Consider the integral \begin{align*} g_f(x)=\int_{\phi=0}^{2\pi} f(\phi) ~e^{x cos(\phi)}~\mathrm{d}\phi, \end{align*} where $f(\phi)$ is a probability density functions defined over $[0,2\pi]$, \begin{align*} \forall \phi: f(\phi) \geq 0,~ \int_{\phi=0}^{2\pi} f(\phi)~ \mathrm{d}\phi=1. \end{align*} In case of uniform distribution, i.e. when $\forall \phi: f(\phi) =\frac{1}{2\pi}$, then we know that $g_u(x)=\mathrm{I}_0(x)$, where $\mathrm{I}_0(x)$ is the modified Bessel function of the first kind.

Does there exist other known functions, for other probability density functions $f(\phi)$?

More precisely, in the case of modified Bessel functions of the first kind, we have the following relation:

$$\int_{x=0}^{\infty}x e^{-ax^2}\cdot\mathrm{I}_0(bx)\cdot\mathrm{I}_0(cx)~\mathrm{d}x=\frac{1}{2a}e^{\frac{b^2+c^2}{4a}}~\mathrm{I}_0\left(\frac{bc}{2a}\right).$$

Do we have similar relation (or upper bounds) on

$$\int_{x=0}^{\infty}x e^{-ax^2}\cdot g_{f_1}(bx)\cdot g_{f_2}(cx)~\mathrm{d}x,$$ for other particular choices of $f_1$ and $f_2$? Even for the simpler case where $f_1=f_2$? The ideal answer for me is something proportional to $g_{f_3}(\frac{bc}{2a})$ (or $g_{f_3}(d\frac{bc}{2a})$, for some parameter $d$), for some pdf $f_3$.

P.S. The question was first asked in math.stackexchange. After some times without any response, I deleted it there, and posted it here.


1 Answer 1


Q: Do there exist other known functions, for other probability density functions $f(\phi)$?

Here are three:

$$f(\phi)=\frac{1}{\pi}\cos^2 \phi\Rightarrow g(x)=\frac{2 [I_1(x)+x I_2(x)]}{x},$$ $$f(\phi)=\frac{4}{\pi}\cos^2 \phi\sin^2\phi\Rightarrow g(x)=\frac{8 (x I_1(x)-3 I_2(x))}{x^2},$$ $$f(\phi)=\frac{1}{2\pi I_0(1)}e^{\sin\phi}\Rightarrow g(x)=\frac{I_0\left(\sqrt{x^2+1}\right)}{I_0(1)}.$$


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

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