Skip to content

6.2-exam3-2002

Quiz 3, 2002 · Section 6.2: Diagonalizing a Matrix Source: exam3_1.pdf, pp. 66, 74

Problem

1 (36 pts.) (a) What are the eigenvalues of the 5 by 5 matrix \(A = ones(5)\) with all entries \(a_{ij} = 1\)? Please look at \(A\), not at \(\det(A - \lambda I)\).

(b) Solve this differential equation to find \(u(t)\):

\[ \frac{du}{dt} = Au \quad \text{starting from} \quad u(0) = (0, 1, 1, 1, 2). \]

First split \(u(0)\) into two eigenvectors of \(A\).

(c) Using part (a), what are the eigenvalues and trace and determinant of the matrix \(B =\) same as \(A\) except zeros on the diagonal.

Solution (a) One eigenvalue of \(A = ones(5)\) is \(\lambda_1 = 5\), corresponding to the eigenvector \(x_1 = (1, 1, 1, 1, 1)\). Since the rank of \(A\) is 1, all the other eigenvalues \(\lambda_2, \dots, \lambda_5\) are zero. Check: the trace of \(A\) is 5. (b) The initial condition \(u(0)\) can be written as a sum of the two eigenvectors \(x_1 = (1, 1, 1, 1, 1)\) and \(x_2 = (-1, 0, 0, 0, 1)\), corresponding to the eigenvalues \(\lambda_1 = 5\) and \(\lambda_2 = 0\): \[ u(0) = (0, 1, 1, 1, 2) = (1, 1, 1, 1, 1) + (-1, 0, 0, 0, 1) = x_1 + x_2. \] The solution to \(\frac{du}{dt} = Au\) is then \[ u(t) = c_1 e^{\lambda_1 t} x_1 + c_2 e^{\lambda_2 t} x_2 = (1, 1, 1, 1, 1)e^{5t} + (-1, 0, 0, 0, 1). \] (c) The eigenvectors of \(B = A - I\) are the same as for \(A\), and the eigenvalues are smaller by 1: \[ Bx = (A - I)x = Ax - x = \lambda x - x = (\lambda - 1)x, \] where \(x, \lambda\) are an eigenvector and an eigenvalue of \(A\). The eigenvalues of \(B\) are then \(4, -1, -1, -1, -1\), the trace is \(\sum_i \lambda_i = 0\), and the determinant is \(\prod_i \lambda_i = 4\).

Keywords: eigenvalues, rank one matrix, differential equation, trace