From the amazing result by Raabe we know that $$LG_1=\int_0^1 \ln \Gamma(x)\,dx = \frac{1}{2}\ln(2\pi) = -\zeta'(0).$$

We also know that

$$LG_2 = \int_0^1 \left(\ln \Gamma(x)\right)^2\,dx = \frac{\gamma^2}{12}+\frac{\pi^2}{48}+\frac{1}{3}\gamma\ln\sqrt{2\pi}+\frac{4}{3}\left(\ln\sqrt{2\pi}\right)^2-\left(\gamma+2\ln\sqrt{2\pi}\right)\frac{\zeta'(2)}{\pi^2}+\frac{\zeta''(2)}{2\pi^2}.$$

Decimal expansion of $LG_1$ is at $\text{A075700}$ and of $LG_2$ at $\text{A102887}$.

According to several authors there is not a known closed-form of

$$\begin{align} LG_3 & = \int_0^1 \left(\ln \Gamma(x)\right)^3\,dx \\ & \approx 5.740388807229474280019571688102461462961\dots \end{align}$$

Any idea to get a closed-form of $LG_3$? A conjectured one also would be nice.


This is by no means an answer, but just too long to fit in a comment. Integration by parts yields a sort-of-interesting result. If I am doing this correctly, then you get that:

$LG_3 = [\int^1_0(\ln(\Gamma(x)))dx]·[(\ln(\Gamma(x)))]^1_0-2·\int_0^1[(\ln(\Gamma(x)))·(\frac{d}{dx}(\ln(\Gamma(x))))·(∫(\ln(\Gamma(x))dx)]dx$

Inside the integral on the right is the product of the function, its derivative, and integral, all being integrated from $0$ to $1$. Particularly there is the product of the digamma function, the log-gamma in question and the log-gamma integral being integrated.

The other term can be calculated in a closed form. If you define $F(x)=\int \ln(\Gamma(x)) dx$, and call $T=[\int^1_0(\ln(\Gamma(x)))dx]·[(\ln(\Gamma(x)))]^1_0$, the Term for which we have a closed form, then

$LG_3=T-2·\int_0^1[F(x)·F'(x)·F''(x)] dx$

Reiterating integration by parts on the new integral leads to an identity which supplies no further information about the object in question, but does provide a somewhat disappointing proof that integration by parts is true.


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.