7

I was messing around with some integrals and series when I arrived at the result (and WolframAlpha agrees):

$$\sum_{n=0}^\infty \frac{(2n)!}{(2^nn!(2n+1))^2} = \sum_{n=0}^\infty \frac{1 }{2^{2n}}\binom{2n}{n} \frac{1}{(2n+1)^2} = \frac{\pi}{2}\ln(2)$$

I therefore have a proof, but I would like to see how a proof would be done the other way around (i.e. starting with this series). Notice that the series could also be expressed with double factorials:

$$\sum_{n=0}^\infty \frac{(2n)!}{(2^nn!(2n+1))^2} = \sum_{n=0}^\infty \frac{(2n-1)!!}{(2n)!!}\frac{1}{(2n+1)^2}$$

My Derivation

My derivation is quite simple. I was considering (and trying to solve) the integral

$$\int_0^1 \frac{\arcsin(x)}{x} \, dx$$

I couldn't work it out, however I found the exact same question here on math exchange and the answers show that the integral equals $\frac{\pi}{2}\ln(2)$. What I did found however, was that I could plug the Taylor series for $\arcsin(x)$ (the proof of the taylor series can be found here) into the original series to only be left with a series. Here's the Taylor series for $\arcsin(x)$:

$$ \arcsin(x) = \sum_{n=0}^\infty \frac{1 }{2^{2n}}\binom{2n}{n} \frac{ x^{2n+1}}{2n+1} $$

So plugging in the series above into the integral gives us:

$$\int_0^1 \frac{1}{x} \sum_{n=0}^\infty \frac{1 }{2^{2n}}\binom{2n}{n} \frac{ x^{2n+1}}{2n+1} \, dx$$

Now here we have to be careful, but since the series works on the integral's interval $[0,1]$ and the summand is only positive, we can interchange the the integral and the sum, leaving us with:

$$\sum_{n=0}^\infty \frac{1 }{2^{2n}}\binom{2n}{n} \frac{1}{2n+1} \, \int_0^1 \frac{x^{2n+1}}{x} \, dx$$

And then solving the integral in the sum gives us the final result:

$$\sum_{n=0}^\infty \frac{1 }{2^{2n}}\binom{2n}{n} \frac{1}{(2n+1)^2} = \frac{\pi}{2}\ln(2)$$

And then expanding the binomial gives the original series in the title.

  • Do you know the classical method of summation of a series using complex variable calculus : summation of residues of $\pi \cot(\pi z)$ at the poles of $f(z)$ ? See paragraph 3 of this document. Of course, the difficulty is to find here the good $f(z)$... – Jean Marie Mar 17 '21 at 18:31
  • @JeanMarie Eh, not really. I have been thinking about learning the residue theorem for a long time, but I haven't had time for it. After all, I'm still a high school student. Thanks, anyway. – Casimir Rönnlöf Mar 17 '21 at 18:51
  • I see... In a general manner, it is rare that the sum of a series isn't somehow connected to integrals that you have already met maybe in your high school. But integrals of some $f(x)$ with $x$ real. I was speaking about a way of doing integration with $f(z)$ functions ($z \in \mathbb{C}$) that you will meet only after, say, 2 years of university specializing in Maths. – Jean Marie Mar 17 '21 at 19:02
  • @JeanMarie Yeah, I'm familiar with the basics of complex analysis (not on a deep level at all) e.g. Cauchy's theorem etc and I understand complex functions to some level. Anyways, I derived this sum using an integral involving the inverse sine and therefore the sum resembles the taylor series for inverse sine a lot. I still think it's a cool series and would like some direct way of proving it. – Casimir Rönnlöf Mar 17 '21 at 19:10
  • @JeanMarie Thanks for your kind comments still! – Casimir Rönnlöf Mar 17 '21 at 19:11
  • Answers show its relation to $\int_0^{\pi/2}t\cot t,dt$. With $\int_0^{\pi/2}t\csc t,dt$ instead, one gets $$\frac{1-\frac{1}{3^2}+\frac{1}{5^2}-\frac{1}{7^2}+\ldots}{1-\frac{1}{3}+\frac{1}{5}-\frac{1}{7}+\ldots}=1+\frac{1^2}{2^2 3}+\frac{1^2 3^2}{2^2 4^2 5}+\frac{1^2 3^2 5^2}{2^2 4^2 6^2 7}+\ldots$$ – metamorphy Mar 17 '21 at 21:05
  • Please let us know the integrals you were "messing around with" so that some approach can be given based on that. Or perhaps you may have found those integrals already present in one of the answers here. In that case also do let us know. – Paramanand Singh Mar 18 '21 at 04:24
  • @ParamanandSingh I added my derivation :) – Casimir Rönnlöf Mar 18 '21 at 06:40
  • 1
    Thanks a lot and +1 – Paramanand Singh Mar 18 '21 at 07:09
  • @ParamanandSingh No problem :) – Casimir Rönnlöf Mar 18 '21 at 07:18

3 Answers3

3

Consider the well-known series $$f(x)=\sum_{n=0}^{\infty}{2n\choose n}x^n=\frac{1}{\sqrt{1-4x}}$$

We will attempt to evaluate the series

$$I(x)=\sum_{n=0}^{\infty}{2n\choose n}\frac{x^n}{(2n+1)^2}$$

A simple relation between these two series exists:

$$\frac{d^2}{dt^2}[e^tI(e^{2t})]=e^t f(e^{2t})=\frac{e^t}{\sqrt{1-4e^{2t}}}$$

To obtain our goal series we just need to integrate twice with an appropriate boundary condition

$$e^t I(e^{2t})=\int_{-\infty}^tdx\int_{-\infty}^{x}dT\frac{e^T}{\sqrt{1-4e^{2T}}}$$

The inner integral is easy to do which leads us to the expression

$$I(x)=\frac{1}{2\sqrt{x}}\int_{-\infty}^{\frac{\ln x}{2}}dt ~ \arcsin(2e^t)$$

which can be further simplified to

$$I(x)=\frac{1}{2\sqrt{x}}\int_{0}^{2\sqrt{x}}\frac{\arcsin u}{u}du$$

Our goal is to evaluate this at $x=1/4$. Performing an integration by parts and afterwards the substitution $u=\cos t$ we reduce the entire thing to a well-known integral that can be calculated in various ways

$$I(1/4)=-\int_0^{\pi/2}\ln \cos t dt=\frac{\pi\ln 2}{2}$$

DinosaurEgg
  • 10,775
2

Start with well-known series: $$ f(x)=\sum_{k=0}^\infty\binom{2n}n\left(\frac x2\right)^{2n}=\frac1{\sqrt{1-x^2}}\tag1 $$ From this one obtains: $$\begin{align} &\sum_{k=0}^\infty\binom{2n}n\frac1{2n+1}\left(\frac x2\right)^{2n}=\frac1x\int_0^x \frac{dt}{\sqrt{1-t^2}}=\frac{\arcsin(x)}x\\ &\sum_{k=0}^\infty\binom{2n}n\frac1{(2n+1)^2}\left(\frac 12\right)^{2n}=\frac12\int_0^1 \frac{\arcsin(t)}tdt.\\ \end{align}$$ The last integral can be evaluated as: $$\begin{align} \int_0^1 \frac{\arcsin(t)}tdt&=\left[\arcsin(t)\log(t)\right]_0^1- \int_0^1 \frac{\log(t)}{\sqrt{1-t^2}}dt\\ &=-\int_0^\pi \log(\sin u)du=\pi\log(2). \end{align}$$

user
  • 26,272
-1

$$\frac{(2n)!}{(2^nn!(2n+1))^2}=\frac{(2n)!}{2^{2n}(n!)^2(2n+1)^2}={2n\choose n}\frac{1}{2^{2n}(2n+1)^2}={2n\choose n}\frac{1}{(2n+1)^2}\left(\frac12\right)^{2n}$$ Now try and find a function whose taylor series fits this summation

Henry Lee
  • 12,215