*> \brief \b CDRVGB * * =========== DOCUMENTATION =========== * * Online html documentation available at * http://www.netlib.org/lapack/explore-html/ * * Definition: * =========== * * SUBROUTINE CDRVGB( DOTYPE, NN, NVAL, NRHS, THRESH, TSTERR, A, LA, * AFB, LAFB, ASAV, B, BSAV, X, XACT, S, WORK, * RWORK, IWORK, NOUT ) * * .. Scalar Arguments .. * LOGICAL TSTERR * INTEGER LA, LAFB, NN, NOUT, NRHS * REAL THRESH * .. * .. Array Arguments .. * LOGICAL DOTYPE( * ) * INTEGER IWORK( * ), NVAL( * ) * REAL RWORK( * ), S( * ) * COMPLEX A( * ), AFB( * ), ASAV( * ), B( * ), BSAV( * ), * $ WORK( * ), X( * ), XACT( * ) * .. * * *> \par Purpose: * ============= *> *> \verbatim *> *> CDRVGB tests the driver routines CGBSV and -SVX. *> \endverbatim * * Arguments: * ========== * *> \param[in] DOTYPE *> \verbatim *> 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. *> \endverbatim *> *> \param[in] NN *> \verbatim *> NN is INTEGER *> The number of values of N contained in the vector NVAL. *> \endverbatim *> *> \param[in] NVAL *> \verbatim *> NVAL is INTEGER array, dimension (NN) *> The values of the matrix column dimension N. *> \endverbatim *> *> \param[in] NRHS *> \verbatim *> NRHS is INTEGER *> The number of right hand side vectors to be generated for *> each linear system. *> \endverbatim *> *> \param[in] THRESH *> \verbatim *> 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. *> \endverbatim *> *> \param[in] TSTERR *> \verbatim *> TSTERR is LOGICAL *> Flag that indicates whether error exits are to be tested. *> \endverbatim *> *> \param[out] A *> \verbatim *> A is COMPLEX array, dimension (LA) *> \endverbatim *> *> \param[in] LA *> \verbatim *> LA is INTEGER *> The length of the array A. LA >= (2*NMAX-1)*NMAX *> where NMAX is the largest entry in NVAL. *> \endverbatim *> *> \param[out] AFB *> \verbatim *> AFB is COMPLEX array, dimension (LAFB) *> \endverbatim *> *> \param[in] LAFB *> \verbatim *> LAFB is INTEGER *> The length of the array AFB. LAFB >= (3*NMAX-2)*NMAX *> where NMAX is the largest entry in NVAL. *> \endverbatim *> *> \param[out] ASAV *> \verbatim *> ASAV is COMPLEX array, dimension (LA) *> \endverbatim *> *> \param[out] B *> \verbatim *> B is COMPLEX array, dimension (NMAX*NRHS) *> \endverbatim *> *> \param[out] BSAV *> \verbatim *> BSAV is COMPLEX array, dimension (NMAX*NRHS) *> \endverbatim *> *> \param[out] X *> \verbatim *> X is COMPLEX array, dimension (NMAX*NRHS) *> \endverbatim *> *> \param[out] XACT *> \verbatim *> XACT is COMPLEX array, dimension (NMAX*NRHS) *> \endverbatim *> *> \param[out] S *> \verbatim *> S is REAL array, dimension (2*NMAX) *> \endverbatim *> *> \param[out] WORK *> \verbatim *> WORK is COMPLEX array, dimension *> (NMAX*max(3,NRHS,NMAX)) *> \endverbatim *> *> \param[out] RWORK *> \verbatim *> RWORK is REAL array, dimension *> (max(NMAX,2*NRHS)) *> \endverbatim *> *> \param[out] IWORK *> \verbatim *> IWORK is INTEGER array, dimension (NMAX) *> \endverbatim *> *> \param[in] NOUT *> \verbatim *> NOUT is INTEGER *> The unit number for output. *> \endverbatim * * Authors: * ======== * *> \author Univ. of Tennessee *> \author Univ. of California Berkeley *> \author Univ. of Colorado Denver *> \author NAG Ltd. * *> \ingroup complex_lin * * ===================================================================== SUBROUTINE CDRVGB( DOTYPE, NN, NVAL, NRHS, THRESH, TSTERR, A, LA, $ AFB, LAFB, ASAV, B, BSAV, X, XACT, S, WORK, $ RWORK, IWORK, NOUT ) * * -- LAPACK test routine -- * -- LAPACK is a software package provided by Univ. of Tennessee, -- * -- Univ. of California Berkeley, Univ. of Colorado Denver and NAG Ltd..-- * * .. Scalar Arguments .. LOGICAL TSTERR INTEGER LA, LAFB, NN, NOUT, NRHS REAL THRESH * .. * .. Array Arguments .. LOGICAL DOTYPE( * ) INTEGER IWORK( * ), NVAL( * ) REAL RWORK( * ), S( * ) COMPLEX A( * ), AFB( * ), ASAV( * ), B( * ), BSAV( * ), $ WORK( * ), X( * ), XACT( * ) * .. * * ===================================================================== * * .. Parameters .. REAL ONE, ZERO PARAMETER ( ONE = 1.0E+0, ZERO = 0.0E+0 ) INTEGER NTYPES PARAMETER ( NTYPES = 8 ) INTEGER NTESTS PARAMETER ( NTESTS = 7 ) INTEGER NTRAN PARAMETER ( NTRAN = 3 ) * .. * .. Local Scalars .. LOGICAL EQUIL, NOFACT, PREFAC, TRFCON, ZEROT CHARACTER DIST, EQUED, FACT, TRANS, TYPE, XTYPE CHARACTER*3 PATH INTEGER I, I1, I2, IEQUED, IFACT, IKL, IKU, IMAT, IN, $ INFO, IOFF, ITRAN, IZERO, J, K, K1, KL, KU, $ LDA, LDAFB, LDB, MODE, N, NB, NBMIN, NERRS, $ NFACT, NFAIL, NIMAT, NKL, NKU, NRUN, NT REAL AINVNM, AMAX, ANORM, ANORMI, ANORMO, ANRMPV, $ CNDNUM, COLCND, RCOND, RCONDC, RCONDI, RCONDO, $ ROLDC, ROLDI, ROLDO, ROWCND, RPVGRW * .. * .. Local Arrays .. CHARACTER EQUEDS( 4 ), FACTS( 3 ), TRANSS( NTRAN ) INTEGER ISEED( 4 ), ISEEDY( 4 ) REAL RDUM( 1 ), RESULT( NTESTS ) * .. * .. External Functions .. LOGICAL LSAME REAL CLANGB, CLANGE, CLANTB, SGET06, SLAMCH EXTERNAL LSAME, CLANGB, CLANGE, CLANTB, SGET06, SLAMCH * .. * .. External Subroutines .. EXTERNAL ALADHD, ALAERH, ALASVM, CERRVX, CGBEQU, CGBSV, $ CGBSVX, CGBT01, CGBT02, CGBT05, CGBTRF, CGBTRS, $ CGET04, CLACPY, CLAQGB, CLARHS, CLASET, CLATB4, $ CLATMS, XLAENV * .. * .. Intrinsic Functions .. INTRINSIC ABS, CMPLX, MAX, MIN * .. * .. Scalars in Common .. LOGICAL LERR, OK CHARACTER*32 SRNAMT INTEGER INFOT, NUNIT * .. * .. Common blocks .. COMMON / INFOC / INFOT, NUNIT, OK, LERR COMMON / SRNAMC / SRNAMT * .. * .. Data statements .. DATA ISEEDY / 1988, 1989, 1990, 1991 / DATA TRANSS / 'N', 'T', 'C' / DATA FACTS / 'F', 'N', 'E' / DATA EQUEDS / 'N', 'R', 'C', 'B' / * .. * .. Executable Statements .. * * Initialize constants and the random number seed. * PATH( 1: 1 ) = 'Complex precision' PATH( 2: 3 ) = 'GB' NRUN = 0 NFAIL = 0 NERRS = 0 DO 10 I = 1, 4 ISEED( I ) = ISEEDY( I ) 10 CONTINUE * * Test the error exits * IF( TSTERR ) $ CALL CERRVX( PATH, NOUT ) INFOT = 0 * * Set the block size and minimum block size for testing. * NB = 1 NBMIN = 2 CALL XLAENV( 1, NB ) CALL XLAENV( 2, NBMIN ) * * Do for each value of N in NVAL * DO 150 IN = 1, NN N = NVAL( IN ) LDB = MAX( N, 1 ) XTYPE = 'N' * * Set limits on the number of loop iterations. * NKL = MAX( 1, MIN( N, 4 ) ) IF( N.EQ.0 ) $ NKL = 1 NKU = NKL NIMAT = NTYPES IF( N.LE.0 ) $ NIMAT = 1 * DO 140 IKL = 1, NKL * * Do for KL = 0, N-1, (3N-1)/4, and (N+1)/4. This order makes * it easier to skip redundant values for small values of N. * IF( IKL.EQ.1 ) THEN KL = 0 ELSE IF( IKL.EQ.2 ) THEN KL = MAX( N-1, 0 ) ELSE IF( IKL.EQ.3 ) THEN KL = ( 3*N-1 ) / 4 ELSE IF( IKL.EQ.4 ) THEN KL = ( N+1 ) / 4 END IF DO 130 IKU = 1, NKU * * Do for KU = 0, N-1, (3N-1)/4, and (N+1)/4. This order * makes it easier to skip redundant values for small * values of N. * IF( IKU.EQ.1 ) THEN KU = 0 ELSE IF( IKU.EQ.2 ) THEN KU = MAX( N-1, 0 ) ELSE IF( IKU.EQ.3 ) THEN KU = ( 3*N-1 ) / 4 ELSE IF( IKU.EQ.4 ) THEN KU = ( N+1 ) / 4 END IF * * Check that A and AFB are big enough to generate this * matrix. * LDA = KL + KU + 1 LDAFB = 2*KL + KU + 1 IF( LDA*N.GT.LA .OR. LDAFB*N.GT.LAFB ) THEN IF( NFAIL.EQ.0 .AND. NERRS.EQ.0 ) $ CALL ALADHD( NOUT, PATH ) IF( LDA*N.GT.LA ) THEN WRITE( NOUT, FMT = 9999 )LA, N, KL, KU, $ N*( KL+KU+1 ) NERRS = NERRS + 1 END IF IF( LDAFB*N.GT.LAFB ) THEN WRITE( NOUT, FMT = 9998 )LAFB, N, KL, KU, $ N*( 2*KL+KU+1 ) NERRS = NERRS + 1 END IF GO TO 130 END IF * DO 120 IMAT = 1, NIMAT * * Do the tests only if DOTYPE( IMAT ) is true. * IF( .NOT.DOTYPE( IMAT ) ) $ GO TO 120 * * Skip types 2, 3, or 4 if the matrix is too small. * ZEROT = IMAT.GE.2 .AND. IMAT.LE.4 IF( ZEROT .AND. N.LT.IMAT-1 ) $ GO TO 120 * * Set up parameters with CLATB4 and generate a * test matrix with CLATMS. * CALL CLATB4( PATH, IMAT, N, N, TYPE, KL, KU, ANORM, $ MODE, CNDNUM, DIST ) RCONDC = ONE / CNDNUM * SRNAMT = 'CLATMS' CALL CLATMS( N, N, DIST, ISEED, TYPE, RWORK, MODE, $ CNDNUM, ANORM, KL, KU, 'Z', A, LDA, WORK, $ INFO ) * * Check the error code from CLATMS. * IF( INFO.NE.0 ) THEN CALL ALAERH( PATH, 'CLATMS', INFO, 0, ' ', N, N, $ KL, KU, -1, IMAT, NFAIL, NERRS, NOUT ) GO TO 120 END IF * * For types 2, 3, and 4, zero one or more columns of * the matrix to test that INFO is returned correctly. * IZERO = 0 IF( ZEROT ) THEN IF( IMAT.EQ.2 ) THEN IZERO = 1 ELSE IF( IMAT.EQ.3 ) THEN IZERO = N ELSE IZERO = N / 2 + 1 END IF IOFF = ( IZERO-1 )*LDA IF( IMAT.LT.4 ) THEN I1 = MAX( 1, KU+2-IZERO ) I2 = MIN( KL+KU+1, KU+1+( N-IZERO ) ) DO 20 I = I1, I2 A( IOFF+I ) = ZERO 20 CONTINUE ELSE DO 40 J = IZERO, N DO 30 I = MAX( 1, KU+2-J ), $ MIN( KL+KU+1, KU+1+( N-J ) ) A( IOFF+I ) = ZERO 30 CONTINUE IOFF = IOFF + LDA 40 CONTINUE END IF END IF * * Save a copy of the matrix A in ASAV. * CALL CLACPY( 'Full', KL+KU+1, N, A, LDA, ASAV, LDA ) * DO 110 IEQUED = 1, 4 EQUED = EQUEDS( IEQUED ) IF( IEQUED.EQ.1 ) THEN NFACT = 3 ELSE NFACT = 1 END IF * DO 100 IFACT = 1, NFACT FACT = FACTS( IFACT ) PREFAC = LSAME( FACT, 'F' ) NOFACT = LSAME( FACT, 'N' ) EQUIL = LSAME( FACT, 'E' ) * IF( ZEROT ) THEN IF( PREFAC ) $ GO TO 100 RCONDO = ZERO RCONDI = ZERO * ELSE IF( .NOT.NOFACT ) THEN * * Compute the condition number for comparison * with the value returned by SGESVX (FACT = * 'N' reuses the condition number from the * previous iteration with FACT = 'F'). * CALL CLACPY( 'Full', KL+KU+1, N, ASAV, LDA, $ AFB( KL+1 ), LDAFB ) IF( EQUIL .OR. IEQUED.GT.1 ) THEN * * Compute row and column scale factors to * equilibrate the matrix A. * CALL CGBEQU( N, N, KL, KU, AFB( KL+1 ), $ LDAFB, S, S( N+1 ), ROWCND, $ COLCND, AMAX, INFO ) IF( INFO.EQ.0 .AND. N.GT.0 ) THEN IF( LSAME( EQUED, 'R' ) ) THEN ROWCND = ZERO COLCND = ONE ELSE IF( LSAME( EQUED, 'C' ) ) THEN ROWCND = ONE COLCND = ZERO ELSE IF( LSAME( EQUED, 'B' ) ) THEN ROWCND = ZERO COLCND = ZERO END IF * * Equilibrate the matrix. * CALL CLAQGB( N, N, KL, KU, AFB( KL+1 ), $ LDAFB, S, S( N+1 ), $ ROWCND, COLCND, AMAX, $ EQUED ) END IF END IF * * Save the condition number of the * non-equilibrated system for use in CGET04. * IF( EQUIL ) THEN ROLDO = RCONDO ROLDI = RCONDI END IF * * Compute the 1-norm and infinity-norm of A. * ANORMO = CLANGB( '1', N, KL, KU, AFB( KL+1 ), $ LDAFB, RWORK ) ANORMI = CLANGB( 'I', N, KL, KU, AFB( KL+1 ), $ LDAFB, RWORK ) * * Factor the matrix A. * CALL CGBTRF( N, N, KL, KU, AFB, LDAFB, IWORK, $ INFO ) * * Form the inverse of A. * CALL CLASET( 'Full', N, N, CMPLX( ZERO ), $ CMPLX( ONE ), WORK, LDB ) SRNAMT = 'CGBTRS' CALL CGBTRS( 'No transpose', N, KL, KU, N, $ AFB, LDAFB, IWORK, WORK, LDB, $ INFO ) * * Compute the 1-norm condition number of A. * AINVNM = CLANGE( '1', N, N, WORK, LDB, $ RWORK ) IF( ANORMO.LE.ZERO .OR. AINVNM.LE.ZERO ) THEN RCONDO = ONE ELSE RCONDO = ( ONE / ANORMO ) / AINVNM END IF * * Compute the infinity-norm condition number * of A. * AINVNM = CLANGE( 'I', N, N, WORK, LDB, $ RWORK ) IF( ANORMI.LE.ZERO .OR. AINVNM.LE.ZERO ) THEN RCONDI = ONE ELSE RCONDI = ( ONE / ANORMI ) / AINVNM END IF END IF * DO 90 ITRAN = 1, NTRAN * * Do for each value of TRANS. * TRANS = TRANSS( ITRAN ) IF( ITRAN.EQ.1 ) THEN RCONDC = RCONDO ELSE RCONDC = RCONDI END IF * * Restore the matrix A. * CALL CLACPY( 'Full', KL+KU+1, N, ASAV, LDA, $ A, LDA ) * * Form an exact solution and set the right hand * side. * SRNAMT = 'CLARHS' CALL CLARHS( PATH, XTYPE, 'Full', TRANS, N, $ N, KL, KU, NRHS, A, LDA, XACT, $ LDB, B, LDB, ISEED, INFO ) XTYPE = 'C' CALL CLACPY( 'Full', N, NRHS, B, LDB, BSAV, $ LDB ) * IF( NOFACT .AND. ITRAN.EQ.1 ) THEN * * --- Test CGBSV --- * * Compute the LU factorization of the matrix * and solve the system. * CALL CLACPY( 'Full', KL+KU+1, N, A, LDA, $ AFB( KL+1 ), LDAFB ) CALL CLACPY( 'Full', N, NRHS, B, LDB, X, $ LDB ) * SRNAMT = 'CGBSV ' CALL CGBSV( N, KL, KU, NRHS, AFB, LDAFB, $ IWORK, X, LDB, INFO ) * * Check error code from CGBSV . * IF( INFO.NE.IZERO ) $ CALL ALAERH( PATH, 'CGBSV ', INFO, $ IZERO, ' ', N, N, KL, KU, $ NRHS, IMAT, NFAIL, NERRS, $ NOUT ) * * Reconstruct matrix from factors and * compute residual. * CALL CGBT01( N, N, KL, KU, A, LDA, AFB, $ LDAFB, IWORK, WORK, $ RESULT( 1 ) ) NT = 1 IF( IZERO.EQ.0 ) THEN * * Compute residual of the computed * solution. * CALL CLACPY( 'Full', N, NRHS, B, LDB, $ WORK, LDB ) CALL CGBT02( 'No transpose', N, N, KL, $ KU, NRHS, A, LDA, X, LDB, $ WORK, LDB, RESULT( 2 ) ) * * Check solution from generated exact * solution. * CALL CGET04( N, NRHS, X, LDB, XACT, $ LDB, RCONDC, RESULT( 3 ) ) NT = 3 END IF * * Print information about the tests that did * not pass the threshold. * DO 50 K = 1, NT IF( RESULT( K ).GE.THRESH ) THEN IF( NFAIL.EQ.0 .AND. NERRS.EQ.0 ) $ CALL ALADHD( NOUT, PATH ) WRITE( NOUT, FMT = 9997 )'CGBSV ', $ N, KL, KU, IMAT, K, RESULT( K ) NFAIL = NFAIL + 1 END IF 50 CONTINUE NRUN = NRUN + NT END IF * * --- Test CGBSVX --- * IF( .NOT.PREFAC ) $ CALL CLASET( 'Full', 2*KL+KU+1, N, $ CMPLX( ZERO ), CMPLX( ZERO ), $ AFB, LDAFB ) CALL CLASET( 'Full', N, NRHS, CMPLX( ZERO ), $ CMPLX( ZERO ), X, LDB ) IF( IEQUED.GT.1 .AND. N.GT.0 ) THEN * * Equilibrate the matrix if FACT = 'F' and * EQUED = 'R', 'C', or 'B'. * CALL CLAQGB( N, N, KL, KU, A, LDA, S, $ S( N+1 ), ROWCND, COLCND, $ AMAX, EQUED ) END IF * * Solve the system and compute the condition * number and error bounds using CGBSVX. * SRNAMT = 'CGBSVX' CALL CGBSVX( FACT, TRANS, N, KL, KU, NRHS, A, $ LDA, AFB, LDAFB, IWORK, EQUED, $ S, S( LDB+1 ), B, LDB, X, LDB, $ RCOND, RWORK, RWORK( NRHS+1 ), $ WORK, RWORK( 2*NRHS+1 ), INFO ) * * Check the error code from CGBSVX. * IF( INFO.NE.IZERO ) $ CALL ALAERH( PATH, 'CGBSVX', INFO, IZERO, $ FACT // TRANS, N, N, KL, KU, $ NRHS, IMAT, NFAIL, NERRS, $ NOUT ) * Compare RWORK(2*NRHS+1) from CGBSVX with the * computed reciprocal pivot growth RPVGRW * IF( INFO.NE.0 .AND. INFO.LE.N) THEN ANRMPV = ZERO DO 70 J = 1, INFO DO 60 I = MAX( KU+2-J, 1 ), $ MIN( N+KU+1-J, KL+KU+1 ) ANRMPV = MAX( ANRMPV, $ ABS( A( I+( J-1 )*LDA ) ) ) 60 CONTINUE 70 CONTINUE RPVGRW = CLANTB( 'M', 'U', 'N', INFO, $ MIN( INFO-1, KL+KU ), $ AFB( MAX( 1, KL+KU+2-INFO ) ), $ LDAFB, RDUM ) IF( RPVGRW.EQ.ZERO ) THEN RPVGRW = ONE ELSE RPVGRW = ANRMPV / RPVGRW END IF ELSE RPVGRW = CLANTB( 'M', 'U', 'N', N, KL+KU, $ AFB, LDAFB, RDUM ) IF( RPVGRW.EQ.ZERO ) THEN RPVGRW = ONE ELSE RPVGRW = CLANGB( 'M', N, KL, KU, A, $ LDA, RDUM ) / RPVGRW END IF END IF RESULT( 7 ) = ABS( RPVGRW-RWORK( 2*NRHS+1 ) ) $ / MAX( RWORK( 2*NRHS+1 ), $ RPVGRW ) / SLAMCH( 'E' ) * IF( .NOT.PREFAC ) THEN * * Reconstruct matrix from factors and * compute residual. * CALL CGBT01( N, N, KL, KU, A, LDA, AFB, $ LDAFB, IWORK, WORK, $ RESULT( 1 ) ) K1 = 1 ELSE K1 = 2 END IF * IF( INFO.EQ.0 ) THEN TRFCON = .FALSE. * * Compute residual of the computed solution. * CALL CLACPY( 'Full', N, NRHS, BSAV, LDB, $ WORK, LDB ) CALL CGBT02( TRANS, N, N, KL, KU, NRHS, $ ASAV, LDA, X, LDB, WORK, LDB, $ RESULT( 2 ) ) * * Check solution from generated exact * solution. * IF( NOFACT .OR. ( PREFAC .AND. $ LSAME( EQUED, 'N' ) ) ) THEN CALL CGET04( N, NRHS, X, LDB, XACT, $ LDB, RCONDC, RESULT( 3 ) ) ELSE IF( ITRAN.EQ.1 ) THEN ROLDC = ROLDO ELSE ROLDC = ROLDI END IF CALL CGET04( N, NRHS, X, LDB, XACT, $ LDB, ROLDC, RESULT( 3 ) ) END IF * * Check the error bounds from iterative * refinement. * CALL CGBT05( TRANS, N, KL, KU, NRHS, ASAV, $ LDA, BSAV, LDB, X, LDB, XACT, $ LDB, RWORK, RWORK( NRHS+1 ), $ RESULT( 4 ) ) ELSE TRFCON = .TRUE. END IF * * Compare RCOND from CGBSVX with the computed * value in RCONDC. * RESULT( 6 ) = SGET06( RCOND, RCONDC ) * * Print information about the tests that did * not pass the threshold. * IF( .NOT.TRFCON ) THEN DO 80 K = K1, NTESTS IF( RESULT( K ).GE.THRESH ) THEN IF( NFAIL.EQ.0 .AND. NERRS.EQ.0 ) $ CALL ALADHD( NOUT, PATH ) IF( PREFAC ) THEN WRITE( NOUT, FMT = 9995 ) $ 'CGBSVX', FACT, TRANS, N, KL, $ KU, EQUED, IMAT, K, $ RESULT( K ) ELSE WRITE( NOUT, FMT = 9996 ) $ 'CGBSVX', FACT, TRANS, N, KL, $ KU, IMAT, K, RESULT( K ) END IF NFAIL = NFAIL + 1 END IF 80 CONTINUE NRUN = NRUN + NTESTS - K1 + 1 ELSE IF( RESULT( 1 ).GE.THRESH .AND. .NOT. $ PREFAC ) THEN IF( NFAIL.EQ.0 .AND. NERRS.EQ.0 ) $ CALL ALADHD( NOUT, PATH ) IF( PREFAC ) THEN WRITE( NOUT, FMT = 9995 )'CGBSVX', $ FACT, TRANS, N, KL, KU, EQUED, $ IMAT, 1, RESULT( 1 ) ELSE WRITE( NOUT, FMT = 9996 )'CGBSVX', $ FACT, TRANS, N, KL, KU, IMAT, 1, $ RESULT( 1 ) END IF NFAIL = NFAIL + 1 NRUN = NRUN + 1 END IF IF( RESULT( 6 ).GE.THRESH ) THEN IF( NFAIL.EQ.0 .AND. NERRS.EQ.0 ) $ CALL ALADHD( NOUT, PATH ) IF( PREFAC ) THEN WRITE( NOUT, FMT = 9995 )'CGBSVX', $ FACT, TRANS, N, KL, KU, EQUED, $ IMAT, 6, RESULT( 6 ) ELSE WRITE( NOUT, FMT = 9996 )'CGBSVX', $ FACT, TRANS, N, KL, KU, IMAT, 6, $ RESULT( 6 ) END IF NFAIL = NFAIL + 1 NRUN = NRUN + 1 END IF IF( RESULT( 7 ).GE.THRESH ) THEN IF( NFAIL.EQ.0 .AND. NERRS.EQ.0 ) $ CALL ALADHD( NOUT, PATH ) IF( PREFAC ) THEN WRITE( NOUT, FMT = 9995 )'CGBSVX', $ FACT, TRANS, N, KL, KU, EQUED, $ IMAT, 7, RESULT( 7 ) ELSE WRITE( NOUT, FMT = 9996 )'CGBSVX', $ FACT, TRANS, N, KL, KU, IMAT, 7, $ RESULT( 7 ) END IF NFAIL = NFAIL + 1 NRUN = NRUN + 1 END IF END IF 90 CONTINUE 100 CONTINUE 110 CONTINUE 120 CONTINUE 130 CONTINUE 140 CONTINUE 150 CONTINUE * * Print a summary of the results. * CALL ALASVM( PATH, NOUT, NFAIL, NRUN, NERRS ) * 9999 FORMAT( ' *** In CDRVGB, LA=', I5, ' is too small for N=', I5, $ ', KU=', I5, ', KL=', I5, / ' ==> Increase LA to at least ', $ I5 ) 9998 FORMAT( ' *** In CDRVGB, LAFB=', I5, ' is too small for N=', I5, $ ', KU=', I5, ', KL=', I5, / $ ' ==> Increase LAFB to at least ', I5 ) 9997 FORMAT( 1X, A, ', N=', I5, ', KL=', I5, ', KU=', I5, ', type ', $ I1, ', test(', I1, ')=', G12.5 ) 9996 FORMAT( 1X, A, '( ''', A1, ''',''', A1, ''',', I5, ',', I5, ',', $ I5, ',...), type ', I1, ', test(', I1, ')=', G12.5 ) 9995 FORMAT( 1X, A, '( ''', A1, ''',''', A1, ''',', I5, ',', I5, ',', $ I5, ',...), EQUED=''', A1, ''', type ', I1, ', test(', I1, $ ')=', G12.5 ) * RETURN * * End of CDRVGB * END