LAPACK 3.3.0
|
00001 SUBROUTINE ZPTCON( N, D, E, ANORM, RCOND, RWORK, INFO ) 00002 * 00003 * -- LAPACK routine (version 3.2) -- 00004 * -- LAPACK is a software package provided by Univ. of Tennessee, -- 00005 * -- Univ. of California Berkeley, Univ. of Colorado Denver and NAG Ltd..-- 00006 * November 2006 00007 * 00008 * .. Scalar Arguments .. 00009 INTEGER INFO, N 00010 DOUBLE PRECISION ANORM, RCOND 00011 * .. 00012 * .. Array Arguments .. 00013 DOUBLE PRECISION D( * ), RWORK( * ) 00014 COMPLEX*16 E( * ) 00015 * .. 00016 * 00017 * Purpose 00018 * ======= 00019 * 00020 * ZPTCON computes the reciprocal of the condition number (in the 00021 * 1-norm) of a complex Hermitian positive definite tridiagonal matrix 00022 * using the factorization A = L*D*L**H or A = U**H*D*U computed by 00023 * ZPTTRF. 00024 * 00025 * Norm(inv(A)) is computed by a direct method, and the reciprocal of 00026 * the condition number is computed as 00027 * RCOND = 1 / (ANORM * norm(inv(A))). 00028 * 00029 * Arguments 00030 * ========= 00031 * 00032 * N (input) INTEGER 00033 * The order of the matrix A. N >= 0. 00034 * 00035 * D (input) DOUBLE PRECISION array, dimension (N) 00036 * The n diagonal elements of the diagonal matrix D from the 00037 * factorization of A, as computed by ZPTTRF. 00038 * 00039 * E (input) COMPLEX*16 array, dimension (N-1) 00040 * The (n-1) off-diagonal elements of the unit bidiagonal factor 00041 * U or L from the factorization of A, as computed by ZPTTRF. 00042 * 00043 * ANORM (input) DOUBLE PRECISION 00044 * The 1-norm of the original matrix A. 00045 * 00046 * RCOND (output) DOUBLE PRECISION 00047 * The reciprocal of the condition number of the matrix A, 00048 * computed as RCOND = 1/(ANORM * AINVNM), where AINVNM is the 00049 * 1-norm of inv(A) computed in this routine. 00050 * 00051 * RWORK (workspace) DOUBLE PRECISION array, dimension (N) 00052 * 00053 * INFO (output) INTEGER 00054 * = 0: successful exit 00055 * < 0: if INFO = -i, the i-th argument had an illegal value 00056 * 00057 * Further Details 00058 * =============== 00059 * 00060 * The method used is described in Nicholas J. Higham, "Efficient 00061 * Algorithms for Computing the Condition Number of a Tridiagonal 00062 * Matrix", SIAM J. Sci. Stat. Comput., Vol. 7, No. 1, January 1986. 00063 * 00064 * ===================================================================== 00065 * 00066 * .. Parameters .. 00067 DOUBLE PRECISION ONE, ZERO 00068 PARAMETER ( ONE = 1.0D+0, ZERO = 0.0D+0 ) 00069 * .. 00070 * .. Local Scalars .. 00071 INTEGER I, IX 00072 DOUBLE PRECISION AINVNM 00073 * .. 00074 * .. External Functions .. 00075 INTEGER IDAMAX 00076 EXTERNAL IDAMAX 00077 * .. 00078 * .. External Subroutines .. 00079 EXTERNAL XERBLA 00080 * .. 00081 * .. Intrinsic Functions .. 00082 INTRINSIC ABS 00083 * .. 00084 * .. Executable Statements .. 00085 * 00086 * Test the input arguments. 00087 * 00088 INFO = 0 00089 IF( N.LT.0 ) THEN 00090 INFO = -1 00091 ELSE IF( ANORM.LT.ZERO ) THEN 00092 INFO = -4 00093 END IF 00094 IF( INFO.NE.0 ) THEN 00095 CALL XERBLA( 'ZPTCON', -INFO ) 00096 RETURN 00097 END IF 00098 * 00099 * Quick return if possible 00100 * 00101 RCOND = ZERO 00102 IF( N.EQ.0 ) THEN 00103 RCOND = ONE 00104 RETURN 00105 ELSE IF( ANORM.EQ.ZERO ) THEN 00106 RETURN 00107 END IF 00108 * 00109 * Check that D(1:N) is positive. 00110 * 00111 DO 10 I = 1, N 00112 IF( D( I ).LE.ZERO ) 00113 $ RETURN 00114 10 CONTINUE 00115 * 00116 * Solve M(A) * x = e, where M(A) = (m(i,j)) is given by 00117 * 00118 * m(i,j) = abs(A(i,j)), i = j, 00119 * m(i,j) = -abs(A(i,j)), i .ne. j, 00120 * 00121 * and e = [ 1, 1, ..., 1 ]'. Note M(A) = M(L)*D*M(L)'. 00122 * 00123 * Solve M(L) * x = e. 00124 * 00125 RWORK( 1 ) = ONE 00126 DO 20 I = 2, N 00127 RWORK( I ) = ONE + RWORK( I-1 )*ABS( E( I-1 ) ) 00128 20 CONTINUE 00129 * 00130 * Solve D * M(L)' * x = b. 00131 * 00132 RWORK( N ) = RWORK( N ) / D( N ) 00133 DO 30 I = N - 1, 1, -1 00134 RWORK( I ) = RWORK( I ) / D( I ) + RWORK( I+1 )*ABS( E( I ) ) 00135 30 CONTINUE 00136 * 00137 * Compute AINVNM = max(x(i)), 1<=i<=n. 00138 * 00139 IX = IDAMAX( N, RWORK, 1 ) 00140 AINVNM = ABS( RWORK( IX ) ) 00141 * 00142 * Compute the reciprocal condition number. 00143 * 00144 IF( AINVNM.NE.ZERO ) 00145 $ RCOND = ( ONE / AINVNM ) / ANORM 00146 * 00147 RETURN 00148 * 00149 * End of ZPTCON 00150 * 00151 END