Skip to content

4.2-final-2012-strang

Final Exam, 2012 (strang) · Section 4.2: Projections onto Lines and Subspaces Source: finalexam_2.pdf, pp. 79-80

Problem

4 (12 pts.) This problem is about \(x + 2y + 2z = 0\), which is the equation of a plane through 0 in \(\mathbb{R}^3\).

(a) That plane is the nullspace of what matrix \(A\)? Find an orthonormal basis for that nullspace (that plane).

(b) That plane is the column space of many matrices \(B\). Give two examples of \(B\).

(c) How would you compute the projection matrix \(P\) onto that plane? (A formula is enough.) What is the rank of \(P\)?

Solution (a) \[ A = \begin{bmatrix} 1 & 2 & 2 \end{bmatrix}. \] We could identify a basis of \(N(A)\) as usual and then apply Gram-Schmidt, but here we can guess two orthonormal vectors in \(N(A)\). First, \[ v_1 = \begin{bmatrix} 0 \\ 1 \\ -1 \end{bmatrix} \in N(A). \] Then anything of the form \([a \quad 1 \quad 1]\) is orthogonal to \(v_1\), and we pick the one that is also in the nullspace: \[ v_2 = \begin{bmatrix} -4 \\ 1 \\ 1 \end{bmatrix} \in N(A). \] An orthonormal basis is \[ q_1 = \frac{1}{\sqrt{2}} \begin{bmatrix} 0 \\ 1 \\ -1 \end{bmatrix}, \qquad q_2 = \frac{1}{3\sqrt{2}} \begin{bmatrix} -4 \\ 1 \\ 1 \end{bmatrix}. \] (b) We can use the basis vectors above as columns, or independent linear combinations of them; a zero column may be filled in: \[ B_1 = [v_1 \quad v_2], \qquad B_2 = [v_1 \quad 2v_1 + v_2], \qquad B_3 = [v_1 \quad v_2 \quad 0]. \] Then \(C(B_i) = N(A)\) for each of them. (c) Using a matrix \(B\) with independent columns (so \(B_1\) or \(B_2\), but not \(B_3\), here), the usual formula gives \[ P = B(B^T B)^{-1} B^T. \] For a projection, \(C(P)\) is always the subspace it projects onto — here the two-dimensional plane — so \[ \mathrm{rank}(P) = \dim C(P) = 2. \]

Keywords: nullspace, orthonormal basis, column space, projection matrix