I remember seeing a closed form for the series : $$\sum_{k=0}^{\infty}\frac{\Gamma(k+z)\psi(k+1)}{k!}x^{k}\;\;\;\; \left | x \right |<1$$ But i don't seem to recall what it was . I tried replacing the gamma function with its Euler's integral , and then changing the order of summation and integration, using the fact that : $$\sum_{k=0}^{\infty}\frac{\psi(k+1)}{k!}y^{k}=e^{y}\left(\log y +\Gamma(0,y\right)$$ Thus : $$\sum_{k=0}^{\infty}\frac{\Gamma(k+z)\psi(k+1)}{k!}x^{k}=\int_{0}^{\infty}y^{z-1}e^{(x-1)y}\left[(\log(xy)+\Gamma(0,xy)) \right ]dy$$ $$=x^{-z}\int_{0}^{\infty}\omega^{z-1}e^{\left(1-\frac{1}{x} \right )\omega}\left[\log \omega +\Gamma(0,\omega) \right ]d\omega$$ Where $\Gamma(\cdot,\cdot)$ is the incomplete gamma function. But i don't know how to do the integral !

  • $\begingroup$ Is $z$ and integer or do you want an expression for $z$ in the reals? $\endgroup$
    – Fabian
    Jan 29 '17 at 8:13
  • $\begingroup$ no actually. $z$ is a complex parameter. $x$ is real. $\endgroup$ Jan 29 '17 at 8:17

Lemma 1. By the extended binomial theorem, $$\sum_{k\geq 0}\frac{\Gamma(k+z)}{k!}\,x^k =\frac{\Gamma(z)}{(1-x)^z}.\tag{1}$$ $\phantom{}$ Lemma 2. The series definition of the digamma function ensures $$\begin{eqnarray*} \psi(k+1) = -\gamma+\sum_{n\geq 1}\left(\frac{1}{n}-\frac{1}{n+k}\right) &=& -\gamma+\sum_{n\geq 1}\int_{0}^{+\infty}e^{-nt}(1-e^{-kt})\,dt\\&=&-\gamma+\int_{0}^{+\infty}\frac{1-e^{-kt}}{e^{t}-1}\,dt\\&=&-\gamma+\int_{0}^{1}\frac{u^k-1}{u-1}\,du.\end{eqnarray*}\tag{2} $$

By Lemma 1 and Lemma 2, $$\sum_{k\geq 0}\frac{\Gamma(k+z)\,\psi(k+1)}{k!}\,x^k = -\frac{\gamma\,\Gamma(z)}{(1-x)^z}+\Gamma(z)\int_{0}^{1}\frac{1}{u-1}\left(\frac{1}{(1-ux)^z}-\frac{1}{(1-x)^z}\right)\,du$$ and by applying integration by parts $$\sum_{k\geq 0}\frac{\Gamma(k+z)\,\psi(k+1)}{k!}\,x^k = -\frac{\gamma\,\Gamma(z)}{(1-x)^z}-\Gamma(z+1)\int_{0}^{1}\frac{x \log(1-u)}{(1-u x)^{z+1}}\,du.\tag{3}$$ Now the last integral can be seen as the derivative of a hypergeometric function (by substituting $u\mapsto(1-u)$ then exploiting $\log(u)=\left.\frac{d}{d\alpha}u^{\alpha}\right|_{\alpha=0^+}$) or expanded as a series by exploiting $\log(1-u)=-\sum_{m\geq 1}\frac{u^m}{m}$.


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.