| 1 |
SUBROUTINE ZGEES( JOBVS, SORT, SELECT, N, A, LDA, SDIM, W, VS, |
| 2 |
$ LDVS, WORK, LWORK, RWORK, BWORK, INFO ) |
| 3 |
* |
| 4 |
* -- LAPACK driver routine (version 3.2) -- |
| 5 |
* -- LAPACK is a software package provided by Univ. of Tennessee, -- |
| 6 |
* -- Univ. of California Berkeley, Univ. of Colorado Denver and NAG Ltd..-- |
| 7 |
* November 2006 |
| 8 |
* |
| 9 |
* .. Scalar Arguments .. |
| 10 |
CHARACTER JOBVS, SORT |
| 11 |
INTEGER INFO, LDA, LDVS, LWORK, N, SDIM |
| 12 |
* .. |
| 13 |
* .. Array Arguments .. |
| 14 |
LOGICAL BWORK( * ) |
| 15 |
DOUBLE PRECISION RWORK( * ) |
| 16 |
COMPLEX*16 A( LDA, * ), VS( LDVS, * ), W( * ), WORK( * ) |
| 17 |
* .. |
| 18 |
* .. Function Arguments .. |
| 19 |
LOGICAL SELECT |
| 20 |
EXTERNAL SELECT |
| 21 |
* .. |
| 22 |
* |
| 23 |
* Purpose |
| 24 |
* ======= |
| 25 |
* |
| 26 |
* ZGEES computes for an N-by-N complex nonsymmetric matrix A, the |
| 27 |
* eigenvalues, the Schur form T, and, optionally, the matrix of Schur |
| 28 |
* vectors Z. This gives the Schur factorization A = Z*T*(Z**H). |
| 29 |
* |
| 30 |
* Optionally, it also orders the eigenvalues on the diagonal of the |
| 31 |
* Schur form so that selected eigenvalues are at the top left. |
| 32 |
* The leading columns of Z then form an orthonormal basis for the |
| 33 |
* invariant subspace corresponding to the selected eigenvalues. |
| 34 |
* |
| 35 |
* A complex matrix is in Schur form if it is upper triangular. |
| 36 |
* |
| 37 |
* Arguments |
| 38 |
* ========= |
| 39 |
* |
| 40 |
* JOBVS (input) CHARACTER*1 |
| 41 |
* = 'N': Schur vectors are not computed; |
| 42 |
* = 'V': Schur vectors are computed. |
| 43 |
* |
| 44 |
* SORT (input) CHARACTER*1 |
| 45 |
* Specifies whether or not to order the eigenvalues on the |
| 46 |
* diagonal of the Schur form. |
| 47 |
* = 'N': Eigenvalues are not ordered: |
| 48 |
* = 'S': Eigenvalues are ordered (see SELECT). |
| 49 |
* |
| 50 |
* SELECT (external procedure) LOGICAL FUNCTION of one COMPLEX*16 argument |
| 51 |
* SELECT must be declared EXTERNAL in the calling subroutine. |
| 52 |
* If SORT = 'S', SELECT is used to select eigenvalues to order |
| 53 |
* to the top left of the Schur form. |
| 54 |
* IF SORT = 'N', SELECT is not referenced. |
| 55 |
* The eigenvalue W(j) is selected if SELECT(W(j)) is true. |
| 56 |
* |
| 57 |
* N (input) INTEGER |
| 58 |
* The order of the matrix A. N >= 0. |
| 59 |
* |
| 60 |
* A (input/output) COMPLEX*16 array, dimension (LDA,N) |
| 61 |
* On entry, the N-by-N matrix A. |
| 62 |
* On exit, A has been overwritten by its Schur form T. |
| 63 |
* |
| 64 |
* LDA (input) INTEGER |
| 65 |
* The leading dimension of the array A. LDA >= max(1,N). |
| 66 |
* |
| 67 |
* SDIM (output) INTEGER |
| 68 |
* If SORT = 'N', SDIM = 0. |
| 69 |
* If SORT = 'S', SDIM = number of eigenvalues for which |
| 70 |
* SELECT is true. |
| 71 |
* |
| 72 |
* W (output) COMPLEX*16 array, dimension (N) |
| 73 |
* W contains the computed eigenvalues, in the same order that |
| 74 |
* they appear on the diagonal of the output Schur form T. |
| 75 |
* |
| 76 |
* VS (output) COMPLEX*16 array, dimension (LDVS,N) |
| 77 |
* If JOBVS = 'V', VS contains the unitary matrix Z of Schur |
| 78 |
* vectors. |
| 79 |
* If JOBVS = 'N', VS is not referenced. |
| 80 |
* |
| 81 |
* LDVS (input) INTEGER |
| 82 |
* The leading dimension of the array VS. LDVS >= 1; if |
| 83 |
* JOBVS = 'V', LDVS >= N. |
| 84 |
* |
| 85 |
* WORK (workspace/output) COMPLEX*16 array, dimension (MAX(1,LWORK)) |
| 86 |
* On exit, if INFO = 0, WORK(1) returns the optimal LWORK. |
| 87 |
* |
| 88 |
* LWORK (input) INTEGER |
| 89 |
* The dimension of the array WORK. LWORK >= max(1,2*N). |
| 90 |
* For good performance, LWORK must generally be larger. |
| 91 |
* |
| 92 |
* If LWORK = -1, then a workspace query is assumed; the routine |
| 93 |
* only calculates the optimal size of the WORK array, returns |
| 94 |
* this value as the first entry of the WORK array, and no error |
| 95 |
* message related to LWORK is issued by XERBLA. |
| 96 |
* |
| 97 |
* RWORK (workspace) DOUBLE PRECISION array, dimension (N) |
| 98 |
* |
| 99 |
* BWORK (workspace) LOGICAL array, dimension (N) |
| 100 |
* Not referenced if SORT = 'N'. |
| 101 |
* |
| 102 |
* INFO (output) INTEGER |
| 103 |
* = 0: successful exit |
| 104 |
* < 0: if INFO = -i, the i-th argument had an illegal value. |
| 105 |
* > 0: if INFO = i, and i is |
| 106 |
* <= N: the QR algorithm failed to compute all the |
| 107 |
* eigenvalues; elements 1:ILO-1 and i+1:N of W |
| 108 |
* contain those eigenvalues which have converged; |
| 109 |
* if JOBVS = 'V', VS contains the matrix which |
| 110 |
* reduces A to its partially converged Schur form. |
| 111 |
* = N+1: the eigenvalues could not be reordered because |
| 112 |
* some eigenvalues were too close to separate (the |
| 113 |
* problem is very ill-conditioned); |
| 114 |
* = N+2: after reordering, roundoff changed values of |
| 115 |
* some complex eigenvalues so that leading |
| 116 |
* eigenvalues in the Schur form no longer satisfy |
| 117 |
* SELECT = .TRUE.. This could also be caused by |
| 118 |
* underflow due to scaling. |
| 119 |
* |
| 120 |
* ===================================================================== |
| 121 |
* |
| 122 |
* .. Parameters .. |
| 123 |
DOUBLE PRECISION ZERO, ONE |
| 124 |
PARAMETER ( ZERO = 0.0D0, ONE = 1.0D0 ) |
| 125 |
* .. |
| 126 |
* .. Local Scalars .. |
| 127 |
LOGICAL LQUERY, SCALEA, WANTST, WANTVS |
| 128 |
INTEGER HSWORK, I, IBAL, ICOND, IERR, IEVAL, IHI, ILO, |
| 129 |
$ ITAU, IWRK, MAXWRK, MINWRK |
| 130 |
DOUBLE PRECISION ANRM, BIGNUM, CSCALE, EPS, S, SEP, SMLNUM |
| 131 |
* .. |
| 132 |
* .. Local Arrays .. |
| 133 |
DOUBLE PRECISION DUM( 1 ) |
| 134 |
* .. |
| 135 |
* .. External Subroutines .. |
| 136 |
EXTERNAL DLABAD, XERBLA, ZCOPY, ZGEBAK, ZGEBAL, ZGEHRD, |
| 137 |
$ ZHSEQR, ZLACPY, ZLASCL, ZTRSEN, ZUNGHR |
| 138 |
* .. |
| 139 |
* .. External Functions .. |
| 140 |
LOGICAL LSAME |
| 141 |
INTEGER ILAENV |
| 142 |
DOUBLE PRECISION DLAMCH, ZLANGE |
| 143 |
EXTERNAL LSAME, ILAENV, DLAMCH, ZLANGE |
| 144 |
* .. |
| 145 |
* .. Intrinsic Functions .. |
| 146 |
INTRINSIC MAX, SQRT |
| 147 |
* .. |
| 148 |
* .. Executable Statements .. |
| 149 |
* |
| 150 |
* Test the input arguments |
| 151 |
* |
| 152 |
INFO = 0 |
| 153 |
LQUERY = ( LWORK.EQ.-1 ) |
| 154 |
WANTVS = LSAME( JOBVS, 'V' ) |
| 155 |
WANTST = LSAME( SORT, 'S' ) |
| 156 |
IF( ( .NOT.WANTVS ) .AND. ( .NOT.LSAME( JOBVS, 'N' ) ) ) THEN |
| 157 |
INFO = -1 |
| 158 |
ELSE IF( ( .NOT.WANTST ) .AND. ( .NOT.LSAME( SORT, 'N' ) ) ) THEN |
| 159 |
INFO = -2 |
| 160 |
ELSE IF( N.LT.0 ) THEN |
| 161 |
INFO = -4 |
| 162 |
ELSE IF( LDA.LT.MAX( 1, N ) ) THEN |
| 163 |
INFO = -6 |
| 164 |
ELSE IF( LDVS.LT.1 .OR. ( WANTVS .AND. LDVS.LT.N ) ) THEN |
| 165 |
INFO = -10 |
| 166 |
END IF |
| 167 |
* |
| 168 |
* Compute workspace |
| 169 |
* (Note: Comments in the code beginning "Workspace:" describe the |
| 170 |
* minimal amount of workspace needed at that point in the code, |
| 171 |
* as well as the preferred amount for good performance. |
| 172 |
* CWorkspace refers to complex workspace, and RWorkspace to real |
| 173 |
* workspace. NB refers to the optimal block size for the |
| 174 |
* immediately following subroutine, as returned by ILAENV. |
| 175 |
* HSWORK refers to the workspace preferred by ZHSEQR, as |
| 176 |
* calculated below. HSWORK is computed assuming ILO=1 and IHI=N, |
| 177 |
* the worst case.) |
| 178 |
* |
| 179 |
IF( INFO.EQ.0 ) THEN |
| 180 |
IF( N.EQ.0 ) THEN |
| 181 |
MINWRK = 1 |
| 182 |
MAXWRK = 1 |
| 183 |
ELSE |
| 184 |
MAXWRK = N + N*ILAENV( 1, 'ZGEHRD', ' ', N, 1, N, 0 ) |
| 185 |
MINWRK = 2*N |
| 186 |
* |
| 187 |
CALL ZHSEQR( 'S', JOBVS, N, 1, N, A, LDA, W, VS, LDVS, |
| 188 |
$ WORK, -1, IEVAL ) |
| 189 |
HSWORK = WORK( 1 ) |
| 190 |
* |
| 191 |
IF( .NOT.WANTVS ) THEN |
| 192 |
MAXWRK = MAX( MAXWRK, HSWORK ) |
| 193 |
ELSE |
| 194 |
MAXWRK = MAX( MAXWRK, N + ( N - 1 )*ILAENV( 1, 'ZUNGHR', |
| 195 |
$ ' ', N, 1, N, -1 ) ) |
| 196 |
MAXWRK = MAX( MAXWRK, HSWORK ) |
| 197 |
END IF |
| 198 |
END IF |
| 199 |
WORK( 1 ) = MAXWRK |
| 200 |
* |
| 201 |
IF( LWORK.LT.MINWRK .AND. .NOT.LQUERY ) THEN |
| 202 |
INFO = -12 |
| 203 |
END IF |
| 204 |
END IF |
| 205 |
* |
| 206 |
IF( INFO.NE.0 ) THEN |
| 207 |
CALL XERBLA( 'ZGEES ', -INFO ) |
| 208 |
RETURN |
| 209 |
ELSE IF( LQUERY ) THEN |
| 210 |
RETURN |
| 211 |
END IF |
| 212 |
* |
| 213 |
* Quick return if possible |
| 214 |
* |
| 215 |
IF( N.EQ.0 ) THEN |
| 216 |
SDIM = 0 |
| 217 |
RETURN |
| 218 |
END IF |
| 219 |
* |
| 220 |
* Get machine constants |
| 221 |
* |
| 222 |
EPS = DLAMCH( 'P' ) |
| 223 |
SMLNUM = DLAMCH( 'S' ) |
| 224 |
BIGNUM = ONE / SMLNUM |
| 225 |
CALL DLABAD( SMLNUM, BIGNUM ) |
| 226 |
SMLNUM = SQRT( SMLNUM ) / EPS |
| 227 |
BIGNUM = ONE / SMLNUM |
| 228 |
* |
| 229 |
* Scale A if max element outside range [SMLNUM,BIGNUM] |
| 230 |
* |
| 231 |
ANRM = ZLANGE( 'M', N, N, A, LDA, DUM ) |
| 232 |
SCALEA = .FALSE. |
| 233 |
IF( ANRM.GT.ZERO .AND. ANRM.LT.SMLNUM ) THEN |
| 234 |
SCALEA = .TRUE. |
| 235 |
CSCALE = SMLNUM |
| 236 |
ELSE IF( ANRM.GT.BIGNUM ) THEN |
| 237 |
SCALEA = .TRUE. |
| 238 |
CSCALE = BIGNUM |
| 239 |
END IF |
| 240 |
IF( SCALEA ) |
| 241 |
$ CALL ZLASCL( 'G', 0, 0, ANRM, CSCALE, N, N, A, LDA, IERR ) |
| 242 |
* |
| 243 |
* Permute the matrix to make it more nearly triangular |
| 244 |
* (CWorkspace: none) |
| 245 |
* (RWorkspace: need N) |
| 246 |
* |
| 247 |
IBAL = 1 |
| 248 |
CALL ZGEBAL( 'P', N, A, LDA, ILO, IHI, RWORK( IBAL ), IERR ) |
| 249 |
* |
| 250 |
* Reduce to upper Hessenberg form |
| 251 |
* (CWorkspace: need 2*N, prefer N+N*NB) |
| 252 |
* (RWorkspace: none) |
| 253 |
* |
| 254 |
ITAU = 1 |
| 255 |
IWRK = N + ITAU |
| 256 |
CALL ZGEHRD( N, ILO, IHI, A, LDA, WORK( ITAU ), WORK( IWRK ), |
| 257 |
$ LWORK-IWRK+1, IERR ) |
| 258 |
* |
| 259 |
IF( WANTVS ) THEN |
| 260 |
* |
| 261 |
* Copy Householder vectors to VS |
| 262 |
* |
| 263 |
CALL ZLACPY( 'L', N, N, A, LDA, VS, LDVS ) |
| 264 |
* |
| 265 |
* Generate unitary matrix in VS |
| 266 |
* (CWorkspace: need 2*N-1, prefer N+(N-1)*NB) |
| 267 |
* (RWorkspace: none) |
| 268 |
* |
| 269 |
CALL ZUNGHR( N, ILO, IHI, VS, LDVS, WORK( ITAU ), WORK( IWRK ), |
| 270 |
$ LWORK-IWRK+1, IERR ) |
| 271 |
END IF |
| 272 |
* |
| 273 |
SDIM = 0 |
| 274 |
* |
| 275 |
* Perform QR iteration, accumulating Schur vectors in VS if desired |
| 276 |
* (CWorkspace: need 1, prefer HSWORK (see comments) ) |
| 277 |
* (RWorkspace: none) |
| 278 |
* |
| 279 |
IWRK = ITAU |
| 280 |
CALL ZHSEQR( 'S', JOBVS, N, ILO, IHI, A, LDA, W, VS, LDVS, |
| 281 |
$ WORK( IWRK ), LWORK-IWRK+1, IEVAL ) |
| 282 |
IF( IEVAL.GT.0 ) |
| 283 |
$ INFO = IEVAL |
| 284 |
* |
| 285 |
* Sort eigenvalues if desired |
| 286 |
* |
| 287 |
IF( WANTST .AND. INFO.EQ.0 ) THEN |
| 288 |
IF( SCALEA ) |
| 289 |
$ CALL ZLASCL( 'G', 0, 0, CSCALE, ANRM, N, 1, W, N, IERR ) |
| 290 |
DO 10 I = 1, N |
| 291 |
BWORK( I ) = SELECT( W( I ) ) |
| 292 |
10 CONTINUE |
| 293 |
* |
| 294 |
* Reorder eigenvalues and transform Schur vectors |
| 295 |
* (CWorkspace: none) |
| 296 |
* (RWorkspace: none) |
| 297 |
* |
| 298 |
CALL ZTRSEN( 'N', JOBVS, BWORK, N, A, LDA, VS, LDVS, W, SDIM, |
| 299 |
$ S, SEP, WORK( IWRK ), LWORK-IWRK+1, ICOND ) |
| 300 |
END IF |
| 301 |
* |
| 302 |
IF( WANTVS ) THEN |
| 303 |
* |
| 304 |
* Undo balancing |
| 305 |
* (CWorkspace: none) |
| 306 |
* (RWorkspace: need N) |
| 307 |
* |
| 308 |
CALL ZGEBAK( 'P', 'R', N, ILO, IHI, RWORK( IBAL ), N, VS, LDVS, |
| 309 |
$ IERR ) |
| 310 |
END IF |
| 311 |
* |
| 312 |
IF( SCALEA ) THEN |
| 313 |
* |
| 314 |
* Undo scaling for the Schur form of A |
| 315 |
* |
| 316 |
CALL ZLASCL( 'U', 0, 0, CSCALE, ANRM, N, N, A, LDA, IERR ) |
| 317 |
CALL ZCOPY( N, A, LDA+1, W, 1 ) |
| 318 |
END IF |
| 319 |
* |
| 320 |
WORK( 1 ) = MAXWRK |
| 321 |
RETURN |
| 322 |
* |
| 323 |
* End of ZGEES |
| 324 |
* |
| 325 |
END |