1
$\begingroup$

I want to show that

$$ f(t) = \int_{\mathbb{R}} e^{x(t+1)} \left(1 + \frac{e^x}{k} \right)^{-(k+1)}dx = k^{t+1} \frac{\Gamma(k-t)\Gamma(1+t)}{\Gamma{(k+1)}}$$

Here, $k,t \in \mathbb{R}$.

I have tried a few things, but none worked out. First, I noticed that the quotient with all the gamma terms can arises from the beta function with $x=k-t$, $y=1+t$, and $x+y=k+1$. But I haven't been able to transform my integral into anything of the required form.

Next, I tried using the fact that we can write $\Gamma(x)\Gamma(y)$ as a double integral with limits of integration $[0,\infty)$, but that did not get me anywhere.

I would greatly appreciate any advice on how to solve this problem.

$\endgroup$
  • $\begingroup$ Shouldn't there be restrictions on $k$ and $t$ such as $k > t > -1$? Otherwise, the right hand side would not make sense when $k = t$. $\endgroup$ – kobe Jan 24 '16 at 17:26
  • $\begingroup$ Good point. There are no restrictions on $k,t$, as far as I know. $\endgroup$ – cascada Jan 24 '16 at 17:30
3
$\begingroup$

Consider the change of variables $\frac{e^x}{k}=z , dx=k\frac{dz}{z}$ , the integral becomes

$$ I(k,t)=k^{t+1}\int_{\mathbb{R\geq0}}\frac{z^{t}}{(1+z)^{k+1}} $$

which equals a integral representation of the Beta function

$$ I(k,t)=k^{t+1}B(t+1,k-t)=k^{t+1}\frac{\Gamma(k-t)\Gamma(t+1)}{\Gamma(k+1)} $$

as long as $t+1>0$ and $k-t>0$

$\endgroup$
  • $\begingroup$ Thanks. I actually tried this change of variables, but rejected it over a mixup I had with with what values $k,t$ should take. Thanks for clearing that up. $\endgroup$ – cascada Jan 24 '16 at 17:54

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.