 |
LAPACK 3.3.1
Linear Algebra PACKage
|
|
LAPACK 3.3.1
Linear Algebra PACKage
|
00001 SUBROUTINE DDRVSP( DOTYPE, NN, NVAL, NRHS, THRESH, TSTERR, NMAX, 00002 $ A, AFAC, AINV, B, X, XACT, WORK, RWORK, IWORK, 00003 $ NOUT ) 00004 * 00005 * -- LAPACK test routine (version 3.1) -- 00006 * Univ. of Tennessee, Univ. of California Berkeley and NAG Ltd.. 00007 * November 2006 00008 * 00009 * .. Scalar Arguments .. 00010 LOGICAL TSTERR 00011 INTEGER NMAX, NN, NOUT, NRHS 00012 DOUBLE PRECISION THRESH 00013 * .. 00014 * .. Array Arguments .. 00015 LOGICAL DOTYPE( * ) 00016 INTEGER IWORK( * ), NVAL( * ) 00017 DOUBLE PRECISION A( * ), AFAC( * ), AINV( * ), B( * ), 00018 $ RWORK( * ), WORK( * ), X( * ), XACT( * ) 00019 * .. 00020 * 00021 * Purpose 00022 * ======= 00023 * 00024 * DDRVSP tests the driver routines DSPSV and -SVX. 00025 * 00026 * Arguments 00027 * ========= 00028 * 00029 * DOTYPE (input) LOGICAL array, dimension (NTYPES) 00030 * The matrix types to be used for testing. Matrices of type j 00031 * (for 1 <= j <= NTYPES) are used for testing if DOTYPE(j) = 00032 * .TRUE.; if DOTYPE(j) = .FALSE., then type j is not used. 00033 * 00034 * NN (input) INTEGER 00035 * The number of values of N contained in the vector NVAL. 00036 * 00037 * NVAL (input) INTEGER array, dimension (NN) 00038 * The values of the matrix dimension N. 00039 * 00040 * NRHS (input) INTEGER 00041 * The number of right hand side vectors to be generated for 00042 * each linear system. 00043 * 00044 * THRESH (input) DOUBLE PRECISION 00045 * The threshold value for the test ratios. A result is 00046 * included in the output file if RESULT >= THRESH. To have 00047 * every test ratio printed, use THRESH = 0. 00048 * 00049 * TSTERR (input) LOGICAL 00050 * Flag that indicates whether error exits are to be tested. 00051 * 00052 * NMAX (input) INTEGER 00053 * The maximum value permitted for N, used in dimensioning the 00054 * work arrays. 00055 * 00056 * A (workspace) DOUBLE PRECISION array, dimension 00057 * (NMAX*(NMAX+1)/2) 00058 * 00059 * AFAC (workspace) DOUBLE PRECISION array, dimension 00060 * (NMAX*(NMAX+1)/2) 00061 * 00062 * AINV (workspace) DOUBLE PRECISION array, dimension 00063 * (NMAX*(NMAX+1)/2) 00064 * 00065 * B (workspace) DOUBLE PRECISION array, dimension (NMAX*NRHS) 00066 * 00067 * X (workspace) DOUBLE PRECISION array, dimension (NMAX*NRHS) 00068 * 00069 * XACT (workspace) DOUBLE PRECISION array, dimension (NMAX*NRHS) 00070 * 00071 * WORK (workspace) DOUBLE PRECISION array, dimension 00072 * (NMAX*max(2,NRHS)) 00073 * 00074 * RWORK (workspace) DOUBLE PRECISION array, dimension (NMAX+2*NRHS) 00075 * 00076 * IWORK (workspace) INTEGER array, dimension (2*NMAX) 00077 * 00078 * NOUT (input) INTEGER 00079 * The unit number for output. 00080 * 00081 * ===================================================================== 00082 * 00083 * .. Parameters .. 00084 DOUBLE PRECISION ONE, ZERO 00085 PARAMETER ( ONE = 1.0D+0, ZERO = 0.0D+0 ) 00086 INTEGER NTYPES, NTESTS 00087 PARAMETER ( NTYPES = 10, NTESTS = 6 ) 00088 INTEGER NFACT 00089 PARAMETER ( NFACT = 2 ) 00090 * .. 00091 * .. Local Scalars .. 00092 LOGICAL ZEROT 00093 CHARACTER DIST, FACT, PACKIT, TYPE, UPLO, XTYPE 00094 CHARACTER*3 PATH 00095 INTEGER I, I1, I2, IFACT, IMAT, IN, INFO, IOFF, IUPLO, 00096 $ IZERO, J, K, K1, KL, KU, LDA, LWORK, MODE, N, 00097 $ NERRS, NFAIL, NIMAT, NPP, NRUN, NT 00098 DOUBLE PRECISION AINVNM, ANORM, CNDNUM, RCOND, RCONDC 00099 * .. 00100 * .. Local Arrays .. 00101 CHARACTER FACTS( NFACT ) 00102 INTEGER ISEED( 4 ), ISEEDY( 4 ) 00103 DOUBLE PRECISION RESULT( NTESTS ) 00104 * .. 00105 * .. External Functions .. 00106 DOUBLE PRECISION DGET06, DLANSP 00107 EXTERNAL DGET06, DLANSP 00108 * .. 00109 * .. External Subroutines .. 00110 EXTERNAL ALADHD, ALAERH, ALASVM, DCOPY, DERRVX, DGET04, 00111 $ DLACPY, DLARHS, DLASET, DLATB4, DLATMS, DPPT02, 00112 $ DPPT05, DSPSV, DSPSVX, DSPT01, DSPTRF, DSPTRI 00113 * .. 00114 * .. Scalars in Common .. 00115 LOGICAL LERR, OK 00116 CHARACTER*32 SRNAMT 00117 INTEGER INFOT, NUNIT 00118 * .. 00119 * .. Common blocks .. 00120 COMMON / INFOC / INFOT, NUNIT, OK, LERR 00121 COMMON / SRNAMC / SRNAMT 00122 * .. 00123 * .. Intrinsic Functions .. 00124 INTRINSIC MAX, MIN 00125 * .. 00126 * .. Data statements .. 00127 DATA ISEEDY / 1988, 1989, 1990, 1991 / 00128 DATA FACTS / 'F', 'N' / 00129 * .. 00130 * .. Executable Statements .. 00131 * 00132 * Initialize constants and the random number seed. 00133 * 00134 PATH( 1: 1 ) = 'Double precision' 00135 PATH( 2: 3 ) = 'SP' 00136 NRUN = 0 00137 NFAIL = 0 00138 NERRS = 0 00139 DO 10 I = 1, 4 00140 ISEED( I ) = ISEEDY( I ) 00141 10 CONTINUE 00142 LWORK = MAX( 2*NMAX, NMAX*NRHS ) 00143 * 00144 * Test the error exits 00145 * 00146 IF( TSTERR ) 00147 $ CALL DERRVX( PATH, NOUT ) 00148 INFOT = 0 00149 * 00150 * Do for each value of N in NVAL 00151 * 00152 DO 180 IN = 1, NN 00153 N = NVAL( IN ) 00154 LDA = MAX( N, 1 ) 00155 NPP = N*( N+1 ) / 2 00156 XTYPE = 'N' 00157 NIMAT = NTYPES 00158 IF( N.LE.0 ) 00159 $ NIMAT = 1 00160 * 00161 DO 170 IMAT = 1, NIMAT 00162 * 00163 * Do the tests only if DOTYPE( IMAT ) is true. 00164 * 00165 IF( .NOT.DOTYPE( IMAT ) ) 00166 $ GO TO 170 00167 * 00168 * Skip types 3, 4, 5, or 6 if the matrix size is too small. 00169 * 00170 ZEROT = IMAT.GE.3 .AND. IMAT.LE.6 00171 IF( ZEROT .AND. N.LT.IMAT-2 ) 00172 $ GO TO 170 00173 * 00174 * Do first for UPLO = 'U', then for UPLO = 'L' 00175 * 00176 DO 160 IUPLO = 1, 2 00177 IF( IUPLO.EQ.1 ) THEN 00178 UPLO = 'U' 00179 PACKIT = 'C' 00180 ELSE 00181 UPLO = 'L' 00182 PACKIT = 'R' 00183 END IF 00184 * 00185 * Set up parameters with DLATB4 and generate a test matrix 00186 * with DLATMS. 00187 * 00188 CALL DLATB4( PATH, IMAT, N, N, TYPE, KL, KU, ANORM, MODE, 00189 $ CNDNUM, DIST ) 00190 * 00191 SRNAMT = 'DLATMS' 00192 CALL DLATMS( N, N, DIST, ISEED, TYPE, RWORK, MODE, 00193 $ CNDNUM, ANORM, KL, KU, PACKIT, A, LDA, WORK, 00194 $ INFO ) 00195 * 00196 * Check error code from DLATMS. 00197 * 00198 IF( INFO.NE.0 ) THEN 00199 CALL ALAERH( PATH, 'DLATMS', INFO, 0, UPLO, N, N, -1, 00200 $ -1, -1, IMAT, NFAIL, NERRS, NOUT ) 00201 GO TO 160 00202 END IF 00203 * 00204 * For types 3-6, zero one or more rows and columns of the 00205 * matrix to test that INFO is returned correctly. 00206 * 00207 IF( ZEROT ) THEN 00208 IF( IMAT.EQ.3 ) THEN 00209 IZERO = 1 00210 ELSE IF( IMAT.EQ.4 ) THEN 00211 IZERO = N 00212 ELSE 00213 IZERO = N / 2 + 1 00214 END IF 00215 * 00216 IF( IMAT.LT.6 ) THEN 00217 * 00218 * Set row and column IZERO to zero. 00219 * 00220 IF( IUPLO.EQ.1 ) THEN 00221 IOFF = ( IZERO-1 )*IZERO / 2 00222 DO 20 I = 1, IZERO - 1 00223 A( IOFF+I ) = ZERO 00224 20 CONTINUE 00225 IOFF = IOFF + IZERO 00226 DO 30 I = IZERO, N 00227 A( IOFF ) = ZERO 00228 IOFF = IOFF + I 00229 30 CONTINUE 00230 ELSE 00231 IOFF = IZERO 00232 DO 40 I = 1, IZERO - 1 00233 A( IOFF ) = ZERO 00234 IOFF = IOFF + N - I 00235 40 CONTINUE 00236 IOFF = IOFF - IZERO 00237 DO 50 I = IZERO, N 00238 A( IOFF+I ) = ZERO 00239 50 CONTINUE 00240 END IF 00241 ELSE 00242 IOFF = 0 00243 IF( IUPLO.EQ.1 ) THEN 00244 * 00245 * Set the first IZERO rows and columns to zero. 00246 * 00247 DO 70 J = 1, N 00248 I2 = MIN( J, IZERO ) 00249 DO 60 I = 1, I2 00250 A( IOFF+I ) = ZERO 00251 60 CONTINUE 00252 IOFF = IOFF + J 00253 70 CONTINUE 00254 ELSE 00255 * 00256 * Set the last IZERO rows and columns to zero. 00257 * 00258 DO 90 J = 1, N 00259 I1 = MAX( J, IZERO ) 00260 DO 80 I = I1, N 00261 A( IOFF+I ) = ZERO 00262 80 CONTINUE 00263 IOFF = IOFF + N - J 00264 90 CONTINUE 00265 END IF 00266 END IF 00267 ELSE 00268 IZERO = 0 00269 END IF 00270 * 00271 DO 150 IFACT = 1, NFACT 00272 * 00273 * Do first for FACT = 'F', then for other values. 00274 * 00275 FACT = FACTS( IFACT ) 00276 * 00277 * Compute the condition number for comparison with 00278 * the value returned by DSPSVX. 00279 * 00280 IF( ZEROT ) THEN 00281 IF( IFACT.EQ.1 ) 00282 $ GO TO 150 00283 RCONDC = ZERO 00284 * 00285 ELSE IF( IFACT.EQ.1 ) THEN 00286 * 00287 * Compute the 1-norm of A. 00288 * 00289 ANORM = DLANSP( '1', UPLO, N, A, RWORK ) 00290 * 00291 * Factor the matrix A. 00292 * 00293 CALL DCOPY( NPP, A, 1, AFAC, 1 ) 00294 CALL DSPTRF( UPLO, N, AFAC, IWORK, INFO ) 00295 * 00296 * Compute inv(A) and take its norm. 00297 * 00298 CALL DCOPY( NPP, AFAC, 1, AINV, 1 ) 00299 CALL DSPTRI( UPLO, N, AINV, IWORK, WORK, INFO ) 00300 AINVNM = DLANSP( '1', UPLO, N, AINV, RWORK ) 00301 * 00302 * Compute the 1-norm condition number of A. 00303 * 00304 IF( ANORM.LE.ZERO .OR. AINVNM.LE.ZERO ) THEN 00305 RCONDC = ONE 00306 ELSE 00307 RCONDC = ( ONE / ANORM ) / AINVNM 00308 END IF 00309 END IF 00310 * 00311 * Form an exact solution and set the right hand side. 00312 * 00313 SRNAMT = 'DLARHS' 00314 CALL DLARHS( PATH, XTYPE, UPLO, ' ', N, N, KL, KU, 00315 $ NRHS, A, LDA, XACT, LDA, B, LDA, ISEED, 00316 $ INFO ) 00317 XTYPE = 'C' 00318 * 00319 * --- Test DSPSV --- 00320 * 00321 IF( IFACT.EQ.2 ) THEN 00322 CALL DCOPY( NPP, A, 1, AFAC, 1 ) 00323 CALL DLACPY( 'Full', N, NRHS, B, LDA, X, LDA ) 00324 * 00325 * Factor the matrix and solve the system using DSPSV. 00326 * 00327 SRNAMT = 'DSPSV ' 00328 CALL DSPSV( UPLO, N, NRHS, AFAC, IWORK, X, LDA, 00329 $ INFO ) 00330 * 00331 * Adjust the expected value of INFO to account for 00332 * pivoting. 00333 * 00334 K = IZERO 00335 IF( K.GT.0 ) THEN 00336 100 CONTINUE 00337 IF( IWORK( K ).LT.0 ) THEN 00338 IF( IWORK( K ).NE.-K ) THEN 00339 K = -IWORK( K ) 00340 GO TO 100 00341 END IF 00342 ELSE IF( IWORK( K ).NE.K ) THEN 00343 K = IWORK( K ) 00344 GO TO 100 00345 END IF 00346 END IF 00347 * 00348 * Check error code from DSPSV . 00349 * 00350 IF( INFO.NE.K ) THEN 00351 CALL ALAERH( PATH, 'DSPSV ', INFO, K, UPLO, N, 00352 $ N, -1, -1, NRHS, IMAT, NFAIL, 00353 $ NERRS, NOUT ) 00354 GO TO 120 00355 ELSE IF( INFO.NE.0 ) THEN 00356 GO TO 120 00357 END IF 00358 * 00359 * Reconstruct matrix from factors and compute 00360 * residual. 00361 * 00362 CALL DSPT01( UPLO, N, A, AFAC, IWORK, AINV, LDA, 00363 $ RWORK, RESULT( 1 ) ) 00364 * 00365 * Compute residual of the computed solution. 00366 * 00367 CALL DLACPY( 'Full', N, NRHS, B, LDA, WORK, LDA ) 00368 CALL DPPT02( UPLO, N, NRHS, A, X, LDA, WORK, LDA, 00369 $ RWORK, RESULT( 2 ) ) 00370 * 00371 * Check solution from generated exact solution. 00372 * 00373 CALL DGET04( N, NRHS, X, LDA, XACT, LDA, RCONDC, 00374 $ RESULT( 3 ) ) 00375 NT = 3 00376 * 00377 * Print information about the tests that did not pass 00378 * the threshold. 00379 * 00380 DO 110 K = 1, NT 00381 IF( RESULT( K ).GE.THRESH ) THEN 00382 IF( NFAIL.EQ.0 .AND. NERRS.EQ.0 ) 00383 $ CALL ALADHD( NOUT, PATH ) 00384 WRITE( NOUT, FMT = 9999 )'DSPSV ', UPLO, N, 00385 $ IMAT, K, RESULT( K ) 00386 NFAIL = NFAIL + 1 00387 END IF 00388 110 CONTINUE 00389 NRUN = NRUN + NT 00390 120 CONTINUE 00391 END IF 00392 * 00393 * --- Test DSPSVX --- 00394 * 00395 IF( IFACT.EQ.2 .AND. NPP.GT.0 ) 00396 $ CALL DLASET( 'Full', NPP, 1, ZERO, ZERO, AFAC, 00397 $ NPP ) 00398 CALL DLASET( 'Full', N, NRHS, ZERO, ZERO, X, LDA ) 00399 * 00400 * Solve the system and compute the condition number and 00401 * error bounds using DSPSVX. 00402 * 00403 SRNAMT = 'DSPSVX' 00404 CALL DSPSVX( FACT, UPLO, N, NRHS, A, AFAC, IWORK, B, 00405 $ LDA, X, LDA, RCOND, RWORK, 00406 $ RWORK( NRHS+1 ), WORK, IWORK( N+1 ), 00407 $ INFO ) 00408 * 00409 * Adjust the expected value of INFO to account for 00410 * pivoting. 00411 * 00412 K = IZERO 00413 IF( K.GT.0 ) THEN 00414 130 CONTINUE 00415 IF( IWORK( K ).LT.0 ) THEN 00416 IF( IWORK( K ).NE.-K ) THEN 00417 K = -IWORK( K ) 00418 GO TO 130 00419 END IF 00420 ELSE IF( IWORK( K ).NE.K ) THEN 00421 K = IWORK( K ) 00422 GO TO 130 00423 END IF 00424 END IF 00425 * 00426 * Check the error code from DSPSVX. 00427 * 00428 IF( INFO.NE.K ) THEN 00429 CALL ALAERH( PATH, 'DSPSVX', INFO, K, FACT // UPLO, 00430 $ N, N, -1, -1, NRHS, IMAT, NFAIL, 00431 $ NERRS, NOUT ) 00432 GO TO 150 00433 END IF 00434 * 00435 IF( INFO.EQ.0 ) THEN 00436 IF( IFACT.GE.2 ) THEN 00437 * 00438 * Reconstruct matrix from factors and compute 00439 * residual. 00440 * 00441 CALL DSPT01( UPLO, N, A, AFAC, IWORK, AINV, LDA, 00442 $ RWORK( 2*NRHS+1 ), RESULT( 1 ) ) 00443 K1 = 1 00444 ELSE 00445 K1 = 2 00446 END IF 00447 * 00448 * Compute residual of the computed solution. 00449 * 00450 CALL DLACPY( 'Full', N, NRHS, B, LDA, WORK, LDA ) 00451 CALL DPPT02( UPLO, N, NRHS, A, X, LDA, WORK, LDA, 00452 $ RWORK( 2*NRHS+1 ), RESULT( 2 ) ) 00453 * 00454 * Check solution from generated exact solution. 00455 * 00456 CALL DGET04( N, NRHS, X, LDA, XACT, LDA, RCONDC, 00457 $ RESULT( 3 ) ) 00458 * 00459 * Check the error bounds from iterative refinement. 00460 * 00461 CALL DPPT05( UPLO, N, NRHS, A, B, LDA, X, LDA, 00462 $ XACT, LDA, RWORK, RWORK( NRHS+1 ), 00463 $ RESULT( 4 ) ) 00464 ELSE 00465 K1 = 6 00466 END IF 00467 * 00468 * Compare RCOND from DSPSVX with the computed value 00469 * in RCONDC. 00470 * 00471 RESULT( 6 ) = DGET06( RCOND, RCONDC ) 00472 * 00473 * Print information about the tests that did not pass 00474 * the threshold. 00475 * 00476 DO 140 K = K1, 6 00477 IF( RESULT( K ).GE.THRESH ) THEN 00478 IF( NFAIL.EQ.0 .AND. NERRS.EQ.0 ) 00479 $ CALL ALADHD( NOUT, PATH ) 00480 WRITE( NOUT, FMT = 9998 )'DSPSVX', FACT, UPLO, 00481 $ N, IMAT, K, RESULT( K ) 00482 NFAIL = NFAIL + 1 00483 END IF 00484 140 CONTINUE 00485 NRUN = NRUN + 7 - K1 00486 * 00487 150 CONTINUE 00488 * 00489 160 CONTINUE 00490 170 CONTINUE 00491 180 CONTINUE 00492 * 00493 * Print a summary of the results. 00494 * 00495 CALL ALASVM( PATH, NOUT, NFAIL, NRUN, NERRS ) 00496 * 00497 9999 FORMAT( 1X, A, ', UPLO=''', A1, ''', N =', I5, ', type ', I2, 00498 $ ', test ', I2, ', ratio =', G12.5 ) 00499 9998 FORMAT( 1X, A, ', FACT=''', A1, ''', UPLO=''', A1, ''', N =', I5, 00500 $ ', type ', I2, ', test ', I2, ', ratio =', G12.5 ) 00501 RETURN 00502 * 00503 * End of DDRVSP 00504 * 00505 END
1.7.3 
