1 *> \brief <b> SSYEVD computes the eigenvalues and, optionally, the left and/or right eigenvectors for SY matrices</b>
3 * =========== DOCUMENTATION ===========
5 * Online html documentation available at
6 * http://www.netlib.org/lapack/explore-html/
9 *> Download SSYEVD + dependencies
10 *> <a href="http://www.netlib.org/cgi-bin/netlibfiles.tgz?format=tgz&filename=/lapack/lapack_routine/ssyevd.f">
12 *> <a href="http://www.netlib.org/cgi-bin/netlibfiles.zip?format=zip&filename=/lapack/lapack_routine/ssyevd.f">
14 *> <a href="http://www.netlib.org/cgi-bin/netlibfiles.txt?format=txt&filename=/lapack/lapack_routine/ssyevd.f">
21 * SUBROUTINE SSYEVD( JOBZ, UPLO, N, A, LDA, W, WORK, LWORK, IWORK,
24 * .. Scalar Arguments ..
25 * CHARACTER JOBZ, UPLO
26 * INTEGER INFO, LDA, LIWORK, LWORK, N
28 * .. Array Arguments ..
30 * REAL A( LDA, * ), W( * ), WORK( * )
39 *> SSYEVD computes all eigenvalues and, optionally, eigenvectors of a
40 *> real symmetric matrix A. If eigenvectors are desired, it uses a
41 *> divide and conquer algorithm.
43 *> The divide and conquer algorithm makes very mild assumptions about
44 *> floating point arithmetic. It will work on machines with a guard
45 *> digit in add/subtract, or on those binary machines without guard
46 *> digits which subtract like the Cray X-MP, Cray Y-MP, Cray C-90, or
47 *> Cray-2. It could conceivably fail on hexadecimal or decimal machines
48 *> without guard digits, but we know of none.
50 *> Because of large use of BLAS of level 3, SSYEVD needs N**2 more
51 *> workspace than SSYEVX.
59 *> JOBZ is CHARACTER*1
60 *> = 'N': Compute eigenvalues only;
61 *> = 'V': Compute eigenvalues and eigenvectors.
66 *> UPLO is CHARACTER*1
67 *> = 'U': Upper triangle of A is stored;
68 *> = 'L': Lower triangle of A is stored.
74 *> The order of the matrix A. N >= 0.
79 *> A is REAL array, dimension (LDA, N)
80 *> On entry, the symmetric matrix A. If UPLO = 'U', the
81 *> leading N-by-N upper triangular part of A contains the
82 *> upper triangular part of the matrix A. If UPLO = 'L',
83 *> the leading N-by-N lower triangular part of A contains
84 *> the lower triangular part of the matrix A.
85 *> On exit, if JOBZ = 'V', then if INFO = 0, A contains the
86 *> orthonormal eigenvectors of the matrix A.
87 *> If JOBZ = 'N', then on exit the lower triangle (if UPLO='L')
88 *> or the upper triangle (if UPLO='U') of A, including the
89 *> diagonal, is destroyed.
95 *> The leading dimension of the array A. LDA >= max(1,N).
100 *> W is REAL array, dimension (N)
101 *> If INFO = 0, the eigenvalues in ascending order.
106 *> WORK is REAL array,
108 *> On exit, if INFO = 0, WORK(1) returns the optimal LWORK.
114 *> The dimension of the array WORK.
115 *> If N <= 1, LWORK must be at least 1.
116 *> If JOBZ = 'N' and N > 1, LWORK must be at least 2*N+1.
117 *> If JOBZ = 'V' and N > 1, LWORK must be at least
120 *> If LWORK = -1, then a workspace query is assumed; the routine
121 *> only calculates the optimal sizes of the WORK and IWORK
122 *> arrays, returns these values as the first entries of the WORK
123 *> and IWORK arrays, and no error message related to LWORK or
124 *> LIWORK is issued by XERBLA.
129 *> IWORK is INTEGER array, dimension (MAX(1,LIWORK))
130 *> On exit, if INFO = 0, IWORK(1) returns the optimal LIWORK.
136 *> The dimension of the array IWORK.
137 *> If N <= 1, LIWORK must be at least 1.
138 *> If JOBZ = 'N' and N > 1, LIWORK must be at least 1.
139 *> If JOBZ = 'V' and N > 1, LIWORK must be at least 3 + 5*N.
141 *> If LIWORK = -1, then a workspace query is assumed; the
142 *> routine only calculates the optimal sizes of the WORK and
143 *> IWORK arrays, returns these values as the first entries of
144 *> the WORK and IWORK arrays, and no error message related to
145 *> LWORK or LIWORK is issued by XERBLA.
151 *> = 0: successful exit
152 *> < 0: if INFO = -i, the i-th argument had an illegal value
153 *> > 0: if INFO = i and JOBZ = 'N', then the algorithm failed
154 *> to converge; i off-diagonal elements of an intermediate
155 *> tridiagonal form did not converge to zero;
156 *> if INFO = i and JOBZ = 'V', then the algorithm failed
157 *> to compute an eigenvalue while working on the submatrix
158 *> lying in rows and columns INFO/(N+1) through
165 *> \author Univ. of Tennessee
166 *> \author Univ. of California Berkeley
167 *> \author Univ. of Colorado Denver
170 *> \date November 2011
172 *> \ingroup realSYeigen
174 *> \par Contributors:
177 *> Jeff Rutter, Computer Science Division, University of California
178 *> at Berkeley, USA \n
179 *> Modified by Francoise Tisseur, University of Tennessee \n
180 *> Modified description of INFO. Sven, 16 Feb 05. \n
182 * =====================================================================
183 SUBROUTINE SSYEVD( JOBZ, UPLO, N, A, LDA, W, WORK, LWORK, IWORK,
186 * -- LAPACK driver routine (version 3.4.0) --
187 * -- LAPACK is a software package provided by Univ. of Tennessee, --
188 * -- Univ. of California Berkeley, Univ. of Colorado Denver and NAG Ltd..--
191 * .. Scalar Arguments ..
193 INTEGER INFO, LDA, LIWORK, LWORK, N
195 * .. Array Arguments ..
197 REAL A( LDA, * ), W( * ), WORK( * )
200 * =====================================================================
204 PARAMETER ( ZERO = 0.0E+0, ONE = 1.0E+0 )
206 * .. Local Scalars ..
208 LOGICAL LOWER, LQUERY, WANTZ
209 INTEGER IINFO, INDE, INDTAU, INDWK2, INDWRK, ISCALE,
210 $ LIOPT, LIWMIN, LLWORK, LLWRK2, LOPT, LWMIN
211 REAL ANRM, BIGNUM, EPS, RMAX, RMIN, SAFMIN, SIGMA,
214 * .. External Functions ..
218 EXTERNAL ILAENV, LSAME, SLAMCH, SLANSY
220 * .. External Subroutines ..
221 EXTERNAL SLACPY, SLASCL, SORMTR, SSCAL, SSTEDC, SSTERF,
224 * .. Intrinsic Functions ..
227 * .. Executable Statements ..
229 * Test the input parameters.
231 WANTZ = LSAME( JOBZ, 'V' )
232 LOWER = LSAME( UPLO, 'L' )
233 LQUERY = ( LWORK.EQ.-1 .OR. LIWORK.EQ.-1 )
236 IF( .NOT.( WANTZ .OR. LSAME( JOBZ, 'N' ) ) ) THEN
238 ELSE IF( .NOT.( LOWER .OR. LSAME( UPLO, 'U' ) ) ) THEN
240 ELSE IF( N.LT.0 ) THEN
242 ELSE IF( LDA.LT.MAX( 1, N ) ) THEN
255 LWMIN = 1 + 6*N + 2*N**2
260 LOPT = MAX( LWMIN, 2*N +
261 $ ILAENV( 1, 'SSYTRD', UPLO, N, -1, -1, -1 ) )
267 IF( LWORK.LT.LWMIN .AND. .NOT.LQUERY ) THEN
269 ELSE IF( LIWORK.LT.LIWMIN .AND. .NOT.LQUERY ) THEN
275 CALL XERBLA( 'SSYEVD', -INFO )
277 ELSE IF( LQUERY ) THEN
281 * Quick return if possible
293 * Get machine constants.
295 SAFMIN = SLAMCH( 'Safe minimum' )
296 EPS = SLAMCH( 'Precision' )
297 SMLNUM = SAFMIN / EPS
298 BIGNUM = ONE / SMLNUM
299 RMIN = SQRT( SMLNUM )
300 RMAX = SQRT( BIGNUM )
302 * Scale matrix to allowable range, if necessary.
304 ANRM = SLANSY( 'M', UPLO, N, A, LDA, WORK )
306 IF( ANRM.GT.ZERO .AND. ANRM.LT.RMIN ) THEN
309 ELSE IF( ANRM.GT.RMAX ) THEN
314 $ CALL SLASCL( UPLO, 0, 0, ONE, SIGMA, N, N, A, LDA, INFO )
316 * Call SSYTRD to reduce symmetric matrix to tridiagonal form.
321 LLWORK = LWORK - INDWRK + 1
322 INDWK2 = INDWRK + N*N
323 LLWRK2 = LWORK - INDWK2 + 1
325 CALL SSYTRD( UPLO, N, A, LDA, W, WORK( INDE ), WORK( INDTAU ),
326 $ WORK( INDWRK ), LLWORK, IINFO )
328 * For eigenvalues only, call SSTERF. For eigenvectors, first call
329 * SSTEDC to generate the eigenvector matrix, WORK(INDWRK), of the
330 * tridiagonal matrix, then call SORMTR to multiply it by the
331 * Householder transformations stored in A.
333 IF( .NOT.WANTZ ) THEN
334 CALL SSTERF( N, W, WORK( INDE ), INFO )
336 CALL SSTEDC( 'I', N, W, WORK( INDE ), WORK( INDWRK ), N,
337 $ WORK( INDWK2 ), LLWRK2, IWORK, LIWORK, INFO )
338 CALL SORMTR( 'L', UPLO, 'N', N, N, A, LDA, WORK( INDTAU ),
339 $ WORK( INDWRK ), N, WORK( INDWK2 ), LLWRK2, IINFO )
340 CALL SLACPY( 'A', N, N, WORK( INDWRK ), N, A, LDA )
343 * If matrix was scaled, then rescale eigenvalues appropriately.
346 $ CALL SSCAL( N, ONE / SIGMA, W, 1 )