3.2-final-2011-strang¶
Final Exam, 2011 (strang) · Section 3.2: Computing the Nullspace by Elimination: A = CR Source: finalexam_2.pdf, p. 50
Problem¶
3. Suppose \(A\) is an \(m\) by \(n\) matrix and its pivot columns (not free columns) are \(c_1, c_2, \dots, c_r\). Put these columns into a matrix \(C\).
(a) Every column of \(A\) is a __ of the columns of \(C\). How would you produce from this a matrix \(R\) so that \(A = CR\)? Explain how to construct \(R\).
(b) Using \(C\) from part (a) factor the following matrix \(A\) into \(CR\), where \(C\) has independent columns and \(R\) has independent rows.
\[
A = \begin{bmatrix} 1 & 2 & 1 \\ 2 & 4 & 2 \\ 1 & 2 & 5 \end{bmatrix}.
\]
Solution
(a) Every column of \(A\) is a linear combination of the columns of \(C\). The matrix \(R\) is \(r\) by \(n\), and its columns are the coefficient vectors: column \(j\) of \(R\) holds the coefficients that express column \(j\) of \(A\) in terms of the pivot columns. In the pivot columns themselves the coefficients are the columns of the identity matrix, so the \(r\) columns of \(R\) at the pivot positions form the identity \(I_r\); hence the rows of \(R\) are independent and \(A = CR\). (b) Elimination: row 2 minus 2 (row 1) and row 3 minus (row 1) give \[ U = \begin{bmatrix} 1 & 2 & 1 \\ 0 & 0 & 0 \\ 0 & 0 & 4 \end{bmatrix}, \] so the pivot columns of \(A\) are columns 1 and 3. Column 2 is 2 (column 1) + 0 (column 3). Therefore \[ C = \begin{bmatrix} 1 & 1 \\ 2 & 2 \\ 1 & 5 \end{bmatrix}, \qquad R = \begin{bmatrix} 1 & 2 & 0 \\ 0 & 0 & 1 \end{bmatrix}, \qquad A = CR. \] The columns of \(C\) are independent and the rows of \(R\) are independent.Keywords: A = CR, pivot columns, rank factorization, independent columns