show that \[\large\sum\limits_{a,b=1}^{\infty} \dfrac{1}{a^2b^2} = \dfrac{5}{2}\] for relatively prime \(a,b\) A not so useful hint can be found @ http://openstudy.com/users/ganeshie8#/updates/54e1ae2ee4b0b0ad88553f82
\(\large \begin{align} \color{black}{\sum\limits_{a,b=1}^{\infty} \dfrac{1}{a^2b^2} = \sum\limits_{a=1}^{\infty} \dfrac{1}{a^2} \times \sum\limits_{b=1}^{\infty} \dfrac{1}{b^2} \hspace{.33em}\\~\\}\end{align}\) so this is equivalent to this ?
thats a good start!
ok so this identity is true alright .
\[\sum\limits_{a,b=1}^{\infty} \dfrac{1}{a^2b^2} = \sum\limits_{a=1}^{\infty} \dfrac{1}{a^2} \times \sum\limits_{b=1}^{\infty} \dfrac{1}{b^2} = \left(\sum\limits_{c=1}^{\infty} \dfrac{1}{c^2} \right)^2 \] the variables are just dummy..
how can we write \(c\) in terms of \(a\) and \(b\)
\(\large \begin{align} \color{black}{\sum\limits_{a,b=1}^{3} \dfrac{1}{a^2b^2} = \sum\limits_{a=1}^{3} \dfrac{1}{a^2} \times \sum\limits_{b=1}^{3} \dfrac{1}{b^2}=\left(\sum\limits_{c=1}^{3} \dfrac{1}{c^2} \right)^2 \hspace{.33em}\\~\\}\end{align}\) if i consider small example like this then how is \(c\) related to \(a\) and \(b\)
Haha I guess I would have to say it's pi^4/36 but apparently I need to look at it differently hmm
i think we need to use that result.. the original sum only involves the relatively prime pairs a,b
Ahhh right, I forgot about that.
@mathmath333 you seem to be on right track.. keep going :)
Hmm I don't know if this is a true statement anymore since a and b are relatively prime hmmm. \[\ \sum\limits_{a=1}^{\infty} \dfrac{1}{a^2} \times \sum\limits_{b=1}^{\infty} \dfrac{1}{b^2} = \left(\sum\limits_{c=1}^{\infty} \dfrac{1}{c^2} \right)^2 \]
Exactly! i think we can start wid that general result and see if we can split the sum based on the given condition..
below is true if we run s,t over all positive integers \[\ \sum\limits_{s=1}^{\infty} \dfrac{1}{s^2} \times \sum\limits_{t=1}^{\infty} \dfrac{1}{t^2} = \left(\sum\limits_{s=1}^{\infty} \dfrac{1}{s^2} \right)^2 \]
i dont understand why u say and b are relatively prime i thought a and b are nothing but indeces
oh i see
yes, in the sum, consider only the indices pairs (a,b) that are relatively prime
it is just a generally accepted notation to put conditions on terms in sum..
i m still unable to understand how will sigma satisfy ur condition
of a and b coprime
\[\large \begin{align} \color{black}{\sum\limits_{a,b=1}^{3} \dfrac{1}{a^2b^2} = \sum\limits_{a=1}^{3} \dfrac{1}{a^2} \times \sum\limits_{b=1}^{3} \dfrac{1}{b^2}=\left(\sum\limits_{c=1}^{3} \dfrac{1}{c^2} \right)^2 \hspace{.33em}\\~\\}\end{align} \] this is tre when there are no conditions on a,b @mathmath333
we need to somehow use this and extract the sum when a,b are relatively prime
yes that is true ,no conditions apply
\[\large \begin{align} \color{black}{\sum\limits_{a,b=1}^{3} \dfrac{1}{a^2b^2} =\left(\sum\limits_{c=1}^{3} \dfrac{1}{c^2} \right)^2 \hspace{.33em}\\~\\}\end{align} \] lets evaluate this and see if we get same value on both sides
yes u know its true when condition aren;t there
yes
but when conditions do apply how will the summation process behave
One way to work it is by subtracting the terms related to a,b not relatively prime from above sum
something like this : \[\sum\limits_{a,b=1}^{\infty} \dfrac{1}{a^2b^2} = \left(\sum\limits_{c=1}^{\infty} \dfrac{1}{c^2} \right)^2\] So \[\sum\limits_{\gcd(a,b)=1, a,b=1}^{\infty} \dfrac{1}{a^2b^2} = \left(\sum\limits_{c=1}^{\infty} \dfrac{1}{c^2} \right)^2 - \sum\limits_{\gcd(a,b)\ne1,~a,b=1}^{\infty} \dfrac{1}{a^2b^2} \]
i think this stuff is either above my reach
found this , might be helpful http://www.wolframalpha.com/input/?i=%28%5Cfrac%7B%5Cpi%5E2%7D%7B6%7D%29%5E2%3D%5Cfrac%7B5%7D%7B2%7D%5Csum%5Climits_%7Bk%3D1%7D%5E%7B%5Cinfty%7D+%5Cdfrac%7B1%7D%7Bk%5E4%7D+
Thats it!
we just need to put all the pieces together..
which is tricky ofc
still trying to understand the process
i got idea from zeta function ok will write down in morning
I'll post a solution by 5 PM IST today incase if this question still remains open :)
OMG CONEXUS BOTS
why are there so many people here
lol are they connexus teachers
they need to get laaaiiidddd XD
that comment made me laugh but il need to remove that haha
i think this will be having some telescooping series .
\(\large (\frac{\pi^2}{6})^2-x=\frac{5}{2}\)
wolfram gave some "digamma function" earlier, which i don't know
it seems from your wolfram link we have \[\sum\limits_{a,b=1}^{\infty} \dfrac{1}{a^2b^2} = \left(\sum\limits_{c=1}^{\infty} \dfrac{1}{c^2} \right)^2 = \left(\frac{\pi^2}{6}\right)^2 = \dfrac{5}{2}\sum\limits_{k=1}^{\infty} \dfrac{1}{k^4} \] ?
yes
so we need to show that \[\sum\limits_{a,b=1}^{\infty} \dfrac{1}{a^2b^2} ~~=~~ \sum\limits_{k=1}^{\infty} \dfrac{1}{k^4} \times\color{Red}{ \dfrac{5}{2}} ~~=~~ \sum\limits_{k=1}^{\infty} \dfrac{1}{k^4} \times \color{Red}{\sum_{\substack{a,b=1\\\gcd(a,b) =1}}^{\infty} \dfrac{1}{a^2b^2} }\]
\(\large \begin{align} \color{black}{\sum_{\substack{a,b=1\\gcd(a,b) =1}}^{3} \dfrac{1}{a^2b^2}\hspace{.33em}\\~\\}\end{align}\) what will be the first 3 values for this
\[\large \begin{align} \color{black}{\sum_{\substack{a,b=1\\gcd(a,b) =1}}^{3} \dfrac{1}{a^2b^2}\hspace{.33em}\\~\\}\end{align}\] is same as \[\large \begin{align} \color{black}{\sum\limits_{a=1}^3\sum_{\substack{b=1\\gcd(a,b) =1}}^{3} \dfrac{1}{a^2b^2}\hspace{.33em}\\~\\}\end{align}\]
\[\begin{align}\sum\limits_{a=1}^3\sum_{\substack{b=1\\gcd(a,b) =1}}^{3} \dfrac{1}{a^2b^2}&= \left(\dfrac{1}{1^21^2} + \dfrac{1}{1^22^2} +\dfrac{1}{1^23^2} \right) \\~\\~\\ &+ \left(\dfrac{1}{2^21^2} +\dfrac{1}{2^23^2} \right) \\~\\~\\ &+ \left(\dfrac{1}{3^21^2} +\dfrac{1}{3^22^2} \right) \\~\\~\\ \end{align}\]
Notice without gcd(a,b)=1 conndition, we will get 9 terms in the sum but with that condition there are two terms missing : \(\dfrac{1}{2^22^2}\) and \(\dfrac{1}{3^23^2}\) because \(\gcd(2,2) \ne 1\) and \(\gcd(3,3)\ne 1\)
Oh u revealed
I cant read fully latex don't appear on time but I might have same idea I thought of it like this for assuming gcd of a,b is not 1 then I would have this (sum 1/a^2)^2 which is the same as (zeta 2 )^2 with expanding some terms assuming ur series S I would got S= (zeta 2/ zeta 4 )^2= 2/5 Its annoying posting from phone :( But looking forward to ur method seems the same :o
Lol hbd kai :p
And yet this (Sum 1/ a^2 * sum 1/b^2 ) = (sum 1/b^2)^2 From kai comment but from gcd (a,b ) is not 1 its for a,b with all possibilities of gcd which is the same as (zeta 2 )^2
Just adding extra counter for all gcd hmmm idk if any one understand me lol but I don't have laptop xD to write neat
the solution i am having similar to that... could you post the detailed solution :)
mystery continous..
Wow very neat :o
It make sense to me and we don't need zeta xD
Like u only used it in final answer u didn't made mass expantion this question made my week lol
That is exactly same as your idea.. I just tried to make it as look as simple as possible by avoiding zeta. Below is true as you said earlier : \[\sum_{\substack{a,b=1\\\gcd(a,b) =1}}^{\infty} \dfrac{1}{a^2b^2} = \dfrac{\zeta(2)^2}{\zeta(4)} = \dfrac{5}{2}\]
Fixed a typo : We know below is true \[ \left(\frac{\pi^2}{6}\right)^2 = \left(\sum\limits_{c=1}^{\infty} \dfrac{1}{c^2} \right)^2 =\sum\limits_{x,y=1}^{\infty} \dfrac{1}{x^2y^2} \tag{1}\] Next notice that all the positive integer pairs can be split into classes based on gcd. Like, all the pairs \((x,y)\) whose gcd is \(2\) exist in a class \(d_2\), all the pairs \((x,y)\) whose gcd is \(3\) exist in a class \(d_3\), etc... each pair exists in one and only one gcd class. So we can split the sum in \((1)\) as below : \[ \left(\frac{\pi^2}{6}\right)^2 = \left(\sum\limits_{c=1}^{\infty} \dfrac{1}{c^2} \right)^2 =\sum\limits_{x,y=1}^{\infty} \dfrac{1}{x^2y^2} = \sum\limits_{d=1}^{\infty}\sum_{\substack{x,y=1\\\gcd(x,y) =d}}^{\infty} \dfrac{1}{x^2y^2} \tag{2}\] Since \(\gcd(x,y) = d\), we can write \(x,y\) as \(x = ad,~ y = bd\) for some relatively prime integers \(a,b\) and \((2)\) becomes : \[\begin{align} \left(\frac{\pi^2}{6}\right)^2 &=\sum\limits_{d=1}^{\infty}\sum_{\substack{a,b=1\\\gcd(a,b) =1}}^{\infty} \dfrac{1}{(ad)^2(bd)^2} \\~\\ &=\sum\limits_{d=1}^{\infty} \sum_{\substack{a,b=1\\\gcd(a,b) =1}}^{\infty} \dfrac{1}{d^4} \dfrac{1}{a^2b^2} \\~\\ &=\left(\sum\limits_{d=1}^{\infty} \dfrac{1}{d^4} \right) \left(\sum_{\substack{a,b=1\\\gcd(a,b) =1}}^{\infty} \dfrac{1}{a^2b^2} \right)~~~\color{red}{\star }\\~\\ &= \left(\dfrac{\pi^4}{90} \right) \left(\sum_{\substack{a,b=1\\\gcd(a,b) =1}}^{\infty} \dfrac{1}{a^2b^2} \right)\\~\\ \end{align}\] So we have \[ \left(\frac{\pi^2}{6}\right)^2 = \left(\dfrac{\pi^4}{90} \right) \left(\sum_{\substack{a,b=1\\\gcd(a,b) =1}}^{\infty} \dfrac{1}{a^2b^2} \right)\] yielding \[\dfrac{5}{2} =\sum_{\substack{a,b=1\\\gcd(a,b) =1}}^{\infty} \dfrac{1}{a^2b^2} \] \(\color{red}{\star}\) : refer to the hint in main question
cool
Join our real-time social learning platform and learn together with your friends!