Skip to content

6.2-final-2009

Final Exam, 2009 · Section 6.2: Diagonalizing a Matrix Source: finalexam_1.pdf, p. 223

Problem

Problem 1:

A sequence of numbers \(f_0, f_1, f_2, \dots\) is defined by the recurrence

\[ f_{k+2} = 3f_{k+1} - f_k, \]

with starting values \(f_0 = 1, f_1 = 1\). (Thus, the first few terms in the sequence are 1, 1, 2, 5, 13, 34, 89, . . ..)

(a) Defining \(u_k = \begin{bmatrix} f_{k+1} \\ f_k \end{bmatrix}\), re-express the above recurrence as \(u_{k+1} = Au_k\), and give the matrix \(A\).

(b) Find the eigenvalues of \(A\), and use these to predict what the ratio \(f_{k+1}/f_k\) of successive terms in the sequence will approach for large \(k\).

(c) The sequence above starts with \(f_0 = f_1 = 1\), and \(|f_k|\) grows rapidly with \(k\). Keep \(f_0 = 1\), but give a different value of \(f_1\) that will make the sequence (with the same recurrence \(f_{k+2} = 3f_{k+1} - f_k\)) approach zero (\(f_k \to 0\)) for large \(k\).

Solution (18 points = 6 + 6 + 6) (a) We have \[ \begin{bmatrix} f_{k+2} \\ f_{k+1} \end{bmatrix} = \begin{bmatrix} 3 & -1 \\ 1 & 0 \end{bmatrix} \begin{bmatrix} f_{k+1} \\ f_k \end{bmatrix} \implies A = \begin{bmatrix} 3 & -1 \\ 1 & 0 \end{bmatrix}. \] (b) Eigenvalues of \(A\) are roots of \(\det(A - \lambda I) = \lambda^2 - 3\lambda + 1 = 0\). They are \(\lambda_1 = \frac{3 + \sqrt{5}}{2}\) and \(\lambda_2 = \frac{3 - \sqrt{5}}{2}\). Note that \(\lambda_1 > \lambda_2\), so the ratio \(f_{k+1}/f_k\) will approach \(\lambda_1 = \frac{3 + \sqrt{5}}{2}\) for large \(k\). (c) Let \(v_1, v_2\) be the eigenvectors with eigenvalues \(\lambda_1\) and \(\lambda_2\), respectively. So, we can write \(u_0 = c_1v_1 + c_2v_2\) and then \(u_k = c_1\lambda_1^kv_1 + c_2\lambda_2^kv_2\). If we need \(f_k \to 0\), we have to make \(c_1 = 0\). In other words, \(u_0\) must be proportional to the eigenvector \(v_2\). \[ A - \lambda_2 I = \begin{bmatrix} \frac{3 + \sqrt{5}}{2} & -1 \\ 1 & -\frac{3 - \sqrt{5}}{2} \end{bmatrix} \Rightarrow v_2 = \begin{bmatrix} \frac{3 - \sqrt{5}}{2} \\ 1 \end{bmatrix}. \] Hence, we need to take \(f_1 = \frac{3 - \sqrt{5}}{2}\) so that \(f_k\) will approach zero for large \(k\).

Keywords: eigenvalues, eigenvectors, difference equations, matrix powers