1 *> \brief \b SLAQPS computes a step of QR factorization with column pivoting of a real m-by-n matrix A by using BLAS level 3.
3 * =========== DOCUMENTATION ===========
5 * Online html documentation available at
6 * http://www.netlib.org/lapack/explore-html/
9 *> Download SLAQPS + dependencies
10 *> <a href="http://www.netlib.org/cgi-bin/netlibfiles.tgz?format=tgz&filename=/lapack/lapack_routine/slaqps.f">
12 *> <a href="http://www.netlib.org/cgi-bin/netlibfiles.zip?format=zip&filename=/lapack/lapack_routine/slaqps.f">
14 *> <a href="http://www.netlib.org/cgi-bin/netlibfiles.txt?format=txt&filename=/lapack/lapack_routine/slaqps.f">
21 * SUBROUTINE SLAQPS( M, N, OFFSET, NB, KB, A, LDA, JPVT, TAU, VN1,
24 * .. Scalar Arguments ..
25 * INTEGER KB, LDA, LDF, M, N, NB, OFFSET
27 * .. Array Arguments ..
29 * REAL A( LDA, * ), AUXV( * ), F( LDF, * ), TAU( * ),
30 * $ VN1( * ), VN2( * )
39 *> SLAQPS computes a step of QR factorization with column pivoting
40 *> of a real M-by-N matrix A by using Blas-3. It tries to factorize
41 *> NB columns from A starting from the row OFFSET+1, and updates all
42 *> of the matrix with Blas-3 xGEMM.
44 *> In some cases, due to catastrophic cancellations, it cannot
45 *> factorize NB columns. Hence, the actual number of factorized
46 *> columns is returned in KB.
48 *> Block A(1:OFFSET,1:N) is accordingly pivoted, but not factorized.
57 *> The number of rows of the matrix A. M >= 0.
63 *> The number of columns of the matrix A. N >= 0
69 *> The number of rows of A that have been factorized in
76 *> The number of columns to factorize.
82 *> The number of columns actually factorized.
87 *> A is REAL array, dimension (LDA,N)
88 *> On entry, the M-by-N matrix A.
89 *> On exit, block A(OFFSET+1:M,1:KB) is the triangular
90 *> factor obtained and block A(1:OFFSET,1:N) has been
91 *> accordingly pivoted, but no factorized.
92 *> The rest of the matrix, block A(OFFSET+1:M,KB+1:N) has
99 *> The leading dimension of the array A. LDA >= max(1,M).
102 *> \param[in,out] JPVT
104 *> JPVT is INTEGER array, dimension (N)
105 *> JPVT(I) = K <==> Column K of the full matrix A has been
106 *> permuted into position I in AP.
111 *> TAU is REAL array, dimension (KB)
112 *> The scalar factors of the elementary reflectors.
115 *> \param[in,out] VN1
117 *> VN1 is REAL array, dimension (N)
118 *> The vector with the partial column norms.
121 *> \param[in,out] VN2
123 *> VN2 is REAL array, dimension (N)
124 *> The vector with the exact column norms.
127 *> \param[in,out] AUXV
129 *> AUXV is REAL array, dimension (NB)
135 *> F is REAL array, dimension (LDF,NB)
136 *> Matrix F**T = L*Y**T*A.
142 *> The leading dimension of the array F. LDF >= max(1,N).
148 *> \author Univ. of Tennessee
149 *> \author Univ. of California Berkeley
150 *> \author Univ. of Colorado Denver
153 *> \date September 2012
155 *> \ingroup realOTHERauxiliary
157 *> \par Contributors:
160 *> G. Quintana-Orti, Depto. de Informatica, Universidad Jaime I, Spain
161 *> X. Sun, Computer Science Dept., Duke University, USA
164 *> Partial column norm updating strategy modified on April 2011
165 *> Z. Drmac and Z. Bujanovic, Dept. of Mathematics,
166 *> University of Zagreb, Croatia.
171 *> LAPACK Working Note 176
174 *> <a href="http://www.netlib.org/lapack/lawnspdf/lawn176.pdf">[PDF]</a>
177 * =====================================================================
178 SUBROUTINE SLAQPS( M, N, OFFSET, NB, KB, A, LDA, JPVT, TAU, VN1,
179 $ VN2, AUXV, F, LDF )
181 * -- LAPACK auxiliary routine (version 3.4.2) --
182 * -- LAPACK is a software package provided by Univ. of Tennessee, --
183 * -- Univ. of California Berkeley, Univ. of Colorado Denver and NAG Ltd..--
186 * .. Scalar Arguments ..
187 INTEGER KB, LDA, LDF, M, N, NB, OFFSET
189 * .. Array Arguments ..
191 REAL A( LDA, * ), AUXV( * ), F( LDF, * ), TAU( * ),
195 * =====================================================================
199 PARAMETER ( ZERO = 0.0E+0, ONE = 1.0E+0 )
201 * .. Local Scalars ..
202 INTEGER ITEMP, J, K, LASTRK, LSTICC, PVT, RK
203 REAL AKK, TEMP, TEMP2, TOL3Z
205 * .. External Subroutines ..
206 EXTERNAL SGEMM, SGEMV, SLARFG, SSWAP
208 * .. Intrinsic Functions ..
209 INTRINSIC ABS, MAX, MIN, NINT, REAL, SQRT
211 * .. External Functions ..
214 EXTERNAL ISAMAX, SLAMCH, SNRM2
216 * .. Executable Statements ..
218 LASTRK = MIN( M, N+OFFSET )
221 TOL3Z = SQRT(SLAMCH('Epsilon'))
223 * Beginning of while loop.
226 IF( ( K.LT.NB ) .AND. ( LSTICC.EQ.0 ) ) THEN
230 * Determine ith pivot column and swap if necessary
232 PVT = ( K-1 ) + ISAMAX( N-K+1, VN1( K ), 1 )
234 CALL SSWAP( M, A( 1, PVT ), 1, A( 1, K ), 1 )
235 CALL SSWAP( K-1, F( PVT, 1 ), LDF, F( K, 1 ), LDF )
237 JPVT( PVT ) = JPVT( K )
239 VN1( PVT ) = VN1( K )
240 VN2( PVT ) = VN2( K )
243 * Apply previous Householder reflectors to column K:
244 * A(RK:M,K) := A(RK:M,K) - A(RK:M,1:K-1)*F(K,1:K-1)**T.
247 CALL SGEMV( 'No transpose', M-RK+1, K-1, -ONE, A( RK, 1 ),
248 $ LDA, F( K, 1 ), LDF, ONE, A( RK, K ), 1 )
251 * Generate elementary reflector H(k).
254 CALL SLARFG( M-RK+1, A( RK, K ), A( RK+1, K ), 1, TAU( K ) )
256 CALL SLARFG( 1, A( RK, K ), A( RK, K ), 1, TAU( K ) )
262 * Compute Kth column of F:
264 * Compute F(K+1:N,K) := tau(K)*A(RK:M,K+1:N)**T*A(RK:M,K).
267 CALL SGEMV( 'Transpose', M-RK+1, N-K, TAU( K ),
268 $ A( RK, K+1 ), LDA, A( RK, K ), 1, ZERO,
272 * Padding F(1:K,K) with zeros.
278 * Incremental updating of F:
279 * F(1:N,K) := F(1:N,K) - tau(K)*F(1:N,1:K-1)*A(RK:M,1:K-1)**T
283 CALL SGEMV( 'Transpose', M-RK+1, K-1, -TAU( K ), A( RK, 1 ),
284 $ LDA, A( RK, K ), 1, ZERO, AUXV( 1 ), 1 )
286 CALL SGEMV( 'No transpose', N, K-1, ONE, F( 1, 1 ), LDF,
287 $ AUXV( 1 ), 1, ONE, F( 1, K ), 1 )
290 * Update the current row of A:
291 * A(RK,K+1:N) := A(RK,K+1:N) - A(RK,1:K)*F(K+1:N,1:K)**T.
294 CALL SGEMV( 'No transpose', N-K, K, -ONE, F( K+1, 1 ), LDF,
295 $ A( RK, 1 ), LDA, ONE, A( RK, K+1 ), LDA )
298 * Update partial column norms.
300 IF( RK.LT.LASTRK ) THEN
302 IF( VN1( J ).NE.ZERO ) THEN
304 * NOTE: The following 4 lines follow from the analysis in
305 * Lapack Working Note 176.
307 TEMP = ABS( A( RK, J ) ) / VN1( J )
308 TEMP = MAX( ZERO, ( ONE+TEMP )*( ONE-TEMP ) )
309 TEMP2 = TEMP*( VN1( J ) / VN2( J ) )**2
310 IF( TEMP2 .LE. TOL3Z ) THEN
311 VN2( J ) = REAL( LSTICC )
314 VN1( J ) = VN1( J )*SQRT( TEMP )
329 * Apply the block reflector to the rest of the matrix:
330 * A(OFFSET+KB+1:M,KB+1:N) := A(OFFSET+KB+1:M,KB+1:N) -
331 * A(OFFSET+KB+1:M,1:KB)*F(KB+1:N,1:KB)**T.
333 IF( KB.LT.MIN( N, M-OFFSET ) ) THEN
334 CALL SGEMM( 'No transpose', 'Transpose', M-RK, N-KB, KB, -ONE,
335 $ A( RK+1, 1 ), LDA, F( KB+1, 1 ), LDF, ONE,
336 $ A( RK+1, KB+1 ), LDA )
339 * Recomputation of difficult columns.
342 IF( LSTICC.GT.0 ) THEN
343 ITEMP = NINT( VN2( LSTICC ) )
344 VN1( LSTICC ) = SNRM2( M-RK, A( RK+1, LSTICC ), 1 )
346 * NOTE: The computation of VN1( LSTICC ) relies on the fact that
347 * SNRM2 does not fail on vectors with norm below the value of
350 VN2( LSTICC ) = VN1( LSTICC )