While doing an assignment on solving a PDE I stumbled into the following inverse Laplace transform question (involving $\cosh$? I can't believe it). Mathematica gives no solution and I have no idea where to start, and don't know how to use $\mathcal L[f_1\star f_2] = \mathcal L[f_1] \mathcal L[f_2]$, if it's of any help.

$$ \mathcal L_{t}^{-1} \left[ \frac{1}{s(s^2+\omega^2)} \frac{\cosh(sx/a)}{\cosh(sl/a)} \right] $$

I have to change $s$ back to $t$, which will allow me to finish off this problem. Please share your ideas.


Usually these require the Residue theorem. The inverse Laplace transform is given by

$$\frac{1}{i 2 \pi} \int_{c-i \infty}^{c+i \infty} ds \frac{\exp{(s t)}}{s (s^2 + \omega^2)} \frac{\cosh{(s \frac{x}{a})}}{\cosh{(s \frac{\ell}{a})}} $$

where $c$ is a real number such that all of the singularities of the integrand lie to the left.

I will refer you to other posts to view a picture of the contour, which involves a semicircle along the line of integration oriented to the left, enclosing all of the poles of the integrand.

The poles of the integrand here are at $s=0$ and $s = \pm i \omega$. The residues corresponding to these poles (i.e., $\lim_{x \rightarrow p} (x-p) f(x)$ for integrand $f(x)$ and pole $p$) are as follows: $\frac{1}{\omega^2}$ and

$$\frac{\exp{(\pm i \omega t)} \cos{(\frac{x}{a})}}{2 \omega^2 \cos{(\frac{\ell}{a})}}$$

respectively. Adding the residues up, we get

$$\frac{1}{\omega^2} \left( 1 - \frac{\cos{(\frac{x}{a})} \cos{(\omega t)}}{\cos{(\frac{\ell}{a})}} \right ) $$

There are also poles resulting from the $\cosh{(s \frac{\ell}{a})}$ term in the integrand at $s=\pm i (2 k+1) (\pi/2) (a/\ell)$ for all integers $k$. The residues from these poles are

$$\sum_{k=-\infty}^{\infty} (-1)^{k+1} \frac{\exp{(i (2 k+1) \frac{\pi}{2} \frac{a}{\ell} t)} \cos{((2 k+1) \frac{\pi}{2} \frac{x}{\ell} t)}}{i (2 k+1) \frac{\pi}{2} \frac{a}{\ell} \left ( \omega^2 - (2 k+1)^2 (\frac{\pi}{2} \frac{a}{\ell})^2 \right )} $$

$$= 2 \sum_{k=0}^{\infty} (-1)^{k+1} \frac{\sin{((2 k+1) \frac{\pi}{2} \frac{a}{\ell} t)} \cos{((2 k+1) \frac{\pi}{2} \frac{x}{\ell} t)}}{(2 k+1) \frac{\pi}{2} \frac{a}{\ell} \left ( \omega^2 - (2 k+1)^2 (\frac{\pi}{2} \frac{a}{\ell})^2 \right )} $$

I do not know of a closed-form expression for this latter sum.

  • $\begingroup$ You may want to check to see if there are singularites of the cosh term in the denominator; I am a little suspicious that there is a pole at $\omega = 0$. $\endgroup$ – Ron Gordon Dec 26 '12 at 14:49
  • $\begingroup$ I can't comprehend becase I have no basics in complex analysis, and it will take me some time to learn it. Also, the Laplace transform of your result is off by 2h's, as in $\cosh$ to $\cos$. Have you switched $\cosh$ to $\cos$ in the middle? $\endgroup$ – Frenzy Li Dec 26 '12 at 15:03
  • $\begingroup$ @rigordonma - The forward Laplace transform of the last expression in your answer gives something that is quite different from the original problem. So something is not quite right. In particular, $\cosh z = 0$ for $z = (k + 1/2) i \pi, k \in \mathbb{Z}$, so there are additional singularities. $\endgroup$ – Hans Engler Dec 26 '12 at 15:20
  • $\begingroup$ @rlgordonma Appreciate your attention, but Hans Engler's statement is true. Please take your time to elaborate on your answer, because I'm going to sleep. $\endgroup$ – Frenzy Li Dec 26 '12 at 15:23
  • $\begingroup$ @Hans, yes, I was aware of this as soon as I hot the send button. (See my previous comment.) I will make a correction. $\endgroup$ – Ron Gordon Dec 26 '12 at 18:22

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.