Gauss and Regular Polygons: Conclusion

2 comments

The Central Result

This is the concluding post in this series and we aim to prove the following result (proved in part by Gauss and finally the converse by Wantzel):
A regular polygon of n, n > 2 sides can be constructed by Euclidean tools if and only if \phi(n) = 2^{k}.

Here \phi(n) denotes the Euler's Totient function. It is clear that a regular polygon of n sides can be constructed if and only if \cos(2\pi / n) is constructible. To do so we find the minimal polynomial of \xi = 2\cos(2\pi / n). We have already established that the minimal polynomial of \zeta = \cos(2\pi / n) + i\sin(2\pi /n) is the n^{th} cyclotomic polynomial \Phi_{n}(z).

We also have \xi = \zeta + \zeta^{-1} and we need to note that \zeta^{-1} is a root of \Phi_{n}(z) = 0 so that the polynomial z^{\phi(n)}\Phi_{n}(1 / z) is the same as \Phi_{n}(z). Since n > 2 we see that \phi(n) is even and we can write \phi(n) = 2m.

It therefore follows we can express z^{-m}\Phi_{n}(z) can be expressed in the form (z^{m} + z^{-m}) + a_{1}(z^{m - 1} + z^{-(m - 1)}) + \cdots + a_{m - 1}(z + z^{-1}) + a_{m} where the a_{i} are integers. Since z^{k} + z^{-k} = (z + z^{-1})(z^{k - 1} + z^{-(k - 1)}) - (z^{k - 2} + z^{-(k - 2)}) we can express z^{-m}\Phi_{n}(z) as a polynomial in (z + z^{-1}) of degree m with integral coefficients only and moreover this polynomial will be monic. Let this polynomial be called \Psi_{m}(z). Then \Phi_{n}(z) = z^{m}\Psi_{m}(z + z^{-1}).

From this relation it follows that \Psi_{m}(z) is irreducible over rationals. For if it were not, we could write \Psi_{m}(z) = f(z)g(z) where f(z) is of degree r < m. Therefore \begin{align} \Phi_{n}(z) &= z^{m}\Psi_{m}(z + z^{-1})\notag\\ &= z^{m}f(z + z^{-1})g(z + z^{-1})\notag\\ &= \{z^{r}f(z + z^{-1})\}\{z^{m - r}g(z + z^{-1})\}\notag \end{align} which leads to a contradiction as \Phi_{n}(z) is irreducible.

It is thus established that \Psi_{m}(z) is the minimal polynomial of degree m = \phi(n) / 2 for \xi. And therefore \xi = 2\cos(2\pi / n) is constructible only if m = \phi(n) / 2 is a power of 2 i.e. when \phi(n) is a power of 2.

Therefore we have established the "only if" part of the main result. We now would like to establish the "if" part. To proceed we need to observe that \phi(n) = 2^{k} if and only if n = 2^{l}p_{1}p_{2}\ldots p_{m} where p_{i} are distinct Fermat primes. If n if of the form above then it is easily seen that \phi(n) is a power of 2. On the other hand if n has the prime factorization n = p_{0}^{k_{0}}p_{1}^{k_{1}}p_{2}^{k_{2}}\ldots p_{m}^{k_{m}} then \phi(n) = p_{0}^{k_{0} - 1}p_{1}^{k_{1} - 1}\ldots p_{m}^{k_{m} - 1}(p_{0} - 1)(p_{1} - 1) \ldots (p_{m} - 1) and it is clear that \phi(n) will be a power of 2 only when one of the primes p_{i} is 2 and the exponents of other primes is 1 and for those primes (p_{i} - 1) is a power of 2. Thus we must have n in the desired form to have \phi(n) a power of 2.

In the expression of n we can obviously ignore the factor 2^{l} because as we have seen, if a polygon of n sides can be constructed by Euclidean tools then a polygon of 2n sides can also be so constructed. So we can assume that n is a product of distinct Fermat primes. We now prove the following result:

If a regular polygon of m sides and a regular polygon of n sides can be constructed by Euclidean tools, and m is coprime to n then a regular polygon of mn sides can also be constructed using Euclidean tools.
Proof: Since m is coprime to n, we have integers a, b such that am + bn = 1 and then we get \begin{align} \frac{a}{n} + \frac{b}{m} &= \frac{1}{mn}\notag\\ \Rightarrow a \cdot \frac{2\pi}{n} + b \cdot \frac{2\pi}{m} &= \frac{2\pi}{mn}\notag \end{align} Since we can construct angles 2\pi / m and 2\pi / n using Euclidean tools, we can therefore construct angle 2\pi / (mn) in this fashion and therefore a regular polygon of mn sides is constructible using Euclidean tools only.

In the previous post we had established that if p is a Fermat prime then a regular polygon of p sides is constructible using Euclidean tools. It now follows that if n is a product of distinct Fermat primes then a regular polygon of n sides is constructible using Euclidean tools.

Therefore we have established the main result concerning the construction of regular polygons. Its now time to show the construction of a regular polygon of 17 sides.

Construction of Regular Polygon of 17 Sides

We will solve the equation \Phi_{17}(z) = 1 + z + z^{2} + \cdots + z^{16} = 0 using the concept of Gaussian periods. Let \alpha = 2\pi / 17 so that \zeta = \cos\alpha + i\sin\alpha is a root of the above cyclotomic equation. Since 3 is a primitive root of 17, we arrange \zeta^{k} in the form \zeta_{m} = \zeta^{3^{m}}. The correspondence between m and k is as follows: \begin{array}{|c|c|c|c|c|c|c|c|c|c|c|c|c|c|c|c|c|} \hline\\ m & 0 & 1 & 2 & 3 & 4 & 5 & 6 & 7 & 8 & 9 & 10 & 11 & 12 & 13 & 14 & 15\\ \hline\\ k & 1 & 3 & 9 & 10 & 13 & 5 & 15 & 11 & 16 & 14 & 8 & 7 & 4 & 12 & 2 & 6\\ \hline \end{array} The periods of 8 terms are given by \begin{align} x_{1} &= \sum_{m\,\, \text{even}}\zeta_{m} = \zeta^{1} + \zeta^{9} + \zeta^{13} + \zeta^{15} + \zeta^{16} + \zeta^{8} + \zeta^{4} + \zeta^{2}\notag\\ x_{2} &= \sum_{m\,\, \text{odd}}\zeta_{m} = \zeta^{3} + \zeta^{10} + \zeta^{5} + \zeta^{11} + \zeta^{14} + \zeta^{7} + \zeta^{12} + \zeta^{6}\notag \end{align} and periods of 4 terms are given by \begin{align} y_{1} &= \sum_{m \equiv 0\pmod 4}\zeta_{m} = \zeta^{1} + \zeta^{13} + \zeta^{16} + \zeta^{4}\notag\\ y_{2} &= \sum_{m \equiv 2 \pmod 4}\zeta_{m} = \zeta^{9} + \zeta^{15} + \zeta^{8} + \zeta^{2}\notag\\ y_{3} &= \sum_{m \equiv 1\pmod 4}\zeta_{m} = \zeta^{3} + \zeta^{5} + \zeta^{14} + \zeta^{12}\notag\\ y_{4} &= \sum_{m \equiv 3 \pmod 4}\zeta_{m} = \zeta^{10} + \zeta^{11} + \zeta^{7} + \zeta^{6}\notag \end{align} Since \zeta^{k} + \zeta^{17 - k} = 2\cos k\alpha, we get \begin{align} x_{1} &= 2(\cos\alpha + \cos 8\alpha + \cos 4\alpha + \cos 2\alpha)\notag\\ x_{2} &= 2(\cos 3\alpha + \cos 7\alpha + \cos 5\alpha + \cos 6\alpha)\notag\\ y_{1} &= 2(\cos\alpha + \cos 4\alpha)\notag\\ y_{2} &= 2(\cos 8\alpha + \cos 2\alpha)\notag\\ y_{3} &= 2(\cos 3\alpha + \cos 5\alpha)\notag\\ y_{4} &= 2(\cos 7\alpha + \cos 6\alpha)\notag \end{align} Clearly x_{1} + x_{2} = -1 (sum of roots) and if we multiply we will find that x_{1}x_{2} = 4(x_{1} + x_{2}) = -4 so that x_{1}, x_{2} are roots of the equation x^{2} + x - 4 = 0\tag{1} where x_{1} > x_{2} (check this yourself). Again by multiplying we get y_{1}y_{2} = -1 and since y_{1} + y_{2} = x_{1}, so that y_{1}, y_{2} are roots of y^{2} - x_{1}y - 1 = 0\tag{2} where y_{1} > y_{2}. Similarly y_{3}, y_{4} are roots of y^{2} - x_{2}y - 1= 0\tag{3} and y_{3} > y_{4}.

Two periods of 2 terms are z_{1} = 2\cos\alpha and z_{2} = 2\cos 4\alpha and then we have z_{1} + z_{2} = y_{1},\,\, z_{1}z_{2} = y_{3} so that z_{1}, z_{2} are roots of equation z^{2} - y_{1}z + y_{3} = 0\tag{4} and z_{1} > z_{2}.

Solving the above equations and taking care of the inequalities involved we can see that z_{1} is given by the somewhat cumbersome expression \begin{align} &2\cos\left(\frac{2\pi}{17}\right)\notag\\ &\,\,\,\,= \frac{-1 + \sqrt{17} + \sqrt{34 - 2 \sqrt{17}} + 2\sqrt{17 + 3 \sqrt{17} - \sqrt{34 - 2 \sqrt{17}} - 2\sqrt{34 + 2 \sqrt{17}}}}{8}\notag \end{align} We should also note that the inequalities which differentiate the above roots are obtained using non-algebraic means. Therefore obtaining an expression of 2 \cos (2 \pi / 17) is not entirely an algebraic problem. In fact, from the view point of algebra, all the roots of \Psi_{8}(z) = 0 are interchangeable.

We now show the construction of a regular polygon of 17 sides graphically and then explain the steps.
Regular 17-gon
Fig 1. Construction of Regular 17-gon

We start with a circle and two perpendicular radii OA and OB. Then a point C is found on OB such that OC = OB / 4. Another point D on OA is found such that \angle OCA = 4 \angle OCD. A point E is located on OA extended towards O such that \angle DCE = \pi / 4. A semicircle with AE as diameter is drawn which intersects OB in F. A circle is drawn with center D and radius DF which intersects OA in G and H. Perpendiculars to OA are drawn from points G and H which meet the initial circle (with center O and radius OA) at points P_{3} and P_{5}. Bisecting the arc P_{3}P_{5} we get P_{4}. Now P_{3}P_{4} is the side of the desired regular polygon and the construction can be completed easily by cutting arcs of that length starting from point A = P_{0}.

The reason the above construction works is not difficult to explain. Let us then assume that \beta is an acute angle such that \tan 4\beta = 4, so that 4\cot 4\beta = 1. Thus equation (1) for x_{1}, x_{2} can be written as x^{2} + 4x\cot 4\beta - 4 = 0 so that the roots are x_{1} = 2\tan 2\beta and x_{2} = - 2\cot 2\beta. Putting these in equations (2) and (3) we get \begin{align} y_{1} &= \tan(\beta + \pi / 4)\notag\\ y_{2} &= \tan(\beta - \pi / 4)\notag\\ y_{3} &= \tan\beta\notag\\ y_{4} &= - \cot\beta\notag \end{align} and therefore we have \begin{align} 2\cos 3\alpha + 2\cos 5\alpha &= y_{3} = \tan\beta\notag\\ 2\cos 3\alpha \cdot 2\cos 5\alpha &= 2\cos 2\alpha + 2\cos 8\alpha\notag\\ &= y_{2} = \tan(\beta - \pi / 4)\notag \end{align} In the given construction we have \angle OCA = 4\beta, \angle OCD = \beta and \angle OCE = \pi / 4 - \beta. Let us assume OA = 1 so that we have \begin{align} 2\cos \angle AOP_{3} + 2\cos \angle AOP_{5} &= 2 (OG - OH)\notag\\ &= 2(OD + DG) - 2(DH - OD)\notag\\ &= 4OD\text{ (since } DH = DG)\notag\\ &= OD / (1 / 4) = OD / OC\notag\\ &= \tan\beta\notag \end{align} and \begin{align} 2\cos \angle AOP_{3} \cdot 2\cos \angle AOP_{5} &= -4 OG \cdot OH\notag\\ &= -4 OF^{2}\notag\\ &= -4 OE \cdot OA\notag\\ &= -OE / (1 / 4) = -OE/OC\notag\\ &= \tan(\beta - \pi / 4)\notag \end{align} It follows that \angle AOP_{3} = 3\alpha and \angle AOP_{5} = 5\alpha.

Note: The construction of regular 17-gon as well as its proof is taken from "An Introduction to the Theory of Numbers" by G. H. Hardy and E. M. Wright. The construction is originally by Richmond and till date is the simplest one available. See MathWorld page for details. Also the wikipedia entry for heptadecagon gives another construction (steps are shown in an animated GIF figure), but it has relatively greater number of steps.

Print/PDF Version

2 comments :: Gauss and Regular Polygons: Conclusion

Post a Comment

  1. Interesting you didnt mention the term “Heptadecagon”. http://en.wikipedia.org/wiki/Heptadecagon has a step by step animation of how you can actually draw the Heptadecagon.

    Here is another interesting link: http://mathworld.wolfram.com/GeometricConstruction.html

  2. SD, the post has been updated and references to Wikipedia and MathWorld have been provided. Thanks for providing the references.