Let $$L = (\operatorname{span}{(1,0,2,1),(2,1,2,3),(0,1,-2,1)})^\perp$$ What is the distance between the point $p = (1,0,0,1)$ and the subspace $L$?

Here is what I have done:

1) calculated the orthogonal base to be:

$$L = \operatorname{span}\{ (-2,2,1,0),(-1,-1,0,1)\}$$

2) calculated the orthonormal base to be:

$$L=\operatorname{span}\left\{\left(-\frac23, \frac23, \frac13, 0\right),\left(-\frac1{\sqrt3},-\frac1{\sqrt3},0,\frac1{\sqrt3}\right)\right\}$$

so it looks like the subspace is of $\dim L = 2$ and therefore a plain. my thoughts are that I need to figure out the projection of $p$ on to the plain spanned by $L$ and then calculate $\| p - \operatorname{proj}(p) \|$.

where I'm stuck is: I'm not sure if to calculate the projection of $p$ onto $L$ by using the orthonormal base or the orthogonal space. I get two different results.

I'm also not sure how to figure out this projection. I know that the projection of $p$ onto another vector $w$ is calculated by $$\frac{\langle p, w\rangle}{\langle w,w\rangle}$$ but onto a plain I'm not sure.

  • $\begingroup$ I have edited your question to make it readable. Please use Mathjax next time. $\endgroup$ – mechanodroid Mar 20 '18 at 13:06

You're almost there.

The ONB for $L$ is

$$\{a_1, a_2\} = \left\{\frac13 (-2,2,1,0), \frac1{\sqrt{3}}(-1,-1,0,1)\right\}$$ The orthogonal projection of $x \in \mathbb{R}^4$ onto $L$ is given by

$$Px = \langle x, a_1\rangle a_1 + \langle x, a_2\rangle a_2$$

In particular, for $x = (1,0,0,1)$ we get

$$Px = \frac19 (4,-4,-2,0)$$

The distance of $x$ to $L$ is given by $\|x - Px\|$.

You can also solve it by projecting onto $L^\perp$. To find the ONB for $L^\perp$, first note that $$(0,1,-2,1) = 2(1,0,2,1)-(2,1,2,3)$$

so in fact $L^\perp = \operatorname{span}\{(1,0,2,1), (2,1,2,3)\}$.

Now apply Gram-Schmidt:

$$e_1 = \frac1{\sqrt{6}}(1,0,2,1)$$ $$(2,1,2,3) - \frac1{6}\Big\langle (2,1,2,3), (1,0,2,1)\Big\rangle (1,0,2,1) = \frac12 (1,2,-2,3) \implies e_2 = \frac{1}{\sqrt{18}}(1,2,-2,3)$$

Thus $\{e_1, e_2\}$ is an ONB for $L^\perp$.

The orthogonal projection of $x \in \mathbb{R}^4$ onto $L^\perp$ can be calculated as $$Qx = \langle x, e_1\rangle e_1 + \langle x, e_2\rangle e_2$$

For $x = (1,0,0,1)$ we get

$$Qx = \frac19(5,4,2,9)$$

and then the distance of $x$ towards $L$ is $\|Qx\|$. To check the result is correct, verify that indeed $Px + Qx = x$.

  • $\begingroup$ Thank you very much for the very detailed explanation $\endgroup$ – eventhorizon02 Mar 20 '18 at 19:30

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.