3
$\begingroup$

I have a system of equations to solve $Ax = b$, but the determinant of matrix $A$ is zero.

$$A=\begin{bmatrix}1&1&0&0\\0&0&1&1\\1&0&1&0\\0&1&0&1\end{bmatrix}$$

To me, none of the rows/columns of this matrix look dependent. I am wondering why the determinant of matrix $A$ is zero?

Also, how should I solve this system or estimate $x$?

|cite|improve this question
$\endgroup$
  • 10
    $\begingroup$ Note that adding the first two rows and subtracting the third yields the fourth, so the row (and column) vectors are not linearly independent. $\endgroup$ – user455343 Jul 28 '17 at 19:37
  • 1
    $\begingroup$ I think I see why you're confused -- there is a nice symmetry in this matrix that would indeed seem to suggest that it is full-rank, since all components of each basis vector contribute equally or "fairly". It definitely defies my intuition as well, so I guess this is yet another lesson that you shouldn't trust your intuition! $\endgroup$ – Mehrdad Jul 28 '17 at 23:43
  • 2
    $\begingroup$ The sum of the first two is equal to the sum of the last two. $\endgroup$ – Alfred Yerger Jul 29 '17 at 4:28
4
$\begingroup$

Suppose $b=\begin{bmatrix}b_1&b_2&b_3&b_4\end{bmatrix}^T$ and consider the Gaussian elimination on the complete matrix: \begin{align} \left[\begin{array}{cccc|c} 1&1&0&0&b_1\\ 0&0&1&1&b_2\\ 1&0&1&0&b_3\\ 0&1&0&1&b_4 \end{array}\right] &\to \left[\begin{array}{cccc|c} 1&1&0&0&b_1\\ 0&0&1&1&b_2\\ 0&-1&1&0&b_3-b_1\\ 0&1&0&1&b_4 \end{array}\right] &&R_3\gets R_3-R_1 \\[6px]&\to \left[\begin{array}{cccc|c} 1&1&0&0&b_1\\ 0&1&-1&0&b_1-b_3\\ 0&0&1&1&b_2\\ 0&1&0&1&b_4 \end{array}\right] &&R_3\leftrightarrow R_2,\quad R_2\gets-R_2 \\[6px]&\to \left[\begin{array}{cccc|c} 1&1&0&0&b_1\\ 0&1&-1&0&b_1-b_3\\ 0&0&1&1&b_2\\ 0&0&1&1&b_4-b_1+b_3 \end{array}\right] &&R_4\gets R_4-R_2 \\[6px]&\to \left[\begin{array}{cccc|c} 1&1&0&0&b_1\\ 0&1&-1&0&b_1-b_3\\ 0&0&1&1&b_2\\ 0&0&0&0&b_4-b_1+b_3-b_2 \end{array}\right] &&R_4\gets R_4-R_3 \tag{*} \\[6px]&\to \left[\begin{array}{cccc|c} 1&1&0&0&b_1\\ 0&1&0&1&b_1-b_3+b_2\\ 0&0&1&1&b_2\\ 0&0&0&0&b_4-b_1+b_3-b_2 \end{array}\right] &&R_2\gets R_2+R_3 \\[6px]&\to \left[\begin{array}{cccc|c} 1&0&0&-1&b_3-b_2\\ 0&1&0&1&b_1-b_3+b_2\\ 0&0&1&1&b_2\\ 0&0&0&0&b_4-b_1+b_3-b_2 \end{array}\right] &&R_1\gets R_1-R_2 \end{align} From the step marked as (*) we deduce that

  1. the matrix $A$ has rank $3$ (hence zero determinant);
  2. the system has solution if and only if $b_4-b_1+b_3-b_2=0$.

When $b_4-b_1+b_3-b_2=0$, the system has infinitely many solutions in the form $$ \begin{bmatrix} b_3-b_2+h\\ b_1-b_3+b_2-h\\ b_2-h\\ h \end{bmatrix} = \begin{bmatrix} b_3-b_2\\ b_1-b_3+b_2\\ b_2\\ 0 \end{bmatrix} + h\begin{bmatrix} 1 \\ -1 \\ -1 \\ 1 \end{bmatrix} $$ where $h$ is any scalar.

|cite|improve this answer
$\endgroup$
13
$\begingroup$

$$ (1,1,0,0)+(0,0,1,1)-(1,0,1,0)-(0,1,0,1)=(0,0,0,0) $$ so the rows are linearly dependent.

|cite|improve this answer
$\endgroup$
3
$\begingroup$

Consider the following block matrix: $$ M=\left[\begin{array}[cc]\\A&B\\ C&D\end{array}\right] $$ where $A$,$B$,$C$ and $D$ have the same order. If $D$ be invertible then we have: $$ \det(M)=\det(D)\cdot \det(A-BD^{-1}C) $$ In your question $C=D=I_2$,the identity matrix of order $2$, which results that: $$ \det(M)=\underbrace{\det(D)}_1\cdot \det(A-B)=\det( \left[ \begin{array}[cc] \\1&1\\ -1&-1 \end{array} \right])=0 $$

|cite|improve this answer
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.