I asked this question a couple days ago, where the solution was brought down to

$$ \lim_{N \to \infty} \left( \pi \left(N-\frac{1}{e}\right) - 2\int_{e^{-1/e}}^{N^N} \frac{\arcsin(\cos u)}{\mathrm{W}(\ln u) + 1} \frac{\mathrm{d}u}{u} \right) $$

by @user3002473. I'm stuck on the integral part,

$$ \int_{e^{-1/e}}^{N^N} \frac{\arcsin(\cos u)}{\mathrm{W}(\ln u) + 1} \frac{\mathrm{d}u}{u} $$

I'd like to find a closed form or its asymptotic behavior at $\infty$ to evaluate the limit. I was incapable of finding accurate numerical approximations via mathematica or scipy due to the rapid oscillations within the integrand. I also tried IBP, but got stuck there as well.


We have that $\arcsin(\cos u)$ is a triangle wave and the term $W(\log u)$ is approximately constant on short intervals, since it behaves like $\log\log u$. The similar integral $$ I= \int_{0}^{+\infty}\frac{\arcsin\sin u}{u}\,du $$ has a simple closed form due to a well-known Fourier series: $$ I = \frac{4}{\pi}\sum_{n\geq 0}\frac{(-1)^n}{(2n+1)^2}\int_{0}^{+\infty}\frac{\sin((2n+1)u)}{u}\,du =2\sum_{n\geq 0}\frac{(-1)^n}{(2n+1)^2}=\color{red}{2G}$$ where $G$ is Catalan's constant. By using integration by parts and this lemma you can achieve good numerical approximation of your integral.


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.