3.4-final-1999-strang¶
Final Exam, 1999 (strang) · Section 3.4: Independence, Basis, and Dimension Source: finalexam_1.pdf, pp. 36, 41
Problem¶
7. In each part, find the required matrix or explain why such a matrix does not exist.
(a) (3 points) The matrices \(A\) and \(A^T\) and \(A + A^T\) have ranks 2 and 1 and 3.
(b) (3 points) The solution to \(Ax = 0\) is unique, but the solution to \(A^Tx = 0\) is not unique.
(c) (3 points) The powers \(A^k\) do not approach the zero matrix as \(k \to \infty\), but the exponential \(e^{At}\) does approach the zero matrix as \(t \to \infty\).
(d) (3 points) The complete solution to
(e) (3 points) The pivots are -1 and -2 but the eigenvalues are +1 and +2. (Symmetric matrix not required, row exchanges not required.)
Solution
(a) Cannot exist because \(A\) and \(A^T\) have the same rank. (b) \(A = \begin{bmatrix} 1 \\ 2 \end{bmatrix}\) or any non-square \(A\) with independent columns. (c) The desired \(A\) has an eigenvalue like -2, outside the unit circle and in the left half-plane. In fact, \(A = [-2]\) is a 1 by 1 example. (d) From the two given nullspace vectors we know that \(A = \begin{bmatrix} v & v & -v \end{bmatrix}\) for some column \(v\). The particular solution \((1, 1, 1)\) determines \(v\): \[ A \begin{bmatrix} 1 \\ 1 \\ 1 \end{bmatrix} = \begin{bmatrix} 1 \\ 1 \\ 2 \end{bmatrix} \quad \text{gives} \quad v + v - v = \begin{bmatrix} 1 \\ 1 \\ 2 \end{bmatrix} \quad \text{so} \quad v = \begin{bmatrix} 1 \\ 1 \\ 2 \end{bmatrix}. \] (e) The first pivot must be \(a_{11} = -1\). The trace \(1 + 2\) requires \(a_{22} = 4\). Then the determinant must be 2, so these matrices will work: \[ A = \begin{bmatrix} -1 & -1 \\ 6 & 4 \end{bmatrix} \quad \text{or any} \quad A = \begin{bmatrix} -1 & -a \\ 6/a & 4 \end{bmatrix} \]Keywords: rank, nullspace, eigenvalues, pivots, matrix construction