File:  [local] / rpl / lapack / lapack / dstedc.f
Revision 1.11: download - view: text, annotated - select for diffs - revision graph
Fri Dec 14 14:22:40 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 DSTEBZ
    2: *
    3: *  =========== DOCUMENTATION ===========
    4: *
    5: * Online html documentation available at 
    6: *            http://www.netlib.org/lapack/explore-html/ 
    7: *
    8: *> \htmlonly
    9: *> Download DSTEDC + dependencies 
   10: *> <a href="http://www.netlib.org/cgi-bin/netlibfiles.tgz?format=tgz&filename=/lapack/lapack_routine/dstedc.f"> 
   11: *> [TGZ]</a> 
   12: *> <a href="http://www.netlib.org/cgi-bin/netlibfiles.zip?format=zip&filename=/lapack/lapack_routine/dstedc.f"> 
   13: *> [ZIP]</a> 
   14: *> <a href="http://www.netlib.org/cgi-bin/netlibfiles.txt?format=txt&filename=/lapack/lapack_routine/dstedc.f"> 
   15: *> [TXT]</a>
   16: *> \endhtmlonly 
   17: *
   18: *  Definition:
   19: *  ===========
   20: *
   21: *       SUBROUTINE DSTEDC( COMPZ, N, D, E, Z, LDZ, WORK, LWORK, IWORK,
   22: *                          LIWORK, INFO )
   23:    24: *       .. Scalar Arguments ..
   25: *       CHARACTER          COMPZ
   26: *       INTEGER            INFO, LDZ, LIWORK, LWORK, N
   27: *       ..
   28: *       .. Array Arguments ..
   29: *       INTEGER            IWORK( * )
   30: *       DOUBLE PRECISION   D( * ), E( * ), WORK( * ), Z( LDZ, * )
   31: *       ..
   32: *  
   33: *
   34: *> \par Purpose:
   35: *  =============
   36: *>
   37: *> \verbatim
   38: *>
   39: *> DSTEDC computes all eigenvalues and, optionally, eigenvectors of a
   40: *> symmetric tridiagonal matrix using the divide and conquer method.
   41: *> The eigenvectors of a full or band real symmetric matrix can also be
   42: *> found if DSYTRD or DSPTRD or DSBTRD has been used to reduce this
   43: *> matrix to tridiagonal form.
   44: *>
   45: *> This code makes very mild assumptions about floating point
   46: *> arithmetic. It will work on machines with a guard digit in
   47: *> add/subtract, or on those binary machines without guard digits
   48: *> which subtract like the Cray X-MP, Cray Y-MP, Cray C-90, or Cray-2.
   49: *> It could conceivably fail on hexadecimal or decimal machines
   50: *> without guard digits, but we know of none.  See DLAED3 for details.
   51: *> \endverbatim
   52: *
   53: *  Arguments:
   54: *  ==========
   55: *
   56: *> \param[in] COMPZ
   57: *> \verbatim
   58: *>          COMPZ is CHARACTER*1
   59: *>          = 'N':  Compute eigenvalues only.
   60: *>          = 'I':  Compute eigenvectors of tridiagonal matrix also.
   61: *>          = 'V':  Compute eigenvectors of original dense symmetric
   62: *>                  matrix also.  On entry, Z contains the orthogonal
   63: *>                  matrix used to reduce the original matrix to
   64: *>                  tridiagonal form.
   65: *> \endverbatim
   66: *>
   67: *> \param[in] N
   68: *> \verbatim
   69: *>          N is INTEGER
   70: *>          The dimension of the symmetric tridiagonal matrix.  N >= 0.
   71: *> \endverbatim
   72: *>
   73: *> \param[in,out] D
   74: *> \verbatim
   75: *>          D is DOUBLE PRECISION array, dimension (N)
   76: *>          On entry, the diagonal elements of the tridiagonal matrix.
   77: *>          On exit, if INFO = 0, the eigenvalues in ascending order.
   78: *> \endverbatim
   79: *>
   80: *> \param[in,out] E
   81: *> \verbatim
   82: *>          E is DOUBLE PRECISION array, dimension (N-1)
   83: *>          On entry, the subdiagonal elements of the tridiagonal matrix.
   84: *>          On exit, E has been destroyed.
   85: *> \endverbatim
   86: *>
   87: *> \param[in,out] Z
   88: *> \verbatim
   89: *>          Z is DOUBLE PRECISION array, dimension (LDZ,N)
   90: *>          On entry, if COMPZ = 'V', then Z contains the orthogonal
   91: *>          matrix used in the reduction to tridiagonal form.
   92: *>          On exit, if INFO = 0, then if COMPZ = 'V', Z contains the
   93: *>          orthonormal eigenvectors of the original symmetric matrix,
   94: *>          and if COMPZ = 'I', Z contains the orthonormal eigenvectors
   95: *>          of the symmetric tridiagonal matrix.
   96: *>          If  COMPZ = 'N', then Z is not referenced.
   97: *> \endverbatim
   98: *>
   99: *> \param[in] LDZ
  100: *> \verbatim
  101: *>          LDZ is INTEGER
  102: *>          The leading dimension of the array Z.  LDZ >= 1.
  103: *>          If eigenvectors are desired, then LDZ >= max(1,N).
  104: *> \endverbatim
  105: *>
  106: *> \param[out] WORK
  107: *> \verbatim
  108: *>          WORK is DOUBLE PRECISION array,
  109: *>                                         dimension (LWORK)
  110: *>          On exit, if INFO = 0, WORK(1) returns the optimal LWORK.
  111: *> \endverbatim
  112: *>
  113: *> \param[in] LWORK
  114: *> \verbatim
  115: *>          LWORK is INTEGER
  116: *>          The dimension of the array WORK.
  117: *>          If COMPZ = 'N' or N <= 1 then LWORK must be at least 1.
  118: *>          If COMPZ = 'V' and N > 1 then LWORK must be at least
  119: *>                         ( 1 + 3*N + 2*N*lg N + 4*N**2 ),
  120: *>                         where lg( N ) = smallest integer k such
  121: *>                         that 2**k >= N.
  122: *>          If COMPZ = 'I' and N > 1 then LWORK must be at least
  123: *>                         ( 1 + 4*N + N**2 ).
  124: *>          Note that for COMPZ = 'I' or 'V', then if N is less than or
  125: *>          equal to the minimum divide size, usually 25, then LWORK need
  126: *>          only be max(1,2*(N-1)).
  127: *>
  128: *>          If LWORK = -1, then a workspace query is assumed; the routine
  129: *>          only calculates the optimal size of the WORK array, returns
  130: *>          this value as the first entry of the WORK array, and no error
  131: *>          message related to LWORK is issued by XERBLA.
  132: *> \endverbatim
  133: *>
  134: *> \param[out] IWORK
  135: *> \verbatim
  136: *>          IWORK is INTEGER array, dimension (MAX(1,LIWORK))
  137: *>          On exit, if INFO = 0, IWORK(1) returns the optimal LIWORK.
  138: *> \endverbatim
  139: *>
  140: *> \param[in] LIWORK
  141: *> \verbatim
  142: *>          LIWORK is INTEGER
  143: *>          The dimension of the array IWORK.
  144: *>          If COMPZ = 'N' or N <= 1 then LIWORK must be at least 1.
  145: *>          If COMPZ = 'V' and N > 1 then LIWORK must be at least
  146: *>                         ( 6 + 6*N + 5*N*lg N ).
  147: *>          If COMPZ = 'I' and N > 1 then LIWORK must be at least
  148: *>                         ( 3 + 5*N ).
  149: *>          Note that for COMPZ = 'I' or 'V', then if N is less than or
  150: *>          equal to the minimum divide size, usually 25, then LIWORK
  151: *>          need only be 1.
  152: *>
  153: *>          If LIWORK = -1, then a workspace query is assumed; the
  154: *>          routine only calculates the optimal size of the IWORK array,
  155: *>          returns this value as the first entry of the IWORK array, and
  156: *>          no error message related to LIWORK is issued by XERBLA.
  157: *> \endverbatim
  158: *>
  159: *> \param[out] INFO
  160: *> \verbatim
  161: *>          INFO is INTEGER
  162: *>          = 0:  successful exit.
  163: *>          < 0:  if INFO = -i, the i-th argument had an illegal value.
  164: *>          > 0:  The algorithm failed to compute an eigenvalue while
  165: *>                working on the submatrix lying in rows and columns
  166: *>                INFO/(N+1) through mod(INFO,N+1).
  167: *> \endverbatim
  168: *
  169: *  Authors:
  170: *  ========
  171: *
  172: *> \author Univ. of Tennessee 
  173: *> \author Univ. of California Berkeley 
  174: *> \author Univ. of Colorado Denver 
  175: *> \author NAG Ltd. 
  176: *
  177: *> \date November 2011
  178: *
  179: *> \ingroup auxOTHERcomputational
  180: *
  181: *> \par Contributors:
  182: *  ==================
  183: *>
  184: *> Jeff Rutter, Computer Science Division, University of California
  185: *> at Berkeley, USA \n
  186: *>  Modified by Francoise Tisseur, University of Tennessee
  187: *>
  188: *  =====================================================================
  189:       SUBROUTINE DSTEDC( COMPZ, N, D, E, Z, LDZ, WORK, LWORK, IWORK,
  190:      $                   LIWORK, INFO )
  191: *
  192: *  -- LAPACK computational routine (version 3.4.0) --
  193: *  -- LAPACK is a software package provided by Univ. of Tennessee,    --
  194: *  -- Univ. of California Berkeley, Univ. of Colorado Denver and NAG Ltd..--
  195: *     November 2011
  196: *
  197: *     .. Scalar Arguments ..
  198:       CHARACTER          COMPZ
  199:       INTEGER            INFO, LDZ, LIWORK, LWORK, N
  200: *     ..
  201: *     .. Array Arguments ..
  202:       INTEGER            IWORK( * )
  203:       DOUBLE PRECISION   D( * ), E( * ), WORK( * ), Z( LDZ, * )
  204: *     ..
  205: *
  206: *  =====================================================================
  207: *
  208: *     .. Parameters ..
  209:       DOUBLE PRECISION   ZERO, ONE, TWO
  210:       PARAMETER          ( ZERO = 0.0D0, ONE = 1.0D0, TWO = 2.0D0 )
  211: *     ..
  212: *     .. Local Scalars ..
  213:       LOGICAL            LQUERY
  214:       INTEGER            FINISH, I, ICOMPZ, II, J, K, LGN, LIWMIN,
  215:      $                   LWMIN, M, SMLSIZ, START, STOREZ, STRTRW
  216:       DOUBLE PRECISION   EPS, ORGNRM, P, TINY
  217: *     ..
  218: *     .. External Functions ..
  219:       LOGICAL            LSAME
  220:       INTEGER            ILAENV
  221:       DOUBLE PRECISION   DLAMCH, DLANST
  222:       EXTERNAL           LSAME, ILAENV, DLAMCH, DLANST
  223: *     ..
  224: *     .. External Subroutines ..
  225:       EXTERNAL           DGEMM, DLACPY, DLAED0, DLASCL, DLASET, DLASRT,
  226:      $                   DSTEQR, DSTERF, DSWAP, XERBLA
  227: *     ..
  228: *     .. Intrinsic Functions ..
  229:       INTRINSIC          ABS, DBLE, INT, LOG, MAX, MOD, SQRT
  230: *     ..
  231: *     .. Executable Statements ..
  232: *
  233: *     Test the input parameters.
  234: *
  235:       INFO = 0
  236:       LQUERY = ( LWORK.EQ.-1 .OR. LIWORK.EQ.-1 )
  237: *
  238:       IF( LSAME( COMPZ, 'N' ) ) THEN
  239:          ICOMPZ = 0
  240:       ELSE IF( LSAME( COMPZ, 'V' ) ) THEN
  241:          ICOMPZ = 1
  242:       ELSE IF( LSAME( COMPZ, 'I' ) ) THEN
  243:          ICOMPZ = 2
  244:       ELSE
  245:          ICOMPZ = -1
  246:       END IF
  247:       IF( ICOMPZ.LT.0 ) THEN
  248:          INFO = -1
  249:       ELSE IF( N.LT.0 ) THEN
  250:          INFO = -2
  251:       ELSE IF( ( LDZ.LT.1 ) .OR.
  252:      $         ( ICOMPZ.GT.0 .AND. LDZ.LT.MAX( 1, N ) ) ) THEN
  253:          INFO = -6
  254:       END IF
  255: *
  256:       IF( INFO.EQ.0 ) THEN
  257: *
  258: *        Compute the workspace requirements
  259: *
  260:          SMLSIZ = ILAENV( 9, 'DSTEDC', ' ', 0, 0, 0, 0 )
  261:          IF( N.LE.1 .OR. ICOMPZ.EQ.0 ) THEN
  262:             LIWMIN = 1
  263:             LWMIN = 1
  264:          ELSE IF( N.LE.SMLSIZ ) THEN
  265:             LIWMIN = 1
  266:             LWMIN = 2*( N - 1 )
  267:          ELSE
  268:             LGN = INT( LOG( DBLE( N ) )/LOG( TWO ) )
  269:             IF( 2**LGN.LT.N )
  270:      $         LGN = LGN + 1
  271:             IF( 2**LGN.LT.N )
  272:      $         LGN = LGN + 1
  273:             IF( ICOMPZ.EQ.1 ) THEN
  274:                LWMIN = 1 + 3*N + 2*N*LGN + 4*N**2
  275:                LIWMIN = 6 + 6*N + 5*N*LGN
  276:             ELSE IF( ICOMPZ.EQ.2 ) THEN
  277:                LWMIN = 1 + 4*N + N**2
  278:                LIWMIN = 3 + 5*N
  279:             END IF
  280:          END IF
  281:          WORK( 1 ) = LWMIN
  282:          IWORK( 1 ) = LIWMIN
  283: *
  284:          IF( LWORK.LT.LWMIN .AND. .NOT. LQUERY ) THEN
  285:             INFO = -8
  286:          ELSE IF( LIWORK.LT.LIWMIN .AND. .NOT. LQUERY ) THEN
  287:             INFO = -10
  288:          END IF
  289:       END IF
  290: *
  291:       IF( INFO.NE.0 ) THEN
  292:          CALL XERBLA( 'DSTEDC', -INFO )
  293:          RETURN
  294:       ELSE IF (LQUERY) THEN
  295:          RETURN
  296:       END IF
  297: *
  298: *     Quick return if possible
  299: *
  300:       IF( N.EQ.0 )
  301:      $   RETURN
  302:       IF( N.EQ.1 ) THEN
  303:          IF( ICOMPZ.NE.0 )
  304:      $      Z( 1, 1 ) = ONE
  305:          RETURN
  306:       END IF
  307: *
  308: *     If the following conditional clause is removed, then the routine
  309: *     will use the Divide and Conquer routine to compute only the
  310: *     eigenvalues, which requires (3N + 3N**2) real workspace and
  311: *     (2 + 5N + 2N lg(N)) integer workspace.
  312: *     Since on many architectures DSTERF is much faster than any other
  313: *     algorithm for finding eigenvalues only, it is used here
  314: *     as the default. If the conditional clause is removed, then
  315: *     information on the size of workspace needs to be changed.
  316: *
  317: *     If COMPZ = 'N', use DSTERF to compute the eigenvalues.
  318: *
  319:       IF( ICOMPZ.EQ.0 ) THEN
  320:          CALL DSTERF( N, D, E, INFO )
  321:          GO TO 50
  322:       END IF
  323: *
  324: *     If N is smaller than the minimum divide size (SMLSIZ+1), then
  325: *     solve the problem with another solver.
  326: *
  327:       IF( N.LE.SMLSIZ ) THEN
  328: *
  329:          CALL DSTEQR( COMPZ, N, D, E, Z, LDZ, WORK, INFO )
  330: *
  331:       ELSE
  332: *
  333: *        If COMPZ = 'V', the Z matrix must be stored elsewhere for later
  334: *        use.
  335: *
  336:          IF( ICOMPZ.EQ.1 ) THEN
  337:             STOREZ = 1 + N*N
  338:          ELSE
  339:             STOREZ = 1
  340:          END IF
  341: *
  342:          IF( ICOMPZ.EQ.2 ) THEN
  343:             CALL DLASET( 'Full', N, N, ZERO, ONE, Z, LDZ )
  344:          END IF
  345: *
  346: *        Scale.
  347: *
  348:          ORGNRM = DLANST( 'M', N, D, E )
  349:          IF( ORGNRM.EQ.ZERO )
  350:      $      GO TO 50
  351: *
  352:          EPS = DLAMCH( 'Epsilon' )
  353: *
  354:          START = 1
  355: *
  356: *        while ( START <= N )
  357: *
  358:    10    CONTINUE
  359:          IF( START.LE.N ) THEN
  360: *
  361: *           Let FINISH be the position of the next subdiagonal entry
  362: *           such that E( FINISH ) <= TINY or FINISH = N if no such
  363: *           subdiagonal exists.  The matrix identified by the elements
  364: *           between START and FINISH constitutes an independent
  365: *           sub-problem.
  366: *
  367:             FINISH = START
  368:    20       CONTINUE
  369:             IF( FINISH.LT.N ) THEN
  370:                TINY = EPS*SQRT( ABS( D( FINISH ) ) )*
  371:      $                    SQRT( ABS( D( FINISH+1 ) ) )
  372:                IF( ABS( E( FINISH ) ).GT.TINY ) THEN
  373:                   FINISH = FINISH + 1
  374:                   GO TO 20
  375:                END IF
  376:             END IF
  377: *
  378: *           (Sub) Problem determined.  Compute its size and solve it.
  379: *
  380:             M = FINISH - START + 1
  381:             IF( M.EQ.1 ) THEN
  382:                START = FINISH + 1
  383:                GO TO 10
  384:             END IF
  385:             IF( M.GT.SMLSIZ ) THEN
  386: *
  387: *              Scale.
  388: *
  389:                ORGNRM = DLANST( 'M', M, D( START ), E( START ) )
  390:                CALL DLASCL( 'G', 0, 0, ORGNRM, ONE, M, 1, D( START ), M,
  391:      $                      INFO )
  392:                CALL DLASCL( 'G', 0, 0, ORGNRM, ONE, M-1, 1, E( START ),
  393:      $                      M-1, INFO )
  394: *
  395:                IF( ICOMPZ.EQ.1 ) THEN
  396:                   STRTRW = 1
  397:                ELSE
  398:                   STRTRW = START
  399:                END IF
  400:                CALL DLAED0( ICOMPZ, N, M, D( START ), E( START ),
  401:      $                      Z( STRTRW, START ), LDZ, WORK( 1 ), N,
  402:      $                      WORK( STOREZ ), IWORK, INFO )
  403:                IF( INFO.NE.0 ) THEN
  404:                   INFO = ( INFO / ( M+1 )+START-1 )*( N+1 ) +
  405:      $                   MOD( INFO, ( M+1 ) ) + START - 1
  406:                   GO TO 50
  407:                END IF
  408: *
  409: *              Scale back.
  410: *
  411:                CALL DLASCL( 'G', 0, 0, ONE, ORGNRM, M, 1, D( START ), M,
  412:      $                      INFO )
  413: *
  414:             ELSE
  415:                IF( ICOMPZ.EQ.1 ) THEN
  416: *
  417: *                 Since QR won't update a Z matrix which is larger than
  418: *                 the length of D, we must solve the sub-problem in a
  419: *                 workspace and then multiply back into Z.
  420: *
  421:                   CALL DSTEQR( 'I', M, D( START ), E( START ), WORK, M,
  422:      $                         WORK( M*M+1 ), INFO )
  423:                   CALL DLACPY( 'A', N, M, Z( 1, START ), LDZ,
  424:      $                         WORK( STOREZ ), N )
  425:                   CALL DGEMM( 'N', 'N', N, M, M, ONE,
  426:      $                        WORK( STOREZ ), N, WORK, M, ZERO,
  427:      $                        Z( 1, START ), LDZ )
  428:                ELSE IF( ICOMPZ.EQ.2 ) THEN
  429:                   CALL DSTEQR( 'I', M, D( START ), E( START ),
  430:      $                         Z( START, START ), LDZ, WORK, INFO )
  431:                ELSE
  432:                   CALL DSTERF( M, D( START ), E( START ), INFO )
  433:                END IF
  434:                IF( INFO.NE.0 ) THEN
  435:                   INFO = START*( N+1 ) + FINISH
  436:                   GO TO 50
  437:                END IF
  438:             END IF
  439: *
  440:             START = FINISH + 1
  441:             GO TO 10
  442:          END IF
  443: *
  444: *        endwhile
  445: *
  446: *        If the problem split any number of times, then the eigenvalues
  447: *        will not be properly ordered.  Here we permute the eigenvalues
  448: *        (and the associated eigenvectors) into ascending order.
  449: *
  450:          IF( M.NE.N ) THEN
  451:             IF( ICOMPZ.EQ.0 ) THEN
  452: *
  453: *              Use Quick Sort
  454: *
  455:                CALL DLASRT( 'I', N, D, INFO )
  456: *
  457:             ELSE
  458: *
  459: *              Use Selection Sort to minimize swaps of eigenvectors
  460: *
  461:                DO 40 II = 2, N
  462:                   I = II - 1
  463:                   K = I
  464:                   P = D( I )
  465:                   DO 30 J = II, N
  466:                      IF( D( J ).LT.P ) THEN
  467:                         K = J
  468:                         P = D( J )
  469:                      END IF
  470:    30             CONTINUE
  471:                   IF( K.NE.I ) THEN
  472:                      D( K ) = D( I )
  473:                      D( I ) = P
  474:                      CALL DSWAP( N, Z( 1, I ), 1, Z( 1, K ), 1 )
  475:                   END IF
  476:    40          CONTINUE
  477:             END IF
  478:          END IF
  479:       END IF
  480: *
  481:    50 CONTINUE
  482:       WORK( 1 ) = LWMIN
  483:       IWORK( 1 ) = LIWMIN
  484: *
  485:       RETURN
  486: *
  487: *     End of DSTEDC
  488: *
  489:       END

CVSweb interface <joel.bertrand@systella.fr>