2
$\begingroup$

Landau in the first equation of Über die Gitterpunkte in einem Kreise uses the following formula for the Bessel function of the first kind: $$\frac{1}{2\pi i } \int_{1-\infty}^{1+i\infty} \frac{\mathrm e^{As-B/s}}{s^4} \, \mathrm d s= (A/B)^{3/2} J_3(2 \sqrt{AB})) $$ valid for all $A,B>0$.

Does anyone know of a reference or a proof of that? Since Landau gives no reference or hint I imagine it is either well-known or completely trivial.

$\endgroup$
1

1 Answer 1

2
$\begingroup$

First, change the variable in the integral $s=kz$, where $k=\sqrt{B/A}$. You obtain $$\left(A/B\right)^{3/2} \int \frac{\exp{\sqrt{AB}\left(z-1/z\right)}}{z^4}dz.$$ Then deform the contour of integration from the vertical line to $|z|=1$, this is possible since the real part of $(z-1/z)$ is bounded in a simply connected region which contains both contours.

And finally refer to the well known generating function for Bessel functions $$\sum_{-\infty}^\infty J_n(x)z^n=\exp\left(\frac{x}{2}\left(z-1/z\right)\right).$$ Combined with Laurent's theorem this implies that our integral is $J_3(2\sqrt{AB})$.

$\endgroup$
2
  • $\begingroup$ I have a silly question: one ends up with $$ \sum_{n\in \mathbb Z} J_n(2\sqrt{AB}) \int_{1-i \infty}^{1+i \infty} \frac{z^n}{z^4}\mathrm d z.$$ I think you are implying that the integrals vanish when $n\neq 3$. But how is that possible? for example, when $n=4$ isn't the integral just the integral of a constant function over an infinite line? I am sorry for I must be doing something wrong. $\endgroup$
    – Dr. Pi
    Commented Mar 16, 2023 at 17:11
  • $\begingroup$ I do not understand the formula you wrote. What does it mean? (there is no equality sign). Where does it come from? It makes to sense because all integrals with $n>2$ diverge. $\endgroup$ Commented Mar 17, 2023 at 5:30

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .