The probability density function (PDF) of Gumbel distribution is given as:

$$f\left(x\right)=\frac{\exp \left(-\left(\exp \left(-\frac{x-\mu}{\beta }\right)+\frac{x-\mu}{\beta }\right)\right)}{\beta },$$

where $\beta>0$. I would like to assume the translation $\mu$ to be $0$, which gives:

$$f\left(x\right)=\frac{\exp \left(-\left(\exp \left(-\frac{x}{\beta }\right)+\frac{x}{\beta }\right)\right)}{\beta }.$$

The Mellin transform of the PDF is therefore:

$$\int_0^{\infty } \frac{x^{s-1} \exp \left(-\left(\exp \left(-\frac{x}{\beta }\right)+\frac{x}{\beta }\right)\right)}{\beta } \ dx,$$ where $s>0$.

To solve this, I input the expression into Mathematica but no luck. Is there any other way to solve this?

  • 1
    $\begingroup$ i might be wrong but this could be a candidat for ramanujans master theorem... $\endgroup$
    – tired
    Oct 7 '15 at 9:12

For a start, I would do

$\begin{array}\\ \int_0^{\infty } \frac{x^{s-1} \exp \left(-\left(\exp \left(-\frac{x}{\beta }\right)+\frac{x}{\beta }\right)\right)}{\beta } \ dx &=\int_0^{\infty } \frac{x^{s-1}\exp(-\frac{x}{\beta }) \exp \left(-\left(\exp \left(-\frac{x}{\beta }\right)\right)\right)}{\beta } \ dx\\ \text{Letting } y =\exp(-\frac{x}{\beta }), dy=-\frac1{\beta}\exp(-\frac{x}{\beta })dx, &x=\beta\ln(1/y)\\ &=\int_1^{0 } -(\beta\ln(1/y))^{s-1} \exp (-y)\ dy\\ &=\beta^{s-1}\int_0^1 (\ln(1/y))^{s-1} \exp (-y)\ dy\\ \end{array} $

This resembles the formula for the $s$-th derivative of $\Gamma(x)$ at $x=1$.

Further than that I know not.

  • 2
    $\begingroup$ I don't think you can say much more than this - except perhaps an elaboration of the fact that it is coming from the lower incomplete gamma function. $\endgroup$ Nov 19 '15 at 12:39
  • $\begingroup$ than you very much for the contribution @martycohen $\endgroup$ Nov 26 '15 at 2:31

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.