There are already proofs out there, however, the problem I am working on requires doing the proof by using the gamma function with $y = \sqrt{2x}$ to show that $\Gamma(\frac{1}{2}) = \sqrt{\pi}$.

Here is what I have so far: $\Gamma(\frac{1}{2}) = \frac{1}{\sqrt{\beta}}\int_{0}^{\infty}x^{-\frac{1}{2}}e^{-\frac{x}{\beta}}dx.$ Substituting $x = \frac{y^2}{2}, dx = ydy$, we get $\Gamma(\frac{1}{2}) = \frac{\sqrt{2}}{\sqrt{\beta}}\int_{0}^{\infty}e^{-\frac{y^2}{2\beta}}dy$.

At this point, the book pulled some trick that I don't understand. enter image description here

Can anyone explain to me what the book did above? Thanks.

  • $\begingroup$ What is $\beta?$ $\endgroup$ – Thomas Andrews Oct 19 '15 at 1:16
  • $\begingroup$ Its a parameter in Gamma distribution function. I think it is the mean time between events. $\endgroup$ – Ansdai Oct 19 '15 at 1:19

The key point is just that, through a change of variable in the integral defining the $\Gamma$ function: $$ \Gamma\left(\frac{1}{2}\right) = \int_{-\infty}^{+\infty}e^{-x^2}\,dx $$ but the integral in the RHS is well-known, it can be computed by using Fubini's theorem and polar coordinates. An alternative proof: $$ \int_{0}^{1}\sqrt{1-x^2}\,dx = \frac{\pi}{4} $$ since the LHS is the area of a quarter-circle. If we set $x=\sqrt{z}$ we have: $$ \frac{1}{2}\int_{0}^{1}z^{-1/2}(1-z)^{1/2}\,dz = \frac{1}{2}\,B\left(\frac{1}{2},\frac{3}{2}\right) = \frac{\Gamma\left(\frac{1}{2}\right)^2}{4}$$ by the properties of the Euler beta function and the identity $\Gamma(\alpha+1)=\alpha\,\Gamma(\alpha)$.


It's using the Gaussian integral or the fact that the area under normal density is 1. Since you have the statistics tag, I'm guessing the latter should be familiar already.


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.