 LAPACK  3.10.1 LAPACK: Linear Algebra PACKage

## ◆ sdrvpo()

 subroutine sdrvpo ( logical, dimension( * ) DOTYPE, integer NN, integer, dimension( * ) NVAL, integer NRHS, real THRESH, logical TSTERR, integer NMAX, real, dimension( * ) A, real, dimension( * ) AFAC, real, dimension( * ) ASAV, real, dimension( * ) B, real, dimension( * ) BSAV, real, dimension( * ) X, real, dimension( * ) XACT, real, dimension( * ) S, real, dimension( * ) WORK, real, dimension( * ) RWORK, integer, dimension( * ) IWORK, integer NOUT )

SDRVPO

SDRVPOX

Purpose:
` SDRVPO tests the driver routines SPOSV and -SVX.`
Parameters
 [in] DOTYPE ``` DOTYPE is LOGICAL array, dimension (NTYPES) The matrix types to be used for testing. Matrices of type j (for 1 <= j <= NTYPES) are used for testing if DOTYPE(j) = .TRUE.; if DOTYPE(j) = .FALSE., then type j is not used.``` [in] NN ``` NN is INTEGER The number of values of N contained in the vector NVAL.``` [in] NVAL ``` NVAL is INTEGER array, dimension (NN) The values of the matrix dimension N.``` [in] NRHS ``` NRHS is INTEGER The number of right hand side vectors to be generated for each linear system.``` [in] THRESH ``` THRESH is REAL The threshold value for the test ratios. A result is included in the output file if RESULT >= THRESH. To have every test ratio printed, use THRESH = 0.``` [in] TSTERR ``` TSTERR is LOGICAL Flag that indicates whether error exits are to be tested.``` [in] NMAX ``` NMAX is INTEGER The maximum value permitted for N, used in dimensioning the work arrays.``` [out] A ` A is REAL array, dimension (NMAX*NMAX)` [out] AFAC ` AFAC is REAL array, dimension (NMAX*NMAX)` [out] ASAV ` ASAV is REAL array, dimension (NMAX*NMAX)` [out] B ` B is REAL array, dimension (NMAX*NRHS)` [out] BSAV ` BSAV is REAL array, dimension (NMAX*NRHS)` [out] X ` X is REAL array, dimension (NMAX*NRHS)` [out] XACT ` XACT is REAL array, dimension (NMAX*NRHS)` [out] S ` S is REAL array, dimension (NMAX)` [out] WORK ``` WORK is REAL array, dimension (NMAX*max(3,NRHS))``` [out] RWORK ` RWORK is REAL array, dimension (NMAX+2*NRHS)` [out] IWORK ` IWORK is INTEGER array, dimension (NMAX)` [in] NOUT ``` NOUT is INTEGER The unit number for output.```
Purpose:
``` SDRVPO tests the driver routines SPOSV, -SVX, and -SVXX.

Note that this file is used only when the XBLAS are available,
otherwise sdrvpo.f defines this subroutine.```
Parameters
 [in] DOTYPE ``` DOTYPE is LOGICAL array, dimension (NTYPES) The matrix types to be used for testing. Matrices of type j (for 1 <= j <= NTYPES) are used for testing if DOTYPE(j) = .TRUE.; if DOTYPE(j) = .FALSE., then type j is not used.``` [in] NN ``` NN is INTEGER The number of values of N contained in the vector NVAL.``` [in] NVAL ``` NVAL is INTEGER array, dimension (NN) The values of the matrix dimension N.``` [in] NRHS ``` NRHS is INTEGER The number of right hand side vectors to be generated for each linear system.``` [in] THRESH ``` THRESH is REAL The threshold value for the test ratios. A result is included in the output file if RESULT >= THRESH. To have every test ratio printed, use THRESH = 0.``` [in] TSTERR ``` TSTERR is LOGICAL Flag that indicates whether error exits are to be tested.``` [in] NMAX ``` NMAX is INTEGER The maximum value permitted for N, used in dimensioning the work arrays.``` [out] A ` A is REAL array, dimension (NMAX*NMAX)` [out] AFAC ` AFAC is REAL array, dimension (NMAX*NMAX)` [out] ASAV ` ASAV is REAL array, dimension (NMAX*NMAX)` [out] B ` B is REAL array, dimension (NMAX*NRHS)` [out] BSAV ` BSAV is REAL array, dimension (NMAX*NRHS)` [out] X ` X is REAL array, dimension (NMAX*NRHS)` [out] XACT ` XACT is REAL array, dimension (NMAX*NRHS)` [out] S ` S is REAL array, dimension (NMAX)` [out] WORK ``` WORK is REAL array, dimension (NMAX*max(3,NRHS))``` [out] RWORK ` RWORK is REAL array, dimension (NMAX+2*NRHS)` [out] IWORK ` IWORK is INTEGER array, dimension (NMAX)` [in] NOUT ``` NOUT is INTEGER The unit number for output.```

Definition at line 161 of file sdrvpo.f.

164 *
165 * -- LAPACK test routine --
166 * -- LAPACK is a software package provided by Univ. of Tennessee, --
167 * -- Univ. of California Berkeley, Univ. of Colorado Denver and NAG Ltd..--
168 *
169 * .. Scalar Arguments ..
170  LOGICAL TSTERR
171  INTEGER NMAX, NN, NOUT, NRHS
172  REAL THRESH
173 * ..
174 * .. Array Arguments ..
175  LOGICAL DOTYPE( * )
176  INTEGER IWORK( * ), NVAL( * )
177  REAL A( * ), AFAC( * ), ASAV( * ), B( * ),
178  \$ BSAV( * ), RWORK( * ), S( * ), WORK( * ),
179  \$ X( * ), XACT( * )
180 * ..
181 *
182 * =====================================================================
183 *
184 * .. Parameters ..
185  REAL ONE, ZERO
186  parameter( one = 1.0e+0, zero = 0.0e+0 )
187  INTEGER NTYPES
188  parameter( ntypes = 9 )
189  INTEGER NTESTS
190  parameter( ntests = 6 )
191 * ..
192 * .. Local Scalars ..
193  LOGICAL EQUIL, NOFACT, PREFAC, ZEROT
194  CHARACTER DIST, EQUED, FACT, TYPE, UPLO, XTYPE
195  CHARACTER*3 PATH
196  INTEGER I, IEQUED, IFACT, IMAT, IN, INFO, IOFF, IUPLO,
197  \$ IZERO, K, K1, KL, KU, LDA, MODE, N, NB, NBMIN,
198  \$ NERRS, NFACT, NFAIL, NIMAT, NRUN, NT
199  REAL AINVNM, AMAX, ANORM, CNDNUM, RCOND, RCONDC,
200  \$ ROLDC, SCOND
201 * ..
202 * .. Local Arrays ..
203  CHARACTER EQUEDS( 2 ), FACTS( 3 ), UPLOS( 2 )
204  INTEGER ISEED( 4 ), ISEEDY( 4 )
205  REAL RESULT( NTESTS )
206 * ..
207 * .. External Functions ..
208  LOGICAL LSAME
209  REAL SGET06, SLANSY
210  EXTERNAL lsame, sget06, slansy
211 * ..
212 * .. External Subroutines ..
213  EXTERNAL aladhd, alaerh, alasvm, serrvx, sget04, slacpy,
216  \$ spotri, xlaenv
217 * ..
218 * .. Intrinsic Functions ..
219  INTRINSIC max
220 * ..
221 * .. Scalars in Common ..
222  LOGICAL LERR, OK
223  CHARACTER*32 SRNAMT
224  INTEGER INFOT, NUNIT
225 * ..
226 * .. Common blocks ..
227  COMMON / infoc / infot, nunit, ok, lerr
228  COMMON / srnamc / srnamt
229 * ..
230 * .. Data statements ..
231  DATA iseedy / 1988, 1989, 1990, 1991 /
232  DATA uplos / 'U', 'L' /
233  DATA facts / 'F', 'N', 'E' /
234  DATA equeds / 'N', 'Y' /
235 * ..
236 * .. Executable Statements ..
237 *
238 * Initialize constants and the random number seed.
239 *
240  path( 1: 1 ) = 'Single precision'
241  path( 2: 3 ) = 'PO'
242  nrun = 0
243  nfail = 0
244  nerrs = 0
245  DO 10 i = 1, 4
246  iseed( i ) = iseedy( i )
247  10 CONTINUE
248 *
249 * Test the error exits
250 *
251  IF( tsterr )
252  \$ CALL serrvx( path, nout )
253  infot = 0
254 *
255 * Set the block size and minimum block size for testing.
256 *
257  nb = 1
258  nbmin = 2
259  CALL xlaenv( 1, nb )
260  CALL xlaenv( 2, nbmin )
261 *
262 * Do for each value of N in NVAL
263 *
264  DO 130 in = 1, nn
265  n = nval( in )
266  lda = max( n, 1 )
267  xtype = 'N'
268  nimat = ntypes
269  IF( n.LE.0 )
270  \$ nimat = 1
271 *
272  DO 120 imat = 1, nimat
273 *
274 * Do the tests only if DOTYPE( IMAT ) is true.
275 *
276  IF( .NOT.dotype( imat ) )
277  \$ GO TO 120
278 *
279 * Skip types 3, 4, or 5 if the matrix size is too small.
280 *
281  zerot = imat.GE.3 .AND. imat.LE.5
282  IF( zerot .AND. n.LT.imat-2 )
283  \$ GO TO 120
284 *
285 * Do first for UPLO = 'U', then for UPLO = 'L'
286 *
287  DO 110 iuplo = 1, 2
288  uplo = uplos( iuplo )
289 *
290 * Set up parameters with SLATB4 and generate a test matrix
291 * with SLATMS.
292 *
293  CALL slatb4( path, imat, n, n, TYPE, KL, KU, ANORM, MODE,
294  \$ CNDNUM, DIST )
295 *
296  srnamt = 'SLATMS'
297  CALL slatms( n, n, dist, iseed, TYPE, RWORK, MODE,
298  \$ CNDNUM, ANORM, KL, KU, UPLO, A, LDA, WORK,
299  \$ INFO )
300 *
301 * Check error code from SLATMS.
302 *
303  IF( info.NE.0 ) THEN
304  CALL alaerh( path, 'SLATMS', info, 0, uplo, n, n, -1,
305  \$ -1, -1, imat, nfail, nerrs, nout )
306  GO TO 110
307  END IF
308 *
309 * For types 3-5, zero one row and column of the matrix to
310 * test that INFO is returned correctly.
311 *
312  IF( zerot ) THEN
313  IF( imat.EQ.3 ) THEN
314  izero = 1
315  ELSE IF( imat.EQ.4 ) THEN
316  izero = n
317  ELSE
318  izero = n / 2 + 1
319  END IF
320  ioff = ( izero-1 )*lda
321 *
322 * Set row and column IZERO of A to 0.
323 *
324  IF( iuplo.EQ.1 ) THEN
325  DO 20 i = 1, izero - 1
326  a( ioff+i ) = zero
327  20 CONTINUE
328  ioff = ioff + izero
329  DO 30 i = izero, n
330  a( ioff ) = zero
331  ioff = ioff + lda
332  30 CONTINUE
333  ELSE
334  ioff = izero
335  DO 40 i = 1, izero - 1
336  a( ioff ) = zero
337  ioff = ioff + lda
338  40 CONTINUE
339  ioff = ioff - izero
340  DO 50 i = izero, n
341  a( ioff+i ) = zero
342  50 CONTINUE
343  END IF
344  ELSE
345  izero = 0
346  END IF
347 *
348 * Save a copy of the matrix A in ASAV.
349 *
350  CALL slacpy( uplo, n, n, a, lda, asav, lda )
351 *
352  DO 100 iequed = 1, 2
353  equed = equeds( iequed )
354  IF( iequed.EQ.1 ) THEN
355  nfact = 3
356  ELSE
357  nfact = 1
358  END IF
359 *
360  DO 90 ifact = 1, nfact
361  fact = facts( ifact )
362  prefac = lsame( fact, 'F' )
363  nofact = lsame( fact, 'N' )
364  equil = lsame( fact, 'E' )
365 *
366  IF( zerot ) THEN
367  IF( prefac )
368  \$ GO TO 90
369  rcondc = zero
370 *
371  ELSE IF( .NOT.lsame( fact, 'N' ) ) THEN
372 *
373 * Compute the condition number for comparison with
374 * the value returned by SPOSVX (FACT = 'N' reuses
375 * the condition number from the previous iteration
376 * with FACT = 'F').
377 *
378  CALL slacpy( uplo, n, n, asav, lda, afac, lda )
379  IF( equil .OR. iequed.GT.1 ) THEN
380 *
381 * Compute row and column scale factors to
382 * equilibrate the matrix A.
383 *
384  CALL spoequ( n, afac, lda, s, scond, amax,
385  \$ info )
386  IF( info.EQ.0 .AND. n.GT.0 ) THEN
387  IF( iequed.GT.1 )
388  \$ scond = zero
389 *
390 * Equilibrate the matrix.
391 *
392  CALL slaqsy( uplo, n, afac, lda, s, scond,
393  \$ amax, equed )
394  END IF
395  END IF
396 *
397 * Save the condition number of the
398 * non-equilibrated system for use in SGET04.
399 *
400  IF( equil )
401  \$ roldc = rcondc
402 *
403 * Compute the 1-norm of A.
404 *
405  anorm = slansy( '1', uplo, n, afac, lda, rwork )
406 *
407 * Factor the matrix A.
408 *
409  CALL spotrf( uplo, n, afac, lda, info )
410 *
411 * Form the inverse of A.
412 *
413  CALL slacpy( uplo, n, n, afac, lda, a, lda )
414  CALL spotri( uplo, n, a, lda, info )
415 *
416 * Compute the 1-norm condition number of A.
417 *
418  ainvnm = slansy( '1', uplo, n, a, lda, rwork )
419  IF( anorm.LE.zero .OR. ainvnm.LE.zero ) THEN
420  rcondc = one
421  ELSE
422  rcondc = ( one / anorm ) / ainvnm
423  END IF
424  END IF
425 *
426 * Restore the matrix A.
427 *
428  CALL slacpy( uplo, n, n, asav, lda, a, lda )
429 *
430 * Form an exact solution and set the right hand side.
431 *
432  srnamt = 'SLARHS'
433  CALL slarhs( path, xtype, uplo, ' ', n, n, kl, ku,
434  \$ nrhs, a, lda, xact, lda, b, lda,
435  \$ iseed, info )
436  xtype = 'C'
437  CALL slacpy( 'Full', n, nrhs, b, lda, bsav, lda )
438 *
439  IF( nofact ) THEN
440 *
441 * --- Test SPOSV ---
442 *
443 * Compute the L*L' or U'*U factorization of the
444 * matrix and solve the system.
445 *
446  CALL slacpy( uplo, n, n, a, lda, afac, lda )
447  CALL slacpy( 'Full', n, nrhs, b, lda, x, lda )
448 *
449  srnamt = 'SPOSV '
450  CALL sposv( uplo, n, nrhs, afac, lda, x, lda,
451  \$ info )
452 *
453 * Check error code from SPOSV .
454 *
455  IF( info.NE.izero ) THEN
456  CALL alaerh( path, 'SPOSV ', info, izero,
457  \$ uplo, n, n, -1, -1, nrhs, imat,
458  \$ nfail, nerrs, nout )
459  GO TO 70
460  ELSE IF( info.NE.0 ) THEN
461  GO TO 70
462  END IF
463 *
464 * Reconstruct matrix from factors and compute
465 * residual.
466 *
467  CALL spot01( uplo, n, a, lda, afac, lda, rwork,
468  \$ result( 1 ) )
469 *
470 * Compute residual of the computed solution.
471 *
472  CALL slacpy( 'Full', n, nrhs, b, lda, work,
473  \$ lda )
474  CALL spot02( uplo, n, nrhs, a, lda, x, lda,
475  \$ work, lda, rwork, result( 2 ) )
476 *
477 * Check solution from generated exact solution.
478 *
479  CALL sget04( n, nrhs, x, lda, xact, lda, rcondc,
480  \$ result( 3 ) )
481  nt = 3
482 *
483 * Print information about the tests that did not
484 * pass the threshold.
485 *
486  DO 60 k = 1, nt
487  IF( result( k ).GE.thresh ) THEN
488  IF( nfail.EQ.0 .AND. nerrs.EQ.0 )
489  \$ CALL aladhd( nout, path )
490  WRITE( nout, fmt = 9999 )'SPOSV ', uplo,
491  \$ n, imat, k, result( k )
492  nfail = nfail + 1
493  END IF
494  60 CONTINUE
495  nrun = nrun + nt
496  70 CONTINUE
497  END IF
498 *
499 * --- Test SPOSVX ---
500 *
501  IF( .NOT.prefac )
502  \$ CALL slaset( uplo, n, n, zero, zero, afac, lda )
503  CALL slaset( 'Full', n, nrhs, zero, zero, x, lda )
504  IF( iequed.GT.1 .AND. n.GT.0 ) THEN
505 *
506 * Equilibrate the matrix if FACT='F' and
507 * EQUED='Y'.
508 *
509  CALL slaqsy( uplo, n, a, lda, s, scond, amax,
510  \$ equed )
511  END IF
512 *
513 * Solve the system and compute the condition number
514 * and error bounds using SPOSVX.
515 *
516  srnamt = 'SPOSVX'
517  CALL sposvx( fact, uplo, n, nrhs, a, lda, afac,
518  \$ lda, equed, s, b, lda, x, lda, rcond,
519  \$ rwork, rwork( nrhs+1 ), work, iwork,
520  \$ info )
521 *
522 * Check the error code from SPOSVX.
523 *
524  IF( info.NE.izero ) THEN
525  CALL alaerh( path, 'SPOSVX', info, izero,
526  \$ fact // uplo, n, n, -1, -1, nrhs,
527  \$ imat, nfail, nerrs, nout )
528  GO TO 90
529  END IF
530 *
531  IF( info.EQ.0 ) THEN
532  IF( .NOT.prefac ) THEN
533 *
534 * Reconstruct matrix from factors and compute
535 * residual.
536 *
537  CALL spot01( uplo, n, a, lda, afac, lda,
538  \$ rwork( 2*nrhs+1 ), result( 1 ) )
539  k1 = 1
540  ELSE
541  k1 = 2
542  END IF
543 *
544 * Compute residual of the computed solution.
545 *
546  CALL slacpy( 'Full', n, nrhs, bsav, lda, work,
547  \$ lda )
548  CALL spot02( uplo, n, nrhs, asav, lda, x, lda,
549  \$ work, lda, rwork( 2*nrhs+1 ),
550  \$ result( 2 ) )
551 *
552 * Check solution from generated exact solution.
553 *
554  IF( nofact .OR. ( prefac .AND. lsame( equed,
555  \$ 'N' ) ) ) THEN
556  CALL sget04( n, nrhs, x, lda, xact, lda,
557  \$ rcondc, result( 3 ) )
558  ELSE
559  CALL sget04( n, nrhs, x, lda, xact, lda,
560  \$ roldc, result( 3 ) )
561  END IF
562 *
563 * Check the error bounds from iterative
564 * refinement.
565 *
566  CALL spot05( uplo, n, nrhs, asav, lda, b, lda,
567  \$ x, lda, xact, lda, rwork,
568  \$ rwork( nrhs+1 ), result( 4 ) )
569  ELSE
570  k1 = 6
571  END IF
572 *
573 * Compare RCOND from SPOSVX with the computed value
574 * in RCONDC.
575 *
576  result( 6 ) = sget06( rcond, rcondc )
577 *
578 * Print information about the tests that did not pass
579 * the threshold.
580 *
581  DO 80 k = k1, 6
582  IF( result( k ).GE.thresh ) THEN
583  IF( nfail.EQ.0 .AND. nerrs.EQ.0 )
584  \$ CALL aladhd( nout, path )
585  IF( prefac ) THEN
586  WRITE( nout, fmt = 9997 )'SPOSVX', fact,
587  \$ uplo, n, equed, imat, k, result( k )
588  ELSE
589  WRITE( nout, fmt = 9998 )'SPOSVX', fact,
590  \$ uplo, n, imat, k, result( k )
591  END IF
592  nfail = nfail + 1
593  END IF
594  80 CONTINUE
595  nrun = nrun + 7 - k1
596  90 CONTINUE
597  100 CONTINUE
598  110 CONTINUE
599  120 CONTINUE
600  130 CONTINUE
601 *
602 * Print a summary of the results.
603 *
604  CALL alasvm( path, nout, nfail, nrun, nerrs )
605 *
606  9999 FORMAT( 1x, a, ', UPLO=''', a1, ''', N =', i5, ', type ', i1,
607  \$ ', test(', i1, ')=', g12.5 )
608  9998 FORMAT( 1x, a, ', FACT=''', a1, ''', UPLO=''', a1, ''', N=', i5,
609  \$ ', type ', i1, ', test(', i1, ')=', g12.5 )
610  9997 FORMAT( 1x, a, ', FACT=''', a1, ''', UPLO=''', a1, ''', N=', i5,
611  \$ ', EQUED=''', a1, ''', type ', i1, ', test(', i1, ') =',
612  \$ g12.5 )
613  RETURN
614 *
615 * End of SDRVPO
616 *
subroutine slaset(UPLO, M, N, ALPHA, BETA, A, LDA)
SLASET initializes the off-diagonal elements and the diagonal elements of a matrix to given values.
Definition: slaset.f:110
subroutine slacpy(UPLO, M, N, A, LDA, B, LDB)
SLACPY copies all or part of one two-dimensional array to another.
Definition: slacpy.f:103
logical function lsame(CA, CB)
LSAME
Definition: lsame.f:53
subroutine alasvm(TYPE, NOUT, NFAIL, NRUN, NERRS)
ALASVM
Definition: alasvm.f:73
subroutine xlaenv(ISPEC, NVALUE)
XLAENV
Definition: xlaenv.f:81
subroutine alaerh(PATH, SUBNAM, INFO, INFOE, OPTS, M, N, KL, KU, N5, IMAT, NFAIL, NERRS, NOUT)
ALAERH
Definition: alaerh.f:147
subroutine slatms(M, N, DIST, ISEED, SYM, D, MODE, COND, DMAX, KL, KU, PACK, A, LDA, WORK, INFO)
SLATMS
Definition: slatms.f:321
subroutine spotri(UPLO, N, A, LDA, INFO)
SPOTRI
Definition: spotri.f:95
subroutine spotrf(UPLO, N, A, LDA, INFO)
SPOTRF
Definition: spotrf.f:107
subroutine spoequ(N, A, LDA, S, SCOND, AMAX, INFO)
SPOEQU
Definition: spoequ.f:112
subroutine sposvx(FACT, UPLO, N, NRHS, A, LDA, AF, LDAF, EQUED, S, B, LDB, X, LDX, RCOND, FERR, BERR, WORK, IWORK, INFO)
SPOSVX computes the solution to system of linear equations A * X = B for PO matrices
Definition: sposvx.f:307
subroutine sposv(UPLO, N, NRHS, A, LDA, B, LDB, INFO)
SPOSV computes the solution to system of linear equations A * X = B for PO matrices
Definition: sposv.f:130
real function slansy(NORM, UPLO, N, A, LDA, WORK)
SLANSY returns the value of the 1-norm, or the Frobenius norm, or the infinity norm,...
Definition: slansy.f:122
subroutine slaqsy(UPLO, N, A, LDA, S, SCOND, AMAX, EQUED)
SLAQSY scales a symmetric/Hermitian matrix, using scaling factors computed by spoequ.
Definition: slaqsy.f:133
subroutine slarhs(PATH, XTYPE, UPLO, TRANS, M, N, KL, KU, NRHS, A, LDA, X, LDX, B, LDB, ISEED, INFO)
SLARHS
Definition: slarhs.f:205
subroutine slatb4(PATH, IMAT, M, N, TYPE, KL, KU, ANORM, MODE, CNDNUM, DIST)
SLATB4
Definition: slatb4.f:120
subroutine spot01(UPLO, N, A, LDA, AFAC, LDAFAC, RWORK, RESID)
SPOT01
Definition: spot01.f:104
subroutine spot05(UPLO, N, NRHS, A, LDA, B, LDB, X, LDX, XACT, LDXACT, FERR, BERR, RESLTS)
SPOT05
Definition: spot05.f:164
subroutine serrvx(PATH, NUNIT)
SERRVX
Definition: serrvx.f:55
subroutine sget04(N, NRHS, X, LDX, XACT, LDXACT, RCOND, RESID)
SGET04
Definition: sget04.f:102
subroutine spot02(UPLO, N, NRHS, A, LDA, X, LDX, B, LDB, RWORK, RESID)
SPOT02
Definition: spot02.f:127
real function sget06(RCOND, RCONDC)
SGET06
Definition: sget06.f:55
Here is the call graph for this function:
Here is the caller graph for this function: