4
$\begingroup$

I was able to show that the integral formula for the gamma function

$$\Gamma(s) = \int_0^\infty x^{s-1}e^{-x}dx$$

is uniformly convergent for $s$ in any closed and bounded interval using the Weierstrass theorem. Now I would like to show that the convergence is not uniform for all $s > 0$ but am not sure how to proceed.

$\endgroup$
  • 1
    $\begingroup$ It's enough to show that it tends to infinity as s tends to 0. and for this, it's enough to show that $\int_0 ^1 x^{s-1} dx$ tends to infinity as s tends to 0. $\endgroup$ – Max Freiburghaus Oct 26 '16 at 21:47
  • $\begingroup$ @WoodWorker What do you have tried so far? Please add your efforts, this will help the community to give you a good answer! $\endgroup$ – the_candyman Oct 26 '16 at 22:01
4
$\begingroup$

The integral representation converges for $Re s > 0$. As an improper integral, convergence is indeed uniform for (real) $s \in [\alpha,\beta]$. However, as you suspect, convergence is not uniform for $s \in (0,\infty)$. Problems arise due to singular behavior both at $x = 0$ when $s < 1$ and as $x \to \infty$ when $s > 1$.

To show non-uniform convergence consider

$$\Gamma(s) = \Phi(s) + \Psi(s),$$

where

$$\Phi(s) = \int_0^1 x^{s-1} e^{-x} \, dx, \\ \Psi(s) = \int_1^\infty x^{s-1} e^{-x} \, dx.$$

We can show that $\Psi(s)$ fails to converge uniformly for $s \in [\alpha, \infty).$ Note that for any sequence $(s_n)$ such that $s_n \to \infty$ with $s_n > 1 + n/\log n,$ we have

$$\left| \int_n^\infty x^{s_n-1} e^{-x} \, dx\right| > n^{s_n-1}\int_n^\infty e^{-x} \, dx = n^{s_n-1}e^{-n} > 1,$$

Thus,

$$\lim_{n \to \infty}\int_n^\infty x^{s_n-1} e^{-x} \, dx \neq 0,$$

and convergence is not uniform.

Similarly, we can show that $\Phi(s)$ fails to converge uniformly for $s \in (0, \beta).$ Note that for $s_n < 1$,

$$\left| \int_{1/2n}^{1/n} x^{s_n-1} e^{-x} \, dx\right| > e^{-1} \left(\frac{1}{n}\right)^{s_n-1}\frac{1}{2n} = \frac{1}{2en^{s_n}}.$$

Now take $s_n = 1/n$. In this case $s_n \to 0$ and $n^{s_n} = n^{1/n} \to 1$. For all $n$ sufficiently large, we have $n^{1/n} < 2$ and

$$\left| \int_{1/2n}^{1/n} x^{s_n-1} e^{-x} \, dx\right| > \frac{1}{4e}.$$

$\endgroup$

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.