1

Can someone explain the following asymptotics, through Watson’s Lemma or through another argument?

\begin{align} \int^\infty_0 e^{-(r+1)u}\cos\left(\frac{\pi}{2}e^{-u}\right)\,du \sim \frac{\pi}{2(r+1)^2},\qquad r\to \infty. \end{align}

Olivier Oloa
  • 122,789
  • 2
    what about reading up something about Watsons Lemma for yourself (wikipedia maybe) and show some own effort? we are no math chatbots here, our time is valuable – tired Jan 01 '18 at 14:16
  • I have tried but I’m not able to understand anything on wikipedia and there are no videos explaining it. I’m currently studying in 12th class we aren’t taught this in school either. @tired – Varun Maddipati Jan 01 '18 at 14:18
  • Because of this? –  Jan 01 '18 at 14:18
  • Yes, I wasn’t able to understand that part in any of the proofs. @Rohan – Varun Maddipati Jan 01 '18 at 14:19

2 Answers2

2

By just integrating by parts twice one gets, as $r \to \infty$, $$ \begin{align} &\int^\infty_0 e^{-(r+1)u}\cos\left(\frac{\pi}{2}e^{-u}\right)\,du \\\\&=\left[ \frac{e^{-(r+1)u}}{-(r+1)}\cos\left(\frac{\pi}{2}e^{-u}\right)\right]_0^\infty +\frac{\pi}{2(r+1)}\int^\infty_0 e^{-(r+2)u}\sin\left(\frac{\pi}{2}e^{-u}\right)\,du \\\\&=\color{red}{0}+\frac{\pi}{2(r+1)}\left(\left[ \frac{e^{-(r+2)u}}{-(r+2)}\sin\left(\frac{\pi}{2}e^{-u}\right)\right]_0^\infty -\frac{\pi}{2(r+2)}\int^\infty_0 e^{-(r+3)u}\cos\left(\frac{\pi}{2}e^{-u}\right)\,du \right) \\\\&=\frac{\pi}{2(r+1)(r+2)}+o\left(\frac1{(r+1)^2}\right) \\\\&=\frac{\pi}{2(r+1)^2}+o\left(\frac1{(r+1)^2}\right). \end{align} $$

Olivier Oloa
  • 122,789
0

If we want to make it match the formula for Watson's lemma on wikipedia, we start by writing

$$ \int^\infty_0 e^{-(r+1)u}\cos\left(\frac{\pi}{2}e^{-u}\right)\,du = \int^\infty_0 e^{-ru} e^{-u}\cos\left(\frac{\pi}{2}e^{-u}\right)\,du = \int^\infty_0 e^{-ru} \varphi(u)\,du, $$

where

$$ \varphi(u) := e^{-u}\cos\left(\frac{\pi}{2}e^{-u}\right). $$

Now $\varphi(0) = 0$ and $\varphi'(0) = \pi/2$, so we can write

$$ \varphi(u) = ug(u), $$

where

$$ g(u) = u^{-1} e^{-u}\cos\left(\frac{\pi}{2}e^{-u}\right) $$

and $g(0) = \pi/2 \neq 0$. In the notation of wikipedia we have $\lambda = 1$.

Thus, taking only the first term ($n=0$) of the asymptotic expansion from the article, we get

$$ \int^\infty_0 e^{-(r+1)u}\cos\left(\frac{\pi}{2}e^{-u}\right)\,du \sim \frac{g^{(0)}(0)\, \Gamma(1+0+1)}{0!\, r^{1+0+1}} = \frac{\pi}{2r^2} $$

as $r \to \infty$ since

$$ g^{(0)}(0) = g(0) = \pi/2 \qquad \text{and} \qquad \Gamma(2) = 1! = 1. $$


To conclude that

$$ \int^\infty_0 e^{-(r+1)u}\cos\left(\frac{\pi}{2}e^{-u}\right)\,du \sim \frac{\pi}{2(r+1)^2} $$

we just need to note that

$$ \frac{\pi}{2r^2} \sim \frac{\pi}{2(r+1)^2} $$

as $r \to \infty$.