I got the following division of two gamma functions with $n>0$: $$ \frac{Γ(\frac{1}{2}(n+2))}{ Γ(\frac{1}{2}(n+3))}. $$ Is there any way to further simplify this expression? I do have the feeling that there must be a way - at least to get rid of the $\frac{1}{2}$ but I am running a bit blank at the moment.
-
Is $n$ an integer? – 5xum Sep 03 '15 at 12:20
-
1Look here: https://en.wikipedia.org/wiki/Student%27s_t-distribution they have expressions essentially equivalent to yours, and they didn't simplify it. So there is unlikely to be an easy simplification. – Stephen Montgomery-Smith Sep 03 '15 at 12:24
-
1$\dfrac{\Gamma\bigg(\dfrac{n+2}2\bigg)}{\Gamma\bigg(\dfrac{n+3}2\bigg)}=\dfrac{B\bigg(\dfrac12,\dfrac{n+2}2\bigg)}{\sqrt\pi}$ – Lucian Sep 03 '15 at 15:29
-
@ 5xum, yes sorry to mention it, n is a real number with 0<n – Frank Sep 03 '15 at 15:38
-
Thanks everyone, @Stephen Montgomery-Smith may be right that the expression cannot be further simplified. I was hoping that there would be a way to, at least to remove the 0.5 from the gamma functions. Or if there would be, for example, a smart way to replace n+2=m and then simplify the expression... but I guess this is something to life with ;) ... unfortunately, n is a real number and hence, an expression for integer values -even though interesting - is not helpful in this case... thanks again everyone! – Frank Sep 04 '15 at 21:57
2 Answers
One way is to use the duplication formula:
$$\Gamma(x)\Gamma(x+{1\over 2}) = 2^{1-2x}\sqrt{\pi}\Gamma(2x)$$
Revised as a response for clarification as to why this helps.
I have been down-voted by presenting just the formula above without providing any clarification. Mea Culpa indeed. Here are some details.
There is nothing really deep here. A direct application of the duplication formula gives the following:
$$ {\sqrt{\pi}(2k+1)!\over {2^{2k+1} k!}}= \begin{cases} \Gamma({n+2\over 2}),\enspace\hbox{if $n=2k+1$} \\ \\ \Gamma({n+3\over 2}),\enspace\hbox{if $n=2k$} \\ \end{cases} $$ For $n=2k+1$ we have $\Gamma({n+3\over 2})=(k+1)!$ and for $n=2k$ we have $\Gamma({n+2\over 2})=k!$, so combining these one gets: $${\Gamma({n+2\over 2})\over \Gamma({n+3\over 2})}= \begin{cases} {\sqrt{\pi}\,n!\over {2^n ({n-1\over 2})!({n+1\over 2})!}},\enspace\hbox{if $n$ is odd} \\ \\ {2^{n+1}\cdot ({n\over 2})!({n\over 2})!\over \sqrt{\pi}(n+1)!},\enspace\hbox{if $n$ is even} \\ \end{cases} $$
Assuming $n$ to be an integer, it seems to me that factorial functions and powers of constants is conceptually simpler than the $\Gamma$ function, at least in the sense that they can be computed numerically much easier.
- 18,541
Sorry I have done a big mistake previously. Thanks to @gammatester for pointing it out.
Please check this one out. I hope that this time I have done no mistake. If I have done any, please let me know about it.
Consider two cases:-
Case-$1$
When $n$ is even, $$\frac{Γ\big(\frac{1}{2}(n+2)\big)}{ Γ\big(\frac{1}{2}(n+3)\big)}$$ $$=\frac{\big(\frac{n}{2}\big)!}{Γ\big(\frac{1}{2}(n+3)\big)}\dots (1).$$ Now, $$Γ\big(\frac{1}{2}(n+3)\big)=\big(\frac{n+1}{2}\big)Γ\big(\frac{1}{2}(n+1)\big)=\big(\frac{n+1}{2}\big)\big(\frac{n-1}{2}\big)\dots \frac{1}{2}Γ\big(\frac{1}{2}\big)$$ $$=\frac{\big(\frac{n+1}{2}\big)\big(\frac{n}{2}\big)\dots \frac{1}{2}Γ\big(\frac{1}{2}\big)}{\big(\frac{n}{2}\big)\dots 1}=\frac{\frac{(n+1)!}{2^{(n+1)}}Γ\big(\frac{1}{2}\big)}{\big(\frac{n}{2}\big)!}\dots(2).$$ Substitute the value of $Γ\big(\frac{1}{2}(n+3)\big)$ in equation $(1)$ from equation $(2)$.
Case-$2$
When $n$ is odd, $$Γ\big(\frac{1}{2}(n+3)\big)=\big(\frac{n+1}{2}\big)!\dots (3).$$ $$Γ\big(\frac{1}{2}(n+2)\big)=\big(\frac{n}{2}\big)Γ\big(\frac{n}{2}\big)=\big(\frac{n}{2}\big)\big(\frac{n-2}{2}\big)\dots \frac{1}{2}Γ\big(\frac{1}{2}\big)$$ $$=\frac{\big(\frac{n}{2}\big)\big(\frac{n-2}{2}\big)\dots \frac{1}{2}Γ\big(\frac{1}{2}\big)}{\big(\frac{n-1}{2}\big)\dots 1}=\frac{\frac{n!}{2^n}Γ\big(\frac{1}{2}\big)}{\big(\frac{n-1}{2}\big)!}\dots(4).$$ From equation $(3) \ \& \ (4)$, you can easily evaluate the desired expression, and put the value of, $Γ\big(\frac{1}{2}\big)=\sqrt \pi$.
- 2,442
-
3Your formula is wrong. It would be correct if the 3 would be a 4 because $$\frac{1}{2}(n+3) = \frac{1}{2}(n+2+1) = \frac{1}{2}(n+2) + \frac{1}{2}$$ but $$\frac{1}{2}(n+4) = \frac{1}{2}(n+2+2) = \frac{1}{2}(n+2) + 1$$ – gammatester Sep 03 '15 at 11:16
-
-