pclaschk.f

Go to the documentation of this file.

      SUBROUTINE pclaschk( SYMM, DIAG, N, NRHS, X, IX, JX, DESCX,
     $                     IASEED, IA, JA, DESCA, IBSEED, ANORM, RESID,
     $                     WORK )
*
*  -- ScaLAPACK auxiliary routine (version 1.7) --
*     University of Tennessee, Knoxville, Oak Ridge National Laboratory,
*     and University of California, Berkeley.
*     May 1, 1997
*
*     .. Scalar Arguments ..
      CHARACTER          DIAG, SYMM
      INTEGER            IA, IASEED, IBSEED, IX, JA, JX, N, NRHS
      REAL               ANORM, RESID
*     ..
*     .. Array Arguments ..
      INTEGER            DESCA( * ), DESCX( * )
      COMPLEX            WORK( * ), X( * )
*     ..
*
*  Purpose
*  =======
*
*  PCLASCHK computes the residual
*  || sub( A )*sub( X ) - B || / (|| sub( A ) ||*|| sub( X ) ||*eps*N)
*  to check the accuracy of the factorization and solve steps in the
*  LU and Cholesky decompositions, where sub( A ) denotes
*  A(IA:IA+N-1,JA,JA+N-1), sub( X ) denotes X(IX:IX+N-1, JX:JX+NRHS-1).
*
*  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
*  =========
*
*  SYMM      (global input) CHARACTER
*          if SYMM = 'H', sub( A ) is a hermitian distributed matrix,
*          otherwise sub( A ) is a general distributed matrix.
*
*  DIAG    (global input) CHARACTER
*          If DIAG = 'D', sub( A ) is diagonally dominant.
*
*  N       (global input) INTEGER
*          The number of columns to be operated on, i.e. the number of
*          columns of the distributed submatrix sub( A ). N >= 0.
*
*  NRHS    (global input) INTEGER
*          The number of right-hand-sides, i.e the number of columns
*          of the distributed matrix sub( X ). NRHS >= 0.
*
*  X       (local input) COMPLEX pointer into the local memory
*          to an array of dimension (LLD_X,LOCc(JX+NRHS-1). This array
*          contains the local pieces of the answer vector(s) sub( X ) of
*          sub( A ) sub( X ) - B, split up over a column of processes.
*
*  IX      (global input) INTEGER
*          The row index in the global array X indicating the first
*          row of sub( X ).
*
*  JX      (global input) INTEGER
*          The column index in the global array X indicating the
*          first column of sub( X ).
*
*  DESCX   (global and local input) INTEGER array of dimension DLEN_.
*          The array descriptor for the distributed matrix X.
*
*  IASEED  (global input) INTEGER
*          The seed number to generate the original matrix Ao.
*
*  IA      (global input) INTEGER
*          The row index in the global array A indicating the first
*          row of sub( A ).
*
*  JA      (global input) INTEGER
*          The column index in the global array A indicating the
*          first column of sub( A ).
*
*  DESCA   (global and local input) INTEGER array of dimension DLEN_.
*          The array descriptor for the distributed matrix A.
*
*  IBSEED  (global input) INTEGER
*          The seed number to generate the original matrix B.
*
*  ANORM   (global input) REAL
*          The 1-norm or infinity norm of the distributed matrix
*          sub( A ).
*
*  RESID   (global output) REAL
*          The residual error:
*          ||sub( A )*sub( X )-B|| / (||sub( A )||*||sub( X )||*eps*N).
*
*  WORK    (local workspace) COMPLEX array, dimension (LWORK)
*          LWORK >= MAX(1,Np)*NB_X + Nq*NB_X + MAX( MAX(NQ*MB_A,2*NB_X),
*          NB_X * NUMROC( NUMROC(N,MB_X,0,0,NPCOL), MB_X, 0, 0, LCMQ ) )
*
*  =====================================================================
*
*     .. 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 )
      COMPLEX            ZERO, ONE
      PARAMETER          ( ONE = ( 1.0e+0, 0.0e+0 ),
     $                     zero = ( 0.0e+0, 0.0e+0 ) )
*     ..
*     .. Local Scalars ..
      INTEGER            IACOL, IAROW, IB, ICOFF, ICTXT, ICURCOL, IDUMM,
     $                   II, IIA, IIX, IOFFX, IPA, IPB, IPW, IPX, IROFF,
     $                   ixcol, ixrow, j, jbrhs, jj, jja, jjx, ldx,
     $                   mycol, myrow, np, npcol, nprow, nq
      REAL               DIVISOR, EPS, RESID1
      COMPLEX            BETA
*     ..
*     .. External Subroutines ..
      EXTERNAL           blacs_gridinfo, cgamx2d, cgemm, cgsum2d,
     $                   claset, pbctran, pcmatgen, sgebr2d,
     $                   sgebs2d, sgerv2d, sgesd2d
*     ..
*     .. External Functions ..
      INTEGER            ICAMAX, NUMROC
      REAL               PSLAMCH
      EXTERNAL           icamax, numroc, pslamch
*     ..
*     .. Intrinsic Functions ..
      INTRINSIC          abs, max, min, mod, real
*     ..
*     .. Executable Statements ..
*
*     Get needed initial parameters
*
      ictxt = desca( ctxt_ )
      CALL blacs_gridinfo( ictxt, nprow, npcol, myrow, mycol )
*
      eps = pslamch( ictxt, 'eps' )
      resid = 0.0e+0
      divisor = anorm * eps * real( n )
*
      CALL infog2l( ia, ja, desca, nprow, npcol, myrow, mycol, iia, jja,
     $              iarow, iacol )
      CALL infog2l( ix, jx, descx, nprow, npcol, myrow, mycol, iix, jjx,
     $              ixrow, ixcol )
      iroff = mod( ia-1, desca( mb_ ) )
      icoff = mod( ja-1, desca( nb_ ) )
      np = numroc( n+iroff, desca( mb_ ), myrow, iarow, nprow )
      nq = numroc( n+icoff, desca( nb_ ), mycol, iacol, npcol )
*
      ldx = max( 1, np )
      ipb = 1
      ipx = ipb + np * descx( nb_ )
      ipa = ipx + nq * descx( nb_ )
*
      IF( myrow.EQ.iarow )
     $   np = np - iroff
      IF( mycol.EQ.iacol )
     $   nq = nq - icoff
*
      icurcol = ixcol
*
*     Loop over the rhs
*
      DO 40 j = 1, nrhs, descx( nb_ )
         jbrhs = min( descx( nb_ ), nrhs-j+1 )
*
*        Transpose x from ICURCOL to all rows
*
         ioffx = iix + ( jjx - 1 ) * descx( lld_ )
         CALL pbctran( ictxt, 'Column', 'Transpose', n, jbrhs,
     $              descx( mb_ ), x( ioffx ), descx( lld_ ), zero,
     $              work( ipx ), jbrhs, ixrow, icurcol, -1, iacol,
     $              work( ipa ) )
*
*        Regenerate B in IXCOL
*
         IF( mycol.EQ.icurcol ) THEN
            CALL pcmatgen( ictxt, 'N', 'N', descx( m_ ), descx( n_ ),
     $                     descx( mb_ ), descx( nb_ ), work( ipb ), ldx,
     $                     ixrow, ixcol, ibseed, iix-1, np, jjx-1,
     $                     jbrhs, myrow, mycol, nprow, npcol )
            beta = one
         ELSE
            beta = zero
         END IF
*
         IF( nq.GT.0 ) THEN
            DO 10 ii = iia, iia+np-1, desca( mb_ )
               ib = min( desca( mb_ ), iia+np-ii )
*
*              Regenerate ib rows of the matrix A(IA:IA+N-1,JA:JA+N-1).
*
               CALL pcmatgen( ictxt, symm, diag, desca( m_ ),
     $                        desca( n_ ), desca( mb_ ), desca( nb_ ),
     $                        work( ipa ), ib, desca( rsrc_ ),
     $                        desca( csrc_ ), iaseed, ii-1, ib,
     $                        jja-1, nq, myrow, mycol, nprow, npcol )
*
*              Compute B <= B - A * X.
*
               CALL cgemm( 'No transpose', 'Transpose', ib, jbrhs, nq,
     $                     -one, work( ipa ), ib, work( ipx ), jbrhs,
     $                     beta, work( ipb+ii-iia ), ldx )
*
   10       CONTINUE
*
         ELSE IF( mycol.NE.icurcol ) THEN
*
            CALL claset( 'All', np, jbrhs, zero, zero, work( ipb ),
     $                   ldx )
*
         END IF
*
*        Add B rowwise to ICURCOL
*
         CALL cgsum2d( ictxt, 'Row', ' ', np, jbrhs, work( ipb ), ldx,
     $                 myrow, icurcol )
*
         IF( mycol.EQ.icurcol ) THEN
*
*           Figure || A * X - B || & || X ||
*
            ipw = ipa + jbrhs
            DO 20 jj = 0, jbrhs - 1
               IF( np.GT.0 ) THEN
                  ii = icamax( np, work( ipb+jj*ldx ), 1 )
                  work( ipa+jj ) = abs( work( ipb+ii-1+jj*ldx ) )
                  work( ipw+jj ) = abs( x( ioffx + icamax( np,
     $            x( ioffx + jj*descx( lld_ ) ), 1 )-1+jj*
     $            descx( lld_ ) ) )
               ELSE
                  work( ipa+jj ) = zero
                  work( ipw+jj ) = zero
               END IF
   20       CONTINUE
*
*           After CGAMX2D computation,
*              WORK(IPB) has the maximum of || Ax - b ||, and
*              WORK(IPX) has the maximum of || X ||.
*
            CALL cgamx2d( ictxt, 'Column', ' ', 1, 2*jbrhs,
     $                    work( ipa ), 1, idumm, idumm, -1, 0, icurcol )
*
*           Calculate residual = ||Ax-b|| / (||x||*||A||*eps*N)
*
            IF( myrow.EQ.0 ) THEN
               DO 30 jj = 0, jbrhs - 1
                  resid1 = real( work( ipa+jj ) ) /
     $                     ( real( work( ipw+jj ) )*divisor )
                  IF( resid.LT.resid1 )
     $               resid = resid1
   30          CONTINUE
               IF( mycol.NE.0 )
     $            CALL sgesd2d( ictxt, 1, 1, resid, 1, 0, 0 )
            END IF
*
         ELSE IF( myrow.EQ.0 .AND. mycol.EQ.0 ) THEN
*
            CALL sgerv2d( ictxt, 1, 1, resid1, 1, 0, icurcol )
            IF( resid.LT.resid1 )
     $         resid = resid1
*
         END IF
*
         IF( mycol.EQ.icurcol )
     $      jjx = jjx + jbrhs
         icurcol = mod( icurcol+1, npcol )
*
   40 CONTINUE
*
      IF( myrow.EQ.0 .AND. mycol.EQ.0 ) THEN
         CALL sgebs2d( ictxt, 'All', ' ', 1, 1, resid, 1 )
      ELSE
         CALL sgebr2d( ictxt, 'All', ' ', 1, 1, resid, 1, 0, 0 )
      END IF
*
      RETURN
*
*     End of PCLASCHK
*
      END