Take the 2-minute tour ×
Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

I would like to have a better form of this series:

$$\sum_{k=0}^{\infty}\,\frac{1}{k!}\,\left(\frac{ab\sin(c)}{\sqrt{2}}\right)^{2k}\,Q_{k+\frac{3}{2}}\left(ab\cos(c),bx\right)$$

where $Q_m(\alpha,\beta)$ is the generalized Marcum Q-function. Does anyone an idea on how to proceed, if possible to make it better (e.g. getting rid of the series)? Thanks!

EDIT: could the result be just proportional to $Q_{\frac{3}{2}}\left(ab,bx\right)$???

share|improve this question

1 Answer 1

up vote 1 down vote accepted

Well, I guess, in some form of. But at the same time you can get a closed form solution.
For simplicity let's set $\alpha=\left(\frac{ab\sin(c)}{\sqrt{2}}\right)^{2}$ and $\beta=ab\cos(c)$
Taking into account the definition of the generalized Marcum Q-function: $$ \begin{eqnarray} \sum_{k=0}^{\infty}\,\frac{\alpha^{k}}{k!}\,\,Q_{k+\frac{3}{2}}\left(\beta,bx\right)&=& \sum_{k=0}^{\infty}\,\frac{\alpha^{k}}{k!}\,\,\int_{bx}^{\infty} t \left( \frac{t}{\beta}\right)^{k+\frac{1}{2}} \exp \left( -\frac{t^2 + \beta^2}{2} \right) I_{k+\frac{1}{2}} \left( \beta t \right) \ \mathrm dt\\ &=&\frac{\exp \left(\!-\!\frac{\beta^2}{2}\! \right)}{\sqrt{\beta}}\int_{bx}^{\infty} t^{\frac{3}{2}}e^{-\frac{t^2}{2}}\sum_{k=0}^{\infty}\frac{\left(\frac{\alpha t}{\beta}\right)^k}{k!}I_{k+\frac{1}{2}}(\beta t) \ \mathrm dt\\ \end{eqnarray} $$ Then one can use Gradshteyn and Ryzhik's Table of Integrals, Series, and Products (Second Part): $$\sum_{k=0}^{\infty}\frac{\xi^k}{k!}I_{k+\nu}(\eta)=\left(\frac{2\xi}{\eta}+1\right)^{-\frac{\nu}{2}}I_{\nu}\left(\sqrt{\eta^2+2\eta\xi}\right), \qquad |2\xi|<|\eta|$$ So $$ \begin{eqnarray} \sum_{k=0}^{\infty}\,\frac{\alpha^{k}}{k!}\,\,Q_{k+\frac{3}{2}}\left(\beta,bx\right) =\frac{\exp \left(\!-\!\frac{\beta^2}{2}\! \right)}{\sqrt{\beta}}\left(1+\frac{2\alpha}{\beta^2}\right)^{-\frac{1}{4}} \int_{bx}^{\infty} t^{\frac{3}{2}}e^{-\frac{t^2}{2}}I_{\frac{1}{2}}(\sqrt{2\alpha+\beta^2}|t|) \ \mathrm dt=\\ =\text{A} \ \left[\sqrt{\pi } \sqrt{2 \alpha +\beta ^2} e^{\alpha +\frac{1}{2} \left(b^2 x^2+\beta ^2\right)+b x \sqrt{2 \alpha +\beta ^2}}\left(\text{erfc}\left(\frac{ \sqrt{2 \alpha +\beta ^2}+b x}{\sqrt{2}}\right)+\text{erfc} \left(\frac{b x-\sqrt{2 \alpha +\beta ^2}}{\sqrt{2}}\right) \right)+\sqrt{2} \left(e^{2 b x \sqrt{2 \alpha +\beta ^2}}-1\right)\right] \end{eqnarray} $$ where $\text{A}=\frac{\exp \left(-\frac{\beta ^2}{2}-\frac{1}{2} b x \left(2 \sqrt{2 \alpha +\beta ^2}+b x\right)\right)}{2 \sqrt{\pi } \sqrt{\left(2 \alpha +\beta ^2\right) \text{sgn}(\beta)}}$ The last equation was obtained via Mathematica (or you can turn to Gradshteyn and Ryzhik ;)). The equation was simplified assuming that $\alpha>0, \ x>0, \ \beta\in\mathbb{R}$. If not the result gets nastier (with lots of square roots). I guess you can turn the last integral into some form of the generalized Marcum Q-function, if you'd like to:

$$\int_{bx}^{\infty} t^{\frac{3}{2}}e^{-\frac{t^2}{2}}I_{\frac{1}{2}}(\sqrt{2\alpha+\beta^2}|t|) \ \mathrm dt=(2\alpha+\beta^2)^{\frac{1}{4}}e^{\frac{|2\alpha+\beta^2|}{2}}Q_{\frac{3}{2}}\left(\sqrt{2\alpha+\beta^2},bx\right)$$ So $$\sum_{k=0}^{\infty}\,\frac{\alpha^{k}}{k!}\,\,Q_{k+\frac{3}{2}}\left(\beta,bx\right)=e^{\alpha}Q_{\frac{3}{2}}\left(\sqrt{2\alpha+\beta^2},bx\right)$$ Noting that $\sqrt{2\alpha+\beta^2}=a \ b $ the result turns out to be even simplier.

share|improve this answer
    
Thanks! there is anyway a mistake in the power of $t$ inside the integral. –  JFNJr Jul 7 '13 at 18:15
    
Correcting the power inside the integral, it turns out to be really what I expected,i.e. proportional to $Q_{\frac{3}{2}}(ab,bx)$. Awesome! Thanks again! –  JFNJr Jul 7 '13 at 18:34
1  
Thanks, I fixed it, guess if I'm not loosing anything else, the answer should be correct. :) –  Caran-d'Ache Jul 7 '13 at 18:43
    
Many thanks! almost correct. The $\frac{1}{\sqrt{\beta}}$ cancels out, instead of giving $\frac{1}{\beta}$. Thanks again! –  JFNJr Jul 7 '13 at 18:47
    
Yep! Overlooked it again. :( Too many constants, too poor eyesight. –  Caran-d'Ache Jul 7 '13 at 18:50

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

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