Euler's solution to the Basel problem

  • Context: Graduate 
  • Thread starter Thread starter hover
  • Start date Start date
Join the discussion
Registration is free. Ask a follow-up in this thread, or start your own.
1 reply · 5K views
hover
Messages
342
Reaction score
0
Can someone find a good explanation of how Euler did it? I can't seem to find anything article or whatnot that carefully explains what he did to solve the problem. From what I can gather, he seems to follow a method of writing out sin(x) as an infinite series(taylor polynomial) and divides by x and from there I get lost in the mishmash. Why does he even choose to use sin(x) and not cos(x) or tan(x) for that matter?
 
Physics news on Phys.org
The function

sin(pi x)/(pi x)

is zero at the nonzero integers, so it should be equal to:

sin(pi x)/(pi x) = product over n of (1-x^2/n^2)

The right hand side is a product that converges and it is zero when x isa nonzero integer. The normalization is correct, because for x = 1 it is 1 while the limit for x to 1 of the left hand side is also 1. So, the above indentity looks correct (but you can't rigorously prove that, Euler only conjectured the identity).

The coefficient of x^2 of the left hand side is:

-pi^2/6

And from the right hand side it is minus the sum of 1/n^2 from n = 1 to infinity. To get an x^2 term, you need to take it from one factor of the infinite product, say the nth, and then you need to take the 1 from all other factors. You then get -1/n^2, and all n from 1 to infinity contribute.

Now, you can just as well do this using cos(pi x). The zeroes are at x = (n+1/2), so you would conjecture that:

cos(pi x) = Product over n of [1-x^2/(n+1/2)^2]

Normalization is correct as can be seen from putting x = 1 on bith sides. Extracting the coefficient of x^2 gives:

-pi^2/2 = -Sum over n from n= 0 to infinity of 1/(n+1/2)^2

You can rewrite this as:

sum over n 1/(2n+1)^2 = pi^2/8

You then use the following trick. If we put:

Zeta(2) = sum from n = 1 to infinity of 1/n^2

Then clearly the sum of the inverse squares of the even numbers only is:

sum from n = 1 to infinity of 1/(2n)^2 = 1/4 Zeta(2)

So, the sum over only the inverse squares if the odd numbers must be
3/4 Zeta(2). So we have:

pi^2/8 = 3/4 Zeta(2) -------->

Zeta(2) = pi^2/6.
 
  • Like
Likes   Reactions: Sergejs Boroviks