[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
## Re: [Help-gsl] fourier integral using QAWF

**From**: |
Brian Gough |

**Subject**: |
Re: [Help-gsl] fourier integral using QAWF |

**Date**: |
Mon, 20 Apr 2009 16:17:24 +0100 |

**User-agent**: |
Wanderlust/2.14.0 (Africa) Emacs/22.1 Mule/5.0 (SAKAKI) |

At Tue, 14 Apr 2009 10:15:36 +0200,
Shuman Serle wrote:
>* In order to see if QAWO does the job, I test my code on the block*
>* function of the FFT example in the gsl manual. I then print the*
>* frequency space result which looks correct. Then when I try to*
>* invert the result to see how close I am to the original block*
>* function, I get the gsl error GSL_EDIVERGE. I've checked the code*
>* and it seems that I'm following the right approach i.e. call*
>* gsl_integration_qawo_table_set in a loop for each omega value then*
>* do the integration & save the result for the given omega. Has anyone*
>* tested this?*
The routines are tested in the "make check" test suite but please send
an example program to address@hidden if it is giving unexpected
results.
>* The other issue that I have is that I do not fully understand (not*
>* clear from the manual) :*
>* *
>* what n should be set to in: gsl_integration_qawo_table_alloc*
>* *
>* what limit should be set to in int gsl_integration_qawo*
>* *
>* in order to get a certain accuracy. Say I have the region from [0,*
>* 100] if I think the minimum quadrature interval should be 1e-6, what*
>* do I set n and limit to. They seem to be correlated.*
The minimum values of these parameters will depend on the behavior of
the function, so there's no easy way to estimate them unfortunately.
In practice it's simplest to pick some sufficiently large values by
trial and error.
--
Brian Gough