Prove $$\int_0^1\prod_{n=1}^\infty(1-x^n)dx=\frac{4\pi\sqrt3}{\sqrt{23}}\frac{\sinh\frac{\pi\sqrt{23}}3}{\cosh\frac{\pi\sqrt{23}}2}.$$ In fact, product $(1-x^n)$ is difficult to compute, I hope you can show me some ideas, thank you!
-
1You should first transform the expression of $\prod_{n=1}^\infty(1-x^n)$ into a (famous !) series given there. – Jean Marie Aug 05 '16 at 07:45
-
@JeanMarie Ohh I didn't know about this thing. Thank you for that :) – Zubzub Aug 05 '16 at 07:46
-
$\int_0^1\prod_{n=1}^\infty(1-x^{24n})dx$ had already been considered here: http://math.stackexchange.com/questions/1541601/closed-form-of-the-integral-large-int-0-infty-e-x-prod-n-1-infty-left/1541666#1541666 – Olivier Oloa Aug 05 '16 at 07:51
-
4This is an ongoing contest problem ! See problem U382 here : https://www.awesomemath.org/wp-pdf-files/math-reflections/mr-2016-04/mr_4_2016_problems_4.pdf – Gabriel Romon Aug 05 '16 at 08:28
-
@LeGrandDODOM Yeah, thanks! – Eufisky Aug 05 '16 at 08:31
-
1Some philosophical question here. Did someone tried to figure out why the number $23$ is there? Why $23$ and not $29$? – Shashi Dec 16 '18 at 10:56
-
This is (4) in Glasser, Some Integrals of the Dedekind $\eta$-Function, specialized to $y = 23/24$. – Eric Towers Jun 02 '21 at 14:09
2 Answers
We may use the pentagonal number theorem and get $$\int_{0}^{1}\prod_{n\geq1}\left(1-x^{n}\right)dx=\sum_{k\in\mathbb{Z}}\left(-1\right)^{k}\int_{0}^{1}x^{k\left(3k-1\right)/2}dx=\sum_{k\in\mathbb{Z}}\left(-1\right)^{k}\frac{2}{3k^{2}-k+2} $$ and now we can use the well known summation formula $$\sum_{n\in\mathbb{Z}}\left(-1\right)^{n}f\left(n\right)=-\sum\left\{ \textrm{residues of }\pi\csc\left(\pi z\right)f(z)\textrm{ at }f\left(z\right)\textrm{'s poles}\right\} $$ and since we have poles at $z=\frac{1}{6}\left(1\pm i\sqrt{23}\right) $ we can conclude.

- 33,062
- 2
- 47
- 93
I didn't manage to find Marco's answer I seemed to recall, so I will just re-do the proof.
By Euler pentagonal number theorem we have $$ \prod_{k=1}^{+\infty}(1-x^k) = \sum_{m=-\infty}^{+\infty}(-1)^m x^{\frac{3m^2-m}{2}} \tag{1}$$ hence by termwise integration the initial integral equals $$ \sum_{m=-\infty}^{+\infty}\frac{2(-1)^m}{3m^2-m+2} \tag{2}$$ and the last series can be computed through the following identity, that comes from the logarithmic derivative of the Weierstrass product for the $\cosh$ function:
$$ \tanh(x) = \sum_{n\geq 0}\frac{8x}{4x^2+(2n+1)^2\pi^2} \tag{3}$$ or through $$ \sum_{n\geq 0}\frac{1}{(n+a)(n+b)}=\frac{\psi(a)-\psi(b)}{a-b} \tag{4}$$ and the dilogarithm reflection and duplication formulas. Putting all together, $$ \boxed{\int_{0}^{1}\prod_{k\geq 1}(1-x^k)\,dx = \color{red}{\frac{4\pi\sqrt{3}\sinh{\frac{\pi\sqrt{23}}{3}}}{{\sqrt{23}\cosh{\frac{\pi\sqrt{23}}{2}}}} }}$$ follows.

- 353,855
-
5
-
1+1 for the use of Euler pentagonal theorem. I was trying to use the link between eta function and elliptic integrals but did not succeed as it led to very cryptic results. – Paramanand Singh Aug 20 '16 at 08:54