1

I found this question : For $x>1$ find $$\lim\limits_{n \to \infty} n\left(n^{x-1 }\left(\zeta(x)-\sum\limits_{k=1}^n \frac{1}{k^x}\right)-\frac{1}{x-1}\right)$$

Using Stolz-Cesàro theorem:

Stolz-Cesàro theorem case $\frac{0}{0}$:- If $b_n$ is a strictly decreasing sequence and $\lim \limits_{n \to \infty} b_n = \lim \limits_{n \to \infty} a_n = 0$, and if $\lim \limits_{n \to \infty} \frac{a_{n+1}-a_n}{b_{n+1}- b_n}= l \in \overline{\mathbb{R}}$, then $\lim \limits_{n \to \infty} \frac{a_n }{b_n}=l$.

It is easy to prove that: $$\lim\limits_{n \to \infty} n^{x-1 } (\zeta(x)-\sum\limits_{k=1}^n \frac{1}{k^x})=\frac{1}{x-1}$$

However it is not easy to prove that $n^{x-1 } \left(\zeta(x)-\sum\limits_{k=1}^n \frac{1}{k^x}\right)$ is a monotone sequence and even if we proved that applying Stolz-Cesàro theorem would be very difficult.


I also want to ask for a generalization: Find a closed form for $a_r(x)$ such that:

$$n^{x-1 } \left(\zeta(x)-\sum\limits_{k=1}^n \frac{1}{k^x}\right)= \frac{1}{x-1} +\sum\limits_{r\ge 1} \frac{a_r(x)}{n^r}$$

pie
  • 8,483
  • 1
    Wolfram Mathematica gives the limit is $\frac{-1}2$. – Riemann Sep 21 '24 at 06:30
  • Use the result $$\zeta\left(s\right)-\sum_{n=1}^{N}\frac{1}{n^{s}}=s\int_{N}^{\infty}\frac{\left[x\right]-x+\frac{1}{2}}{x^{s+1}}dx+\frac{N^{s-1}}{s-1}-\frac{1}{2N^{s}},$$ you can get the limit is $\frac{-1}2$. For detail , you can refer @Marco Cantarini's answer https://math.stackexchange.com/questions/1694904/analytic-continuation-for-zetas-using-finite-sums?noredirect=1 – Riemann Sep 21 '24 at 06:45
  • @Riemann, This is strange, I thought that it would be a function of $x$. – pie Sep 21 '24 at 06:49
  • When computing the two cases $x=2$ and $x=3$, I find in both cases the limits are $\frac{-1}2$. – Riemann Sep 21 '24 at 06:52

1 Answers1

1

Too long for a comment

This is an ideal problem for the Euler-Maclaurin summation formula application $$\sum_{k=1}^n\frac1{k^x}=\sum_{k=1}^{n-1}\frac1{k^x}+\frac1{n^x}=\zeta(x)-\sum_{k=n}^\infty\frac1{k^x}+\frac1{n^x}$$ Using the formula $$\sum_{k=n}^\infty\frac1{k^x}\sim\int_n^\infty\frac{dk}{k^x}+\frac1{2n^x}+\sum_{l=1}^\infty\frac{B_{2l}}{(2l)!}\frac{x(x+1)...(x+2l-2)}{n^{x+2l-1}}$$ $$\sim\frac1{(x-1)n^{x-1}}+\frac1{2n^x}+\sum_{l=1}^\infty\frac{B_{2l}}{(2l)!}\frac{\Gamma(x+2l-1)}{\Gamma(x)}\frac1{n^{x+2l-1}}$$ and $$\left(n^{x-1 }\left(\zeta(x)-\sum\limits_{k=1}^n \frac{1}{k^x}\right)-\frac{1}{x-1}\right)\sim-\frac1{2n}+\sum_{l=1}^\infty\frac{B_{2l}}{(2l)!}\frac{\Gamma(x+2l-1)}{\Gamma(x)}\frac1{n^{2l}}$$ The limit follows.

Svyatoslav
  • 20,502
  • May I ask for books that have Euler-Maclaurin summation formula? I didn't learn it or stumble upon it in any book that I have read. – pie Sep 21 '24 at 07:03
  • 1
    As a former physicist, I would recommend MATHEMATICAL METHODS FOR PHYSICISTS by George B. Arfken and other, seventh edition, p. 567 ; 12.3 EULER-MACLAURIN INTEGRATION FORMULA. This book is an interesting and comprehensive guide to many mathematics sections and contains lots of interesting problems :) – Svyatoslav Sep 21 '24 at 07:19