4
$\begingroup$

I need to solve a certain definite integral, and several places (for example Wikipedia) I've come across the following formula:

$\displaystyle\int_0^{\infty}\frac{x^mdx}{(x^n+a^n)^r}=\frac{(-1)^{r-1}\pi\ a^{m+1-nr}\ \Gamma\big((m+1)/n\big)}{n\sin((m+1)\pi/n)(r-1)!\ \Gamma\big((m+1)/n-r+1\big)}$

with the sole condition $0<m+1<nr$. However this is seems to me undefined for $(m+1)/n=3,5,7,\ldots$ because of the sine, which are exactly the cases I need. Since this is not contained in the conditions, am I mistaken? And if not how would you go about solving the integral?

As an example mathematica gives me the answer $\frac{\Gamma(A-3)}{\Gamma(A)}$ when I use $m=5,\ a=1,\ n=2,\ r=A$, which is one of the cases I need.

$\endgroup$
0
2
$\begingroup$

You are not mistaken - rather, Wikipedia does not give the answer in its simplest form. It is actually $$\int_{0}^{\infty}\frac{x^mdx}{(x^n+a^n)^r}=a^{m+1-nr}\frac{\Gamma\left(\frac{m+1}{n}\right)\Gamma\left(r-\frac{m+1}{n}\right)}{n\,\Gamma(r)}.\tag{1}$$ The formula you have seen before is obtained from this one using gamma function reflection relation $\Gamma(z)\Gamma(1-z)=\frac{\pi}{\sin\pi z}$ with $z=r-\frac{m+1}{n}$. However, (1) is ok for your parameter values.

$\endgroup$
1
$\begingroup$

You can consider this expression valid as a limit. Let's denote the RHS of your expression $I$. Then for e.g. $r=10,m=3,n=0.8,a=1$ you have: $$\int_0^\infty \frac{x^mdx}{(x^n+a^n)^r}=\lim_{m\to3}I=\frac{1}{504}$$, although substituting $m=3$ directly wouldn't work.

$\endgroup$
3
  • $\begingroup$ Hmmm I don't completely understand, do you mean ((m+1)/n) -> 3? Because using ((3+1)/0.8) would be well defined anyway. Also it seems to me for the limit to exist the denominator should somehow also tend to zero, which I can't see how would happen. $\endgroup$ – jorgen May 9 '13 at 17:15
  • 1
    $\begingroup$ No, the point is that as $m\to3$, $\sin((m+1)\pi/n)\to0$. Thus you have a $0$ in denominator. But at the same time, $\Gamma$ function in that same denominator becomes infinite. So the limit is needed to solve the $0\cdot\infty$ indeterminate, which appears to be finite. $\endgroup$ – Ruslan May 9 '13 at 17:18
  • $\begingroup$ Oh I see, my mistake, thanks. $\endgroup$ – jorgen May 9 '13 at 17:21

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.