6.1-exam3-1998-b¶
Quiz 3, 1998 · Section 6.1: Introduction to Eigenvalues : Ax = λx Source: exam3_1.pdf, pp. 16, 21
Problem¶
2. (20 pts) Let \(A\) have eigenvalues \(\lambda_1, \dots, \lambda_n\) (all nonzero) and corresponding eigenvectors \(x_1, \dots, x_n\) forming a basis for \(R^n\). Let \(C\) be its cofactor matrix. (The answers to the questions below should be in terms of the \(\lambda_i\).)
(a) (5 pts) What is \(trace(A^{-1})\)? \(det(A^{-1})\)?
(b) (15 pts) What is \(trace(C)\)? What is \(det(C)\)? (Hint: \(A^{-1} = \frac{C^T}{\det A}\))
Solution
(a) \(A^{-1}\) has eigenvalues \(\frac{1}{\lambda_1}, \dots, \frac{1}{\lambda_n}\). Therefore \[ trace(A^{-1}) = \frac{1}{\lambda_1} + \dots + \frac{1}{\lambda_n}, \qquad det(A^{-1}) = \frac{1}{\lambda_1} \frac{1}{\lambda_2} \dots \frac{1}{\lambda_n}. \] (b) The eigenvalues of \(C^T\) are the same as those of \(C\), which are \(\det(A)\) times those of \(A^{-1}\): \[ \mu_i = \frac{\lambda_1 \lambda_2 \dots \lambda_n}{\lambda_i}. \] Hence \[ trace(C) = \lambda_1 \lambda_2 \dots \lambda_n \left( \frac{1}{\lambda_1} + \dots + \frac{1}{\lambda_n} \right), \qquad det(C) = (\lambda_1 \dots \lambda_n)^{n-1}. \]Keywords: eigenvalues, trace, determinant, cofactor matrix