4

What are the dihedral angles in a disphenoid with four identical triangles, each having one edge of length $2$ and two edges of length $\sqrt{3}$? Tried to look it up, but couldn't find it...

4 Answers4

3

Here's a hedronometric approach ...

Let $a$ and $d$ be the (necessarily opposite) edges of length $2$. These edges are orthogonal to each other in space; moreover, their projections into a mutually-parallel plane form the diagonals of a square of side-length $\sqrt{2}$. This square is a "pseudo-face" of the tetrahedron, and has area $H = 2$.

The other pairs of opposite edges ---say, $b$ & $e$ and $c$ & $f$--- determine the remaining two pseudo-face areas $J$ and $K$. Although the tetrahedron's symmetry dictates that these areas must be equal, the values are a little harder to intuit than $H$. However, the Sum of Squares formula tells us that $$W^2 + X^2 + Y^2 + Z^2 = H^2 + J^2 + K^2$$ where $W$, $X$, $Y$, $Z$ are the areas of the tetrahedron's (ordinary) faces. In this case, by Heron's Formula, $W = X = Y = Z = \sqrt{2}$. We deduce that $J = K = \sqrt{2}$.

Finally, if we assume that faces $Y$ and $Z$ meet along edge $a$, bounding dihedral angle $A$, and that $Z$ and $X$ meet along edge $b$, bounding dihedral angle $B$, then the tetrahedral Law of Cosines says $$\begin{align} H^2 &= Y^2 + Z^2 - 2 Y Z \cos A \qquad\to\qquad \cos A = \,\,0 \qquad\to\qquad A = \frac{\pi}{2} \\[4pt] J^2 &= Z^2 + X^2 - 2 Z X \cos B \qquad\to\qquad \cos B = \frac{1}{2} \qquad\to\qquad B = \frac{\pi}{3} \end{align}$$

Blue
  • 75,673
  • much of this is over my head. I especially find it strange that: areas (rather than lengths) can be used in the law of cosines... and you are using the areas of the rectangular prism that this tetrahedron resides in... I found this (http://math.stackexchange.com/questions/49330/the-dihedral-angles-of-a-tetrahedron-in-terms-of-its-edge-lengths) where you explain the law of cosines of tetrahedral in greater depth, I will need to review this further. – Jasand Pruski Aug 30 '15 at 03:35
  • 1
    @JasandPruski: "I especially find it strange that: areas [...] can be used in the law of cosines". I find it strange, too, but that's what makes it really interesting! :) Your observation about "the rectangular prism that this tetrahedron resides in" is insightful, but not quite accurate; instead, you want to consider the parallelepiped whose faces have the tetrahedron's edges as diagonals. The face areas of this parallelepiped match the $H$, $J$, $K$ above, although those are not my usual geometric interpretations of pseudo-faces. (Perhaps they should be. They're easier to describe! :) – Blue Aug 30 '15 at 09:06
  • This was also very helpful, the part about the parallelepiped... I wonder if the rhombic faces are the same as those on the rhombic dodecahedra... I will need to return your post to see if they are identical... inface it was the relationship between the internal angles reaching for the vertices on the rhombic dodecahedra that made me see the faces of this disphenoid... I will formally accept one of these answers one I am able to grasp them fully... I need more time for background research... I apologize for the delay. – Jasand Pruski Aug 30 '15 at 12:53
  • 1
    @JasandPruski: Take your time. By the way: I was inaccurate to say that your rectangular prism observation was inaccurate. :) In general, a tetrahedron's bounding parallelepiped is not a rectangular prism, but for your tetrahedron it is. (In all cases, though, the parallelepiped faces have areas $H$, $J$, $K$.) I need to stop commenting in the middle of the night ... :) – Blue Aug 30 '15 at 13:22
  • it's a rectangular prism because there are two dihedral angles of 90? I have no idea. I figure that you were refering to how the edges of a regular tetrahedra are the diagonals of a cube and because this is an irregular tetrahedra the edges must be diagonals of a rectangular prism... but then you said it's a parallelipiped so I realized that my simply assumign the rectangular prism would be right angles for an irregular tetrahedral was misguided and it must be a parallelepiped instead. My source on most of this is Amy Edmondson's "A Fuller Explaination", her book on synergetics. I'm remedial – Jasand Pruski Aug 30 '15 at 13:48
  • 1
    @JasandPruski: It's a rectangular prism because one pair of opposite faces consists of squares, and the segment joining the centers of those faces (and, therefore also, each segment joining corresponding corners of those faces) is perpendicular to them. The tetrahedron's dihedral angles are not a factor in that deduction. – Blue Aug 30 '15 at 14:00
  • https://en.wikipedia.org/wiki/Law_of_cosines#Analog_for_tetrahedra – Jasand Pruski Sep 07 '15 at 22:11
2

Say the four vertices of the shape are $ABCD$, with $\overline{AB} = \overline{CD} = 2$, and the rest of the segments having length $\sqrt{3}$. Let $O$ denote the midpoint of $AB$, then the triangle $AOC$ is a right triangle with hypotenuse $\sqrt{3}$ and one of the legs $1$. We get that $\overline{OC} = \sqrt{2}$. Similarly, $\overline{OD} = \sqrt{2}$. So the triangle $COD$ is an isosceles triangle with two sides of length $\sqrt{2}$ and the third side of length $2$, which implies that it is a right triangle(!).

So one of the dihedral angles is $90^{\circ}$. Now we need to compute the angle between two faces meeting on a side of length $\sqrt{3}$- for example, between $ABC$ and $DBC$. One way to do this is to assign coordinates to the vertices. From the previous analysis, a particularly nice choice is $A = (-1,0,0)$, $B = (1,0,0)$, $C = (0,\sqrt{2},0)$, $D = (0,0,\sqrt{2})$.

The plane containing $ABC$ is given by the equation $z = 0$, and the plane containing $DBC$ is given by $x\sqrt{2} + y + z = \sqrt{2}$. Then the angle between the planes is equal to the angle between the normal vectors $\mathbf{n}_1 = (0,0,1)$ and $\mathbf{n}_2 = (\sqrt{2},1,1)$:

$$\cos(\theta) = \frac{\mathbf{n}_1 \cdot \mathbf{n}_2}{|\mathbf{n}_1||\mathbf{n}_2|} = \frac{1}{1 \cdot 2} = \frac{1}{2}$$

It follows that the other dihedral angle is $60^{\circ}$.

Alex Zorn
  • 4,304
2

Let $\alpha$ be the vertex angle of a face, and $\beta$ a base angle. Then, by the Law of Cosines: $$\cos\alpha = \frac{-2^2 + \sqrt{3}^2 + \sqrt{3}^2}{2\cdot\sqrt{3}\sqrt{3}} = \frac{1}{3} \qquad \cos\beta = \frac{2^2 - \sqrt{3}^2 + \sqrt{3}^2}{2\cdot 2\sqrt{3}} = \frac{\sqrt{3}}{3}$$ Also, $$\sin\alpha = \frac{2\sqrt{2}}{3} \qquad \sin\beta = \frac{\sqrt{6}}{3}$$

Each vertex of your tetrahedron is surrounded by one $\alpha$ angle and two $\beta$ angles. Let $A$ be the dihedral angle opposite face angle $\alpha$ at a vertex; and let $B$ be opposite a $\beta$. By the Spherical Law of Cosines, we can calculate $$\begin{align} \cos A &= \frac{\cos\alpha-\cos^2\beta}{\sin^2\beta} = \frac{1/3-1/3}{2/3} = 0 \\[4pt] \cos B &= \frac{\cos\beta - \cos\alpha\cos\beta}{\sin\alpha\sin\beta} = \frac{\sqrt{3}/3\,(1-1/3)}{2\sqrt{12}/9} = \frac{1}{2} \end{align}$$

Therefore,

$$A = \frac{\pi}{2} \qquad\qquad B = \frac{\pi}{3}$$

Blue
  • 75,673
  • this seems like the most reasonable answer for me to understand... I am stuck at the "spherical law of cosines", with some more research this seems like it will be the answer I will accept... – Jasand Pruski Sep 04 '15 at 23:00
1

Here's a filthy coordinate-based approach — but one that can be done by just plugging through the coordinate geometry, without needing to remember any formulas substantially more complicated than Pythagoras. As with other answers, note that (by easy symmetry arguments, if nothing else) the edges of length 2 are orthogonal to each other; this means that projecting along the axis mutually orthogonal to both of them, we can use them as axes for the orthogonal plane — in other words (choosing the $Z$ axis as our projection axis and putting one of these edges on the $XY$ plane), the vertices of the tetrahedron are $(\pm1, 0, 0)$ and $(0, \pm1, C)$ for some $C$. Now, Pythagoras gives us $C$: $(\pm1)^2+(\pm1)^2+C^2=(\sqrt3)^2$, or in other words $C=1$. This means the four vertices of the disphenoid can be placed at $(1,0,0)$, $(-1,0,0)$, $(0,1,1)$ and $(0,-1,1)$.

Now, for instance, we can look at one of the faces — e.g., the one with verts $A,B = (\pm1,0,0)$ and $C=(0,1,1)$. Two vectors within this face are $(2,0,0)$ (representing the edge $AB$) and $(1,1,1)$ (representing the edge $BC$), so an orthogonal vector to this face is $(2,0,0)\times(1,1,1)=(0,-2,2)$ and a normal is $\hat{n}_{ABC}=(0, -\frac12\sqrt2,\frac12\sqrt2)$. Similarly, a normal to the face $ABD$ (where $D=(0,-1,1)$) is $\hat{n}_{ABD}=(0, \frac12\sqrt2,\frac12\sqrt2)$, and so the dihedral angle across the edge $AB$ is $\arccos(\hat{n}_{ABC}\cdot\hat{n}_{ABD})=\arccos(0)=\frac\pi2$. Obviously the dihedral across the edge $CD$ is the same, and the dihedrals across any of the 'vertical' edges can be gotten in similar fashion, after finding e.g. a normal to the face $ACD$.

  • I understand paragraph 1, but not the 2nd paragraph. Your coordinates are most helpful. Using law of cosines on the midpoint on the edges I can get one of the dihedral angles to be pi/2, but the other comes out to 62.9643082106 [arcos(5/11)]. Midpoint on (1,0,0) & (0,1,1) is (0.5,0.5,0.5), length from previously mentioned midpoint to (−1,0,0) & (0,−1,1) is sqrt(2.75) in both cases, side opposite to midpoint is length sqrt(3). Calculating angle on midpoint opposite sqrt(3): 3=2(2.75) - 2(2.75)cosX. Where have I gone wrong, since the other answers here and Wikipedia all say it's pi/3 and not ~63 – Jasand Pruski Sep 06 '15 at 15:42
  • so it seems my problem is that becase it's an issocales triangle, that method (law of cos done on midpoint of edge) only works when I have bilateral symmetry, so it only works to figure out the dihedral angle of pi/2... but for the other one... arg... looks like I have to deal with your "normals" after all – Jasand Pruski Sep 06 '15 at 15:59