sort
  107: *>          to the top left of the Schur form.
  108: *>          An eigenvalue ALPHA(j)/BETA(j) is selected if
  109: *>          SELCTG(ALPHA(j),BETA(j)) is true.
  110: *>
  111: *>          Note that a selected complex eigenvalue may no longer satisfy
  112: *>          SELCTG(ALPHA(j),BETA(j)) = .TRUE. after ordering, since
  113: *>          ordering may change the value of complex eigenvalues
  114: *>          (especially if the eigenvalue is ill-conditioned), in this
  115: *>          case INFO is set to N+2 (See INFO below).
  116: *> \endverbatim
  117: *>
  118: *> \param[in] N
  119: *> \verbatim
  120: *>          N is INTEGER
  121: *>          The order of the matrices A, B, VSL, and VSR.  N >= 0.
  122: *> \endverbatim
  123: *>
  124: *> \param[in,out] A
  125: *> \verbatim
  126: *>          A is COMPLEX*16 array, dimension (LDA, N)
  127: *>          On entry, the first of the pair of matrices.
  128: *>          On exit, A has been overwritten by its generalized Schur
  129: *>          form S.
  130: *> \endverbatim
  131: *>
  132: *> \param[in] LDA
  133: *> \verbatim
  134: *>          LDA is INTEGER
  135: *>          The leading dimension of A.  LDA >= max(1,N).
  136: *> \endverbatim
  137: *>
  138: *> \param[in,out] B
  139: *> \verbatim
  140: *>          B is COMPLEX*16 array, dimension (LDB, N)
  141: *>          On entry, the second of the pair of matrices.
  142: *>          On exit, B has been overwritten by its generalized Schur
  143: *>          form T.
  144: *> \endverbatim
  145: *>
  146: *> \param[in] LDB
  147: *> \verbatim
  148: *>          LDB is INTEGER
  149: *>          The leading dimension of B.  LDB >= max(1,N).
  150: *> \endverbatim
  151: *>
  152: *> \param[out] SDIM
  153: *> \verbatim
  154: *>          SDIM is INTEGER
  155: *>          If SORT = 'N', SDIM = 0.
  156: *>          If SORT = 'S', SDIM = number of eigenvalues (after sorting)
  157: *>          for which SELCTG is true.
  158: *> \endverbatim
  159: *>
  160: *> \param[out] ALPHA
  161: *> \verbatim
  162: *>          ALPHA is COMPLEX*16 array, dimension (N)
  163: *> \endverbatim
  164: *>
  165: *> \param[out] BETA
  166: *> \verbatim
  167: *>          BETA is COMPLEX*16 array, dimension (N)
  168: *>          On exit,  ALPHA(j)/BETA(j), j=1,...,N, will be the
  169: *>          generalized eigenvalues.  ALPHA(j), j=1,...,N  and  BETA(j),
  170: *>          j=1,...,N  are the diagonals of the complex Schur form (A,B)
  171: *>          output by ZGGES. The  BETA(j) will be non-negative real.
  172: *>
  173: *>          Note: the quotients ALPHA(j)/BETA(j) may easily over- or
  174: *>          underflow, and BETA(j) may even be zero.  Thus, the user
  175: *>          should avoid naively computing the ratio alpha/beta.
  176: *>          However, ALPHA will be always less than and usually
  177: *>          comparable with norm(A) in magnitude, and BETA always less
  178: *>          than and usually comparable with norm(B).
  179: *> \endverbatim
  180: *>
  181: *> \param[out] VSL
  182: *> \verbatim
  183: *>          VSL is COMPLEX*16 array, dimension (LDVSL,N)
  184: *>          If JOBVSL = 'V', VSL will contain the left Schur vectors.
  185: *>          Not referenced if JOBVSL = 'N'.
  186: *> \endverbatim
  187: *>
  188: *> \param[in] LDVSL
  189: *> \verbatim
  190: *>          LDVSL is INTEGER
  191: *>          The leading dimension of the matrix VSL. LDVSL >= 1, and
  192: *>          if JOBVSL = 'V', LDVSL >= N.
  193: *> \endverbatim
  194: *>
  195: *> \param[out] VSR
  196: *> \verbatim
  197: *>          VSR is COMPLEX*16 array, dimension (LDVSR,N)
  198: *>          If JOBVSR = 'V', VSR will contain the right Schur vectors.
  199: *>          Not referenced if JOBVSR = 'N'.
  200: *> \endverbatim
  201: *>
  202: *> \param[in] LDVSR
  203: *> \verbatim
  204: *>          LDVSR is INTEGER
  205: *>          The leading dimension of the matrix VSR. LDVSR >= 1, and
  206: *>          if JOBVSR = 'V', LDVSR >= N.
  207: *> \endverbatim
  208: *>
  209: *> \param[out] WORK
  210: *> \verbatim
  211: *>          WORK is COMPLEX*16 array, dimension (MAX(1,LWORK))
  212: *>          On exit, if INFO = 0, WORK(1) returns the optimal LWORK.
  213: *> \endverbatim
  214: *>
  215: *> \param[in] LWORK
  216: *> \verbatim
  217: *>          LWORK is INTEGER
  218: *>          The dimension of the array WORK.  LWORK >= max(1,2*N).
  219: *>          For good performance, LWORK must generally be larger.
  220: *>
  221: *>          If LWORK = -1, then a workspace query is assumed; the routine
  222: *>          only calculates the optimal size of the WORK array, returns
  223: *>          this value as the first entry of the WORK array, and no error
  224: *>          message related to LWORK is issued by XERBLA.
  225: *> \endverbatim
  226: *>
  227: *> \param[out] RWORK
  228: *> \verbatim
  229: *>          RWORK is DOUBLE PRECISION array, dimension (8*N)
  230: *> \endverbatim
  231: *>
  232: *> \param[out] BWORK
  233: *> \verbatim
  234: *>          BWORK is LOGICAL array, dimension (N)
  235: *>          Not referenced if SORT = 'N'.
  236: *> \endverbatim
  237: *>
  238: *> \param[out] INFO
  239: *> \verbatim
  240: *>          INFO is INTEGER
  241: *>          = 0:  successful exit
  242: *>          < 0:  if INFO = -i, the i-th argument had an illegal value.
  243: *>          =1,...,N:
  244: *>                The QZ iteration failed.  (A,B) are not in Schur
  245: *>                form, but ALPHA(j) and BETA(j) should be correct for
  246: *>                j=INFO+1,...,N.
  247: *>          > N:  =N+1: other than QZ iteration failed in ZHGEQZ
  248: *>                =N+2: after reordering, roundoff changed values of
  249: *>                      some complex eigenvalues so that leading
  250: *>                      eigenvalues in the Generalized Schur form no
  251: *>                      longer satisfy SELCTG=.TRUE.  This could also
  252: *>                      be caused due to scaling.
  253: *>                =N+3: reordering falied in ZTGSEN.
  254: *> \endverbatim
  255: *
  256: *  Authors:
  257: *  ========
  258: *
  259: *> \author Univ. of Tennessee 
  260: *> \author Univ. of California Berkeley 
  261: *> \author Univ. of Colorado Denver 
  262: *> \author NAG Ltd. 
  263: *
  264: *> \date November 2011
  265: *
  266: *> \ingroup complex16GEeigen
  267: *
  268: *  =====================================================================
  269:       SUBROUTINE ZGGES( JOBVSL, JOBVSR, SORT, SELCTG, N, A, LDA, B, LDB,
  270:      $                  SDIM, ALPHA, BETA, VSL, LDVSL, VSR, LDVSR, WORK,
  271:      $                  LWORK, RWORK, BWORK, INFO )
  272: *
  273: *  -- LAPACK driver routine (version 3.4.0) --
  274: *  -- LAPACK is a software package provided by Univ. of Tennessee,    --
  275: *  -- Univ. of California Berkeley, Univ. of Colorado Denver and NAG Ltd..--
  276: *     November 2011
  277: *
  278: *     .. Scalar Arguments ..
  279:       CHARACTER          JOBVSL, JOBVSR, SORT
  280:       INTEGER            INFO, LDA, LDB, LDVSL, LDVSR, LWORK, N, SDIM
  281: *     ..
  282: *     .. Array Arguments ..
  283:       LOGICAL            BWORK( * )
  284:       DOUBLE PRECISION   RWORK( * )
  285:       COMPLEX*16         A( LDA, * ), ALPHA( * ), B( LDB, * ),
  286:      $                   BETA( * ), VSL( LDVSL, * ), VSR( LDVSR, * ),
  287:      $                   WORK( * )
  288: *     ..
  289: *     .. Function Arguments ..
  290:       LOGICAL            SELCTG
  291:       EXTERNAL           SELCTG
  292: *     ..
  293: *
  294: *  =====================================================================
  295: *
  296: *     .. Parameters ..
  297:       DOUBLE PRECISION   ZERO, ONE
  298:       PARAMETER          ( ZERO = 0.0D0, ONE = 1.0D0 )
  299:       COMPLEX*16         CZERO, CONE
  300:       PARAMETER          ( CZERO = ( 0.0D0, 0.0D0 ),
  301:      $                   CONE = ( 1.0D0, 0.0D0 ) )
  302: *     ..
  303: *     .. Local Scalars ..
  304:       LOGICAL            CURSL, ILASCL, ILBSCL, ILVSL, ILVSR, LASTSL,
  305:      $                   LQUERY, WANTST
  306:       INTEGER            I, ICOLS, IERR, IHI, IJOBVL, IJOBVR, ILEFT,
  307:      $                   ILO, IRIGHT, IROWS, IRWRK, ITAU, IWRK, LWKMIN,
  308:      $                   LWKOPT
  309:       DOUBLE PRECISION   ANRM, ANRMTO, BIGNUM, BNRM, BNRMTO, EPS, PVSL,
  310:      $                   PVSR, SMLNUM
  311: *     ..
  312: *     .. Local Arrays ..
  313:       INTEGER            IDUM( 1 )
  314:       DOUBLE PRECISION   DIF( 2 )
  315: *     ..
  316: *     .. External Subroutines ..
  317:       EXTERNAL           DLABAD, XERBLA, ZGEQRF, ZGGBAK, ZGGBAL, ZGGHRD,
  318:      $                   ZHGEQZ, ZLACPY, ZLASCL, ZLASET, ZTGSEN, ZUNGQR,
  319:      $                   ZUNMQR
  320: *     ..
  321: *     .. External Functions ..
  322:       LOGICAL            LSAME
  323:       INTEGER            ILAENV
  324:       DOUBLE PRECISION   DLAMCH, ZLANGE
  325:       EXTERNAL           LSAME, ILAENV, DLAMCH, ZLANGE
  326: *     ..
  327: *     .. Intrinsic Functions ..
  328:       INTRINSIC          MAX, SQRT
  329: *     ..
  330: *     .. Executable Statements ..
  331: *
  332: *     Decode the input arguments
  333: *
  334:       IF( LSAME( JOBVSL, 'N' ) ) THEN
  335:          IJOBVL = 1
  336:          ILVSL = .FALSE.
  337:       ELSE IF( LSAME( JOBVSL, 'V' ) ) THEN
  338:          IJOBVL = 2
  339:          ILVSL = .TRUE.
  340:       ELSE
  341:          IJOBVL = -1
  342:          ILVSL = .FALSE.
  343:       END IF
  344: *
  345:       IF( LSAME( JOBVSR, 'N' ) ) THEN
  346:          IJOBVR = 1
  347:          ILVSR = .FALSE.
  348:       ELSE IF( LSAME( JOBVSR, 'V' ) ) THEN
  349:          IJOBVR = 2
  350:          ILVSR = .TRUE.
  351:       ELSE
  352:          IJOBVR = -1
  353:          ILVSR = .FALSE.
  354:       END IF
  355: *
  356:       WANTST = LSAME( SORT, 'S' )
  357: *
  358: *     Test the input arguments
  359: *
  360:       INFO = 0
  361:       LQUERY = ( LWORK.EQ.-1 )
  362:       IF( IJOBVL.LE.0 ) THEN
  363:          INFO = -1
  364:       ELSE IF( IJOBVR.LE.0 ) THEN
  365:          INFO = -2
  366:       ELSE IF( ( .NOT.WANTST ) .AND. ( .NOT.LSAME( SORT, 'N' ) ) ) THEN
  367:          INFO = -3
  368:       ELSE IF( N.LT.0 ) THEN
  369:          INFO = -5
  370:       ELSE IF( LDA.LT.MAX( 1, N ) ) THEN
  371:          INFO = -7
  372:       ELSE IF( LDB.LT.MAX( 1, N ) ) THEN
  373:          INFO = -9
  374:       ELSE IF( LDVSL.LT.1 .OR. ( ILVSL .AND. LDVSL.LT.N ) ) THEN
  375:          INFO = -14
  376:       ELSE IF( LDVSR.LT.1 .OR. ( ILVSR .AND. LDVSR.LT.N ) ) THEN
  377:          INFO = -16
  378:       END IF
  379: *
  380: *     Compute workspace
  381: *      (Note: Comments in the code beginning "Workspace:" describe the
  382: *       minimal amount of workspace needed at that point in the code,
  383: *       as well as the preferred amount for good performance.
  384: *       NB refers to the optimal block size for the immediately
  385: *       following subroutine, as returned by ILAENV.)
  386: *
  387:       IF( INFO.EQ.0 ) THEN
  388:          LWKMIN = MAX( 1, 2*N )
  389:          LWKOPT = MAX( 1, N + N*ILAENV( 1, 'ZGEQRF', ' ', N, 1, N, 0 ) )
  390:          LWKOPT = MAX( LWKOPT, N +
  391:      $                 N*ILAENV( 1, 'ZUNMQR', ' ', N, 1, N, -1 ) )
  392:          IF( ILVSL ) THEN
  393:             LWKOPT = MAX( LWKOPT, N +
  394:      $                    N*ILAENV( 1, 'ZUNGQR', ' ', N, 1, N, -1 ) )
  395:          END IF
  396:          WORK( 1 ) = LWKOPT
  397: *
  398:          IF( LWORK.LT.LWKMIN .AND. .NOT.LQUERY )
  399:      $      INFO = -18
  400:       END IF
  401: *
  402:       IF( INFO.NE.0 ) THEN
  403:          CALL XERBLA( 'ZGGES ', -INFO )
  404:          RETURN
  405:       ELSE IF( LQUERY ) THEN
  406:          RETURN
  407:       END IF
  408: *
  409: *     Quick return if possible
  410: *
  411:       IF( N.EQ.0 ) THEN
  412:          SDIM = 0
  413:          RETURN
  414:       END IF
  415: *
  416: *     Get machine constants
  417: *
  418:       EPS = DLAMCH( 'P' )
  419:       SMLNUM = DLAMCH( 'S' )
  420:       BIGNUM = ONE / SMLNUM
  421:       CALL DLABAD( SMLNUM, BIGNUM )
  422:       SMLNUM = SQRT( SMLNUM ) / EPS
  423:       BIGNUM = ONE / SMLNUM
  424: *
  425: *     Scale A if max element outside range [SMLNUM,BIGNUM]
  426: *
  427:       ANRM = ZLANGE( 'M', N, N, A, LDA, RWORK )
  428:       ILASCL = .FALSE.
  429:       IF( ANRM.GT.ZERO .AND. ANRM.LT.SMLNUM ) THEN
  430:          ANRMTO = SMLNUM
  431:          ILASCL = .TRUE.
  432:       ELSE IF( ANRM.GT.BIGNUM ) THEN
  433:          ANRMTO = BIGNUM
  434:          ILASCL = .TRUE.
  435:       END IF
  436: *
  437:       IF( ILASCL )
  438:      $   CALL ZLASCL( 'G', 0, 0, ANRM, ANRMTO, N, N, A, LDA, IERR )
  439: *
  440: *     Scale B if max element outside range [SMLNUM,BIGNUM]
  441: *
  442:       BNRM = ZLANGE( 'M', N, N, B, LDB, RWORK )
  443:       ILBSCL = .FALSE.
  444:       IF( BNRM.GT.ZERO .AND. BNRM.LT.SMLNUM ) THEN
  445:          BNRMTO = SMLNUM
  446:          ILBSCL = .TRUE.
  447:       ELSE IF( BNRM.GT.BIGNUM ) THEN
  448:          BNRMTO = BIGNUM
  449:          ILBSCL = .TRUE.
  450:       END IF
  451: *
  452:       IF( ILBSCL )
  453:      $   CALL ZLASCL( 'G', 0, 0, BNRM, BNRMTO, N, N, B, LDB, IERR )
  454: *
  455: *     Permute the matrix to make it more nearly triangular
  456: *     (Real Workspace: need 6*N)
  457: *
  458:       ILEFT = 1
  459:       IRIGHT = N + 1
  460:       IRWRK = IRIGHT + N
  461:       CALL ZGGBAL( 'P', N, A, LDA, B, LDB, ILO, IHI, RWORK( ILEFT ),
  462:      $             RWORK( IRIGHT ), RWORK( IRWRK ), IERR )
  463: *
  464: *     Reduce B to triangular form (QR decomposition of B)
  465: *     (Complex Workspace: need N, prefer N*NB)
  466: *
  467:       IROWS = IHI + 1 - ILO
  468:       ICOLS = N + 1 - ILO
  469:       ITAU = 1
  470:       IWRK = ITAU + IROWS
  471:       CALL ZGEQRF( IROWS, ICOLS, B( ILO, ILO ), LDB, WORK( ITAU ),
  472:      $             WORK( IWRK ), LWORK+1-IWRK, IERR )
  473: *
  474: *     Apply the orthogonal transformation to matrix A
  475: *     (Complex Workspace: need N, prefer N*NB)
  476: *
  477:       CALL ZUNMQR( 'L', 'C', IROWS, ICOLS, IROWS, B( ILO, ILO ), LDB,
  478:      $             WORK( ITAU ), A( ILO, ILO ), LDA, WORK( IWRK ),
  479:      $             LWORK+1-IWRK, IERR )
  480: *
  481: *     Initialize VSL
  482: *     (Complex Workspace: need N, prefer N*NB)
  483: *
  484:       IF( ILVSL ) THEN
  485:          CALL ZLASET( 'Full', N, N, CZERO, CONE, VSL, LDVSL )
  486:          IF( IROWS.GT.1 ) THEN
  487:             CALL ZLACPY( 'L', IROWS-1, IROWS-1, B( ILO+1, ILO ), LDB,
  488:      $                   VSL( ILO+1, ILO ), LDVSL )
  489:          END IF
  490:          CALL ZUNGQR( IROWS, IROWS, IROWS, VSL( ILO, ILO ), LDVSL,
  491:      $                WORK( ITAU ), WORK( IWRK ), LWORK+1-IWRK, IERR )
  492:       END IF
  493: *
  494: *     Initialize VSR
  495: *
  496:       IF( ILVSR )
  497:      $   CALL ZLASET( 'Full', N, N, CZERO, CONE, VSR, LDVSR )
  498: *
  499: *     Reduce to generalized Hessenberg form
  500: *     (Workspace: none needed)
  501: *
  502:       CALL ZGGHRD( JOBVSL, JOBVSR, N, ILO, IHI, A, LDA, B, LDB, VSL,
  503:      $             LDVSL, VSR, LDVSR, IERR )
  504: *
  505:       SDIM = 0
  506: *
  507: *     Perform QZ algorithm, computing Schur vectors if desired
  508: *     (Complex Workspace: need N)
  509: *     (Real Workspace: need N)
  510: *
  511:       IWRK = ITAU
  512:       CALL ZHGEQZ( 'S', JOBVSL, JOBVSR, N, ILO, IHI, A, LDA, B, LDB,
  513:      $             ALPHA, BETA, VSL, LDVSL, VSR, LDVSR, WORK( IWRK ),
  514:      $             LWORK+1-IWRK, RWORK( IRWRK ), IERR )
  515:       IF( IERR.NE.0 ) THEN
  516:          IF( IERR.GT.0 .AND. IERR.LE.N ) THEN
  517:             INFO = IERR
  518:          ELSE IF( IERR.GT.N .AND. IERR.LE.2*N ) THEN
  519:             INFO = IERR - N
  520:          ELSE
  521:             INFO = N + 1
  522:          END IF
  523:          GO TO 30
  524:       END IF
  525: *
  526: *     Sort eigenvalues ALPHA/BETA if desired
  527: *     (Workspace: none needed)
  528: *
  529:       IF( WANTST ) THEN
  530: *
  531: *        Undo scaling on eigenvalues before selecting
  532: *
  533:          IF( ILASCL )
  534:      $      CALL ZLASCL( 'G', 0, 0, ANRM, ANRMTO, N, 1, ALPHA, N, IERR )
  535:          IF( ILBSCL )
  536:      $      CALL ZLASCL( 'G', 0, 0, BNRM, BNRMTO, N, 1, BETA, N, IERR )
  537: *
  538: *        Select eigenvalues
  539: *
  540:          DO 10 I = 1, N
  541:             BWORK( I ) = SELCTG( ALPHA( I ), BETA( I ) )
  542:    10    CONTINUE
  543: *
  544:          CALL ZTGSEN( 0, ILVSL, ILVSR, BWORK, N, A, LDA, B, LDB, ALPHA,
  545:      $                BETA, VSL, LDVSL, VSR, LDVSR, SDIM, PVSL, PVSR,
  546:      $                DIF, WORK( IWRK ), LWORK-IWRK+1, IDUM, 1, IERR )
  547:          IF( IERR.EQ.1 )
  548:      $      INFO = N + 3
  549: *
  550:       END IF
  551: *
  552: *     Apply back-permutation to VSL and VSR
  553: *     (Workspace: none needed)
  554: *
  555:       IF( ILVSL )
  556:      $   CALL ZGGBAK( 'P', 'L', N, ILO, IHI, RWORK( ILEFT ),
  557:      $                RWORK( IRIGHT ), N, VSL, LDVSL, IERR )
  558:       IF( ILVSR )
  559:      $   CALL ZGGBAK( 'P', 'R', N, ILO, IHI, RWORK( ILEFT ),
  560:      $                RWORK( IRIGHT ), N, VSR, LDVSR, IERR )
  561: *
  562: *     Undo scaling
  563: *
  564:       IF( ILASCL ) THEN
  565:          CALL ZLASCL( 'U', 0, 0, ANRMTO, ANRM, N, N, A, LDA, IERR )
  566:          CALL ZLASCL( 'G', 0, 0, ANRMTO, ANRM, N, 1, ALPHA, N, IERR )
  567:       END IF
  568: *
  569:       IF( ILBSCL ) THEN
  570:          CALL ZLASCL( 'U', 0, 0, BNRMTO, BNRM, N, N, B, LDB, IERR )
  571:          CALL ZLASCL( 'G', 0, 0, BNRMTO, BNRM, N, 1, BETA, N, IERR )
  572:       END IF
  573: *
  574:       IF( WANTST ) THEN
  575: *
  576: *        Check if reordering is correct
  577: *
  578:          LASTSL = .TRUE.
  579:          SDIM = 0
  580:          DO 20 I = 1, N
  581:             CURSL = SELCTG( ALPHA( I ), BETA( I ) )
  582:             IF( CURSL )
  583:      $         SDIM = SDIM + 1
  584:             IF( CURSL .AND. .NOT.LASTSL )
  585:      $         INFO = N + 2
  586:             LASTSL = CURSL
  587:    20    CONTINUE
  588: *
  589:       END IF
  590: *
  591:    30 CONTINUE
  592: *
  593:       WORK( 1 ) = LWKOPT
  594: *
  595:       RETURN
  596: *
  597: *     End of ZGGES
  598: *
  599:       END

CVSweb interface <joel.bertrand@systella.fr>