Well, I have this subspace: $V = \operatorname{span}\left\{ \begin{pmatrix}\frac{1}{3} \\\frac{2}{3} \\\frac{2}{3}\end{pmatrix},\begin{pmatrix}1 \\3 \\4\end{pmatrix}\right\}$ and the vector $v = \begin{pmatrix}9 \\0 \\0\end{pmatrix}$

How can I find the orthogonal projection of $v$ on $V$?
This is what I did so far:

\begin{align}&P_v(v)=\langle v,v_1\rangle v_1+\langle v,v_2\rangle v_2 =\\=& \left\langle\begin{pmatrix}9 \\0 \\0\end{pmatrix},\begin{pmatrix}\frac{1}{3} \\\frac{2}{3} \\\frac{2}{3}\end{pmatrix}\right\rangle\begin{pmatrix}\frac{1}{3} \\\frac{2}{3} \\\frac{2}{3}\end{pmatrix}+\left<\begin{pmatrix}9 \\0 \\0\end{pmatrix},\begin{pmatrix}1 \\3 \\4\end{pmatrix}\right>\begin{pmatrix}1 \\3 \\4\end{pmatrix} = \begin{pmatrix}10 \\29 \\38\end{pmatrix}\end{align}

Is this the right method to compute this?


4 Answers 4


That would be the correct method...if $v_1$ and $v_2$ were orthogonal and unit length. Unfortunately, they're not.

Three alternatives:

  1. Compute $w = v_1 \times v_2$, and the projection of $v$ onto $w$ -- call it $q$. Then compute $v - q$, which will be the desired projection.

  2. Orthgonalize $v_1$ and $v_2$ using the gram-schmidt process, and then apply your method.

  3. Write $q = av_1 + bv_2$ as the proposed projection vector. You then want $v - q$ to the orthogonal to both $v_1$ and $v_2$. This gives you two equations in the unknowns $a$ adn $b$, which you can solve.



You have to construct by the Gram Schmidt procedure an orthonormal basis $(e_1,e_2)$ from the given basis of $V$ and then $$P_v(V)=\langle v,e_1\rangle e_1+\langle v,e_2\rangle e_2$$


Since codimension of $V$ is one, in this case it is probably easier to calculate projection onto $V^\bot$ first.

In this specific case you get $V^\bot = \operatorname{span}\{(2,-2,1)^T\}$. To calculate projection onto one-dimensional subspace space, you can simply take unit vector $u$ generating this subspace and then and calculate $\langle \vec v,\vec u \rangle \vec u$. In this case you get $\vec u = \frac13 (2,-2,1)^T$, $\langle \vec v,\vec u\rangle = 6$ and he projection onto $V^\bot$ is $$\vec q=\langle \vec v,\vec u \rangle \vec u = (4,-4,2)^T.$$

Projection onto $V$ is $$\vec p = \vec v - \vec q = (5,4,-2)^T.$$

In the above $(a,b,c)^T$ denotes transpose. (So that I can write vectors in a row. Based on the question it seems that you are using notation for vectors as columns.)



1) Compute an orthonormal base $v_1,v_2$ of $V$ using Gram-Schmidt

2) Consider the projector $p_V(x) = \sum_{i=1}^2 \langle v_i,x\rangle v_i$

3) Compute $p_V(v)$

So what you did is wrong because $\begin{pmatrix}\frac{1}{3} \\\frac{2}{3} \\\frac{2}{3}\end{pmatrix}$,$\begin{pmatrix}1 \\3 \\4\end{pmatrix}$ are not orthonormal vectors.


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.