LAPACK  3.6.1
LAPACK: Linear Algebra PACKage
zdrvge.f
Go to the documentation of this file.
1 *> \brief \b ZDRVGE
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 ZDRVGE( DOTYPE, NN, NVAL, NRHS, THRESH, TSTERR, NMAX,
12 * A, AFAC, ASAV, B, BSAV, X, XACT, S, WORK,
13 * RWORK, IWORK, NOUT )
14 *
15 * .. Scalar Arguments ..
16 * LOGICAL TSTERR
17 * INTEGER NMAX, NN, NOUT, NRHS
18 * DOUBLE PRECISION THRESH
19 * ..
20 * .. Array Arguments ..
21 * LOGICAL DOTYPE( * )
22 * INTEGER IWORK( * ), NVAL( * )
23 * DOUBLE PRECISION RWORK( * ), S( * )
24 * COMPLEX*16 A( * ), AFAC( * ), ASAV( * ), B( * ),
25 * $ BSAV( * ), WORK( * ), X( * ), XACT( * )
26 * ..
27 *
28 *
29 *> \par Purpose:
30 * =============
31 *>
32 *> \verbatim
33 *>
34 *> ZDRVGE tests the driver routines ZGESV and -SVX.
35 *> \endverbatim
36 *
37 * Arguments:
38 * ==========
39 *
40 *> \param[in] DOTYPE
41 *> \verbatim
42 *> DOTYPE is LOGICAL array, dimension (NTYPES)
43 *> The matrix types to be used for testing. Matrices of type j
44 *> (for 1 <= j <= NTYPES) are used for testing if DOTYPE(j) =
45 *> .TRUE.; if DOTYPE(j) = .FALSE., then type j is not used.
46 *> \endverbatim
47 *>
48 *> \param[in] NN
49 *> \verbatim
50 *> NN is INTEGER
51 *> The number of values of N contained in the vector NVAL.
52 *> \endverbatim
53 *>
54 *> \param[in] NVAL
55 *> \verbatim
56 *> NVAL is INTEGER array, dimension (NN)
57 *> The values of the matrix column dimension N.
58 *> \endverbatim
59 *>
60 *> \param[in] NRHS
61 *> \verbatim
62 *> NRHS is INTEGER
63 *> The number of right hand side vectors to be generated for
64 *> each linear system.
65 *> \endverbatim
66 *>
67 *> \param[in] THRESH
68 *> \verbatim
69 *> THRESH is DOUBLE PRECISION
70 *> The threshold value for the test ratios. A result is
71 *> included in the output file if RESULT >= THRESH. To have
72 *> every test ratio printed, use THRESH = 0.
73 *> \endverbatim
74 *>
75 *> \param[in] TSTERR
76 *> \verbatim
77 *> TSTERR is LOGICAL
78 *> Flag that indicates whether error exits are to be tested.
79 *> \endverbatim
80 *>
81 *> \param[in] NMAX
82 *> \verbatim
83 *> NMAX is INTEGER
84 *> The maximum value permitted for N, used in dimensioning the
85 *> work arrays.
86 *> \endverbatim
87 *>
88 *> \param[out] A
89 *> \verbatim
90 *> A is COMPLEX*16 array, dimension (NMAX*NMAX)
91 *> \endverbatim
92 *>
93 *> \param[out] AFAC
94 *> \verbatim
95 *> AFAC is COMPLEX*16 array, dimension (NMAX*NMAX)
96 *> \endverbatim
97 *>
98 *> \param[out] ASAV
99 *> \verbatim
100 *> ASAV is COMPLEX*16 array, dimension (NMAX*NMAX)
101 *> \endverbatim
102 *>
103 *> \param[out] B
104 *> \verbatim
105 *> B is COMPLEX*16 array, dimension (NMAX*NRHS)
106 *> \endverbatim
107 *>
108 *> \param[out] BSAV
109 *> \verbatim
110 *> BSAV is COMPLEX*16 array, dimension (NMAX*NRHS)
111 *> \endverbatim
112 *>
113 *> \param[out] X
114 *> \verbatim
115 *> X is COMPLEX*16 array, dimension (NMAX*NRHS)
116 *> \endverbatim
117 *>
118 *> \param[out] XACT
119 *> \verbatim
120 *> XACT is COMPLEX*16 array, dimension (NMAX*NRHS)
121 *> \endverbatim
122 *>
123 *> \param[out] S
124 *> \verbatim
125 *> S is DOUBLE PRECISION array, dimension (2*NMAX)
126 *> \endverbatim
127 *>
128 *> \param[out] WORK
129 *> \verbatim
130 *> WORK is COMPLEX*16 array, dimension
131 *> (NMAX*max(3,NRHS))
132 *> \endverbatim
133 *>
134 *> \param[out] RWORK
135 *> \verbatim
136 *> RWORK is DOUBLE PRECISION array, dimension (2*NRHS+NMAX)
137 *> \endverbatim
138 *>
139 *> \param[out] IWORK
140 *> \verbatim
141 *> IWORK is INTEGER array, dimension (NMAX)
142 *> \endverbatim
143 *>
144 *> \param[in] NOUT
145 *> \verbatim
146 *> NOUT is INTEGER
147 *> The unit number for output.
148 *> \endverbatim
149 *
150 * Authors:
151 * ========
152 *
153 *> \author Univ. of Tennessee
154 *> \author Univ. of California Berkeley
155 *> \author Univ. of Colorado Denver
156 *> \author NAG Ltd.
157 *
158 *> \date November 2015
159 *
160 *> \ingroup complex16_lin
161 *
162 * =====================================================================
163  SUBROUTINE zdrvge( DOTYPE, NN, NVAL, NRHS, THRESH, TSTERR, NMAX,
164  $ a, afac, asav, b, bsav, x, xact, s, work,
165  $ rwork, iwork, nout )
166 *
167 * -- LAPACK test routine (version 3.6.0) --
168 * -- LAPACK is a software package provided by Univ. of Tennessee, --
169 * -- Univ. of California Berkeley, Univ. of Colorado Denver and NAG Ltd..--
170 * November 2015
171 *
172 * .. Scalar Arguments ..
173  LOGICAL TSTERR
174  INTEGER NMAX, NN, NOUT, NRHS
175  DOUBLE PRECISION THRESH
176 * ..
177 * .. Array Arguments ..
178  LOGICAL DOTYPE( * )
179  INTEGER IWORK( * ), NVAL( * )
180  DOUBLE PRECISION RWORK( * ), S( * )
181  COMPLEX*16 A( * ), AFAC( * ), ASAV( * ), B( * ),
182  $ bsav( * ), work( * ), x( * ), xact( * )
183 * ..
184 *
185 * =====================================================================
186 *
187 * .. Parameters ..
188  DOUBLE PRECISION ONE, ZERO
189  parameter ( one = 1.0d+0, zero = 0.0d+0 )
190  INTEGER NTYPES
191  parameter ( ntypes = 11 )
192  INTEGER NTESTS
193  parameter ( ntests = 7 )
194  INTEGER NTRAN
195  parameter ( ntran = 3 )
196 * ..
197 * .. Local Scalars ..
198  LOGICAL EQUIL, NOFACT, PREFAC, TRFCON, ZEROT
199  CHARACTER DIST, EQUED, FACT, TRANS, TYPE, XTYPE
200  CHARACTER*3 PATH
201  INTEGER I, IEQUED, IFACT, IMAT, IN, INFO, IOFF, ITRAN,
202  $ izero, k, k1, kl, ku, lda, lwork, mode, n, nb,
203  $ nbmin, nerrs, nfact, nfail, nimat, nrun, nt
204  DOUBLE PRECISION AINVNM, AMAX, ANORM, ANORMI, ANORMO, CNDNUM,
205  $ colcnd, rcond, rcondc, rcondi, rcondo, roldc,
206  $ roldi, roldo, rowcnd, rpvgrw
207 * ..
208 * .. Local Arrays ..
209  CHARACTER EQUEDS( 4 ), FACTS( 3 ), TRANSS( ntran )
210  INTEGER ISEED( 4 ), ISEEDY( 4 )
211  DOUBLE PRECISION RDUM( 1 ), RESULT( ntests )
212 * ..
213 * .. External Functions ..
214  LOGICAL LSAME
215  DOUBLE PRECISION DGET06, DLAMCH, ZLANGE, ZLANTR
216  EXTERNAL lsame, dget06, dlamch, zlange, zlantr
217 * ..
218 * .. External Subroutines ..
219  EXTERNAL aladhd, alaerh, alasvm, xlaenv, zerrvx, zgeequ,
222  $ zlatb4, zlatms
223 * ..
224 * .. Intrinsic Functions ..
225  INTRINSIC abs, dcmplx, max
226 * ..
227 * .. Scalars in Common ..
228  LOGICAL LERR, OK
229  CHARACTER*32 SRNAMT
230  INTEGER INFOT, NUNIT
231 * ..
232 * .. Common blocks ..
233  COMMON / infoc / infot, nunit, ok, lerr
234  COMMON / srnamc / srnamt
235 * ..
236 * .. Data statements ..
237  DATA iseedy / 1988, 1989, 1990, 1991 /
238  DATA transs / 'N', 'T', 'C' /
239  DATA facts / 'F', 'N', 'E' /
240  DATA equeds / 'N', 'R', 'C', 'B' /
241 * ..
242 * .. Executable Statements ..
243 *
244 * Initialize constants and the random number seed.
245 *
246  path( 1: 1 ) = 'Zomplex precision'
247  path( 2: 3 ) = 'GE'
248  nrun = 0
249  nfail = 0
250  nerrs = 0
251  DO 10 i = 1, 4
252  iseed( i ) = iseedy( i )
253  10 CONTINUE
254 *
255 * Test the error exits
256 *
257  IF( tsterr )
258  $ CALL zerrvx( path, nout )
259  infot = 0
260 *
261 * Set the block size and minimum block size for testing.
262 *
263  nb = 1
264  nbmin = 2
265  CALL xlaenv( 1, nb )
266  CALL xlaenv( 2, nbmin )
267 *
268 * Do for each value of N in NVAL
269 *
270  DO 90 in = 1, nn
271  n = nval( in )
272  lda = max( n, 1 )
273  xtype = 'N'
274  nimat = ntypes
275  IF( n.LE.0 )
276  $ nimat = 1
277 *
278  DO 80 imat = 1, nimat
279 *
280 * Do the tests only if DOTYPE( IMAT ) is true.
281 *
282  IF( .NOT.dotype( imat ) )
283  $ GO TO 80
284 *
285 * Skip types 5, 6, or 7 if the matrix size is too small.
286 *
287  zerot = imat.GE.5 .AND. imat.LE.7
288  IF( zerot .AND. n.LT.imat-4 )
289  $ GO TO 80
290 *
291 * Set up parameters with ZLATB4 and generate a test matrix
292 * with ZLATMS.
293 *
294  CALL zlatb4( path, imat, n, n, TYPE, KL, KU, ANORM, MODE,
295  $ cndnum, dist )
296  rcondc = one / cndnum
297 *
298  srnamt = 'ZLATMS'
299  CALL zlatms( n, n, dist, iseed, TYPE, RWORK, MODE, CNDNUM,
300  $ anorm, kl, ku, 'No packing', a, lda, work,
301  $ info )
302 *
303 * Check error code from ZLATMS.
304 *
305  IF( info.NE.0 ) THEN
306  CALL alaerh( path, 'ZLATMS', info, 0, ' ', n, n, -1, -1,
307  $ -1, imat, nfail, nerrs, nout )
308  GO TO 80
309  END IF
310 *
311 * For types 5-7, zero one or more columns of the matrix to
312 * test that INFO is returned correctly.
313 *
314  IF( zerot ) THEN
315  IF( imat.EQ.5 ) THEN
316  izero = 1
317  ELSE IF( imat.EQ.6 ) THEN
318  izero = n
319  ELSE
320  izero = n / 2 + 1
321  END IF
322  ioff = ( izero-1 )*lda
323  IF( imat.LT.7 ) THEN
324  DO 20 i = 1, n
325  a( ioff+i ) = zero
326  20 CONTINUE
327  ELSE
328  CALL zlaset( 'Full', n, n-izero+1, dcmplx( zero ),
329  $ dcmplx( zero ), a( ioff+1 ), lda )
330  END IF
331  ELSE
332  izero = 0
333  END IF
334 *
335 * Save a copy of the matrix A in ASAV.
336 *
337  CALL zlacpy( 'Full', n, n, a, lda, asav, lda )
338 *
339  DO 70 iequed = 1, 4
340  equed = equeds( iequed )
341  IF( iequed.EQ.1 ) THEN
342  nfact = 3
343  ELSE
344  nfact = 1
345  END IF
346 *
347  DO 60 ifact = 1, nfact
348  fact = facts( ifact )
349  prefac = lsame( fact, 'F' )
350  nofact = lsame( fact, 'N' )
351  equil = lsame( fact, 'E' )
352 *
353  IF( zerot ) THEN
354  IF( prefac )
355  $ GO TO 60
356  rcondo = zero
357  rcondi = zero
358 *
359  ELSE IF( .NOT.nofact ) THEN
360 *
361 * Compute the condition number for comparison with
362 * the value returned by ZGESVX (FACT = 'N' reuses
363 * the condition number from the previous iteration
364 * with FACT = 'F').
365 *
366  CALL zlacpy( 'Full', n, n, asav, lda, afac, lda )
367  IF( equil .OR. iequed.GT.1 ) THEN
368 *
369 * Compute row and column scale factors to
370 * equilibrate the matrix A.
371 *
372  CALL zgeequ( n, n, afac, lda, s, s( n+1 ),
373  $ rowcnd, colcnd, amax, info )
374  IF( info.EQ.0 .AND. n.GT.0 ) THEN
375  IF( lsame( equed, 'R' ) ) THEN
376  rowcnd = zero
377  colcnd = one
378  ELSE IF( lsame( equed, 'C' ) ) THEN
379  rowcnd = one
380  colcnd = zero
381  ELSE IF( lsame( equed, 'B' ) ) THEN
382  rowcnd = zero
383  colcnd = zero
384  END IF
385 *
386 * Equilibrate the matrix.
387 *
388  CALL zlaqge( n, n, afac, lda, s, s( n+1 ),
389  $ rowcnd, colcnd, amax, equed )
390  END IF
391  END IF
392 *
393 * Save the condition number of the non-equilibrated
394 * system for use in ZGET04.
395 *
396  IF( equil ) THEN
397  roldo = rcondo
398  roldi = rcondi
399  END IF
400 *
401 * Compute the 1-norm and infinity-norm of A.
402 *
403  anormo = zlange( '1', n, n, afac, lda, rwork )
404  anormi = zlange( 'I', n, n, afac, lda, rwork )
405 *
406 * Factor the matrix A.
407 *
408  srnamt = 'ZGETRF'
409  CALL zgetrf( n, n, afac, lda, iwork, info )
410 *
411 * Form the inverse of A.
412 *
413  CALL zlacpy( 'Full', n, n, afac, lda, a, lda )
414  lwork = nmax*max( 3, nrhs )
415  srnamt = 'ZGETRI'
416  CALL zgetri( n, a, lda, iwork, work, lwork, info )
417 *
418 * Compute the 1-norm condition number of A.
419 *
420  ainvnm = zlange( '1', n, n, a, lda, rwork )
421  IF( anormo.LE.zero .OR. ainvnm.LE.zero ) THEN
422  rcondo = one
423  ELSE
424  rcondo = ( one / anormo ) / ainvnm
425  END IF
426 *
427 * Compute the infinity-norm condition number of A.
428 *
429  ainvnm = zlange( 'I', n, n, a, lda, rwork )
430  IF( anormi.LE.zero .OR. ainvnm.LE.zero ) THEN
431  rcondi = one
432  ELSE
433  rcondi = ( one / anormi ) / ainvnm
434  END IF
435  END IF
436 *
437  DO 50 itran = 1, ntran
438 *
439 * Do for each value of TRANS.
440 *
441  trans = transs( itran )
442  IF( itran.EQ.1 ) THEN
443  rcondc = rcondo
444  ELSE
445  rcondc = rcondi
446  END IF
447 *
448 * Restore the matrix A.
449 *
450  CALL zlacpy( 'Full', n, n, asav, lda, a, lda )
451 *
452 * Form an exact solution and set the right hand side.
453 *
454  srnamt = 'ZLARHS'
455  CALL zlarhs( path, xtype, 'Full', trans, n, n, kl,
456  $ ku, nrhs, a, lda, xact, lda, b, lda,
457  $ iseed, info )
458  xtype = 'C'
459  CALL zlacpy( 'Full', n, nrhs, b, lda, bsav, lda )
460 *
461  IF( nofact .AND. itran.EQ.1 ) THEN
462 *
463 * --- Test ZGESV ---
464 *
465 * Compute the LU factorization of the matrix and
466 * solve the system.
467 *
468  CALL zlacpy( 'Full', n, n, a, lda, afac, lda )
469  CALL zlacpy( 'Full', n, nrhs, b, lda, x, lda )
470 *
471  srnamt = 'ZGESV '
472  CALL zgesv( n, nrhs, afac, lda, iwork, x, lda,
473  $ info )
474 *
475 * Check error code from ZGESV .
476 *
477  IF( info.NE.izero )
478  $ CALL alaerh( path, 'ZGESV ', info, izero,
479  $ ' ', n, n, -1, -1, nrhs, imat,
480  $ nfail, nerrs, nout )
481 *
482 * Reconstruct matrix from factors and compute
483 * residual.
484 *
485  CALL zget01( n, n, a, lda, afac, lda, iwork,
486  $ rwork, result( 1 ) )
487  nt = 1
488  IF( izero.EQ.0 ) THEN
489 *
490 * Compute residual of the computed solution.
491 *
492  CALL zlacpy( 'Full', n, nrhs, b, lda, work,
493  $ lda )
494  CALL zget02( 'No transpose', n, n, nrhs, a,
495  $ lda, x, lda, work, lda, rwork,
496  $ result( 2 ) )
497 *
498 * Check solution from generated exact solution.
499 *
500  CALL zget04( n, nrhs, x, lda, xact, lda,
501  $ rcondc, result( 3 ) )
502  nt = 3
503  END IF
504 *
505 * Print information about the tests that did not
506 * pass the threshold.
507 *
508  DO 30 k = 1, nt
509  IF( result( k ).GE.thresh ) THEN
510  IF( nfail.EQ.0 .AND. nerrs.EQ.0 )
511  $ CALL aladhd( nout, path )
512  WRITE( nout, fmt = 9999 )'ZGESV ', n,
513  $ imat, k, result( k )
514  nfail = nfail + 1
515  END IF
516  30 CONTINUE
517  nrun = nrun + nt
518  END IF
519 *
520 * --- Test ZGESVX ---
521 *
522  IF( .NOT.prefac )
523  $ CALL zlaset( 'Full', n, n, dcmplx( zero ),
524  $ dcmplx( zero ), afac, lda )
525  CALL zlaset( 'Full', n, nrhs, dcmplx( zero ),
526  $ dcmplx( zero ), x, lda )
527  IF( iequed.GT.1 .AND. n.GT.0 ) THEN
528 *
529 * Equilibrate the matrix if FACT = 'F' and
530 * EQUED = 'R', 'C', or 'B'.
531 *
532  CALL zlaqge( n, n, a, lda, s, s( n+1 ), rowcnd,
533  $ colcnd, amax, equed )
534  END IF
535 *
536 * Solve the system and compute the condition number
537 * and error bounds using ZGESVX.
538 *
539  srnamt = 'ZGESVX'
540  CALL zgesvx( fact, trans, n, nrhs, a, lda, afac,
541  $ lda, iwork, equed, s, s( n+1 ), b,
542  $ lda, x, lda, rcond, rwork,
543  $ rwork( nrhs+1 ), work,
544  $ rwork( 2*nrhs+1 ), info )
545 *
546 * Check the error code from ZGESVX.
547 *
548  IF( info.NE.izero )
549  $ CALL alaerh( path, 'ZGESVX', info, izero,
550  $ fact // trans, n, n, -1, -1, nrhs,
551  $ imat, nfail, nerrs, nout )
552 *
553 * Compare RWORK(2*NRHS+1) from ZGESVX with the
554 * computed reciprocal pivot growth factor RPVGRW
555 *
556  IF( info.NE.0 .AND. info.LE.n) THEN
557  rpvgrw = zlantr( 'M', 'U', 'N', info, info,
558  $ afac, lda, rdum )
559  IF( rpvgrw.EQ.zero ) THEN
560  rpvgrw = one
561  ELSE
562  rpvgrw = zlange( 'M', n, info, a, lda,
563  $ rdum ) / rpvgrw
564  END IF
565  ELSE
566  rpvgrw = zlantr( 'M', 'U', 'N', n, n, afac, lda,
567  $ rdum )
568  IF( rpvgrw.EQ.zero ) THEN
569  rpvgrw = one
570  ELSE
571  rpvgrw = zlange( 'M', n, n, a, lda, rdum ) /
572  $ rpvgrw
573  END IF
574  END IF
575  result( 7 ) = abs( rpvgrw-rwork( 2*nrhs+1 ) ) /
576  $ max( rwork( 2*nrhs+1 ), rpvgrw ) /
577  $ dlamch( 'E' )
578 *
579  IF( .NOT.prefac ) THEN
580 *
581 * Reconstruct matrix from factors and compute
582 * residual.
583 *
584  CALL zget01( n, n, a, lda, afac, lda, iwork,
585  $ rwork( 2*nrhs+1 ), result( 1 ) )
586  k1 = 1
587  ELSE
588  k1 = 2
589  END IF
590 *
591  IF( info.EQ.0 ) THEN
592  trfcon = .false.
593 *
594 * Compute residual of the computed solution.
595 *
596  CALL zlacpy( 'Full', n, nrhs, bsav, lda, work,
597  $ lda )
598  CALL zget02( trans, n, n, nrhs, asav, lda, x,
599  $ lda, work, lda, rwork( 2*nrhs+1 ),
600  $ result( 2 ) )
601 *
602 * Check solution from generated exact solution.
603 *
604  IF( nofact .OR. ( prefac .AND. lsame( equed,
605  $ 'N' ) ) ) THEN
606  CALL zget04( n, nrhs, x, lda, xact, lda,
607  $ rcondc, result( 3 ) )
608  ELSE
609  IF( itran.EQ.1 ) THEN
610  roldc = roldo
611  ELSE
612  roldc = roldi
613  END IF
614  CALL zget04( n, nrhs, x, lda, xact, lda,
615  $ roldc, result( 3 ) )
616  END IF
617 *
618 * Check the error bounds from iterative
619 * refinement.
620 *
621  CALL zget07( trans, n, nrhs, asav, lda, b, lda,
622  $ x, lda, xact, lda, rwork, .true.,
623  $ rwork( nrhs+1 ), result( 4 ) )
624  ELSE
625  trfcon = .true.
626  END IF
627 *
628 * Compare RCOND from ZGESVX with the computed value
629 * in RCONDC.
630 *
631  result( 6 ) = dget06( rcond, rcondc )
632 *
633 * Print information about the tests that did not pass
634 * the threshold.
635 *
636  IF( .NOT.trfcon ) THEN
637  DO 40 k = k1, ntests
638  IF( result( k ).GE.thresh ) THEN
639  IF( nfail.EQ.0 .AND. nerrs.EQ.0 )
640  $ CALL aladhd( nout, path )
641  IF( prefac ) THEN
642  WRITE( nout, fmt = 9997 )'ZGESVX',
643  $ fact, trans, n, equed, imat, k,
644  $ result( k )
645  ELSE
646  WRITE( nout, fmt = 9998 )'ZGESVX',
647  $ fact, trans, n, imat, k, result( k )
648  END IF
649  nfail = nfail + 1
650  END IF
651  40 CONTINUE
652  nrun = nrun + ntests - k1 + 1
653  ELSE
654  IF( result( 1 ).GE.thresh .AND. .NOT.prefac )
655  $ THEN
656  IF( nfail.EQ.0 .AND. nerrs.EQ.0 )
657  $ CALL aladhd( nout, path )
658  IF( prefac ) THEN
659  WRITE( nout, fmt = 9997 )'ZGESVX', fact,
660  $ trans, n, equed, imat, 1, result( 1 )
661  ELSE
662  WRITE( nout, fmt = 9998 )'ZGESVX', fact,
663  $ trans, n, imat, 1, result( 1 )
664  END IF
665  nfail = nfail + 1
666  nrun = nrun + 1
667  END IF
668  IF( result( 6 ).GE.thresh ) THEN
669  IF( nfail.EQ.0 .AND. nerrs.EQ.0 )
670  $ CALL aladhd( nout, path )
671  IF( prefac ) THEN
672  WRITE( nout, fmt = 9997 )'ZGESVX', fact,
673  $ trans, n, equed, imat, 6, result( 6 )
674  ELSE
675  WRITE( nout, fmt = 9998 )'ZGESVX', fact,
676  $ trans, n, imat, 6, result( 6 )
677  END IF
678  nfail = nfail + 1
679  nrun = nrun + 1
680  END IF
681  IF( result( 7 ).GE.thresh ) THEN
682  IF( nfail.EQ.0 .AND. nerrs.EQ.0 )
683  $ CALL aladhd( nout, path )
684  IF( prefac ) THEN
685  WRITE( nout, fmt = 9997 )'ZGESVX', fact,
686  $ trans, n, equed, imat, 7, result( 7 )
687  ELSE
688  WRITE( nout, fmt = 9998 )'ZGESVX', fact,
689  $ trans, n, imat, 7, result( 7 )
690  END IF
691  nfail = nfail + 1
692  nrun = nrun + 1
693  END IF
694 *
695  END IF
696 *
697  50 CONTINUE
698  60 CONTINUE
699  70 CONTINUE
700  80 CONTINUE
701  90 CONTINUE
702 *
703 * Print a summary of the results.
704 *
705  CALL alasvm( path, nout, nfail, nrun, nerrs )
706 *
707  9999 FORMAT( 1x, a, ', N =', i5, ', type ', i2, ', test(', i2, ') =',
708  $ g12.5 )
709  9998 FORMAT( 1x, a, ', FACT=''', a1, ''', TRANS=''', a1, ''', N=', i5,
710  $ ', type ', i2, ', test(', i1, ')=', g12.5 )
711  9997 FORMAT( 1x, a, ', FACT=''', a1, ''', TRANS=''', a1, ''', N=', i5,
712  $ ', EQUED=''', a1, ''', type ', i2, ', test(', i1, ')=',
713  $ g12.5 )
714  RETURN
715 *
716 * End of ZDRVGE
717 *
718  END
subroutine alasvm(TYPE, NOUT, NFAIL, NRUN, NERRS)
ALASVM
Definition: alasvm.f:75
subroutine alaerh(PATH, SUBNAM, INFO, INFOE, OPTS, M, N, KL, KU, N5, IMAT, NFAIL, NERRS, NOUT)
ALAERH
Definition: alaerh.f:149
subroutine zlacpy(UPLO, M, N, A, LDA, B, LDB)
ZLACPY copies all or part of one two-dimensional array to another.
Definition: zlacpy.f:105
subroutine zget04(N, NRHS, X, LDX, XACT, LDXACT, RCOND, RESID)
ZGET04
Definition: zget04.f:104
subroutine zdrvge(DOTYPE, NN, NVAL, NRHS, THRESH, TSTERR, NMAX, A, AFAC, ASAV, B, BSAV, X, XACT, S, WORK, RWORK, IWORK, NOUT)
ZDRVGE
Definition: zdrvge.f:166
subroutine zgetrf(M, N, A, LDA, IPIV, INFO)
ZGETRF VARIANT: Crout Level 3 BLAS version of the algorithm.
Definition: zgetrf.f:102
subroutine zlarhs(PATH, XTYPE, UPLO, TRANS, M, N, KL, KU, NRHS, A, LDA, X, LDX, B, LDB, ISEED, INFO)
ZLARHS
Definition: zlarhs.f:211
subroutine zgesv(N, NRHS, A, LDA, IPIV, B, LDB, INFO)
ZGESV computes the solution to system of linear equations A * X = B for GE matrices (simple driver) ...
Definition: zgesv.f:124
subroutine zlaset(UPLO, M, N, ALPHA, BETA, A, LDA)
ZLASET initializes the off-diagonal elements and the diagonal elements of a matrix to given values...
Definition: zlaset.f:108
subroutine xlaenv(ISPEC, NVALUE)
XLAENV
Definition: xlaenv.f:83
subroutine zget07(TRANS, N, NRHS, A, LDA, B, LDB, X, LDX, XACT, LDXACT, FERR, CHKFERR, BERR, RESLTS)
ZGET07
Definition: zget07.f:168
subroutine zlatb4(PATH, IMAT, M, N, TYPE, KL, KU, ANORM, MODE, CNDNUM, DIST)
ZLATB4
Definition: zlatb4.f:123
subroutine zgesvx(FACT, TRANS, N, NRHS, A, LDA, AF, LDAF, IPIV, EQUED, R, C, B, LDB, X, LDX, RCOND, FERR, BERR, WORK, RWORK, INFO)
ZGESVX computes the solution to system of linear equations A * X = B for GE matrices ...
Definition: zgesvx.f:352
subroutine aladhd(IOUNIT, PATH)
ALADHD
Definition: aladhd.f:80
subroutine zgetri(N, A, LDA, IPIV, WORK, LWORK, INFO)
ZGETRI
Definition: zgetri.f:116
subroutine zerrvx(PATH, NUNIT)
ZERRVX
Definition: zerrvx.f:57
subroutine zlatms(M, N, DIST, ISEED, SYM, D, MODE, COND, DMAX, KL, KU, PACK, A, LDA, WORK, INFO)
ZLATMS
Definition: zlatms.f:334
subroutine zgeequ(M, N, A, LDA, R, C, ROWCND, COLCND, AMAX, INFO)
ZGEEQU
Definition: zgeequ.f:142
subroutine zlaqge(M, N, A, LDA, R, C, ROWCND, COLCND, AMAX, EQUED)
ZLAQGE scales a general rectangular matrix, using row and column scaling factors computed by sgeequ...
Definition: zlaqge.f:145
subroutine zget01(M, N, A, LDA, AFAC, LDAFAC, IPIV, RWORK, RESID)
ZGET01
Definition: zget01.f:110
subroutine zget02(TRANS, M, N, NRHS, A, LDA, X, LDX, B, LDB, RWORK, RESID)
ZGET02
Definition: zget02.f:135