1       SUBROUTINE DLARZ( SIDE, M, N, L, V, INCV, TAU, C, LDC, WORK )
  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          SIDE
 10       INTEGER            INCV, L, LDC, M, N
 11       DOUBLE PRECISION   TAU
 12 *     ..
 13 *     .. Array Arguments ..
 14       DOUBLE PRECISION   C( LDC, * ), V( * ), WORK( * )
 15 *     ..
 16 *
 17 *  Purpose
 18 *  =======
 19 *
 20 *  DLARZ applies a real elementary reflector H to a real M-by-N
 21 *  matrix C, from either the left or the right. H is represented in the
 22 *  form
 23 *
 24 *        H = I - tau * v * v**T
 25 *
 26 *  where tau is a real scalar and v is a real vector.
 27 *
 28 *  If tau = 0, then H is taken to be the unit matrix.
 29 *
 30 *
 31 *  H is a product of k elementary reflectors as returned by DTZRZF.
 32 *
 33 *  Arguments
 34 *  =========
 35 *
 36 *  SIDE    (input) CHARACTER*1
 37 *          = 'L': form  H * C
 38 *          = 'R': form  C * H
 39 *
 40 *  M       (input) INTEGER
 41 *          The number of rows of the matrix C.
 42 *
 43 *  N       (input) INTEGER
 44 *          The number of columns of the matrix C.
 45 *
 46 *  L       (input) INTEGER
 47 *          The number of entries of the vector V containing
 48 *          the meaningful part of the Householder vectors.
 49 *          If SIDE = 'L', M >= L >= 0, if SIDE = 'R', N >= L >= 0.
 50 *
 51 *  V       (input) DOUBLE PRECISION array, dimension (1+(L-1)*abs(INCV))
 52 *          The vector v in the representation of H as returned by
 53 *          DTZRZF. V is not used if TAU = 0.
 54 *
 55 *  INCV    (input) INTEGER
 56 *          The increment between elements of v. INCV <> 0.
 57 *
 58 *  TAU     (input) DOUBLE PRECISION
 59 *          The value tau in the representation of H.
 60 *
 61 *  C       (input/output) DOUBLE PRECISION array, dimension (LDC,N)
 62 *          On entry, the M-by-N matrix C.
 63 *          On exit, C is overwritten by the matrix H * C if SIDE = 'L',
 64 *          or C * H if SIDE = 'R'.
 65 *
 66 *  LDC     (input) INTEGER
 67 *          The leading dimension of the array C. LDC >= max(1,M).
 68 *
 69 *  WORK    (workspace) DOUBLE PRECISION array, dimension
 70 *                         (N) if SIDE = 'L'
 71 *                      or (M) if SIDE = 'R'
 72 *
 73 *  Further Details
 74 *  ===============
 75 *
 76 *  Based on contributions by
 77 *    A. Petitet, Computer Science Dept., Univ. of Tenn., Knoxville, USA
 78 *
 79 *  =====================================================================
 80 *
 81 *     .. Parameters ..
 82       DOUBLE PRECISION   ONE, ZERO
 83       PARAMETER          ( ONE = 1.0D+0, ZERO = 0.0D+0 )
 84 *     ..
 85 *     .. External Subroutines ..
 86       EXTERNAL           DAXPY, DCOPY, DGEMV, DGER
 87 *     ..
 88 *     .. External Functions ..
 89       LOGICAL            LSAME
 90       EXTERNAL           LSAME
 91 *     ..
 92 *     .. Executable Statements ..
 93 *
 94       IF( LSAME( SIDE, 'L' ) ) THEN
 95 *
 96 *        Form  H * C
 97 *
 98          IF( TAU.NE.ZERO ) THEN
 99 *
100 *           w( 1:n ) = C( 1, 1:n )
101 *
102             CALL DCOPY( N, C, LDC, WORK, 1 )
103 *
104 *           w( 1:n ) = w( 1:n ) + C( m-l+1:m, 1:n )**T * v( 1:l )
105 *
106             CALL DGEMV( 'Transpose', L, N, ONE, C( M-L+11 ), LDC, V,
107      $                  INCV, ONE, WORK, 1 )
108 *
109 *           C( 1, 1:n ) = C( 1, 1:n ) - tau * w( 1:n )
110 *
111             CALL DAXPY( N, -TAU, WORK, 1, C, LDC )
112 *
113 *           C( m-l+1:m, 1:n ) = C( m-l+1:m, 1:n ) - ...
114 *                               tau * v( 1:l ) * w( 1:n )**T
115 *
116             CALL DGER( L, N, -TAU, V, INCV, WORK, 1, C( M-L+11 ),
117      $                 LDC )
118          END IF
119 *
120       ELSE
121 *
122 *        Form  C * H
123 *
124          IF( TAU.NE.ZERO ) THEN
125 *
126 *           w( 1:m ) = C( 1:m, 1 )
127 *
128             CALL DCOPY( M, C, 1, WORK, 1 )
129 *
130 *           w( 1:m ) = w( 1:m ) + C( 1:m, n-l+1:n, 1:n ) * v( 1:l )
131 *
132             CALL DGEMV( 'No transpose', M, L, ONE, C( 1, N-L+1 ), LDC,
133      $                  V, INCV, ONE, WORK, 1 )
134 *
135 *           C( 1:m, 1 ) = C( 1:m, 1 ) - tau * w( 1:m )
136 *
137             CALL DAXPY( M, -TAU, WORK, 1, C, 1 )
138 *
139 *           C( 1:m, n-l+1:n ) = C( 1:m, n-l+1:n ) - ...
140 *                               tau * w( 1:m ) * v( 1:l )**T
141 *
142             CALL DGER( M, L, -TAU, WORK, 1, V, INCV, C( 1, N-L+1 ),
143      $                 LDC )
144 *
145          END IF
146 *
147       END IF
148 *
149       RETURN
150 *
151 *     End of DLARZ
152 *
153       END