22.09.2013 Views

SSL II USER'S GUIDE - Lahey Computer Systems

SSL II USER'S GUIDE - Lahey Computer Systems

SSL II USER'S GUIDE - Lahey Computer Systems

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

CHAPTER 4<br />

EIGENVALUES AND EIGENVECTORS<br />

4.1 OUTLINE<br />

Eigenvalue problems can be organized as show in Table<br />

4.1 according to the type of problem (Ax =λx, Ax =λBx)<br />

and the shape (dense, band), type (real, complex), and<br />

form (symmetric, unsymmetric) of the matrices. The<br />

reader should refer to the appropriate section specified<br />

in the table.<br />

Table 4.1 Organization of eigenvalue problem<br />

Shape of<br />

matrix<br />

Dense<br />

matrix<br />

Band<br />

matrix<br />

Type of<br />

problem<br />

Matrix type and form<br />

Ax =λx Real matrix<br />

Complex matrix<br />

Real symmetric matrix<br />

Hermitian matrix<br />

Explanation<br />

section<br />

4.2<br />

4.3<br />

4.4<br />

4.5<br />

Ax =λBx Real symmetric matrix 4.7<br />

Ax =λx Real symmetric band<br />

matrix<br />

4.6<br />

Ax =λBx Real symmetric band<br />

matrix<br />

4.8<br />

Note:<br />

Refer to the section on a real symmetric matrix concerning a<br />

real symmetric tridiagonal matrix.<br />

Table 4.2 Subroutines used for standard eigenproblem of a real matrix<br />

4.2 EIGENVALUES AND<br />

EIGENVECTORS OF A REAL MATRIX<br />

A standard sequenses of procedures are shown here<br />

when <strong>SSL</strong> <strong>II</strong> routines are used to solve the eigenvalue<br />

problems.<br />

<strong>SSL</strong> <strong>II</strong> provides the following:<br />

− Standard routines by which the entire procedures for<br />

obtaining eigenvalues and eigenvectors of real<br />

matrices may be performed at one time.<br />

− Component routines performing component functions.<br />

For details, see Table 4.2.<br />

User problems are classified as follows :<br />

• Obtaining all eigenvalues<br />

• Obtaining all eigenvalues and corresponding<br />

eigenvectors (or selected eigenvectors)<br />

In the first and second items that follow, the use of<br />

component routines and standard routines is explained<br />

by describing their procedures. Further comments on<br />

processing are in the third item.<br />

When obtaining eigenvalues and eigenvectors of a real<br />

matrix, the choice of calling the various component<br />

routines or calling the standard routine is up to the user.<br />

However, the standard routine, which is easier to use, is<br />

recommended to be called.<br />

Level Function Subroutine name<br />

Standard routines Eigenvalues and eigenvectors of a real matrix<br />

EIG1<br />

(B21-11-0101)<br />

Balancing of a real matrix<br />

BLNC<br />

(B21-11-0202)<br />

Reduction of a real matrix to a Hessenberg matrix<br />

HES1<br />

(B21-11-0302)<br />

Component<br />

Obtaining the eigenvalues of a real Hessenberg<br />

matrix<br />

HSQR<br />

(B21-11-0402)<br />

routines<br />

Obtaining the eigenvectors of a real Hessenberg<br />

matrix<br />

HVEC<br />

(B21-11-0502)<br />

Back transformation to and normalization of the<br />

HBK1<br />

eigenvectors of a real matrix<br />

(B21-11-0602)<br />

Normalization of the eigenvectors of a real matrix<br />

NRML<br />

(B21-11-0702)<br />

35

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!