[Bug-gsl] bug in gsl_sf_bessel_jl

[Mario Santos]

There is a possible big problem with the spherical bessel routine:

gsl_sf_bessel_jl(L, x)

When running with L<1000 I can see a second peak at x=1001.
For instance, try L=100 and plot the output from x=90 to x=1100...

Code:

int main(void) {

  double result, x, xmax, xmin;
  int L;

  L=100;
  xmin=L-L/20.;
  xmax=40.*L;
  for(x=xmin;x<xmax;x++) {
    result=gsl_sf_bessel_jl(L, x);
    printf("%f       %E\n",x,result);
  }


}