LAPACK 3.3.0
|
00001 SUBROUTINE CTRT01( UPLO, DIAG, N, A, LDA, AINV, LDAINV, RCOND, 00002 $ RWORK, RESID ) 00003 * 00004 * -- LAPACK test routine (version 3.1) -- 00005 * Univ. of Tennessee, Univ. of California Berkeley and NAG Ltd.. 00006 * November 2006 00007 * 00008 * .. Scalar Arguments .. 00009 CHARACTER DIAG, UPLO 00010 INTEGER LDA, LDAINV, N 00011 REAL RCOND, RESID 00012 * .. 00013 * .. Array Arguments .. 00014 REAL RWORK( * ) 00015 COMPLEX A( LDA, * ), AINV( LDAINV, * ) 00016 * .. 00017 * 00018 * Purpose 00019 * ======= 00020 * 00021 * CTRT01 computes the residual for a triangular matrix A times its 00022 * inverse: 00023 * RESID = norm( A*AINV - I ) / ( N * norm(A) * norm(AINV) * EPS ), 00024 * where EPS is the machine epsilon. 00025 * 00026 * Arguments 00027 * ========== 00028 * 00029 * UPLO (input) CHARACTER*1 00030 * Specifies whether the matrix A is upper or lower triangular. 00031 * = 'U': Upper triangular 00032 * = 'L': Lower triangular 00033 * 00034 * DIAG (input) CHARACTER*1 00035 * Specifies whether or not the matrix A is unit triangular. 00036 * = 'N': Non-unit triangular 00037 * = 'U': Unit triangular 00038 * 00039 * N (input) INTEGER 00040 * The order of the matrix A. N >= 0. 00041 * 00042 * A (input) COMPLEX array, dimension (LDA,N) 00043 * The triangular matrix A. If UPLO = 'U', the leading n by n 00044 * upper triangular part of the array A contains the upper 00045 * triangular matrix, and the strictly lower triangular part of 00046 * A is not referenced. If UPLO = 'L', the leading n by n lower 00047 * triangular part of the array A contains the lower triangular 00048 * matrix, and the strictly upper triangular part of A is not 00049 * referenced. If DIAG = 'U', the diagonal elements of A are 00050 * also not referenced and are assumed to be 1. 00051 * 00052 * LDA (input) INTEGER 00053 * The leading dimension of the array A. LDA >= max(1,N). 00054 * 00055 * AINV (input) COMPLEX array, dimension (LDAINV,N) 00056 * On entry, the (triangular) inverse of the matrix A, in the 00057 * same storage format as A. 00058 * On exit, the contents of AINV are destroyed. 00059 * 00060 * LDAINV (input) INTEGER 00061 * The leading dimension of the array AINV. LDAINV >= max(1,N). 00062 * 00063 * RCOND (output) REAL 00064 * The reciprocal condition number of A, computed as 00065 * 1/(norm(A) * norm(AINV)). 00066 * 00067 * RWORK (workspace) REAL array, dimension (N) 00068 * 00069 * RESID (output) REAL 00070 * norm(A*AINV - I) / ( N * norm(A) * norm(AINV) * EPS ) 00071 * 00072 * ===================================================================== 00073 * 00074 * .. Parameters .. 00075 REAL ZERO, ONE 00076 PARAMETER ( ZERO = 0.0E+0, ONE = 1.0E+0 ) 00077 * .. 00078 * .. Local Scalars .. 00079 INTEGER J 00080 REAL AINVNM, ANORM, EPS 00081 * .. 00082 * .. External Functions .. 00083 LOGICAL LSAME 00084 REAL CLANTR, SLAMCH 00085 EXTERNAL LSAME, CLANTR, SLAMCH 00086 * .. 00087 * .. External Subroutines .. 00088 EXTERNAL CTRMV 00089 * .. 00090 * .. Intrinsic Functions .. 00091 INTRINSIC REAL 00092 * .. 00093 * .. Executable Statements .. 00094 * 00095 * Quick exit if N = 0 00096 * 00097 IF( N.LE.0 ) THEN 00098 RCOND = ONE 00099 RESID = ZERO 00100 RETURN 00101 END IF 00102 * 00103 * Exit with RESID = 1/EPS if ANORM = 0 or AINVNM = 0. 00104 * 00105 EPS = SLAMCH( 'Epsilon' ) 00106 ANORM = CLANTR( '1', UPLO, DIAG, N, N, A, LDA, RWORK ) 00107 AINVNM = CLANTR( '1', UPLO, DIAG, N, N, AINV, LDAINV, RWORK ) 00108 IF( ANORM.LE.ZERO .OR. AINVNM.LE.ZERO ) THEN 00109 RCOND = ZERO 00110 RESID = ONE / EPS 00111 RETURN 00112 END IF 00113 RCOND = ( ONE / ANORM ) / AINVNM 00114 * 00115 * Set the diagonal of AINV to 1 if AINV has unit diagonal. 00116 * 00117 IF( LSAME( DIAG, 'U' ) ) THEN 00118 DO 10 J = 1, N 00119 AINV( J, J ) = ONE 00120 10 CONTINUE 00121 END IF 00122 * 00123 * Compute A * AINV, overwriting AINV. 00124 * 00125 IF( LSAME( UPLO, 'U' ) ) THEN 00126 DO 20 J = 1, N 00127 CALL CTRMV( 'Upper', 'No transpose', DIAG, J, A, LDA, 00128 $ AINV( 1, J ), 1 ) 00129 20 CONTINUE 00130 ELSE 00131 DO 30 J = 1, N 00132 CALL CTRMV( 'Lower', 'No transpose', DIAG, N-J+1, A( J, J ), 00133 $ LDA, AINV( J, J ), 1 ) 00134 30 CONTINUE 00135 END IF 00136 * 00137 * Subtract 1 from each diagonal element to form A*AINV - I. 00138 * 00139 DO 40 J = 1, N 00140 AINV( J, J ) = AINV( J, J ) - ONE 00141 40 CONTINUE 00142 * 00143 * Compute norm(A*AINV - I) / (N * norm(A) * norm(AINV) * EPS) 00144 * 00145 RESID = CLANTR( '1', UPLO, 'Non-unit', N, N, AINV, LDAINV, RWORK ) 00146 * 00147 RESID = ( ( RESID*RCOND ) / REAL( N ) ) / EPS 00148 * 00149 RETURN 00150 * 00151 * End of CTRT01 00152 * 00153 END