Skip to content

6.3-exam3-2011

Quiz 3, 2011 · Section 6.3: Symmetric Positive Definite Matrices Source: exam3_2.pdf, pp. 26-27

Problem

4. (30 pts.) Consider the matrix \(A = \begin{bmatrix} x & 1 & 1 \\ 1 & 1 & 1 \\ 1 & 1 & 1 \end{bmatrix}\) with parameter \(x\) in the (1,1) position.

  1. (10 pts) Specify all numbers \(x\), if any, for which \(A\) is positive definite. (Explain briefly.)

  2. (10 pts) Specify all numbers \(x\), if any, for which \(e^A\) is positive definite. (Explain briefly.)

  3. (10 pts) Find an \(x\), if any, for which \(4I - A\) is positive definite. (Explain briefly.)

Solution 1. No such \(x\): the matrix is clearly singular, with two equal rows and two equal columns. 2. Every \(x\) works. The eigenvalues of \(e^A\) are the exponentials of the eigenvalues of \(A\). Since \(A\) is symmetric the eigenvalues are real, and thus the exponentials are positive. A symmetric matrix with positive eigenvalues is positive definite. 3. One can take any \(x < 3\). The easiest choice is \(x = 1\): with this guess the matrix \(A\) has two eigenvalues 0 and one eigenvalue 3 (so \(4I - A\) has eigenvalues 4, 4, 1), all less than 4, so \(4 - \lambda > 0\) for all three eigenvalues of \(A\). Systematically, one can consider the three upper left determinants of \(4I - A\), which are \(4 - x\), \(11 - 3x\), and \(24 - 8x\). They are all positive if and only if \(x < 3\).

Keywords: positive definite, positive semidefinite, matrix exponential, eigenvalues, determinant