1 *> \brief \b CLANHB returns the value of the 1-norm, or the Frobenius norm, or the infinity norm, or the element of largest absolute value of a Hermitian band matrix.
3 * =========== DOCUMENTATION ===========
5 * Online html documentation available at
6 * http://www.netlib.org/lapack/explore-html/
9 *> Download CLANHB + dependencies
10 *> <a href="http://www.netlib.org/cgi-bin/netlibfiles.tgz?format=tgz&filename=/lapack/lapack_routine/clanhb.f">
12 *> <a href="http://www.netlib.org/cgi-bin/netlibfiles.zip?format=zip&filename=/lapack/lapack_routine/clanhb.f">
14 *> <a href="http://www.netlib.org/cgi-bin/netlibfiles.txt?format=txt&filename=/lapack/lapack_routine/clanhb.f">
21 * REAL FUNCTION CLANHB( NORM, UPLO, N, K, AB, LDAB,
24 * .. Scalar Arguments ..
25 * CHARACTER NORM, UPLO
28 * .. Array Arguments ..
30 * COMPLEX AB( LDAB, * )
39 *> CLANHB returns the value of the one norm, or the Frobenius norm, or
40 *> the infinity norm, or the element of largest absolute value of an
41 *> n by n hermitian band matrix A, with k super-diagonals.
47 *> CLANHB = ( max(abs(A(i,j))), NORM = 'M' or 'm'
49 *> ( norm1(A), NORM = '1', 'O' or 'o'
51 *> ( normI(A), NORM = 'I' or 'i'
53 *> ( normF(A), NORM = 'F', 'f', 'E' or 'e'
55 *> where norm1 denotes the one norm of a matrix (maximum column sum),
56 *> normI denotes the infinity norm of a matrix (maximum row sum) and
57 *> normF denotes the Frobenius norm of a matrix (square root of sum of
58 *> squares). Note that max(abs(A(i,j))) is not a consistent matrix norm.
66 *> NORM is CHARACTER*1
67 *> Specifies the value to be returned in CLANHB as described
73 *> UPLO is CHARACTER*1
74 *> Specifies whether the upper or lower triangular part of the
75 *> band matrix A is supplied.
76 *> = 'U': Upper triangular
77 *> = 'L': Lower triangular
83 *> The order of the matrix A. N >= 0. When N = 0, CLANHB is
90 *> The number of super-diagonals or sub-diagonals of the
91 *> band matrix A. K >= 0.
96 *> AB is COMPLEX array, dimension (LDAB,N)
97 *> The upper or lower triangle of the hermitian band matrix A,
98 *> stored in the first K+1 rows of AB. The j-th column of A is
99 *> stored in the j-th column of the array AB as follows:
100 *> if UPLO = 'U', AB(k+1+i-j,j) = A(i,j) for max(1,j-k)<=i<=j;
101 *> if UPLO = 'L', AB(1+i-j,j) = A(i,j) for j<=i<=min(n,j+k).
102 *> Note that the imaginary parts of the diagonal elements need
103 *> not be set and are assumed to be zero.
109 *> The leading dimension of the array AB. LDAB >= K+1.
114 *> WORK is REAL array, dimension (MAX(1,LWORK)),
115 *> where LWORK >= N when NORM = 'I' or '1' or 'O'; otherwise,
116 *> WORK is not referenced.
122 *> \author Univ. of Tennessee
123 *> \author Univ. of California Berkeley
124 *> \author Univ. of Colorado Denver
127 *> \date September 2012
129 *> \ingroup complexOTHERauxiliary
131 * =====================================================================
132 REAL FUNCTION CLANHB( NORM, UPLO, N, K, AB, LDAB,
135 * -- LAPACK auxiliary routine (version 3.4.2) --
136 * -- LAPACK is a software package provided by Univ. of Tennessee, --
137 * -- Univ. of California Berkeley, Univ. of Colorado Denver and NAG Ltd..--
140 * .. Scalar Arguments ..
144 * .. Array Arguments ..
146 COMPLEX AB( LDAB, * )
149 * =====================================================================
153 PARAMETER ( ONE = 1.0E+0, ZERO = 0.0E+0 )
155 * .. Local Scalars ..
157 REAL ABSA, SCALE, SUM, VALUE
159 * .. External Functions ..
160 LOGICAL LSAME, SISNAN
161 EXTERNAL LSAME, SISNAN
163 * .. External Subroutines ..
166 * .. Intrinsic Functions ..
167 INTRINSIC ABS, MAX, MIN, REAL, SQRT
169 * .. Executable Statements ..
173 ELSE IF( LSAME( NORM, 'M' ) ) THEN
175 * Find max(abs(A(i,j))).
178 IF( LSAME( UPLO, 'U' ) ) THEN
180 DO 10 I = MAX( K+2-J, 1 ), K
181 SUM = ABS( AB( I, J ) )
182 IF( VALUE .LT. SUM .OR. SISNAN( SUM ) ) VALUE = SUM
184 SUM = ABS( REAL( AB( K+1, J ) ) )
185 IF( VALUE .LT. SUM .OR. SISNAN( SUM ) ) VALUE = SUM
189 SUM = ABS( REAL( AB( 1, J ) ) )
190 IF( VALUE .LT. SUM .OR. SISNAN( SUM ) ) VALUE = SUM
191 DO 30 I = 2, MIN( N+1-J, K+1 )
192 SUM = ABS( AB( I, J ) )
193 IF( VALUE .LT. SUM .OR. SISNAN( SUM ) ) VALUE = SUM
197 ELSE IF( ( LSAME( NORM, 'I' ) ) .OR. ( LSAME( NORM, 'O' ) ) .OR.
198 $ ( NORM.EQ.'1' ) ) THEN
200 * Find normI(A) ( = norm1(A), since A is hermitian).
203 IF( LSAME( UPLO, 'U' ) ) THEN
207 DO 50 I = MAX( 1, J-K ), J - 1
208 ABSA = ABS( AB( L+I, J ) )
210 WORK( I ) = WORK( I ) + ABSA
212 WORK( J ) = SUM + ABS( REAL( AB( K+1, J ) ) )
216 IF( VALUE .LT. SUM .OR. SISNAN( SUM ) ) VALUE = SUM
223 SUM = WORK( J ) + ABS( REAL( AB( 1, J ) ) )
225 DO 90 I = J + 1, MIN( N, J+K )
226 ABSA = ABS( AB( L+I, J ) )
228 WORK( I ) = WORK( I ) + ABSA
230 IF( VALUE .LT. SUM .OR. SISNAN( SUM ) ) VALUE = SUM
233 ELSE IF( ( LSAME( NORM, 'F' ) ) .OR. ( LSAME( NORM, 'E' ) ) ) THEN
240 IF( LSAME( UPLO, 'U' ) ) THEN
242 CALL CLASSQ( MIN( J-1, K ), AB( MAX( K+2-J, 1 ), J ),
248 CALL CLASSQ( MIN( N-J, K ), AB( 2, J ), 1, SCALE,
258 IF( REAL( AB( L, J ) ).NE.ZERO ) THEN
259 ABSA = ABS( REAL( AB( L, J ) ) )
260 IF( SCALE.LT.ABSA ) THEN
261 SUM = ONE + SUM*( SCALE / ABSA )**2
264 SUM = SUM + ( ABSA / SCALE )**2
268 VALUE = SCALE*SQRT( SUM )