6.5-exam3-2015¶
Exam III, 2015 · Section 6.5: Solving Linear Differential Equations Source: exam3_2.pdf, pp. 103, 106
Problem¶
2. (33 points)
(a) Show that \(\lambda_1 = 0\) is an eigenvalue of \(A\) and find an eigenvector \(x_1\) with that zero eigenvalue:
(b) Find the other eigenvalues \(\lambda_2\) and \(\lambda_3\) of this symmetric matrix. Does \(A\) have two more independent eigenvectors \(x_2\) and \(x_3\)? Give a reason why or why not. (Not required to find \(x_2\) and \(x_3\).)
(c) Suppose \(\frac{du}{dt} = Au\) starts from \(u(0) = \begin{bmatrix} 1 \\ 2 \\ 3 \end{bmatrix}\). Explain why this \(u(t)\) approaches a steady state \(u(\infty)\) as \(t \to \infty\). You can use the general formula \(u(t) = c_1e^{\lambda_1 t}x_1 + c_2e^{\lambda_2 t}x_2 + c_3e^{\lambda_3 t}x_3\) or \(e^{At} = Se^{\Lambda t}S^{-1}\) without putting in all eigenvectors. Find that steady state \(u(\infty)\).
Solution
(a) \(A(1, 1, 1)^T = 0\), and so \(x_1 = (1, 1, 1)^T\) is an eigenvector with eigenvalue \(\lambda_1 = 0\). (b) Each of the columns of \(A + 3I\) is \((1, 1, 1)^T\), and so it has rank 1. In particular, the nullspace of \(A + 3I\) has dimension 2, and so the other eigenvalues are \(\lambda_2 = -3 = \lambda_3\). Since the nullspace of \(A + 3I\) has dimension 2, \(A\) does have two more independent eigenvectors \(x_2\) and \(x_3\). (c) \[ u(t) = c_1e^{0t}x_1 + c_2e^{-3t}x_2 + c_3e^{-3t}x_3 \to u(\infty) = c_1x_1 \quad \text{as } t \to \infty. \] We just need to find \(c_1\). But \(u(0) = c_1x_1 + c_2x_2 + c_3x_3 = (1, 2, 3)^T\). Since \(x_1\) is orthogonal to \(x_2\) and \(x_3\), we see, by dotting with \(x_1\), that \(c_1x_1 \cdot x_1 = (1, 2, 3)^T \cdot x_1\). Remembering that \(x_1 = (1, 1, 1)^T\), we obtain \(3c_1 = 6\), so that \(c_1 = 2\), and \[ u(\infty) = (2, 2, 2)^T. \]Keywords: eigenvalues, symmetric matrix, differential equation, steady state, nullspace