The question I am currently working on is, "...find $a$ and $b$ such that $\vec{v} = a \vec{u} + b \vec{w}$, where $ \vec{u} = \langle 1,2 \rangle$, $\vec{w} = \langle 1,-1 \rangle$, and $\vec{v} = \langle 2,1 \rangle$ Before I asked a question on this forum, I did a google search to see if it had been previously asked. Indeed, I found one on yahoo answers: http://answers.yahoo.com/question/index?qid=20080719094705AAiMeCG. However, it wasn't exactly helpful. For the best answer, I understand the solution up until the part, "comparing LHS and RHS." What did the person do, why can someone just split up the equation like the person did?

  • $\begingroup$ Read more about vectors? $\endgroup$ – hjpotter92 Jan 27 '13 at 0:48
  • $\begingroup$ To him/her, $i$ and $j$ are the vectors from the basis. So you can do that. $\endgroup$ – Sigur Jan 27 '13 at 0:48

$\alpha \vec{u} + \beta \vec{w}=\langle \alpha+\beta,2 \alpha-\beta\rangle=\langle2,1\rangle.$ and now $\alpha+\beta =2$ and $2\alpha-\beta=1$. Now it's simple.


Putting in the values for $\vec{v}, \vec{u}$ and $\vec{w}$ in the equation, we get:

$$ \langle 2, 1 \rangle = a * \langle 1, 2 \rangle + b * \langle 1, -1 \rangle $$

Here, $a$ and $b$ are just scalar multipliers and nothing else. Using the property of vectors and their scalaar multiplication:

$$ f * \vec{v} = f * \left( v_1, v_2, \dots v_n \right) = \left( f*v_1, f*v_2, \dots f*v_n \right) $$

You'll arrive at two linear equations in $a$ and $b$ as follows:

  • $$ a + b = 2 $$
  • $$ 2a - b = 1 $$

Solving which, you get $a = 1$ and $b = 1$.


Your question is basically one of solving a system $y=Ax$ for a fixed $A$ and $y$. That's exactly what the multiplication of an $n\times n$ matrix with a vector of length $n$ is, a linear combination of the vectors in the matrix.

We'll work explicitly in the 2 dimensional case, as that is the context of your problem. We're given two vectors $\vec v = <a,b>,$ and $\vec w = <c,d>$. We're trying to find $\alpha$ and $\beta$ such that $$\alpha \vec v + \beta \vec w = <e,f>$$

How do we do this? Well we use a wonderful piece of notation called a matrix. This can be represented by the following: $$\left( \begin{array}{ccc} a & c \\ b & d \end{array} \right) \left( \begin{array}{ccc} \alpha \\ \beta \end{array} \right)= \left( \begin{array}{ccc} e \\ f \end{array} \right)$$

or in your case: $$\left( \begin{array}{ccc} 1 & 1 \\ 2 & -1 \end{array} \right) \left( \begin{array}{ccc} \alpha \\ \beta \end{array} \right)= \left( \begin{array}{ccc} 2 \\ 1 \end{array} \right)$$

Then you can invert the matrix using Cramer's rule, which solves the system $y=Ax$ by finding a matrix $A^{-1}$ such that $A^{-1}y=x$. This only works when the determinant of the matrix is non-zero, because the formula for the inverse is: $$A^{-1} = \frac{1}{\det A}\operatorname{adj}A$$

Where $\det$ gives the determinant of $A$, and $\operatorname{adj}$ gives the adjugate matrix. Calculating these values allows us to find the inverse matrix: $$\left( \begin{array}{ccc} \frac{1}{3} & \frac{1}{3} \\ \frac{2}{3} & \frac{-1}{3} \end{array} \right)$$ And to find $\alpha, \beta$ you simply multiply this by the vector $\vec y$, and get $$\left( \begin{array}{ccc} \alpha \\ \beta \end{array} \right) = \left( \begin{array}{ccc} \frac{1}{3} & \frac{1}{3} \\ \frac{2}{3} & \frac{-1}{3} \end{array} \right)\left( \begin{array}{ccc} 2 \\ 1 \end{array} \right)\\ \left( \begin{array}{ccc} \alpha \\ \beta \end{array} \right)= \left( \begin{array}{ccc} 1 \\ 1 \end{array} \right)$$

Which gives your solution. There are easier solutions for such a simple case, but this generalizes far better.


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.