1
$\begingroup$

How can we calculate the following ?: $$ \lim_{r \to \infty}\,\sqrt{\,{r \over 2}\,}\,\ {\Gamma\left(\,\left(r - 1\right)/2\,\right) \over \Gamma\left(\,r/2\,\right)} = 1 $$

$\endgroup$
  • $\begingroup$ Did you try using Stirling's approximation? $\endgroup$ – Theorem Oct 21 '16 at 6:11
  • $\begingroup$ @Stoke No... Can Stirling formula be applied to non-integers? But for advance, Thanks. If I use the formula, I can handle it. $\endgroup$ – kayak Oct 21 '16 at 6:15
  • $\begingroup$ The approximation is also good for the Gamma function. You can use it by converting back to factorial by $\Gamma(n)=(n-1)!$ $\endgroup$ – Theorem Oct 21 '16 at 6:18
2
$\begingroup$

By Gautschi's inequality with $x+1=\frac{r}{2}$ and $s=\frac{1}{2}$,

$$\sqrt{x}\leq \frac{\Gamma(x+1)}{\Gamma(x+s)}\leq \sqrt{x+1} \tag{1}$$ the claim immediately follows by squeezing.

$\endgroup$
  • 2
    $\begingroup$ I just saw this question, but I see that you've given the answer I was going to give. (+1) $\endgroup$ – robjohn Oct 24 '16 at 22:45
1
$\begingroup$

Hint

For this kind of problems, Stirling approximation is the key.

Consider $$A=\sqrt\frac{r}{2} \frac{\Gamma(\frac{r-1}{2})}{\Gamma(\frac{r}{2})}\implies \log(A)=\frac 12 \log(r)-\frac 12 \log(2)+\log\left(\Gamma(\frac{r-1}{2})\right)-\log\left(\Gamma(\frac{r}{2})\right)$$ Stirling approximation write $$\log\left(\Gamma(m)\right)=m (\log (m)-1)+\frac{1}{2} \left(-\log (m)+\log (2 \pi )\right)+O\left(\frac{1}{m}\right)$$ Apply to each factorial and simplify.

If you use it and continue with Taylor series for infinitely large values of $r$, you should find $$\log(A)=\frac{3}{4 r}+O\left(\frac{1}{r}\right)$$ and now, remembering that $A=e^{\log(A)}$ and Taylor again $$A=1+\frac{3}{4 r}+O\left(\frac{1}{r}\right)$$ For illustration purposes, using $r=100$, the exact value is $\approx 1.00758$ while the simple asymptotics gives $1.00750$.

$\endgroup$
0
$\begingroup$

$\newcommand{\bbx}[1]{\,\bbox[8px,border:1px groove navy]{{#1}}\,} \newcommand{\braces}[1]{\left\lbrace\,{#1}\,\right\rbrace} \newcommand{\bracks}[1]{\left\lbrack\,{#1}\,\right\rbrack} \newcommand{\dd}{\mathrm{d}} \newcommand{\ds}[1]{\displaystyle{#1}} \newcommand{\expo}[1]{\,\mathrm{e}^{#1}\,} \newcommand{\ic}{\mathrm{i}} \newcommand{\mc}[1]{\mathcal{#1}} \newcommand{\mrm}[1]{\mathrm{#1}} \newcommand{\pars}[1]{\left(\,{#1}\,\right)} \newcommand{\partiald}[3][]{\frac{\partial^{#1} #2}{\partial #3^{#1}}} \newcommand{\root}[2][]{\,\sqrt[#1]{\,{#2}\,}\,} \newcommand{\totald}[3][]{\frac{\mathrm{d}^{#1} #2}{\mathrm{d} #3^{#1}}} \newcommand{\verts}[1]{\left\vert\,{#1}\,\right\vert}$ \begin{align} &\lim_{r \to \infty}\braces{\root{r \over 2}\, {\Gamma\pars{\bracks{r - 1}/2} \over \Gamma\pars{r/2}}} = \lim_{r \to \infty}\bracks{\root{r + 1}\, {\Gamma\pars{r + 1/2} \over \Gamma\pars{r + 1}}}\qquad \pars{~{r \over 2}\ \mapsto\ r + 1~} \\[5mm] = &\ \root{2\pi}\lim_{r \to \infty}\bracks{{2^{1/2 - 2r} \over \root{r + 1}}\, {\Gamma\pars{2r} \over \Gamma^{\, 2}\pars{r}}}\qquad\qquad\qquad \pars{~\Gamma\mbox{-}Duplication\ Formula\ \mbox{and}\ Recurrence~} \\[5mm] = &\ \root{2\pi}\lim_{r \to \infty}\braces{{2^{1/2 - 2r} \over \root{r + 1}}\, {\root{2\pi}\pars{2r - 1}^{2r - 1/2}\expo{-2r + 1} \over \bracks{\root{2\pi}\pars{r - 1}^{r - 1/2}\expo{-r + 1}}^{\, 2}}} \quad\pars{~Stirling\ Asymptotic\ Expansion~} \\[5mm] = &\ \lim_{r \to \infty}\bracks{{2^{1/2 - 2r} \over \root{r + 1}}\, {\pars{2r - 1}^{2r - 1/2}\expo{-2r + 1} \over \pars{r - 1}^{2r - 1}\expo{-2r + 2}}} = \lim_{r \to \infty}\braces{{r^{1/2} \over \root{r + 1}}\, {\bracks{1 - \pars{1/2}/r}^{2r} \over\pars{1 - 1/r}^{2r}\expo{}}} \\[5mm] & = {\pars{\expo{-1/2}}^{2} \over \pars{\expo{-1}}^{2}\expo{}} = \bbox[#ffd,10px,border:1px dotted navy]{1} \qquad\qquad\qquad\qquad\qquad\qquad \pars{~\mbox{Note that}\ \lim_{n \to \infty}\pars{1 + {x \over n}}^{n} = \expo{x}~} \end{align}

We used the $\ds{\Gamma}$-Duplication Formula to get rid of $\ds{1/2}$-factors. In this way, the $\ds{\Gamma^{\, 2}\pars{r}}$ function in the denominator is quite convenient.

$\endgroup$

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.