1
$\begingroup$

I get the result from Wolfram that $$\lim_{a\to \infty}a-\frac{\Gamma(a+1/2)^2}{\Gamma(a)^2}=\frac{1}{4}.$$

I am trying to prove it. It seems the Stirling's formula cannot be used here. Can anyone help me on this?

Thank you very much!

I tried to use Stirling's approximation as:

\begin{equation} \begin{aligned} a-\frac{\Gamma(a+1/2)^2}{\Gamma(a)^2} &\to a-\frac{(a-1/2)^{2a}}{(a-1)^{2a-1}}e^{-1}\\ &=a-\big(\frac{a-1+1/2}{(a-1)}\big)^{2(a-1)}e^{-1}\frac{a^2-a+1/4}{a-1}\\ &\to a-a-\frac{1}{4(a-1)} \end{aligned} \end{equation}

$\endgroup$
  • $\begingroup$ I'd have thought Stirling was the way to go. What goes wrong? $\endgroup$ – Lord Shark the Unknown Apr 6 '18 at 21:46
  • $\begingroup$ @LordSharktheUnknown I put my Stirling's try here. Thanks for your reply. $\endgroup$ – Faith Bian Apr 6 '18 at 22:03
0
$\begingroup$

Considering Stirling approximation $$\log (\Gamma (p))=p (\log (p)-1)-\frac 12 \log(p)+\frac{1}{2} \log (2 \pi )+\frac{1}{12 p}-\frac{1}{360 p^3}+O\left(\frac{1}{p^5}\right)$$ apply it to $$A=\frac{\Gamma \left(a+\frac{1}{2}\right)}{\Gamma (a)}\implies \log(A)=\log \left(\Gamma \left(a+\frac{1}{2}\right)\right)-\log (\Gamma (a))$$ and continue with Taylor expansion for large values of $a$ to get $$\log(A)=\frac{1}{2} \log \left({a}\right)-\frac{1}{8 a}+\frac{1}{192 a^3}+O\left(\frac{1}{a^4}\right)$$ $$\log(A^2)=2\log(A)= \log \left({a}\right)-\frac{1}{4 a}+\frac{1}{96 a^3}+O\left(\frac{1}{a^4}\right)$$ Continue with Taylor, using $A^2=e^{\log(A^2)}$ to get $$A^2=a-\frac{1}{4}+\frac{1}{32 a}+\frac{1}{128 a^2}+O\left(\frac{1}{a^3}\right)$$ which makes $$a-\left(\frac{\Gamma \left(a+\frac{1}{2}\right)}{\Gamma (a)} \right)^2=\frac{1}{4}-\frac{1}{32 a}-\frac{1}{128 a^2}+O\left(\frac{1}{a^3}\right)$$ which shows the limit and how it is approached.

Moreover, it is a quite good approxiamtion of the function even for small values of $a$ as shown in the table $$\left( \begin{array}{ccc} a & \text{approximation} & \text{exact} \\ 1 & 0.210938 & 0.214602 \\ 2 & 0.232422 & 0.232854 \\ 3 & 0.238715 & 0.238835 \\ 4 & 0.241699 & 0.241747 \\ 5 & 0.243438 & 0.243461 \\ 6 & 0.244575 & 0.244588 \\ 7 & 0.245376 & 0.245384 \\ 8 & 0.245972 & 0.245977 \\ 9 & 0.246431 & 0.246435 \\ 10 & 0.246797 & 0.246800 \end{array} \right)$$

$\endgroup$
1
$\begingroup$

$\Gamma$ is log-convex by the Bohr-Mollerup theorem/characterization, hence it is enough to prove $$ \lim_{n\to +\infty} n-\left(\frac{n\sqrt{\pi}}{4^n}\binom{2n}{n}\right)^2 = \frac{1}{4}.\tag{1} $$ On the other hand $$ \frac{1}{4^n}\binom{2n}{n} = \prod_{k=1}^{n}\left(1-\frac{1}{2k}\right),\qquad \left(\frac{1}{4^n}\binom{2n}{n}\right)^2=\frac{1}{4}\prod_{k=2}^{n}\left(1-\frac{1}{k}\right)\prod_{k=2}^{n}\left(1+\frac{1}{4k(k-1)}\right)\tag{2}$$ and $$\pi\left(\frac{n}{4^n}\binom{2n}{n}\right)^2=n\prod_{k>n}\left(1+\frac{1}{4k(k-1)}\right)^{-1}\tag{3}$$ by Wallis' product. We also have $$\prod_{k>n}\left(1+\frac{1}{4k(k-1)}\right)^{-1}=\exp\sum_{k>n}\left[-\frac{1}{4k(k-1)}+O\left(\frac{1}{k^4}\right)\right]=\exp\left(-\frac{1}{4n}\right)\left(1+O\left(\frac{1}{n^3}\right)\right) $$ and the RHS of the last line equals $$\left(1-\frac{1}{4n}+O\left(\frac{1}{n^2}\right)\right)\left(1+O\left(\frac{1}{n^3}\right)\right)=1-\frac{1}{4n}+O\left(\frac{1}{n^2}\right)\tag{4}$$ so the claim is proved without resorting to the full power of Stirling's approximation/inequality.

$\endgroup$
0
$\begingroup$

$\newcommand{\bbx}[1]{\,\bbox[15px,border:1px groove navy]{\displaystyle{#1}}\,} \newcommand{\braces}[1]{\left\lbrace\,{#1}\,\right\rbrace} \newcommand{\bracks}[1]{\left\lbrack\,{#1}\,\right\rbrack} \newcommand{\dd}{\mathrm{d}} \newcommand{\ds}[1]{\displaystyle{#1}} \newcommand{\expo}[1]{\,\mathrm{e}^{#1}\,} \newcommand{\ic}{\mathrm{i}} \newcommand{\mc}[1]{\mathcal{#1}} \newcommand{\mrm}[1]{\mathrm{#1}} \newcommand{\pars}[1]{\left(\,{#1}\,\right)} \newcommand{\partiald}[3][]{\frac{\partial^{#1} #2}{\partial #3^{#1}}} \newcommand{\root}[2][]{\,\sqrt[#1]{\,{#2}\,}\,} \newcommand{\totald}[3][]{\frac{\mathrm{d}^{#1} #2}{\mathrm{d} #3^{#1}}} \newcommand{\verts}[1]{\left\vert\,{#1}\,\right\vert}$

With A & S Table $\mathbf{\color{black}{6.1.47}}$ identity:

\begin{align} a^{-1/2}\,{\Gamma\pars{a + 1/2} \over \Gamma\pars{a}} & \,\,\,\stackrel{\mrm{as}\ a\ \to\ \infty}{\sim}\,\,\, 1 - {1 \over 8a} + {1 \over 128a^{2}} \\[5mm] \implies \bracks{\Gamma\pars{a + 1/2} \over \Gamma\pars{a}}^{2} & \,\,\,\stackrel{\mrm{as}\ a\ \to\ \infty}{\sim}\,\,\, a\pars{1 - {1 \over 8a} + {1 \over 128a^{2}}}^{2} \,\,\,\stackrel{\mrm{as}\ a\ \to\ \infty}{\sim}\,\,\, \bbx{a - \color{red}{1 \over 4} + {1 \over 32a}} \end{align}


$$ \bbx{a - \bracks{\Gamma\pars{a + 1/2} \over \Gamma\pars{a}}^{2} \,\,\,\stackrel{\mrm{as}\ a\ \to\ \infty}{\sim}\,\,\, \color{red}{1 \over 4} - {1 \over 32a}} $$

$\endgroup$

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.