I'm trying to evaluate the integral for an interference problem in Physics. When $y=0$, this reduces to the Bessel Function of the first kind, and when $l=1$, I can differentiate under the integral w.r.t. $x$ and evaluate the integral (which gives a first order Bessel function of the first kind). However, I'm looking for a more general answer, where $l$ is any integer, and $A$ is an arbitrary constant.

Here's a link to a similar question posted 2 years ago. How to solve integral $\int_0^{2\pi} e^{i(a\cos\phi + b\sin\phi)} \cos\phi\ d\phi$

  • $\begingroup$ Why don't you simply define a new coordinate system by rotation in which $(x,y)$ lies strictly on one axis. $\endgroup$ – Hamed Aug 23 '18 at 16:05

Let $\rho = \sqrt {x^2 + y^2}, \phi = \arctan(x, y)$. Then $$\int_0^{2 \pi} e^{-i A (x \cos t + y \sin t)} \cos l t \,dt = \\ \int_0^{2 \pi} e^{-i A \rho \cos(t - \phi)} \cos l t \,dt = \cos l \phi \int_0^{2 \pi} e^{-i A \rho \cos t} \cos l t \,dt = \\ 2 \pi (-i)^l \cos l \phi \,J_l(A \rho).$$

  • $\begingroup$ Hey! Thanks for your answer. Before posting this question, I was able to get to $\int_{0}^{2\pi}e^{-iA\rho\cos(t-\phi)}\cos(lt)\,dt$, but it's the next step where you factored out $\cos(l\phi)$ that I couldn't get to, and still don't understand. Can you please show the working? $\endgroup$ – Pooja Jayachandran Aug 23 '18 at 22:12
  • 1
    $\begingroup$ $\int_{-\phi}^{2 \pi - \phi} e^{-i A \rho \cos t} \cos l (t + \phi) dt = \int_0^{2 \pi} e^{-i A \rho \cos t} \cos l (t + \phi) dt$ because the integrand is periodic with period $2 \pi$. After expanding $\cos l (t + \phi)$, the integral of $e^{-i A \rho \cos t} \sin l t$ is zero. Similar to this question. $\endgroup$ – Maxim Aug 23 '18 at 23:18
  • $\begingroup$ That makes sense, thank you! $\endgroup$ – Pooja Jayachandran Aug 24 '18 at 10:42

Let's recall an integral form of Bessel function of first kind $J_\ell(y)$ for $\ell$ integer: $$ J_\ell(y)=\frac{1}{2\pi}\int_0^{2\pi} e^{i(y\sin \varphi-\ell\varphi)}d\varphi $$ which is a real number.

Now for your question, define the unit vector $\vec{n}(\varphi)=(\cos\varphi, \sin\varphi)$ and $\vec{r}=(x,y)$. Note that the exponent in the integrand is now $$ -i(x\cos \varphi+y\sin\varphi)=-i\vec{n}\cdot \vec{r} $$ Define the following integrals Similarly, define two more integrals $$\begin{aligned} I_\ell(\vec{r})&=\int_0^{2\pi} e^{-i\vec{n}(\varphi)\cdot\vec{r}}\cos (\ell \varphi) d\phi\\ I'_\ell(\vec{r})&=\int_0^{2\pi} e^{-i\vec{n}(\varphi)\cdot \vec{r}}\sin \ell \varphi d\varphi\\ \widetilde{I_\ell}(\vec{r})&= \int_0^{2\pi} e^{-i[\vec{n}(\varphi)\cdot \vec{r}-\ell\varphi]} d\varphi = I_\ell(\vec{r})+iI'_\ell(\vec{r}) \end{aligned} $$ Note that $I_\ell$ and $I'_\ell$ are both real if $\ell$ is even, and purely imaginary otherwise (use the fact that $\vec{n}\to -\vec{n}$ means $\varphi\to \pi+\varphi$). This means $I_\ell$ is the real part of $\widetilde{I}$ if $\ell$ is even, and it is equal to $i\mathfrak{I}(\widetilde{I_\ell})$ if $\ell$ is odd.

Moving on... Let $R_{\theta}$ be a rotation matrix (around $z$-axis, i.e. a 2D rotation) by an amount $\theta$. Note that in general $\vec{u}\cdot R_\theta \vec{v}=R_{-\theta}\vec{u}\cdot \vec{v}$. Moreover, $R_\theta \vec{n}(\varphi)=\vec{n}(\varphi+\theta)$. Using these facts, you find (I'll leave the details to you) $$ \widetilde{I_\ell}(R_\theta\vec{r})= e^{-i\theta \ell} \widetilde{I_\ell}(\vec{r}) $$ At the same time, let $\theta(\vec{r})$ be such that $R_\theta \vec{r}=(0,Y)$ for $Y(\vec{r})=\sqrt{x^2+y^2}$ (no $x$-component). If you need a formula $\theta(\vec{r})=\pi/2-\arctan y/x$. Then $$ \widetilde{I_\ell}(\vec{r})=e^{i\theta(\vec{r})\ell}\int_0^{2\pi} e^{-i(Y(\vec{r})\sin \varphi-\ell\varphi)}d\varphi= 2\pi e^{i\theta(\vec{r})\ell} J_\ell(Y(\vec{r})) $$ Therefore $$ \boxed{ I_\ell(\vec{r})=\begin{cases} 2\pi \cos [\theta(\vec{r})\ell] J_\ell(Y(\vec{r})) & \ell\text{ even}\\ 2\pi i \sin [\theta(\vec{r})\ell] J_\ell(Y(\vec{r})) & \ell\text{ odd} \end{cases} } $$

  • $\begingroup$ Note that the integral is not zero for $\ell = 0, y = 0$. The factor depending on $\phi = \arctan(x, y)$ is $\cos \ell \phi$. $\endgroup$ – Maxim Aug 23 '18 at 20:45
  • $\begingroup$ Thanks, it was a propagating typo. $\endgroup$ – Hamed Aug 23 '18 at 20:58
  • $\begingroup$ The sign for odd $\ell$ is reversed, $I_1((x, 0)) = -2 \pi i J_1(x)$. $\endgroup$ – Maxim Aug 23 '18 at 21:18

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.