[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Bug-gsl] Bug in Gumbel Distribution
From: |
Brian Gough |
Subject: |
Re: [Bug-gsl] Bug in Gumbel Distribution |
Date: |
Sun, 19 Jul 2009 12:57:09 +0100 |
User-agent: |
Wanderlust/2.14.0 (Africa) Emacs/22.2 Mule/5.0 (SAKAKI) |
At Mon, 13 Jul 2009 17:43:48 -0700 (PDT),
Mojtaba Mahsuli wrote:
> In the function that evaluates the CDF of the Gumbel Distribution,
> namely "gsl_cdf_gumbel1_P" and "gsl_cdf_gumbel1_Q", if the variable
> "a" has a small value (due to a large standard deviation) and
> variable b has a large value, the returned CDF is Zero for all
> X. This is because in the implementation, first, exp(-b) and
> exp(-ax) are evaluated and then, using "pow" function, the CDF is
> evaluated. If instead, first, A = -b * exp(-ax) is evaluated and
> then, exp(A) is calculated, the problem will be solved.
Thanks for the bug report. I will change the function to make
its behaviour more uniform for all values of a and b.
--
Brian Gough