Statistiques
| Révision :

root / src / lapack / double / dlarz.f @ 10

Historique | Voir | Annoter | Télécharger (4,14 ko)

1
      SUBROUTINE DLARZ( SIDE, M, N, L, V, INCV, TAU, C, LDC, WORK )
2
*
3
*  -- LAPACK routine (version 3.2) --
4
*  -- LAPACK is a software package provided by Univ. of Tennessee,    --
5
*  -- Univ. of California Berkeley, Univ. of Colorado Denver and NAG Ltd..--
6
*     November 2006
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'
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 )' * v( 1:l )
105
*
106
            CALL DGEMV( 'Transpose', L, N, ONE, C( M-L+1, 1 ), 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 )'
115
*
116
            CALL DGER( L, N, -TAU, V, INCV, WORK, 1, C( M-L+1, 1 ),
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 )'
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