Is there a simple upper bound of the following fraction of gamma functions for any $a,b\geq1/2$: $$\left(\frac{\Gamma(a+b)}{a\Gamma(a)\Gamma(b)}\right)^{1/a}$$ An upper bound in the following form is enough: there exists a constant $C>0$ and a function $f(a,b)$ such that: $$\left(\frac{\Gamma(a+b)}{a\Gamma(a)\Gamma(b)}\right)^{1/a}\leq C\cdot f(a,b), \forall a,b\geq1/2$$

I have a guess of $f$, which is $\frac{a+b}{a}$. Is it true?



The inequality in question trivially holds if e.g. $C=1$ and $f(a,b)$ equals the left-hand side of the inequality.

More informatively, the bound with $f(a,b)=\frac{a+b}a$ indeed holds: \begin{equation} \frac{\Ga(a+b)}{a\Ga(a)\Ga(b)}\le C^a\Big(\frac{a+b}a\Big)^a \end{equation} for some real $C>0$ and all $a,b\ge1/2$. Indeed, by Stirling's formula for the gamma function, \begin{equation} \Ga(x)\asymp\frac1{\sqrt x}\,\Big(\frac xe\Big)^x \end{equation} for $x\ge1/2$, where $\asymp$ means "equals up to a universal positive constant factor". So, \begin{multline} \frac{\Ga(a+b)}{a\Ga(a)\Ga(b)}\asymp\frac1a\,\sqrt{\frac{ab}{a+b}}\frac{(a+b)^{a+b}}{a^a b^b} \ll \Big(\frac{a+b}a\Big)^a \Big(\frac{a+b}b\Big)^b \\ = \Big(\frac{a+b}a\Big)^a \Big(1+\frac ab\Big)^b < e^a\Big(\frac{a+b}a\Big)^a \end{multline} for $a,b\ge1/2$.

  • $\begingroup$ Shouldn't $\Gamma(x)\asymp\sqrt{x}(x/e)^x$? $\endgroup$ – neverevernever Oct 18 '18 at 16:21
  • $\begingroup$ @neverevernever : No, it is as I wrote. You may be confusing it with the Stirling formula for $n!=\Gamma(n+1)$. I have added the reference to Stirling's formula; see the last displayed formula in Section "Stirling's formula for the gamma function" there. $\endgroup$ – Iosif Pinelis Oct 18 '18 at 21:10

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.