Skip to content

6.1-exam3-2007-johnson

Quiz 3, 2007 (johnson) · Section 6.1: Introduction to Eigenvalues : Ax = λx Source: exam3_1.pdf, p. 124

Problem

4 (20 pts.) Some \(3 \times 3\) real matrix \(A\) has eigenvalues \(\lambda_1 = 0\), \(\lambda_2 = 1\), and \(\lambda_3 = 2\), with the corresponding eigenvectors \(\vec{x}_1 = (1, 0, 0)^T\), \(\vec{x}_2 = (0, 1, 2)^T\), and \(\vec{x}_3 = (0, 1, 1)^T\).

(a) Give a basis for: (i) the nullspace \(N(A)\), (ii) the column space \(C(A)\), and (iii) the row space \(C(A^H)\).

(b) Find all solutions \(\vec{x}\) to \(A\vec{x} = \vec{x}_2 - 3\vec{x}_3\).

(c) Is \(A\) (i) real-symmetric, (ii) orthogonal, (iii) Markov, or (iv) none of the above?

Solution (a) The nullspace is just the span of the \(\lambda = 0\) eigenvector \(\vec{x}_1\). If we act \(A\) on any vector, we only get multiples of the \(\lambda \neq 0\) eigenvectors, so \(C(A)\) is the span of \(\vec{x}_2\) and \(\vec{x}_3\). The row space is the orthogonal complement of the nullspace, and here this is spanned by (e.g.) the vectors \((0, 1, 0)^T\) and \((0, 0, 1)^T\). (b) The right hand side is clearly in the column space. Since we have expanded the right hand side in the \(\lambda \neq 0\) eigenvectors, we can get a particular solution just by dividing them by the corresponding eigenvalues: remember, \(A\) acts just like a number on these vectors. Hence a particular solution is \(\vec{x}_p = \vec{x}_2/1 - 3\vec{x}_3/2 = (0, -1/2, 1/2)^T\). To get all the solutions we must add the nullspace, obtaining \[ \vec{x} = (a, -1/2, 1/2)^T \quad \text{for any constant } a. \] Equivalently, expand \(\vec{x}\) in the eigenvectors, \(\vec{x} = a\vec{x}_1 + b\vec{x}_2 + c\vec{x}_3\), and plug in to \(A\vec{x} = b\vec{x}_2 + 2c\vec{x}_3 = \vec{x}_2 - 3\vec{x}_3\) to find \(a =\) arbitrary, \(b = 1\), and \(c = -3/2\). (c) (iv) None of the above. It's clearly not Markov or orthogonal since there is a \(\lambda = 2\) eigenvalue. Although the eigenvalues are real, it's not real-symmetric since the eigenvectors are not orthogonal.

Keywords: eigenvalues, eigenvectors, nullspace, column space, row space