24.04.2013 Views

Nonlinear Control Sy.. - Free

Nonlinear Control Sy.. - Free

Nonlinear Control Sy.. - Free

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.

2.4. MATRICES 41<br />

Proof: By definition, the columns of the matrix S are the eigenvectors of A. Thus, we have<br />

AS=A :l vn<br />

and we can re write the last matrix in the following form:<br />

Alvl ... Anon<br />

Al<br />

Alvl ... Anvn = V1 ... vn ...<br />

Therefore, we have that AS = SD. The columns of the matrix S are, by assumption,<br />

linearly independent. It follows that S is invertible and we can write<br />

or also<br />

A = SDS-l<br />

D = S-IAS.<br />

This completes the proof of the theorem. El<br />

Special Case: <strong>Sy</strong>mmetric Matrices<br />

<strong>Sy</strong>mmetric matrices have several important properties. Here we mention two of them,<br />

without proof:<br />

(i) The eigenvalues of a symmetric matrix A E IIF"' are all real.<br />

(ii) Every symmetric matrix A is diagonalizable. Moreover, if A is symmetric, then the<br />

diagonalizing matrix S can be chosen to be an orthogonal matrix P; that is, if A = AT,<br />

then there exist a matrix P satisfying PT = P-l such that<br />

2.4.2 Quadratic Forms<br />

P-'AP =PTAP=D.<br />

Given a matrix A E R""n, a function q : lR - IR of the form<br />

q(x) = xT Ax, x E R"<br />

is called a quadratic form. The matrix A in this definition can be any real matrix. There is,<br />

however, no loss of generality in restricting this matrix to be symmetric. To see this, notice<br />

An

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

Saved successfully!

Ooh no, something went wrong!