Skip to content

4.2-final-2011-strang

Final Exam, 2011 (strang) · Section 4.2: Projections onto Lines and Subspaces Source: finalexam_2.pdf, p. 49

Problem

2. (a) Compute the 4 by 4 matrix \(P\) that projects every vector in \(\mathbb{R}^4\) onto the column space of \(A\):

\[ A = \begin{bmatrix} 1 & -1 \\ 1 & 1 \\ 1 & 1 \\ 1 & -1 \end{bmatrix}. \]

(b) What are the four eigenvalues of \(P\)? Explain your reasoning.

(c) Find a unit vector \(u\) (length 1) that is as far away as possible from the column space of \(A\).

Solution (a) The two columns of \(A\) are orthogonal (their dot product is \(-1 + 1 + 1 - 1 = 0\)) and each has length squared \(4\). Hence \(A^T A = 4I\) and \[ P = A(A^T A)^{-1}A^T = \frac{1}{4} AA^T = \frac{1}{4} \begin{bmatrix} 2 & 0 & 0 & 2 \\ 0 & 2 & 2 & 0 \\ 0 & 2 & 2 & 0 \\ 2 & 0 & 0 & 2 \end{bmatrix}. \] (b) \(P\) is a projection onto the 2-dimensional column space of \(A\), so its eigenvalues are \(1, 1, 0, 0\): \(\lambda = 1\) with multiplicity 2 (vectors in the column space are fixed) and \(\lambda = 0\) with multiplicity 2 (vectors in the orthogonal complement \(N(A^T)\) are sent to zero). (c) The vector farthest from the column space is a unit vector in the orthogonal complement \(N(A^T)\), perpendicular to both columns \((1, 1, 1, 1)\) and \((-1, 1, 1, -1)\). One such unit vector is \[ u = \frac{1}{2} \begin{bmatrix} 1 \\ 1 \\ -1 \\ -1 \end{bmatrix}. \] Its distance from the column space is the full length \(\|u\| = 1\).

Keywords: projection matrix, eigenvalues, orthogonal complement, distance