Skip to content

4.2-exam2-2015

Exam II, 2015 · Section 4.2: Projections onto Lines and Subspaces Source: exam2_2.pdf, pp. 97, 100-101

Problem

2. (34 points) Suppose \(q_1 = (c, d, e)\) and \(q_2 = (f, g, h)\) are orthonormal column vectors in \(R^3\). They span a subspace \(S\).

(a) Find the (1, 1) entry in the projection matrix \(P\) that projects each vector in \(R^3\) onto that subspace \(S\).

(b) For this projection matrix \(P\), describe 3 independent eigenvectors (vectors for which \(Px\) is a number \(\lambda\) times \(x\)). What are the 3 eigenvalues of \(P\)? What is its determinant?

(c) For some vectors \(v\) and \(w\) in \(R^3\) the Gram-Schmidt orthonormalization process (applied to \(v\) and \(w\)) will produce those particular vectors \(q_1\) and \(q_2\). Describe the vectors \(v\) and \(w\) that lead to this \(q_1\) and \(q_2\).

Solution (a) Denote by \(Q\) the matrix with columns \(q_1\) and \(q_2\): \(Q = \begin{bmatrix} c & f \\ d & g \\ e & h \end{bmatrix}\). The projection matrix is \(P = Q(Q^TQ)^{-1}Q^T\). As the column vectors are orthonormal, we know that \(Q^TQ\) is the 2 by 2 identity matrix. Thus, \(P = QQ^T\), and the first entry is \(c^2 + f^2\). (b) The projection matrix \(P\) projects onto a 2-dimensional plane. That means its eigenvalues are \((1, 1, 0)\) and the determinant is 0. The eigenvector corresponding to the eigenvalue 0 is orthogonal to the projection plane, that is orthogonal to both vectors \(q_1\) and \(q_2\). The independent vectors corresponding to eigenvalue 1 are any two independent vectors in the projection plane. We can choose \(q_1\) and \(q_2\) as such vectors. (c) Vector \(v\) is on the same line as \(q_1\) and in the same direction. Therefore, \(v = aq_1\), where \(a\) is a positive number. The second vector \(w\) has to be in the same plane as \(q_1\) and \(q_2\), on the same side of the line drawn through \(q_1\) as \(q_2\), and has to be independent of \(v\).

Keywords: projection matrix, orthonormal columns, eigenvalues, Gram-Schmidt