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
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
|
<HTML>
<HEAD><TITLE>MB03UD - SLICOT Library Routine Documentation</TITLE>
</HEAD>
<BODY>
<H2><A Name="MB03UD">MB03UD</A></H2>
<H3>
Computation of the singular value decomposition of a real upper triangular matrix
</H3>
<A HREF ="#Specification"><B>[Specification]</B></A>
<A HREF ="#Arguments"><B>[Arguments]</B></A>
<A HREF ="#Method"><B>[Method]</B></A>
<A HREF ="#References"><B>[References]</B></A>
<A HREF ="#Comments"><B>[Comments]</B></A>
<A HREF ="#Example"><B>[Example]</B></A>
<P>
<B><FONT SIZE="+1">Purpose</FONT></B>
<PRE>
To compute all, or part, of the singular value decomposition of a
real upper triangular matrix.
The N-by-N upper triangular matrix A is factored as A = Q*S*P',
where Q and P are N-by-N orthogonal matrices and S is an
N-by-N diagonal matrix with non-negative diagonal elements,
SV(1), SV(2), ..., SV(N), ordered such that
SV(1) >= SV(2) >= ... >= SV(N) >= 0.
The columns of Q are the left singular vectors of A, the diagonal
elements of S are the singular values of A and the columns of P
are the right singular vectors of A.
Either or both of Q and P' may be requested.
When P' is computed, it is returned in A.
</PRE>
<A name="Specification"><B><FONT SIZE="+1">Specification</FONT></B></A>
<PRE>
SUBROUTINE MB03UD( JOBQ, JOBP, N, A, LDA, Q, LDQ, SV, DWORK,
$ LDWORK, INFO )
C .. Scalar Arguments ..
CHARACTER JOBP, JOBQ
INTEGER INFO, LDA, LDQ, LDWORK, N
C .. Array Arguments ..
DOUBLE PRECISION A(LDA,*), DWORK(*), Q(LDQ,*), SV(*)
</PRE>
<A name="Arguments"><B><FONT SIZE="+1">Arguments</FONT></B></A>
<P>
<B>Mode Parameters</B>
<PRE>
JOBQ CHARACTER*1
Specifies whether the user wishes to compute the matrix Q
of left singular vectors as follows:
= 'V': Left singular vectors are computed;
= 'N': No left singular vectors are computed.
JOBP CHARACTER*1
Specifies whether the user wishes to compute the matrix P'
of right singular vectors as follows:
= 'V': Right singular vectors are computed;
= 'N': No right singular vectors are computed.
</PRE>
<B>Input/Output Parameters</B>
<PRE>
N (input) INTEGER
The order of the matrix A. N >= 0.
A (input/output) DOUBLE PRECISION array, dimension (LDA,N)
On entry, the leading N-by-N upper triangular part of this
array must contain the upper triangular matrix A.
On exit, if JOBP = 'V', the leading N-by-N part of this
array contains the N-by-N orthogonal matrix P'; otherwise
the N-by-N upper triangular part of A is used as internal
workspace. The strictly lower triangular part of A is set
internally to zero before the reduction to bidiagonal form
is performed.
LDA INTEGER
The leading dimension of array A. LDA >= MAX(1,N).
Q (output) DOUBLE PRECISION array, dimension (LDQ,N)
If JOBQ = 'V', the leading N-by-N part of this array
contains the orthogonal matrix Q.
If JOBQ = 'N', Q is not referenced.
LDQ INTEGER
The leading dimension of array Q.
LDQ >= 1, and when JOBQ = 'V', LDQ >= MAX(1,N).
SV (output) DOUBLE PRECISION array, dimension (N)
The N singular values of the matrix A, sorted in
descending order.
</PRE>
<B>Workspace</B>
<PRE>
DWORK DOUBLE PRECISION array, dimension (LDWORK)
On exit, if INFO = 0, DWORK(1) returns the optimal LDWORK;
if INFO > 0, DWORK(2:N) contains the unconverged
superdiagonal elements of an upper bidiagonal matrix B
whose diagonal is in SV (not necessarily sorted).
B satisfies A = Q*B*P', so it has the same singular
values as A, and singular vectors related by Q and P'.
LDWORK INTEGER
The length of the array DWORK.
LDWORK >= MAX(1,5*N).
For optimum performance LDWORK should be larger.
If LDWORK = -1, then a workspace query is assumed;
the routine only calculates the optimal size of the
DWORK array, returns this value as the first entry of
the DWORK array, and no error message related to LDWORK
is issued by XERBLA.
</PRE>
<B>Error Indicator</B>
<PRE>
INFO INTEGER
= 0: successful exit;
< 0: if INFO = -i, the i-th argument had an illegal
value;
> 0: the QR algorithm has failed to converge. In this
case INFO specifies how many superdiagonals did not
converge (see the description of DWORK).
This failure is not likely to occur.
</PRE>
<A name="Method"><B><FONT SIZE="+1">Method</FONT></B></A>
<PRE>
The routine reduces A to bidiagonal form by means of elementary
reflectors and then uses the QR algorithm on the bidiagonal form.
</PRE>
<A name="Comments"><B><FONT SIZE="+1">Further Comments</FONT></B></A>
<PRE>
None
</PRE>
<A name="Example"><B><FONT SIZE="+1">Example</FONT></B></A>
<P>
<B>Program Text</B>
<PRE>
* MB03UD EXAMPLE PROGRAM TEXT
*
* .. Parameters ..
INTEGER NIN, NOUT
PARAMETER ( NIN = 5, NOUT = 6 )
INTEGER NMAX
PARAMETER ( NMAX = 10 )
INTEGER LDA, LDQ
PARAMETER ( LDA = NMAX, LDQ = NMAX )
INTEGER LDWORK
PARAMETER ( LDWORK = MAX( 1, 5*NMAX ) )
* .. Local Scalars ..
CHARACTER*1 JOBQ, JOBP
INTEGER I, INFO, J, N
* .. Local Arrays ..
DOUBLE PRECISION A(LDA,NMAX), DWORK(LDWORK), Q(LDQ,NMAX),
$ SV(NMAX)
* .. External Functions ..
LOGICAL LSAME
* .. External Subroutines ..
EXTERNAL MB03UD
* .. Intrinsic Functions ..
INTRINSIC MAX
* .. Executable Statements ..
*
WRITE ( NOUT, FMT = 99999 )
* Skip the heading in the data file and read the data.
READ ( NIN, FMT = '()' )
READ ( NIN, FMT = * ) N, JOBQ, JOBP
IF ( N.LT.0 .OR. N.GT.NMAX ) THEN
WRITE ( NOUT, FMT = 99993 ) N
ELSE
READ ( NIN, FMT = * ) ( ( A(I,J), J = 1,N ), I = 1,N )
* Compute the singular values and vectors.
CALL MB03UD( JOBQ, JOBP, N, A, LDA, Q, LDQ, SV, DWORK,
$ LDWORK, INFO )
IF ( INFO.NE.0 ) THEN
WRITE ( NOUT, FMT = 99998 ) INFO
ELSE
WRITE ( NOUT, FMT = 99997 )
WRITE ( NOUT, FMT = 99995 ) ( SV(I), I = 1,N )
IF ( LSAME( JOBP, 'V' ) ) THEN
WRITE ( NOUT, FMT = 99996 )
DO 10 I = 1, N
WRITE ( NOUT, FMT = 99995 ) ( A(I,J), J = 1,N )
10 CONTINUE
END IF
IF ( LSAME( JOBQ, 'V' ) ) THEN
WRITE ( NOUT, FMT = 99994 )
DO 20 I = 1, N
WRITE ( NOUT, FMT = 99995 ) ( Q(I,J), J = 1,N )
20 CONTINUE
END IF
END IF
END IF
*
STOP
*
99999 FORMAT (' MB03UD EXAMPLE PROGRAM RESULTS',/1X)
99998 FORMAT (' INFO on exit from MB03UD = ',I2)
99997 FORMAT (' Singular values are ',I5)
99996 FORMAT (/' The transpose of the right singular vectors matrix is '
$ )
99995 FORMAT (8X,20(1X,F8.4))
99994 FORMAT (/' The left singular vectors matrix is ')
99993 FORMAT (/' N is out of range.',/' N = ',I5)
END
</PRE>
<B>Program Data</B>
<PRE>
MB03UD EXAMPLE PROGRAM DATA
4 V V
-1.0 37.0 -12.0 -12.0
0.0 -10.0 0.0 4.0
0.0 0.0 7.0 -6.0
0.0 0.0 0.0 -9.0
</PRE>
<B>Program Results</B>
<PRE>
MB03UD EXAMPLE PROGRAM RESULTS
Singular values are
42.0909 11.7764 5.4420 0.2336
The transpose of the right singular vectors matrix is
0.0230 -0.9084 0.2759 0.3132
0.0075 -0.1272 0.5312 -0.8376
0.0092 0.3978 0.8009 0.4476
0.9997 0.0182 -0.0177 -0.0050
The left singular vectors matrix is
-0.9671 -0.0882 -0.0501 -0.2335
0.2456 -0.1765 -0.4020 -0.8643
0.0012 0.7425 0.5367 -0.4008
-0.0670 0.6401 -0.7402 0.1945
</PRE>
<HR>
<p>
<A HREF=..\libindex.html><B>Return to index</B></A></BODY>
</HTML>
|
