Skip to content

6.2-final-2001

Final Exam, 2001 · Section 6.2: Diagonalizing a Matrix Source: finalexam_1.pdf, p. 55

Problem

3 (36 pts.) (a) Let

\[ A = \begin{bmatrix} 4 & 2 \\ 0 & 4 \end{bmatrix}. \]

Find all diagonal matrices \(D\) for which \(DAD^{-1}\) is a Jordan matrix.

(b) Find all pairs \(a, b \in R\) such that

\[ A = \begin{bmatrix} 1/\sqrt{5} & 2/\sqrt{5} \\ a & b \end{bmatrix} \]

is an orthogonal matrix.

(c) Find all orthogonal upper-triangular matrices of size 3-by-3.

Solution (a) Write \(D = \begin{bmatrix} d_1 & 0 \\ 0 & d_2 \end{bmatrix}\) with \(d_1d_2 \neq 0\). Then \[ DAD^{-1} = \begin{bmatrix} 4 & 2d_1/d_2 \\ 0 & 4 \end{bmatrix}. \] This is a single 2-by-2 Jordan block with eigenvalue 4 whenever the superdiagonal entry is nonzero, i.e. \(2d_1/d_2 \neq 0\), which holds for every invertible diagonal matrix. (Under the strict convention that the superdiagonal of a Jordan block is exactly 1, the condition is \(2d_1/d_2 = 1\), i.e. \(D = t\begin{bmatrix} 1 & 0 \\ 0 & 2 \end{bmatrix}\) for \(t \neq 0\).) (b) The columns must be orthonormal: \(1/5 + a^2 = 1\) gives \(a = \pm 2/\sqrt{5}\), and orthogonality of the columns gives \((1/\sqrt{5})(2/\sqrt{5}) + ab = 0\), so \(ab = -2/5\). Thus the two solutions are \[ (a, b) = \left(\frac{2}{\sqrt{5}}, -\frac{1}{\sqrt{5}}\right) \quad \text{or} \quad \left(-\frac{2}{\sqrt{5}}, \frac{1}{\sqrt{5}}\right). \] (c) Let \(R = \begin{bmatrix} r_{11} & r_{12} & r_{13} \\ 0 & r_{22} & r_{23} \\ 0 & 0 & r_{33} \end{bmatrix}\) be orthogonal. The first column \((r_{11}, 0, 0)\) has length 1, so \(r_{11} = \pm 1\). Orthogonality of the columns then forces \(r_{12} = r_{13} = 0\), and the second column has length 1, so \(r_{22} = \pm 1\); orthogonality forces \(r_{23} = 0\), and the third column gives \(r_{33} = \pm 1\). Hence \[ R = \begin{bmatrix} \pm 1 & 0 & 0 \\ 0 & \pm 1 & 0 \\ 0 & 0 & \pm 1 \end{bmatrix}, \] i.e. the orthogonal upper-triangular matrices are exactly the diagonal matrices with diagonal entries \(\pm 1\) (all 8 of them).

Keywords: Jordan form, orthogonal matrix, similarity, upper triangular