Let $T$ be the triangle in $\mathbb{R}^2$ with vertices $(0,0)$, $(1,0)$, $(\frac12, \frac12)$. Find the area of the inverse image of $T$ under the orthogonal projection $\phi(x,y,z)=(x,y)$.

So I find the following:

$\psi(0,0) = (0,0,1)$, $\psi(1,0) = (1,0,0)$, $\psi(\frac12,\frac12) = (\frac12,\frac12,\frac1{\sqrt2})$

Then the area is $\frac12\cdot$(length of one side)$\cdot$(length of another side)$\cdot\sin\theta$ and $\theta = \cos^{-1}\left(\frac{\text{dot product of two sides}}{\text{norms of two sides}}\right)$. However, I get some mess with lots of square roots.

What am I doing wrong?

  • $\begingroup$ There’s an entire line that gets projected onto the origin by $\phi$, another line that gets projected onto $(1,0)$ and yet another that is projected onto $\left(\frac12,\frac12\right)$. How did you decide on the particular inverse images that you’re using? $\endgroup$ – amd Jul 12 '17 at 17:24

Given $A(0,0,1);\;B(1,0,0);\;C (\frac12,\frac12,\frac1{\sqrt2})$

define two vectors $\vec u=B-A;\;\vec v=C-B$

we get $$\vec u=(1,0,-1);\;\vec v=(-\frac12,\frac12,\frac1{\sqrt2})$$ the area of the triangle is $$Area(\Delta)=\frac12||\vec u\times \vec v||$$ the cross product can be computed by the determinant of this formal matrix $$\left( \begin{array}{ccc} \vec i & \vec j & \vec k \\ 1 & 0 & -1 \\ -\frac{1}{2} & \frac{1}{2} & \frac{1}{\sqrt{2}} \\ \end{array} \right)$$ giving $$\vec u\times \vec v= \frac{\vec i}{2}+\frac{\vec j}{2}-\frac{\vec j}{\sqrt{2}}+\frac{\vec k}{2}$$ where $\{\vec i,\vec j,\vec k\}$ is a base of $\mathbb{R}^3$.

In components $$\vec u\times \vec v= \left(\frac12,\frac12-\frac{1}{\sqrt 2},\frac12\right)$$ and $$Area(\Delta)=\frac12||\vec u\times \vec v||=\frac{1}{4} \sqrt{5-2 \sqrt{2}}\approx 0.3684$$


Erect vertical stalks on the three given points. Then any triangle $\triangle\subset{\mathbb R}^3$ with its vertices on the three stalks can be viewed as an inverse image of $T$ under $\phi$. Therefore you cannot talk about "the" inverse image of $T$.

Now any such $\triangle$ is lying in a plane $\Pi$ not parallel to the $z$-axis. If you want to know the area of $\triangle$ you have to know the angle $\theta$ between $\Pi$ and the $(x,y)$-plane, or what amounts to the same thing, the value $\cos\theta=\bigl|n_\Pi\cdot(0,0,1)\bigr|$, whereby $n_\Pi$ is the unit normal of $\Pi$. You then have $${\rm area}(\triangle)={1\over\cos\theta}\>{\rm area}(T)\ .$$

  • $\begingroup$ So can $\Pi$ be just any plane as long as it is not parallel to the z-axis? $\endgroup$ – sequence Jul 12 '17 at 8:36
  • $\begingroup$ @sequence: Given the first two lines of your question: Yes. If there are additional data not reveiled in your question things might be different. Note that even the $T$ could be such a $\triangle$, in which case $\theta=0$. $\endgroup$ – Christian Blatter Jul 12 '17 at 18:44

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.