4
$\begingroup$

I am trying to evaluate the following integral \begin{eqnarray} I(\phi, \beta, \kappa) = \int_{T}^{\infty} \exp\left[\beta - \phi t \right]t^{-\frac{3}{2}} \log ^{\kappa}(t) \mbox{d}t \end{eqnarray} where $\phi$ and $\beta$ are positive constants and $\kappa > -4$.

Integration by parts for $I$

Let $u = t^{-\frac{3}{2}} \log ^{\kappa}(t)$ and $\frac{dv}{dt} = \exp\left[\beta - \phi t \right]$ then using integration by parts formula I got \begin{eqnarray} - t^{-\frac{3}{2}} \log ^{\kappa}(t) \phi^{-1} \exp\left[\beta - \phi t \right] \mid_T^{\infty} + \phi^{-1} \int_{T}^{\infty} \exp\left[\beta - \phi t \right]\frac{(2 \kappa -3 \log (t)) \log ^{\kappa -1}(t)}{2 t^{5/2}} \mbox{d}t \end{eqnarray} I do n't think this new integral is simpler.

Taylor Series expansion attempt for $I$

Using Taylor Series expansion for the exponential term I get \begin{eqnarray}\nonumber &=& \int_{T}^{\infty} \sum_{n=0}^{\infty} \frac{\left[\beta - \phi t \right]^n}{n!} t^{-\frac{3}{2}} \log ^{\kappa}(t) \mbox{d}t\\\nonumber &=& \int_{T}^{\infty} \sum_{n=0}^{\infty} \sum_{k=0}^{n} \frac{1}{n!} {{n}\choose{k}} \beta^{n - k} (- \phi t)^k t^{-\frac{3}{2}} \log^{\kappa}(t) \mbox{d}t\\ &=& \sum_{n=0}^{\infty} \sum_{k=0}^{n} \frac{1}{n!} {{n}\choose{k}} \beta^{n - k} (- \phi)^k \int_{T}^{\infty} t^{-\frac{3}{2} + k} \log^{\kappa}(t) \mbox{d}t \end{eqnarray} For $k \neq 0$ the last integral does not converge, but i think my original integral should converge, even if conditionally.

Any suggestions how to evaluate this integral? Thank you

$\endgroup$
1
  • $\begingroup$ No I haven't I am not quite sure how to solve this problem with the laplace transform $\endgroup$ Jan 8, 2018 at 3:28

1 Answer 1

3
$\begingroup$

Your integral $I\left( \phi ,\beta ,\kappa \right) $ can be evaluated for $\kappa =n\in N>0$:

$$I\left( \phi ,\beta ,n\right) =\exp \left( \beta \right) \int_{T}^{\infty }\exp \left( -\phi \,t\right) t^{-\frac{3}{2}}\log \left( t\right) ^{n}dt$$

with the trick of [Jack D'Aurizio] (Integrating $\int_0^{\frac{\pi}{2}} x (\log\tan x)^{2n+1}\;dx$)

$$I\left( \phi ,\beta ,n\right) =\exp \left( \beta \right) \underset{\alpha \rightarrow 0}{\lim }\frac{d^{n}}{d\alpha ^{n}}\int_{0}^{\infty }u^{\alpha -% \frac{3}{2}}\exp \left( -\phi \,t\right) dt$$

Performing the integration:

$$I\left( \phi ,\beta ,n\right) =\exp \left( \beta \right) \sqrt{\phi }% \underset{\alpha \rightarrow 0}{\lim }\frac{d^{2n+1}}{d\alpha ^{2n+1}}\phi ^{-\alpha }\Gamma \left( \alpha -\frac{1}{2},T~\phi \right)$$

the differentiation and take the limit as $\alpha \rightarrow 0$ leads to:

$$I\left( \phi ,\beta ,n\right) =\exp \left( \beta \right) \sqrt{\phi }% \sum_{k=0}^{n}\binom{n}{k}\left( -1\right) ^{n-k}\left( \log [\phi ]\right) ^{n-k}\times $$ $$\times \Gamma \left( -\frac{1}{2},T~\phi \right) \left( \log [T~\phi ]\right) ^{k}+k!\sum_{m=1}^{k}\frac{\left( \log \left( T~\phi \right) \right) ^{k-m}}{\left( k-m\right) !}G_{m+1,m+2}^{m+2,0}\left( T~\phi \left\vert \begin{array}{c} 1,1,...,1 \\ 0,0,...,0,-\frac{1}{2}% \end{array}% \right. \right)$$

where the general Leibniz rule (https://en.wikipedia.org/wiki/Product_rule) and the Wolfram Function side is used.

$\endgroup$
7
  • $\begingroup$ Thank you I will check the result and accept in the near future. $\endgroup$ Jan 8, 2018 at 23:36
  • $\begingroup$ I evaluate your integral for $\kappa$ = n, because I have problems to find the Mellin-Transformation of M{exp(-exp(y)}, but anyway then lead to some kind of complex H-Fox-Function. If anyone have a hint to M{exp(-exp(y)}, we can go further. $\endgroup$
    – stocha
    Jan 9, 2018 at 8:10
  • $\begingroup$ I have a bounty open over a similar integral, it would be awesome if you can please take a look at that. In the bounty question. I am not surprised to hear that this showed up some connection to H Fox function if you will like to share that calculation that will be educational for me and others as well, I think. Thank you once again $\endgroup$ Jan 9, 2018 at 10:57
  • $\begingroup$ Thank you for your bounty, I have now an idea for the exact solution, I will try in the evening. Also I will consider the other question. Thanks $\endgroup$
    – stocha
    Jan 9, 2018 at 17:43
  • $\begingroup$ The idea does not work, the solution looked like a Mittag-Leffler-Function, so I'm left with $\kappa $ = $n$, I can take a look at your other integral, but I have no idea to solve this for any $\alpha$ $\endgroup$
    – stocha
    Jan 9, 2018 at 21:42

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.