There was a similar question in the past, which was not resolved. The Laplace Transform pair is well known: $$ \mathcal{L}_{t \mapsto s}: \frac{e^{-\frac{x^2}{4t}}}{2t} \div K_0( x \sqrt{s}) $$ Is it possible to give meaning to the Inverse Laplace Transform of the other modified Bessel function $I_0(\sqrt{s})$? Since $s$ is complex a complex rotation $ I_0(\sqrt{s}) \mapsto J_0(\sqrt{s})$, which is oscillating, so may be integration along the Browmwitch contour is possible.

Explanations are welcome.

  • $\begingroup$ I just changed \frac{{{ e}^{-\frac{{{x}^{2}}}{4 t}}}}{2 \, t} to \frac{e^{-\frac{x^2}{4t}}}{2t}. All the purposeless complications in the former are absurd. $\endgroup$ – Michael Hardy Jul 9 '18 at 0:47

The integral of $I_0(\sqrt {\sigma + i \omega}) e^{t(\sigma + i \omega)}$ from $\omega=-A$ to $\omega=A$ will behave as a subexponentially growing factor times an oscillating factor, thus the Bromwich integral diverges. Roughly speaking, this is because the absolute value of $I_0(\sqrt {\sigma + i \omega})$ grows as $|\omega|^{-1/4} \exp(\sqrt {|\omega|/2})/\sqrt {2\pi}$, while the oscillation frequency does not grow fast enough to compensate for that.

If the Laplace transform is defined through the Fourier transform, the inverse transform will be given by $$\mathcal L_{s \to t}^{-1}[F] = e^{\sigma t} \mathcal F_{\omega \to t}^{-1}[F], \\ s = \sigma + i \omega,$$ showing that the inverse transform of $I_0(\sqrt s)$ doesn't exist in terms of tempered distributions either.

  • $\begingroup$ How did you get the last claim? Can you elaborate? $\endgroup$ – user48672 Jul 9 '18 at 15:09
  • $\begingroup$ The reasoning is that $F(\omega) = I_0(\sqrt {\sigma + i \omega})$ does not induce a tempered distribution, because of the subexponentially growing factor. $\endgroup$ – Maxim Jul 9 '18 at 16:15

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.