Skip to main content

A silly little derivation of ζ(2)

(This is a cleaned-up and somewhat expanded version of this Twitter thread.)

What follows is a silly little proof that

ζ(2)=n=11n2=π26

where ζ is the Riemann zeta function.

Consider the integral

I:=01log(1x+x2)x(x1)dx.

We have, by using partial fractions and performing some other algebraic manipulations,

I=01log(1x+x2)xdx01log(1x+x2)1xdx=201log(1x+x2)x(x1x)=2(01log(1+x)xdx01log(1+x3)xdx)=4301log(1+x)xdx(xx1/3).

To evaluate this integral, we take the Maclaurin series:

01log(1+x)xdx=01n=1(1)nxn1ndx

Since for all positive integers N,

|n=1N(1)nxn1n|n=1Nxn1nn=1xn1n=log(1x)x

on [0,1) and

01log(1x)xdx<01/2log(1x)xdx21/21log(1x)dx<,

we can invoke the dominated convergence theorem to switch summation and limit processes. We then have

01log(1+x)xdx=n=101(x)n1ndx=n=1(1)n1n2=η(2)

where η(s) is the Dirichlet eta function.

Now, we have η(s)=21sζ(s); for s>1, this can be seen by separating and rearranging all even terms in the summation. Thus, we have

I=43η(2)=23ζ(2).

We look at another way to evaluate I. Noticing that 1x+x2=1x(1x), we can write the integrand as a power series in x(1x):

I=01log(1x(1x))x(1x)dx=01n=1xn1(1x)n1ndx=n=101xn1(1x)n1n

where we justify the switch of integral and sum by the monotone convergence theorem (since the summand is nonnegative on [0,1]). Recall the Euler Beta function given by

B(m,n)=01xm1(1x)n1dx=Γ(m)Γ(n)Γ(m+n).

In the case that m and n are integers, we get

B(m,n)=(m1)!(n1)!(m+n1)!

and in particular

01xn1(1x)n1ndx=B(n,n)n=((n1)!)2n(2n1)!=2n2(2nn).

Thus,

I=2n=11n2(2nn).

The next part may strike you as something I pulled out of nowhere. We invoke the identity

arcsin2x=12n=0(2x)2nn2(2nn),x[1,1],

a nice proof of which can be found by following the links starting here. (Okay, I'll admit it's not a very commonly-taught series, and the only reason I recognized it is that I used to spend too much time on AoPS. As such, I feel bad about blackboxing it like this. But it's cute!) We deduce that

I=4arcsin2(12)=π29.

Thus, (2/3)ζ(2)=π2/9, and so ζ(2)=π2/6. QED.

Comments

Popular posts from this blog

On infinite decimal expansions, missing numbers, and generating functions

(This post is a cleaned up and expanded version of this thread .) A cool fact I've seen shared around the internet   a few times : The decimal expansion of 1/998001 starts with 1998001=0.000001002003996997999 That is, it begins with three-digit strings from 000 to 999, in order, except that it skips 998 for some reason. The first thing to observe is that 998001=9992. Recall the formula for the infinite geometric series: n=0rn=11r. If we differentiate both sides with respect to r, we get n=1nrn1=1(1r)2, and multiplying by r gives n=1nrn=r(1r)2. (This can also be obtained by some series manipulations.) Now, take r=0.001. We have 0.001(10.001)2=1000998001=0.001+0.000002+0.000000003+ From here, the appearance of the numbers from 001 to \( 997...

100 is the only square that is the sum of 4 consecutive (positive) cubes.

The OEIS article on the number 100  opens with an interesting factoid:  "100 is the square of 10, and the smallest square that is the sum of four [positive] consecutive cubes: 13+23+33+43=100." In fact, it is the only one. To see this, let's look at the equation y2=x3+(x+1)3+(x+2)3+(x+3)3=4x3+18x2+42x+36. Let X=4x+6,Y=4y; the above equation then reduces to Y2=X3+60X. Note that a positive integer solution in (x,y) will give a positive integer solution in (X,Y), though the converse is not true. Now, generally speaking, whenever one sees an equation of the form Y2=X3+aX+b, one has an elliptic curve . Well, we have the extra condition 4a3+27b20 to get rid of problematic cases like y2=x3, which are referred to as singular curves ; we'll see the logic of this later on. I will not go too...