LAPACK 3.12.0 LAPACK: Linear Algebra PACKage
Searching...
No Matches

## ◆ dchkpp()

 subroutine dchkpp ( logical, dimension( * ) dotype, integer nn, integer, dimension( * ) nval, integer nns, integer, dimension( * ) nsval, double precision thresh, logical tsterr, integer nmax, double precision, dimension( * ) a, double precision, dimension( * ) afac, double precision, dimension( * ) ainv, double precision, dimension( * ) b, double precision, dimension( * ) x, double precision, dimension( * ) xact, double precision, dimension( * ) work, double precision, dimension( * ) rwork, integer, dimension( * ) iwork, integer nout )

DCHKPP

Purpose:
` DCHKPP tests DPPTRF, -TRI, -TRS, -RFS, and -CON`
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] NNS ``` NNS is INTEGER The number of values of NRHS contained in the vector NSVAL.``` [in] NSVAL ``` NSVAL is INTEGER array, dimension (NNS) The values of the number of right hand sides NRHS.``` [in] THRESH ``` THRESH is DOUBLE PRECISION 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 DOUBLE PRECISION array, dimension (NMAX*(NMAX+1)/2)``` [out] AFAC ``` AFAC is DOUBLE PRECISION array, dimension (NMAX*(NMAX+1)/2)``` [out] AINV ``` AINV is DOUBLE PRECISION array, dimension (NMAX*(NMAX+1)/2)``` [out] B ``` B is DOUBLE PRECISION array, dimension (NMAX*NSMAX) where NSMAX is the largest entry in NSVAL.``` [out] X ` X is DOUBLE PRECISION array, dimension (NMAX*NSMAX)` [out] XACT ` XACT is DOUBLE PRECISION array, dimension (NMAX*NSMAX)` [out] WORK ``` WORK is DOUBLE PRECISION array, dimension (NMAX*max(3,NSMAX))``` [out] RWORK ``` RWORK is DOUBLE PRECISION array, dimension (max(NMAX,2*NSMAX))``` [out] IWORK ` IWORK is INTEGER array, dimension (NMAX)` [in] NOUT ``` NOUT is INTEGER The unit number for output.```

Definition at line 160 of file dchkpp.f.

163*
164* -- LAPACK test routine --
165* -- LAPACK is a software package provided by Univ. of Tennessee, --
166* -- Univ. of California Berkeley, Univ. of Colorado Denver and NAG Ltd..--
167*
168* .. Scalar Arguments ..
169 LOGICAL TSTERR
170 INTEGER NMAX, NN, NNS, NOUT
171 DOUBLE PRECISION THRESH
172* ..
173* .. Array Arguments ..
174 LOGICAL DOTYPE( * )
175 INTEGER IWORK( * ), NSVAL( * ), NVAL( * )
176 DOUBLE PRECISION A( * ), AFAC( * ), AINV( * ), B( * ),
177 \$ RWORK( * ), WORK( * ), X( * ), XACT( * )
178* ..
179*
180* =====================================================================
181*
182* .. Parameters ..
183 DOUBLE PRECISION ZERO
184 parameter( zero = 0.0d+0 )
185 INTEGER NTYPES
186 parameter( ntypes = 9 )
187 INTEGER NTESTS
188 parameter( ntests = 8 )
189* ..
190* .. Local Scalars ..
191 LOGICAL ZEROT
192 CHARACTER DIST, PACKIT, TYPE, UPLO, XTYPE
193 CHARACTER*3 PATH
194 INTEGER I, IMAT, IN, INFO, IOFF, IRHS, IUPLO, IZERO, K,
195 \$ KL, KU, LDA, MODE, N, NERRS, NFAIL, NIMAT, NPP,
196 \$ NRHS, NRUN
197 DOUBLE PRECISION ANORM, CNDNUM, RCOND, RCONDC
198* ..
199* .. Local Arrays ..
200 CHARACTER PACKS( 2 ), UPLOS( 2 )
201 INTEGER ISEED( 4 ), ISEEDY( 4 )
202 DOUBLE PRECISION RESULT( NTESTS )
203* ..
204* .. External Functions ..
205 DOUBLE PRECISION DGET06, DLANSP
206 EXTERNAL dget06, dlansp
207* ..
208* .. External Subroutines ..
209 EXTERNAL alaerh, alahd, alasum, dcopy, derrpo, dget04,
212 \$ dpptrs
213* ..
214* .. Scalars in Common ..
215 LOGICAL LERR, OK
216 CHARACTER*32 SRNAMT
217 INTEGER INFOT, NUNIT
218* ..
219* .. Common blocks ..
220 COMMON / infoc / infot, nunit, ok, lerr
221 COMMON / srnamc / srnamt
222* ..
223* .. Intrinsic Functions ..
224 INTRINSIC max
225* ..
226* .. Data statements ..
227 DATA iseedy / 1988, 1989, 1990, 1991 /
228 DATA uplos / 'U', 'L' / , packs / 'C', 'R' /
229* ..
230* .. Executable Statements ..
231*
232* Initialize constants and the random number seed.
233*
234 path( 1: 1 ) = 'Double precision'
235 path( 2: 3 ) = 'PP'
236 nrun = 0
237 nfail = 0
238 nerrs = 0
239 DO 10 i = 1, 4
240 iseed( i ) = iseedy( i )
241 10 CONTINUE
242*
243* Test the error exits
244*
245 IF( tsterr )
246 \$ CALL derrpo( path, nout )
247 infot = 0
248*
249* Do for each value of N in NVAL
250*
251 DO 110 in = 1, nn
252 n = nval( in )
253 lda = max( n, 1 )
254 xtype = 'N'
255 nimat = ntypes
256 IF( n.LE.0 )
257 \$ nimat = 1
258*
259 DO 100 imat = 1, nimat
260*
261* Do the tests only if DOTYPE( IMAT ) is true.
262*
263 IF( .NOT.dotype( imat ) )
264 \$ GO TO 100
265*
266* Skip types 3, 4, or 5 if the matrix size is too small.
267*
268 zerot = imat.GE.3 .AND. imat.LE.5
269 IF( zerot .AND. n.LT.imat-2 )
270 \$ GO TO 100
271*
272* Do first for UPLO = 'U', then for UPLO = 'L'
273*
274 DO 90 iuplo = 1, 2
275 uplo = uplos( iuplo )
276 packit = packs( iuplo )
277*
278* Set up parameters with DLATB4 and generate a test matrix
279* with DLATMS.
280*
281 CALL dlatb4( path, imat, n, n, TYPE, KL, KU, ANORM, MODE,
282 \$ CNDNUM, DIST )
283*
284 srnamt = 'DLATMS'
285 CALL dlatms( n, n, dist, iseed, TYPE, RWORK, MODE,
286 \$ CNDNUM, ANORM, KL, KU, PACKIT, A, LDA, WORK,
287 \$ INFO )
288*
289* Check error code from DLATMS.
290*
291 IF( info.NE.0 ) THEN
292 CALL alaerh( path, 'DLATMS', info, 0, uplo, n, n, -1,
293 \$ -1, -1, imat, nfail, nerrs, nout )
294 GO TO 90
295 END IF
296*
297* For types 3-5, zero one row and column of the matrix to
298* test that INFO is returned correctly.
299*
300 IF( zerot ) THEN
301 IF( imat.EQ.3 ) THEN
302 izero = 1
303 ELSE IF( imat.EQ.4 ) THEN
304 izero = n
305 ELSE
306 izero = n / 2 + 1
307 END IF
308*
309* Set row and column IZERO of A to 0.
310*
311 IF( iuplo.EQ.1 ) THEN
312 ioff = ( izero-1 )*izero / 2
313 DO 20 i = 1, izero - 1
314 a( ioff+i ) = zero
315 20 CONTINUE
316 ioff = ioff + izero
317 DO 30 i = izero, n
318 a( ioff ) = zero
319 ioff = ioff + i
320 30 CONTINUE
321 ELSE
322 ioff = izero
323 DO 40 i = 1, izero - 1
324 a( ioff ) = zero
325 ioff = ioff + n - i
326 40 CONTINUE
327 ioff = ioff - izero
328 DO 50 i = izero, n
329 a( ioff+i ) = zero
330 50 CONTINUE
331 END IF
332 ELSE
333 izero = 0
334 END IF
335*
336* Compute the L*L' or U'*U factorization of the matrix.
337*
338 npp = n*( n+1 ) / 2
339 CALL dcopy( npp, a, 1, afac, 1 )
340 srnamt = 'DPPTRF'
341 CALL dpptrf( uplo, n, afac, info )
342*
343* Check error code from DPPTRF.
344*
345 IF( info.NE.izero ) THEN
346 CALL alaerh( path, 'DPPTRF', info, izero, uplo, n, n,
347 \$ -1, -1, -1, imat, nfail, nerrs, nout )
348 GO TO 90
349 END IF
350*
351* Skip the tests if INFO is not 0.
352*
353 IF( info.NE.0 )
354 \$ GO TO 90
355*
356*+ TEST 1
357* Reconstruct matrix from factors and compute residual.
358*
359 CALL dcopy( npp, afac, 1, ainv, 1 )
360 CALL dppt01( uplo, n, a, ainv, rwork, result( 1 ) )
361*
362*+ TEST 2
363* Form the inverse and compute the residual.
364*
365 CALL dcopy( npp, afac, 1, ainv, 1 )
366 srnamt = 'DPPTRI'
367 CALL dpptri( uplo, n, ainv, info )
368*
369* Check error code from DPPTRI.
370*
371 IF( info.NE.0 )
372 \$ CALL alaerh( path, 'DPPTRI', info, 0, uplo, n, n, -1,
373 \$ -1, -1, imat, nfail, nerrs, nout )
374*
375 CALL dppt03( uplo, n, a, ainv, work, lda, rwork, rcondc,
376 \$ result( 2 ) )
377*
378* Print information about the tests that did not pass
379* the threshold.
380*
381 DO 60 k = 1, 2
382 IF( result( k ).GE.thresh ) THEN
383 IF( nfail.EQ.0 .AND. nerrs.EQ.0 )
384 \$ CALL alahd( nout, path )
385 WRITE( nout, fmt = 9999 )uplo, n, imat, k,
386 \$ result( k )
387 nfail = nfail + 1
388 END IF
389 60 CONTINUE
390 nrun = nrun + 2
391*
392 DO 80 irhs = 1, nns
393 nrhs = nsval( irhs )
394*
395*+ TEST 3
396* Solve and compute residual for A * X = B.
397*
398 srnamt = 'DLARHS'
399 CALL dlarhs( path, xtype, uplo, ' ', n, n, kl, ku,
400 \$ nrhs, a, lda, xact, lda, b, lda, iseed,
401 \$ info )
402 CALL dlacpy( 'Full', n, nrhs, b, lda, x, lda )
403*
404 srnamt = 'DPPTRS'
405 CALL dpptrs( uplo, n, nrhs, afac, x, lda, info )
406*
407* Check error code from DPPTRS.
408*
409 IF( info.NE.0 )
410 \$ CALL alaerh( path, 'DPPTRS', info, 0, uplo, n, n,
411 \$ -1, -1, nrhs, imat, nfail, nerrs,
412 \$ nout )
413*
414 CALL dlacpy( 'Full', n, nrhs, b, lda, work, lda )
415 CALL dppt02( uplo, n, nrhs, a, x, lda, work, lda,
416 \$ rwork, result( 3 ) )
417*
418*+ TEST 4
419* Check solution from generated exact solution.
420*
421 CALL dget04( n, nrhs, x, lda, xact, lda, rcondc,
422 \$ result( 4 ) )
423*
424*+ TESTS 5, 6, and 7
425* Use iterative refinement to improve the solution.
426*
427 srnamt = 'DPPRFS'
428 CALL dpprfs( uplo, n, nrhs, a, afac, b, lda, x, lda,
429 \$ rwork, rwork( nrhs+1 ), work, iwork,
430 \$ info )
431*
432* Check error code from DPPRFS.
433*
434 IF( info.NE.0 )
435 \$ CALL alaerh( path, 'DPPRFS', info, 0, uplo, n, n,
436 \$ -1, -1, nrhs, imat, nfail, nerrs,
437 \$ nout )
438*
439 CALL dget04( n, nrhs, x, lda, xact, lda, rcondc,
440 \$ result( 5 ) )
441 CALL dppt05( uplo, n, nrhs, a, b, lda, x, lda, xact,
442 \$ lda, rwork, rwork( nrhs+1 ),
443 \$ result( 6 ) )
444*
445* Print information about the tests that did not pass
446* the threshold.
447*
448 DO 70 k = 3, 7
449 IF( result( k ).GE.thresh ) THEN
450 IF( nfail.EQ.0 .AND. nerrs.EQ.0 )
451 \$ CALL alahd( nout, path )
452 WRITE( nout, fmt = 9998 )uplo, n, nrhs, imat,
453 \$ k, result( k )
454 nfail = nfail + 1
455 END IF
456 70 CONTINUE
457 nrun = nrun + 5
458 80 CONTINUE
459*
460*+ TEST 8
461* Get an estimate of RCOND = 1/CNDNUM.
462*
463 anorm = dlansp( '1', uplo, n, a, rwork )
464 srnamt = 'DPPCON'
465 CALL dppcon( uplo, n, afac, anorm, rcond, work, iwork,
466 \$ info )
467*
468* Check error code from DPPCON.
469*
470 IF( info.NE.0 )
471 \$ CALL alaerh( path, 'DPPCON', info, 0, uplo, n, n, -1,
472 \$ -1, -1, imat, nfail, nerrs, nout )
473*
474 result( 8 ) = dget06( rcond, rcondc )
475*
476* Print the test ratio if greater than or equal to THRESH.
477*
478 IF( result( 8 ).GE.thresh ) THEN
479 IF( nfail.EQ.0 .AND. nerrs.EQ.0 )
480 \$ CALL alahd( nout, path )
481 WRITE( nout, fmt = 9999 )uplo, n, imat, 8,
482 \$ result( 8 )
483 nfail = nfail + 1
484 END IF
485 nrun = nrun + 1
486 90 CONTINUE
487 100 CONTINUE
488 110 CONTINUE
489*
490* Print a summary of the results.
491*
492 CALL alasum( path, nout, nfail, nrun, nerrs )
493*
494 9999 FORMAT( ' UPLO = ''', a1, ''', N =', i5, ', type ', i2, ', test ',
495 \$ i2, ', ratio =', g12.5 )
496 9998 FORMAT( ' UPLO = ''', a1, ''', N =', i5, ', NRHS=', i3, ', type ',
497 \$ i2, ', test(', i2, ') =', g12.5 )
498 RETURN
499*
500* End of DCHKPP
501*
subroutine alasum(type, nout, nfail, nrun, nerrs)
ALASUM
Definition alasum.f:73
subroutine dlarhs(path, xtype, uplo, trans, m, n, kl, ku, nrhs, a, lda, x, ldx, b, ldb, iseed, info)
DLARHS
Definition dlarhs.f:205
subroutine alaerh(path, subnam, info, infoe, opts, m, n, kl, ku, n5, imat, nfail, nerrs, nout)
ALAERH
Definition alaerh.f:147
subroutine alahd(iounit, path)
ALAHD
Definition alahd.f:107
subroutine derrpo(path, nunit)
DERRPO
Definition derrpo.f:55
subroutine dget04(n, nrhs, x, ldx, xact, ldxact, rcond, resid)
DGET04
Definition dget04.f:102
double precision function dget06(rcond, rcondc)
DGET06
Definition dget06.f:55
subroutine dlatb4(path, imat, m, n, type, kl, ku, anorm, mode, cndnum, dist)
DLATB4
Definition dlatb4.f:120
subroutine dlatms(m, n, dist, iseed, sym, d, mode, cond, dmax, kl, ku, pack, a, lda, work, info)
DLATMS
Definition dlatms.f:321
subroutine dppt01(uplo, n, a, afac, rwork, resid)
DPPT01
Definition dppt01.f:93
subroutine dppt02(uplo, n, nrhs, a, x, ldx, b, ldb, rwork, resid)
DPPT02
Definition dppt02.f:122
subroutine dppt03(uplo, n, a, ainv, work, ldwork, rwork, rcond, resid)
DPPT03
Definition dppt03.f:110
subroutine dppt05(uplo, n, nrhs, ap, b, ldb, x, ldx, xact, ldxact, ferr, berr, reslts)
DPPT05
Definition dppt05.f:156
subroutine dcopy(n, dx, incx, dy, incy)
DCOPY
Definition dcopy.f:82
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
double precision function dlansp(norm, uplo, n, ap, work)
DLANSP returns the value of the 1-norm, or the Frobenius norm, or the infinity norm,...
Definition dlansp.f:114
subroutine dppcon(uplo, n, ap, anorm, rcond, work, iwork, info)
DPPCON
Definition dppcon.f:118
subroutine dpprfs(uplo, n, nrhs, ap, afp, b, ldb, x, ldx, ferr, berr, work, iwork, info)
DPPRFS
Definition dpprfs.f:171
subroutine dpptrf(uplo, n, ap, info)
DPPTRF
Definition dpptrf.f:119
subroutine dpptri(uplo, n, ap, info)
DPPTRI
Definition dpptri.f:93
subroutine dpptrs(uplo, n, nrhs, ap, b, ldb, info)
DPPTRS
Definition dpptrs.f:108
Here is the call graph for this function:
Here is the caller graph for this function: