Skip to content

4.4-exam2-1997-strang

Quiz 2, 1997 (strang) · Section 4.4: Orthonormal Bases and Gram-Schmidt Source: exam2_1.pdf, pp. 9, 11-12

Problem

3. (a) Suppose \(q_1, q_2, q_3\) are orthonormal vectors in \(R^6\). Under what condition on the vector \(v\) will there be a fourth orthonormal vector \(q_4\) that is a combination of \(v, q_1, q_2, q_3\)?

(b) Give a formula for that fourth orthonormal vector \(q_4\).

(c) Suppose \(q_1, \dots, q_n\) is an orthonormal basis for \(R^n\). Define the \(n \times n\) matrix \(A = q_1 q_1^T + \dots + q_n q_n^T\). What does \(Aq_1\) equal? What does \(Aq_i\) equal? What is \(A\)?

Solution 3a. The vectors \(v, q_1, q_2, q_3\) must be linearly independent (equivalently, \(v\) must not lie in the span of \(q_1, q_2, q_3\)). 3b. \[ q_4 = \frac{v - (v^T q_1)q_1 - (v^T q_2)q_2 - (v^T q_3)q_3}{\| v - (v^T q_1)q_1 - (v^T q_2)q_2 - (v^T q_3)q_3 \|}. \] It is always OK to write \(q^T v\) instead of \(v^T q\) for real vectors. 3c. \[ Aq_1 = q_1 q_1^T q_1 + q_2 q_2^T q_1 + \dots + q_n q_n^T q_1 = q_1 \cdot 1 + q_2 \cdot 0 + \dots + q_n \cdot 0 = q_1. \] Similarly \(Aq_i = q_i\) for every \(i\). Since the \(q_i\)'s are a basis for \(R^n\), we get \(A = I\). Other method (columns \(\times\) rows): \[ A = [q_1 \dots q_n] \begin{bmatrix} q_1^T \\ \vdots \\ q_n^T \end{bmatrix} = QQ^T = QQ^{-1} = I. \]

Keywords: orthonormal vectors, Gram-Schmidt, projection, identity matrix