LAPACK 3.12.0
LAPACK: Linear Algebra PACKage
Loading...
Searching...
No Matches
dgeev.f
Go to the documentation of this file.
1*> \brief <b> DGEEV computes the eigenvalues and, optionally, the left and/or right eigenvectors for GE matrices</b>
2*
3* =========== DOCUMENTATION ===========
4*
5* Online html documentation available at
6* http://www.netlib.org/lapack/explore-html/
7*
8*> \htmlonly
9*> Download DGEEV + dependencies
10*> <a href="http://www.netlib.org/cgi-bin/netlibfiles.tgz?format=tgz&filename=/lapack/lapack_routine/dgeev.f">
11*> [TGZ]</a>
12*> <a href="http://www.netlib.org/cgi-bin/netlibfiles.zip?format=zip&filename=/lapack/lapack_routine/dgeev.f">
13*> [ZIP]</a>
14*> <a href="http://www.netlib.org/cgi-bin/netlibfiles.txt?format=txt&filename=/lapack/lapack_routine/dgeev.f">
15*> [TXT]</a>
16*> \endhtmlonly
17*
18* Definition:
19* ===========
20*
21* SUBROUTINE DGEEV( JOBVL, JOBVR, N, A, LDA, WR, WI, VL, LDVL, VR,
22* LDVR, WORK, LWORK, INFO )
23*
24* .. Scalar Arguments ..
25* CHARACTER JOBVL, JOBVR
26* INTEGER INFO, LDA, LDVL, LDVR, LWORK, N
27* ..
28* .. Array Arguments ..
29* DOUBLE PRECISION A( LDA, * ), VL( LDVL, * ), VR( LDVR, * ),
30* $ WI( * ), WORK( * ), WR( * )
31* ..
32*
33*
34*> \par Purpose:
35* =============
36*>
37*> \verbatim
38*>
39*> DGEEV computes for an N-by-N real nonsymmetric matrix A, the
40*> eigenvalues and, optionally, the left and/or right eigenvectors.
41*>
42*> The right eigenvector v(j) of A satisfies
43*> A * v(j) = lambda(j) * v(j)
44*> where lambda(j) is its eigenvalue.
45*> The left eigenvector u(j) of A satisfies
46*> u(j)**H * A = lambda(j) * u(j)**H
47*> where u(j)**H denotes the conjugate-transpose of u(j).
48*>
49*> The computed eigenvectors are normalized to have Euclidean norm
50*> equal to 1 and largest component real.
51*> \endverbatim
52*
53* Arguments:
54* ==========
55*
56*> \param[in] JOBVL
57*> \verbatim
58*> JOBVL is CHARACTER*1
59*> = 'N': left eigenvectors of A are not computed;
60*> = 'V': left eigenvectors of A are computed.
61*> \endverbatim
62*>
63*> \param[in] JOBVR
64*> \verbatim
65*> JOBVR is CHARACTER*1
66*> = 'N': right eigenvectors of A are not computed;
67*> = 'V': right eigenvectors of A are computed.
68*> \endverbatim
69*>
70*> \param[in] N
71*> \verbatim
72*> N is INTEGER
73*> The order of the matrix A. N >= 0.
74*> \endverbatim
75*>
76*> \param[in,out] A
77*> \verbatim
78*> A is DOUBLE PRECISION array, dimension (LDA,N)
79*> On entry, the N-by-N matrix A.
80*> On exit, A has been overwritten.
81*> \endverbatim
82*>
83*> \param[in] LDA
84*> \verbatim
85*> LDA is INTEGER
86*> The leading dimension of the array A. LDA >= max(1,N).
87*> \endverbatim
88*>
89*> \param[out] WR
90*> \verbatim
91*> WR is DOUBLE PRECISION array, dimension (N)
92*> \endverbatim
93*>
94*> \param[out] WI
95*> \verbatim
96*> WI is DOUBLE PRECISION array, dimension (N)
97*> WR and WI contain the real and imaginary parts,
98*> respectively, of the computed eigenvalues. Complex
99*> conjugate pairs of eigenvalues appear consecutively
100*> with the eigenvalue having the positive imaginary part
101*> first.
102*> \endverbatim
103*>
104*> \param[out] VL
105*> \verbatim
106*> VL is DOUBLE PRECISION array, dimension (LDVL,N)
107*> If JOBVL = 'V', the left eigenvectors u(j) are stored one
108*> after another in the columns of VL, in the same order
109*> as their eigenvalues.
110*> If JOBVL = 'N', VL is not referenced.
111*> If the j-th eigenvalue is real, then u(j) = VL(:,j),
112*> the j-th column of VL.
113*> If the j-th and (j+1)-st eigenvalues form a complex
114*> conjugate pair, then u(j) = VL(:,j) + i*VL(:,j+1) and
115*> u(j+1) = VL(:,j) - i*VL(:,j+1).
116*> \endverbatim
117*>
118*> \param[in] LDVL
119*> \verbatim
120*> LDVL is INTEGER
121*> The leading dimension of the array VL. LDVL >= 1; if
122*> JOBVL = 'V', LDVL >= N.
123*> \endverbatim
124*>
125*> \param[out] VR
126*> \verbatim
127*> VR is DOUBLE PRECISION array, dimension (LDVR,N)
128*> If JOBVR = 'V', the right eigenvectors v(j) are stored one
129*> after another in the columns of VR, in the same order
130*> as their eigenvalues.
131*> If JOBVR = 'N', VR is not referenced.
132*> If the j-th eigenvalue is real, then v(j) = VR(:,j),
133*> the j-th column of VR.
134*> If the j-th and (j+1)-st eigenvalues form a complex
135*> conjugate pair, then v(j) = VR(:,j) + i*VR(:,j+1) and
136*> v(j+1) = VR(:,j) - i*VR(:,j+1).
137*> \endverbatim
138*>
139*> \param[in] LDVR
140*> \verbatim
141*> LDVR is INTEGER
142*> The leading dimension of the array VR. LDVR >= 1; if
143*> JOBVR = 'V', LDVR >= N.
144*> \endverbatim
145*>
146*> \param[out] WORK
147*> \verbatim
148*> WORK is DOUBLE PRECISION array, dimension (MAX(1,LWORK))
149*> On exit, if INFO = 0, WORK(1) returns the optimal LWORK.
150*> \endverbatim
151*>
152*> \param[in] LWORK
153*> \verbatim
154*> LWORK is INTEGER
155*> The dimension of the array WORK. LWORK >= max(1,3*N), and
156*> if JOBVL = 'V' or JOBVR = 'V', LWORK >= 4*N. For good
157*> performance, LWORK must generally be larger.
158*>
159*> If LWORK = -1, then a workspace query is assumed; the routine
160*> only calculates the optimal size of the WORK array, returns
161*> this value as the first entry of the WORK array, and no error
162*> message related to LWORK is issued by XERBLA.
163*> \endverbatim
164*>
165*> \param[out] INFO
166*> \verbatim
167*> INFO is INTEGER
168*> = 0: successful exit
169*> < 0: if INFO = -i, the i-th argument had an illegal value.
170*> > 0: if INFO = i, the QR algorithm failed to compute all the
171*> eigenvalues, and no eigenvectors have been computed;
172*> elements i+1:N of WR and WI contain eigenvalues which
173*> have converged.
174*> \endverbatim
175*
176* Authors:
177* ========
178*
179*> \author Univ. of Tennessee
180*> \author Univ. of California Berkeley
181*> \author Univ. of Colorado Denver
182*> \author NAG Ltd.
183*
184*
185* @precisions fortran d -> s
186*
187*> \ingroup geev
188*
189* =====================================================================
190 SUBROUTINE dgeev( JOBVL, JOBVR, N, A, LDA, WR, WI, VL, LDVL, VR,
191 $ LDVR, WORK, LWORK, INFO )
192 implicit none
193*
194* -- LAPACK driver routine --
195* -- LAPACK is a software package provided by Univ. of Tennessee, --
196* -- Univ. of California Berkeley, Univ. of Colorado Denver and NAG Ltd..--
197*
198* .. Scalar Arguments ..
199 CHARACTER JOBVL, JOBVR
200 INTEGER INFO, LDA, LDVL, LDVR, LWORK, N
201* ..
202* .. Array Arguments ..
203 DOUBLE PRECISION A( LDA, * ), VL( LDVL, * ), VR( LDVR, * ),
204 $ wi( * ), work( * ), wr( * )
205* ..
206*
207* =====================================================================
208*
209* .. Parameters ..
210 DOUBLE PRECISION ZERO, ONE
211 parameter( zero = 0.0d0, one = 1.0d0 )
212* ..
213* .. Local Scalars ..
214 LOGICAL LQUERY, SCALEA, WANTVL, WANTVR
215 CHARACTER SIDE
216 INTEGER HSWORK, I, IBAL, IERR, IHI, ILO, ITAU, IWRK, K,
217 $ lwork_trevc, maxwrk, minwrk, nout
218 DOUBLE PRECISION ANRM, BIGNUM, CS, CSCALE, EPS, R, SCL, SMLNUM,
219 $ sn
220* ..
221* .. Local Arrays ..
222 LOGICAL SELECT( 1 )
223 DOUBLE PRECISION DUM( 1 )
224* ..
225* .. External Subroutines ..
226 EXTERNAL dgebak, dgebal, dgehrd, dhseqr, dlacpy, dlartg,
228* ..
229* .. External Functions ..
230 LOGICAL LSAME
231 INTEGER IDAMAX, ILAENV
232 DOUBLE PRECISION DLAMCH, DLANGE, DLAPY2, DNRM2
233 EXTERNAL lsame, idamax, ilaenv, dlamch, dlange, dlapy2,
234 $ dnrm2
235* ..
236* .. Intrinsic Functions ..
237 INTRINSIC max, sqrt
238* ..
239* .. Executable Statements ..
240*
241* Test the input arguments
242*
243 info = 0
244 lquery = ( lwork.EQ.-1 )
245 wantvl = lsame( jobvl, 'V' )
246 wantvr = lsame( jobvr, 'V' )
247 IF( ( .NOT.wantvl ) .AND. ( .NOT.lsame( jobvl, 'N' ) ) ) THEN
248 info = -1
249 ELSE IF( ( .NOT.wantvr ) .AND. ( .NOT.lsame( jobvr, 'N' ) ) ) THEN
250 info = -2
251 ELSE IF( n.LT.0 ) THEN
252 info = -3
253 ELSE IF( lda.LT.max( 1, n ) ) THEN
254 info = -5
255 ELSE IF( ldvl.LT.1 .OR. ( wantvl .AND. ldvl.LT.n ) ) THEN
256 info = -9
257 ELSE IF( ldvr.LT.1 .OR. ( wantvr .AND. ldvr.LT.n ) ) THEN
258 info = -11
259 END IF
260*
261* Compute workspace
262* (Note: Comments in the code beginning "Workspace:" describe the
263* minimal amount of workspace needed at that point in the code,
264* as well as the preferred amount for good performance.
265* NB refers to the optimal block size for the immediately
266* following subroutine, as returned by ILAENV.
267* HSWORK refers to the workspace preferred by DHSEQR, as
268* calculated below. HSWORK is computed assuming ILO=1 and IHI=N,
269* the worst case.)
270*
271 IF( info.EQ.0 ) THEN
272 IF( n.EQ.0 ) THEN
273 minwrk = 1
274 maxwrk = 1
275 ELSE
276 maxwrk = 2*n + n*ilaenv( 1, 'DGEHRD', ' ', n, 1, n, 0 )
277 IF( wantvl ) THEN
278 minwrk = 4*n
279 maxwrk = max( maxwrk, 2*n + ( n - 1 )*ilaenv( 1,
280 $ 'DORGHR', ' ', n, 1, n, -1 ) )
281 CALL dhseqr( 'S', 'V', n, 1, n, a, lda, wr, wi, vl, ldvl,
282 $ work, -1, info )
283 hswork = int( work(1) )
284 maxwrk = max( maxwrk, n + 1, n + hswork )
285 CALL dtrevc3( 'L', 'B', SELECT, n, a, lda,
286 $ vl, ldvl, vr, ldvr, n, nout,
287 $ work, -1, ierr )
288 lwork_trevc = int( work(1) )
289 maxwrk = max( maxwrk, n + lwork_trevc )
290 maxwrk = max( maxwrk, 4*n )
291 ELSE IF( wantvr ) THEN
292 minwrk = 4*n
293 maxwrk = max( maxwrk, 2*n + ( n - 1 )*ilaenv( 1,
294 $ 'DORGHR', ' ', n, 1, n, -1 ) )
295 CALL dhseqr( 'S', 'V', n, 1, n, a, lda, wr, wi, vr, ldvr,
296 $ work, -1, info )
297 hswork = int( work(1) )
298 maxwrk = max( maxwrk, n + 1, n + hswork )
299 CALL dtrevc3( 'R', 'B', SELECT, n, a, lda,
300 $ vl, ldvl, vr, ldvr, n, nout,
301 $ work, -1, ierr )
302 lwork_trevc = int( work(1) )
303 maxwrk = max( maxwrk, n + lwork_trevc )
304 maxwrk = max( maxwrk, 4*n )
305 ELSE
306 minwrk = 3*n
307 CALL dhseqr( 'E', 'N', n, 1, n, a, lda, wr, wi, vr, ldvr,
308 $ work, -1, info )
309 hswork = int( work(1) )
310 maxwrk = max( maxwrk, n + 1, n + hswork )
311 END IF
312 maxwrk = max( maxwrk, minwrk )
313 END IF
314 work( 1 ) = maxwrk
315*
316 IF( lwork.LT.minwrk .AND. .NOT.lquery ) THEN
317 info = -13
318 END IF
319 END IF
320*
321 IF( info.NE.0 ) THEN
322 CALL xerbla( 'DGEEV ', -info )
323 RETURN
324 ELSE IF( lquery ) THEN
325 RETURN
326 END IF
327*
328* Quick return if possible
329*
330 IF( n.EQ.0 )
331 $ RETURN
332*
333* Get machine constants
334*
335 eps = dlamch( 'P' )
336 smlnum = dlamch( 'S' )
337 bignum = one / smlnum
338 smlnum = sqrt( smlnum ) / eps
339 bignum = one / smlnum
340*
341* Scale A if max element outside range [SMLNUM,BIGNUM]
342*
343 anrm = dlange( 'M', n, n, a, lda, dum )
344 scalea = .false.
345 IF( anrm.GT.zero .AND. anrm.LT.smlnum ) THEN
346 scalea = .true.
347 cscale = smlnum
348 ELSE IF( anrm.GT.bignum ) THEN
349 scalea = .true.
350 cscale = bignum
351 END IF
352 IF( scalea )
353 $ CALL dlascl( 'G', 0, 0, anrm, cscale, n, n, a, lda, ierr )
354*
355* Balance the matrix
356* (Workspace: need N)
357*
358 ibal = 1
359 CALL dgebal( 'B', n, a, lda, ilo, ihi, work( ibal ), ierr )
360*
361* Reduce to upper Hessenberg form
362* (Workspace: need 3*N, prefer 2*N+N*NB)
363*
364 itau = ibal + n
365 iwrk = itau + n
366 CALL dgehrd( n, ilo, ihi, a, lda, work( itau ), work( iwrk ),
367 $ lwork-iwrk+1, ierr )
368*
369 IF( wantvl ) THEN
370*
371* Want left eigenvectors
372* Copy Householder vectors to VL
373*
374 side = 'L'
375 CALL dlacpy( 'L', n, n, a, lda, vl, ldvl )
376*
377* Generate orthogonal matrix in VL
378* (Workspace: need 3*N-1, prefer 2*N+(N-1)*NB)
379*
380 CALL dorghr( n, ilo, ihi, vl, ldvl, work( itau ), work( iwrk ),
381 $ lwork-iwrk+1, ierr )
382*
383* Perform QR iteration, accumulating Schur vectors in VL
384* (Workspace: need N+1, prefer N+HSWORK (see comments) )
385*
386 iwrk = itau
387 CALL dhseqr( 'S', 'V', n, ilo, ihi, a, lda, wr, wi, vl, ldvl,
388 $ work( iwrk ), lwork-iwrk+1, info )
389*
390 IF( wantvr ) THEN
391*
392* Want left and right eigenvectors
393* Copy Schur vectors to VR
394*
395 side = 'B'
396 CALL dlacpy( 'F', n, n, vl, ldvl, vr, ldvr )
397 END IF
398*
399 ELSE IF( wantvr ) THEN
400*
401* Want right eigenvectors
402* Copy Householder vectors to VR
403*
404 side = 'R'
405 CALL dlacpy( 'L', n, n, a, lda, vr, ldvr )
406*
407* Generate orthogonal matrix in VR
408* (Workspace: need 3*N-1, prefer 2*N+(N-1)*NB)
409*
410 CALL dorghr( n, ilo, ihi, vr, ldvr, work( itau ), work( iwrk ),
411 $ lwork-iwrk+1, ierr )
412*
413* Perform QR iteration, accumulating Schur vectors in VR
414* (Workspace: need N+1, prefer N+HSWORK (see comments) )
415*
416 iwrk = itau
417 CALL dhseqr( 'S', 'V', n, ilo, ihi, a, lda, wr, wi, vr, ldvr,
418 $ work( iwrk ), lwork-iwrk+1, info )
419*
420 ELSE
421*
422* Compute eigenvalues only
423* (Workspace: need N+1, prefer N+HSWORK (see comments) )
424*
425 iwrk = itau
426 CALL dhseqr( 'E', 'N', n, ilo, ihi, a, lda, wr, wi, vr, ldvr,
427 $ work( iwrk ), lwork-iwrk+1, info )
428 END IF
429*
430* If INFO .NE. 0 from DHSEQR, then quit
431*
432 IF( info.NE.0 )
433 $ GO TO 50
434*
435 IF( wantvl .OR. wantvr ) THEN
436*
437* Compute left and/or right eigenvectors
438* (Workspace: need 4*N, prefer N + N + 2*N*NB)
439*
440 CALL dtrevc3( side, 'B', SELECT, n, a, lda, vl, ldvl, vr, ldvr,
441 $ n, nout, work( iwrk ), lwork-iwrk+1, ierr )
442 END IF
443*
444 IF( wantvl ) THEN
445*
446* Undo balancing of left eigenvectors
447* (Workspace: need N)
448*
449 CALL dgebak( 'B', 'L', n, ilo, ihi, work( ibal ), n, vl, ldvl,
450 $ ierr )
451*
452* Normalize left eigenvectors and make largest component real
453*
454 DO 20 i = 1, n
455 IF( wi( i ).EQ.zero ) THEN
456 scl = one / dnrm2( n, vl( 1, i ), 1 )
457 CALL dscal( n, scl, vl( 1, i ), 1 )
458 ELSE IF( wi( i ).GT.zero ) THEN
459 scl = one / dlapy2( dnrm2( n, vl( 1, i ), 1 ),
460 $ dnrm2( n, vl( 1, i+1 ), 1 ) )
461 CALL dscal( n, scl, vl( 1, i ), 1 )
462 CALL dscal( n, scl, vl( 1, i+1 ), 1 )
463 DO 10 k = 1, n
464 work( iwrk+k-1 ) = vl( k, i )**2 + vl( k, i+1 )**2
465 10 CONTINUE
466 k = idamax( n, work( iwrk ), 1 )
467 CALL dlartg( vl( k, i ), vl( k, i+1 ), cs, sn, r )
468 CALL drot( n, vl( 1, i ), 1, vl( 1, i+1 ), 1, cs, sn )
469 vl( k, i+1 ) = zero
470 END IF
471 20 CONTINUE
472 END IF
473*
474 IF( wantvr ) THEN
475*
476* Undo balancing of right eigenvectors
477* (Workspace: need N)
478*
479 CALL dgebak( 'B', 'R', n, ilo, ihi, work( ibal ), n, vr, ldvr,
480 $ ierr )
481*
482* Normalize right eigenvectors and make largest component real
483*
484 DO 40 i = 1, n
485 IF( wi( i ).EQ.zero ) THEN
486 scl = one / dnrm2( n, vr( 1, i ), 1 )
487 CALL dscal( n, scl, vr( 1, i ), 1 )
488 ELSE IF( wi( i ).GT.zero ) THEN
489 scl = one / dlapy2( dnrm2( n, vr( 1, i ), 1 ),
490 $ dnrm2( n, vr( 1, i+1 ), 1 ) )
491 CALL dscal( n, scl, vr( 1, i ), 1 )
492 CALL dscal( n, scl, vr( 1, i+1 ), 1 )
493 DO 30 k = 1, n
494 work( iwrk+k-1 ) = vr( k, i )**2 + vr( k, i+1 )**2
495 30 CONTINUE
496 k = idamax( n, work( iwrk ), 1 )
497 CALL dlartg( vr( k, i ), vr( k, i+1 ), cs, sn, r )
498 CALL drot( n, vr( 1, i ), 1, vr( 1, i+1 ), 1, cs, sn )
499 vr( k, i+1 ) = zero
500 END IF
501 40 CONTINUE
502 END IF
503*
504* Undo scaling if necessary
505*
506 50 CONTINUE
507 IF( scalea ) THEN
508 CALL dlascl( 'G', 0, 0, cscale, anrm, n-info, 1, wr( info+1 ),
509 $ max( n-info, 1 ), ierr )
510 CALL dlascl( 'G', 0, 0, cscale, anrm, n-info, 1, wi( info+1 ),
511 $ max( n-info, 1 ), ierr )
512 IF( info.GT.0 ) THEN
513 CALL dlascl( 'G', 0, 0, cscale, anrm, ilo-1, 1, wr, n,
514 $ ierr )
515 CALL dlascl( 'G', 0, 0, cscale, anrm, ilo-1, 1, wi, n,
516 $ ierr )
517 END IF
518 END IF
519*
520 work( 1 ) = maxwrk
521 RETURN
522*
523* End of DGEEV
524*
525 END
subroutine xerbla(srname, info)
Definition cblat2.f:3285
subroutine dgebak(job, side, n, ilo, ihi, scale, m, v, ldv, info)
DGEBAK
Definition dgebak.f:130
subroutine dgebal(job, n, a, lda, ilo, ihi, scale, info)
DGEBAL
Definition dgebal.f:163
subroutine dgeev(jobvl, jobvr, n, a, lda, wr, wi, vl, ldvl, vr, ldvr, work, lwork, info)
DGEEV computes the eigenvalues and, optionally, the left and/or right eigenvectors for GE matrices
Definition dgeev.f:192
subroutine dgehrd(n, ilo, ihi, a, lda, tau, work, lwork, info)
DGEHRD
Definition dgehrd.f:167
subroutine dhseqr(job, compz, n, ilo, ihi, h, ldh, wr, wi, z, ldz, work, lwork, info)
DHSEQR
Definition dhseqr.f:316
subroutine dlacpy(uplo, m, n, a, lda, b, ldb)
DLACPY copies all or part of one two-dimensional array to another.
Definition dlacpy.f:103
subroutine dlartg(f, g, c, s, r)
DLARTG generates a plane rotation with real cosine and real sine.
Definition dlartg.f90:111
subroutine dlascl(type, kl, ku, cfrom, cto, m, n, a, lda, info)
DLASCL multiplies a general rectangular matrix by a real scalar defined as cto/cfrom.
Definition dlascl.f:143
subroutine drot(n, dx, incx, dy, incy, c, s)
DROT
Definition drot.f:92
subroutine dscal(n, da, dx, incx)
DSCAL
Definition dscal.f:79
subroutine dtrevc3(side, howmny, select, n, t, ldt, vl, ldvl, vr, ldvr, mm, m, work, lwork, info)
DTREVC3
Definition dtrevc3.f:237
subroutine dorghr(n, ilo, ihi, a, lda, tau, work, lwork, info)
DORGHR
Definition dorghr.f:126