4
$\begingroup$

How to evaluate the following limit? $$ \lim_{m\to\infty}m\left[ \left(\sum_{n=1}^{m}\frac{1}{n^2}\right)^{\pi^2/6}-{(\pi^2/6)}^{\sum_{n=1}^{m}\frac{1}{n^2}} \right] $$

This limit is of the form $\infty \cdot 0$. I generally solve such problem by taking the term that equals $0$ to the denominator and then using l'Hôpital rule. However, that won't work here.

$\endgroup$
  • $\begingroup$ is that second summation really in the exponent? $\endgroup$ – T_M Nov 16 '18 at 18:38
  • 1
    $\begingroup$ It probably helps to know that $\sum_{n=1}^m\frac1n-\ln(m)\to\gamma$ where $\gamma$ is a constant. The Euler Mascheroni constant to be precise. $\endgroup$ – SmileyCraft Nov 16 '18 at 18:44
  • 3
    $\begingroup$ Do you mean $1/n^2$ instead of $1/n$? Otherwise the latter part of that difference grows much faster than the former, and you have a limit of the form $\infty \cdot -\infty$. $\endgroup$ – Reese Nov 16 '18 at 18:46
  • $\begingroup$ Sorry, it is $1/n^2$, I forgot the ^2 $\endgroup$ – Arsh Nov 16 '18 at 18:48
  • 3
    $\begingroup$ Please don't use display style in title. I've edited this away. $\endgroup$ – GNUSupporter 8964民主女神 地下教會 Nov 16 '18 at 19:01
6
$\begingroup$

Write $c = \frac{\pi^2}{6} = \sum_{k=1}^{\infty} \frac{1}{k^2}$ and $\epsilon_n = \sum_{k=n+1}^{\infty} \frac{1}{k^2}$. Then

  • $ c^{c - \epsilon_n} = c^c \left[ 1 - \epsilon_n \log c + \mathcal{O}(\epsilon_n) \right] $,

  • $ (c - \epsilon_n)^c = c^c \left( 1 - \frac{\epsilon_n}{c} \right)^c = c^c \left[ 1 - \epsilon_n + \mathcal{O}(\epsilon_n^2) \right] $,

  • $\epsilon_n = \int_{n}^{\infty} \frac{dx}{x^2} + \mathcal{O}\left(\frac{1}{n^2}\right) = \frac{1}{n} + \mathcal{O}\left(\frac{1}{n^2}\right) $.

Combining altogether,

$$ n \left( c^{c-\epsilon_n} - (c-\epsilon_n)^c \right) = c^c (1 - \log c) + \mathcal{O}\left(\frac{1}{n}\right), $$

which converges to $c^c (1 - \log c) = \zeta(2)^{\zeta(2)} \left( 1 - \log \zeta(2) \right)$ as $n\to\infty$.

$\endgroup$
  • $\begingroup$ Why does your answer differ by a minus sign, have I done anything wrong? $\endgroup$ – Zacky Nov 16 '18 at 19:36
  • $\begingroup$ @Zacky, I made a mistake by switching two terms at the beginning. My answer itself is not wrong, but it computed the negative of OP's question. $\endgroup$ – Sangchul Lee Nov 16 '18 at 19:40
3
$\begingroup$

The factor $m$ can be moved into denominator as $1/m$ and then we can apply Cesaro-Stolz. The numerator of the expression after applying Cesaro-Stolz is $$(a+h) ^k-a^k-k^{a+h} +k^a$$ where $$k=\frac{\pi^2}{6},a=\sum_{i=1}^{m}i^{-2}\to k,h=(m+1)^{-2}\to 0$$ Thus $k$ is constant and $a, h$ are functions of $m$. The denominator of the expression after applying Cesaro-Stolz is $-1/m(m+1)$ and this can be replaced by $-h$. Thus we need to find the limit $$\lim_{h\to 0}\frac{a^k-(a+h)^k+k^{a+h}-k^a}{h}$$ and this is easily evaluated as $k^k\log k-k^k$ where $k=\pi^2/6$.

$\endgroup$

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.