1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
|
SUBROUTINE DPBT01( UPLO, N, KD, A, LDA, AFAC, LDAFAC, RWORK,
$ RESID )
*
* -- LAPACK test routine (version 3.0) --
* Univ. of Tennessee, Univ. of California Berkeley, NAG Ltd.,
* Courant Institute, Argonne National Lab, and Rice University
* February 29, 1992
*
* .. Scalar Arguments ..
CHARACTER UPLO
INTEGER KD, LDA, LDAFAC, N
DOUBLE PRECISION RESID
* ..
* .. Array Arguments ..
DOUBLE PRECISION A( LDA, * ), AFAC( LDAFAC, * ), RWORK( * )
* ..
*
* Purpose
* =======
*
* DPBT01 reconstructs a symmetric positive definite band matrix A from
* its L*L' or U'*U factorization and computes the residual
* norm( L*L' - A ) / ( N * norm(A) * EPS ) or
* norm( U'*U - A ) / ( N * norm(A) * EPS ),
* where EPS is the machine epsilon, L' is the conjugate transpose of
* L, and U' is the conjugate transpose of U.
*
* Arguments
* =========
*
* UPLO (input) CHARACTER*1
* Specifies whether the upper or lower triangular part of the
* symmetric matrix A is stored:
* = 'U': Upper triangular
* = 'L': Lower triangular
*
* N (input) INTEGER
* The number of rows and columns of the matrix A. N >= 0.
*
* KD (input) INTEGER
* The number of super-diagonals of the matrix A if UPLO = 'U',
* or the number of sub-diagonals if UPLO = 'L'. KD >= 0.
*
* A (input) DOUBLE PRECISION array, dimension (LDA,N)
* The original symmetric band matrix A. If UPLO = 'U', the
* upper triangular part of A is stored as a band matrix; if
* UPLO = 'L', the lower triangular part of A is stored. The
* columns of the appropriate triangle are stored in the columns
* of A and the diagonals of the triangle are stored in the rows
* of A. See DPBTRF for further details.
*
* LDA (input) INTEGER.
* The leading dimension of the array A. LDA >= max(1,KD+1).
*
* AFAC (input) DOUBLE PRECISION array, dimension (LDAFAC,N)
* The factored form of the matrix A. AFAC contains the factor
* L or U from the L*L' or U'*U factorization in band storage
* format, as computed by DPBTRF.
*
* LDAFAC (input) INTEGER
* The leading dimension of the array AFAC.
* LDAFAC >= max(1,KD+1).
*
* RWORK (workspace) DOUBLE PRECISION array, dimension (N)
*
* RESID (output) DOUBLE PRECISION
* If UPLO = 'L', norm(L*L' - A) / ( N * norm(A) * EPS )
* If UPLO = 'U', norm(U'*U - A) / ( N * norm(A) * EPS )
*
* =====================================================================
*
*
* .. Parameters ..
DOUBLE PRECISION ZERO, ONE
PARAMETER ( ZERO = 0.0D+0, ONE = 1.0D+0 )
* ..
* .. Local Scalars ..
INTEGER I, J, K, KC, KLEN, ML, MU
DOUBLE PRECISION ANORM, EPS, T
* ..
* .. External Functions ..
LOGICAL LSAME
DOUBLE PRECISION DDOT, DLAMCH, DLANSB
EXTERNAL LSAME, DDOT, DLAMCH, DLANSB
* ..
* .. External Subroutines ..
EXTERNAL DSCAL, DSYR, DTRMV
* ..
* .. Intrinsic Functions ..
INTRINSIC DBLE, MAX, MIN
* ..
* .. Executable Statements ..
*
* Quick exit if N = 0.
*
IF( N.LE.0 ) THEN
RESID = ZERO
RETURN
END IF
*
* Exit with RESID = 1/EPS if ANORM = 0.
*
EPS = DLAMCH( 'Epsilon' )
ANORM = DLANSB( '1', UPLO, N, KD, A, LDA, RWORK )
IF( ANORM.LE.ZERO ) THEN
RESID = ONE / EPS
RETURN
END IF
*
* Compute the product U'*U, overwriting U.
*
IF( LSAME( UPLO, 'U' ) ) THEN
DO 10 K = N, 1, -1
KC = MAX( 1, KD+2-K )
KLEN = KD + 1 - KC
*
* Compute the (K,K) element of the result.
*
T = DDOT( KLEN+1, AFAC( KC, K ), 1, AFAC( KC, K ), 1 )
AFAC( KD+1, K ) = T
*
* Compute the rest of column K.
*
IF( KLEN.GT.0 )
$ CALL DTRMV( 'Upper', 'Transpose', 'Non-unit', KLEN,
$ AFAC( KD+1, K-KLEN ), LDAFAC-1,
$ AFAC( KC, K ), 1 )
*
10 CONTINUE
*
* UPLO = 'L': Compute the product L*L', overwriting L.
*
ELSE
DO 20 K = N, 1, -1
KLEN = MIN( KD, N-K )
*
* Add a multiple of column K of the factor L to each of
* columns K+1 through N.
*
IF( KLEN.GT.0 )
$ CALL DSYR( 'Lower', KLEN, ONE, AFAC( 2, K ), 1,
$ AFAC( 1, K+1 ), LDAFAC-1 )
*
* Scale column K by the diagonal element.
*
T = AFAC( 1, K )
CALL DSCAL( KLEN+1, T, AFAC( 1, K ), 1 )
*
20 CONTINUE
END IF
*
* Compute the difference L*L' - A or U'*U - A.
*
IF( LSAME( UPLO, 'U' ) ) THEN
DO 40 J = 1, N
MU = MAX( 1, KD+2-J )
DO 30 I = MU, KD + 1
AFAC( I, J ) = AFAC( I, J ) - A( I, J )
30 CONTINUE
40 CONTINUE
ELSE
DO 60 J = 1, N
ML = MIN( KD+1, N-J+1 )
DO 50 I = 1, ML
AFAC( I, J ) = AFAC( I, J ) - A( I, J )
50 CONTINUE
60 CONTINUE
END IF
*
* Compute norm( L*L' - A ) / ( N * norm(A) * EPS )
*
RESID = DLANSB( 'I', UPLO, N, KD, AFAC, LDAFAC, RWORK )
*
RESID = ( ( RESID / DBLE( N ) ) / ANORM ) / EPS
*
RETURN
*
* End of DPBT01
*
END
|
