Along the lines of my lines of my previous question about irrational angles "$45^\circ$ Rubik's Cube: proving $\arccos ( \frac{\sqrt{2}}{2} - \frac{1}{4} )$ is an irrational angle?", I was working on a puzzle and I ran into an interesting question about an irrational angle.

Take a puzzle made out of a triangular bipyramid that turns on its vertices: constellation 6 puzzle

The piece in the center must be a circle because it gets rotated by an irrational amount after some combination of moves. Here is the path a point takes after repeatedly turning the purple axis followed by the green axis 90 degrees: point path after 90L 90R

The rotation matrix for this operation is $$\left( \begin{array}{ccc} \frac{1}{8} & \frac{3}{4} & \frac{3}{8}\sqrt{3} \\ \frac{-3}{4} & \frac{1}{2} & \frac{-1}{4}\sqrt{3} \\ \frac{-3}{8}\sqrt{3} & \frac{-1}{4}\sqrt{3} & \frac{5}{8} \end{array} \right)$$

From there you can find the axis that points are rotating around is $[0, \sqrt{3}/2, -1]$ And the angle $\theta$ they're rotating through is $\arccos{\frac{1}{8}}$.

$\arccos{\frac{1}{8}}$ is approximately $82.819244218541^{\circ}$ and it's an irrational angle because the only rational values that correspond to rational angles for $\cos{\theta}$ are $\{-1, \frac{-1}{2}, 0, \frac{1}{2}, 1\}$.

So what do we know about $\theta = \arccos{\frac{1}{8}}$? It's irrational angle but is it an algebraic angle? In general if $\cos \theta$ is rational does that tell us anything (beyond rationality) about $\theta$?

  • $\begingroup$ "because the only rational values that correspond to rational angles for $\cos\theta$ are ..." this is completely false. The rationals are dense in $[-1,1]$, which is the image of $\cos$. $\endgroup$
    – hjhjhj57
    Aug 3, 2015 at 2:31
  • 2
    $\begingroup$ @hjhjhj57 There are plenty of angles that have rational cosines, but they are not rational multiples of $\pi$. $\endgroup$ Aug 3, 2015 at 2:33
  • $\begingroup$ @RobertIsrael agreed, but that's not what the OP is saying. $\endgroup$
    – hjhjhj57
    Aug 3, 2015 at 2:34
  • 3
    $\begingroup$ I think that is exactly what the OP means by "the only rational values that correspond to rational angles". $\endgroup$ Aug 3, 2015 at 2:37
  • 1
    $\begingroup$ OP actually means $\cos(\mathbb{Q}) \cap \mathbb{Q} = \{-1,\frac{-1}{2},0,\frac{1}{2},1\}$ if you reads carefully. $\endgroup$
    – Empiricist
    Aug 3, 2015 at 2:42

1 Answer 1


The Gelfond-Schneider theorem says if $a\notin \{0,1\}$ is algebraic and $b$ is algebraic and irrational, then $a^b$ is transcendental. Apply this to $a = e^{i\pi/180}$, and you find that if $b$ is an irrational algebraic, $\omega = \cos(b^\circ) + i \sin(b^\circ)$ is transcendental. Then $\cos(b^\circ) = (\omega + \omega^{-1})/2$ and $\sin(b^\circ) = (\omega - \omega^{-1})/(2i)$ are also transcendental.

  • $\begingroup$ Your answer is a bit over my head so I'm going to have to do more reading before I understand it. Thank you, it's going to take me a while to accept though :-) $\endgroup$ Aug 3, 2015 at 3:38
  • $\begingroup$ Alright I think I follow the first part. You choose $a = e^{i \pi / 180}$ because it corresponds to $\cos({\frac{\pi}{180}}) + i\sin({\frac{\pi}{180}})$ and $\frac{\pi}{180}$ is a rational angle so $a$ would have to be algebraic. Then $a^b$ would be $e^{i\pi b / 180}$ which treats $b$ as irrational degrees and establishes the transcendence of $\cos(b^{\circ}) + i\sin(b^{\circ})$. I don't follow how you eliminated the $i\sin(b^{\circ})$ term though. How does raising $\omega$ to the $-1$ power do that? $\endgroup$ Aug 3, 2015 at 3:57
  • $\begingroup$ @BrandonEnright: $(\cos(x) + i \sin(x))(\cos(x) - i \sin(x)) = \cos^2(x) + \sin^2(x) = 1$. $\endgroup$ Aug 3, 2015 at 4:35
  • $\begingroup$ Got it. You multiplied the numerator and denominator of $\omega^{-1}$ by $\cos(b^{\circ}) - i\sin(b^{\circ})$ which eliminates the denominator and leaves $\cos(b^{\circ}) - i\sin(b^{\circ})$ in the numerator and then the $i\sin(b^{\circ})$ terms cancel leaving $2\cos(b^{\circ})$ which you fix by dividing by 2. Whew I understand you answer now! $\endgroup$ Aug 3, 2015 at 4:57

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.