2
$\begingroup$

I need the following Laplace transform to solve the Differential Equation

$$\int_{0}^{\infty}e^{-pt}\sin\sqrt{t}\, dt, \quad \text{where} \ \ \ p>0$$

I tried Integration by parts after substituting $t=x^2$, but didn't work.


\begin{align} \int_{0}^{\infty}e^{-pt}\sin\sqrt{t}dt &\overset{t=x^2}= \int_{0}^{\infty}e^{-px^2}2x\sin xdx \ = \ \text{I}\\ & = \sin x \ \frac{e^{-px^2}}{-p} + \frac{1}{p}\int_{0}^{\infty}e^{-px^2}\cos xdx \\ & = \sin x \ \frac{e^{-px^2}}{-p} + \frac{1}{p}\left(-e^{-px^2}\sin x - \int_{0}^{\infty}e^{-px^2}(-2px)(-\sin x)dx\right) \\ & = -\sin x \ \frac{e^{-px^2}}{p} + \frac{1}{p}\left(-\sin xe^{-px^2} - p\int_{0}^{\infty}e^{-px^2}(2x)(\sin x)dx\right) \\ & = -\sin x \ \frac{e^{-px^2}}{p} + \frac{1}{p}\left(-\sin xe^{-px^2} - p\text{I}\right) \\ \end{align} \begin{align} & \text{I} = -\sin x \ \frac{e^{-px^2}}{p} + \frac{1}{p}\left(-\sin xe^{-px^2} - p\text{I}\right) \\ & 2\text{I} = -2\sin x \ \frac{e^{-px^2}}{p}\Big|_0^\infty \\ & 2\text{I} = 0 \end{align}

$\endgroup$
  • 1
    $\begingroup$ Have you tried using Moivre's relationship? $\sin(\theta)=\frac{ e^{i\theta}-e^{-i\theta}}{2i}$, It could help $\endgroup$ – RScrlli Dec 31 '18 at 12:30
  • 1
    $\begingroup$ Substitute $\sqrt{t}=y$, to get $$ \int_{0}^{\infty}e^{-pt}\sin\sqrt{t}~dt=2\int_{0}^{\infty}e^{-py^2}y\sin y~dy=2\mathrm{Im}\int_{0}^{\infty}e^{-py^2+\mathrm{i}y}y~dy\ , $$ where Im stands for imaginary part. $\endgroup$ – Pierpaolo Vivo Dec 31 '18 at 12:32
  • $\begingroup$ Why did you delete your answer? It was a great method, and if you're not familiar with contour integration, it's fine to just leave that link to justify your calculations. $\endgroup$ – Zachary Dec 31 '18 at 15:04
  • $\begingroup$ Are you sure that is fine? I really don't have knowledge for that. $\endgroup$ – LeBlanc Dec 31 '18 at 15:05
  • 1
    $\begingroup$ Well I'm not a moderator, but I would assume that posting such an answer, which in this case is mathematically correct and relevant, would be acceptable - even if you leave a link to justify a certain computation. The answer may help somebody who may not see it in the comments. $\endgroup$ – Zachary Dec 31 '18 at 15:13
2
$\begingroup$

$$I=\int_0^\infty \sin\left(\sqrt t \right)e^{-pt}dt\overset{\sqrt t=x}=2\int_0^\infty x\sin x e^{-px^2}dx=\int_0^\infty \sin x\left(-\frac1pe^{-px^2}\right)'dx$$ $$\overset{IBP}=\underbrace{-\frac1p\sin xe^{-px^2}\bigg|_0^\infty}_{=0}+\frac1p\int_0^\infty \cos x\,e^{-px^2}dx=\frac1{2p}\int_{-\infty}^\infty \cos x\,e^{-px^2}dx$$ We can also make use of the fact that $\cos x$ is the real part of $e^{ix}=\cos x+i\sin x$. $$I=\frac1{2p}\Re \left(\int_{-\infty}^\infty e^{ix}e^{-px^2}dx\right)=\frac1{2p}\Re \left(\int_{-\infty}^\infty e^{\large-(px^2-ix)+\frac{1}{4p}-\frac{1}{4p}}dx\right)$$ $$=\frac1{2p}\Re \left(\int_{-\infty}^\infty e^{-\large\left(\sqrt{p}x-\frac{i}{2\sqrt p}\right)^2 -\frac{1}{4p}}dx\right)=\frac{e^{-\frac{1}{4p}}}{2p}\Re \left(\int_{-\infty}^\infty e^{-\large\left(\sqrt{p}x-\frac{i}{2\sqrt p}\right)^2}dx\right)$$ Substituting $\,\displaystyle{\sqrt{p}x-\frac{i}{2\sqrt p}=t\Rightarrow dx=\frac{dt}{\sqrt p}}$ gives: $$I=\frac{e^{-\frac{1}{4p}}}{2p} \frac{1}{\sqrt p}\Re\left(\int_{-\infty-\large\frac{i}{2\sqrt p}}^{\infty-\large\frac{i}{2\sqrt p}} e^{-t^2}dt\right)=\frac{e^{-\frac{1}{4p}}}{2p} \frac{1}{\sqrt p}\Re\left(\int_{-\infty}^\infty e^{-t^2}dt\right)=\frac{e^{-\frac{1}{4p}}}{2p} \sqrt{\frac{\pi}{p}}$$ For the last line see here and here.

$\endgroup$
  • 1
    $\begingroup$ Careful there. After your substitution $\sqrt{p}x-i/2\sqrt{p}=t$, your new bounds are $\infty - i/2\sqrt{p}$, and $-\infty - i/2\sqrt{p}$. You'd have to show that the extra imaginary part does not matter in the calculation of the integral. $\endgroup$ – Zachary Dec 31 '18 at 14:57
  • 1
    $\begingroup$ Yup. Are you familiar with Contour integration? Use Cauchy's theorem, as in this post: math.stackexchange.com/questions/648043/…. $\endgroup$ – Zachary Dec 31 '18 at 14:59
  • $\begingroup$ @Zachary thanks alot! Over time I alway struggled with contour integration and I tend to rely on real analysis as much as I can. I have also done the same mistake in the past: math.stackexchange.com/q/2861708/515527 $\endgroup$ – LeBlanc Dec 31 '18 at 15:14
2
$\begingroup$

Hint: Use the expansion of $\sin$ $$\sin\sqrt{t}=\sum_{n=0}^{\infty}(-1)^n\dfrac{t^{n+1/2}}{\Gamma(2n+2)}$$ Edit: $${\cal L}(\sin\sqrt{t})=\sum_{n=0}^{\infty}(-1)^n\dfrac{\Gamma(n+3/2)}{\Gamma(2n+2)p^{n+3/2}}$$ with Legendre Duplication Formula we have $${\cal L}(\sin\sqrt{t})=\dfrac{1}{p^{3/2}}\sum_{n=0}^{\infty}(-1)^n\dfrac{\Gamma(n+3/2)}{\sqrt{\pi}^{-1}2^{2n+1}\Gamma(n+1)\Gamma(n+3/2)p^n}$$ $$=\dfrac{\sqrt{\pi}}{p^{3/2}}\sum_{n=0}^{\infty}\dfrac{\left(\frac{-1}{4p}\right)^{n}}{n!}=\color{blue}{\dfrac{\sqrt{\pi}}{p^{3/2}}e^{\frac{-1}{4p}}}$$

$\endgroup$
  • $\begingroup$ I know it can be solved by using the expansion. But I was looking for some other methods to do the same. Can you see if I applied by parts correctly? $\endgroup$ – Ajay Choudhary Dec 31 '18 at 13:30
  • $\begingroup$ How exactly would you use that? $\endgroup$ – Zachary Dec 31 '18 at 13:31
  • $\begingroup$ @AjayChoudhary you should say that in your question. $\endgroup$ – Nosrati Dec 31 '18 at 13:39
0
$\begingroup$

Hint: Use integration by parts two times. The result should be $$\frac{\sqrt{\pi}e^{-1/(4p)}}{2p^{2/3}}$$

$\endgroup$
  • $\begingroup$ I've added my attempt of Integration by parts, don't know where I'm going wrong. Have a look. $\endgroup$ – Ajay Choudhary Dec 31 '18 at 13:04

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.