Skip to content

6.2-exam2-2014-b

Exam 2, 2014 · Section 6.2: Diagonalizing a Matrix Source: exam2_2.pdf, pp. 76, 82

Problem

Problem 4. Consider the Markov matrix \(A = \begin{bmatrix} 0 & 1/3 & 1/3 & 0 \\ 1/2 & 0 & 1/3 & 1/2 \\ 1/2 & 1/3 & 0 & 1/2 \\ 0 & 1/3 & 1/3 & 0 \end{bmatrix}\).

(a) Three of the eigenvalues of \(A\) are \(1, 0, -1/3\). Find the fourth eigenvalue of \(A\).

(b) Find the determinant \(\det(A)\).

(c) Find the eigenvector of the transposed matrix \(A^T\) with the eigenvalue \(\lambda_1 = 1\).

(d) Find the eigenvector of the matrix \(A\) with the eigenvalue \(\lambda_1 = 1\). (Hint: Notice that nonzero entries in each column of \(A\) are the same.)

(e) Find the limit of \(A^k (1 \ 0 \ 0 \ 0)^T\) as \(k \to +\infty\).

Solution (a) Since \(\mathrm{tr}(A) = 0\), the sum of the eigenvalues is 0. Thus, the fourth eigenvalue must be \(-2/3\). (b) The determinant is the product of the eigenvalues, which is 0. (c) \((1, 1, 1, 1)A = (1, 1, 1, 1)\), and so the eigenvector of \(A^T\) with eigenvalue \(\lambda_1 = 1\) is \((1, 1, 1, 1)^T\). (d) The Markov matrix \(A\) corresponds to a random walk on the graph with four nodes 1, 2, 3, 4 connected by the edges (1, 2), (1, 3), (2, 3), (2, 4), (3, 4). The degrees of the nodes are 2, 3, 3, 2. Thus the vector \((2, 3, 3, 2)^T\) is an eigenvector with eigenvalue \(\lambda_1 = 1\). (e) Let \(v_1 = (2, 3, 3, 2)^T\) and let \(v_2, v_3\) and \(v_4\) be eigenvectors for \(0, -1/3, -2/3\), respectively. Then there exist \(c_1, \dots, c_4 \in R\) with \[ (1, 0, 0, 0)^T = c_1v_1 + c_2v_2 + c_3v_3 + c_4v_4. \] Thus \[ A^k(1, 0, 0, 0)^T = c_1v_1 + \frac{(-1)^kc_3}{3^k}v_3 + \frac{(-2)^kc_4}{3^k}v_4 \longrightarrow c_1v_1 \quad \text{as} \ k \longrightarrow +\infty. \] To find \(c_1\) we recall that \((1, 1, 1, 1)A = (1, 1, 1, 1)\). By induction we obtain \[ (1, 1, 1, 1)A^k = (1, 1, 1, 1), \] and so \((1, 1, 1, 1)A^k(1, 0, 0, 0)^T = (1, 1, 1, 1)(1, 0, 0, 0)^T = 1\). Letting \(k \longrightarrow +\infty\) we obtain \[ (1, 1, 1, 1)c_1v_1 = 1, \] so that \(c_1 = 1/((1, 1, 1, 1)v_1) = 1/10\). The answer to the question is \((2, 3, 3, 2)^T/10\).

Keywords: Markov matrix, eigenvalues, eigenvectors, matrix powers, steady state