MathOverflow is a question and answer site for professional mathematicians. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Referring to a question I posted on MS, I post it here, as I didn't get an answer:

By analogy with the Jacobi–Anger expansion, one expects that $e^{iz\cot(x)}$ has a Fourier expansion of the form : $$e^{iz\cot(\theta)}=\sum_{n=-\infty}^{\infty}\Lambda_{n}(z)e^{in\theta}$$ $\Lambda_{n}(z)$ is given by: $$\Lambda_{n}(z)=\frac{1}{\pi}\int_{-\pi/2}^{\pi/2}e^{iz\cot(\theta)-in\theta}d\theta$$ A simple calculation yields: $$\Lambda_{n}(z)=\frac{2}{n\pi}\sin\left(\frac{\pi n}{2} \right )\sum_{m=0}^{\infty}\frac{z^{m}}{m!}\text{F}_{1}\left(-\frac{n}{2};-m,m;1-\frac{n}{2};1,-1 \right )$$

Where $\text{F}_{1}(\alpha;\beta,\beta^{'};\gamma;x,y)$ is the Appell Hypergeometric Function. Now, I have two questions:

1-For a purely imaginary $z$,$\;\;$ $e^{iz\cot(\theta)}$ has essential singularities at $\theta=\pm n\pi$. How is that reflected in the Fourier expansion?

2-Can we express the infinite sum in terms of other special functions?

share|cite|improve this question
up vote 1 down vote accepted

I will try to answer your two questions in reverse order.

$2.$ For real $z$ your Fourier coefficient


can be expressed in terms of elementary functions for even $n$ and in terms of the special functions $I_{m}(z)$ (modified Bessel function of the first kind) and $L_{-m}(z)$ (modified Struve function) for odd $n$. The first few Fourier coefficients are


$$\Lambda_1(z)=\Lambda_{-1}(-z)=z L_{-1}(z)+zL_0(z)-|z|I_0(z)-|z|I_1(z)+2/\pi$$


For even $n$ the Fourier coefficient $\Lambda_n(z)\equiv 0$ for $nz>0$ [***]

$1.$ The Fourier coefficients, given above along the real $z$-axis, can be continued analytically into the complex-$z$ plane, but not along the imaginary $z$-axis. The branch cut for imaginary $z$ is a direct consequence of the essential singularity you noticed.


added November 2014 in response to a request for clarification.

since $\Lambda_{n}(z)=\Lambda_{-n}(-z)$, we may take $n>0$. Suppose $n=2m$, $m=1,2,\ldots$ and $z>0$. Then $$ \Lambda_{2m}(z)=\frac{1}{2\pi}{\rm Re}\,\int_{-\pi}^{\pi}e^{im\theta}e^{-iz\cot(\theta/2)}\,d\theta =\frac{1}{2\pi i}{\rm Re}\,\oint_{|\xi|=1^-} \xi^{m-1}e^{-z(1+\xi)/(1-\xi)}\,d\xi=0 $$ For $z<0$ the singularity at $\xi\equiv e^{i\theta}=1$ contributes, so the contour integral does not vanish.

share|cite|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.