File:  [local] / rpl / lapack / lapack / dpstf2.f
Revision 1.17: download - view: text, annotated - select for diffs - revision graph
Mon Aug 7 08:39:05 2023 UTC (8 months, 3 weeks ago) by bertrand
Branches: MAIN
CVS tags: rpl-4_1_35, rpl-4_1_34, HEAD
Première mise à jour de lapack et blas.

    1: *> \brief \b DPSTF2 computes the Cholesky factorization with complete pivoting of a real symmetric positive semidefinite matrix.
    2: *
    3: *  =========== DOCUMENTATION ===========
    4: *
    5: * Online html documentation available at
    6: *            http://www.netlib.org/lapack/explore-html/
    7: *
    8: *> \htmlonly
    9: *> Download DPSTF2 + dependencies
   10: *> <a href="http://www.netlib.org/cgi-bin/netlibfiles.tgz?format=tgz&filename=/lapack/lapack_routine/dpstf2.f">
   11: *> [TGZ]</a>
   12: *> <a href="http://www.netlib.org/cgi-bin/netlibfiles.zip?format=zip&filename=/lapack/lapack_routine/dpstf2.f">
   13: *> [ZIP]</a>
   14: *> <a href="http://www.netlib.org/cgi-bin/netlibfiles.txt?format=txt&filename=/lapack/lapack_routine/dpstf2.f">
   15: *> [TXT]</a>
   16: *> \endhtmlonly
   17: *
   18: *  Definition:
   19: *  ===========
   20: *
   21: *       SUBROUTINE DPSTF2( UPLO, N, A, LDA, PIV, RANK, TOL, WORK, INFO )
   22: *
   23: *       .. Scalar Arguments ..
   24: *       DOUBLE PRECISION   TOL
   25: *       INTEGER            INFO, LDA, N, RANK
   26: *       CHARACTER          UPLO
   27: *       ..
   28: *       .. Array Arguments ..
   29: *       DOUBLE PRECISION   A( LDA, * ), WORK( 2*N )
   30: *       INTEGER            PIV( N )
   31: *       ..
   32: *
   33: *
   34: *> \par Purpose:
   35: *  =============
   36: *>
   37: *> \verbatim
   38: *>
   39: *> DPSTF2 computes the Cholesky factorization with complete
   40: *> pivoting of a real symmetric positive semidefinite matrix A.
   41: *>
   42: *> The factorization has the form
   43: *>    P**T * A * P = U**T * U ,  if UPLO = 'U',
   44: *>    P**T * A * P = L  * L**T,  if UPLO = 'L',
   45: *> where U is an upper triangular matrix and L is lower triangular, and
   46: *> P is stored as vector PIV.
   47: *>
   48: *> This algorithm does not attempt to check that A is positive
   49: *> semidefinite. This version of the algorithm calls level 2 BLAS.
   50: *> \endverbatim
   51: *
   52: *  Arguments:
   53: *  ==========
   54: *
   55: *> \param[in] UPLO
   56: *> \verbatim
   57: *>          UPLO is CHARACTER*1
   58: *>          Specifies whether the upper or lower triangular part of the
   59: *>          symmetric matrix A is stored.
   60: *>          = 'U':  Upper triangular
   61: *>          = 'L':  Lower triangular
   62: *> \endverbatim
   63: *>
   64: *> \param[in] N
   65: *> \verbatim
   66: *>          N is INTEGER
   67: *>          The order of the matrix A.  N >= 0.
   68: *> \endverbatim
   69: *>
   70: *> \param[in,out] A
   71: *> \verbatim
   72: *>          A is DOUBLE PRECISION array, dimension (LDA,N)
   73: *>          On entry, the symmetric matrix A.  If UPLO = 'U', the leading
   74: *>          n by n upper triangular part of A contains the upper
   75: *>          triangular part of the matrix A, and the strictly lower
   76: *>          triangular part of A is not referenced.  If UPLO = 'L', the
   77: *>          leading n by n lower triangular part of A contains the lower
   78: *>          triangular part of the matrix A, and the strictly upper
   79: *>          triangular part of A is not referenced.
   80: *>
   81: *>          On exit, if INFO = 0, the factor U or L from the Cholesky
   82: *>          factorization as above.
   83: *> \endverbatim
   84: *>
   85: *> \param[out] PIV
   86: *> \verbatim
   87: *>          PIV is INTEGER array, dimension (N)
   88: *>          PIV is such that the nonzero entries are P( PIV(K), K ) = 1.
   89: *> \endverbatim
   90: *>
   91: *> \param[out] RANK
   92: *> \verbatim
   93: *>          RANK is INTEGER
   94: *>          The rank of A given by the number of steps the algorithm
   95: *>          completed.
   96: *> \endverbatim
   97: *>
   98: *> \param[in] TOL
   99: *> \verbatim
  100: *>          TOL is DOUBLE PRECISION
  101: *>          User defined tolerance. If TOL < 0, then N*U*MAX( A( K,K ) )
  102: *>          will be used. The algorithm terminates at the (K-1)st step
  103: *>          if the pivot <= TOL.
  104: *> \endverbatim
  105: *>
  106: *> \param[in] LDA
  107: *> \verbatim
  108: *>          LDA is INTEGER
  109: *>          The leading dimension of the array A.  LDA >= max(1,N).
  110: *> \endverbatim
  111: *>
  112: *> \param[out] WORK
  113: *> \verbatim
  114: *>          WORK is DOUBLE PRECISION array, dimension (2*N)
  115: *>          Work space.
  116: *> \endverbatim
  117: *>
  118: *> \param[out] INFO
  119: *> \verbatim
  120: *>          INFO is INTEGER
  121: *>          < 0: If INFO = -K, the K-th argument had an illegal value,
  122: *>          = 0: algorithm completed successfully, and
  123: *>          > 0: the matrix A is either rank deficient with computed rank
  124: *>               as returned in RANK, or is not positive semidefinite. See
  125: *>               Section 7 of LAPACK Working Note #161 for further
  126: *>               information.
  127: *> \endverbatim
  128: *
  129: *  Authors:
  130: *  ========
  131: *
  132: *> \author Univ. of Tennessee
  133: *> \author Univ. of California Berkeley
  134: *> \author Univ. of Colorado Denver
  135: *> \author NAG Ltd.
  136: *
  137: *> \ingroup doubleOTHERcomputational
  138: *
  139: *  =====================================================================
  140:       SUBROUTINE DPSTF2( UPLO, N, A, LDA, PIV, RANK, TOL, WORK, INFO )
  141: *
  142: *  -- LAPACK computational routine --
  143: *  -- LAPACK is a software package provided by Univ. of Tennessee,    --
  144: *  -- Univ. of California Berkeley, Univ. of Colorado Denver and NAG Ltd..--
  145: *
  146: *     .. Scalar Arguments ..
  147:       DOUBLE PRECISION   TOL
  148:       INTEGER            INFO, LDA, N, RANK
  149:       CHARACTER          UPLO
  150: *     ..
  151: *     .. Array Arguments ..
  152:       DOUBLE PRECISION   A( LDA, * ), WORK( 2*N )
  153:       INTEGER            PIV( N )
  154: *     ..
  155: *
  156: *  =====================================================================
  157: *
  158: *     .. Parameters ..
  159:       DOUBLE PRECISION   ONE, ZERO
  160:       PARAMETER          ( ONE = 1.0D+0, ZERO = 0.0D+0 )
  161: *     ..
  162: *     .. Local Scalars ..
  163:       DOUBLE PRECISION   AJJ, DSTOP, DTEMP
  164:       INTEGER            I, ITEMP, J, PVT
  165:       LOGICAL            UPPER
  166: *     ..
  167: *     .. External Functions ..
  168:       DOUBLE PRECISION   DLAMCH
  169:       LOGICAL            LSAME, DISNAN
  170:       EXTERNAL           DLAMCH, LSAME, DISNAN
  171: *     ..
  172: *     .. External Subroutines ..
  173:       EXTERNAL           DGEMV, DSCAL, DSWAP, XERBLA
  174: *     ..
  175: *     .. Intrinsic Functions ..
  176:       INTRINSIC          MAX, SQRT, MAXLOC
  177: *     ..
  178: *     .. Executable Statements ..
  179: *
  180: *     Test the input parameters
  181: *
  182:       INFO = 0
  183:       UPPER = LSAME( UPLO, 'U' )
  184:       IF( .NOT.UPPER .AND. .NOT.LSAME( UPLO, 'L' ) ) THEN
  185:          INFO = -1
  186:       ELSE IF( N.LT.0 ) THEN
  187:          INFO = -2
  188:       ELSE IF( LDA.LT.MAX( 1, N ) ) THEN
  189:          INFO = -4
  190:       END IF
  191:       IF( INFO.NE.0 ) THEN
  192:          CALL XERBLA( 'DPSTF2', -INFO )
  193:          RETURN
  194:       END IF
  195: *
  196: *     Quick return if possible
  197: *
  198:       IF( N.EQ.0 )
  199:      $   RETURN
  200: *
  201: *     Initialize PIV
  202: *
  203:       DO 100 I = 1, N
  204:          PIV( I ) = I
  205:   100 CONTINUE
  206: *
  207: *     Compute stopping value
  208: *
  209:       PVT = 1
  210:       AJJ = A( PVT, PVT )
  211:       DO I = 2, N
  212:          IF( A( I, I ).GT.AJJ ) THEN
  213:             PVT = I
  214:             AJJ = A( PVT, PVT )
  215:          END IF
  216:       END DO
  217:       IF( AJJ.LE.ZERO.OR.DISNAN( AJJ ) ) THEN
  218:          RANK = 0
  219:          INFO = 1
  220:          GO TO 170
  221:       END IF
  222: *
  223: *     Compute stopping value if not supplied
  224: *
  225:       IF( TOL.LT.ZERO ) THEN
  226:          DSTOP = N * DLAMCH( 'Epsilon' ) * AJJ
  227:       ELSE
  228:          DSTOP = TOL
  229:       END IF
  230: *
  231: *     Set first half of WORK to zero, holds dot products
  232: *
  233:       DO 110 I = 1, N
  234:          WORK( I ) = 0
  235:   110 CONTINUE
  236: *
  237:       IF( UPPER ) THEN
  238: *
  239: *        Compute the Cholesky factorization P**T * A * P = U**T * U
  240: *
  241:          DO 130 J = 1, N
  242: *
  243: *        Find pivot, test for exit, else swap rows and columns
  244: *        Update dot products, compute possible pivots which are
  245: *        stored in the second half of WORK
  246: *
  247:             DO 120 I = J, N
  248: *
  249:                IF( J.GT.1 ) THEN
  250:                   WORK( I ) = WORK( I ) + A( J-1, I )**2
  251:                END IF
  252:                WORK( N+I ) = A( I, I ) - WORK( I )
  253: *
  254:   120       CONTINUE
  255: *
  256:             IF( J.GT.1 ) THEN
  257:                ITEMP = MAXLOC( WORK( (N+J):(2*N) ), 1 )
  258:                PVT = ITEMP + J - 1
  259:                AJJ = WORK( N+PVT )
  260:                IF( AJJ.LE.DSTOP.OR.DISNAN( AJJ ) ) THEN
  261:                   A( J, J ) = AJJ
  262:                   GO TO 160
  263:                END IF
  264:             END IF
  265: *
  266:             IF( J.NE.PVT ) THEN
  267: *
  268: *              Pivot OK, so can now swap pivot rows and columns
  269: *
  270:                A( PVT, PVT ) = A( J, J )
  271:                CALL DSWAP( J-1, A( 1, J ), 1, A( 1, PVT ), 1 )
  272:                IF( PVT.LT.N )
  273:      $            CALL DSWAP( N-PVT, A( J, PVT+1 ), LDA,
  274:      $                        A( PVT, PVT+1 ), LDA )
  275:                CALL DSWAP( PVT-J-1, A( J, J+1 ), LDA, A( J+1, PVT ), 1 )
  276: *
  277: *              Swap dot products and PIV
  278: *
  279:                DTEMP = WORK( J )
  280:                WORK( J ) = WORK( PVT )
  281:                WORK( PVT ) = DTEMP
  282:                ITEMP = PIV( PVT )
  283:                PIV( PVT ) = PIV( J )
  284:                PIV( J ) = ITEMP
  285:             END IF
  286: *
  287:             AJJ = SQRT( AJJ )
  288:             A( J, J ) = AJJ
  289: *
  290: *           Compute elements J+1:N of row J
  291: *
  292:             IF( J.LT.N ) THEN
  293:                CALL DGEMV( 'Trans', J-1, N-J, -ONE, A( 1, J+1 ), LDA,
  294:      $                     A( 1, J ), 1, ONE, A( J, J+1 ), LDA )
  295:                CALL DSCAL( N-J, ONE / AJJ, A( J, J+1 ), LDA )
  296:             END IF
  297: *
  298:   130    CONTINUE
  299: *
  300:       ELSE
  301: *
  302: *        Compute the Cholesky factorization P**T * A * P = L * L**T
  303: *
  304:          DO 150 J = 1, N
  305: *
  306: *        Find pivot, test for exit, else swap rows and columns
  307: *        Update dot products, compute possible pivots which are
  308: *        stored in the second half of WORK
  309: *
  310:             DO 140 I = J, N
  311: *
  312:                IF( J.GT.1 ) THEN
  313:                   WORK( I ) = WORK( I ) + A( I, J-1 )**2
  314:                END IF
  315:                WORK( N+I ) = A( I, I ) - WORK( I )
  316: *
  317:   140       CONTINUE
  318: *
  319:             IF( J.GT.1 ) THEN
  320:                ITEMP = MAXLOC( WORK( (N+J):(2*N) ), 1 )
  321:                PVT = ITEMP + J - 1
  322:                AJJ = WORK( N+PVT )
  323:                IF( AJJ.LE.DSTOP.OR.DISNAN( AJJ ) ) THEN
  324:                   A( J, J ) = AJJ
  325:                   GO TO 160
  326:                END IF
  327:             END IF
  328: *
  329:             IF( J.NE.PVT ) THEN
  330: *
  331: *              Pivot OK, so can now swap pivot rows and columns
  332: *
  333:                A( PVT, PVT ) = A( J, J )
  334:                CALL DSWAP( J-1, A( J, 1 ), LDA, A( PVT, 1 ), LDA )
  335:                IF( PVT.LT.N )
  336:      $            CALL DSWAP( N-PVT, A( PVT+1, J ), 1, A( PVT+1, PVT ),
  337:      $                        1 )
  338:                CALL DSWAP( PVT-J-1, A( J+1, J ), 1, A( PVT, J+1 ), LDA )
  339: *
  340: *              Swap dot products and PIV
  341: *
  342:                DTEMP = WORK( J )
  343:                WORK( J ) = WORK( PVT )
  344:                WORK( PVT ) = DTEMP
  345:                ITEMP = PIV( PVT )
  346:                PIV( PVT ) = PIV( J )
  347:                PIV( J ) = ITEMP
  348:             END IF
  349: *
  350:             AJJ = SQRT( AJJ )
  351:             A( J, J ) = AJJ
  352: *
  353: *           Compute elements J+1:N of column J
  354: *
  355:             IF( J.LT.N ) THEN
  356:                CALL DGEMV( 'No Trans', N-J, J-1, -ONE, A( J+1, 1 ), LDA,
  357:      $                     A( J, 1 ), LDA, ONE, A( J+1, J ), 1 )
  358:                CALL DSCAL( N-J, ONE / AJJ, A( J+1, J ), 1 )
  359:             END IF
  360: *
  361:   150    CONTINUE
  362: *
  363:       END IF
  364: *
  365: *     Ran to completion, A has full rank
  366: *
  367:       RANK = N
  368: *
  369:       GO TO 170
  370:   160 CONTINUE
  371: *
  372: *     Rank is number of steps completed.  Set INFO = 1 to signal
  373: *     that the factorization cannot be used to solve a system.
  374: *
  375:       RANK = J - 1
  376:       INFO = 1
  377: *
  378:   170 CONTINUE
  379:       RETURN
  380: *
  381: *     End of DPSTF2
  382: *
  383:       END

CVSweb interface <joel.bertrand@systella.fr>