Skip to content

4.4-exam2-2009-postnikov

Quiz 2, 2009 (postnikov) · Section 4.4: Orthonormal Bases and Gram-Schmidt Source: exam2_1.pdf, pp. 183-184

Problem

Problem 1. Consider the matrix \(A = \begin{bmatrix} 1 & 1 \\ 1 & 3 \\ 1 & 5 \end{bmatrix}\).

(a) Find an orthogonal basis of the column space of the matrix \(A\).

(b) Find a non-zero vector \(v\) which is orthogonal to the column space of \(A\).

(c) Does this vector \(v\) belong to one of the four fundamental subspaces of \(A\)? Which subspace? Explain why.

(d) Find a 3 by 2 matrix \(Q\) with \(Q^TQ = I\) such that \(Q\) has the same column space as the matrix \(A\).

Solution (a) Use Gram-Schmidt on the columns of \(A = (a_1, a_2)\): \[ q_1 = \frac{a_1}{\|a_1\|} = \frac{1}{\sqrt{3}} \begin{bmatrix} 1 \\ 1 \\ 1 \end{bmatrix}, \qquad q_2 = \frac{a_2 - (a_2^T q_1) q_1}{\|a_2 - (a_2^T q_1) q_1\|}. \] Since \(a_2^T q_1 = \frac{9}{\sqrt{3}}\), the projection of \(a_2\) onto \(q_1\) is \(\frac{9}{\sqrt{3}} \cdot \frac{1}{\sqrt{3}} (1, 1, 1)^T = 3(1, 1, 1)^T\), so \[ q_2 = \frac{\begin{bmatrix} 1 \\ 3 \\ 5 \end{bmatrix} - 3 \begin{bmatrix} 1 \\ 1 \\ 1 \end{bmatrix}}{\left\| \begin{bmatrix} 1 \\ 3 \\ 5 \end{bmatrix} - 3 \begin{bmatrix} 1 \\ 1 \\ 1 \end{bmatrix} \right\|} = \frac{\begin{bmatrix} -2 \\ 0 \\ 2 \end{bmatrix}}{\sqrt{8}} = \frac{1}{\sqrt{2}} \begin{bmatrix} -1 \\ 0 \\ 1 \end{bmatrix}. \] An orthogonal basis is given by \(\left\{ \frac{1}{\sqrt{3}} \begin{bmatrix} 1 \\ 1 \\ 1 \end{bmatrix}, \frac{1}{\sqrt{2}} \begin{bmatrix} -1 \\ 0 \\ 1 \end{bmatrix} \right\}\). (b) Must find \(v\) such that \(a_1^T v = 0\) and \(a_2^T v = 0\), or similarly \(q_1^T v = 0\) and \(q_2^T v = 0\) (since \(q_1\) and \(q_2\) are a basis for \(C(A)\)). This is the same as solving \[ \begin{bmatrix} q_1^T \\ q_2^T \end{bmatrix} v = 0, \quad \text{i.e.} \quad \begin{bmatrix} \frac{1}{\sqrt{3}}(1 \ 1 \ 1) \\ \frac{1}{\sqrt{2}}(-1 \ 0 \ 1) \end{bmatrix} \begin{bmatrix} v_1 \\ v_2 \\ v_3 \end{bmatrix} = \begin{bmatrix} 0 \\ 0 \\ 0 \end{bmatrix}, \] which gives \(v_1 - v_3 = 0\) and \(v_1 + v_2 + v_3 = 0\). So \(v_1 = v_3\), \(v_2 = -2v_1\), and \(v = \begin{bmatrix} 1 \\ -2 \\ 1 \end{bmatrix}\). (c) \(v \in N(A^T)\). Why? Let \(Q\) be the matrix with columns given by \(q_1\) and \(q_2\) from part (a). Then \(C(A) = C(Q) \implies N(A^T) = N(Q^T)\). And from part (b), we saw that \(v\) was a solution to \(Q^T v = 0\). Since \(C(Q)\) is 2-dimensional, \(N(Q^T)\) is 1-dimensional, so it is actually spanned by \(v\). (d) Take \(Q\) to be the matrix from part (a) with columns \(q_1\) and \(q_2\). Then \[ Q^T Q = \begin{bmatrix} q_1^T \\ q_2^T \end{bmatrix} \begin{bmatrix} q_1 & q_2 \end{bmatrix} = \begin{bmatrix} q_1^T q_1 & q_1^T q_2 \\ q_2^T q_1 & q_2^T q_2 \end{bmatrix} = \begin{bmatrix} 1 & 0 \\ 0 & 1 \end{bmatrix} = I, \] and \(Q\) has the same column space as the matrix \(A\).

Keywords: Gram-Schmidt, orthogonal basis, column space, left nullspace