Skip to content

4.3-exam2-1997

Hour Exam II, 1997 · Section 4.3: Least Squares Approximations Source: exam2_1.pdf, pp. 5-6

Problem

3. Let \(A = \begin{bmatrix} 1 & -2 & -5 & 1 \\ 2 & -4 & -10 & 3 \end{bmatrix}\).

(3a) (16 points) Find the solution to \(Ax = \begin{bmatrix} 0 \\ 0 \end{bmatrix}\) that is closest to \(\begin{bmatrix} 5 \\ 5 \\ 11 \\ 11 \end{bmatrix}\).

(3b) (12 points) Give an orthonormal basis for the nullspace of \(A = \begin{bmatrix} 1 & -2 & -5 & 1 \\ 2 & -4 & -10 & 3 \end{bmatrix}\).

Solution 3a. \(x = \begin{bmatrix} 7 \\ 1 \\ 1 \\ 0 \end{bmatrix}\). Gaussian elimination reveals that the nullspace of \(A\) spans a 2-dimensional subspace of \(R^4\). In particular, the special solutions \(\begin{bmatrix} 2 \\ 1 \\ 0 \\ 0 \end{bmatrix}\), \(\begin{bmatrix} 5 \\ 0 \\ 1 \\ 0 \end{bmatrix}\) give a basis for this subspace. The vector \(\begin{bmatrix} 5 \\ 5 \\ 11 \\ 11 \end{bmatrix}\) is closest to \(c_1 \begin{bmatrix} 2 \\ 1 \\ 0 \\ 0 \end{bmatrix} + c_2 \begin{bmatrix} 5 \\ 0 \\ 1 \\ 0 \end{bmatrix}\), where \(\begin{bmatrix} c_1 \\ c_2 \end{bmatrix}\) satisfies the normal equations \[ \begin{bmatrix} 2 & 1 & 0 & 0 \\ 5 & 0 & 1 & 0 \end{bmatrix} \begin{bmatrix} 2 & 5 \\ 1 & 0 \\ 0 & 1 \\ 0 & 0 \end{bmatrix} \begin{bmatrix} c_1 \\ c_2 \end{bmatrix} = \begin{bmatrix} 2 & 1 & 0 & 0 \\ 5 & 0 & 1 & 0 \end{bmatrix} \begin{bmatrix} 5 \\ 5 \\ 11 \\ 11 \end{bmatrix}, \] i.e. \[ \begin{bmatrix} 5 & 10 \\ 10 & 26 \end{bmatrix} \begin{bmatrix} c_1 \\ c_2 \end{bmatrix} = \begin{bmatrix} 15 \\ 36 \end{bmatrix}. \] The coefficients are \(c_1 = 1\) and \(c_2 = 1\), giving \(x = [7, 1, 1, 0]^T\). 3b. The special solutions to \(A = \begin{bmatrix} 1 & -2 & -5 & 1 \\ 2 & -4 & -10 & 3 \end{bmatrix} x = \begin{bmatrix} 0 \\ 0 \end{bmatrix}\) give a linearly independent basis for the nullspace of \(A\) (see Problem 3a). Next, we make an orthogonal basis for the nullspace. The orthogonal vectors are denoted as \(A\), \(B\): \[ A = \begin{bmatrix} 2 \\ 1 \\ 0 \\ 0 \end{bmatrix}, \qquad B = \begin{bmatrix} 5 \\ 0 \\ 1 \\ 0 \end{bmatrix} - \left(\frac{10}{5}\right) \underbrace{\begin{bmatrix} 2 \\ 1 \\ 0 \\ 0 \end{bmatrix}}_{A} = \begin{bmatrix} 1 \\ -2 \\ 1 \\ 0 \end{bmatrix}. \] We make an orthonormal basis by scaling these vectors to have unit length: \(\|A\| = \sqrt{5}\) and \(\|B\| = \sqrt{6}\). So \[ q_1 = \frac{1}{\sqrt{5}} \begin{bmatrix} 2 \\ 1 \\ 0 \\ 0 \end{bmatrix}, \qquad q_2 = \frac{1}{\sqrt{6}} \begin{bmatrix} 1 \\ -2 \\ 1 \\ 0 \end{bmatrix}. \]

Keywords: nullspace, projection, normal equations, orthonormal basis