Skip to content

4.2-exam2-2010

Quiz 2, 2010 · Section 4.2: Projections onto Lines and Subspaces Source: exam2_2.pdf, pp. 1-3

Problem

1. (33 points)

(a) Find the matrix \(P\) that projects every vector \(b\) in \(R^3\) onto the line in the direction of \(a = (2, 1, 3)\).

(b) What are the column space and nullspace of \(P\)? Describe them geometrically and also give a basis for each space.

(c) What are all the eigenvectors of \(P\) and their corresponding eigenvalues? (You can use the geometry of projections, not a messy calculation.) The diagonal entries of \(P\) add up to __.

Solution (a) The general formula for the orthogonal projection onto the column space of a matrix \(A\) is \[ P = A(A^TA)^{-1}A^T. \] Here \[ A = \begin{bmatrix} 2 \\ 1 \\ 3 \end{bmatrix} \quad \text{so that} \quad P = \frac{1}{14} \begin{bmatrix} 4 & 2 & 6 \\ 2 & 1 & 3 \\ 6 & 3 & 9 \end{bmatrix}. \] Remarks: since we're projecting onto a one-dimensional space, \(A^TA\) is just a number and we can write things like \(P = (AA^T)/(A^TA)\). This won't work in general. You don't have to know the formula to do this: the \(i\)-th column of \(P\) is, pretty much by definition, the projection of \(e_i\) (\(e_1 = (1, 0, 0)\), \(e_2 = (0, 1, 0)\), \(e_3 = (0, 0, 1)\)) onto the line in the direction of \(a\). And this is something you should know how to do without a formula. RUBRIC: There was some leniency for computational errors, but otherwise there weren't many opportunities for partial credit. (b) The column space is the line in \(R^3\) in the direction of \(a = (2, 1, 3)\). One basis for it is \[ \begin{bmatrix} 2 \\ 1 \\ 3 \end{bmatrix}, \] and there's not really much choice in giving this basis (you can rescale by a non-zero constant). The nullspace is the plane in \(R^3\) that is perpendicular to \(a = (2, 1, 3)\) (i.e., \(2x + y + 3z = 0\)). One basis for it is \[ \begin{bmatrix} 3 \\ 0 \\ -2 \end{bmatrix}, \quad \begin{bmatrix} -1 \\ 2 \\ 0 \end{bmatrix}, \] though there are a lot of different looking choices for it (any two vectors that are perpendicular to \(a\) and not in the same line will work). RUBRIC: 6 points for giving a correct basis, and 4 points for giving the complete geometric description. Note that it is not correct to say e.g., \(N(P) = R^2\). It is correct to say that \(N(P)\) is a (2-dimensional) plane in \(R^3\), but this is not a complete geometric description unless you say (geometrically) which plane it is: the one perpendicular to \(a\)/to the line through \(a\). (c) The diagonal entries of \(P\) add up to \(1\) = the sum of the eigenvalues. Since \(P\) is a projection, its only possible eigenvalues are \(\lambda = 0\) (with multiplicity equal to the dimension of the nullspace, here 2) and \(\lambda = 1\) (with multiplicity equal to the dimension of the column space, here 1). So a complete list of eigenvectors and eigenvalues is: - \(\lambda = 0\) with multiplicity 2. The eigenvectors for \(\lambda = 0\) are precisely the vectors in the nullspace, i.e., all linear combinations of \(\begin{bmatrix} 3 & 0 & -2 \end{bmatrix}^T\) and \(\begin{bmatrix} -1 & 2 & 0 \end{bmatrix}^T\). - \(\lambda = 1\) with multiplicity 1. The eigenvectors for \(\lambda = 1\) are precisely the vectors in the column space, i.e., all multiples of \(\begin{bmatrix} 2 & 1 & 3 \end{bmatrix}^T\). RUBRIC: 2 points for the sum of eigenvalues, 4 points for a full list (with multiplicities) of eigenvalues, and 4 points for a complete description of all eigenvectors. In light of the emphasized "all," you'd lose 1 point if you gave two eigenvectors for \(\lambda = 0\) and didn't say that all (at least non-zero) linear combinations were also eigenvectors for \(\lambda = 0\).

Keywords: projection matrix, column space, nullspace, eigenvectors