Skip to content

6.2-exam3-2006

Quiz 3, 2006 · Section 6.2: Diagonalizing a Matrix Source: exam3_1.pdf, pp. 102, 109-110

Problem

2 (33 pts.) (a) Find three eigenvalues of \(A\), and an eigenvector matrix \(S\):

\[ A = \begin{bmatrix} -1 & 2 & 4 \\ 0 & 0 & 5 \\ 0 & 0 & 1 \end{bmatrix} \]

(b) Explain why \(A^{1001} = A\). Is \(A^{1000} = I\)? Find the three diagonal entries of \(e^{At}\).

(c) The matrix \(A^{T}A\) (for the same \(A\)) is

\[ A^{T}A = \begin{bmatrix} 1 & -2 & -4 \\ -2 & 4 & 8 \\ -4 & 8 & 42 \end{bmatrix}. \]

How many eigenvalues of \(A^{T}A\) are positive? zero? negative? (Don't compute them but explain your answer.) Does \(A^{T}A\) have the same eigenvectors as \(A\)?

Solution (a) The eigenvalues are \(-1, 0, 1\) since \(A\) is triangular: \[ \lambda = -1 \text{ has } x = \begin{bmatrix} 1 \\ 0 \\ 0 \end{bmatrix}, \qquad \lambda = 0 \text{ has } x = \begin{bmatrix} 2 \\ 1 \\ 0 \end{bmatrix}, \qquad \lambda = 1 \text{ has } x = \begin{bmatrix} 7 \\ 5 \\ 1 \end{bmatrix}. \] Those vectors \(x\) are the columns of \(S\) (upper triangular!). (b) \(A = S\Lambda S^{-1}\) and \(A^{1001} = S\Lambda^{1001}S^{-1}\). Notice that \(\Lambda^{1001} = \Lambda\). Also \(A^{1000} \neq I\): \(A\) is singular, and \(0^{1000} = 0 \neq 1\). \(e^{At}\) has \(e^{-t}\), \(e^{0t} = 1\), \(e^t\) on its diagonal. *Proof using the series:* \(\sum_{0}^{\infty}(At)^n/n!\) has triangular matrices, so the diagonal has \(\sum (-t)^n/n! = e^{-t}\), \(\sum 0^n/n! = 1\), \(\sum t^n/n! = e^t\). *Proof using \(S\Lambda S^{-1}\):* \[ e^{At} = Se^{\Lambda t}S^{-1} = \begin{bmatrix} 1 & \times & \times \\ 0 & 1 & \times \\ 0 & 0 & 1 \end{bmatrix} \begin{bmatrix} e^{-t} & & \\ & 1 & \\ & & e^t \end{bmatrix} \begin{bmatrix} 1 & \times & \times \\ 0 & 1 & \times \\ 0 & 0 & 1 \end{bmatrix}. \] (c) \(A^{T}A\) has 2 positive eigenvalues: it has rank 2, and its eigenvalues can never be negative. One eigenvalue is zero because \(A^{T}A\) is singular. And 0 of them are negative. (Or: \(A^{T}A\) is symmetric, so the eigenvalues have the same signs as the pivots. Do elimination: the pivots are 1, 0, and \(42 - 16 = 26\).)

Keywords: eigenvalues, eigenvectors, triangular matrix, matrix powers, matrix exponential