LAPACK  3.11.0
LAPACK: Linear Algebra PACKage
cunbdb4.f
1 *> \brief \b CUNBDB4
2 *
3 * =========== DOCUMENTATION ===========
4 *
5 * Online html documentation available at
6 * http://www.netlib.org/lapack/explore-html/
7 *
8 *> \htmlonly
9 *> Download CUNBDB4 + dependencies
10 *> <a href="http://www.netlib.org/cgi-bin/netlibfiles.tgz?format=tgz&filename=/lapack/lapack_routine/cunbdb4.f">
11 *> [TGZ]</a>
12 *> <a href="http://www.netlib.org/cgi-bin/netlibfiles.zip?format=zip&filename=/lapack/lapack_routine/cunbdb4.f">
13 *> [ZIP]</a>
14 *> <a href="http://www.netlib.org/cgi-bin/netlibfiles.txt?format=txt&filename=/lapack/lapack_routine/cunbdb4.f">
15 *> [TXT]</a>
16 *> \endhtmlonly
17 *
18 * Definition:
19 * ===========
20 *
21 * SUBROUTINE CUNBDB4( M, P, Q, X11, LDX11, X21, LDX21, THETA, PHI,
22 * TAUP1, TAUP2, TAUQ1, PHANTOM, WORK, LWORK,
23 * INFO )
24 *
25 * .. Scalar Arguments ..
26 * INTEGER INFO, LWORK, M, P, Q, LDX11, LDX21
27 * ..
28 * .. Array Arguments ..
29 * REAL PHI(*), THETA(*)
30 * COMPLEX PHANTOM(*), TAUP1(*), TAUP2(*), TAUQ1(*),
31 * $ WORK(*), X11(LDX11,*), X21(LDX21,*)
32 * ..
33 *
34 *
35 *> \par Purpose:
36 * =============
37 *>
38 *>\verbatim
39 *>
40 *> CUNBDB4 simultaneously bidiagonalizes the blocks of a tall and skinny
41 *> matrix X with orthonomal columns:
42 *>
43 *> [ B11 ]
44 *> [ X11 ] [ P1 | ] [ 0 ]
45 *> [-----] = [---------] [-----] Q1**T .
46 *> [ X21 ] [ | P2 ] [ B21 ]
47 *> [ 0 ]
48 *>
49 *> X11 is P-by-Q, and X21 is (M-P)-by-Q. M-Q must be no larger than P,
50 *> M-P, or Q. Routines CUNBDB1, CUNBDB2, and CUNBDB3 handle cases in
51 *> which M-Q is not the minimum dimension.
52 *>
53 *> The unitary matrices P1, P2, and Q1 are P-by-P, (M-P)-by-(M-P),
54 *> and (M-Q)-by-(M-Q), respectively. They are represented implicitly by
55 *> Householder vectors.
56 *>
57 *> B11 and B12 are (M-Q)-by-(M-Q) bidiagonal matrices represented
58 *> implicitly by angles THETA, PHI.
59 *>
60 *>\endverbatim
61 *
62 * Arguments:
63 * ==========
64 *
65 *> \param[in] M
66 *> \verbatim
67 *> M is INTEGER
68 *> The number of rows X11 plus the number of rows in X21.
69 *> \endverbatim
70 *>
71 *> \param[in] P
72 *> \verbatim
73 *> P is INTEGER
74 *> The number of rows in X11. 0 <= P <= M.
75 *> \endverbatim
76 *>
77 *> \param[in] Q
78 *> \verbatim
79 *> Q is INTEGER
80 *> The number of columns in X11 and X21. 0 <= Q <= M and
81 *> M-Q <= min(P,M-P,Q).
82 *> \endverbatim
83 *>
84 *> \param[in,out] X11
85 *> \verbatim
86 *> X11 is COMPLEX array, dimension (LDX11,Q)
87 *> On entry, the top block of the matrix X to be reduced. On
88 *> exit, the columns of tril(X11) specify reflectors for P1 and
89 *> the rows of triu(X11,1) specify reflectors for Q1.
90 *> \endverbatim
91 *>
92 *> \param[in] LDX11
93 *> \verbatim
94 *> LDX11 is INTEGER
95 *> The leading dimension of X11. LDX11 >= P.
96 *> \endverbatim
97 *>
98 *> \param[in,out] X21
99 *> \verbatim
100 *> X21 is COMPLEX array, dimension (LDX21,Q)
101 *> On entry, the bottom block of the matrix X to be reduced. On
102 *> exit, the columns of tril(X21) specify reflectors for P2.
103 *> \endverbatim
104 *>
105 *> \param[in] LDX21
106 *> \verbatim
107 *> LDX21 is INTEGER
108 *> The leading dimension of X21. LDX21 >= M-P.
109 *> \endverbatim
110 *>
111 *> \param[out] THETA
112 *> \verbatim
113 *> THETA is REAL array, dimension (Q)
114 *> The entries of the bidiagonal blocks B11, B21 are defined by
115 *> THETA and PHI. See Further Details.
116 *> \endverbatim
117 *>
118 *> \param[out] PHI
119 *> \verbatim
120 *> PHI is REAL array, dimension (Q-1)
121 *> The entries of the bidiagonal blocks B11, B21 are defined by
122 *> THETA and PHI. See Further Details.
123 *> \endverbatim
124 *>
125 *> \param[out] TAUP1
126 *> \verbatim
127 *> TAUP1 is COMPLEX array, dimension (M-Q)
128 *> The scalar factors of the elementary reflectors that define
129 *> P1.
130 *> \endverbatim
131 *>
132 *> \param[out] TAUP2
133 *> \verbatim
134 *> TAUP2 is COMPLEX array, dimension (M-Q)
135 *> The scalar factors of the elementary reflectors that define
136 *> P2.
137 *> \endverbatim
138 *>
139 *> \param[out] TAUQ1
140 *> \verbatim
141 *> TAUQ1 is COMPLEX array, dimension (Q)
142 *> The scalar factors of the elementary reflectors that define
143 *> Q1.
144 *> \endverbatim
145 *>
146 *> \param[out] PHANTOM
147 *> \verbatim
148 *> PHANTOM is COMPLEX array, dimension (M)
149 *> The routine computes an M-by-1 column vector Y that is
150 *> orthogonal to the columns of [ X11; X21 ]. PHANTOM(1:P) and
151 *> PHANTOM(P+1:M) contain Householder vectors for Y(1:P) and
152 *> Y(P+1:M), respectively.
153 *> \endverbatim
154 *>
155 *> \param[out] WORK
156 *> \verbatim
157 *> WORK is COMPLEX array, dimension (LWORK)
158 *> \endverbatim
159 *>
160 *> \param[in] LWORK
161 *> \verbatim
162 *> LWORK is INTEGER
163 *> The dimension of the array WORK. LWORK >= M-Q.
164 *>
165 *> If LWORK = -1, then a workspace query is assumed; the routine
166 *> only calculates the optimal size of the WORK array, returns
167 *> this value as the first entry of the WORK array, and no error
168 *> message related to LWORK is issued by XERBLA.
169 *> \endverbatim
170 *>
171 *> \param[out] INFO
172 *> \verbatim
173 *> INFO is INTEGER
174 *> = 0: successful exit.
175 *> < 0: if INFO = -i, the i-th argument had an illegal value.
176 *> \endverbatim
177 *
178 * Authors:
179 * ========
180 *
181 *> \author Univ. of Tennessee
182 *> \author Univ. of California Berkeley
183 *> \author Univ. of Colorado Denver
184 *> \author NAG Ltd.
185 *
186 *> \ingroup unbdb4
187 *
188 *> \par Further Details:
189 * =====================
190 
191 *> \verbatim
192 *>
193 *> The upper-bidiagonal blocks B11, B21 are represented implicitly by
194 *> angles THETA(1), ..., THETA(Q) and PHI(1), ..., PHI(Q-1). Every entry
195 *> in each bidiagonal band is a product of a sine or cosine of a THETA
196 *> with a sine or cosine of a PHI. See [1] or CUNCSD for details.
197 *>
198 *> P1, P2, and Q1 are represented as products of elementary reflectors.
199 *> See CUNCSD2BY1 for details on generating P1, P2, and Q1 using CUNGQR
200 *> and CUNGLQ.
201 *> \endverbatim
202 *
203 *> \par References:
204 * ================
205 *>
206 *> [1] Brian D. Sutton. Computing the complete CS decomposition. Numer.
207 *> Algorithms, 50(1):33-65, 2009.
208 *>
209 * =====================================================================
210  SUBROUTINE cunbdb4( M, P, Q, X11, LDX11, X21, LDX21, THETA, PHI,
211  $ TAUP1, TAUP2, TAUQ1, PHANTOM, WORK, LWORK,
212  $ INFO )
213 *
214 * -- LAPACK computational routine --
215 * -- LAPACK is a software package provided by Univ. of Tennessee, --
216 * -- Univ. of California Berkeley, Univ. of Colorado Denver and NAG Ltd..--
217 *
218 * .. Scalar Arguments ..
219  INTEGER INFO, LWORK, M, P, Q, LDX11, LDX21
220 * ..
221 * .. Array Arguments ..
222  REAL PHI(*), THETA(*)
223  COMPLEX PHANTOM(*), TAUP1(*), TAUP2(*), TAUQ1(*),
224  $ work(*), x11(ldx11,*), x21(ldx21,*)
225 * ..
226 *
227 * ====================================================================
228 *
229 * .. Parameters ..
230  COMPLEX NEGONE, ONE, ZERO
231  parameter( negone = (-1.0e0,0.0e0), one = (1.0e0,0.0e0),
232  $ zero = (0.0e0,0.0e0) )
233 * ..
234 * .. Local Scalars ..
235  REAL C, S
236  INTEGER CHILDINFO, I, ILARF, IORBDB5, J, LLARF,
237  $ lorbdb5, lworkmin, lworkopt
238  LOGICAL LQUERY
239 * ..
240 * .. External Subroutines ..
241  EXTERNAL clarf, clarfgp, cunbdb5, csrot, cscal, clacgv,
242  $ xerbla
243 * ..
244 * .. External Functions ..
245  REAL SCNRM2
246  EXTERNAL scnrm2
247 * ..
248 * .. Intrinsic Function ..
249  INTRINSIC atan2, cos, max, sin, sqrt
250 * ..
251 * .. Executable Statements ..
252 *
253 * Test input arguments
254 *
255  info = 0
256  lquery = lwork .EQ. -1
257 *
258  IF( m .LT. 0 ) THEN
259  info = -1
260  ELSE IF( p .LT. m-q .OR. m-p .LT. m-q ) THEN
261  info = -2
262  ELSE IF( q .LT. m-q .OR. q .GT. m ) THEN
263  info = -3
264  ELSE IF( ldx11 .LT. max( 1, p ) ) THEN
265  info = -5
266  ELSE IF( ldx21 .LT. max( 1, m-p ) ) THEN
267  info = -7
268  END IF
269 *
270 * Compute workspace
271 *
272  IF( info .EQ. 0 ) THEN
273  ilarf = 2
274  llarf = max( q-1, p-1, m-p-1 )
275  iorbdb5 = 2
276  lorbdb5 = q
277  lworkopt = ilarf + llarf - 1
278  lworkopt = max( lworkopt, iorbdb5 + lorbdb5 - 1 )
279  lworkmin = lworkopt
280  work(1) = lworkopt
281  IF( lwork .LT. lworkmin .AND. .NOT.lquery ) THEN
282  info = -14
283  END IF
284  END IF
285  IF( info .NE. 0 ) THEN
286  CALL xerbla( 'CUNBDB4', -info )
287  RETURN
288  ELSE IF( lquery ) THEN
289  RETURN
290  END IF
291 *
292 * Reduce columns 1, ..., M-Q of X11 and X21
293 *
294  DO i = 1, m-q
295 *
296  IF( i .EQ. 1 ) THEN
297  DO j = 1, m
298  phantom(j) = zero
299  END DO
300  CALL cunbdb5( p, m-p, q, phantom(1), 1, phantom(p+1), 1,
301  $ x11, ldx11, x21, ldx21, work(iorbdb5),
302  $ lorbdb5, childinfo )
303  CALL cscal( p, negone, phantom(1), 1 )
304  CALL clarfgp( p, phantom(1), phantom(2), 1, taup1(1) )
305  CALL clarfgp( m-p, phantom(p+1), phantom(p+2), 1, taup2(1) )
306  theta(i) = atan2( REAL( PHANTOM(1) ), REAL( PHANTOM(P+1) ) )
307  c = cos( theta(i) )
308  s = sin( theta(i) )
309  phantom(1) = one
310  phantom(p+1) = one
311  CALL clarf( 'L', p, q, phantom(1), 1, conjg(taup1(1)), x11,
312  $ ldx11, work(ilarf) )
313  CALL clarf( 'L', m-p, q, phantom(p+1), 1, conjg(taup2(1)),
314  $ x21, ldx21, work(ilarf) )
315  ELSE
316  CALL cunbdb5( p-i+1, m-p-i+1, q-i+1, x11(i,i-1), 1,
317  $ x21(i,i-1), 1, x11(i,i), ldx11, x21(i,i),
318  $ ldx21, work(iorbdb5), lorbdb5, childinfo )
319  CALL cscal( p-i+1, negone, x11(i,i-1), 1 )
320  CALL clarfgp( p-i+1, x11(i,i-1), x11(i+1,i-1), 1, taup1(i) )
321  CALL clarfgp( m-p-i+1, x21(i,i-1), x21(i+1,i-1), 1,
322  $ taup2(i) )
323  theta(i) = atan2( REAL( X11(I,I-1) ), REAL( X21(I,I-1) ) )
324  c = cos( theta(i) )
325  s = sin( theta(i) )
326  x11(i,i-1) = one
327  x21(i,i-1) = one
328  CALL clarf( 'L', p-i+1, q-i+1, x11(i,i-1), 1,
329  $ conjg(taup1(i)), x11(i,i), ldx11, work(ilarf) )
330  CALL clarf( 'L', m-p-i+1, q-i+1, x21(i,i-1), 1,
331  $ conjg(taup2(i)), x21(i,i), ldx21, work(ilarf) )
332  END IF
333 *
334  CALL csrot( q-i+1, x11(i,i), ldx11, x21(i,i), ldx21, s, -c )
335  CALL clacgv( q-i+1, x21(i,i), ldx21 )
336  CALL clarfgp( q-i+1, x21(i,i), x21(i,i+1), ldx21, tauq1(i) )
337  c = REAL( X21(I,I) )
338  x21(i,i) = one
339  CALL clarf( 'R', p-i, q-i+1, x21(i,i), ldx21, tauq1(i),
340  $ x11(i+1,i), ldx11, work(ilarf) )
341  CALL clarf( 'R', m-p-i, q-i+1, x21(i,i), ldx21, tauq1(i),
342  $ x21(i+1,i), ldx21, work(ilarf) )
343  CALL clacgv( q-i+1, x21(i,i), ldx21 )
344  IF( i .LT. m-q ) THEN
345  s = sqrt( scnrm2( p-i, x11(i+1,i), 1 )**2
346  $ + scnrm2( m-p-i, x21(i+1,i), 1 )**2 )
347  phi(i) = atan2( s, c )
348  END IF
349 *
350  END DO
351 *
352 * Reduce the bottom-right portion of X11 to [ I 0 ]
353 *
354  DO i = m - q + 1, p
355  CALL clacgv( q-i+1, x11(i,i), ldx11 )
356  CALL clarfgp( q-i+1, x11(i,i), x11(i,i+1), ldx11, tauq1(i) )
357  x11(i,i) = one
358  CALL clarf( 'R', p-i, q-i+1, x11(i,i), ldx11, tauq1(i),
359  $ x11(i+1,i), ldx11, work(ilarf) )
360  CALL clarf( 'R', q-p, q-i+1, x11(i,i), ldx11, tauq1(i),
361  $ x21(m-q+1,i), ldx21, work(ilarf) )
362  CALL clacgv( q-i+1, x11(i,i), ldx11 )
363  END DO
364 *
365 * Reduce the bottom-right portion of X21 to [ 0 I ]
366 *
367  DO i = p + 1, q
368  CALL clacgv( q-i+1, x21(m-q+i-p,i), ldx21 )
369  CALL clarfgp( q-i+1, x21(m-q+i-p,i), x21(m-q+i-p,i+1), ldx21,
370  $ tauq1(i) )
371  x21(m-q+i-p,i) = one
372  CALL clarf( 'R', q-i, q-i+1, x21(m-q+i-p,i), ldx21, tauq1(i),
373  $ x21(m-q+i-p+1,i), ldx21, work(ilarf) )
374  CALL clacgv( q-i+1, x21(m-q+i-p,i), ldx21 )
375  END DO
376 *
377  RETURN
378 *
379 * End of CUNBDB4
380 *
381  END
382 
xerbla
subroutine xerbla(SRNAME, INFO)
XERBLA
Definition: xerbla.f:60
cscal
subroutine cscal(N, CA, CX, INCX)
CSCAL
Definition: cscal.f:78
clarf
subroutine clarf(SIDE, M, N, V, INCV, TAU, C, LDC, WORK)
CLARF applies an elementary reflector to a general rectangular matrix.
Definition: clarf.f:128
clarfgp
subroutine clarfgp(N, ALPHA, X, INCX, TAU)
CLARFGP generates an elementary reflector (Householder matrix) with non-negative beta.
Definition: clarfgp.f:104
csrot
subroutine csrot(N, CX, INCX, CY, INCY, C, S)
CSROT
Definition: csrot.f:98
cunbdb5
subroutine cunbdb5(M1, M2, N, X1, INCX1, X2, INCX2, Q1, LDQ1, Q2, LDQ2, WORK, LWORK, INFO)
CUNBDB5
Definition: cunbdb5.f:156
cunbdb4
subroutine cunbdb4(M, P, Q, X11, LDX11, X21, LDX21, THETA, PHI, TAUP1, TAUP2, TAUQ1, PHANTOM, WORK, LWORK, INFO)
CUNBDB4
Definition: cunbdb4.f:213
clacgv
subroutine clacgv(N, X, INCX)
CLACGV conjugates a complex vector.
Definition: clacgv.f:74