2
$\begingroup$

I am wondering if someone could tell me whether or not the following integral has a closed form representation:

$$\int_0^1 t^n\log\Gamma(t+a)dt$$

In Srivastava's and Choi's wonderful book Zeta and q-Zeta Functions and Associated Series and Integrals, they give a closed form for the following Digamma integral:

$$\int_0^1 t^n\psi(t)dt$$

and the also list specific cases for the integral I am searching for, but not the general case.

The formulae for the integrals containing $n>2$ become extremely complex, which makes me wonder whether a closed form exists but I was hoping someone could settle this question for me.

$\endgroup$
  • $\begingroup$ Almost certainly not. $\endgroup$ – Alex M. Jan 6 '16 at 16:34
  • 2
    $\begingroup$ The 1998 paper by Adamchik "Polygamma functions of negative order" may interest you. The answer is given in function of $\psi^{(-n)}(t)$ (integrals of polygamma or "Negapolygamma" in Adamchik 'parlance'). An example using Alpha. Another paper by Espinosa and Moll. $\endgroup$ – Raymond Manzoni Jan 6 '16 at 17:02
  • $\begingroup$ Raymond: Thank you so much for the references, I will take a look! $\endgroup$ – FofX Jan 6 '16 at 17:18
  • $\begingroup$ Glad it interested you @FofX. Note that Gosper's 'negapolygamma' is in fact defined by $$\psi^{(-n-2)}=(z):=\frac 1{n!}\int_0^z (z-t)^{n}\log\Gamma(t)\,dt$$ $\endgroup$ – Raymond Manzoni Jan 6 '16 at 17:26
1
$\begingroup$

This is purely guessing. For integer $n$, we seem to have something like $$ \int_0^1 t^n\log\Gamma(t+a)dt = -n! \psi^{(-n-2)}(a)+n!\psi^{(-n-2)}(1+a)-\frac{n!}{1!}\psi^{(-n-1)}(1+a)+\frac{n!}{2!}\psi^{(-n)}(1+a)-\frac{n!}{3!}\psi^{(-n+1)}(1+a)+\frac{n!}{4!}\psi^{(-n+2)}(1+a)-\cdots $$ so I suspect that $$ I(a,n)=\int_0^1 t^n\log\Gamma(t+a)dt = n!(-1)^{n+1}\left( \psi^{(-n-2)}(a) -\sum_{k=0}^n \frac{(-1)^k}{k!}\psi^{(-n-2+k)}(1+a) \right) $$ this seems to work numerically for integer $n\ge 0$ and any $a$ I have tried. This gives some sort of closed forms, for example $$ I(1,1) = -\frac{1}{4} - 2 \ln A + \psi^{(-3)}(1)\\ I(1,2) = \frac{1}{4}\ln\left(\frac{2\pi}{A^4}\right)\\ $$ with $A$ the Glaisher–Kinkelin constant.

$\endgroup$
  • 1
    $\begingroup$ Yes that looks good to me. Awhile after asking this question, I came across a semi-closed form in Henri Cohen's Number Theory Vol 2 (pg. 130). It is essentially expressible in terms of the Hurwitz zeta function and it's derivative at negative integers. $\endgroup$ – FofX Jul 26 '17 at 0:54

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.