 |
LAPACK 3.3.1
Linear Algebra PACKage
|
|
LAPACK 3.3.1
Linear Algebra PACKage
|
00001 SUBROUTINE DCHKGK( NIN, NOUT ) 00002 * 00003 * -- LAPACK test routine (version 3.1) -- 00004 * Univ. of Tennessee, Univ. of California Berkeley and NAG Ltd.. 00005 * November 2006 00006 * 00007 * .. Scalar Arguments .. 00008 INTEGER NIN, NOUT 00009 * .. 00010 * 00011 * Purpose 00012 * ======= 00013 * 00014 * DCHKGK tests DGGBAK, a routine for backward balancing of 00015 * a matrix pair (A, B). 00016 * 00017 * Arguments 00018 * ========= 00019 * 00020 * NIN (input) INTEGER 00021 * The logical unit number for input. NIN > 0. 00022 * 00023 * NOUT (input) INTEGER 00024 * The logical unit number for output. NOUT > 0. 00025 * 00026 * ===================================================================== 00027 * 00028 * .. Parameters .. 00029 INTEGER LDA, LDB, LDVL, LDVR 00030 PARAMETER ( LDA = 50, LDB = 50, LDVL = 50, LDVR = 50 ) 00031 INTEGER LDE, LDF, LDWORK 00032 PARAMETER ( LDE = 50, LDF = 50, LDWORK = 50 ) 00033 DOUBLE PRECISION ZERO, ONE 00034 PARAMETER ( ZERO = 0.0D+0, ONE = 1.0D+0 ) 00035 * .. 00036 * .. Local Scalars .. 00037 INTEGER I, IHI, ILO, INFO, J, KNT, M, N, NINFO 00038 DOUBLE PRECISION ANORM, BNORM, EPS, RMAX, VMAX 00039 * .. 00040 * .. Local Arrays .. 00041 INTEGER LMAX( 4 ) 00042 DOUBLE PRECISION A( LDA, LDA ), AF( LDA, LDA ), B( LDB, LDB ), 00043 $ BF( LDB, LDB ), E( LDE, LDE ), F( LDF, LDF ), 00044 $ LSCALE( LDA ), RSCALE( LDA ), VL( LDVL, LDVL ), 00045 $ VLF( LDVL, LDVL ), VR( LDVR, LDVR ), 00046 $ VRF( LDVR, LDVR ), WORK( LDWORK, LDWORK ) 00047 * .. 00048 * .. External Functions .. 00049 DOUBLE PRECISION DLAMCH, DLANGE 00050 EXTERNAL DLAMCH, DLANGE 00051 * .. 00052 * .. External Subroutines .. 00053 EXTERNAL DGEMM, DGGBAK, DGGBAL, DLACPY 00054 * .. 00055 * .. Intrinsic Functions .. 00056 INTRINSIC ABS, MAX 00057 * .. 00058 * .. Executable Statements .. 00059 * 00060 * Initialization 00061 * 00062 LMAX( 1 ) = 0 00063 LMAX( 2 ) = 0 00064 LMAX( 3 ) = 0 00065 LMAX( 4 ) = 0 00066 NINFO = 0 00067 KNT = 0 00068 RMAX = ZERO 00069 * 00070 EPS = DLAMCH( 'Precision' ) 00071 * 00072 10 CONTINUE 00073 READ( NIN, FMT = * )N, M 00074 IF( N.EQ.0 ) 00075 $ GO TO 100 00076 * 00077 DO 20 I = 1, N 00078 READ( NIN, FMT = * )( A( I, J ), J = 1, N ) 00079 20 CONTINUE 00080 * 00081 DO 30 I = 1, N 00082 READ( NIN, FMT = * )( B( I, J ), J = 1, N ) 00083 30 CONTINUE 00084 * 00085 DO 40 I = 1, N 00086 READ( NIN, FMT = * )( VL( I, J ), J = 1, M ) 00087 40 CONTINUE 00088 * 00089 DO 50 I = 1, N 00090 READ( NIN, FMT = * )( VR( I, J ), J = 1, M ) 00091 50 CONTINUE 00092 * 00093 KNT = KNT + 1 00094 * 00095 ANORM = DLANGE( 'M', N, N, A, LDA, WORK ) 00096 BNORM = DLANGE( 'M', N, N, B, LDB, WORK ) 00097 * 00098 CALL DLACPY( 'FULL', N, N, A, LDA, AF, LDA ) 00099 CALL DLACPY( 'FULL', N, N, B, LDB, BF, LDB ) 00100 * 00101 CALL DGGBAL( 'B', N, A, LDA, B, LDB, ILO, IHI, LSCALE, RSCALE, 00102 $ WORK, INFO ) 00103 IF( INFO.NE.0 ) THEN 00104 NINFO = NINFO + 1 00105 LMAX( 1 ) = KNT 00106 END IF 00107 * 00108 CALL DLACPY( 'FULL', N, M, VL, LDVL, VLF, LDVL ) 00109 CALL DLACPY( 'FULL', N, M, VR, LDVR, VRF, LDVR ) 00110 * 00111 CALL DGGBAK( 'B', 'L', N, ILO, IHI, LSCALE, RSCALE, M, VL, LDVL, 00112 $ INFO ) 00113 IF( INFO.NE.0 ) THEN 00114 NINFO = NINFO + 1 00115 LMAX( 2 ) = KNT 00116 END IF 00117 * 00118 CALL DGGBAK( 'B', 'R', N, ILO, IHI, LSCALE, RSCALE, M, VR, LDVR, 00119 $ INFO ) 00120 IF( INFO.NE.0 ) THEN 00121 NINFO = NINFO + 1 00122 LMAX( 3 ) = KNT 00123 END IF 00124 * 00125 * Test of DGGBAK 00126 * 00127 * Check tilde(VL)'*A*tilde(VR) - VL'*tilde(A)*VR 00128 * where tilde(A) denotes the transformed matrix. 00129 * 00130 CALL DGEMM( 'N', 'N', N, M, N, ONE, AF, LDA, VR, LDVR, ZERO, WORK, 00131 $ LDWORK ) 00132 CALL DGEMM( 'T', 'N', M, M, N, ONE, VL, LDVL, WORK, LDWORK, ZERO, 00133 $ E, LDE ) 00134 * 00135 CALL DGEMM( 'N', 'N', N, M, N, ONE, A, LDA, VRF, LDVR, ZERO, WORK, 00136 $ LDWORK ) 00137 CALL DGEMM( 'T', 'N', M, M, N, ONE, VLF, LDVL, WORK, LDWORK, ZERO, 00138 $ F, LDF ) 00139 * 00140 VMAX = ZERO 00141 DO 70 J = 1, M 00142 DO 60 I = 1, M 00143 VMAX = MAX( VMAX, ABS( E( I, J )-F( I, J ) ) ) 00144 60 CONTINUE 00145 70 CONTINUE 00146 VMAX = VMAX / ( EPS*MAX( ANORM, BNORM ) ) 00147 IF( VMAX.GT.RMAX ) THEN 00148 LMAX( 4 ) = KNT 00149 RMAX = VMAX 00150 END IF 00151 * 00152 * Check tilde(VL)'*B*tilde(VR) - VL'*tilde(B)*VR 00153 * 00154 CALL DGEMM( 'N', 'N', N, M, N, ONE, BF, LDB, VR, LDVR, ZERO, WORK, 00155 $ LDWORK ) 00156 CALL DGEMM( 'T', 'N', M, M, N, ONE, VL, LDVL, WORK, LDWORK, ZERO, 00157 $ E, LDE ) 00158 * 00159 CALL DGEMM( 'N', 'N', N, M, N, ONE, B, LDB, VRF, LDVR, ZERO, WORK, 00160 $ LDWORK ) 00161 CALL DGEMM( 'T', 'N', M, M, N, ONE, VLF, LDVL, WORK, LDWORK, ZERO, 00162 $ F, LDF ) 00163 * 00164 VMAX = ZERO 00165 DO 90 J = 1, M 00166 DO 80 I = 1, M 00167 VMAX = MAX( VMAX, ABS( E( I, J )-F( I, J ) ) ) 00168 80 CONTINUE 00169 90 CONTINUE 00170 VMAX = VMAX / ( EPS*MAX( ANORM, BNORM ) ) 00171 IF( VMAX.GT.RMAX ) THEN 00172 LMAX( 4 ) = KNT 00173 RMAX = VMAX 00174 END IF 00175 * 00176 GO TO 10 00177 * 00178 100 CONTINUE 00179 * 00180 WRITE( NOUT, FMT = 9999 ) 00181 9999 FORMAT( 1X, '.. test output of DGGBAK .. ' ) 00182 * 00183 WRITE( NOUT, FMT = 9998 )RMAX 00184 9998 FORMAT( ' value of largest test error =', D12.3 ) 00185 WRITE( NOUT, FMT = 9997 )LMAX( 1 ) 00186 9997 FORMAT( ' example number where DGGBAL info is not 0 =', I4 ) 00187 WRITE( NOUT, FMT = 9996 )LMAX( 2 ) 00188 9996 FORMAT( ' example number where DGGBAK(L) info is not 0 =', I4 ) 00189 WRITE( NOUT, FMT = 9995 )LMAX( 3 ) 00190 9995 FORMAT( ' example number where DGGBAK(R) info is not 0 =', I4 ) 00191 WRITE( NOUT, FMT = 9994 )LMAX( 4 ) 00192 9994 FORMAT( ' example number having largest error =', I4 ) 00193 WRITE( NOUT, FMT = 9993 )NINFO 00194 9993 FORMAT( ' number of examples where info is not 0 =', I4 ) 00195 WRITE( NOUT, FMT = 9992 )KNT 00196 9992 FORMAT( ' total number of examples tested =', I4 ) 00197 * 00198 RETURN 00199 * 00200 * End of DCHKGK 00201 * 00202 END
1.7.3 
