Re: [Bug-gsl] Complex: gsl_complex_div_real bug

[Chris]

Hi brian,
Actually after having tried to implement such a bug i've realised that
it was impossible. The fn — Function: gsl_complex gsl_complex_div_real
(gsl_complex a,float b)

gives you this: answer = z / b

However i was imagining a situation where you want to do b / z which
would if Im(Z) = 0 give you a spurious divide by zero error. However
since there is no function for doing this it's not an issue.

I must confess that i came across this bug in my own cuda
re-implementation of the gsl complex class, i just assumed it'd be
present if you actually used the gsl code as it should be.

Finally however, if you do use the gsl_complex_inverse on a complex
number z which has a zero imaginary part then the imaginary part
becomes negative zero instead of still being zero.

Viz the following program, it's not the original bug but it's another.

Thanks


#include "stdio.h"
#include "stdlib.h"
#include "gsl/gsl_complex.h"
#include "gsl/gsl_complex_math.h"

int main (void){
        gsl_complex complex_number;
        gsl_complex answer;
        
        GSL_SET_COMPLEX(&complex_number, 4, 0);
        
        complex_number = gsl_complex_inverse(complex_number);   
        
        printf("%g,%g\n", GSL_REAL(complex_number), GSL_IMAG(complex_number));
}







On Tue, May 6, 2008 at 1:30 PM, Brian Gough <address@hidden> wrote:
> At Mon, 5 May 2008 13:37:08 -0700,
> chris coleman-smith wrote:
>> There is a bug in the implementation of gsl_complex_div_real
>> (gsl_complex a, double x), from the 1.9 package in the complex/math.c
>> file.
>
> Can you send a small program which demonstrates the problem.  Thank
> you.
>
> --
> Brian Gough
>
>