F08HEF (DSBTRD) (PDF version)
F08 Chapter Contents
F08 Chapter Introduction
NAG Library Manual

NAG Library Routine Document

F08HEF (DSBTRD)

Note:  before using this routine, please read the Users' Note for your implementation to check the interpretation of bold italicised terms and other implementation-dependent details.

+ Contents

    1  Purpose
    7  Accuracy

1  Purpose

F08HEF (DSBTRD) reduces a real symmetric band matrix to tridiagonal form.

2  Specification

SUBROUTINE F08HEF ( VECT, UPLO, N, KD, AB, LDAB, D, E, Q, LDQ, WORK, INFO)
INTEGER  N, KD, LDAB, LDQ, INFO
REAL (KIND=nag_wp)  AB(LDAB,*), D(N), E(N-1), Q(LDQ,*), WORK(N)
CHARACTER(1)  VECT, UPLO
The routine may be called by its LAPACK name dsbtrd.

3  Description

F08HEF (DSBTRD) reduces a symmetric band matrix A to symmetric tridiagonal form T by an orthogonal similarity transformation:
T = QT A Q .
The orthogonal matrix Q is determined as a product of Givens rotation matrices, and may be formed explicitly by the routine if required.
The routine uses a vectorizable form of the reduction, due to Kaufman (1984).

4  References

Kaufman L (1984) Banded eigenvalue solvers on vector machines ACM Trans. Math. Software 10 73–86
Parlett B N (1998) The Symmetric Eigenvalue Problem SIAM, Philadelphia

5  Parameters

1:     VECT – CHARACTER(1)Input
Constraint: VECT='V', 'U' or 'N'.
2:     UPLO – CHARACTER(1)Input
On entry: indicates whether the upper or lower triangular part of A is stored.
UPLO='U'
The upper triangular part of A is stored.
UPLO='L'
The lower triangular part of A is stored.
Constraint: UPLO='U' or 'L'.
3:     N – INTEGERInput
On entry: n, the order of the matrix A.
Constraint: N0.
4:     KD – INTEGERInput
5:     AB(LDAB,*) – REAL (KIND=nag_wp) arrayInput/Output
Note: the second dimension of the array AB must be at least max1,N.
On entry: the upper or lower triangle of the n by n symmetric band matrix A.
The matrix is stored in rows 1 to kd+1, more precisely,
  • if UPLO='U', the elements of the upper triangle of A within the band must be stored with element Aij in ABkd+1+i-jj​ for ​max1,j-kdij;
  • if UPLO='L', the elements of the lower triangle of A within the band must be stored with element Aij in AB1+i-jj​ for ​jiminn,j+kd.
On exit: AB is overwritten by values generated during the reduction to tridiagonal form.
The first superdiagonal and the diagonal of the tridiagonal matrix T are returned in AB using the same storage format as described above.
6:     LDAB – INTEGERInput
On entry: the first dimension of the array AB as declared in the (sub)program from which F08HEF (DSBTRD) is called.
Constraint: LDABmax1,KD+1.
7:     D(N) – REAL (KIND=nag_wp) arrayOutput
On exit: the diagonal elements of the tridiagonal matrix T.
8:     E(N-1) – REAL (KIND=nag_wp) arrayOutput
9:     Q(LDQ,*) – REAL (KIND=nag_wp) arrayInput/Output
Note: the second dimension of the array Q must be at least max1,N if VECT='V' or 'U' and at least 1 if VECT='N'.
On entry: if VECT='U', Q must contain the matrix formed in a previous stage of the reduction (for example, the reduction of a banded symmetric-definite generalized eigenproblem); otherwise Q need not be set.
On exit: if VECT='V' or 'U', the n by n matrix Q.
If VECT='N', Q is not referenced.
10:   LDQ – INTEGERInput
On entry: the first dimension of the array Q as declared in the (sub)program from which F08HEF (DSBTRD) is called.
Constraints:
  • if VECT='V' or 'U', LDQmax1,N;
  • if VECT='N', LDQ1.
11:   WORK(N) – REAL (KIND=nag_wp) arrayWorkspace
12:   INFO – INTEGEROutput

6  Error Indicators and Warnings

Errors or warnings detected by the routine:
INFO<0
If INFO=-i, argument i had an illegal value. An explanatory message is output, and execution of the program is terminated.

7  Accuracy

The computed tridiagonal matrix T is exactly similar to a nearby matrix A+E, where
E2 c n ε A2 ,
cn is a modestly increasing function of n, and ε is the machine precision.
The elements of T themselves may be sensitive to small perturbations in A or to rounding errors in the computation, but this does not affect the stability of the eigenvalues and eigenvectors.
The computed matrix Q differs from an exactly orthogonal matrix by a matrix E such that
E2 = Oε ,
where ε is the machine precision.

8  Further Comments

The total number of floating point operations is approximately 6n2k if VECT='N' with 3n3k-1/k additional operations if VECT='V'.
The complex analogue of this routine is F08HSF (ZHBTRD).

9  Example

This example computes all the eigenvalues and eigenvectors of the matrix A, where
A = 4.99 0.04 0.22 0.00 0.04 1.05 -0.79 1.04 0.22 -0.79 -2.31 -1.30 0.00 1.04 -1.30 -0.43 .
Here A is symmetric and is treated as a band matrix. The program first calls F08HEF (DSBTRD) to reduce A to tridiagonal form T, and to form the orthogonal matrix Q; the results are then passed to
F08JEF (DSTEQR) which computes the eigenvalues and eigenvectors of A.

9.1  Program Text

Program Text (f08hefe.f90)

9.2  Program Data

Program Data (f08hefe.d)

9.3  Program Results

Program Results (f08hefe.r)


F08HEF (DSBTRD) (PDF version)
F08 Chapter Contents
F08 Chapter Introduction
NAG Library Manual

© The Numerical Algorithms Group Ltd, Oxford, UK. 2011