Skip to content

6.2-exam3-2009-spring

Quiz 3, 2009 (spring) · Section 6.2: Diagonalizing a Matrix Source: exam3_1.pdf, p. 160

Problem

Problem 2: You are given the matrix

\[ A = \begin{bmatrix} 0.5 & 0.2 & 0.2 \\ 0.1 & 0.5 & 0.5 \\ 0.4 & 0.3 & 0.3 \end{bmatrix}. \]

(i) What are the eigenvalues of \(A\)? [Hint: Very little calculation required! You should be able to see two eigenvalues by inspection of the form of \(A\), and the third by an easy calculation. You shouldn't need to compute \(\det(A - \lambda I)\) unless you really want to do it the hard way.]

(ii) The vector \(u(t)\) solves the system

\[ \frac{du}{dt} = Au \]

for some initial condition \(u(0)\). If you are told that \(u(t)\) approaches some constant vector as \(t \to \infty\), give as much true information as possible regarding the initial condition \(u(0)\).

[Note: be sure you understand that this is not the same thing as solving the recurrence \(u_{k+1} = Au_k\)! Imagine how you would find \(u(t)\) if you knew what \(u(0)\) was.]

Solution (i) First, the last two columns of \(A\) are the same. Hence \(A\) is singular and it must have an eigenvalue \(\lambda_1 = 0\). Also, we observe that \(A\) is a Markov matrix. This means that \(\lambda_2 = 1\) is an eigenvalue of \(A\). Finally, we know the trace of \(A\) is the sum of its three eigenvalues. So, \(Tr(A) = 0.5 + 0.5 + 0.3 = 1.3\) and the last eigenvalue is \(\lambda_3 = 1.3 - 1 - 0 = 0.3\). (ii) We can write \(u(0) = c_1v_1 + c_2v_2 + c_3v_3\) using three eigenvectors \(v_1, v_2, v_3\), which correspond to \(\lambda_1 = 0, \lambda_2 = 1\), and \(\lambda_3 = 0.3\), respectively. We know that this system has solution \(u(t) = c_1v_1 + c_2e^tv_2 + c_3e^{0.3t}v_3\). So, if either one of \(c_2\) and \(c_3\) is nonzero, the system would blow up as \(t \to \infty\). Therefore, the only possibility for \(u(t)\) to approach some constant is to have \(c_2 = c_3 = 0\), that is to say that \(u(0)\) is a multiple of the eigenvector \(v_1 = (0, -1, 1)^T\). In this case, \(u(t) = u(0) = c_1v_1\) is a constant.

Keywords: Markov matrix, eigenvalues, trace, differential equation, eigenvectors