Skip to content

3.2-final-2006

Final Exam, 2006 · Section 3.2: Computing the Nullspace by Elimination: A = CR Source: finalexam_1.pdf, p. 143

Problem

3 (3+3+5=11 pts.) (a) \(A\) and \(B\) are any matrices with the same number of rows. What can you say (and explain why it is true) about the comparison of

\[ \text{rank of } A \quad \quad \quad \text{rank of the block matrix } \begin{bmatrix} A & B \end{bmatrix} \]

(b) Suppose \(B = A^2\). How do those ranks compare? Explain your reasoning.

(c) If \(A\) is \(m\) by \(n\) of rank \(r\), what are the dimensions of these nullspaces?

\[ \text{Nullspace of } A \quad \quad \quad \text{Nullspace of } \begin{bmatrix} A & A \end{bmatrix} \]
Solution (a) All you can say is that \(\text{rank } A \leq \text{rank } \begin{bmatrix} A & B \end{bmatrix}\). (\(A\) can have any number \(r\) of pivot columns, and these will all be pivot columns for \([A \ B]\); but there could be more pivot columns among the columns of \(B\).) (b) Now \(\text{rank } A = \text{rank } \begin{bmatrix} A & A^2 \end{bmatrix}\). (Every column of \(A^2\) is a linear combination of columns of \(A\). For instance, if we call \(A\)'s first column \(a_1\), then \(Aa_1\) is the first column of \(A^2\). So there are no new pivot columns in the \(A^2\) part of \([A \ A^2]\).) (c) The nullspace \(N(A)\) has dimension \(n - r\), as always. Since \([A \ A]\) only has \(r\) pivot columns—the \(n\) columns we added are all duplicates—\([A \ A]\) is an \(m\)-by-\(2n\) matrix of rank \(r\), and its nullspace \(N([A \ A])\) has dimension \(2n - r\).

Keywords: rank, block matrix, nullspace, dimension