10
$\begingroup$

I ran across this Euler sum while trying to evaluate an integral. I mentioned it in another thread, but though perhaps asking about it separate may be a good idea.

Is there a closed form for this Euler sum? $$\sum_{n=1}^{\infty}\frac{H_{2n}}{n(6n+1)}.$$

Numerically, it converges to around $0.502788$

I found that sum while trying to evaluate

$$\int_{0}^{1}\log(1+x^{3})\log(1-x^{3})dx=\int_{0}^{1}\sum_{n=1}^{\infty}\frac{1}{n}\left(H_{n}-H_{2n}-\frac{1}{2n}\right)dx$$

I managed to obtain two of the sums. This is the only one giving me trouble.

It comes from the identity:

$$\log(1+x)\log(1-x)=\sum_{n=1}^{\infty}\frac{1}{n}\left(H_{n}-H_{2n}-\frac{1}{2n}\right)x^{2n}$$

Note that:

$$\sum_{n=1}^{\infty}\frac{H_{n}}{n(6n+1)}=1/4\left(-72+2\gamma^{2}+\pi^{2}+4\gamma\psi(1/6)+2\psi(1/6)-2\psi_{1}(7/6)\right)$$

and

$$\sum_{n=1}^{\infty}\frac{1}{n^{2}(6n+1)}=3\sqrt{3}\pi+\zeta(2)+9\log(3)+12\log(2)-36.$$

I thought maybe the residue method would work, but I am not so sure.

By considering $f(z)=\frac{\pi cot(\pi z)(\gamma+\psi(-2z))}{z(6z+1)}$

The residue at the origin is $18-\frac{\pi^{2}}{2}$

The residue at $-1/6$ is $3\pi (\gamma+\psi(1/3))$

But, there does not appear to be a residue for the positive half integers.

The series for the positive integers is:

$$\frac{1}{2}\cdot \frac{1}{(z-n)^{2}}+\frac{H_{2n}}{z-n}+...$$

which gives a residue of :

$$\frac{-1}{2}\sum_{n=1}^{\infty}\frac{12n+1}{n^{2}(6n+1)^{2}}+\sum_{n=1}^{\infty}\frac{H_{2n}}{n(6n+1)}+....=\frac{1}{2}\psi_{1}(7/6)-\frac{\pi^{2}}{12}+H$$

At the negative integers:

$$\frac{H_{2n}}{z+n}+....$$

giving a residue of:

$$\sum_{n=1}^{\infty}\frac{H_{2n-1}}{n(6n+1)}$$

I do not think it quite adds up though. I could have easily went astray in all of that, though.

$\endgroup$
  • $\begingroup$ What is your question? $\endgroup$ – Wojowu Dec 6 '15 at 14:34
  • $\begingroup$ What is your question exactly? $\endgroup$ – Jan Dec 6 '15 at 14:45
  • 2
    $\begingroup$ The question is (I guess) : is there a closed form for the sum ? $\endgroup$ – Claude Leibovici Dec 6 '15 at 14:47
  • $\begingroup$ Yes, of course. Can anyone find a closed form for said Euler sum?. $\endgroup$ – Cody Dec 6 '15 at 16:15
  • $\begingroup$ I emended the heading to be more specific about what I was asking. $\endgroup$ – Cody Dec 6 '15 at 16:19
8
$\begingroup$

This is a partial answer but I believe I am making progress. Let $S$ denote the sum. Then $$ \begin{align*} S=\sum_{n=1}^\infty \frac{H_{2n}}{n(6n+1)} &= \sum_{n=1}^\infty\frac{H_{2n}}{n}\int_0^1 x^{6n}\; dx \\ &= \int_0^1\left( \sum_{n=1}^\infty\frac{H_{2n}}{n}x^{6n}\right)dx \tag{1}\end{align*} $$

Let $f(x)=\sum_{n=1}^\infty \frac{H_n}{n}x^n$ where $|x|<1$. It can be shown that $$f(x)=\text{Li}_2(x)+\frac{1}{2}\log^2(1-x) \tag{2}$$ Then, we can write $$\begin{align*} \sum_{n=1}^\infty\frac{H_{2n}}{n}x^{6n} &= f(x^3)+f(-x^3) \\ &= \text{Li}_2(x^3)+\text{Li}_2(-x^3)+\frac{\log^2(1-x^3)+\log^2(1+x^3)}{2}\tag{3} \end{align*} $$ Substitute (3) into (1) to get $$ \begin{align*} S=\int_0^1 \left(\text{Li}_2(x^3)+\text{Li}_2(-x^3)+\frac{\log^2(1-x^3)+\log^2(1+x^3)}{2} \right) dx \tag{4} \end{align*} $$ Note that $$\begin{align*} \int_0^1\left( \text{Li}_2(x^3)+\text{Li}_2(-x^3)\right)dx &= \frac{1}{2}\int_0^1\sum_{n=1}^\infty\frac{x^{6n}}{n^2} \; dx \\ &=\frac{1}{2}\sum_{n=1}^\infty\frac{1}{n^2(6n+1)} \\ &= \frac{1}{2}\sum_{n=1}^\infty \left(\frac{1}{n^2}-\frac{6}{n}+\frac{36}{1+6n} \right) \\ &= \frac{1}{2}\left(\frac{\pi^2}{6} -6\psi_0\left(\frac{1}{6} \right)-6\gamma_0-36\right) \\ &= \frac{\pi^2}{12}+\frac{3\pi\sqrt{3}}{2}+6\log 2+\frac{9}{2}\log 3-18 \tag{5} \end{align*} $$

$$ \begin{align*} \frac{1}{2}\int_0^1\log^3(1-x^3)dx &= \frac{1}{6}\int_0^1t^{-2/3}\log^2(1-t)dt \quad (t=x^3)\\ &= \frac{1}{6}\left[\frac{\partial^2}{\partial y^2} B(x,y)\right]_{x=1/3,y=1} \\ &= \frac{\pi ^2}{8}-\frac{\sqrt{3} \pi }{2}+\frac{9}{2}+\frac{9}{8} \log^2 3-\frac{9 \log 3}{2}+\frac{1}{4} \sqrt{3} \pi \log 3-\frac{\psi_1\left(\frac{4}{3}\right)}{2}\tag{6} \end{align*} $$ Substitute (5) and (6) into equation (4) to get $$S=-\frac{27}{2}+\frac{5\pi^2}{24}+\frac{9}{8}\log^2 3+\frac{\pi\sqrt{3}}{4}(4+\log 3)+6\log 2-\frac{1}{2}\psi_1\left(\frac{4}{3} \right)+\frac{1}{2}\int_0^1 \log^2(1+x^3)dx \tag{7}$$

Now, it remains to calculate $\int_0^1\log^2(1+x^3)dx$.

ADDED

Use integration by parts, $$ \begin{align*} \int_0^1\log^2(1+x^3) dx &= \left[x\log^2(1+x^3) \right]_0^1-6\int_0^1 \frac{x^3\log(1+x^3)}{1+x^3}dx \\ &= \log^2 2-6 \int_0^1 \log(1+x^3)dx +6\int_0^1\frac{\log(1+x^3)}{1+x^3}dx\tag{8} \end{align*} $$ The first integral is "elementary": $$\int_0^1\log(1+x^3)dx = -3+\frac{\pi}{\sqrt{3}}+2\log 2 \tag{9}$$

Evaluating $\int_0^1\frac{\log(1+x^3)}{1+x^3}$ is more tedious. Let $\omega=e^{i\pi /3}$. Then note that $1+x^3=(x+1)(x-\omega)(x-\bar{\omega})$ and

$$\frac{1}{1+x^3}=\frac{1}{3}\frac{1}{1+x}+\frac{1+\bar{\omega}}{3\sqrt{3}}\frac{1}{x-\omega}-\frac{1+\omega}{3\sqrt{3}}\frac{1}{x-\bar{\omega}}$$

Expanding the logarithms and using the above partial fraction decomposition in $\int_0^1\frac{\log(1+x^3)}{1+x^3}$ will result in 9 pieces which can be evaluated in terms of dilogarithms. Ultimately, all this will give a closed form for $\int_0^1 \log^2(1+x^3)dx$ and $\sum_{n=1}^\infty\frac{H_{2n}}{n(6n+1)}$.

$\endgroup$
  • 1
    $\begingroup$ according to Mathematica, that last integral equals $18-\frac{5}{36}\pi^2+\frac{\log^23}{4}+(\log2-4)\log8+\frac{\log\frac{2187}{16}-12}{2\sqrt{3}}\pi+\text{Li}_2(-1/3)-(1+i\sqrt{3})\text{Li}_2\left(\frac{3-i \sqrt{3}}{6}\right)$ $+(1-i\sqrt{3})\text{Li}_2\left(\frac{3-i\sqrt{3}}{4}\right)-(1-i\sqrt{3})\text{Li}_2\left(\frac{3+i\sqrt{3}}{6}\right)+(1+i\sqrt{3})\text{Li}_2\left(\frac{3+ i \sqrt{3}}{4}\right)=0.0837342\dots$ $\endgroup$ – AccidentalFourierTransform Feb 12 '16 at 16:22
  • $\begingroup$ @AccidentalFourierTransform: Great! Now we have the closed form for the sum. $\endgroup$ – Shobhit Bhatnagar Feb 12 '16 at 16:28
  • $\begingroup$ Wow, look at you I&S!. It is surprising and cool to see a reply to this problem after all this time. I had forgotten about this sum. Nice indeed. btw...Shobit...is that you?. That is one wacky result Mathematica gave. Maybe it can be simplified. $\endgroup$ – Cody Feb 13 '16 at 14:40
  • $\begingroup$ @Cody: Yes, that is me. $\endgroup$ – Shobhit Bhatnagar Feb 14 '16 at 11:12
  • $\begingroup$ Wow, dude, where ya' been?. Good to hear from you. Good work! $\endgroup$ – Cody Feb 14 '16 at 14:33

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.