0
$\begingroup$

I am trying to evaluate
$$P(X>x) = \int_x^{\infty } t^{\kappa } \exp{\left(-\rho t^{\alpha\kappa + 1}\right)} \, dt$$ where $\kappa$, $\rho$ and $\alpha$ are all constants. I have tried some substitutions, integrating by parts to solve the integral, but it did not seem to converge.

I also tried to compute the laplace transform, inverse Fourier transform of the density using Mathematica, but it was n't able to work it out.

After having spent 6-7 hours trying to solve this I am still hopelessly stuck.

Any help would be much appreciated.

$\endgroup$
  • $\begingroup$ What did you spend "6-7 hours" at, exactly? Did you note that the RHS cannot be P(X>x) for every x, for any random variable X? $\endgroup$ – Did Sep 20 '14 at 7:48
  • $\begingroup$ @Did No I did not notice anything like that. Can you please explain more why it can not be? $\endgroup$ – Comic Book Guy Sep 20 '14 at 20:41
  • $\begingroup$ To begin with, P(X>0) might be greater than 1... $\endgroup$ – Did Sep 20 '14 at 21:35
  • $\begingroup$ @Did I think what you are getting at is that in order to compute this probability I need to rewrite the rhs as 1 - the integral with lower limit negative infinity going to x, using the complement rule. Is that correct ? $\endgroup$ – Comic Book Guy Sep 20 '14 at 21:48
  • 1
    $\begingroup$ Then you might want to mention in the question the regime you have in mind, ensuring that P(X>0) is not greater than 1. If P(X>0)<1, this means there is a Dirac delta at x=0? (FYI, I just flagged your last comment.) $\endgroup$ – Did Sep 21 '14 at 11:56
1
$\begingroup$

Did you try the substitution $u=t^{\alpha\kappa+1}$? As far as I see, you get then something like $k\cdot u^{\beta}\exp(-\rho u)$ ($k,\beta$ constants) as integrand, the integral is then similar to the incomplete Gamma function. https://en.wikipedia.org/wiki/Incomplete_gamma_function

Hope that it helps.

$\endgroup$
1
$\begingroup$

As answered by Karl $$\int t^{\kappa } \exp{\left(-\rho t^{\alpha\kappa + 1}\right)} \, dt=-\frac{t^{\kappa +1} \left(\rho t^{\alpha \kappa +1}\right)^{-\frac{\kappa +1}{\alpha \kappa +1}} \Gamma \left(\frac{\kappa +1}{\alpha \kappa +1},t^{\alpha \kappa +1} \rho \right)}{\alpha \kappa +1}$$ where appears the incomplete gamma function.

This can also write $$\int t^{\kappa } \exp{\left(-\rho t^{\alpha\kappa + 1}\right)} \, dt=-\frac{t^{\kappa +1} E_{\frac{(\alpha -1) \kappa }{\alpha \kappa +1}}\left(t^{\alpha \kappa +1} \rho \right)}{\alpha \kappa +1}$$ where appears the exponential integral function.

$\endgroup$

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.