Skip to content

3.5-exam1-2010-fall

Exam 1, 2010 (fall) · Section 3.5: Dimensions of the Four Subspaces Source: exam1_1.pdf, pp. 221-222

Problem

3 (20 pts.) (In the questions below, you can choose any \(n\) that works for an example, or prove that for all \(n\), there is no example.)

(a) (10 pts.) Can you find independent vectors \(v, w, x\) and \(y\) in some space \(\mathbb{R}^n\) and where \(A = vw^T + xy^T\) is invertible? or prove that no such example exists?

(b) (10 pts.) Can you find vectors \(v, w, x\) and \(y\) that span some space \(\mathbb{R}^n\) and where \(A = vw^T + xy^T\) is invertible? or prove that no such example exists?

Solution (a) No. Matrices of the form \(vw^T\) always have rank \(\leq 1\). The sum of two matrices of rank \(\leq 1\) has rank \(\leq 2\). So if \(A\) is invertible we need \(n \leq 2\). But we also have the requirement that \(v, w, x, y\) are independent vectors, which forces \(n \geq 4\). So we can't pick any \(n\) to make both inequalities happen. (b) Yes. There are many possibilities. The easiest is to take \(n = 1\). Then \(v, w, x, y\) are just numbers and \(A\) being invertible means it is nonzero. So we could take \(v = w = 1\) and \(x = y = 0\), for example.

Keywords: rank, outer product, invertible matrix, independence