2
$\begingroup$

I want to evaluate an integral $I_1$ defined in $Eq.(1)$ as

\begin{align} I_1=\int_{0}^{\infty}\frac{x\exp(-\beta x)K_1(\alpha x)}{1+x}dx\tag{1} \end{align} Where $\alpha\geq0$, $\beta\geq0$, and $K_1(.)$ is modified first order Bessel's function of second kind

In order to evaluate it, I went through transforming $\frac{x}{1+x}$ and $K_1(\alpha x)$ in terms of Meijer functions, which has been verified in MATHEMATICA and MUPAD

\begin{align} \frac{x}{1+x}=G_{1,1}^{1,1}\bigg(x\bigg|_{1}^{1}\bigg)\tag{2} \end{align} and \begin{align} K_1(\alpha x)=\exp (\alpha x)\sqrt{\pi}G_{1,2}^{2,0}\bigg(2\alpha x\bigg|_{1, -1}^{1/2}\bigg)\tag{3} \end{align}

thereby rewriting $Eq.(1)$ as $I_2$ from $Eq.(2)$ and $Eq.(3)$ \begin{align} I_2=\int_{0}^{\infty}\exp(-\phi x)G_{1,1}^{1,1}\bigg(x\bigg|_{1}^{1}\bigg)G_{1,2}^{2,0}\bigg(2\alpha x\bigg|_{1, -1}^{1/2}\bigg)dx\tag{4} \end{align} where $\phi = \beta - \alpha$

Now in order to proceed further I need to have a single Meijer function for the product of above two Meijer's function i.e.

\begin{align} G_{pq}^{mn}\bigg(\theta x\bigg|_{b_1,b_2,...,b_q}^{a_1,a_2,...,a_p}\bigg)=G_{1,1}^{1,1}\bigg(x\bigg|_{1}^{1}\bigg)G_{1,2}^{2,0}\bigg(2\alpha x\bigg|_{1, -1}^{1/2}\bigg)\tag{5} \end{align}

From $Eq.(5)$ and $Eq.(4)$, I can evaluate the above integral as \begin{align} I=\int_{0}^{\infty}\exp(-\phi x)G_{p,q}^{m,n}\bigg(\theta x\bigg|_{b_1,b_2,...,b_q}^{a_1,a_2,...,a_p}\bigg)dx =\frac{1}{\phi}G_{p+1,q}^{m,n+1}\bigg(\frac{\theta}{\phi} \bigg|_{b_1,b_2,...,b_q}^{0,a_1,a_2,...,a_p}\bigg)\tag{6} \end{align}

Solution to integral is from $[Eq.7.813.1]$, Table of Integrals, Series, and Products, I. S. Gradshteyn and I. M. Ryzhik, 8e

Here what I've tried at MATHEMATICA

MeijerG[{{1}, {}}, {{1}, {}}, x] Exp[\[Alpha]x] Sqrt[\[Pi]] MeijerG[{{}, {1/2}}, {{1, -1}, {}}, 2 \[Alpha]x]

which results into

\begin{align} \frac{xK_1(\alpha x)}{1+x} \end{align}

Instead of this I want a single Meijer's G function representation of it.

$\endgroup$
5
  • $\begingroup$ Is this a better question for the math stack exchange? Do you know of a formula for the general product of two MeijerG functions that you are trying to get Mathematica to find? $\endgroup$
    – Jason B.
    Commented Feb 4, 2016 at 14:18
  • $\begingroup$ Definitely Jason B. After not getting response I tried do solve in Mathematica. It would be enough for me if some reference for the general product of two MeijerG functions can be pointed out. $\endgroup$
    – Meet
    Commented Feb 4, 2016 at 14:24
  • $\begingroup$ Related Q&A. Similar: 103737. Reverse problem: 102270 $\endgroup$
    – Michael E2
    Commented Feb 4, 2016 at 14:25
  • $\begingroup$ What are the domains of $\alpha$ and $\beta$? Are there any restrictions on their values? $\endgroup$ Commented Feb 4, 2016 at 15:31
  • $\begingroup$ Where $\alpha\geq0$ and $\beta\geq0$ $\endgroup$
    – Meet
    Commented Feb 4, 2016 at 15:44

1 Answer 1

2
$\begingroup$

The general case seems to be insoluble for Mathematica, but in the special case of $\alpha=\beta$, one can obtain a result in terms of Meijer $G$. Note that the integral for $\alpha=\beta$ can be recast as a Mellin convolution:

$$\int_0^\infty x\exp(-\beta x)K_1(\beta x)\frac1{1+\frac1x}\frac{\mathrm dx}{x}$$

and thus:

MellinConvolve[x Exp[-b x] BesselK[1, b x], 1/(1 + x), x, s] /. s -> 1
   Sqrt[π] MeijerG[{{-1}, {1/2}}, {{-1, -1, 1}, {}}, 2 b]

Compare:

With[{b = 5}, 
     NIntegrate[(x E^(-b x) BesselK[1, b x])/(1 + x), {x, 0, ∞}, WorkingPrecision -> 20]]
   0.024024943996405476775

With[{b = 5}, Sqrt[π] MeijerG[{{-1}, {1/2}}, {{-1, -1, 1}, {}}, 2 b]]
   0.024024943996405476768
$\endgroup$

Your Answer

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

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