Treasures buried by pi in "the" Euler identity

An enlightening discussion about pi and tau.

Treasures buried by pi in "the" Euler identity

by Kodegadulo » Tue Jul 10, 2012 3:01 pm

There's even more to "Euler Identities" than meets the ... eye. ;) The Tau Manifesto made the point about how silly it is that there is all this hype over the equation

[tex]e^{i\pi} + 1 = 0[/tex]

otherwise known as "the" Euler Identity (so-called, as if there's only one). People marvel over how it combines the "five most important numbers in mathematics" and the operations of "exponentiation, multiplication, and addition" in one equation -- yet they do so without gaining any new insight or understanding of what the equation means. It amounts to nothing more than a piece of curious pop numerological mumbo-jumbo. But if we can just peel off the obscurity caused by [tex]\pi[/tex] there are interesting treasures we can dig up.

First off, looking at this "Euler Identity" is far less edifying than looking at what it's derived from, the Euler formula:

[tex]e^{i\theta} = \cos \theta + i \sin \theta[/tex]

This formula reveals that complex exponentiation is equivalent to a rotation around the unit circle, mapping an angle [tex]\theta[/tex] of polar coordinates to real and imaginary rectilinear coordinates via the circle functions sin and cos. In that light, plugging in interesting values for [tex]\theta[/tex] can be instructive:

[tex]\begin{tabular}{lll}
\underline{\operatorname{Identity}} & & \underline{\operatorname{Meaning}} \\
e^{i\tau} = 1 & & \operatorname{A rotation of a full turn is unity.} \\
e^{i\tau/2} = -1 & & \operatorname{A rotation of a half turn is negation. } \\
e^{i\tau/4} = i & & \operatorname{A rotation of a quarter turn is perpendicular.} \\
e^{i\tau \cdot k} = 1^k = 1 & & \operatorname{Any integer number of whole turns is unity.} \\
\end{tabular}[/tex]

I find each of these equations interesting and revealing. But for some reason the second of these is considered "ugly" because somehow division by two and negation are "inelegant" operations. But why should division and negation be disparaged as operations? Aren't they nearly as fundamental as multiplication and addition, and aren't they just as vital to mathematics? And why should 2 be disparaged? It's a rather important number in and of itself, being the first prime number and the only even prime. Yet it supposedly makes the equation more "beautiful" and "elegant" if we perform a bit of algebraic sleight-of-hand, burying the half in a [tex]\pi[/tex], and making a negative appear as if it's a positive. Hiding the dirty laundry as it were. But the only thing we accomplish in doing that is to mask the true importance of this identity:

What is the significance of a division when it appears within an exponentiation? In other words, what is the meaning of [tex]z^{1/n}[/tex]? The answer is:

[tex]z^{1/n} = \sqrt[n]{z}[/tex]

That means dividing an exponent by [tex]n[/tex] is the same as taking the [tex]n[/tex]-th root. But the Euler formula reveals that a complex exponentiation is equivalent to a rotation around the origin in the complex plane. So taking an [tex]n[/tex]-th root of a complex number is equivalent to dividing its rotation by [tex]n[/tex]. What are the consequences of that?

If we start with a full circle of rotation

[tex]e^{i\tau} = 1[/tex]

and divide the rotation in half we get

[tex]e^{i\tau\cdot 1/2} = -1 = 1^{1/2} = \sqrt[2]{1}[/tex]

In other words, this reveals that the square root of unity is negation. Or rather, the first square root of unity is negation. Because in fact if we take any number of whole turns

[tex]e^{i\tau \cdot k} = 1[/tex]

and divide their rotations in half

[tex]e^{i\tau \cdot k/2} = 1^{1/2}[/tex]

we see that the square roots of unity must include both

[tex]\begin{tabular}{lllll}
e^{i\tau \cdot 1/2} = -1 & & \operatorname{and} & & e^{i\tau \cdot 2/2} = 1 \\
\end{tabular}[/tex]

We can confirm that these are square roots by squaring them to get 1:

[tex]\begin{tabular}{lllll}
\left(e^{i\tau \cdot 1/2}\right)^2 = \left(-1\right)^2 = 1 & & \operatorname{and} & & \left(e^{i\tau \cdot 2/2}\right)^2 = \left(1\right)^2 = 1 \\
\end{tabular}[/tex]

But this means there is a very interesting bit of mathematical treasure buried unnoticed in the formula

[tex]e^{i\pi} + 1 = 0[/tex]

because we can substitute equivalent terms to yield

[tex]e^{i\tau \cdot 1/2} + e^{i\tau \cdot 2/2} = 0[/tex]

in other words, the sum of the square roots of unity (the multiplicative identity) is zero (the additive identity). Let's plot that on a unit circle:

Image

That makes sense: The negative unity counterbalances the positive unity, to average out to zero.

Hmm. What about the cube roots of unity? What would those be? And do they also add up to zero? Dividing up whole turns by 3 we get:

[tex]e^{i\tau \cdot k/3} = 1^{1/3}[/tex]

[tex]\begin{tabular}{lllll}
e^{i\tau \cdot 1/3} = -\frac{1}{2} + \frac{\sqrt{3}}{2} i & & e^{i\tau \cdot 2/3} = -\frac{1}{2} - \frac{\sqrt{3}}{2} i & & e^{i\tau \cdot 3/3} = 1\\
\end{tabular}[/tex]

Let's plot them:

Image

Does this look familiar? That's right, those are the vertices of the equilateral triangle.

Let's confirm they're actually cube roots:

[tex]\left(e^{i\tau\cdot 1/3}\left)^3 = \left(-\frac{1}{2} + \frac{\sqrt{3}}{2}i\right)^3 = \left(-\frac{1}{2}\right)^3 + 3\left(-\frac{1}{2}\right)^2\left(\frac{\sqrt{3}}{2}\right) + 3\left(-\frac{1}{2}\right)\left(\frac{\sqrt{3}}{2}\right)^2 + \left(\frac{\sqrt{3}}{2}\right)^3 = -\frac{1}{8} + \frac{3\sqrt{3}}{2} i + \frac{9}{8} - \frac{3\sqrt{3}}{2}i = \frac{8}{8} = 1[/tex]

[tex]\left(e^{i\tau\cdot 2/3}\left)^3 = \left(-\frac{1}{2} - \frac{\sqrt{3}}{2}i\right)^3 = \left(-\frac{1}{2}\right)^3 + 3\left(-\frac{1}{2}\right)^2\left(-\frac{\sqrt{3}}{2}\right) + 3\left(-\frac{1}{2}\right)\left(-\frac{\sqr{3}}{2}\right)^2 + \left(-\frac{\sqrt{3}}{2}\right)^3 = -\frac{1}{8} - \frac{3\sqrt{3}}{2} i + \frac{9}{8} + \frac{3\sqrt{3}}{2}i = \frac{8}{8} = 1[/tex]

[tex]\left(e^{i\tau\cdot 3/3}\left)^3 = \left(1\right)^3 = 1[/tex]

Yes, that works. And what do they add up to?

[tex]e^{i\tau\cdot 1/3} + e^{i\tau\cdot 2/3} + e^{i\tau\cdot 3/3} = \left(-\frac{1}{2} + \frac{\sqrt{3}}{2}i\right) + \left(-\frac{1}{2} - \frac{\sqrt{3}}{2}i\right) + \left(1\right) = \left(-\frac{1}{2} - \frac{1}{2}\right) + \left(\frac{\sqrt{3}}{2}i - \frac{\sqrt{3}}{2}i\right) + 1 = -1 + 0 + 1 = 0[/tex]

They do add up to zero! But that makes sense too, because in the diagram, the two vectors on the left are mirror images of each other vertically, so they must cancel each out each other's imaginary component. And they each contribute a negative half in their real components, so they cancel out the unity on the right, to yield zero.

How about the fourth roots, where we divide up whole turns by 4? These are a bit easier:

[tex]e^{i\tau \cdot k/4} = 1^{1/4}[/tex]

[tex]\begin{tabular}{lllllll}
e^{i\tau \cdot 1/4} = i & & e^{i\tau \cdot 2/4} = -1 & & e^{i\tau \cdot 3/4} = -i & & e^{i\tau \cdot 4/4} = 1\\
\end{tabular}[/tex]

Image

And here we have the vertices of the square (disguised as its alter-ego, the diamond). They certainly look like they'll cancel each other out.

Let's confirm they're actually fourth roots:

[tex]\left( e^{i\tau \cdot 1/4}\right)^4 = \left(i\right)^4 = \left(-1\right)^2 = 1[/tex]
[tex]\left( e^{i\tau \cdot 2/4}\right)^4 = \left(-1\right)^4 = \left(1\right)^2 = 1[/tex]
[tex]\left( e^{i\tau \cdot 3/4}\right)^4 = \left(-i\right)^4 = \left(-1\right)^2 = 1[/tex]
[tex]\left( e^{i\tau \cdot 4/4}\right)^4 = \left(1\right)^4 = \left(1\right)^2 = 1[/tex]

And do they add up to zero?

[tex]e^{i\tau \cdot 1/4} + e^{i\tau \cdot 2/4} + e^{i\tau \cdot 3/4} + e^{i\tau \cdot 4/4} = \left( i\right) + \left(-1\right) + \left(-i\right) + \left(1\right) = \left(i - i\right) + \left(-1 + 1\right) = 0 + 0 = 0[/tex]

They do, as expected.

Are you beginning to see a pattern here? For every natural number [tex]n \ge 2[/tex] there are [tex]n[/tex] complex roots of unity. The first such [tex]n[/tex]-th root is defined as

[tex]\zeta_n = e^{i\tau_n}[/tex] where [tex]\tau_n = \frac{\tau}{n}[/tex]

In other words, it's positioned on the unit circle at the [tex]n[/tex]-th of the "candidate circle constants", which I identified as "special angles" in an earlier post, i.e., the angle that is an [tex]n[/tex]-th of a full turn from unity. And the entire set of [tex]n[/tex]-th roots are defined as the powers of this first [tex]n[/tex]-th root, up to and including unity itself:

[tex]{\zeta_n}^k = e^{i\tau_n\cdot k}[/tex] for all [tex]k \in \left[1, n \right][/tex]

These roots are spaced around the unit circle at multiples of the special angle, each position being an [tex]n[/tex]-th part of [tex]k[/tex] whole turns from unity, and each corresponding to a vertex of the regular [tex]n[/tex]-gon plotted on the unit circle. And the sum of these [tex]n[/tex]-th roots is zero:

[tex]\sum_{k=1}^{n} {\zeta_n}^k = 0[/tex]

You can try it out with the fifth roots and the vertices of the pentagon, using a calculator for those multiples of [tex]72^{\circ}[/tex] angles. Or a little more easily with the sixth roots at those sextant angles. But you'll see it works out for every [tex]n[/tex] you try. It makes sense geometrically: The [tex]n[/tex]th roots of unity are unit vectors centered at the origin and distributed evenly around the circle, so they must counterbalance each other to add up to that center point. But we can prove this rigorously by noting that [tex]{\zeta_n}^k[/tex] constitutes a geometric series. The formula for the sum of a geometric series [tex]a^k[/tex] is given as:

[tex]\sum_{k=0}^{n-1} a^k = \frac{a^n - 1}{a - 1}[/tex]

So if we substitute [tex]a = \zeta_n[/tex] and note that [tex]{\zeta_n}^n = {\zeta_n}^0 = 1[/tex]:

[tex]\sum_{k=1}^{n} {\zeta_n}^k = \sum_{k=0}^{n-1} {\zeta_n}^k = \frac{{\zeta_n}^n - 1}{{\zeta_n} - 1} = \frac{1 - 1}{{\zeta_n} - 1} = 0[/tex] Q.E.D.

Bottom line, ask yourself: Would any of this be any clearer or more "beautiful" or "elegant", if it were cast in terms of [tex]\pi[/tex]? Wouldn't it be incrementally more ugly, and therefore incrementally more obscure, and therefore incrementally harder for students to grasp, if it were encrusted with [tex]2\pi[/tex] everywhere? If you saw a [tex]\pi/3[/tex] in the Euler formula, would you immediately grasp that it was the rotation for a 6th-root of unity, and not a cube root? If you saw [tex]\pi \cdot 2/7[/tex] and [tex]\pi \cdot 4/7[/tex], would you immediately grasp they these were the first and second of the 7th-roots of unity, and not the second and fourth? If you are a confirmed [tex]\pi[/tex]-ist, are you really willing to go through all those mental gymnastics, just for the sake of [tex]\pi[/tex]?

We are dealing with complex numbers plotted on the unit circle. Isn't this just confirmation that the most fundamental constant associated with circles, and with radians as the ideal angular measure to use with circle functions such as [tex]\sin \theta[/tex], [tex]\cos \theta[/tex], and [tex]e^{i \theta}[/tex], is the number identified as [tex]\tau[/tex]? And isn't [tex]\pi[/tex], at best, just one of many possible numbers derivable from [tex]\tau[/tex]?

And given all this insight we can derive, doesn't "the" Euler Identity [tex]e^{i\pi} + 1 = 0[/tex] become more beautiful, not less, when we re-cast it as [tex]e^{i\tau/2} = -1[/tex]?
Kodegadulo
Kindergarten
 
Posts: 6
Joined: Sun Jun 10, 2012 3:04 am

Re: Treasures buried by pi in "the" Euler identity

by josephlindenberg » Sat Jul 21, 2012 6:31 pm

Excellent post! I'm glad to see more people using the sum-of-the-nth-roots-of-unity argument for tau in Euler's Identity, especially when it's expressed this well.

By the way, there's a great way to appeal to all those numerologists fawning over the traditional form of Euler's Identity. Show them this conjugate form of Euler's Identity using tau:

[tex]0\, =\, 1\, +\, e^{-i\cdot{\frac\tau{2}}}[/tex]

All four basic arithmetic operations appear exactly once, and in their standard order (+,−,*,/)
0,1,2 appear in numerical order
e,i,tau appear in alphabetical order
PI is the SEMICIRCLE constant, not the circle constant . . . http://sites.google.com/site/taubeforeitwascool
josephlindenberg
Elementary School
Elementary School
 
Posts: 18
Joined: Wed Jul 06, 2011 2:34 am

Re: Treasures buried by pi in "the" Euler identity

by 1=0 » Fri Mar 24, 2017 11:56 am

Wonderful post! This even turns the Pi Manifesto's own argument on it because we get nice identities when using any fraction of tau. Sure there's a nice identity with τ/2, but there are also ones with τ/4 and τ/8 and any other fraction of tau, but that doesn't mean we should start worshipping them. The Pi Manifesto claims you get a nice identity for 3π, but in reality, it is the same as the identity for π, because π+τ=3π, and e^ix has a period of tau. In fact, the Euler identity for nπ depends on the parity of n, and we see many similar things with pi due to it being only half of the period of just about anything that has a period. The identity for nκ, where κ is is τ/α alternates between α different values, the αth roots of unity. The identity for tau is always e^(inτ)=1, no matter what integer n is equal to. That makes it more general than identities with any other constant. I'm always looking for generality, as it is one of the things that makes math beautiful, but using pi destroys it.
e^(iτ)=1 is also much more elegant than e^(iπ)+1=0 because it comes directly from Euler's formula, while in the pi version, 1 is arbitrarily added to each side for no reason other than to disguise the negative sign and make the equation look more elegant than it actually is. In reality both e^(iτ)=1 and e^(iπ)=-1 (the version without rearrangement) are elegant, and so are the examples with other angle constants, but the version with tau is the most elegant, because it represents a full revolution. However, the most elegant formula of all of them is the more general formula from which they are all derived, e^(iθ)=cosθ+i*sinθ.

Another thing that makes Euler's identity with tau more elegant than the pi version is that it actually matters. With pi, they show you this supposedly important equation...and then it never shows up anywhere in mathematics. I can't even think of a single place where e^iπ shows up without a two other than Euler's identity. The tau version shows up all the time though. It, of course, shows up in roots of unity, which only make sense if you know that e^iτ=1, showing another way pi makes things less intuitive. It also shows up in very important places like the Fourier transform and many, many more areas. Even e^τ without an i shows up in some places, including the Reimann zeta function. Meanwhile e^iπ is nowhere to be found because it's useless. Do πists really think an identity that shows up nowhere is more elegant than one that shows up throughout mathematics?

Since it is clear that e^(iτ)=1 is more elegant than e^(iπ)=-1, most piists will try to disguise the negative one by adding one to both sides, yielding e^(iπ)+1=0. However, even if I did support pi, I would find this version of Euler's identity incredibly inelegant, and, even before learning about tau, I always thought it was silly to rearrange Euler's identity like that, and I considered e^(iπ)=-1 to be the more useful and more elegant form. So why is it that piists will use a version of Euler's identity that is inconsistent with Euler's formula, has an unclear meaning, comes from rearranging the terms in another equation, moves the result off the unit circle, and is less simple? The only justification is that it unites "the five most important numbers in mathematics." This is a pretty ridiculous reason though, because, by changing the pi to τ/2 in the formula, you could unite the six most important numbers. However, piists will claim that somehow this version isn't better with tau because it's simpler with pi. Hold on just a second! They rearranged Euler's identity making it LESS SIMPLE, to get a one and a zero in it so they could unite more constants, but if we make it less simple by adding in another fundamental constant that's somehow bad? In order to be consistent with their own logic, those who say e^(iπ)+1=0 is better than e^(iτ)=1 must also say that e^(iτ/2)+1=0 is better than the pi version. There is simply no way ti justify the use of pi here, no matter which form you use.
However out of both forms, I would definitely prefer e^(iτ)=1, just like how I used to prefer e^(iπ)=-1 because the beauty of a formula comes not from how many constants we can shove in there, but from what the formula means. The rearranged formulas have no elegant meaning, while the non-rearranged ones tell us where we end up when we rotate the complex plane by pi or tau radians.
τ>π
1=0
Mathlete
Mathlete
 
Posts: 47
Joined: Sun Mar 19, 2017 11:01 am


Return to The Pi Manifesto



Who is online

Users browsing this forum: No registered users and 2 guests

cron
Fatal: Not able to open ./cache/data_global.php