7
$\begingroup$

$$ \lim_{n\to\infty}\left(\int_a^b((x-a)(b-x))^n\mathrm{d}x\right)^{\frac{1}{n}}$$ I have found the form of the integrand to be $$\frac{(n!)^2}{(2n+1)!}(b-a)^{2n+1}$$ Now, splitting the limit into two different parts i would need to solve $$\lim_{n\to\infty}(\frac{(n!)^2}{(2n+1)!} )^{\frac{1}{n}}$$ My mind went to Stirling's approximation formula but i don't think it could help.

$\endgroup$
3
  • 3
    $\begingroup$ why did you delete the question? $\endgroup$
    – supinf
    Commented Apr 30, 2021 at 8:51
  • 2
    $\begingroup$ i didn't write the limit with the factorial right.I placed the exponent wrong at first. $\endgroup$
    – Jack
    Commented Apr 30, 2021 at 9:31
  • $\begingroup$ @Jack Oh I see, that's unfortunate, you'll have to be careful next time. $\endgroup$ Commented Apr 30, 2021 at 9:55

2 Answers 2

6
$\begingroup$

You want to find $$\lim_{n\to\infty} \left(\frac{(n!)^2}{(2n+1)!} (b-a)^{2n+1}\right)^{\frac{1}{n}}$$ Clearly, $\lim_{n\to\infty}(b-a)^{\frac{2n+1}{n}} = (b-a)^2$. I claim that $$\lim_{n\to\infty} \left(\frac{(n!)^2}{(2n+1)!}\right)^\frac{1}{n} = \frac{1}{4}$$ One way to do this is, as you say, to use Stirling's approximation: $$ n! \sim \frac{1}{\sqrt{2\pi n}}\left(\frac{n}{e}\right)^n$$ $$ (2n+1)! \sim \frac{1}{\sqrt{2\pi (2n+1)}}\left(\frac{2n+1}{e}\right)^{2n+1}$$ Note that after taking $n$-th roots, the terms $\frac{1}{\sqrt{2\pi n}},\frac{1}{\sqrt{2\pi (2n+1)}}$ and the dangling factor of $\frac{2n+1}{e}$ don't really matter, so $$\lim_{n\to\infty} \left(\frac{(n!)^2}{(2n+1)!}\right)^\frac{1}{n} = \lim_{n\to\infty} \left(\frac{n}{e}\right)^2\left(\frac{e}{2n+1}\right)^2 = \frac{1}{4}$$ as advertised.

More generally, it is a fact that $\lim_{p\to\infty}\left\lVert f\right\rVert_p = \left\lVert f\right\rVert_\infty$, where $\left\lVert f\right\rVert_p = \left(\int_{[a,b]} \left|f\right|^p\right)^{\frac{1}{p}}$, and for continuous functions, $\left\lVert f\right\rVert_\infty$ is equal to $\max_{[a,b]}|f|$. Indeed, in this example, the function $f(x)=(x-a)(b-x)$ attains a maximum value $\frac{(b-a)^2}{4}$ at $x=\frac{a+b}{2}$.

$\endgroup$
7
  • 5
    $\begingroup$ What a coincidence! Just today morning I found a proof that $\lim_{n \to \infty} \left(\frac{(n!)^2}{(2n+1)!}\right)^{\frac 1n} = \frac 14$ which did not use Stirling's approximation. $\endgroup$ Commented Apr 30, 2021 at 9:13
  • $\begingroup$ @TeresaLisbon. Was it on the site ? I should be very interested. Cheers :-) $\endgroup$ Commented Apr 30, 2021 at 9:25
  • $\begingroup$ Thank you! Cauchy D'Alambert' criterion also works pretty well. $\endgroup$
    – Jack
    Commented Apr 30, 2021 at 9:33
  • $\begingroup$ @Ic2r43, i look forward to prove that fact.Does it use some integral inequalities? $\endgroup$
    – Jack
    Commented Apr 30, 2021 at 9:36
  • 1
    $\begingroup$ @ClaudeLeibovici Not on the site, but I'll give you a hint , it's basically the fact that you can create, using the binomial expansion of $(2+2)^n$, a squeeze theorem kind of bound for $\frac{(n!)^2}{(2n+1)!}$. On both sides will be functionals of $4^{-n}$. Then you take everything to the power $\frac 1n$ and notice both sides of the squeeze will go to $\frac 14$. $\endgroup$ Commented Apr 30, 2021 at 9:53
2
$\begingroup$

$$y=\left(\frac{(n!)^2}{(2n+1)!} (b-a)^{2n+1}\right)^{\frac{1}{n}}$$ $$\log(y)=\frac{1}{n}\Bigg[2\log(n!)+(2n+1)\log(b-a)-\log((2n+1)!) \Bigg]$$ Using Stirling approximation and continuing with Taylor expansions $$n\,\log(y)=\log\Big[\frac 14{(b-a)^2} \Big]n+\frac{1}{2} \log \left(\frac{\pi (b-a)^2}{4 n}\right)-\frac 3 {8n}+O\left(\frac{1}{n^2}\right)$$ $$\log(y)=\log\Big[\frac 14{(b-a)^2} \Big]+\frac{1}{2n} \log \left(\frac{\pi (b-a)^2}{4 n}\right)+O\left(\frac{1}{n^2}\right)$$ $$y=e^{\log(y)}\sim\frac 14{(b-a)^2}\exp\Bigg[\frac{1}{2n} \log \left(\frac{\pi (b-a)^2}{4 n}\right) \Bigg]$$

$\endgroup$
2
  • $\begingroup$ thank you! Combining Stirling with Taylor, i wouldn't have thought of that. $\endgroup$
    – Jack
    Commented Apr 30, 2021 at 10:15
  • $\begingroup$ @Jack. This is a classical trick when you have to compose series involving different factorials. Remember it since it is extremely useful. $\endgroup$ Commented Apr 30, 2021 at 10:38

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .