pdlarzb.f

Go to the documentation of this file.

      SUBROUTINE pdlarzb( SIDE, TRANS, DIRECT, STOREV, M, N, K, L, V,
     $                    IV, JV, DESCV, T, C, IC, JC, DESCC, WORK )
*
*  -- ScaLAPACK auxiliary routine (version 2.0.2) --
*     Univ. of Tennessee, Univ. of California Berkeley, Univ. of Colorado Denver
*     May 1 2012
*
*     .. Scalar Arguments ..
      CHARACTER          DIRECT, SIDE, STOREV, TRANS
      INTEGER            IC, IV, JC, JV, K, L, M, N
*     ..
*     .. Array Arguments ..
      INTEGER            DESCC( * ), DESCV( * )
      DOUBLE PRECISION   C( * ), T( * ), V( * ), WORK( * )
*     ..
*
*  Purpose
*  =======
*
*  PDLARZB applies a real block reflector Q or its transpose Q**T to
*  a real distributed M-by-N matrix sub( C ) = C(IC:IC+M-1,JC:JC+N-1)
*  from the left or the right.
*
*  Q is a product of k elementary reflectors as returned by PDTZRZF.
*
*  Currently, only STOREV = 'R' and DIRECT = 'B' are supported.
*
*  Notes
*  =====
*
*  Each global data object is described by an associated description
*  vector.  This vector stores the information required to establish
*  the mapping between an object element and its corresponding process
*  and memory location.
*
*  Let A be a generic term for any 2D block cyclicly distributed array.
*  Such a global array has an associated description vector DESCA.
*  In the following comments, the character _ should be read as
*  "of the global array".
*
*  NOTATION        STORED IN      EXPLANATION
*  --------------- -------------- --------------------------------------
*  DTYPE_A(global) DESCA( DTYPE_ )The descriptor type.  In this case,
*                                 DTYPE_A = 1.
*  CTXT_A (global) DESCA( CTXT_ ) The BLACS context handle, indicating
*                                 the BLACS process grid A is distribu-
*                                 ted over. The context itself is glo-
*                                 bal, but the handle (the integer
*                                 value) may vary.
*  M_A    (global) DESCA( M_ )    The number of rows in the global
*                                 array A.
*  N_A    (global) DESCA( N_ )    The number of columns in the global
*                                 array A.
*  MB_A   (global) DESCA( MB_ )   The blocking factor used to distribute
*                                 the rows of the array.
*  NB_A   (global) DESCA( NB_ )   The blocking factor used to distribute
*                                 the columns of the array.
*  RSRC_A (global) DESCA( RSRC_ ) The process row over which the first
*                                 row of the array A is distributed.
*  CSRC_A (global) DESCA( CSRC_ ) The process column over which the
*                                 first column of the array A is
*                                 distributed.
*  LLD_A  (local)  DESCA( LLD_ )  The leading dimension of the local
*                                 array.  LLD_A >= MAX(1,LOCr(M_A)).
*
*  Let K be the number of rows or columns of a distributed matrix,
*  and assume that its process grid has dimension p x q.
*  LOCr( K ) denotes the number of elements of K that a process
*  would receive if K were distributed over the p processes of its
*  process column.
*  Similarly, LOCc( K ) denotes the number of elements of K that a
*  process would receive if K were distributed over the q processes of
*  its process row.
*  The values of LOCr() and LOCc() may be determined via a call to the
*  ScaLAPACK tool function, NUMROC:
*          LOCr( M ) = NUMROC( M, MB_A, MYROW, RSRC_A, NPROW ),
*          LOCc( N ) = NUMROC( N, NB_A, MYCOL, CSRC_A, NPCOL ).
*  An upper bound for these quantities may be computed by:
*          LOCr( M ) <= ceil( ceil(M/MB_A)/NPROW )*MB_A
*          LOCc( N ) <= ceil( ceil(N/NB_A)/NPCOL )*NB_A
*
*  Arguments
*  =========
*
*  SIDE    (global input) CHARACTER
*          = 'L': apply Q or Q**T from the Left;
*          = 'R': apply Q or Q**T from the Right.
*
*  TRANS   (global input) CHARACTER
*          = 'N':  No transpose, apply Q;
*          = 'T':  Transpose, apply Q**T.
*
*  DIRECT  (global input) CHARACTER
*          Indicates how H is formed from a product of elementary
*          reflectors
*          = 'F': H = H(1) H(2) . . . H(k) (Forward, not supported yet)
*          = 'B': H = H(k) . . . H(2) H(1) (Backward)
*
*  STOREV  (global input) CHARACTER
*          Indicates how the vectors which define the elementary
*          reflectors are stored:
*          = 'C': Columnwise                        (not supported yet)
*          = 'R': Rowwise
*
*  M       (global input) INTEGER
*          The number of rows to be operated on i.e the number of rows
*          of the distributed submatrix sub( C ). M >= 0.
*
*  N       (global input) INTEGER
*          The number of columns to be operated on i.e the number of
*          columns of the distributed submatrix sub( C ). N >= 0.
*
*  K       (global input) INTEGER
*          The order of the matrix T (= the number of elementary
*          reflectors whose product defines the block reflector).
*
*  L       (global input) INTEGER
*          The columns of the distributed submatrix sub( A ) containing
*          the meaningful part of the Householder reflectors.
*          If SIDE = 'L', M >= L >= 0, if SIDE = 'R', N >= L >= 0.
*
*  V       (local input) DOUBLE PRECISION pointer into the local memory
*          to an array of dimension (LLD_V, LOCc(JV+M-1)) if SIDE = 'L',
*          (LLD_V, LOCc(JV+N-1)) if SIDE = 'R'. It contains the local
*          pieces of the distributed vectors V representing the
*          Householder transformation as returned by PDTZRZF.
*          LLD_V >= LOCr(IV+K-1).
*
*  IV      (global input) INTEGER
*          The row index in the global array V indicating the first
*          row of sub( V ).
*
*  JV      (global input) INTEGER
*          The column index in the global array V indicating the
*          first column of sub( V ).
*
*  DESCV   (global and local input) INTEGER array of dimension DLEN_.
*          The array descriptor for the distributed matrix V.
*
*  T       (local input) DOUBLE PRECISION array, dimension MB_V by MB_V
*          The lower triangular matrix T in the representation of the
*          block reflector.
*
*  C       (local input/local output) DOUBLE PRECISION pointer into the
*          local memory to an array of dimension (LLD_C,LOCc(JC+N-1)).
*          On entry, the M-by-N distributed matrix sub( C ). On exit,
*          sub( C ) is overwritten by Q*sub( C ) or Q'*sub( C ) or
*          sub( C )*Q or sub( C )*Q'.
*
*  IC      (global input) INTEGER
*          The row index in the global array C indicating the first
*          row of sub( C ).
*
*  JC      (global input) INTEGER
*          The column index in the global array C indicating the
*          first column of sub( C ).
*
*  DESCC   (global and local input) INTEGER array of dimension DLEN_.
*          The array descriptor for the distributed matrix C.
*
*  WORK    (local workspace) DOUBLE PRECISION array, dimension (LWORK)
*          If STOREV = 'C',
*            if SIDE = 'L',
*              LWORK >= ( NqC0 + MpC0 ) * K
*            else if SIDE = 'R',
*              LWORK >= ( NqC0 + MAX( NpV0 + NUMROC( NUMROC( N+ICOFFC,
*                         NB_V, 0, 0, NPCOL ), NB_V, 0, 0, LCMQ ),
*                         MpC0 ) ) * K
*            end if
*          else if STOREV = 'R',
*            if SIDE = 'L',
*              LWORK >= ( MpC0 + MAX( MqV0 + NUMROC( NUMROC( M+IROFFC,
*                         MB_V, 0, 0, NPROW ), MB_V, 0, 0, LCMP ),
*                         NqC0 ) ) * K
*            else if SIDE = 'R',
*              LWORK >= ( MpC0 + NqC0 ) * K
*            end if
*          end if
*
*          where LCMQ = LCM / NPCOL with LCM = ICLM( NPROW, NPCOL ),
*
*          IROFFV = MOD( IV-1, MB_V ), ICOFFV = MOD( JV-1, NB_V ),
*          IVROW = INDXG2P( IV, MB_V, MYROW, RSRC_V, NPROW ),
*          IVCOL = INDXG2P( JV, NB_V, MYCOL, CSRC_V, NPCOL ),
*          MqV0 = NUMROC( M+ICOFFV, NB_V, MYCOL, IVCOL, NPCOL ),
*          NpV0 = NUMROC( N+IROFFV, MB_V, MYROW, IVROW, NPROW ),
*
*          IROFFC = MOD( IC-1, MB_C ), ICOFFC = MOD( JC-1, NB_C ),
*          ICROW = INDXG2P( IC, MB_C, MYROW, RSRC_C, NPROW ),
*          ICCOL = INDXG2P( JC, NB_C, MYCOL, CSRC_C, NPCOL ),
*          MpC0 = NUMROC( M+IROFFC, MB_C, MYROW, ICROW, NPROW ),
*          NpC0 = NUMROC( N+ICOFFC, MB_C, MYROW, ICROW, NPROW ),
*          NqC0 = NUMROC( N+ICOFFC, NB_C, MYCOL, ICCOL, NPCOL ),
*
*          ILCM, INDXG2P and NUMROC are ScaLAPACK tool functions;
*          MYROW, MYCOL, NPROW and NPCOL can be determined by calling
*          the subroutine BLACS_GRIDINFO.
*
*  Alignment requirements
*  ======================
*
*  The distributed submatrices V(IV:*, JV:*) and C(IC:IC+M-1,JC:JC+N-1)
*  must verify some alignment properties, namely the following
*  expressions should be true:
*
*  If STOREV = 'Columnwise'
*    If SIDE = 'Left',
*      ( MB_V.EQ.MB_C .AND. IROFFV.EQ.IROFFC .AND. IVROW.EQ.ICROW )
*    If SIDE = 'Right',
*      ( MB_V.EQ.NB_C .AND. IROFFV.EQ.ICOFFC )
*  else if STOREV = 'Rowwise'
*    If SIDE = 'Left',
*      ( NB_V.EQ.MB_C .AND. ICOFFV.EQ.IROFFC )
*    If SIDE = 'Right',
*      ( NB_V.EQ.NB_C .AND. ICOFFV.EQ.ICOFFC .AND. IVCOL.EQ.ICCOL )
*  end if
*
*  =====================================================================
*
*     .. Parameters ..
      INTEGER            BLOCK_CYCLIC_2D, CSRC_, CTXT_, DLEN_, DTYPE_,
     $                   lld_, mb_, m_, nb_, n_, rsrc_
      parameter( block_cyclic_2d = 1, dlen_ = 9, dtype_ = 1,
     $                     ctxt_ = 2, m_ = 3, n_ = 4, mb_ = 5, nb_ = 6,
     $                     rsrc_ = 7, csrc_ = 8, lld_ = 9 )
      DOUBLE PRECISION   ONE, ZERO
      parameter( one = 1.0d+0, zero = 0.0d+0 )
*     ..
*     .. Local Scalars ..
      LOGICAL            LEFT
      CHARACTER          COLBTOP, TRANST
      INTEGER            ICCOL1, ICCOL2, ICOFFC1, ICOFFC2, ICOFFV,
     $                   icrow1, icrow2, ictxt, iibeg, iic1, iic2,
     $                   iiend, iinxt, iiv, ileft, info, ioffc2, ioffv,
     $                   ipt, ipv, ipw, iroffc1, iroffc2, itop, ivcol,
     $                   ivrow, jjbeg, jjend, jjnxt, jjc1, jjc2, jjv,
     $                   ldc, ldv, lv, lw, mbc, mbv, mpc1, mpc2, mpc20,
     $                   mqv, mqv0, mycol, mydist, myrow, nbc, nbv,
     $                   npcol, nprow, nqc1, nqc2, nqcall, nqv
*     ..
*     .. External Subroutines ..
      EXTERNAL           blacs_abort, blacs_gridinfo, dgebr2d,
     $                   dgebs2d,dgemm, dgsum2d, dlamov,
     $                   dlaset, dtrbr2d, dtrbs2d, dtrmm,
     $                   infog2l, pbdmatadd, pbdtran, pb_topget, pxerbla
*     ..
*     .. Intrinsic Functions ..
      INTRINSIC          max, min, mod
*     ..
*     .. External Functions ..
      LOGICAL            LSAME
      INTEGER            ICEIL, NUMROC
      EXTERNAL           iceil, lsame, numroc
*     ..
*     .. Executable Statements ..
*
*     Quick return if possible
*
      IF( m.LE.0 .OR. n.LE.0 .OR. k.LE.0 )
     $   RETURN
*
*     Get grid parameters
*
      ictxt = descc( ctxt_ )
      CALL blacs_gridinfo( ictxt, nprow, npcol, myrow, mycol )
*
*     Check for currently supported options
*
      info = 0
      IF( .NOT.lsame( direct, 'B' ) ) THEN
         info = -3
      ELSE IF( .NOT.lsame( storev, 'R' ) ) THEN
         info = -4
      END IF
      IF( info.NE.0 ) THEN
         CALL pxerbla( ictxt, 'PDLARZB', -info )
         CALL blacs_abort( ictxt, 1 )
         RETURN
      END IF
*
      left = lsame( side, 'L' )
      IF( lsame( trans, 'N' ) ) THEN
          transt = 'T'
      ELSE
          transt = 'N'
      END IF
*
      CALL infog2l( iv, jv, descv, nprow, npcol, myrow, mycol, iiv, jjv,
     $              ivrow, ivcol )
      mbv = descv( mb_ )
      nbv = descv( nb_ )
      icoffv = mod( jv-1, nbv )
      nqv = numroc( l+icoffv, nbv, mycol, ivcol, npcol )
      IF( mycol.EQ.ivcol )
     $   nqv = nqv - icoffv
      ldv = descv( lld_ )
      iiv = min( iiv, ldv )
      jjv = min( jjv, max( 1, numroc( descv( n_ ), nbv, mycol,
     $                                descv( csrc_ ), npcol ) ) )
      ioffv = iiv + ( jjv-1 ) * ldv
      mbc = descc( mb_ )
      nbc = descc( nb_ )
      nqcall = numroc( descc( n_ ), nbc, mycol, descc( csrc_ ), npcol )
      CALL infog2l( ic, jc, descc, nprow, npcol, myrow, mycol, iic1,
     $              jjc1, icrow1, iccol1 )
      ldc = descc( lld_ )
      iic1 = min( iic1, ldc )
      jjc1 = min( jjc1, max( 1, nqcall ) )
*
      IF( left ) THEN
         iroffc1 = mod( ic-1, mbc )
         mpc1 = numroc( k+iroffc1, mbc, myrow, icrow1, nprow )
         IF( myrow.EQ.icrow1 )
     $      mpc1 = mpc1 - iroffc1
         icoffc1 = mod( jc-1, nbc )
         nqc1 = numroc( n+icoffc1, nbc, mycol, iccol1, npcol )
         IF( mycol.EQ.iccol1 )
     $      nqc1 = nqc1 - icoffc1
         CALL infog2l( ic+m-l, jc, descc, nprow, npcol, myrow, mycol,
     $                 iic2, jjc2, icrow2, iccol2 )
         iroffc2 = mod( ic+m-l-1, mbc )
         mpc2 = numroc( l+iroffc2, mbc, myrow, icrow2, nprow )
         IF( myrow.EQ.icrow2 )
     $      mpc2 = mpc2 - iroffc2
         icoffc2 = icoffc1
         nqc2 = nqc1
      ELSE
         iroffc1 = mod( ic-1, mbc )
         mpc1 = numroc( m+iroffc1, mbc, myrow, icrow1, nprow )
         IF( myrow.EQ.icrow1 )
     $      mpc1 = mpc1 - iroffc1
         icoffc1 = mod( jc-1, nbc )
         nqc1 = numroc( k+icoffc1, nbc, mycol, iccol1, npcol )
         IF( mycol.EQ.iccol1 )
     $      nqc1 = nqc1 - icoffc1
         CALL infog2l( ic, jc+n-l, descc, nprow, npcol, myrow, mycol,
     $                 iic2, jjc2, icrow2, iccol2 )
         iroffc2 = iroffc1
         mpc2 = mpc1
         icoffc2 = mod( jc+n-l-1, nbc )
         nqc2 = numroc( l+icoffc2, nbc, mycol, iccol2, npcol )
         IF( mycol.EQ.iccol2 )
     $      nqc2 = nqc2 - icoffc2
      END IF
      iic2 = min( iic2, ldc )
      jjc2 = min( jjc2, nqcall )
      ioffc2 = iic2 + ( jjc2-1 ) * ldc
*
      IF( lsame( side, 'L' ) ) THEN
*
*        Form Q*sub( C ) or Q'*sub( C )
*
*        IROFFC2 = ICOFFV is required by the current transposition
*        routine PBDTRAN
*
         mqv0 = numroc( m+icoffv, nbv, mycol, ivcol, npcol )
         IF( mycol.EQ.ivcol ) THEN
            mqv = mqv0 - icoffv
         ELSE
            mqv = mqv0
         END IF
         IF( myrow.EQ.icrow2 ) THEN
            mpc20 = mpc2 + iroffc2
         ELSE
            mpc20 = mpc2
         END IF
*
*        Locally V( IOFFV ) is K x MQV, C( IOFFC2 ) is MPC2 x NQC2
*        WORK( IPV ) is MPC20 x K = [ . V( IOFFV ) ]'
*        WORK( IPW ) is K x MQV0  = [ . V( IOFFV ) ]
*        WORK( IPT ) is the workspace for PBDTRAN
*
         ipv = 1
         ipw = ipv + mpc20 * k
         ipt = ipw + k * mqv0
         lv = max( 1, mpc20 )
         lw = max( 1, k )
*
         IF( myrow.EQ.ivrow ) THEN
            IF( mycol.EQ.ivcol ) THEN
               CALL dlamov( 'All', k, mqv, v( ioffv ), ldv,
     $                      work( ipw+icoffv*lw ), lw )
            ELSE
               CALL dlamov( 'All', k, mqv, v( ioffv ), ldv,
     $                      work( ipw ), lw )
            END IF
         END IF
*
*        WORK( IPV ) = WORK( IPW )' (replicated) is MPC20 x K
*
         CALL pbdtran( ictxt, 'Rowwise', 'Transpose', k, m+icoffv,
     $                 descv( nb_ ), work( ipw ), lw, zero,
     $                 work( ipv ), lv, ivrow, ivcol, icrow2, -1,
     $                 work( ipt ) )
*
*        WORK( IPV ) = ( . V )' -> WORK( IPV ) = V' is MPC2 x K
*
         IF( myrow.EQ.icrow2 )
     $      ipv = ipv + iroffc2
*
*        WORK( IPW ) becomes NQC2 x K = C( IOFFC2 )' * V'
*        WORK( IPW ) = C( IOFFC2 )' * V'  (NQC2 x MPC2 x K) -> NQC2 x K
*
         lw = max( 1, nqc2 )
*
         IF( mpc2.GT.0 ) THEN
            CALL dgemm( 'Transpose', 'No transpose', nqc2, k, mpc2,
     $                  one, c( ioffc2 ), ldc, work( ipv ), lv, zero,
     $                  work( ipw ), lw )
         ELSE
            CALL dlaset( 'All', nqc2, k, zero, zero, work( ipw ), lw )
         END IF
*
*        WORK( IPW ) = WORK( IPW ) + C1 ( NQC1 = NQC2 )
*
         IF( mpc1.GT.0 ) THEN
            mydist = mod( myrow-icrow1+nprow, nprow )
            itop = max( 0, mydist * mbc - iroffc1 )
            iibeg = iic1
            iiend = iic1 + mpc1 - 1
            iinxt = min( iceil( iibeg, mbc ) * mbc, iiend )
*
   10       CONTINUE
            IF( iibeg.LE.iinxt ) THEN
               CALL pbdmatadd( ictxt, 'Transpose', nqc2, iinxt-iibeg+1,
     $                         one, c( iibeg+(jjc1-1)*ldc ), ldc, one,
     $                         work( ipw+itop ), lw )
               mydist = mydist + nprow
               itop = mydist * mbc - iroffc1
               iibeg = iinxt +1
               iinxt = min( iinxt+mbc, iiend )
               GO TO 10
            END IF
         END IF
*
         CALL dgsum2d( ictxt, 'Columnwise', ' ', nqc2, k, work( ipw ),
     $                 lw, ivrow, mycol )
*
*        WORK( IPW ) = WORK( IPW ) * T' or WORK( IPW ) * T
*
         IF( myrow.EQ.ivrow ) THEN
            IF( mycol.EQ.ivcol ) THEN
*
*              Broadcast the block reflector to the other columns.
*
               CALL dtrbs2d( ictxt, 'Rowwise', ' ', 'Lower', 'Non unit',
     $                       k, k, t, mbv )
            ELSE
               CALL dtrbr2d( ictxt, 'Rowwise', ' ', 'Lower', 'Non unit',
     $                       k, k, t, mbv, myrow, ivcol )
            END IF
            CALL dtrmm( 'Right', 'Lower', transt, 'Non unit', nqc2, k,
     $                  one, t, mbv, work( ipw ), lw )
*
            CALL dgebs2d( ictxt, 'Columnwise', ' ', nqc2, k,
     $                    work( ipw ), lw )
         ELSE
            CALL dgebr2d( ictxt, 'Columnwise', ' ', nqc2, k,
     $                    work( ipw ), lw, ivrow, mycol )
         END IF
*
*        C1 = C1 - WORK( IPW )
*
         IF( mpc1.GT.0 ) THEN
            mydist = mod( myrow-icrow1+nprow, nprow )
            itop = max( 0, mydist * mbc - iroffc1 )
            iibeg = iic1
            iiend = iic1 + mpc1 - 1
            iinxt = min( iceil( iibeg, mbc ) * mbc, iiend )
*
   20       CONTINUE
            IF( iibeg.LE.iinxt ) THEN
               CALL pbdmatadd( ictxt, 'Transpose', iinxt-iibeg+1, nqc2,
     $                         -one, work( ipw+itop ), lw, one,
     $                         c( iibeg+(jjc1-1)*ldc ), ldc )
               mydist = mydist + nprow
               itop = mydist * mbc - iroffc1
               iibeg = iinxt +1
               iinxt = min( iinxt+mbc, iiend )
               GO TO 20
            END IF
         END IF
*
*            C2            C2      -     V'      *     W'
*        C( IOFFC2 ) = C( IOFFC2 ) - WORK( IPV ) * WORK( IPW )'
*                      MPC2 x NQC2    MPC2 x K      K x NQC2
*
         CALL dgemm( 'No transpose', 'Transpose', mpc2, nqc2, k, -one,
     $               work( ipv ), lv, work( ipw ), lw, one,
     $               c( ioffc2 ), ldc )
*
      ELSE
*
*        Form sub( C ) * Q or sub( C ) * Q'
*
*        Locally V( IOFFV ) is K x NQV, C( IOFFC2 ) is MPC2 x NQC2
*        WORK( IPV ) is K x NQV = V( IOFFV ), NQV = NQC2
*        WORK( IPW ) is MPC2 x K = C( IOFFC2 ) * V( IOFFV )'
*
         ipv = 1
         ipw = ipv + k * nqc2
         lv = max( 1, k )
         lw = max( 1, mpc2 )
*
*        Broadcast V to the other process rows.
*
         CALL pb_topget( ictxt, 'Broadcast', 'Columnwise', colbtop )
         IF( myrow.EQ.ivrow ) THEN
            CALL dgebs2d( ictxt, 'Columnwise', colbtop, k, nqc2,
     $                    v( ioffv ), ldv )
            IF( mycol.EQ.ivcol )
     $         CALL dtrbs2d( ictxt, 'Columnwise', colbtop, 'Lower',
     $                       'Non unit', k, k, t, mbv )
            CALL dlamov( 'All', k, nqc2, v( ioffv ), ldv, work( ipv ),
     $                   lv )
         ELSE
            CALL dgebr2d( ictxt, 'Columnwise', colbtop, k, nqc2,
     $                    work( ipv ), lv, ivrow, mycol )
            IF( mycol.EQ.ivcol )
     $         CALL dtrbr2d( ictxt, 'Columnwise', colbtop, 'Lower',
     $                       'Non unit', k, k, t, mbv, ivrow, mycol )
         END IF
*
*        WORK( IPV ) is K x NQC2 = V = V( IOFFV )
*        WORK( IPW ) = C( IOFFC2 ) * V'  (MPC2 x NQC2 x K) -> MPC2 x K
*
         IF( nqc2.GT.0 ) THEN
            CALL dgemm( 'No Transpose', 'Transpose', mpc2, k, nqc2,
     $                 one, c( ioffc2 ), ldc, work( ipv ), lv, zero,
     $                 work( ipw ), lw )
         ELSE
            CALL dlaset( 'All', mpc2, k, zero, zero, work( ipw ), lw )
         END IF
*
*        WORK( IPW ) = WORK( IPW ) + C1 ( MPC1 = MPC2 )
*
         IF( nqc1.GT.0 ) THEN
            mydist = mod( mycol-iccol1+npcol, npcol )
            ileft = max( 0, mydist * nbc - icoffc1 )
            jjbeg = jjc1
            jjend = jjc1 + nqc1 - 1
            jjnxt = min( iceil( jjbeg, nbc ) * nbc, jjend )
*
   30       CONTINUE
            IF( jjbeg.LE.jjnxt ) THEN
               CALL pbdmatadd( ictxt, 'No transpose', mpc2,
     $                         jjnxt-jjbeg+1, one,
     $                         c( iic1+(jjbeg-1)*ldc ), ldc, one,
     $                         work( ipw+ileft*lw ), lw )
               mydist = mydist + npcol
               ileft = mydist * nbc - icoffc1
               jjbeg = jjnxt +1
               jjnxt = min( jjnxt+nbc, jjend )
               GO TO 30
            END IF
         END IF
*
         CALL dgsum2d( ictxt, 'Rowwise', ' ', mpc2, k, work( ipw ),
     $                 lw, myrow, ivcol )
*
*        WORK( IPW ) = WORK( IPW ) * T' or WORK( IPW ) * T
*
         IF( mycol.EQ.ivcol ) THEN
            CALL dtrmm( 'Right', 'Lower', trans, 'Non unit', mpc2, k,
     $                  one, t, mbv, work( ipw ), lw )
            CALL dgebs2d( ictxt, 'Rowwise', ' ', mpc2, k, work( ipw ),
     $                    lw )
         ELSE
            CALL dgebr2d( ictxt, 'Rowwise', ' ', mpc2, k, work( ipw ),
     $                    lw, myrow, ivcol )
         END IF
*
*        C1 = C1 - WORK( IPW )
*
         IF( nqc1.GT.0 ) THEN
            mydist = mod( mycol-iccol1+npcol, npcol )
            ileft = max( 0, mydist * nbc - icoffc1 )
            jjbeg = jjc1
            jjend = jjc1 + nqc1 - 1
            jjnxt = min( iceil( jjbeg, nbc ) * nbc, jjend )
*
   40       CONTINUE
            IF( jjbeg.LE.jjnxt ) THEN
               CALL pbdmatadd( ictxt, 'No transpose', mpc2,
     $                         jjnxt-jjbeg+1, -one,
     $                         work( ipw+ileft*lw ), lw, one,
     $                         c( iic1+(jjbeg-1)*ldc ), ldc )
               mydist = mydist + npcol
               ileft = mydist * nbc - icoffc1
               jjbeg = jjnxt +1
               jjnxt = min( jjnxt+nbc, jjend )
               GO TO 40
            END IF
         END IF
*
*            C2           C2      -     W       *    V
*        C( IOFFC ) = C( IOFFC )  - WORK( IPW ) * WORK( IPV )
*                     MPC2 x NQC2    MPC2 x K      K x NQC2
*
         IF( ioffc2.GT.0 )
     $      CALL dgemm( 'No transpose', 'No transpose', mpc2, nqc2, k,
     $                  -one, work( ipw ), lw, work( ipv ), lv, one,
     $                  c( ioffc2 ), ldc )
*
      END IF
*
      RETURN
*
*     End of PDLARZB
*
      END