1
$\begingroup$

Create a tetrahedron with corners $A=(1,2,4)$, $B = (1,0,2)$, $C = (2,1,3)$ and $D = (4,1,1)$. Determine the angle $\alpha$ between the edge $AB$ and the side $BCD$.

The first step I make is displacing the tetrahedron to place $B$ in origin. Thus we get the new coordinates $A' = (0,2,2)$, $B=(0,0,0)$, $C'=(1,1,1)$, $D'=(3,1,-1)$. Let $v_1$, $v_{2}$ and $v_3$ be the vectors from $B'$ to $C'$, $D'$ and $A'$ respectively.The crossproduct $u=v_1 \times v_2$ gives us the normal vector to the side $BCD$. Notice how the angle $\beta$ between $u$ and $v_3$ added with $\alpha$ gives a perpendicular angle (that is, $\pi/2$).

We can easily compute $\beta$: $$cos(\beta)=\frac{u \cdot v_3}{|u| |v_3|}$$

Here's the interesting part. The numerator is the triple product $v_1 \times v_2 \cdot v_3$, also the volume of the parallelepipedum formed by the vectors, is negative! More precisely, it is equal to $-\frac{1}{2\sqrt{3}}$. Notice how this yields a negative angle $\alpha$: $$\alpha + \beta = \pi/2 \leftrightarrow \alpha = \pi/2 - arccos(-\frac{1}{2\sqrt{3}}) = arcsin(-\frac{1}{2\sqrt{3}})$$ The book however says the angle should be $arcsin(\frac{1}{2\sqrt{3}})$. Have I reached the same result or how do I interpret it (geometrically) that the triple product is negative?

$\endgroup$
  • 1
    $\begingroup$ Have you checked that $v_1 \times v_2$ is indeed the normal vector to $(BCD)$ that points inside of the tetrahedron, and not outside ? $\endgroup$ – Dimitri Feb 3 '16 at 13:48
  • $\begingroup$ @Dimitri How can I do this without plotting? $\endgroup$ – Lozansky Feb 3 '16 at 15:37
  • $\begingroup$ Explicitly calculate the vector $u = v_1 \times v_2$ in cartesian coordinates. Then for instance, you can calculate the scalar product $u .BA$. Depending on the sign of this scalar product, you might have to take $v_2 \times v_1$ as a normal vector for your calculation to make sense, and it fixes your sign problem. Draw a little sketch to convince yourself. $\endgroup$ – Dimitri Feb 3 '16 at 15:43
  • $\begingroup$ @Dimitri The scalar product is -4. So I should reverse the order! I still don't understand geometrically why cross-product is anti-commutative but now I know how to fix the error. Thanks! $\endgroup$ – Lozansky Feb 3 '16 at 21:18
0
$\begingroup$

Since you do not know in advance if the points of the surface are order in clockwise or counterclockwise order when viewed from $AB$ in general. So it would be better to generalize the way one finds this angle, such that it gives the same result in both cases.

One way to this is by projecting $A$ on to the plane defined by $BCD$. I will define $\vec{v}$, $\vec{u}_1$ and $\vec{u}_2$ as the vector pointing from $B$ to $A$, from $B$ to $C$ and from $B$ to $D$ respectively. Then the projection of $A$ onto the plane can be found with

$$ \vec{n} = \vec{u}_1 \times \vec{u}_2, $$

$$ A_p = B + \vec{v} - \frac{\vec{v} \cdot \vec{n}}{\vec{n} \cdot \vec{n}} \vec{n}. $$

The angle $\angle ABA_p$ should then always be between 0° and 90°.

$\endgroup$

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.