3
$\begingroup$

$$\int\limits_0^{2\pi} \int\limits_0^{2\pi} \min \{ x-y, 2\pi- (x-y) \} e^{a\cos(x)} e^{a\cos(y)} \, dx \, dy, \qquad a\in\mathbb R. $$ I tried to find the value of the integral following the method proposed in this example: however I didn't succeed so I posted the question here looking forward to your experience.

$\endgroup$
1
  • $\begingroup$ How about turning this into a contour integral using the Euler identity (math.stackexchange.com/questions/816763/…)? First you would need to remove the minimum and the variables x,y by splitting the integral and doing some integration by parts so that you are only left with trig functions. $\endgroup$ Commented Dec 18, 2021 at 20:48

1 Answer 1

1
$\begingroup$

I don't have enough reputation to comment so I am leaving this as an answer. You are integrating over the square of side-length $2\pi$, a reasonable first step (that might make the minimum more easy to handle) is to shift to a coordinate system that is rotated at 45 degrees to the one you're currently using, something like

\begin{align} u &= \frac{1}{\sqrt{2}}(x+y)\\ v &= \frac{1}{\sqrt{2}}(x-y), \end{align}

the limits of integration are a bit annoying with this subsitution, one of them is simple you just go from the origin to the top right corner of the square so the interval is $[0, 2\pi\sqrt{2}]$. The second limit has to between the edges of the square, orthogonal to the diagonal. Its simplest to break the first integral when we reach the center of the square so we can express this relatively simply

\begin{align} I = &\int_{0}^{\pi\sqrt{2}}\int_{-v}^v \min\{v\sqrt{2}, 2\pi-v\sqrt{2}\} e^{a\cos\left(\frac{u+v}{\sqrt{2}}\right)} e^{a\cos\left(\frac{u-v}{\sqrt{2}}\right)} \, du \, dv \\&+ \int_{\pi\sqrt{2}}^{2\pi\sqrt{2}}\int_{v-2\pi\sqrt{2}}^{2\pi\sqrt{2}-v} \min\{v\sqrt{2}, 2\pi-v\sqrt{2}\} e^{a\cos\left(\frac{u+v}{\sqrt{2}}\right)} e^{a\cos\left(\frac{u-v}{\sqrt{2}}\right)} \, du \, dv. \end{align}

At this point you can split up the $v$ integrals further to get rid of the minimums and (hopefully!) make an integral you can deal with.

$\endgroup$
3
  • $\begingroup$ I am afraid the your result isn't any easier to work with than the original integral, in particular because now you have a combination of $u$ and $v$ inside the trigonometric functions, which in turn are inside the exponential. $\endgroup$
    – Alex M.
    Commented Dec 18, 2021 at 20:29
  • 1
    $\begingroup$ The (simple) integral of $\exp(a\cos(x))$ is equal to $2\pi I_0(x)$ I believe. Is there an "explicit" formula for the integral of $x\exp(a\cos(x))$ ? This would be a first step. $\endgroup$ Commented Dec 18, 2021 at 20:33
  • $\begingroup$ Next: $$ \begin{align} & \int_{-v}^v \min\{v\sqrt{2}, 2\pi-v\sqrt{2}\} e^{a\cos\left(\frac{u+v}{\sqrt{2}}\right)} e^{a\cos\left(\frac{u-v}{\sqrt{2}}\right)} \, du \\ {} \\ = {} & \min\{v\sqrt{2}, 2\pi-v\sqrt{2}\} \int_{-v}^v e^{a\cos\left(\frac{u+v}{\sqrt{2}}\right)} e^{a\cos\left(\frac{u-v}{\sqrt{2}}\right)} \, du \end{align}$$ Then think about a trigonometric identity whose left side is the sum of two cosines. $\endgroup$ Commented May 7 at 13:22

You must log in to answer this question.

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