4.2-exam2-1998-strang¶
Exam 2, 1998 (strang) · Section 4.2: Projections onto Lines and Subspaces Source: exam2_1.pdf, pp. 22, 24
Problem¶
4 (a) (10) Give the formula for the projection matrix \(P\) onto the column space of a matrix \(A\). Where does the formula assume that \(A\) has independent columns?
(b) (5) The two properties of all these projection matrices are \(P^2 = P\) and \(P^T = P\). Suppose \(v^T\) is the first row of \(P\) and \(v_1\) is the first entry in that row. Prove that \(v^T v = v_1\).
Solution
4a. \(P = A(A^T A)^{-1}A^T\): the matrix \(A^T A\) is invertible if and only if \(A\) has independent columns, and that is where the formula assumes the independence. 4b. The properties give \(PP^T = P\). Compare the \((1, 1)\) entry on both sides of this equation: the \((1, 1)\) entry of \(PP^T\) is the first row \(v^T\) times the first column \(v\), i.e. \(v^T v\), and the \((1, 1)\) entry of \(P\) is \(v_1\). Hence \(v^T v = v_1\).Keywords: projection matrix, independent columns, symmetric, idempotent