Sum over Q + \mathbb Q^+

Calculus Level 3

For a rational number x x that equals a b \frac ab in lowest terms, let f ( x ) = a b . f(x)=ab.

What is the value of x Q + 1 f ( x ) 2 , \large \sum_{x\in\mathbb Q^+}\frac 1{f(x)^2}, where the sum extends over all positive rationals?


Bonus: More generally, exhibit an infinite sequence of distinct rational exponents s s such that x Q + f ( x ) s \displaystyle \sum_{x\in\mathbb Q^+}f(x)^{-s} is rational.


The answer is 2.5.

This section requires Javascript.
You are seeing this because something didn't load right. We suggest you, (a) try refreshing the page, (b) enabling javascript if it is disabled on your browser and, finally, (c) loading the non-javascript version of this page . We're sorry about the hassle.

6 solutions

Otto Bretscher
Nov 9, 2018

Relevant wiki: Principle of Inclusion and Exclusion Problem Solving

If we allow all positive integers a , b a,b , we find a , b 1 a 2 b 2 = ζ ( 2 ) 2 \sum_{a,b}\frac{1}{a^2b^2}=\zeta(2)^2 . If we exclude all a , b a,b with a fixed common prime divisor p p , we find a , b 1 a 2 b 2 = ( 1 1 p 4 ) ζ ( 2 ) 2 \sum_{a,b}\frac{1}{a^2b^2}=(1-\frac{1}{p^4})\zeta(2)^2 . If we exclude all a , b a,b with gcd ( a , b ) 1 \gcd(a,b)\neq 1 , we find gcd ( a , b ) = 1 1 a 2 b 2 = p ( 1 1 p 4 ) ζ ( 2 ) 2 = ζ ( 2 ) 2 ζ ( 4 ) = 2.5 \sum_{\gcd(a,b) = 1}\frac{1}{a^2b^2}=\prod_{p}(1-\frac{1}{p^4})\zeta(2)^2=\frac{\zeta(2)^2}{\zeta(4)}=\boxed{2.5} , by the inclusion-exclusion principle.

Indeed x Q + 1 f ( x ) n = ζ ( n ) 2 ζ ( 2 n ) \sum_{x \in \mathbb{Q}^+} \frac{1}{f(x)^n} \; = \; \frac{\zeta(n)^2}{\zeta(2n)} for any n 2 n \ge 2 , and hence this sum is rational whenever n n is an even positive integer.

Mark Hennings - 2 years, 7 months ago

What does that symbol u used before 2^2 mean??

Mr. India - 2 years, 6 months ago

Log in to reply

It's the Riemann Zeta Function . See formula (9) in the link.

Otto Bretscher - 2 years, 6 months ago

How inclusion exclusion gives the desired result ? How it is 1-1/p^4 ????

Jitender Sharma - 2 years, 5 months ago
Brian Lie
Nov 18, 2018

We first note that F ( s ) = x Q + 1 f ( x ) s = gcd ( a , b ) = 1 a , b = 1 1 ( a b ) s . F(s)=\sum_{x\in\mathbb Q^+}\frac 1{f(x)^s}=\sum_{\stackrel{a,b=1}{\text{gcd}(a,b)=1}}^\infty\frac 1{(a\cdot b)^s}. Moreover for s > 1 s>1 we have that ζ ( s ) 2 = ( a = 1 1 a s ) 2 = a , b = 1 1 ( a b ) s = d = 1 gcd ( a , b ) = d a , b = 1 1 ( a b ) s = d = 1 1 d 2 s gcd ( a , b ) = 1 a , b = 1 1 ( a b ) s = ζ ( 2 s ) F ( s ) \begin{aligned} \zeta(s)^2&=\left(\sum_{a=1}^\infty\frac 1{a^s}\right)^2=\sum_{a,b=1}^\infty\frac 1{(a\cdot b)^s}=\sum_{d=1}^\infty\sum_{\stackrel{a,b=1}{\text{gcd}(a,b)=d}}^\infty\frac 1{(a\cdot b)^s}\\ &=\sum_{d=1}^\infty\frac 1{d^{2s}}\cdot\sum_{\stackrel{a,b=1}{\text{gcd}(a,b)=1}}^\infty\frac 1{(a\cdot b)^s}=\zeta(2s)\cdot F(s) \end{aligned} where ζ ( s ) \zeta(s) is Riemann zeta function . Therefore if s s is a positive even number 2 n 2n then the sum converges to F ( 2 n ) = ζ ( 2 n ) 2 ζ ( 4 n ) = ( ( 1 ) n 1 2 2 n 1 B 2 n π 2 n ( 2 n ) ! ) 2 ( 1 ) 2 n 1 2 4 n 1 B 4 n π 4 n ( 4 n ) ! = ( 4 n 2 n ) B 2 n 2 2 B 4 n Q \begin{aligned} F(2n)&=\frac {\zeta(2n)^2}{\zeta(4n)}=\frac {\left((-1)^{n-1}\dfrac {2^{2n-1}B_{2n}\pi^{2n}}{(2n)!}\right)^2}{(-1)^{2n-1}\dfrac {2^{4n-1}B_{4n}\pi^{4n}}{(4n)!}} \\&=\binom{4n}{2n}\frac {B_{2n}^2}{2|B_{4n}|}\in\mathbb Q \end{aligned} where B k B_k is the k k -th Bernoulli number (it is rational!). Here there are some values of the function F ( 2 ) = 5 2 , F ( 4 ) = 7 6 , F ( 6 ) = 715 691 , F ( 8 ) = 7293 7234 . F(2)=\boxed{\dfrac 52},\, F(4)=\frac 76,\, F(6)=\frac {715}{691},\, F(8)=\frac {7293}{7234}.

Arjen Vreugdenhil
Nov 20, 2018

We are looking for x = ( α , β ) = 1 1 α 2 β 2 . x = \sum_{(\alpha,\beta) = 1} \frac 1{\alpha^2\beta^2}.

Define c = gcd ( a , b ) , a = c α , b = c β c = \text{gcd}(a,b),\ a = c\alpha,\ b = c\beta . Then

( a 1 a 2 ) ( b 1 b 2 ) = a , b 1 a 2 b 2 = c ( α , β ) = 1 1 ( c α ) 2 ( c β ) 2 = ( c 1 c 4 ) ( ( α , β ) = 1 1 α 2 β 2 ) . \left(\sum_a \frac 1{a^2}\right)\left(\sum_b \frac 1{b^2}\right) = \sum_{a,b} \frac 1{a^2b^2} = \sum_c \sum_{(\alpha,\beta) = 1} \frac 1{(c\alpha)^2(c\beta)^2} = \left(\sum_c \frac 1{c^4}\right)\left(\sum_{(\alpha,\beta) = 1} \frac 1{\alpha^2\beta^2}\right). Thus ζ ( 2 ) 2 = ζ ( 4 ) x , \zeta(2)^2 = \zeta(4)x, and x = ζ ( 2 ) 2 ζ ( 4 ) = ( π 2 / 6 ) 2 π 4 / 90 = 90 36 = 2 1 2 . x = \frac{\zeta(2)^2}{\zeta(4)} = \frac{(\pi^2/6)^2}{\pi^4/90} = \frac{90}{36} = \boxed{2\tfrac12}.

The result is rigth and awesome. But the the produ t of 2 infinites series is not always the sum of the product of the arguments Please can you proof that(probably it should start from the cauchy product of 2 series)in order to make your resolution complete ?

Davide Lombardi - 1 year, 6 months ago
Antoine Crbs
Nov 19, 2018

I'm not sure if we can consider that as a solution

First , I've observed that when a b ab become higher, 1 f ( x ) 2 \frac{1}{f(x)^2} become smaller.

So with a simple Python script, I calculated the 50th first terms for a and b

1
2
3
4
5
6
arr, s = [0.0], 0
for a in range(1,50):
    for b in [d for d in range(1,50) if (a/float(d)) not in arr]:
        arr.append(a/float(b)) # useful to don't compute several times the same number (ex: 1/1, 2/2, ...)
        s+=1/float(((a*b)**2))
print(s)

output: 2.44505876736

So we can conclude that the result is close to 2.5 \boxed{2.5}

That's quite the leap. I used the first 300 terms of https://oeis.org/A034444 to get 2.497709839 from which I was much more certain of 2.5

Jeremy Galvagni - 2 years, 6 months ago

Log in to reply

I did this with subsequently higher upper bounds to infer that the result approached 2.5 as the upper bound approached infinity

Jake Shannin - 2 years, 6 months ago
Julian Poon
Nov 20, 2018

Right, I'm a little late to the party

The number of x x that satisfies f ( x ) = q f(x) = q is the number of ways to partition the distinct prime factors of q q into 2 ordered groups: The number of such x x equals 2 ω ( q ) 2^{\omega(q)} , where ω ( n ) \omega(n) is the number of distinct prime factors of n n . Hence, using the substitution f ( x ) = n f(x) = n , the sum equates: n 0 2 ω ( n ) n s = n 0 μ 1 ( n ) n s = ζ ( s ) ζ ( s ) ζ ( 2 s ) = ζ ( s ) 2 ζ ( 2 s ) \large \sum_{n \ge 0}\frac{2^{\omega(n)}}{n^s} = \sum_{n \ge 0}\frac{|\mu|*1(n)}{n^s} = \zeta(s)\frac{\zeta(s)}{\zeta(2s)} = \frac{\zeta(s)^2}{\zeta(2s)}

Vinod Kumar
Nov 23, 2018

Find ratio of

[double sum {(1/n^2)(1/m^2)} from n & m =1 to Inf ] = (π^4/36)

with

[sum {1/n^4} from n=1 to Inf ] = (π^4/90) ,

using WolframAlpha.

Answer=2.5

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...