7.1-final-2011-strang¶
Final Exam, 2011 (strang) · Section 7.1: Singular Values and Singular Vectors Source: finalexam_2.pdf, p. 55
Problem¶
8. Suppose the 2 by 3 matrix \(A\) has \(Av_1 = 3u_1\) and \(Av_2 = 5u_2\) with orthonormal \(v_1, v_2\) in \(\mathbb{R}^3\) and orthonormal \(u_1, u_2\) in \(\mathbb{R}^2\).
(a) Describe the nullspace of \(A\).
(b) Find the eigenvalues of \(A^T A\).
(c) Find the eigenvalues and eigenvectors of \(AA^T\).
Solution
(a) Since \(Av_1 = 3u_1\) and \(Av_2 = 5u_2\) are independent, the rank of \(A\) is 2, so the nullspace has dimension \(3 - 2 = 1\). The nullspace is the orthogonal complement of the row space, and the row space is spanned by \(v_1, v_2\) (these are the eigenvectors of \(A^T A\) with nonzero eigenvalues). So \[ N(A) = \text{span}\{v_3\}, \] where \(v_3 = v_1 \times v_2\) is the unit vector orthogonal to both \(v_1\) and \(v_2\). (Indeed \(\|Av_3\|^2 = v_3^T A^T A v_3 = 0\) since \(A^T A v_3 = 0\).) (b) \(v_1, v_2\) are orthonormal eigenvectors of \(A^T A\): \[ A^T A v_1 = A^T(3u_1) = 9v_1, \qquad A^T A v_2 = 25v_2, \] so the eigenvalues of \(A^T A\) are \(9, 25, 0\) (the third eigenvalue is 0 because the rank is 2). (c) The nonzero eigenvalues of \(AA^T\) are the same \(9\) and \(25\), with eigenvectors \(u_1\) and \(u_2\): \[ AA^T u_1 = 9u_1, \qquad AA^T u_2 = 25u_2. \] (\(AA^T\) is 2 by 2, so these are all of its eigenvalues and eigenvectors.)Keywords: singular values, SVD, nullspace, eigenvalues of A^T A