I would like to compute the Fourier transform of the product of $\tanh(x)$ and the Heaviside step function $H(x)$, i.e.

$$\int_{-\infty}^{\infty} H(x)\tanh(x)e^{-ikx}dx = \int_{0}^{\infty} \tanh(x)e^{-ikx}dx$$

For the Fourier transform of $\tanh(x)$ alone, I have read that the use of differentiation:

$$ik\mathcal{F}\left[ \tanh(x)\right](k) = \mathcal{F}\left[ \text{sech}^2(x)\right](k)$$

is a possible way of proceding, as the integral does not exist as a classical Riemann integral. I have then read derivations of the Fourier transform of $\text{sech}^2(x)$ using contour integration.

Adapting for the new problem I have supposed:

$$ik\mathcal{F}\left[ H(x)\tanh(x)\right](k) = \mathcal{F}\left[H(x) \text{sech}^2(x)\right](k)$$

However, I cannot work out how to adapt contour integration to the half line integral from 0 to $\infty$ instead.

Can someone explain how to compute the Fourier transform either by the method outlined above or a different method?

Thank you!

  • $\begingroup$ You cant try to add and subtract $\frac{1}{2} sech^2 (x)$ inside the fourier transform. Then you can pull out terms that look more like the Fourier transform of $sech^2$ and a Hilbert transform. $\endgroup$
    – AHusain
    Apr 23, 2020 at 23:01
  • $\begingroup$ The regular part is $$\int_{\mathbb R} (\tanh(x) H(x) - H(x)) \, e^{-i k x} dx = \frac 1 2 \psi {\left( \frac {i k} 4 + \frac 1 2 \right)} - \frac 1 2 \psi {\left( \frac {i k} 4 + 1 \right)},$$ which leaves the singular part $\mathcal F[H](k)$. $\endgroup$
    – Maxim
    Apr 24, 2020 at 7:10
  • $\begingroup$ What is $\psi$ here? $\endgroup$
    – mshaw
    Apr 28, 2020 at 20:15
  • $\begingroup$ $\psi$ is the polygamma function. $\endgroup$
    – Maxim
    May 6, 2020 at 20:50

1 Answer 1


notice that: $$I=\int_{-\infty}^\infty \text{H}(x)\tanh(x)e^{-ikx}dx=\int_0^\infty\frac{e^{2x}-1}{e^{2x}+1}e^{-ikx}dx=\int_0^\infty\frac{u^{-ik}(u-1)}{u^2+1}$$ now notice that: $$I=\int_0^\infty\frac{u^{1-ik}}{u^2+1}-\frac{u^{-ik}}{u^2+1}du$$ if we let: $$J(a)=\int_0^\infty\frac{u^a}{u^2+1}du=\frac{1}{2}\int_1^\infty(v-1)^{\frac{a-1}{2}}v^{-1}dv=\frac{(-1)^{\frac{a-1}{2}}}{2}\int_1^\infty(1-v)^{\frac{a-1}{2}}v^{-1}dv$$ $$J(a)=\lim_{z\to\infty}\frac{(-1)^{\frac{a-1}{2}}}{2}\left[B\left(z;0,\frac{a+1}{2}\right)-B\left(1;0,\frac{a+1}{2}\right)\right]$$ and we know that: $$I=J(1-ik)-J(-ik)$$ where $B$ is the incomplete beta function

  • $\begingroup$ Shouldn't it read $\int_0^\infty\frac{u^{-ik}(u-u^{-1})}{u^2+1}du$ at the end of the first line? $\endgroup$
    – mshaw
    Apr 28, 2020 at 13:41

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.