[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
## Re: [Help-gsl] Exponential calculation

**From**: |
Kevin Channon |

**Subject**: |
Re: [Help-gsl] Exponential calculation |

**Date**: |
Mon, 16 Feb 2009 08:37:21 +0000 |

Hi Ashish,
>* As structure of probability density function is as follows.*
>* Pdf = (expression_1) * (exponential (expression_2).*
>
>* Now during our calculation we are encountering the situation where*
>* expression_2 is getting evaluated to such values like -800.0.*
>* In this case exponential (-800.0) is zero.*
>
>* So how do we deal with such situations what will be pdf in this case.*
>
Like you say, in this situation the result is zero. This is down to the
limits of floating point precision. If you can't live with this, then I guess
there are three things that I can think of right now:
1. Take logs and work with those:
Ln(PDF) = Ln(expression1) + expression2
2. Separate the exponent:
You know exp(-100) = 3.7e+44, so
exp(-800) = 3.7^8 *10^(-44*8) = 35125e-352
3. Find a library that someone has written to deal with large and small
exponentials
Hope that helps,
Kevin
--
Kevin Channon
School of Chemistry
University of Bristol
UK