This time remembering to CC the mailing list... On Tue, Mar 7, 2023, 9:57 PM Rhys Ulerich <[email protected]> wrote:
> On Tue, Mar 7, 2023, 5:11 PM Simon Wiesheier <[email protected]> > wrote: > >> After reading the manual, it is not clear to me how GNU internally >> constructs the knot vector. >> There are the functions, >> gsl_bspline_knots >> gsl_bspline_knots_uniform, >> that create the knot vector based on given breakpoints. >> > > I encourage you to initialize a cubic workspace (k=4, pick nbreak) then to > use gsl_bspline_knots_uniform to have the GSL construct the knot vector for > you given some [a, b]. You will be able to observe the multiplicity of the > various knots in the resulting w->knots. The multiplicity is a consequence > of the chosen k meaning that if you opted for quadratic or quintic k you > will see a different knot multiplicity. Play around a bit. > > You may (or may not) find the routines at > https://github.com/RhysU/suzerain/blob/master/suzerain/bspline.h to be > useful worked examples. Those include forming linear combinations of the > basis. > > - Rhys > >>
