1
$\begingroup$

In this post, the author claims that given a tetrahedron $ABCD$ with known base ABC (i.e. the side lengths $AB$, $BC$, $CA$ are known), and also known angles $\angle ADB, \angle ADC, \angle BDC $, that they can solve for the dihedral angles between the planes of $\triangle ADB$ and $\triangle ADC$, and $\triangle BDC$.


My question is: How is this possible? How is it possible to solve for the dihedral angles, without first solving for the side lengths $AD$, $BD$, $CD$?


As suggested in a solution added to the above post, to solve the tetrahedron, I would assign coordinates to the known base triangle $\triangle ABC$. Then I would define the unknown quantities $x_1, x_2, x_3$ where $x_1 = DA , x_2 = DB, x_3 = DC $ and apply the law of cosines to the various lateral triangles, which would result in three quadratic equations in $x_1, x_2, x_3$. These equations take the form

$ \dfrac{x_1^2 + x_2^2 - AB^2 }{2 x_1 x_2} = \cos \angle ADB $

$ \dfrac{x_1^2 + x_3^2 - AC^2 }{2 x_1 x_3} = \cos \angle ADC $

$ \dfrac{x_2^2 + x_3^2 - BC^2 }{2 x_2 x_3} = \cos \angle BDC $

These three equations can be solved numerically for $x_1, x_2, x_3$.

The next step is to find the coordinates of vertex $D$. So let $D = (d_1, d_2, d_3) $, then it follows that

$ (a_1 - d_1)^2 + (a_2 - d_2)^2 + (a_3 - d_3)^2 = x_1^2$

$ (b_1 - d_1)^2 + (b_2 - d_2)^2 + (b_3 - d_3)^2 = x_2^2$

$ (c_1 - d_1)^2 + (c_2 - d_2)^2 + (c_3 - d_3)^2 = x_3^2$

These equations are easy to solve due to their special structure, where taking the differences between the first and second equations and the first and the third equations, results in two planar equations whose intersection gives a line along which the point $(d_1, d_2, d_3)$ lies. Substituting in one of the original three equations, gives two possible coordinates for the vertex $D$.

Once the coordinates of $D$ are known, then we have solved the tetrahedron completely and all the sides and angles and dihedral angles can be computed in a straight forward manner.

$\endgroup$
2
  • $\begingroup$ We can think of it like this, the last point (the vertex of the tetrahedon) requires 3 coordinates to define it uniquely. Those three coordinates can be $(x,y,z)$. Another choice for the three coordinates can be the distances from any two points on the base and the height from the base or $(d_i,d_j,h)$. In the post, the three angles are used to characterize the vertex in order to uniquely determine the tetrahedron. I did have some trouble comprehending the law of sines used in the post. What are the trigonometric rules that apply to a tetrahedron? $\endgroup$
    – ananta
    Commented Jul 3, 2022 at 10:34
  • $\begingroup$ You did not answer my question. How can you solve for the dihedral angles given only the three side lengths of the base $ABC$ and the three angles that point $D$ makes with the vertices $A, B$ and $C$ ? $\endgroup$
    – disgraced
    Commented Jul 3, 2022 at 11:13

1 Answer 1

4
$\begingroup$

As it happens, the "face-angles" surrounding a tetrahedral vertex are all you need to know to determine the dihedral angles at that vertex (and vice versa). What happens at the vertex doesn't care about the opposite face.

The connection is made through what's called the "law of cosines for spherical geometry", although that terminology makes the results seem more exotic and scary than they really are. They follow from straightforward (if slightly tedious) Euclidean trig or vector methods.

Given face-angles $\alpha := \angle BDC$, $\beta := \angle CDA$, $\gamma := \angle ADB$, and writing simply $A$, $B$, $C$ for dihedral angles along edges $DA$, $DB$, $DC$, the relations are

$$\begin{align} \cos\alpha &= \cos\beta\cos\gamma+\sin\beta\sin\gamma\cos A \\ \cos\beta &= \cos\gamma\cos\alpha+\sin\gamma\sin\alpha\cos B \\ \cos\gamma &= \cos\alpha\cos\beta +\sin\alpha\sin\beta\cos C \end{align}$$

As a sanity check: Consider a corner of a room where two walls and meet the floor. The floor need not be square; let's say there's a face-angle $\alpha$ between the two wall-floor edges. We'll assume the walls themselves are upright in the sense that they meet along an edge that makes right face-angles with the wall-floor edges (so, $\beta=\gamma=90^\circ$). Substituting this information into the above relations gives

$$\begin{align} \cos\alpha &= \cos90^\circ\cos90^\circ+\sin90^\circ\sin90^\circ\cos A \;\quad\to\quad &\cos\alpha &= \cos A\\ \cos90^\circ&= \cos90^\circ\cos\alpha\;\;+\sin90^\circ\sin\alpha\;\;\,\cos B \;\quad\to\quad &0 &= \cos B\\ \cos90^\circ&= \;\;\cos\alpha\cos90^\circ+\;\sin\alpha\;\sin90^\circ\;\cos C \,\quad\to\quad &0&=\cos C \end{align}$$ These confirm that the walls are upright in another sense, making right dihedral angles with the floor: $B=C=90^\circ$. Moreover, the dihedral angle between the walls exactly matches the face-angle on the floor: $A=\alpha$.

These relations have counterparts that switch the roles of face- and dihedral angles (and change one sign, so be careful!): $$\begin{align} \cos A &= -\cos B\cos C+\sin B\sin C\cos \alpha \\ \cos B &= -\cos C\cos A+\sin C\sin A\cos \beta \\ \cos C &= -\cos A\cos B+\sin A\sin B\cos \gamma \end{align}$$ Note that substituting dihedral angle information from the corner of our room ($B=C=90^\circ$) would give us the face-angle information ($\beta=\gamma=90^\circ$, $\alpha=A$).

Importantly, these calculations require no knowledge of the shape of the rest of the room. In the tetrahedral context: edge-lengths are unnecessary; face-angles alone determine dihedral angles, and vice-versa.

(Of course, edge-lengths can help you find any missing face-angles you might need, but this isn't the scenario described in the question.)

$\endgroup$

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .