I encountered the following integral in a physical problem $$I=\int r^2 \text{J}_0(\alpha r) \text{I}_1(\beta r)\text{d}r$$ where $\text{J}_0$ is the Bessel function of first kind of order $0$ and $\text{I}_1$ is the modified Bessel function of order $1$. Also, $\alpha$ and $\beta$ are arbitrary real numbers.

It seems that MAPLE and WOLFRAM are not able to find the primitive. However, I think that there should be a tidy one in terms of Bessel functions.

My Thought

My instinct in integration tells me to use integration by parts and the recursive relations for Bessel functions. But I couldn't get any where yet.

  • 1
    $\begingroup$ u might find this site quiet helful: dlmf.nist.gov/10.22 $\endgroup$ – tired Dec 12 '15 at 15:44
  • $\begingroup$ @tired: Thanks, how can I search this integral there!? :) $\endgroup$ – H. R. Dec 12 '15 at 15:46
  • $\begingroup$ i just say that u might something useful there. I guess there is a closed form but im not sure. $\endgroup$ – tired Dec 12 '15 at 16:14
  • $\begingroup$ @tired: The most similar one there is the relation $10.22.7$. But they have the same multiplier $a$ in the argument! :) $\endgroup$ – H. R. Dec 12 '15 at 16:16

First note that

$$ r\text{I}_1(\beta r)=\frac{\partial}{\partial\beta} \text{I}_0\left(\beta r\right).$$

Then we can write \begin{align} \mathcal{I}(t):=&\int_0^t r^2 \text{J}_0\left(\alpha r\right)\text{I}_1\left(\beta r\right) dr\\ =&\frac{\partial}{\partial\beta}\int_0^t r\text{J}_0\left(\alpha r\right)\text{I}_0\left(\beta r\right)dr\\ =& \frac{\partial}{\partial\beta}\left[\frac{t\left(\alpha \text{J}_1(\alpha t)\text{I}_0(\beta t)+\beta \text{J}_0(\alpha t)\text{I}_1(\beta t)\right)}{\alpha^2+\beta^2}\right]. \end{align} The last line follows from the formula in Vol. II of Prudnikov-Brychkov-Marychev (and can of course be verified by direct differentiation).

  • $\begingroup$ (+1) Oh, GOD! That was GREAT! :) I would be happier if you included the computation of the last integral! :) $\endgroup$ – H. R. Dec 12 '15 at 21:13
  • $\begingroup$ @H.R. For the proof of the last integral, one only needs to differentiate the expression in square brackets with respect to $t$ and use recursion relations for Bessel functions. However if the answer were unknown, it would be unclear to me how to proceed. $\endgroup$ – Start wearing purple Dec 12 '15 at 21:21
  • $\begingroup$ Of course, taking derivative is much easier than finding primitive! :) $\endgroup$ – H. R. Dec 12 '15 at 21:24

According to the Start Wearing Purple's answer, I just took the computations and wanted to put the final answer here

$$\eqalign{ & I = {1 \over {{\alpha ^2} + {\beta ^2}}}\,\,\,\,\left( {\alpha {\text{I}_1}\left( {\beta r} \right){\text{J}_1}\left( {\alpha r} \right) + \beta {\text{I}_0}\left( {\beta r} \right){\text{J}_0}\left( {\alpha r} \right)} \right){r^2} \cr & \,\,\, - {{2\beta } \over {{{\left( {{\alpha ^2} + {\beta ^2}} \right)}^2}}}\left( {\alpha {\text{I}_0}\left( {\beta r} \right){\text{J}_1}\left( {\alpha r} \right) + \beta {\text{I}_1}\left( {\beta r} \right){\text{J}_0}\left( {\alpha r} \right)} \right)r \cr} $$

Hope that it may be useful for future readers. There may still be other heuristic ways to compute this integral so do not hesitate to write a new answer with a different approach. :)


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.