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
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
|
<HTML>
<HEAD><TITLE>MB04UD - SLICOT Library Routine Documentation</TITLE>
</HEAD>
<BODY>
<H2><A Name="MB04UD">MB04UD</A></H2>
<H3>
Column echelon form by unitary transformations for a rectangular matrix (added functionality)
</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 orthogonal transformations Q and Z such that the
transformed pencil Q'(sE-A)Z has the E matrix in column echelon
form, where E and A are M-by-N matrices.
</PRE>
<A name="Specification"><B><FONT SIZE="+1">Specification</FONT></B></A>
<PRE>
SUBROUTINE MB04UD( JOBQ, JOBZ, M, N, A, LDA, E, LDE, Q, LDQ,
$ Z, LDZ, RANKE, ISTAIR, TOL, DWORK, INFO )
C .. Scalar Arguments ..
CHARACTER JOBQ, JOBZ
INTEGER INFO, LDA, LDE, LDQ, LDZ, M, N, RANKE
DOUBLE PRECISION TOL
C .. Array Arguments ..
INTEGER ISTAIR(*)
DOUBLE PRECISION A(LDA,*), DWORK(*), E(LDE,*), Q(LDQ,*), Z(LDZ,*)
</PRE>
<A name="Arguments"><B><FONT SIZE="+1">Arguments</FONT></B></A>
<P>
<B>Mode Parameters</B>
<PRE>
JOBQ CHARACTER*1
Indicates whether the user wishes to accumulate in a
matrix Q the unitary row permutations, as follows:
= 'N': Do not form Q;
= 'I': Q is initialized to the unit matrix and the
unitary row permutation matrix Q is returned;
= 'U': The given matrix Q is updated by the unitary
row permutations used in the reduction.
JOBZ CHARACTER*1
Indicates whether the user wishes to accumulate in a
matrix Z the unitary column transformations, as follows:
= 'N': Do not form Z;
= 'I': Z is initialized to the unit matrix and the
unitary transformation matrix Z is returned;
= 'U': The given matrix Z is updated by the unitary
transformations used in the reduction.
</PRE>
<B>Input/Output Parameters</B>
<PRE>
M (input) INTEGER
The number of rows in the matrices A, E and the order of
the matrix Q. M >= 0.
N (input) INTEGER
The number of columns in the matrices A, E and the order
of the matrix Z. N >= 0.
A (input/output) DOUBLE PRECISION array, dimension (LDA,N)
On entry, the leading M-by-N part of this array must
contain the A matrix of the pencil sE-A.
On exit, the leading M-by-N part of this array contains
the unitary transformed matrix Q' * A * Z.
LDA INTEGER
The leading dimension of array A. LDA >= MAX(1,M).
E (input/output) DOUBLE PRECISION array, dimension (LDE,N)
On entry, the leading M-by-N part of this array must
contain the E matrix of the pencil sE-A, to be reduced to
column echelon form.
On exit, the leading M-by-N part of this array contains
the unitary transformed matrix Q' * E * Z, which is in
column echelon form.
LDE INTEGER
The leading dimension of array E. LDE >= MAX(1,M).
Q (input/output) DOUBLE PRECISION array, dimension (LDQ,*)
On entry, if JOBQ = 'U', then the leading M-by-M part of
this array must contain a given matrix Q (e.g. from a
previous call to another SLICOT routine), and on exit, the
leading M-by-M part of this array contains the product of
the input matrix Q and the row permutation matrix used to
transform the rows of matrix E.
On exit, if JOBQ = 'I', then the leading M-by-M part of
this array contains the matrix of accumulated unitary
row transformations performed.
If JOBQ = 'N', the array Q is not referenced and can be
supplied as a dummy array (i.e. set parameter LDQ = 1 and
declare this array to be Q(1,1) in the calling program).
LDQ INTEGER
The leading dimension of array Q. If JOBQ = 'U' or
JOBQ = 'I', LDQ >= MAX(1,M); if JOBQ = 'N', LDQ >= 1.
Z (input/output) DOUBLE PRECISION array, dimension (LDZ,*)
On entry, if JOBZ = 'U', then the leading N-by-N part of
this array must contain a given matrix Z (e.g. from a
previous call to another SLICOT routine), and on exit, the
leading N-by-N part of this array contains the product of
the input matrix Z and the column transformation matrix
used to transform the columns of matrix E.
On exit, if JOBZ = 'I', then the leading N-by-N part of
this array contains the matrix of accumulated unitary
column transformations performed.
If JOBZ = 'N', the array Z is not referenced and can be
supplied as a dummy array (i.e. set parameter LDZ = 1 and
declare this array to be Z(1,1) in the calling program).
LDZ INTEGER
The leading dimension of array Z. If JOBZ = 'U' or
JOBZ = 'I', LDZ >= MAX(1,N); if JOBZ = 'N', LDZ >= 1.
RANKE (output) INTEGER
The computed rank of the unitary transformed matrix E.
ISTAIR (output) INTEGER array, dimension (M)
This array contains information on the column echelon form
of the unitary transformed matrix E. Specifically,
ISTAIR(i) = +j if the first non-zero element E(i,j)
is a corner point and -j otherwise, for i = 1,2,...,M.
</PRE>
<B>Tolerances</B>
<PRE>
TOL DOUBLE PRECISION
A tolerance below which matrix elements are considered
to be zero. If the user sets TOL to be less than (or
equal to) zero then the tolerance is taken as
EPS * MAX(ABS(E(I,J))), where EPS is the machine
precision (see LAPACK Library routine DLAMCH),
I = 1,2,...,M and J = 1,2,...,N.
</PRE>
<B>Workspace</B>
<PRE>
DWORK DOUBLE PRECISION array, dimension (MAX(M,N))
</PRE>
<B>Error Indicator</B>
<PRE>
INFO INTEGER
= 0: successful exit;
< 0: if INFO = -i, the i-th argument had an illegal
value.
</PRE>
<A name="Method"><B><FONT SIZE="+1">Method</FONT></B></A>
<PRE>
Given an M-by-N matrix pencil sE-A with E not necessarily regular,
the routine computes a unitary transformed pencil Q'(sE-A)Z such
that the matrix Q' * E * Z is in column echelon form (trapezoidal
form). Further details can be found in [1].
[An M-by-N matrix E with rank(E) = r is said to be in column
echelon form if the following conditions are satisfied:
(a) the first (N - r) columns contain only zero elements; and
(b) if E(i(k),k) is the last nonzero element in column k for
k = N-r+1,...,N, i.e. E(i(k),k) <> 0 and E(j,k) = 0 for
j > i(k), then 1 <= i(N-r+1) < i(N-r+2) < ... < i(N) <= M.]
</PRE>
<A name="References"><B><FONT SIZE="+1">References</FONT></B></A>
<PRE>
[1] Beelen, Th. and Van Dooren, P.
An improved algorithm for the computation of Kronecker's
canonical form of a singular pencil.
Linear Algebra and Applications, 105, pp. 9-65, 1988.
</PRE>
<A name="Numerical Aspects"><B><FONT SIZE="+1">Numerical Aspects</FONT></B></A>
<PRE>
It is shown in [1] that the algorithm is numerically backward
stable. The operations count is proportional to (MAX(M,N))**3.
</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>
* MB04UD EXAMPLE PROGRAM TEXT
*
* .. Parameters ..
INTEGER NIN, NOUT
PARAMETER ( NIN = 5, NOUT = 6 )
INTEGER MMAX, NMAX
PARAMETER ( MMAX = 20, NMAX = 20 )
INTEGER LDA, LDE, LDQ, LDZ
PARAMETER ( LDA = MMAX, LDE = MMAX, LDQ = MMAX,
$ LDZ = NMAX )
INTEGER LDWORK
PARAMETER ( LDWORK = MAX( NMAX,MMAX ) )
* PARAMETER ( LDWORK = NMAX+MMAX )
* .. Local Scalars ..
DOUBLE PRECISION TOL
INTEGER I, INFO, J, M, N, RANKE
CHARACTER*1 JOBQ, JOBZ
* .. Local Arrays ..
DOUBLE PRECISION A(LDA,NMAX), DWORK(LDWORK), E(LDE,NMAX),
$ Q(LDQ,MMAX), Z(LDZ,NMAX)
INTEGER ISTAIR(MMAX)
* .. External Subroutines ..
EXTERNAL MB04UD
* .. 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 = * ) M, N, TOL
IF ( M.LT.0 .OR. M.GT.MMAX ) THEN
WRITE ( NOUT, FMT = 99993 ) M
ELSE IF ( N.LT.0 .OR. N.GT.NMAX ) THEN
WRITE ( NOUT, FMT = 99992 ) N
ELSE
READ ( NIN, FMT = * ) ( ( A(I,J), J = 1,N ), I = 1,M )
READ ( NIN, FMT = * ) ( ( E(I,J), J = 1,N ), I = 1,M )
JOBQ = 'N'
JOBZ = 'N'
* Reduce E to column echelon form and compute Q'*A*Z.
CALL MB04UD( JOBQ, JOBZ, M, N, A, LDA, E, LDE, Q, LDQ, Z, LDZ,
$ RANKE, ISTAIR, TOL, DWORK, INFO )
*
IF ( INFO.NE.0 ) THEN
WRITE ( NOUT, FMT = 99998 ) INFO
ELSE
WRITE ( NOUT, FMT = 99991 )
DO 10 I = 1, M
WRITE ( NOUT, FMT = 99996 ) ( A(I,J), J = 1,N )
10 CONTINUE
WRITE ( NOUT, FMT = 99997 )
DO 100 I = 1, M
WRITE ( NOUT, FMT = 99996 ) ( E(I,J), J = 1,N )
100 CONTINUE
WRITE ( NOUT, FMT = 99995 ) RANKE
WRITE ( NOUT, FMT = 99994 ) ( ISTAIR(I), I = 1,M )
END IF
END IF
STOP
*
99999 FORMAT (' MB04UD EXAMPLE PROGRAM RESULTS',/1X)
99998 FORMAT (' INFO on exit from MB04UD = ',I2)
99997 FORMAT (' The transformed matrix E is ')
99996 FORMAT (20(1X,F8.4))
99995 FORMAT (/' The computed rank of E = ',I2)
99994 FORMAT (/' ISTAIR is ',/20(1X,I5))
99993 FORMAT (/' M is out of range.',/' M = ',I5)
99992 FORMAT (/' N is out of range.',/' N = ',I5)
99991 FORMAT (' The transformed matrix A is ')
END
</PRE>
<B>Program Data</B>
<PRE>
MB04UD EXAMPLE PROGRAM DATA
4 4 0.0
2.0 0.0 2.0 -2.0
0.0 -2.0 0.0 2.0
2.0 0.0 -2.0 0.0
2.0 -2.0 0.0 2.0
1.0 0.0 1.0 -1.0
0.0 -1.0 0.0 1.0
1.0 0.0 -1.0 0.0
1.0 -1.0 0.0 1.0
</PRE>
<B>Program Results</B>
<PRE>
MB04UD EXAMPLE PROGRAM RESULTS
The transformed matrix A is
0.5164 1.0328 1.1547 -2.3094
0.0000 -2.5820 0.0000 -1.1547
0.0000 0.0000 -3.4641 0.0000
0.0000 0.0000 0.0000 -3.4641
The transformed matrix E is
0.2582 0.5164 0.5774 -1.1547
0.0000 -1.2910 0.0000 -0.5774
0.0000 0.0000 -1.7321 0.0000
0.0000 0.0000 0.0000 -1.7321
The computed rank of E = 4
ISTAIR is
1 2 3 4
</PRE>
<HR>
<p>
<A HREF=..\libindex.html><B>Return to index</B></A></BODY>
</HTML>
|
