6.3-final-2003¶
Final Exam, 2003 · Section 6.3: Symmetric Positive Definite Matrices Source: finalexam_1.pdf, p. 104
Problem¶
6. True or False. Circle the appropriate answer. "True" means "always true", and "false" means "sometimes false". Justify each answer briefly.
(a) The product of the pivots when performing Gauss-Jordan is equal to the determinant if we do not have to permute rows. True or False.
(b) Let \(A\) be an \(m \times n\) matrix whose columns are independent. Then \(AA^T\) is positive definite. True or False.
(c) If \(A\) is a symmetric matrix then the singular values are the absolute values of the nonzero eigenvalues. True or False.
(d) There exists a \(5 \times 5\) unitary matrix with eigenvalues \(1, 1 + i, 1 - i, i\) and \(-i\). True or False.
(e) Suppose \(V\) and \(W\) are two vector spaces of dimension \(n\). If \(T\) is a linear transformation from \(V\) to \(W\) with only the 0 vector in the kernel, then for any basis of \(V\), there exists an orthonormal basis of \(W\) such that the resulting matrix representing \(T\) is upper triangular. True or False.
Solution
(a) True. With no row exchanges, Gaussian elimination factors \(A = LU\) with the pivots on the diagonal of \(U\), and \(\det A = \det U\) = product of the pivots. (If the elimination runs into a zero pivot without a row exchange, the matrix is singular and the product of pivots is 0, consistent with \(\det A = 0\).) (b) False. \(A\) is \(m \times n\) of rank \(n\); \(AA^T\) is \(m \times m\) of rank \(n\). When \(m > n\) it is singular (has a nullspace), so it cannot be positive definite. Example: \(A = \begin{bmatrix} 1 \\ 0 \end{bmatrix}\) has independent columns but \(AA^T = \begin{bmatrix} 1 & 0 \\ 0 & 0 \end{bmatrix}\) is not positive definite. (What is true: \(A^TA\) is positive definite.) (c) True. A symmetric \(A\) diagonalizes orthogonally, \(A = Q\Lambda Q^T\), so \(A^TA = Q\Lambda^2Q^T\) and the eigenvalues of \(A^TA\) are \(\lambda_i^2\). Hence the singular values are \(\sigma_i = |\lambda_i|\). (d) False. Eigenvalues of a unitary matrix have absolute value 1, but \(|1 + i| = \sqrt{2} \neq 1\). (e) True. \(\ker T = \{0\}\) with \(\dim V = \dim W = n\) makes \(T\) an isomorphism, so for any basis \(v_1, \dots, v_n\) of \(V\), the vectors \(Tv_1, \dots, Tv_n\) are linearly independent. Apply Gram-Schmidt to them to get an orthonormal basis \(w_1, \dots, w_n\) of \(W\) with \(\text{span}\{w_1, \dots, w_j\} = \text{span}\{Tv_1, \dots, Tv_j\}\) for each \(j\); the matrix of \(T\) in \((v\)-basis, \(w\)-basis) is then upper triangular (column \(j\) has zeroes below entry \(j\)).Keywords: pivots, positive definite, singular values, unitary matrix