1*> \brief \b CGET22
2*
3*  =========== DOCUMENTATION ===========
4*
5* Online html documentation available at
6*            http://www.netlib.org/lapack/explore-html/
7*
8*  Definition:
9*  ===========
10*
11*       SUBROUTINE CGET22( TRANSA, TRANSE, TRANSW, N, A, LDA, E, LDE, W,
12*                          WORK, RWORK, RESULT )
13*
14*       .. Scalar Arguments ..
15*       CHARACTER          TRANSA, TRANSE, TRANSW
16*       INTEGER            LDA, LDE, N
17*       ..
18*       .. Array Arguments ..
19*       REAL               RESULT( 2 ), RWORK( * )
20*       COMPLEX            A( LDA, * ), E( LDE, * ), W( * ), WORK( * )
21*       ..
22*
23*
24*> \par Purpose:
25*  =============
26*>
27*> \verbatim
28*>
29*> CGET22 does an eigenvector check.
30*>
31*> The basic test is:
32*>
33*>    RESULT(1) = | A E  -  E W | / ( |A| |E| ulp )
34*>
35*> using the 1-norm.  It also tests the normalization of E:
36*>
37*>    RESULT(2) = max | m-norm(E(j)) - 1 | / ( n ulp )
38*>                 j
39*>
40*> where E(j) is the j-th eigenvector, and m-norm is the max-norm of a
41*> vector.  The max-norm of a complex n-vector x in this case is the
42*> maximum of |re(x(i)| + |im(x(i)| over i = 1, ..., n.
43*> \endverbatim
44*
45*  Arguments:
46*  ==========
47*
48*> \param[in] TRANSA
49*> \verbatim
50*>          TRANSA is CHARACTER*1
51*>          Specifies whether or not A is transposed.
52*>          = 'N':  No transpose
53*>          = 'T':  Transpose
54*>          = 'C':  Conjugate transpose
55*> \endverbatim
56*>
57*> \param[in] TRANSE
58*> \verbatim
59*>          TRANSE is CHARACTER*1
60*>          Specifies whether or not E is transposed.
61*>          = 'N':  No transpose, eigenvectors are in columns of E
62*>          = 'T':  Transpose, eigenvectors are in rows of E
63*>          = 'C':  Conjugate transpose, eigenvectors are in rows of E
64*> \endverbatim
65*>
66*> \param[in] TRANSW
67*> \verbatim
68*>          TRANSW is CHARACTER*1
69*>          Specifies whether or not W is transposed.
70*>          = 'N':  No transpose
71*>          = 'T':  Transpose, same as TRANSW = 'N'
72*>          = 'C':  Conjugate transpose, use -WI(j) instead of WI(j)
73*> \endverbatim
74*>
75*> \param[in] N
76*> \verbatim
77*>          N is INTEGER
78*>          The order of the matrix A.  N >= 0.
79*> \endverbatim
80*>
81*> \param[in] A
82*> \verbatim
83*>          A is COMPLEX array, dimension (LDA,N)
84*>          The matrix whose eigenvectors are in E.
85*> \endverbatim
86*>
87*> \param[in] LDA
88*> \verbatim
89*>          LDA is INTEGER
90*>          The leading dimension of the array A.  LDA >= max(1,N).
91*> \endverbatim
92*>
93*> \param[in] E
94*> \verbatim
95*>          E is COMPLEX array, dimension (LDE,N)
96*>          The matrix of eigenvectors. If TRANSE = 'N', the eigenvectors
97*>          are stored in the columns of E, if TRANSE = 'T' or 'C', the
98*>          eigenvectors are stored in the rows of E.
99*> \endverbatim
100*>
101*> \param[in] LDE
102*> \verbatim
103*>          LDE is INTEGER
104*>          The leading dimension of the array E.  LDE >= max(1,N).
105*> \endverbatim
106*>
107*> \param[in] W
108*> \verbatim
109*>          W is COMPLEX array, dimension (N)
110*>          The eigenvalues of A.
111*> \endverbatim
112*>
113*> \param[out] WORK
114*> \verbatim
115*>          WORK is COMPLEX array, dimension (N*N)
116*> \endverbatim
117*>
118*> \param[out] RWORK
119*> \verbatim
120*>          RWORK is REAL array, dimension (N)
121*> \endverbatim
122*>
123*> \param[out] RESULT
124*> \verbatim
125*>          RESULT is REAL array, dimension (2)
126*>          RESULT(1) = | A E  -  E W | / ( |A| |E| ulp )
127*>          RESULT(2) = max | m-norm(E(j)) - 1 | / ( n ulp )
128*>                       j
129*> \endverbatim
130*
131*  Authors:
132*  ========
133*
134*> \author Univ. of Tennessee
135*> \author Univ. of California Berkeley
136*> \author Univ. of Colorado Denver
137*> \author NAG Ltd.
138*
139*> \ingroup complex_eig
140*
141*  =====================================================================
142      SUBROUTINE CGET22( TRANSA, TRANSE, TRANSW, N, A, LDA, E, LDE, W,
143     $                   WORK, RWORK, RESULT )
144*
145*  -- LAPACK test routine --
146*  -- LAPACK is a software package provided by Univ. of Tennessee,    --
147*  -- Univ. of California Berkeley, Univ. of Colorado Denver and NAG Ltd..--
148*
149*     .. Scalar Arguments ..
150      CHARACTER          TRANSA, TRANSE, TRANSW
151      INTEGER            LDA, LDE, N
152*     ..
153*     .. Array Arguments ..
154      REAL               RESULT( 2 ), RWORK( * )
155      COMPLEX            A( LDA, * ), E( LDE, * ), W( * ), WORK( * )
156*     ..
157*
158*  =====================================================================
159*
160*     .. Parameters ..
161      REAL               ZERO, ONE
162      PARAMETER          ( ZERO = 0.0E+0, ONE = 1.0E+0 )
163      COMPLEX            CZERO, CONE
164      PARAMETER          ( CZERO = ( 0.0E+0, 0.0E+0 ),
165     $                   CONE = ( 1.0E+0, 0.0E+0 ) )
166*     ..
167*     .. Local Scalars ..
168      CHARACTER          NORMA, NORME
169      INTEGER            ITRNSE, ITRNSW, J, JCOL, JOFF, JROW, JVEC
170      REAL               ANORM, ENORM, ENRMAX, ENRMIN, ERRNRM, TEMP1,
171     $                   ULP, UNFL
172      COMPLEX            WTEMP
173*     ..
174*     .. External Functions ..
175      LOGICAL            LSAME
176      REAL               CLANGE, SLAMCH
177      EXTERNAL           LSAME, CLANGE, SLAMCH
178*     ..
179*     .. External Subroutines ..
180      EXTERNAL           CGEMM, CLASET
181*     ..
182*     .. Intrinsic Functions ..
183      INTRINSIC          ABS, AIMAG, CONJG, MAX, MIN, REAL
184*     ..
185*     .. Executable Statements ..
186*
187*     Initialize RESULT (in case N=0)
188*
189      RESULT( 1 ) = ZERO
190      RESULT( 2 ) = ZERO
191      IF( N.LE.0 )
192     $   RETURN
193*
194      UNFL = SLAMCH( 'Safe minimum' )
195      ULP = SLAMCH( 'Precision' )
196*
197      ITRNSE = 0
198      ITRNSW = 0
199      NORMA = 'O'
200      NORME = 'O'
201*
202      IF( LSAME( TRANSA, 'T' ) .OR. LSAME( TRANSA, 'C' ) ) THEN
203         NORMA = 'I'
204      END IF
205*
206      IF( LSAME( TRANSE, 'T' ) ) THEN
207         ITRNSE = 1
208         NORME = 'I'
209      ELSE IF( LSAME( TRANSE, 'C' ) ) THEN
210         ITRNSE = 2
211         NORME = 'I'
212      END IF
213*
214      IF( LSAME( TRANSW, 'C' ) ) THEN
215         ITRNSW = 1
216      END IF
217*
218*     Normalization of E:
219*
220      ENRMIN = ONE / ULP
221      ENRMAX = ZERO
222      IF( ITRNSE.EQ.0 ) THEN
223         DO 20 JVEC = 1, N
224            TEMP1 = ZERO
225            DO 10 J = 1, N
226               TEMP1 = MAX( TEMP1, ABS( REAL( E( J, JVEC ) ) )+
227     $                 ABS( AIMAG( E( J, JVEC ) ) ) )
228   10       CONTINUE
229            ENRMIN = MIN( ENRMIN, TEMP1 )
230            ENRMAX = MAX( ENRMAX, TEMP1 )
231   20    CONTINUE
232      ELSE
233         DO 30 JVEC = 1, N
234            RWORK( JVEC ) = ZERO
235   30    CONTINUE
236*
237         DO 50 J = 1, N
238            DO 40 JVEC = 1, N
239               RWORK( JVEC ) = MAX( RWORK( JVEC ),
240     $                         ABS( REAL( E( JVEC, J ) ) )+
241     $                         ABS( AIMAG( E( JVEC, J ) ) ) )
242   40       CONTINUE
243   50    CONTINUE
244*
245         DO 60 JVEC = 1, N
246            ENRMIN = MIN( ENRMIN, RWORK( JVEC ) )
247            ENRMAX = MAX( ENRMAX, RWORK( JVEC ) )
248   60    CONTINUE
249      END IF
250*
251*     Norm of A:
252*
253      ANORM = MAX( CLANGE( NORMA, N, N, A, LDA, RWORK ), UNFL )
254*
255*     Norm of E:
256*
257      ENORM = MAX( CLANGE( NORME, N, N, E, LDE, RWORK ), ULP )
258*
259*     Norm of error:
260*
261*     Error =  AE - EW
262*
263      CALL CLASET( 'Full', N, N, CZERO, CZERO, WORK, N )
264*
265      JOFF = 0
266      DO 100 JCOL = 1, N
267         IF( ITRNSW.EQ.0 ) THEN
268            WTEMP = W( JCOL )
269         ELSE
270            WTEMP = CONJG( W( JCOL ) )
271         END IF
272*
273         IF( ITRNSE.EQ.0 ) THEN
274            DO 70 JROW = 1, N
275               WORK( JOFF+JROW ) = E( JROW, JCOL )*WTEMP
276   70       CONTINUE
277         ELSE IF( ITRNSE.EQ.1 ) THEN
278            DO 80 JROW = 1, N
279               WORK( JOFF+JROW ) = E( JCOL, JROW )*WTEMP
280   80       CONTINUE
281         ELSE
282            DO 90 JROW = 1, N
283               WORK( JOFF+JROW ) = CONJG( E( JCOL, JROW ) )*WTEMP
284   90       CONTINUE
285         END IF
286         JOFF = JOFF + N
287  100 CONTINUE
288*
289      CALL CGEMM( TRANSA, TRANSE, N, N, N, CONE, A, LDA, E, LDE, -CONE,
290     $            WORK, N )
291*
292      ERRNRM = CLANGE( 'One', N, N, WORK, N, RWORK ) / ENORM
293*
294*     Compute RESULT(1) (avoiding under/overflow)
295*
296      IF( ANORM.GT.ERRNRM ) THEN
297         RESULT( 1 ) = ( ERRNRM / ANORM ) / ULP
298      ELSE
299         IF( ANORM.LT.ONE ) THEN
300            RESULT( 1 ) = ONE / ULP
301         ELSE
302            RESULT( 1 ) = MIN( ERRNRM / ANORM, ONE ) / ULP
303         END IF
304      END IF
305*
306*     Compute RESULT(2) : the normalization error in E.
307*
308      RESULT( 2 ) = MAX( ABS( ENRMAX-ONE ), ABS( ENRMIN-ONE ) ) /
309     $              ( REAL( N )*ULP )
310*
311      RETURN
312*
313*     End of CGET22
314*
315      END
316