Proving $\dfrac{4}{\pi}\sum_{k = 1}^{\infty}\dfrac{(-1)^{k - 1}k}{B_{2k}(2k - 1)}\left(\dfrac{\log n}{2\pi}\right)^{2k - 1} = \dfrac{n}{\text{In}n}$

118 Views Asked by At

How can we go about proving: As $n \to \infty$

$$\dfrac{4}{\pi}\sum_{k = 1}^{\infty}\dfrac{(-1)^{k - 1}k}{B_{2k}(2k - 1)}\left(\dfrac{\log n}{2\pi}\right)^{2k - 1} \approx \dfrac{n}{\text{ln} \text{ } n}\tag 1\label{1}$$

where $B_m$ denotes the $m$-th Bernoulli number for all $m \geqslant 2$. I have tried this problem for quite considerable amount of time. Indeed, I was able to simplify the LHS more using some clever substitutions and some analytic number theory.

I was able to prove that: As $n \to \infty$

$$\int_{0}^{\infty}\dfrac{(\log n)^\tau d\tau}{\Gamma(\tau + 1)\zeta(\tau + 1)\tau} \approx \dfrac{4}{\pi}\sum_{k = 1}^{\infty}\dfrac{(-1)^{k - 1}k}{B_{2k}(2k - 1)}\left(\dfrac{\log n}{2\pi}\right)^{2k - 1} \tag 2\label{2}$$

however I'm not quite satisfied with my proof for the same. I am wondering if there's a nice way to prove $\eqref{1}$ and $\eqref{2}$ that doesn't make use of complex analysis. Your help for proving $\eqref{2}$ would be highly appreciated. I'm looking for a detailed real analytic answer.

Your help would be highly appreciated. Thanks.

1

There are 1 best solutions below

6
On

No idea of what you are trying to do.

The asymptotic of Bernouilli numbers can be obtained directly from that $\frac{x}{e^x-1}-\frac{2i\pi}{x-2i\pi}-\frac{-2i\pi}{x+2i\pi}$ is analytic on a larger disk $|x|<4\pi$, but for clarity it is convenient to know that

$$\zeta(2k) = (-1)^{k+1}\frac{(2\pi)^{2k}B_{2k}}{2(2k)!} = 1+O(2^{-2k}))$$

Which gives that

$$\sum_{k \ge 1} \frac{2(-1)^{k+1}}{(2\pi)^{2k}B_{2k}} x^{2k}\frac{k}{2k-1}=\sum_{k\ge 1} \frac{x^{2k}}{\zeta(2k) (2k)!}\frac{k}{2k-1}=\sum_{k\ge 1} \frac{x^{2k}}{ (2k)!}\frac{k}{2k-1} (1+O(2^{-2k})$$ as $x\to \infty$ it is $$=\frac14 e^x+O(e^x/x)+O(e^{x/2})$$