Skip to content

6.3-final-2000

Final Exam, 2000 · Section 6.3: Symmetric Positive Definite Matrices Source: finalexam_1.pdf, p. 45

Problem

2. (a) (4 points) Suppose \(A\) is a symmetric matrix. If you first subtract 3 times row 1 from row 3, and after that you subtract 3 times column 1 from column 3, is the resulting matrix \(B\) still symmetric? Yes or not necessarily, with a reason.

(b) (4 points) Create a symmetric positive definite matrix (but not diagonal) with eigenvalues 1, 2, 4.

(c) (4 points) Create a nonsymmetric matrix (if possible) with those eigenvalues. Create a rank-one matrix (if possible) with those eigenvalues.

Solution (a) Yes, \(B\) is still symmetric. The row operation is \(B_1 = EA\) with \(E = I - 3e_3e_1^T\) and the column operation is \(B = B_1E^T\), so \(B = EAE^T\). Then \(B^T = (EAE^T)^T = E A^T E^T = EAE^T = B\), since \(A^T = A\). (Equivalently: the symmetric entry \((i, 3) = (3, i)\) of \(A\) is changed in the same way by the row and the column operation.) (b) Take \(A = \begin{bmatrix} 3 & 1 & 0 \\ 1 & 3 & 0 \\ 0 & 0 & 1 \end{bmatrix}\): the 2 by 2 block \(\begin{bmatrix} 3 & 1 \\ 1 & 3 \end{bmatrix}\) has eigenvalues \(3 \pm 1 = 4, 2\), and the remaining eigenvalue is 1. \(A\) is symmetric, positive definite (all eigenvalues positive), and not diagonal. (c) A nonsymmetric matrix: \(A = \begin{bmatrix} 4 & 1 & 0 \\ 0 & 2 & 1 \\ 0 & 0 & 1 \end{bmatrix}\) is upper triangular, so its eigenvalues are the diagonal entries 4, 2, 1. A rank-one matrix is impossible: a rank-one matrix has nullspace of dimension 2, hence eigenvalues \(0, 0, \text{trace}\), i.e., at most one nonzero eigenvalue. It cannot have three nonzero eigenvalues 1, 2, 4.

Keywords: symmetric matrix, positive definite, eigenvalues, rank one matrix