Re: [Help-gsl] Possible bug on gsl_bspline_knot constructor

[Foivos Diakogiannis]

Hi Rhys,

to be honest, I have no idea. It just compiles and runs. With some other
breakpts I tried, I was getting inf values as result of B_i(x), but not
always.  I noticed that if the inconsistent (non increasing) break points
vector values are close to the begining, they are found, but if they are
close to the end of the breakpts they are not. I haven't tested this
thoroughly though.







On Tue, Jul 22, 2014 at 10:14 PM, Rhys Ulerich <address@hidden>
wrote:

> Thank you for the bug report.
>
> > I found the following problem which I think is a bug on the construction
> of
> > B-spline basis from non uniform breakpts
>
> How does
>
> vector<double> breakpts  {0.0,1.0,2.,3.,4.,5.,6.,7.,
> xmax,8.,9.,10.,11.}; // With this compiles/runs
>
> not exhibit the bug when xmax is set to 15?
>
> - Rhys
>