13
$\begingroup$

I saw the beta function:

$$ \frac{\Gamma(r)\Gamma(s)}{\Gamma(r+s)}= \int_0^1 t^{(r-1)}(1-t)^{(s-1)} dt $$

and got me wondering if I could do something similar the product of 3 or more gamma functions. What I mean is is there a nice form to express:

$$ \frac{\Gamma(r)\Gamma(s)\Gamma(k)}{\Gamma(r+s+k)} $$

as a nice integral?

$\endgroup$
  • $\begingroup$ $$\int_0^1\int_0^{1-x}x^{r-1}y^{s-1}(1-x-y)^{k-1}dydx$$ $\endgroup$ – Did Mar 29 '15 at 21:49
  • $\begingroup$ Can you show how you got that? Like how you knew which substitutions to make and all? $\endgroup$ – drewdles Mar 29 '15 at 21:51
  • $\begingroup$ This is one way of getting integrals of this type, for any number of arguments: en.wikipedia.org/wiki/Feynman_parametrization $\endgroup$ – Chappers Mar 29 '15 at 21:55
9
$\begingroup$

Yes. In general, let $\Sigma = \{(x_1,\ldots, x_n) \in \Bbb R^n \, |\, x_i\ge 0, \sum x_i \le 1\}$. If $\operatorname{Re}(s_i) > 0$ for all $i$, then

$$\int\cdots \int_\Sigma x_1^{s_1 - 1}x_2^{s_2 - 1}\cdots x_n^{s_n - 1}\, dx_1\cdots\, dx_n = \frac{\Gamma(s_1)\cdots \Gamma(s_n)}{\Gamma(1 + s_1 + \cdots + s_n)}.$$

This can be proven by induction.

Using the relation $\Gamma(z + 1) = z \Gamma(z)$, we represent

$$\frac{\Gamma(s_1) \cdots \Gamma(s_n)}{\Gamma(s_1 + \cdots + s_n)} = s\int \cdots \int_{\Sigma} x_1^{s_1 - 1}\cdots x_n^{s_n - 1} dx_1\ldots dx_n$$

where $s = s_1 + \cdots + s_n$.

$\endgroup$

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.