2
$\begingroup$

I have stumbled upon Gamma functions when dealing with Gamma distributions on my studies with basic statistics. However, I have not understood how its computation expands factorials to real and complex numbers through:

$$\Gamma(x+1)=\int_0^1(-\ln t)^x\;\mathrm dt=\int_0^\infty t^xe^{-t}\;\mathrm dt = x!$$

with x as a non-negative integer. Mine is sort of a straightforward question: your input is a non-negative integer and you get the factorials of all real and complex number between the integers as the output? If that's the point, what does it even mean to have the factorial of a real/complex number? Are they arbitrary definitions derived of the Gamma function?

P.S.: I have searched the forum and found some interesting discussions on the topic, but they're generally at a math level still inaccessible to me:

$\endgroup$
  • $\begingroup$ Indeed, they are just considered $\Gamma$ values at that point, they're no longer heavily combinatorial like with postive integer factorials. $\endgroup$ – Adam Hughes Aug 10 '14 at 17:43
  • $\begingroup$ The article "Is the Gamma function mis-defined?" (appearing too in your last link) proposed different variants. $\endgroup$ – Raymond Manzoni Aug 11 '14 at 17:49
1
$\begingroup$

For real $x$, the gamma function provides a continuous (indeed, even smooth) function that gives the factorials at the positive integers. However, you can choose points in between the positive integers and define your function to be whatever value you want there, and still find a smooth function that fits those values plus gives factorials at the positive integers. So we want some restrictions to define a "good" gamma function. One property of factorials is that $x! = x \cdot (x-1)!$ so it makes sense to insist that $\Gamma (x+1) = x \Gamma (x)$. However, we can still define an essentially arbitrary smooth function between $x=1$ and $x=2$ that takes on factorial values at those end points, and then extend the function to all positive $x$ by using the relation $\Gamma (x+1) = x \Gamma (x)$. So yes, in a sense, the values that the Gamma function assigns to non-integer positive reals are essentially arbitrary. The main reason why the Gamma function is preferred is that it appears so often in other mathematical formulas and studies.

$\endgroup$

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.