File:  [local] / rpl / lapack / lapack / zlahef.f
Revision 1.8: download - view: text, annotated - select for diffs - revision graph
Fri Jul 22 07:38:17 2011 UTC (12 years, 9 months ago) by bertrand
Branches: MAIN
CVS tags: rpl-4_1_3, rpl-4_1_2, rpl-4_1_1, HEAD
En route vers la 4.4.1.

    1:       SUBROUTINE ZLAHEF( UPLO, N, NB, KB, A, LDA, IPIV, W, LDW, INFO )
    2: *
    3: *  -- LAPACK routine (version 3.3.1) --
    4: *  -- LAPACK is a software package provided by Univ. of Tennessee,    --
    5: *  -- Univ. of California Berkeley, Univ. of Colorado Denver and NAG Ltd..--
    6: *  -- April 2011                                                      --
    7: *
    8: *     .. Scalar Arguments ..
    9:       CHARACTER          UPLO
   10:       INTEGER            INFO, KB, LDA, LDW, N, NB
   11: *     ..
   12: *     .. Array Arguments ..
   13:       INTEGER            IPIV( * )
   14:       COMPLEX*16         A( LDA, * ), W( LDW, * )
   15: *     ..
   16: *
   17: *  Purpose
   18: *  =======
   19: *
   20: *  ZLAHEF computes a partial factorization of a complex Hermitian
   21: *  matrix A using the Bunch-Kaufman diagonal pivoting method. The
   22: *  partial factorization has the form:
   23: *
   24: *  A  =  ( I  U12 ) ( A11  0  ) (  I      0     )  if UPLO = 'U', or:
   25: *        ( 0  U22 ) (  0   D  ) ( U12**H U22**H )
   26: *
   27: *  A  =  ( L11  0 ) (  D   0  ) ( L11**H L21**H )  if UPLO = 'L'
   28: *        ( L21  I ) (  0  A22 ) (  0      I     )
   29: *
   30: *  where the order of D is at most NB. The actual order is returned in
   31: *  the argument KB, and is either NB or NB-1, or N if N <= NB.
   32: *  Note that U**H denotes the conjugate transpose of U.
   33: *
   34: *  ZLAHEF is an auxiliary routine called by ZHETRF. It uses blocked code
   35: *  (calling Level 3 BLAS) to update the submatrix A11 (if UPLO = 'U') or
   36: *  A22 (if UPLO = 'L').
   37: *
   38: *  Arguments
   39: *  =========
   40: *
   41: *  UPLO    (input) CHARACTER*1
   42: *          Specifies whether the upper or lower triangular part of the
   43: *          Hermitian matrix A is stored:
   44: *          = 'U':  Upper triangular
   45: *          = 'L':  Lower triangular
   46: *
   47: *  N       (input) INTEGER
   48: *          The order of the matrix A.  N >= 0.
   49: *
   50: *  NB      (input) INTEGER
   51: *          The maximum number of columns of the matrix A that should be
   52: *          factored.  NB should be at least 2 to allow for 2-by-2 pivot
   53: *          blocks.
   54: *
   55: *  KB      (output) INTEGER
   56: *          The number of columns of A that were actually factored.
   57: *          KB is either NB-1 or NB, or N if N <= NB.
   58: *
   59: *  A       (input/output) COMPLEX*16 array, dimension (LDA,N)
   60: *          On entry, the Hermitian matrix A.  If UPLO = 'U', the leading
   61: *          n-by-n upper triangular part of A contains the upper
   62: *          triangular part of the matrix A, and the strictly lower
   63: *          triangular part of A is not referenced.  If UPLO = 'L', the
   64: *          leading n-by-n lower triangular part of A contains the lower
   65: *          triangular part of the matrix A, and the strictly upper
   66: *          triangular part of A is not referenced.
   67: *          On exit, A contains details of the partial factorization.
   68: *
   69: *  LDA     (input) INTEGER
   70: *          The leading dimension of the array A.  LDA >= max(1,N).
   71: *
   72: *  IPIV    (output) INTEGER array, dimension (N)
   73: *          Details of the interchanges and the block structure of D.
   74: *          If UPLO = 'U', only the last KB elements of IPIV are set;
   75: *          if UPLO = 'L', only the first KB elements are set.
   76: *
   77: *          If IPIV(k) > 0, then rows and columns k and IPIV(k) were
   78: *          interchanged and D(k,k) is a 1-by-1 diagonal block.
   79: *          If UPLO = 'U' and IPIV(k) = IPIV(k-1) < 0, then rows and
   80: *          columns k-1 and -IPIV(k) were interchanged and D(k-1:k,k-1:k)
   81: *          is a 2-by-2 diagonal block.  If UPLO = 'L' and IPIV(k) =
   82: *          IPIV(k+1) < 0, then rows and columns k+1 and -IPIV(k) were
   83: *          interchanged and D(k:k+1,k:k+1) is a 2-by-2 diagonal block.
   84: *
   85: *  W       (workspace) COMPLEX*16 array, dimension (LDW,NB)
   86: *
   87: *  LDW     (input) INTEGER
   88: *          The leading dimension of the array W.  LDW >= max(1,N).
   89: *
   90: *  INFO    (output) INTEGER
   91: *          = 0: successful exit
   92: *          > 0: if INFO = k, D(k,k) is exactly zero.  The factorization
   93: *               has been completed, but the block diagonal matrix D is
   94: *               exactly singular.
   95: *
   96: *  =====================================================================
   97: *
   98: *     .. Parameters ..
   99:       DOUBLE PRECISION   ZERO, ONE
  100:       PARAMETER          ( ZERO = 0.0D+0, ONE = 1.0D+0 )
  101:       COMPLEX*16         CONE
  102:       PARAMETER          ( CONE = ( 1.0D+0, 0.0D+0 ) )
  103:       DOUBLE PRECISION   EIGHT, SEVTEN
  104:       PARAMETER          ( EIGHT = 8.0D+0, SEVTEN = 17.0D+0 )
  105: *     ..
  106: *     .. Local Scalars ..
  107:       INTEGER            IMAX, J, JB, JJ, JMAX, JP, K, KK, KKW, KP,
  108:      $                   KSTEP, KW
  109:       DOUBLE PRECISION   ABSAKK, ALPHA, COLMAX, R1, ROWMAX, T
  110:       COMPLEX*16         D11, D21, D22, Z
  111: *     ..
  112: *     .. External Functions ..
  113:       LOGICAL            LSAME
  114:       INTEGER            IZAMAX
  115:       EXTERNAL           LSAME, IZAMAX
  116: *     ..
  117: *     .. External Subroutines ..
  118:       EXTERNAL           ZCOPY, ZDSCAL, ZGEMM, ZGEMV, ZLACGV, ZSWAP
  119: *     ..
  120: *     .. Intrinsic Functions ..
  121:       INTRINSIC          ABS, DBLE, DCONJG, DIMAG, MAX, MIN, SQRT
  122: *     ..
  123: *     .. Statement Functions ..
  124:       DOUBLE PRECISION   CABS1
  125: *     ..
  126: *     .. Statement Function definitions ..
  127:       CABS1( Z ) = ABS( DBLE( Z ) ) + ABS( DIMAG( Z ) )
  128: *     ..
  129: *     .. Executable Statements ..
  130: *
  131:       INFO = 0
  132: *
  133: *     Initialize ALPHA for use in choosing pivot block size.
  134: *
  135:       ALPHA = ( ONE+SQRT( SEVTEN ) ) / EIGHT
  136: *
  137:       IF( LSAME( UPLO, 'U' ) ) THEN
  138: *
  139: *        Factorize the trailing columns of A using the upper triangle
  140: *        of A and working backwards, and compute the matrix W = U12*D
  141: *        for use in updating A11 (note that conjg(W) is actually stored)
  142: *
  143: *        K is the main loop index, decreasing from N in steps of 1 or 2
  144: *
  145: *        KW is the column of W which corresponds to column K of A
  146: *
  147:          K = N
  148:    10    CONTINUE
  149:          KW = NB + K - N
  150: *
  151: *        Exit from loop
  152: *
  153:          IF( ( K.LE.N-NB+1 .AND. NB.LT.N ) .OR. K.LT.1 )
  154:      $      GO TO 30
  155: *
  156: *        Copy column K of A to column KW of W and update it
  157: *
  158:          CALL ZCOPY( K-1, A( 1, K ), 1, W( 1, KW ), 1 )
  159:          W( K, KW ) = DBLE( A( K, K ) )
  160:          IF( K.LT.N ) THEN
  161:             CALL ZGEMV( 'No transpose', K, N-K, -CONE, A( 1, K+1 ), LDA,
  162:      $                  W( K, KW+1 ), LDW, CONE, W( 1, KW ), 1 )
  163:             W( K, KW ) = DBLE( W( K, KW ) )
  164:          END IF
  165: *
  166:          KSTEP = 1
  167: *
  168: *        Determine rows and columns to be interchanged and whether
  169: *        a 1-by-1 or 2-by-2 pivot block will be used
  170: *
  171:          ABSAKK = ABS( DBLE( W( K, KW ) ) )
  172: *
  173: *        IMAX is the row-index of the largest off-diagonal element in
  174: *        column K, and COLMAX is its absolute value
  175: *
  176:          IF( K.GT.1 ) THEN
  177:             IMAX = IZAMAX( K-1, W( 1, KW ), 1 )
  178:             COLMAX = CABS1( W( IMAX, KW ) )
  179:          ELSE
  180:             COLMAX = ZERO
  181:          END IF
  182: *
  183:          IF( MAX( ABSAKK, COLMAX ).EQ.ZERO ) THEN
  184: *
  185: *           Column K is zero: set INFO and continue
  186: *
  187:             IF( INFO.EQ.0 )
  188:      $         INFO = K
  189:             KP = K
  190:             A( K, K ) = DBLE( A( K, K ) )
  191:          ELSE
  192:             IF( ABSAKK.GE.ALPHA*COLMAX ) THEN
  193: *
  194: *              no interchange, use 1-by-1 pivot block
  195: *
  196:                KP = K
  197:             ELSE
  198: *
  199: *              Copy column IMAX to column KW-1 of W and update it
  200: *
  201:                CALL ZCOPY( IMAX-1, A( 1, IMAX ), 1, W( 1, KW-1 ), 1 )
  202:                W( IMAX, KW-1 ) = DBLE( A( IMAX, IMAX ) )
  203:                CALL ZCOPY( K-IMAX, A( IMAX, IMAX+1 ), LDA,
  204:      $                     W( IMAX+1, KW-1 ), 1 )
  205:                CALL ZLACGV( K-IMAX, W( IMAX+1, KW-1 ), 1 )
  206:                IF( K.LT.N ) THEN
  207:                   CALL ZGEMV( 'No transpose', K, N-K, -CONE,
  208:      $                        A( 1, K+1 ), LDA, W( IMAX, KW+1 ), LDW,
  209:      $                        CONE, W( 1, KW-1 ), 1 )
  210:                   W( IMAX, KW-1 ) = DBLE( W( IMAX, KW-1 ) )
  211:                END IF
  212: *
  213: *              JMAX is the column-index of the largest off-diagonal
  214: *              element in row IMAX, and ROWMAX is its absolute value
  215: *
  216:                JMAX = IMAX + IZAMAX( K-IMAX, W( IMAX+1, KW-1 ), 1 )
  217:                ROWMAX = CABS1( W( JMAX, KW-1 ) )
  218:                IF( IMAX.GT.1 ) THEN
  219:                   JMAX = IZAMAX( IMAX-1, W( 1, KW-1 ), 1 )
  220:                   ROWMAX = MAX( ROWMAX, CABS1( W( JMAX, KW-1 ) ) )
  221:                END IF
  222: *
  223:                IF( ABSAKK.GE.ALPHA*COLMAX*( COLMAX / ROWMAX ) ) THEN
  224: *
  225: *                 no interchange, use 1-by-1 pivot block
  226: *
  227:                   KP = K
  228:                ELSE IF( ABS( DBLE( W( IMAX, KW-1 ) ) ).GE.ALPHA*ROWMAX )
  229:      $                   THEN
  230: *
  231: *                 interchange rows and columns K and IMAX, use 1-by-1
  232: *                 pivot block
  233: *
  234:                   KP = IMAX
  235: *
  236: *                 copy column KW-1 of W to column KW
  237: *
  238:                   CALL ZCOPY( K, W( 1, KW-1 ), 1, W( 1, KW ), 1 )
  239:                ELSE
  240: *
  241: *                 interchange rows and columns K-1 and IMAX, use 2-by-2
  242: *                 pivot block
  243: *
  244:                   KP = IMAX
  245:                   KSTEP = 2
  246:                END IF
  247:             END IF
  248: *
  249:             KK = K - KSTEP + 1
  250:             KKW = NB + KK - N
  251: *
  252: *           Updated column KP is already stored in column KKW of W
  253: *
  254:             IF( KP.NE.KK ) THEN
  255: *
  256: *              Copy non-updated column KK to column KP
  257: *
  258:                A( KP, KP ) = DBLE( A( KK, KK ) )
  259:                CALL ZCOPY( KK-1-KP, A( KP+1, KK ), 1, A( KP, KP+1 ),
  260:      $                     LDA )
  261:                CALL ZLACGV( KK-1-KP, A( KP, KP+1 ), LDA )
  262:                CALL ZCOPY( KP-1, A( 1, KK ), 1, A( 1, KP ), 1 )
  263: *
  264: *              Interchange rows KK and KP in last KK columns of A and W
  265: *
  266:                IF( KK.LT.N )
  267:      $            CALL ZSWAP( N-KK, A( KK, KK+1 ), LDA, A( KP, KK+1 ),
  268:      $                        LDA )
  269:                CALL ZSWAP( N-KK+1, W( KK, KKW ), LDW, W( KP, KKW ),
  270:      $                     LDW )
  271:             END IF
  272: *
  273:             IF( KSTEP.EQ.1 ) THEN
  274: *
  275: *              1-by-1 pivot block D(k): column KW of W now holds
  276: *
  277: *              W(k) = U(k)*D(k)
  278: *
  279: *              where U(k) is the k-th column of U
  280: *
  281: *              Store U(k) in column k of A
  282: *
  283:                CALL ZCOPY( K, W( 1, KW ), 1, A( 1, K ), 1 )
  284:                R1 = ONE / DBLE( A( K, K ) )
  285:                CALL ZDSCAL( K-1, R1, A( 1, K ), 1 )
  286: *
  287: *              Conjugate W(k)
  288: *
  289:                CALL ZLACGV( K-1, W( 1, KW ), 1 )
  290:             ELSE
  291: *
  292: *              2-by-2 pivot block D(k): columns KW and KW-1 of W now
  293: *              hold
  294: *
  295: *              ( W(k-1) W(k) ) = ( U(k-1) U(k) )*D(k)
  296: *
  297: *              where U(k) and U(k-1) are the k-th and (k-1)-th columns
  298: *              of U
  299: *
  300:                IF( K.GT.2 ) THEN
  301: *
  302: *                 Store U(k) and U(k-1) in columns k and k-1 of A
  303: *
  304:                   D21 = W( K-1, KW )
  305:                   D11 = W( K, KW ) / DCONJG( D21 )
  306:                   D22 = W( K-1, KW-1 ) / D21
  307:                   T = ONE / ( DBLE( D11*D22 )-ONE )
  308:                   D21 = T / D21
  309:                   DO 20 J = 1, K - 2
  310:                      A( J, K-1 ) = D21*( D11*W( J, KW-1 )-W( J, KW ) )
  311:                      A( J, K ) = DCONJG( D21 )*
  312:      $                           ( D22*W( J, KW )-W( J, KW-1 ) )
  313:    20             CONTINUE
  314:                END IF
  315: *
  316: *              Copy D(k) to A
  317: *
  318:                A( K-1, K-1 ) = W( K-1, KW-1 )
  319:                A( K-1, K ) = W( K-1, KW )
  320:                A( K, K ) = W( K, KW )
  321: *
  322: *              Conjugate W(k) and W(k-1)
  323: *
  324:                CALL ZLACGV( K-1, W( 1, KW ), 1 )
  325:                CALL ZLACGV( K-2, W( 1, KW-1 ), 1 )
  326:             END IF
  327:          END IF
  328: *
  329: *        Store details of the interchanges in IPIV
  330: *
  331:          IF( KSTEP.EQ.1 ) THEN
  332:             IPIV( K ) = KP
  333:          ELSE
  334:             IPIV( K ) = -KP
  335:             IPIV( K-1 ) = -KP
  336:          END IF
  337: *
  338: *        Decrease K and return to the start of the main loop
  339: *
  340:          K = K - KSTEP
  341:          GO TO 10
  342: *
  343:    30    CONTINUE
  344: *
  345: *        Update the upper triangle of A11 (= A(1:k,1:k)) as
  346: *
  347: *        A11 := A11 - U12*D*U12**H = A11 - U12*W**H
  348: *
  349: *        computing blocks of NB columns at a time (note that conjg(W) is
  350: *        actually stored)
  351: *
  352:          DO 50 J = ( ( K-1 ) / NB )*NB + 1, 1, -NB
  353:             JB = MIN( NB, K-J+1 )
  354: *
  355: *           Update the upper triangle of the diagonal block
  356: *
  357:             DO 40 JJ = J, J + JB - 1
  358:                A( JJ, JJ ) = DBLE( A( JJ, JJ ) )
  359:                CALL ZGEMV( 'No transpose', JJ-J+1, N-K, -CONE,
  360:      $                     A( J, K+1 ), LDA, W( JJ, KW+1 ), LDW, CONE,
  361:      $                     A( J, JJ ), 1 )
  362:                A( JJ, JJ ) = DBLE( A( JJ, JJ ) )
  363:    40       CONTINUE
  364: *
  365: *           Update the rectangular superdiagonal block
  366: *
  367:             CALL ZGEMM( 'No transpose', 'Transpose', J-1, JB, N-K,
  368:      $                  -CONE, A( 1, K+1 ), LDA, W( J, KW+1 ), LDW,
  369:      $                  CONE, A( 1, J ), LDA )
  370:    50    CONTINUE
  371: *
  372: *        Put U12 in standard form by partially undoing the interchanges
  373: *        in columns k+1:n
  374: *
  375:          J = K + 1
  376:    60    CONTINUE
  377:          JJ = J
  378:          JP = IPIV( J )
  379:          IF( JP.LT.0 ) THEN
  380:             JP = -JP
  381:             J = J + 1
  382:          END IF
  383:          J = J + 1
  384:          IF( JP.NE.JJ .AND. J.LE.N )
  385:      $      CALL ZSWAP( N-J+1, A( JP, J ), LDA, A( JJ, J ), LDA )
  386:          IF( J.LE.N )
  387:      $      GO TO 60
  388: *
  389: *        Set KB to the number of columns factorized
  390: *
  391:          KB = N - K
  392: *
  393:       ELSE
  394: *
  395: *        Factorize the leading columns of A using the lower triangle
  396: *        of A and working forwards, and compute the matrix W = L21*D
  397: *        for use in updating A22 (note that conjg(W) is actually stored)
  398: *
  399: *        K is the main loop index, increasing from 1 in steps of 1 or 2
  400: *
  401:          K = 1
  402:    70    CONTINUE
  403: *
  404: *        Exit from loop
  405: *
  406:          IF( ( K.GE.NB .AND. NB.LT.N ) .OR. K.GT.N )
  407:      $      GO TO 90
  408: *
  409: *        Copy column K of A to column K of W and update it
  410: *
  411:          W( K, K ) = DBLE( A( K, K ) )
  412:          IF( K.LT.N )
  413:      $      CALL ZCOPY( N-K, A( K+1, K ), 1, W( K+1, K ), 1 )
  414:          CALL ZGEMV( 'No transpose', N-K+1, K-1, -CONE, A( K, 1 ), LDA,
  415:      $               W( K, 1 ), LDW, CONE, W( K, K ), 1 )
  416:          W( K, K ) = DBLE( W( K, K ) )
  417: *
  418:          KSTEP = 1
  419: *
  420: *        Determine rows and columns to be interchanged and whether
  421: *        a 1-by-1 or 2-by-2 pivot block will be used
  422: *
  423:          ABSAKK = ABS( DBLE( W( K, K ) ) )
  424: *
  425: *        IMAX is the row-index of the largest off-diagonal element in
  426: *        column K, and COLMAX is its absolute value
  427: *
  428:          IF( K.LT.N ) THEN
  429:             IMAX = K + IZAMAX( N-K, W( K+1, K ), 1 )
  430:             COLMAX = CABS1( W( IMAX, K ) )
  431:          ELSE
  432:             COLMAX = ZERO
  433:          END IF
  434: *
  435:          IF( MAX( ABSAKK, COLMAX ).EQ.ZERO ) THEN
  436: *
  437: *           Column K is zero: set INFO and continue
  438: *
  439:             IF( INFO.EQ.0 )
  440:      $         INFO = K
  441:             KP = K
  442:             A( K, K ) = DBLE( A( K, K ) )
  443:          ELSE
  444:             IF( ABSAKK.GE.ALPHA*COLMAX ) THEN
  445: *
  446: *              no interchange, use 1-by-1 pivot block
  447: *
  448:                KP = K
  449:             ELSE
  450: *
  451: *              Copy column IMAX to column K+1 of W and update it
  452: *
  453:                CALL ZCOPY( IMAX-K, A( IMAX, K ), LDA, W( K, K+1 ), 1 )
  454:                CALL ZLACGV( IMAX-K, W( K, K+1 ), 1 )
  455:                W( IMAX, K+1 ) = DBLE( A( IMAX, IMAX ) )
  456:                IF( IMAX.LT.N )
  457:      $            CALL ZCOPY( N-IMAX, A( IMAX+1, IMAX ), 1,
  458:      $                        W( IMAX+1, K+1 ), 1 )
  459:                CALL ZGEMV( 'No transpose', N-K+1, K-1, -CONE, A( K, 1 ),
  460:      $                     LDA, W( IMAX, 1 ), LDW, CONE, W( K, K+1 ),
  461:      $                     1 )
  462:                W( IMAX, K+1 ) = DBLE( W( IMAX, K+1 ) )
  463: *
  464: *              JMAX is the column-index of the largest off-diagonal
  465: *              element in row IMAX, and ROWMAX is its absolute value
  466: *
  467:                JMAX = K - 1 + IZAMAX( IMAX-K, W( K, K+1 ), 1 )
  468:                ROWMAX = CABS1( W( JMAX, K+1 ) )
  469:                IF( IMAX.LT.N ) THEN
  470:                   JMAX = IMAX + IZAMAX( N-IMAX, W( IMAX+1, K+1 ), 1 )
  471:                   ROWMAX = MAX( ROWMAX, CABS1( W( JMAX, K+1 ) ) )
  472:                END IF
  473: *
  474:                IF( ABSAKK.GE.ALPHA*COLMAX*( COLMAX / ROWMAX ) ) THEN
  475: *
  476: *                 no interchange, use 1-by-1 pivot block
  477: *
  478:                   KP = K
  479:                ELSE IF( ABS( DBLE( W( IMAX, K+1 ) ) ).GE.ALPHA*ROWMAX )
  480:      $                   THEN
  481: *
  482: *                 interchange rows and columns K and IMAX, use 1-by-1
  483: *                 pivot block
  484: *
  485:                   KP = IMAX
  486: *
  487: *                 copy column K+1 of W to column K
  488: *
  489:                   CALL ZCOPY( N-K+1, W( K, K+1 ), 1, W( K, K ), 1 )
  490:                ELSE
  491: *
  492: *                 interchange rows and columns K+1 and IMAX, use 2-by-2
  493: *                 pivot block
  494: *
  495:                   KP = IMAX
  496:                   KSTEP = 2
  497:                END IF
  498:             END IF
  499: *
  500:             KK = K + KSTEP - 1
  501: *
  502: *           Updated column KP is already stored in column KK of W
  503: *
  504:             IF( KP.NE.KK ) THEN
  505: *
  506: *              Copy non-updated column KK to column KP
  507: *
  508:                A( KP, KP ) = DBLE( A( KK, KK ) )
  509:                CALL ZCOPY( KP-KK-1, A( KK+1, KK ), 1, A( KP, KK+1 ),
  510:      $                     LDA )
  511:                CALL ZLACGV( KP-KK-1, A( KP, KK+1 ), LDA )
  512:                IF( KP.LT.N )
  513:      $            CALL ZCOPY( N-KP, A( KP+1, KK ), 1, A( KP+1, KP ), 1 )
  514: *
  515: *              Interchange rows KK and KP in first KK columns of A and W
  516: *
  517:                CALL ZSWAP( KK-1, A( KK, 1 ), LDA, A( KP, 1 ), LDA )
  518:                CALL ZSWAP( KK, W( KK, 1 ), LDW, W( KP, 1 ), LDW )
  519:             END IF
  520: *
  521:             IF( KSTEP.EQ.1 ) THEN
  522: *
  523: *              1-by-1 pivot block D(k): column k of W now holds
  524: *
  525: *              W(k) = L(k)*D(k)
  526: *
  527: *              where L(k) is the k-th column of L
  528: *
  529: *              Store L(k) in column k of A
  530: *
  531:                CALL ZCOPY( N-K+1, W( K, K ), 1, A( K, K ), 1 )
  532:                IF( K.LT.N ) THEN
  533:                   R1 = ONE / DBLE( A( K, K ) )
  534:                   CALL ZDSCAL( N-K, R1, A( K+1, K ), 1 )
  535: *
  536: *                 Conjugate W(k)
  537: *
  538:                   CALL ZLACGV( N-K, W( K+1, K ), 1 )
  539:                END IF
  540:             ELSE
  541: *
  542: *              2-by-2 pivot block D(k): columns k and k+1 of W now hold
  543: *
  544: *              ( W(k) W(k+1) ) = ( L(k) L(k+1) )*D(k)
  545: *
  546: *              where L(k) and L(k+1) are the k-th and (k+1)-th columns
  547: *              of L
  548: *
  549:                IF( K.LT.N-1 ) THEN
  550: *
  551: *                 Store L(k) and L(k+1) in columns k and k+1 of A
  552: *
  553:                   D21 = W( K+1, K )
  554:                   D11 = W( K+1, K+1 ) / D21
  555:                   D22 = W( K, K ) / DCONJG( D21 )
  556:                   T = ONE / ( DBLE( D11*D22 )-ONE )
  557:                   D21 = T / D21
  558:                   DO 80 J = K + 2, N
  559:                      A( J, K ) = DCONJG( D21 )*
  560:      $                           ( D11*W( J, K )-W( J, K+1 ) )
  561:                      A( J, K+1 ) = D21*( D22*W( J, K+1 )-W( J, K ) )
  562:    80             CONTINUE
  563:                END IF
  564: *
  565: *              Copy D(k) to A
  566: *
  567:                A( K, K ) = W( K, K )
  568:                A( K+1, K ) = W( K+1, K )
  569:                A( K+1, K+1 ) = W( K+1, K+1 )
  570: *
  571: *              Conjugate W(k) and W(k+1)
  572: *
  573:                CALL ZLACGV( N-K, W( K+1, K ), 1 )
  574:                CALL ZLACGV( N-K-1, W( K+2, K+1 ), 1 )
  575:             END IF
  576:          END IF
  577: *
  578: *        Store details of the interchanges in IPIV
  579: *
  580:          IF( KSTEP.EQ.1 ) THEN
  581:             IPIV( K ) = KP
  582:          ELSE
  583:             IPIV( K ) = -KP
  584:             IPIV( K+1 ) = -KP
  585:          END IF
  586: *
  587: *        Increase K and return to the start of the main loop
  588: *
  589:          K = K + KSTEP
  590:          GO TO 70
  591: *
  592:    90    CONTINUE
  593: *
  594: *        Update the lower triangle of A22 (= A(k:n,k:n)) as
  595: *
  596: *        A22 := A22 - L21*D*L21**H = A22 - L21*W**H
  597: *
  598: *        computing blocks of NB columns at a time (note that conjg(W) is
  599: *        actually stored)
  600: *
  601:          DO 110 J = K, N, NB
  602:             JB = MIN( NB, N-J+1 )
  603: *
  604: *           Update the lower triangle of the diagonal block
  605: *
  606:             DO 100 JJ = J, J + JB - 1
  607:                A( JJ, JJ ) = DBLE( A( JJ, JJ ) )
  608:                CALL ZGEMV( 'No transpose', J+JB-JJ, K-1, -CONE,
  609:      $                     A( JJ, 1 ), LDA, W( JJ, 1 ), LDW, CONE,
  610:      $                     A( JJ, JJ ), 1 )
  611:                A( JJ, JJ ) = DBLE( A( JJ, JJ ) )
  612:   100       CONTINUE
  613: *
  614: *           Update the rectangular subdiagonal block
  615: *
  616:             IF( J+JB.LE.N )
  617:      $         CALL ZGEMM( 'No transpose', 'Transpose', N-J-JB+1, JB,
  618:      $                     K-1, -CONE, A( J+JB, 1 ), LDA, W( J, 1 ),
  619:      $                     LDW, CONE, A( J+JB, J ), LDA )
  620:   110    CONTINUE
  621: *
  622: *        Put L21 in standard form by partially undoing the interchanges
  623: *        in columns 1:k-1
  624: *
  625:          J = K - 1
  626:   120    CONTINUE
  627:          JJ = J
  628:          JP = IPIV( J )
  629:          IF( JP.LT.0 ) THEN
  630:             JP = -JP
  631:             J = J - 1
  632:          END IF
  633:          J = J - 1
  634:          IF( JP.NE.JJ .AND. J.GE.1 )
  635:      $      CALL ZSWAP( J, A( JP, 1 ), LDA, A( JJ, 1 ), LDA )
  636:          IF( J.GE.1 )
  637:      $      GO TO 120
  638: *
  639: *        Set KB to the number of columns factorized
  640: *
  641:          KB = K - 1
  642: *
  643:       END IF
  644:       RETURN
  645: *
  646: *     End of ZLAHEF
  647: *
  648:       END

CVSweb interface <joel.bertrand@systella.fr>