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
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
|
SUBROUTINE CLAHILB(N, NRHS, A, LDA, X, LDX, B, LDB, WORK,
$ INFO, PATH)
!
! -- LAPACK auxiliary test routine (version 3.0) --
! Univ. of Tennessee, Univ. of California Berkeley, NAG Ltd.,
! Courant Institute, Argonne National Lab, and Rice University
! 28 August, 2006
!
! David Vu <dtv@cs.berkeley.edu>
! Yozo Hida <yozo@cs.berkeley.edu>
! Jason Riedy <ejr@cs.berkeley.edu>
! D. Halligan <dhalligan@berkeley.edu>
!
IMPLICIT NONE
! .. Scalar Arguments ..
INTEGER N, NRHS, LDA, LDX, LDB, INFO
! .. Array Arguments ..
REAL WORK(N)
COMPLEX A(LDA,N), X(LDX, NRHS), B(LDB, NRHS)
CHARACTER*3 PATH
! ..
!
! Purpose
! =======
!
! CLAHILB generates an N by N scaled Hilbert matrix in A along with
! NRHS right-hand sides in B and solutions in X such that A*X=B.
!
! The Hilbert matrix is scaled by M = LCM(1, 2, ..., 2*N-1) so that all
! entries are integers. The right-hand sides are the first NRHS
! columns of M * the identity matrix, and the solutions are the
! first NRHS columns of the inverse Hilbert matrix.
!
! The condition number of the Hilbert matrix grows exponentially with
! its size, roughly as O(e ** (3.5*N)). Additionally, the inverse
! Hilbert matrices beyond a relatively small dimension cannot be
! generated exactly without extra precision. Precision is exhausted
! when the largest entry in the inverse Hilbert matrix is greater than
! 2 to the power of the number of bits in the fraction of the data type
! used plus one, which is 24 for single precision.
!
! In single, the generated solution is exact for N <= 6 and has
! small componentwise error for 7 <= N <= 11.
!
! Arguments
! =========
!
! N (input) INTEGER
! The dimension of the matrix A.
!
! NRHS (input) NRHS
! The requested number of right-hand sides.
!
! A (output) COMPLEX array, dimension (LDA, N)
! The generated scaled Hilbert matrix.
!
! LDA (input) INTEGER
! The leading dimension of the array A. LDA >= N.
!
! X (output) COMPLEX array, dimension (LDX, NRHS)
! The generated exact solutions. Currently, the first NRHS
! columns of the inverse Hilbert matrix.
!
! LDX (input) INTEGER
! The leading dimension of the array X. LDX >= N.
!
! B (output) REAL array, dimension (LDB, NRHS)
! The generated right-hand sides. Currently, the first NRHS
! columns of LCM(1, 2, ..., 2*N-1) * the identity matrix.
!
! LDB (input) INTEGER
! The leading dimension of the array B. LDB >= N.
!
! WORK (workspace) REAL array, dimension (N)
!
!
! INFO (output) INTEGER
! = 0: successful exit
! = 1: N is too large; the data is still generated but may not
! be not exact.
! < 0: if INFO = -i, the i-th argument had an illegal value
!
! =====================================================================
! .. Local Scalars ..
INTEGER TM, TI, R
INTEGER M
INTEGER I, J
COMPLEX TMP
CHARACTER*2 C2
! .. Parameters ..
! NMAX_EXACT the largest dimension where the generated data is
! exact.
! NMAX_APPROX the largest dimension where the generated data has
! a small componentwise relative error.
! ??? complex uses how many bits ???
INTEGER NMAX_EXACT, NMAX_APPROX, SIZE_D
PARAMETER (NMAX_EXACT = 6, NMAX_APPROX = 11, SIZE_D = 8)
! d's are generated from random permuation of those eight elements.
COMPLEX D1(8), D2(8), INVD1(8), INVD2(8)
DATA D1 /(-1,0),(0,1),(-1,-1),(0,-1),(1,0),(-1,1),(1,1),(1,-1)/
DATA D2 /(-1,0),(0,-1),(-1,1),(0,1),(1,0),(-1,-1),(1,-1),(1,1)/
DATA INVD1 /(-1,0),(0,-1),(-.5,.5),(0,1),(1,0),
$ (-.5,-.5),(.5,-.5),(.5,.5)/
DATA INVD2 /(-1,0),(0,1),(-.5,-.5),(0,-1),(1,0),
$ (-.5,.5),(.5,.5),(.5,-.5)/
! ..
! .. External Functions
EXTERNAL CLASET, LSAMEN
INTRINSIC REAL
LOGICAL LSAMEN
! ..
! .. Executable Statements ..
C2 = PATH( 2: 3 )
!
! Test the input arguments
!
INFO = 0
IF (N .LT. 0 .OR. N .GT. NMAX_APPROX) THEN
INFO = -1
ELSE IF (NRHS .LT. 0) THEN
INFO = -2
ELSE IF (LDA .LT. N) THEN
INFO = -4
ELSE IF (LDX .LT. N) THEN
INFO = -6
ELSE IF (LDB .LT. N) THEN
INFO = -8
END IF
IF (INFO .LT. 0) THEN
CALL XERBLA('CLAHILB', -INFO)
RETURN
END IF
IF (N .GT. NMAX_EXACT) THEN
INFO = 1
END IF
! Compute M = the LCM of the integers [1, 2*N-1]. The largest
! reasonable N is small enough that integers suffice (up to N = 11).
M = 1
DO I = 2, (2*N-1)
TM = M
TI = I
R = MOD(TM, TI)
DO WHILE (R .NE. 0)
TM = TI
TI = R
R = MOD(TM, TI)
END DO
M = (M / TI) * I
END DO
! Generate the scaled Hilbert matrix in A
! If we are testing SY routines, take
! D1_i = D2_i, else, D1_i = D2_i*
IF ( LSAMEN( 2, C2, 'SY' ) ) THEN
DO J = 1, N
DO I = 1, N
A(I, J) = D1(MOD(J,SIZE_D)+1) * (REAL(M) / (I + J - 1))
$ * D1(MOD(I,SIZE_D)+1)
END DO
END DO
ELSE
DO J = 1, N
DO I = 1, N
A(I, J) = D1(MOD(J,SIZE_D)+1) * (REAL(M) / (I + J - 1))
$ * D2(MOD(I,SIZE_D)+1)
END DO
END DO
END IF
! Generate matrix B as simply the first NRHS columns of M * the
! identity.
TMP = REAL(M)
CALL CLASET('Full', N, NRHS, (0.0,0.0), TMP, B, LDB)
! Generate the true solutions in X. Because B = the first NRHS
! columns of M*I, the true solutions are just the first NRHS columns
! of the inverse Hilbert matrix.
WORK(1) = N
DO J = 2, N
WORK(J) = ( ( (WORK(J-1)/(J-1)) * (J-1 - N) ) /(J-1) )
$ * (N +J -1)
END DO
! If we are testing SY routines,
! take D1_i = D2_i, else, D1_i = D2_i*
IF ( LSAMEN( 2, C2, 'SY' ) ) THEN
DO J = 1, NRHS
DO I = 1, N
X(I, J) =
$ INVD1(MOD(J,SIZE_D)+1) *
$ ((WORK(I)*WORK(J)) / (I + J - 1))
$ * INVD1(MOD(I,SIZE_D)+1)
END DO
END DO
ELSE
DO J = 1, NRHS
DO I = 1, N
X(I, J) =
$ INVD2(MOD(J,SIZE_D)+1) *
$ ((WORK(I)*WORK(J)) / (I + J - 1))
$ * INVD1(MOD(I,SIZE_D)+1)
END DO
END DO
END IF
END
|
