6.2-final-2013-strang¶
Final Exam, 2013 (strang) · Section 6.2: Diagonalizing a Matrix Source: finalexam_2.pdf, p. 120
Problem¶
7 (3+3+3+3 pts.) The equation \(y_{n+2} + By_{n+1} + Cy_n = 0\) has the solution \(y_n = \lambda^n\) if \(\lambda^2 + B\lambda + C = 0\). In most cases this will give two roots \(\lambda_1\), \(\lambda_2\) and the complete solution is \(y_n = c_1 \lambda_1^n + c_2 \lambda_2^n\).
Now solve the same problem the matrix way (slower). Create this vector unknown and vector equation \(u_{n+1} = A u_n\):
(a) What is the matrix \(A\) in that equation?
(b) What equation gives the eigenvalues \(\lambda_1\) and \(\lambda_2\) of \(A\)?
(c) If \(\lambda_1\) is an eigenvalue, show directly that
(d) If \(\lambda_1 \neq \lambda_2\), what is now the complete solution \(u_n\) (including constants \(c_1\) and \(c_2\)) to our equation \(u_{n+1} = A u_n\)? (Then \(y_n\) is the first component of \(u_n\).)
Solution
(a) The recurrence says \(y_{n+2} = -By_{n+1} - Cy_n\), so \[ \begin{bmatrix} y_{n+1} \\ y_{n+2} \end{bmatrix} = \begin{bmatrix} 0 & 1 \\ -C & -B \end{bmatrix} \begin{bmatrix} y_n \\ y_{n+1} \end{bmatrix}, \qquad A = \begin{bmatrix} 0 & 1 \\ -C & -B \end{bmatrix}. \] (b) The eigenvalues satisfy \(\det(A - \lambda I) = 0\): \[ \det \begin{bmatrix} -\lambda & 1 \\ -C & -B - \lambda \end{bmatrix} = \lambda^2 + B\lambda + C = 0, \] the same quadratic as for the scalar equation. (c) Using the equation from (b): \[ A \begin{bmatrix} 1 \\ \lambda_1 \end{bmatrix} = \begin{bmatrix} 0 & 1 \\ -C & -B \end{bmatrix} \begin{bmatrix} 1 \\ \lambda_1 \end{bmatrix} = \begin{bmatrix} \lambda_1 \\ -C - B\lambda_1 \end{bmatrix} = \begin{bmatrix} \lambda_1 \\ \lambda_1^2 \end{bmatrix} = \lambda_1 \begin{bmatrix} 1 \\ \lambda_1 \end{bmatrix}, \] since \(\lambda_1^2 + B\lambda_1 + C = 0\). (d) If \(\lambda_1 \neq \lambda_2\), the eigenvectors \(\begin{bmatrix} 1 \\ \lambda_1 \end{bmatrix}\) and \(\begin{bmatrix} 1 \\ \lambda_2 \end{bmatrix}\) are independent, so the complete solution is \[ u_n = c_1 \lambda_1^n \begin{bmatrix} 1 \\ \lambda_1 \end{bmatrix} + c_2 \lambda_2^n \begin{bmatrix} 1 \\ \lambda_2 \end{bmatrix}, \] and taking first components, \[ y_n = c_1 \lambda_1^n + c_2 \lambda_2^n. \]Keywords: recurrence equation, eigenvalues, eigenvectors, companion matrix