Skip to content

6.5-final-2008

Final Exam, 2008 · Section 6.5: Solving Linear Differential Equations Source: finalexam_1.pdf, p. 189

Problem

2 (12 pts.) The matrix

\[ A = \begin{bmatrix} 0 & 1 & 0 \\ 0 & 0 & 1 \\ 0 & 0 & p \end{bmatrix}. \]

(a) What are the eigenvalues of \(A\) (possibly in terms of \(p\))?

(b) If \(p\) is not 0, find an eigenvector that is not in the nullspace.

(c) What are the singular values of \(A\) (possibly in terms of \(p\))?

(d) Find a nonzero solution \(u(t)\) to \(du/dt = (A + 2009I)u\). Check that your answer is correct. (Note that \(A + 2009I\) is the matrix above with the upcoming new year added to the diagonal elements.)

Solution (a) To find the eigenvalues of \(A\), we compute \[ \det(A - \lambda I) = \det \begin{bmatrix} -\lambda & 1 & 0 \\ 0 & -\lambda & 1 \\ 0 & 0 & p - \lambda \end{bmatrix} = \lambda^2(p - \lambda) \] (because \(A\) is upper triangular). So the eigenvalues are 0 and \(p\). (b) If \(p \neq 0\), we wish to find \(\begin{bmatrix} a \\ b \\ c \end{bmatrix}\) so that \(A \begin{bmatrix} a \\ b \\ c \end{bmatrix} = p \begin{bmatrix} a \\ b \\ c \end{bmatrix}\). But \[ A \begin{bmatrix} a \\ b \\ c \end{bmatrix} = \begin{bmatrix} b \\ c \\ pc \end{bmatrix}, \] so we need \(b = pa\) and \(c = pb\); and so \(\begin{bmatrix} a \\ b \\ c \end{bmatrix} = \begin{bmatrix} 1 \\ p \\ p^2 \end{bmatrix}\) works. (c) The singular values of \(A\) are found by first computing the eigenvalues of \[ A^TA = \begin{bmatrix} 0 & 0 & 0 \\ 1 & 0 & 0 \\ 0 & 1 & p \end{bmatrix} \begin{bmatrix} 0 & 1 & 0 \\ 0 & 0 & 1 \\ 0 & 0 & p \end{bmatrix} = \begin{bmatrix} 0 & 0 & 0 \\ 0 & 1 & 0 \\ 0 & 0 & 1 + p^2 \end{bmatrix}. \] As this is upper triangular, the eigenvalues are 0, 1 and \(1 + p^2\). So the singular values are 0, 1 and \(\sqrt{1 + p^2}\). (d) In general, \(du/dt = Bu\) is solved by \(e^{Bt}u(0)\). For us, \[ B = \begin{bmatrix} 2009 & 1 & 0 \\ 0 & 2009 & 1 \\ 0 & 0 & 2009 + p \end{bmatrix} = A + 2009I. \] To compute \(e^{Bt}\), we note that \[ e^{Bt} = \exp\left(t \begin{bmatrix} 2009 & 0 & 0 \\ 0 & 2009 & 0 \\ 0 & 0 & 2009 + p \end{bmatrix} + t \begin{bmatrix} 0 & 1 & 0 \\ 0 & 0 & 1 \\ 0 & 0 & 0 \end{bmatrix}\right) \] and the two matrices commute, so this is \[ \begin{bmatrix} e^{2009t} & 0 & 0 \\ 0 & e^{2009t} & 0 \\ 0 & 0 & e^{(2009+p)t} \end{bmatrix} \exp\left(t \begin{bmatrix} 0 & 1 & 0 \\ 0 & 0 & 1 \\ 0 & 0 & 0 \end{bmatrix}\right). \] To compute \(\exp\left(t \begin{bmatrix} 0 & 1 & 0 \\ 0 & 0 & 1 \\ 0 & 0 & 0 \end{bmatrix}\right)\), we note that \[ \begin{bmatrix} 0 & 1 & 0 \\ 0 & 0 & 1 \\ 0 & 0 & 0 \end{bmatrix}^2 = \begin{bmatrix} 0 & 0 & 1 \\ 0 & 0 & 0 \\ 0 & 0 & 0 \end{bmatrix} \quad \text{and} \quad \begin{bmatrix} 0 & 1 & 0 \\ 0 & 0 & 1 \\ 0 & 0 & 0 \end{bmatrix}^3 = 0. \] So \[ \exp\left(t \begin{bmatrix} 0 & 1 & 0 \\ 0 & 0 & 1 \\ 0 & 0 & 0 \end{bmatrix}\right) = I + t \begin{bmatrix} 0 & 1 & 0 \\ 0 & 0 & 1 \\ 0 & 0 & 0 \end{bmatrix} + \frac{t^2}{2} \begin{bmatrix} 0 & 0 & 1 \\ 0 & 0 & 0 \\ 0 & 0 & 0 \end{bmatrix} = \begin{bmatrix} 1 & t & t^2/2 \\ 0 & 1 & t \\ 0 & 0 & 1 \end{bmatrix}, \] and finally \[ e^{Bt} = \begin{bmatrix} e^{2009t} & te^{2009t} & e^{2009t}t^2/2 \\ 0 & e^{2009t} & te^{2009t} \\ 0 & 0 & e^{(2009+p)t} \end{bmatrix}. \] So our answer is \[ u(t) = e^{Bt} \begin{bmatrix} 1 \\ 0 \\ 0 \end{bmatrix} = \begin{bmatrix} e^{2009t} \\ 0 \\ 0 \end{bmatrix}. \] Check: \(\frac{du}{dt} = 2009e^{2009t}(1, 0, 0)^T = (A + 2009I)u(t)\). (The printed solutions omit the factor \(t\) in the exponents; the correct entries are \(e^{2009t}\), \(e^{(2009+p)t}\).)

Keywords: eigenvalues, singular values, matrix exponential, differential equations