LAPACK  3.4.2
LAPACK: Linear Algebra PACKage
 All Files Functions Groups
dchkpt.f
Go to the documentation of this file.
1 *> \brief \b DCHKPT
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 DCHKPT( DOTYPE, NN, NVAL, NNS, NSVAL, THRESH, TSTERR,
12 * A, D, E, B, X, XACT, WORK, RWORK, NOUT )
13 *
14 * .. Scalar Arguments ..
15 * LOGICAL TSTERR
16 * INTEGER NN, NNS, NOUT
17 * DOUBLE PRECISION THRESH
18 * ..
19 * .. Array Arguments ..
20 * LOGICAL DOTYPE( * )
21 * INTEGER NSVAL( * ), NVAL( * )
22 * DOUBLE PRECISION A( * ), B( * ), D( * ), E( * ), RWORK( * ),
23 * $ WORK( * ), X( * ), XACT( * )
24 * ..
25 *
26 *
27 *> \par Purpose:
28 * =============
29 *>
30 *> \verbatim
31 *>
32 *> DCHKPT tests DPTTRF, -TRS, -RFS, and -CON
33 *> \endverbatim
34 *
35 * Arguments:
36 * ==========
37 *
38 *> \param[in] DOTYPE
39 *> \verbatim
40 *> DOTYPE is LOGICAL array, dimension (NTYPES)
41 *> The matrix types to be used for testing. Matrices of type j
42 *> (for 1 <= j <= NTYPES) are used for testing if DOTYPE(j) =
43 *> .TRUE.; if DOTYPE(j) = .FALSE., then type j is not used.
44 *> \endverbatim
45 *>
46 *> \param[in] NN
47 *> \verbatim
48 *> NN is INTEGER
49 *> The number of values of N contained in the vector NVAL.
50 *> \endverbatim
51 *>
52 *> \param[in] NVAL
53 *> \verbatim
54 *> NVAL is INTEGER array, dimension (NN)
55 *> The values of the matrix dimension N.
56 *> \endverbatim
57 *>
58 *> \param[in] NNS
59 *> \verbatim
60 *> NNS is INTEGER
61 *> The number of values of NRHS contained in the vector NSVAL.
62 *> \endverbatim
63 *>
64 *> \param[in] NSVAL
65 *> \verbatim
66 *> NSVAL is INTEGER array, dimension (NNS)
67 *> The values of the number of right hand sides NRHS.
68 *> \endverbatim
69 *>
70 *> \param[in] THRESH
71 *> \verbatim
72 *> THRESH is DOUBLE PRECISION
73 *> The threshold value for the test ratios. A result is
74 *> included in the output file if RESULT >= THRESH. To have
75 *> every test ratio printed, use THRESH = 0.
76 *> \endverbatim
77 *>
78 *> \param[in] TSTERR
79 *> \verbatim
80 *> TSTERR is LOGICAL
81 *> Flag that indicates whether error exits are to be tested.
82 *> \endverbatim
83 *>
84 *> \param[out] A
85 *> \verbatim
86 *> A is DOUBLE PRECISION array, dimension (NMAX*2)
87 *> \endverbatim
88 *>
89 *> \param[out] D
90 *> \verbatim
91 *> D is DOUBLE PRECISION array, dimension (NMAX*2)
92 *> \endverbatim
93 *>
94 *> \param[out] E
95 *> \verbatim
96 *> E is DOUBLE PRECISION array, dimension (NMAX*2)
97 *> \endverbatim
98 *>
99 *> \param[out] B
100 *> \verbatim
101 *> B is DOUBLE PRECISION array, dimension (NMAX*NSMAX)
102 *> where NSMAX is the largest entry in NSVAL.
103 *> \endverbatim
104 *>
105 *> \param[out] X
106 *> \verbatim
107 *> X is DOUBLE PRECISION array, dimension (NMAX*NSMAX)
108 *> \endverbatim
109 *>
110 *> \param[out] XACT
111 *> \verbatim
112 *> XACT is DOUBLE PRECISION array, dimension (NMAX*NSMAX)
113 *> \endverbatim
114 *>
115 *> \param[out] WORK
116 *> \verbatim
117 *> WORK is DOUBLE PRECISION array, dimension
118 *> (NMAX*max(3,NSMAX))
119 *> \endverbatim
120 *>
121 *> \param[out] RWORK
122 *> \verbatim
123 *> RWORK is DOUBLE PRECISION array, dimension
124 *> (max(NMAX,2*NSMAX))
125 *> \endverbatim
126 *>
127 *> \param[in] NOUT
128 *> \verbatim
129 *> NOUT is INTEGER
130 *> The unit number for output.
131 *> \endverbatim
132 *
133 * Authors:
134 * ========
135 *
136 *> \author Univ. of Tennessee
137 *> \author Univ. of California Berkeley
138 *> \author Univ. of Colorado Denver
139 *> \author NAG Ltd.
140 *
141 *> \date November 2011
142 *
143 *> \ingroup double_lin
144 *
145 * =====================================================================
146  SUBROUTINE dchkpt( DOTYPE, NN, NVAL, NNS, NSVAL, THRESH, TSTERR,
147  $ a, d, e, b, x, xact, work, rwork, nout )
148 *
149 * -- LAPACK test routine (version 3.4.0) --
150 * -- LAPACK is a software package provided by Univ. of Tennessee, --
151 * -- Univ. of California Berkeley, Univ. of Colorado Denver and NAG Ltd..--
152 * November 2011
153 *
154 * .. Scalar Arguments ..
155  LOGICAL tsterr
156  INTEGER nn, nns, nout
157  DOUBLE PRECISION thresh
158 * ..
159 * .. Array Arguments ..
160  LOGICAL dotype( * )
161  INTEGER nsval( * ), nval( * )
162  DOUBLE PRECISION a( * ), b( * ), d( * ), e( * ), rwork( * ),
163  $ work( * ), x( * ), xact( * )
164 * ..
165 *
166 * =====================================================================
167 *
168 * .. Parameters ..
169  DOUBLE PRECISION one, zero
170  parameter( one = 1.0d+0, zero = 0.0d+0 )
171  INTEGER ntypes
172  parameter( ntypes = 12 )
173  INTEGER ntests
174  parameter( ntests = 7 )
175 * ..
176 * .. Local Scalars ..
177  LOGICAL zerot
178  CHARACTER dist, type
179  CHARACTER*3 path
180  INTEGER i, ia, imat, in, info, irhs, ix, izero, j, k,
181  $ kl, ku, lda, mode, n, nerrs, nfail, nimat,
182  $ nrhs, nrun
183  DOUBLE PRECISION ainvnm, anorm, cond, dmax, rcond, rcondc
184 * ..
185 * .. Local Arrays ..
186  INTEGER iseed( 4 ), iseedy( 4 )
187  DOUBLE PRECISION result( ntests ), z( 3 )
188 * ..
189 * .. External Functions ..
190  INTEGER idamax
191  DOUBLE PRECISION dasum, dget06, dlanst
192  EXTERNAL idamax, dasum, dget06, dlanst
193 * ..
194 * .. External Subroutines ..
195  EXTERNAL alaerh, alahd, alasum, dcopy, derrgt, dget04,
196  $ dlacpy, dlaptm, dlarnv, dlatb4, dlatms, dptcon,
197  $ dptrfs, dptt01, dptt02, dptt05, dpttrf, dpttrs,
198  $ dscal
199 * ..
200 * .. Intrinsic Functions ..
201  INTRINSIC abs, max
202 * ..
203 * .. Scalars in Common ..
204  LOGICAL lerr, ok
205  CHARACTER*32 srnamt
206  INTEGER infot, nunit
207 * ..
208 * .. Common blocks ..
209  common / infoc / infot, nunit, ok, lerr
210  common / srnamc / srnamt
211 * ..
212 * .. Data statements ..
213  DATA iseedy / 0, 0, 0, 1 /
214 * ..
215 * .. Executable Statements ..
216 *
217  path( 1: 1 ) = 'Double precision'
218  path( 2: 3 ) = 'PT'
219  nrun = 0
220  nfail = 0
221  nerrs = 0
222  DO 10 i = 1, 4
223  iseed( i ) = iseedy( i )
224  10 continue
225 *
226 * Test the error exits
227 *
228  IF( tsterr )
229  $ CALL derrgt( path, nout )
230  infot = 0
231 *
232  DO 110 in = 1, nn
233 *
234 * Do for each value of N in NVAL.
235 *
236  n = nval( in )
237  lda = max( 1, n )
238  nimat = ntypes
239  IF( n.LE.0 )
240  $ nimat = 1
241 *
242  DO 100 imat = 1, nimat
243 *
244 * Do the tests only if DOTYPE( IMAT ) is true.
245 *
246  IF( n.GT.0 .AND. .NOT.dotype( imat ) )
247  $ go to 100
248 *
249 * Set up parameters with DLATB4.
250 *
251  CALL dlatb4( path, imat, n, n, type, kl, ku, anorm, mode,
252  $ cond, dist )
253 *
254  zerot = imat.GE.8 .AND. imat.LE.10
255  IF( imat.LE.6 ) THEN
256 *
257 * Type 1-6: generate a symmetric tridiagonal matrix of
258 * known condition number in lower triangular band storage.
259 *
260  srnamt = 'DLATMS'
261  CALL dlatms( n, n, dist, iseed, type, rwork, mode, cond,
262  $ anorm, kl, ku, 'B', a, 2, work, info )
263 *
264 * Check the error code from DLATMS.
265 *
266  IF( info.NE.0 ) THEN
267  CALL alaerh( path, 'DLATMS', info, 0, ' ', n, n, kl,
268  $ ku, -1, imat, nfail, nerrs, nout )
269  go to 100
270  END IF
271  izero = 0
272 *
273 * Copy the matrix to D and E.
274 *
275  ia = 1
276  DO 20 i = 1, n - 1
277  d( i ) = a( ia )
278  e( i ) = a( ia+1 )
279  ia = ia + 2
280  20 continue
281  IF( n.GT.0 )
282  $ d( n ) = a( ia )
283  ELSE
284 *
285 * Type 7-12: generate a diagonally dominant matrix with
286 * unknown condition number in the vectors D and E.
287 *
288  IF( .NOT.zerot .OR. .NOT.dotype( 7 ) ) THEN
289 *
290 * Let D and E have values from [-1,1].
291 *
292  CALL dlarnv( 2, iseed, n, d )
293  CALL dlarnv( 2, iseed, n-1, e )
294 *
295 * Make the tridiagonal matrix diagonally dominant.
296 *
297  IF( n.EQ.1 ) THEN
298  d( 1 ) = abs( d( 1 ) )
299  ELSE
300  d( 1 ) = abs( d( 1 ) ) + abs( e( 1 ) )
301  d( n ) = abs( d( n ) ) + abs( e( n-1 ) )
302  DO 30 i = 2, n - 1
303  d( i ) = abs( d( i ) ) + abs( e( i ) ) +
304  $ abs( e( i-1 ) )
305  30 continue
306  END IF
307 *
308 * Scale D and E so the maximum element is ANORM.
309 *
310  ix = idamax( n, d, 1 )
311  dmax = d( ix )
312  CALL dscal( n, anorm / dmax, d, 1 )
313  CALL dscal( n-1, anorm / dmax, e, 1 )
314 *
315  ELSE IF( izero.GT.0 ) THEN
316 *
317 * Reuse the last matrix by copying back the zeroed out
318 * elements.
319 *
320  IF( izero.EQ.1 ) THEN
321  d( 1 ) = z( 2 )
322  IF( n.GT.1 )
323  $ e( 1 ) = z( 3 )
324  ELSE IF( izero.EQ.n ) THEN
325  e( n-1 ) = z( 1 )
326  d( n ) = z( 2 )
327  ELSE
328  e( izero-1 ) = z( 1 )
329  d( izero ) = z( 2 )
330  e( izero ) = z( 3 )
331  END IF
332  END IF
333 *
334 * For types 8-10, set one row and column of the matrix to
335 * zero.
336 *
337  izero = 0
338  IF( imat.EQ.8 ) THEN
339  izero = 1
340  z( 2 ) = d( 1 )
341  d( 1 ) = zero
342  IF( n.GT.1 ) THEN
343  z( 3 ) = e( 1 )
344  e( 1 ) = zero
345  END IF
346  ELSE IF( imat.EQ.9 ) THEN
347  izero = n
348  IF( n.GT.1 ) THEN
349  z( 1 ) = e( n-1 )
350  e( n-1 ) = zero
351  END IF
352  z( 2 ) = d( n )
353  d( n ) = zero
354  ELSE IF( imat.EQ.10 ) THEN
355  izero = ( n+1 ) / 2
356  IF( izero.GT.1 ) THEN
357  z( 1 ) = e( izero-1 )
358  e( izero-1 ) = zero
359  z( 3 ) = e( izero )
360  e( izero ) = zero
361  END IF
362  z( 2 ) = d( izero )
363  d( izero ) = zero
364  END IF
365  END IF
366 *
367  CALL dcopy( n, d, 1, d( n+1 ), 1 )
368  IF( n.GT.1 )
369  $ CALL dcopy( n-1, e, 1, e( n+1 ), 1 )
370 *
371 *+ TEST 1
372 * Factor A as L*D*L' and compute the ratio
373 * norm(L*D*L' - A) / (n * norm(A) * EPS )
374 *
375  CALL dpttrf( n, d( n+1 ), e( n+1 ), info )
376 *
377 * Check error code from DPTTRF.
378 *
379  IF( info.NE.izero ) THEN
380  CALL alaerh( path, 'DPTTRF', info, izero, ' ', n, n, -1,
381  $ -1, -1, imat, nfail, nerrs, nout )
382  go to 100
383  END IF
384 *
385  IF( info.GT.0 ) THEN
386  rcondc = zero
387  go to 90
388  END IF
389 *
390  CALL dptt01( n, d, e, d( n+1 ), e( n+1 ), work,
391  $ result( 1 ) )
392 *
393 * Print the test ratio if greater than or equal to THRESH.
394 *
395  IF( result( 1 ).GE.thresh ) THEN
396  IF( nfail.EQ.0 .AND. nerrs.EQ.0 )
397  $ CALL alahd( nout, path )
398  WRITE( nout, fmt = 9999 )n, imat, 1, result( 1 )
399  nfail = nfail + 1
400  END IF
401  nrun = nrun + 1
402 *
403 * Compute RCONDC = 1 / (norm(A) * norm(inv(A))
404 *
405 * Compute norm(A).
406 *
407  anorm = dlanst( '1', n, d, e )
408 *
409 * Use DPTTRS to solve for one column at a time of inv(A),
410 * computing the maximum column sum as we go.
411 *
412  ainvnm = zero
413  DO 50 i = 1, n
414  DO 40 j = 1, n
415  x( j ) = zero
416  40 continue
417  x( i ) = one
418  CALL dpttrs( n, 1, d( n+1 ), e( n+1 ), x, lda, info )
419  ainvnm = max( ainvnm, dasum( n, x, 1 ) )
420  50 continue
421  rcondc = one / max( one, anorm*ainvnm )
422 *
423  DO 80 irhs = 1, nns
424  nrhs = nsval( irhs )
425 *
426 * Generate NRHS random solution vectors.
427 *
428  ix = 1
429  DO 60 j = 1, nrhs
430  CALL dlarnv( 2, iseed, n, xact( ix ) )
431  ix = ix + lda
432  60 continue
433 *
434 * Set the right hand side.
435 *
436  CALL dlaptm( n, nrhs, one, d, e, xact, lda, zero, b,
437  $ lda )
438 *
439 *+ TEST 2
440 * Solve A*x = b and compute the residual.
441 *
442  CALL dlacpy( 'Full', n, nrhs, b, lda, x, lda )
443  CALL dpttrs( n, nrhs, d( n+1 ), e( n+1 ), x, lda, info )
444 *
445 * Check error code from DPTTRS.
446 *
447  IF( info.NE.0 )
448  $ CALL alaerh( path, 'DPTTRS', info, 0, ' ', n, n, -1,
449  $ -1, nrhs, imat, nfail, nerrs, nout )
450 *
451  CALL dlacpy( 'Full', n, nrhs, b, lda, work, lda )
452  CALL dptt02( n, nrhs, d, e, x, lda, work, lda,
453  $ result( 2 ) )
454 *
455 *+ TEST 3
456 * Check solution from generated exact solution.
457 *
458  CALL dget04( n, nrhs, x, lda, xact, lda, rcondc,
459  $ result( 3 ) )
460 *
461 *+ TESTS 4, 5, and 6
462 * Use iterative refinement to improve the solution.
463 *
464  srnamt = 'DPTRFS'
465  CALL dptrfs( n, nrhs, d, e, d( n+1 ), e( n+1 ), b, lda,
466  $ x, lda, rwork, rwork( nrhs+1 ), work, info )
467 *
468 * Check error code from DPTRFS.
469 *
470  IF( info.NE.0 )
471  $ CALL alaerh( path, 'DPTRFS', info, 0, ' ', n, n, -1,
472  $ -1, nrhs, imat, nfail, nerrs, nout )
473 *
474  CALL dget04( n, nrhs, x, lda, xact, lda, rcondc,
475  $ result( 4 ) )
476  CALL dptt05( n, nrhs, d, e, b, lda, x, lda, xact, lda,
477  $ rwork, rwork( nrhs+1 ), result( 5 ) )
478 *
479 * Print information about the tests that did not pass the
480 * threshold.
481 *
482  DO 70 k = 2, 6
483  IF( result( k ).GE.thresh ) THEN
484  IF( nfail.EQ.0 .AND. nerrs.EQ.0 )
485  $ CALL alahd( nout, path )
486  WRITE( nout, fmt = 9998 )n, nrhs, imat, k,
487  $ result( k )
488  nfail = nfail + 1
489  END IF
490  70 continue
491  nrun = nrun + 5
492  80 continue
493 *
494 *+ TEST 7
495 * Estimate the reciprocal of the condition number of the
496 * matrix.
497 *
498  90 continue
499  srnamt = 'DPTCON'
500  CALL dptcon( n, d( n+1 ), e( n+1 ), anorm, rcond, rwork,
501  $ info )
502 *
503 * Check error code from DPTCON.
504 *
505  IF( info.NE.0 )
506  $ CALL alaerh( path, 'DPTCON', info, 0, ' ', n, n, -1, -1,
507  $ -1, imat, nfail, nerrs, nout )
508 *
509  result( 7 ) = dget06( rcond, rcondc )
510 *
511 * Print the test ratio if greater than or equal to THRESH.
512 *
513  IF( result( 7 ).GE.thresh ) THEN
514  IF( nfail.EQ.0 .AND. nerrs.EQ.0 )
515  $ CALL alahd( nout, path )
516  WRITE( nout, fmt = 9999 )n, imat, 7, result( 7 )
517  nfail = nfail + 1
518  END IF
519  nrun = nrun + 1
520  100 continue
521  110 continue
522 *
523 * Print a summary of the results.
524 *
525  CALL alasum( path, nout, nfail, nrun, nerrs )
526 *
527  9999 format( ' N =', i5, ', type ', i2, ', test ', i2, ', ratio = ',
528  $ g12.5 )
529  9998 format( ' N =', i5, ', NRHS=', i3, ', type ', i2, ', test(', i2,
530  $ ') = ', g12.5 )
531  return
532 *
533 * End of DCHKPT
534 *
535  END