Proving $\sum_{n=1}^{\infty}\binom{2n}{n}^2\frac{4H_{2n}-3H_n}{n2^{4n}}=\zeta(2)$

binomial-coefficientscalculusharmonic-numbersintegrationsummation

While trying to solve Prove $\int_{0}^{1}\frac1k K(k)\ln\left[\frac{\left(1+k \right)^3}{1-k} \right]\text{d}k=\frac{\pi^3}{4}$

I was able to reduce it to the following form,
$$\sum_{n=1}^{\infty}\binom{2n}{n}^2\frac{4H_{2n}-3H_n}{n2^{4n}}=\zeta(2)$$
Where, $H_n$ is the $n$th Harmonic Number.

This should be correct if I did all the steps correctly, but I am unable to prove this further.

I tried searching more about series of this form which have the Central Binomial Coefficient Squared with Harmonic Numbers but was not able to find any.

Any help would be appreciated.

EDIT:

Considering,
$$M(x)=\int_{0}^{x}\left(\frac{2K\left(\sqrt{s}\right)-\pi}{s}\right)ds$$
where $K(k)$ is the Complete Elliptical Integral of the First Kind.

Then the question is reducible to,
$$\int_{0}^{1}\frac{M\left(1\right)+3M\left(x\right)-4M\left(x^{2}\right)}{1-x}dx=\frac{\pi^3}{6}$$
Though no idea comes to mind after this.

Here are some related results:
$$M(1)=4\pi\ln2-8G$$
$$\int_{0}^{1}M\left(x\right)dx=-8G-4+\pi+4\pi\ln2$$

where $G$ is Catalan's Constant.

EDIT:

Let,
$$A=\Im \operatorname{Li_{3}}\left(\frac{1+i}{2}\right)$$
where $\operatorname{Li_{3}}(.)$ is the trilogarithm.

and $G$ be the Catalan's Constant.

Using the Twin Integrals Mentioned in the Post at the Beginning, was able to arrive at these.
$$\sum_{r=1}^{\infty}\binom{2r}{r}^2\frac{H_{2n}}{n2^{4n}}=24\frac{G\ln2}{\pi}+48\frac{A}{\pi}-\frac{3}{2}\ln\left(2\right)^{2}-\frac{29}{24}\pi^{2}$$
$$\sum_{r=1}^{\infty}\binom{2r}{r}^2\frac{H_{n}}{n2^{4n}}=32\frac{G\ln2}{\pi}+64\frac{A}{\pi}-2\ln\left(2\right)^{2}-\frac{5}{3}\pi^{2}$$

Best Answer

A fast and simple solution idea by Cornel Ioan Valean

We will use the power of the ideas and strategies from the books (Almost) Impossible Integrals, Sums, and Series (2019) and More (Almost) Impossible Integrals, Sums, and Series (2023).

Okay, let's start!

From the book, More (Almost) Impossible Integrals, Sums, and Series (2023), Sect. $4.5$, pages $396$-$398$, we have that $$\displaystyle \color{blue}{-\frac{\log(1-x^2)}{\sqrt{1-x^2}}=\sum_{n=0}^{\infty}x^{2n} \frac{1}{4^n}\binom{2n}{n}(2 H_{2n}-H_n),\ |x|<1} ,$$ and exploiting this fact and turning the left-hand side into a double integral, $\displaystyle \int _0^{\pi/2}\left(\int _0^{\sin (x)}\frac{\log \left(1-y^2\right)}{y\sqrt{1-y^2}}\textrm{d}y\right)\textrm{d}x$, immediately reveals (after changing the integration order) that $$\sum _{n=1}^{\infty } \binom{2 n}{n}^2 \frac{2 H_{2 n}-H_n}{n 2^{4 n}}$$ $$ =\frac{\pi^2}{2}+\frac{16}{\pi }\int_0^1 \frac{\arctan(x) \log(1-x)}{x} \textrm{d}x+\frac{16}{\pi }\int_0^1 \frac{\arctan(x) \log(1+x)}{x} \textrm{d}x$$ $$-\frac{16}{\pi }\int_0^1 \frac{\arctan(x) \log(1+x^2)}{x} \textrm{d}x. \tag1$$

On the other hand, exploiting that $$\int_0^{\pi/2} \log(\sin(x)) \sin^{2n}(x) \textrm{d}x=\frac{\pi}{2}\frac{1}{2^{2n}}\binom{2n}{n}\left(H_{2n}-H_n-\log(2)\right),$$ which is also found in the sequel, page $191$, multiplying both sides by $\displaystyle \frac{1}{n 2^{2n}}\binom{2n}{n}$, making the summation from $n=1$ to $\infty$, and rearranging, we get that $$\sum_{n=1}^{\infty}\binom{2n}{n}^2\frac{H_{2n}-H_n-\log(2)}{n 2^{4n}}$$ $$\small =-\frac{8 }{\pi }\int_0^1 \frac{\log ^2(1+x^2)}{1+x^2} \textrm{d}x+\frac{8 }{\pi } \log (2)\int_0^1 \frac{ \log(1+x^2)}{1+x^2}\textrm{d}x$$ $$+\frac{8 }{\pi }\int_0^1 \frac{\log(x) \log(1+x^2)}{1+x^2} \textrm{d}x.\tag2$$

The rest is known and trivial, and the separate series involving the numbers $H_{2n}$ and $H_n$ are extracted by a system of relation consisting of $(1)$ and $(2)$.

End of story

Thank you so much Cornel for your work and your life-changing books, (Almost) Impossible Integrals, Sums, and Series (2019) and More (Almost) Impossible Integrals, Sums, and Series (2023).

(Very) important: Exploiting such techniques involving the series in blue for building systems of relations, one can derive extremely difficult series. Some (simple) paper(s) showing such examples will be written soon.

Update 1: Indeed, the series taken apart also contain a Trilogarithm with a complex argument. The forms of the two key identities (that is, using integrals) avoid that appearance, and to get directly the desired value without touching the Trilogarithm with a complex argument, all we need is to turn the integrals from the identity in $(2)$ into arctan-log integrals. Then we need to exploit the arctan-log integral identities given in the sequel, Sects. $1.36$-$1.38$, pages $48$-$51$. That's all (and done)!

Update 2: A full solution with all the details of this problem will be found soon in a separate paper (the main series plus the two separated series).

Update 3: Binoharmonic Series with the Squared Central Binomial Coefficient And Their Integral Transformation Using Elliptic Integrals by Cornel Ioan Valean