8

Using Maple I am obtaining the numerical approximation

$$0.5902373619$$

Please, let me know what is the closed form. Many thanks.

Olivier Oloa
  • 122,789
Juan Ospina
  • 2,257

2 Answers2

11

Hint.

Observe that

$$ J_0(n)=\frac 1\pi \int_0^\pi \cos (n \sin x)\:{\rm d}x \tag1 $$

and that $$\sum_{n=1}^\infty\frac{\cos nt}{n^4}=\frac{\pi^4}{90}-\frac{\pi^2 t^2}{12}+\frac{\pi t^3}{12}-\frac{t^4}{48},\quad 0\leq t\leq 2\pi. \tag2 $$ Then, due to normal convergence of the series on $[0,2\pi] $, we may write $$ \begin{align} \sum_{n=1}^\infty\frac{J_0^2(n)}{n^4} & =\frac{1}{\pi^2} \int_0^\pi \!\!\int_0^\pi \sum_{n=1}^\infty\frac{\cos (n \sin x)\cos (n \sin y)}{n^4}{\rm d}x\:{\rm d}y .\tag3 \end{align} $$ We may plug $$ 2\cos (n \sin x)\cos (n \sin y)=\cos (n(\sin x+\sin y))+\cos (n(\sin x-\sin y)) \tag4 $$ into $(3)$ and integrate as here.

Hence we obtain

$$ \sum_{n=1}^\infty\frac{J_0^2(n)}{n^4}=\frac{\pi ^4}{90}-\frac{\pi ^2}{12}-\frac{3}{64}+\frac{32}{27 \pi }. \tag5 $$

A numerical value is $$ \sum_{n=1}^\infty\frac{J_0^2(n)}{n^4} =0.5902373616900361395467798486 \ldots. $$


Some details.

Identity $(2)$ may be rewritten as $$\sum_{n=1}^\infty\frac{\cos nt}{n^4}=\frac{\pi^4}{90}-\frac{\pi^2 t^2}{12}+\frac{\pi |t|^3}{12}-\frac{t^4}{48},\quad -\pi \leq t\leq \pi. \tag6 $$

Then using $(3)$, $(4)$ and $(6)$, we get

$$ \begin{align} \sum_{n=1}^\infty\frac{J_0^2(n)}{n^4} =\frac{1}{2\pi^2}\int_0^\pi \!\!\int_0^\pi \left[\left(\frac{\pi^4}{90}-\frac{\pi^2}{12}(\sin x+\sin y)^2+\frac{\pi }{12}(\sin x+\sin y)^3-\frac{1}{48}(\sin x+\sin y)^4\right)+\left(\frac{\pi ^4}{90}-\frac{\pi ^2}{12}(\sin x-\sin y)^2+\frac{\pi }{12}\left|\sin x-\sin y\right|^3-\frac{1}{48}(\sin x-\sin y)^4\right)\right]{\rm d}x\:{\rm d}y .\tag7 \end{align} $$ We just have to be careful with the computation of the term ($|t|^3=\left({\rm Abs} (t)\right)^3$) $$ \begin{align} \frac{1}{2\pi^2}\times \frac{\pi }{12}\times \int_0^\pi \!\!\int_0^\pi \left|\sin x-\sin y\right|^3 {\rm d}x\:{\rm d}y & = \frac{1}{2\pi^2}\times\frac{\pi }{12} \times 4\int_0^{\pi/2} \!\!\int_0^{\pi/2} \left|\sin x-\sin y\right|^3 {\rm d}x\:{\rm d}y \\\\ &=-\frac{13}{36}+\frac{32}{27 \pi }. \end{align} $$

Olivier Oloa
  • 122,789
  • 4
    @JuanOspina My method is correct, you just have to be careful with the sign of the argument $(\sin x-\sin y)$ in the series, which is not always positive, leading to consider instead $$\int_0^\pi !!\int_0^\pi ({\rm Abs}(\sin x-\sin y))^3{\rm d}x:{\rm d}y$$ probably you did not see this. Can you take it from there? Hoping to be clear. Thank you! – Olivier Oloa Sep 20 '14 at 21:49
  • 1
    @JuanOspina OK, I'm going to show the computations. Thanks. – Olivier Oloa Sep 20 '14 at 22:23
  • 2
    Hi Olivier, you are a STAR. Super very nice computation, many, many thanks. – Juan Ospina Sep 21 '14 at 00:01
  • 1
    Mr. @OlivierOloa, how did you get equation 2 or 6? Could you please elaborate? To me, it looks like $\Re,\text{Li}_4(e^{int})$. Thank you (ô‿ô) – Anastasiya-Romanova 秀 Sep 21 '14 at 08:19
  • @JuanOspina You are very welcome. – Olivier Oloa Sep 21 '14 at 09:47
  • @Anastasiya-Romanova This is a classic result [http://dlmf.nist.gov/24.8#i], it may be seen as: [The Expansion of Bernoulli Polynomials in Fourier Series.] For a proof you have [www.math.tulane.edu/~vhm/fourier.pdf] Equation (2) or (6) in the answer above is equivalent to (45) of the paper. Thanks. – Olivier Oloa Sep 21 '14 at 10:06
  • Thanks Mr. @OlivierOloa for the link of paper. I haven't learned Fourier Series yet. +1 for your answer – Anastasiya-Romanova 秀 Sep 21 '14 at 10:15
  • @Anastasiya-Romanova. You are right it is possible to compute (6) using polylogarithm functions, specifically $Li_4$: $$\sum _{n=1}^{\infty }{\frac {\cos \left( nt \right) }{{n}^{4}}}={\frac {1}{2}},{ \it polylog} \left( 4,{{\rm e}^{it}} \right) +{\frac {1}{2}},{\it polylog} \left( 4,{{\rm e}^{-it}} \right)$$ – Juan Ospina Sep 21 '14 at 12:47
  • 2
    @JuanOspina In general, we can use this identity $$\text{Li}_n(e^{2\pi ix})+(-1)^n, \text{Li}_n(e^{-2\pi ix})=-\frac{(2\pi i)^n}{n!}B_n(x)$$ where $B_n(x)$ is the Bernoulli polynomials – Anastasiya-Romanova 秀 Sep 21 '14 at 13:27
  • Hi @Anastasiya-Romanova, you are really a Princess of Mathematics. Using your identity and the definition of the fourth Bernoulli polynomial you are able to derive the equation (2) without using Fourier series. Congratulations. – Juan Ospina Sep 21 '14 at 13:44
  • Hello Mr. @JuanOspina, I look at your profile & you're from Medellin, Colombia. Wow! That's dangerous place in South America. I watch from CNN that drugs cartel are everywhere there, is that true? – Anastasiya-Romanova 秀 Sep 21 '14 at 13:49
  • Hi @Anastasiya-Romanova, Medellin is a beautiful place in the land. It is a city with many problems but at general the people are good and very kind. All the best, Princess of Mathematics. – Juan Ospina Sep 21 '14 at 13:59
  • 2
    Mr. @JuanOspina: Indeed, Andres Escobar is also from there. Say hi to James Rodriguez! ≧◠◡◠≦✌ – Anastasiya-Romanova 秀 Sep 21 '14 at 14:06
  • 1
    Hi @Anastasiya-Romanova, you are very kind, many thanks. – Juan Ospina Sep 21 '14 at 14:13
4

Using the wonderful method of Olivier implemented by Maple it is possible to prove that

$$\sum _{n=1}^{\infty }{\frac { {{\it J}_{0}\left(\,\alpha\,n\right)} ^{2}}{{n}^{4}}}=-{\frac {3 }{64}}\,{\alpha}^{4}+{\frac {1}{90}}\,{\pi }^{4}-{\frac {1}{12}}\,{\alpha}^{2}{ \pi }^{2}+{\frac {32}{27}}\,{\frac {{\alpha}^{3}}{\pi }} $$

$$\sum _{n=1}^{\infty }{\frac { {{\it J}_{0}\left(\,\alpha\,n\right)} ^{2}}{{n}^{2}}}=\frac{1}{4}\,{ \alpha}^{2}+\frac{1}{6}\,{\pi }^{2}-{\frac {4\alpha}{\pi }} $$

For this last sum, when $\alpha =2 $ we obtain

$$\sum _{n=1}^{\infty }{\frac { {{\it J}_{0}\left(\,2\,n\right)} ^{2}}{{n}^{2}}}=1+\frac{1}{6}\,{\pi }^{2}-8\,{\pi }^{-1} $$

that is justly the result derived here.

Other results:

$$\sum _{n=1}^{\infty }{\frac { {{\it J}_{0}\left(\,\alpha\,n\right)} ^{2}}{{n}^{6}}}={\frac {1} {945}}\,{\pi }^{6}+{\frac {1}{64}}\,{\pi }^{2}{\alpha}^{4}-{\frac {1}{ 180}}\,{\pi }^{4}{\alpha}^{2}+{\frac {5}{1152}}\,{\alpha}^{6}-{\frac { 512}{3375}}\,{\frac {{\alpha}^{5}}{\pi }} $$

$$\sum _{n=1}^{\infty }{\frac { {{\it J}_{0}\left(\,\alpha\,n\right)} ^{2}}{{n}^{8}}}=-{\frac {{\pi }^{6}{\alpha}^{2} }{1890}}\,+{\frac {1}{960}}\,{\pi }^{4}{\alpha}^ {4}-{\frac {35}{147456}}\,{\alpha}^{8}-{\frac {5}{3456}}\,{\pi }^{2}{ \alpha}^{6}+{\frac {1}{9450}}\,{\pi }^{8}+{\frac {4096}{385875}}\,{ \frac {{\alpha}^{7}}{\pi }} $$

Juan Ospina
  • 2,257
  • 1
    Nice results! Thank you. +1 – Olivier Oloa Sep 21 '14 at 10:09
  • Hi Olivier, thanks. These results were obtained using Maple with your equations (1), (3) and (7). Your equation (4) was not used. I am using the evaluation of (3) in terms of the polylogaritm function given by Maple. Anastasiya-Romanova is noting that point. Again, many thanks. All the best. – Juan Ospina Sep 21 '14 at 12:57