This is the problem.

The solution is below.

How did they row reduce the matrix like that below? How did the $x_i$ variables become $1$s and $0$s? I'd like someone to further clarify the solution.


You want to have :

$$t_1L_1 + t_2L_2 = \vec{0}$$

By substituting in, this translates to the vector equation :

$$t_1(7x_1,x_2-5x_3) + t_2(x_1+6x_2+x_3,x_3)=0$$ $$\Rightarrow$$

$$\begin{cases} 7t_1x_1 + t_2(x_1+6x_2 + x_3)=0 \\ (x_2-5x_3)t_1 + t_2x_3 = 0\end{cases}$$


$$\begin{cases} (7t_1+t_2)x_1 + 6t_2x_2 + t_2x_3=0 \\ t_1x_2 + (t_2-5t_1)x_3 = 0 \end{cases}$$

In order for your system to be equal to $0$ for any pair of coordinates, you need the coefficients to be zero, hence :

$$\begin{cases} 7t_1 + t_2= 0, t_2 = 0 , t_2 = 0 \\ t_1 = 0 , t_2 - 5t_1 = 0 \end{cases}$$


$$t_1=t_2 = 0$$

What the author is doing there is that he is converting the pair of equations to a matrix expression factoring out $t_1,t_2$ and applying row operations (alike Gauss' Elimination) to the matrix of the system that gets formed to elementary show that the $X,Y$ corresponding parts must be equal to $0$, since from the last matrix he derives $1\cdot X + 0 = 0$ and $0 + 1\cdot Y = 0$, which ultimately means that $t_1=t_2=0$ as proved above.

  • $\begingroup$ "In order for your system to be equal to 0 for any pair of coordinates, you need the coefficients to be zero, hence :" This is an unjustified assumption since the question is asking whether or not you need the coefficients to be zero, or if there is a non-zero solution. Hence you have to justify this statement as it essentially answers the question. $\endgroup$ – Max Li Feb 20 '18 at 22:09
  • $\begingroup$ Try solving the system with respect to one unknown and substituting. In this case, you will have a random variable with values over the reals. The only way this random expression which will depend on a variable to be zero would be for the coefficients to be zero. I didn't elaborate that but you can easily figure it out, as you have a 2×2 system with 3 unknowns! $\endgroup$ – Rebellos Feb 20 '18 at 22:13
  • $\begingroup$ What's the explanation for the matrix reductions they did in the provided solution in my original post? $\endgroup$ – Max Li Feb 20 '18 at 22:16
  • $\begingroup$ The system derived by the way I described or by the linearity of the solutions giving you the zero vector 8s reduced into a Gaussian form of elimination, as I mentioned in the post. $\endgroup$ – Rebellos Feb 20 '18 at 22:19
  • $\begingroup$ When I used your method and reduced it into a Gaussian form, I didn't have any "x"s in the matrix to start with since I was solving for (c1t1+c2t2+...) = 0, so I just put the coefficients c1, c2... into the matrix. How would using your method and then putting it into a matrix get "x" variables in the matrix? And how would we "eliminate" the "x" variables as shown in the solution I posted when there are only two rows and so many variables? $\endgroup$ – Max Li Feb 20 '18 at 22:24

Maybe the notation was confusing...

t1 * $\mathbf{L1}$ + t2 * $\mathbf{L2}$ = $\vec{0}$


$\begin{pmatrix}\mathbf{L1} & \mathbf{L2}\end{pmatrix} \begin{pmatrix} t1 \\ t2 \end{pmatrix}$ = $\vec{0}$


$\left(\begin{array}{cc|c}\mathbf{L1} & \mathbf{L2} & \vec{0} \end{array}\right)$

Then to row reduce this matrix ...

Add $-((x_2-5x_3)/7x_1)*row1$ to $row2$:

$\left(\begin{array}{cc|c}\begin{matrix} 7x_1 \\ 0 \end{matrix} & \begin{matrix} x_1 + 6x_2 + x_3 \\ BLOB_1 \end{matrix} & \vec{0} \end{array}\right)$

Add $-((x_1+6x_2+x_3)/BLOB_1)*row2$ to $row1$:

$\left(\begin{array}{cc|c}\begin{matrix} 7x_1 \\ 0 \end{matrix} & \begin{matrix} 0 \\ BLOB_1 \end{matrix} & \vec{0} \end{array}\right)$

Then, normalize $row1$ and $row2$

$\left(\begin{array}{cc|c}\begin{matrix} 1 \\ 0 \end{matrix} & \begin{matrix} 0 \\ 1 \end{matrix} & \vec{0} \end{array}\right)$

  • $\begingroup$ No, I understand that. What I don't understand is how the $x_i$s disappear after that, and how that matrix was reduced. $\endgroup$ – Max Li Feb 20 '18 at 22:07
  • $\begingroup$ Sorry for the delay... $\endgroup$ – Victor Feb 21 '18 at 12:40

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.