6.2-final-2001-spring-conflict¶
Final Exam (Conflict), 2001 (spring-conflict) · Section 6.2: Diagonalizing a Matrix Source: finalexam_1.pdf, p. 73
Problem¶
3. (15pts.) The following are some quick questions. Give only brief reasoning for your answers, no detailed proofs.
(a) Let \(A, B, C\) and \(D\) be four 3 by 3 matrices. Let \(E\) be the 6 by 6 matrix
Is it necessarily true that \(\det(E) = \det(A) \cdot \det C - \det B \cdot \det D\)?
(b) Let \(A\) be a 3 by 4 matrix, and \(B\) be a 4 by 3 matrix. Can you say anything about the determinant of their product, \(BA\)? How about \(AB\)?
(c) Do similar matrices have the same
i. eigenvalues;
ii. eigenvectors;
iii. rank;
iv. column space;
v. determinant?
(d) Does an \(n \times n\) matrix with \(n\) distinct eigenvalues have an orthogonal set of eigenvectors?
(e) Is the product of two symmetric matrices symmetric?
Solution
(a) No. For \(A = B = C = I\), \(D = 2I\), the formula gives \(1 \cdot 1 - 1 \cdot 4 = -3\), but by the block determinant formula \(\det E = \det A \cdot \det(D - CA^{-1}B) = 1 \cdot \det(2I - I) = 1\). The stated formula only holds for special blocks (e.g. triangular blocks). (b) \(BA\) is 4 by 4 of rank at most 3, so \(\det(BA) = 0\) always. \(AB\) is 3 by 3 and may be invertible: for example \(A = \begin{bmatrix} I_3 & 0 \end{bmatrix}\), \(B = \begin{bmatrix} I_3 \\ 0 \end{bmatrix}\) gives \(AB = I_3\), so \(\det(AB) = 1\). Nothing general can be said about \(\det(AB)\). (c) i. eigenvalues: yes (similar matrices have the same characteristic polynomial). ii. eigenvectors: no, in general they change (e.g. \(A\) and \(S^{-1}AS\) have eigenvectors \(S^{-1}x\)). iii. rank: yes (similarity by invertible \(S\) preserves rank). iv. column space: no, in general it changes. v. determinant: yes. (d) No. Distinct eigenvalues give linearly independent eigenvectors, but not orthogonal ones in general (orthogonality holds when the matrix is symmetric, or more generally normal). For example, \(\begin{bmatrix} 1 & 1 \\ 0 & 2 \end{bmatrix}\) has distinct eigenvalues 1, 2 but eigenvectors \((1, 0)\), \((1, 1)\), which are not orthogonal. (e) No in general. \((AB)^T = B^TA^T = BA\), which equals \(AB\) only when \(A\) and \(B\) commute. For example \(A = \begin{bmatrix} 1 & 1 \\ 1 & 1 \end{bmatrix}\), \(B = \begin{bmatrix} 1 & 0 \\ 0 & 0 \end{bmatrix}\) are symmetric but \(AB = \begin{bmatrix} 1 & 0 \\ 1 & 0 \end{bmatrix}\) is not.Keywords: determinant, similar matrices, eigenvectors, symmetric matrices