Skip to content

6.3-final-2013-strang

Final Exam, 2013 (strang) · Section 6.3: Symmetric Positive Definite Matrices Source: finalexam_2.pdf, p. 114

Problem

4 (3+4+3 pts.) The matrix \(A\) is symmetric and also orthogonal.

(a) How is \(A^{-1}\) related to \(A\)?

(b) What number(s) can be eigenvalues of \(A\) and why?

(c) Here is an example of \(A\). What are the eigenvalues of this matrix? I don't recommend computing with \(\det(A - \lambda I)\)! Find a way to use part (b).

\[ A = \begin{bmatrix} .5 & -.5 & -.1 & -.7 \\ -.5 & .5 & -.1 & -.7 \\ -.1 & -.1 & .98 & -.14 \\ -.7 & -.7 & -.14 & .02 \end{bmatrix}. \]
Solution (a) Since \(A\) is orthogonal, \(A^{-1} = A^T\). Since \(A\) is symmetric, \(A^T = A\). Therefore \[ A^{-1} = A^T = A. \] (b) The only possible eigenvalues are \[ \lambda = \pm 1. \] The eigenvalues are real because \(A\) is symmetric, and they satisfy \(|\lambda| = 1\) because \(A\) is orthogonal: if \(Ax = \lambda x\) then \(\|x\| = \|A^T Ax\| = \|A^2 x\| = |\lambda|^2 \|x\|\), or directly \(x^T x = x^T A^T A x = \lambda^2 x^T x\). (c) The trace is \(\mathrm{trace}(A) = .5 + .5 + .98 + .02 = 2\). By part (b) the eigenvalues are four numbers \(\pm 1\) whose sum is 2, so they must be \[ \lambda = 1, 1, 1, -1. \]

Keywords: symmetric matrix, orthogonal matrix, eigenvalues, trace