Page 90 - Geometric Modeling and Algebraic Geometry
P. 90

88     R. Krasauskas and S. Zube
                           We use homogeneous control points, denoting by e 0 ,...,e 4 the standard frame in

                           RP . Consider the usual parametrization of a half of C: c(t)=  2  p i B (t) with
                                                                                       2
                              4
                                                                                       i
                                                                                 i=0
                           control points
                                            (p 0 ,p 1 ,p 2 )=(e 0 − be 2 ,ae 1 ,e 0 + be 2 ).

                           Define a biquadratic parametrization h(t, u)=  2  j=0 ij B (t)B (u), where
                                                                           r
                                                                        2
                                                                                    2
                                                                               2
                                                                                    j
                                                                               i
                                                                   i=0
                                   ⎛                                                  ⎞
                                          ab(e 3 + e 4 )               ab(−e 3 + e 4 )
                                        √                   ae 0     √
                             (r ij )=  ⎝  −b a − b (e 3 + e 4 )(a − b )e 1 b a − b (−e 3 + e 4 )  ⎠  . (5.9)
                                                           2
                                                               2
                                                                           2
                                           2
                                               2
                                                                       2
                                          ab(e 3 + e 4 )    ae 0       ab(−e 3 + e 4 )
                           This parametrization is derived from Example 5: take the parametrization F(t, u) of
                           Γ(C) and intersect with the hyperplane x 2 =0.
                              Then another rational parametrization of Γ(C) can be defined by drawing lines
                           through the ellipse points c(t) and the hyperboloid points h(t, u). The parametriza-
                           tion of the intersection of Γ(C) with a hyperplane Π(x)=0
                                           f(t, u)= Π(h(t, u))c(t) − Π(c(t))h(t, u).     (5.10)
                           Now it is easy to calculate B´ ezier control points of this parametrization of bi-degree
                           (4, 2). Indeed, just use conversion formulas from products B (t)B (t) to B 4 i+j (t).
                                                                                2
                                                                            2
                                                                                j
                                                                           i
                              If we choose a hyperplane Π(x)= x 4 − rx 0 then the section is a pipe surface
                           shown in Fig. 5.5(left).
                           5.6 Implicit equations and double points
                           It will be convenient to use affine coordinates X =(x, y, z, r)=(x 1 /x 0 ,...,x 4 /x 0 )
                           in R . We start from the case E ++ of ellipse. By the definition Γ(E) is the envelope
                               4
                               1
                           of the family of isotropic cones Γ(E(t)) =  X − E(t),X − E(t) , where E(t)=
                           (a(1 − t )/(1 + t ), 2bt/(1 + t ), 0, 0). The equation of the envelope is obtained by
                                         2
                                  2
                                                    2
                           elimination of parameter t from the system

                                             f 1 (X, t)=  X − E(t),X − E(t)  =0,         (5.11)
                                                        ˙
                                             f 2 (X, t)=  E(t),X − E(t)  =0.
                           Here f 1 ,f 2 are rational functions in the variable t, i.e. f 1 = g 1 (X, t)/p(t),f 2 =
                           g 2 (X, t)/q(t). Let

                                              F(X)=Res g 1 (X, t),g 2 (X, t),t           (5.12)
                           be the resultant of two polynomials g 1 ,g 2 (here we assume that degree of poly-
                           nomials p, q are minimal). A priori F may be a reducible polynomial, i.e. F =
                           F 1 F 2 ··· F n then one factor, assume F 1 , is the equation of Γ(E). After easy compu-
                           tation with MAPLE package we see that degree of F is 8.
                                               √
                              For example, if a =  2 and b =1 the equation of Γ(E) is
   85   86   87   88   89   90   91   92   93   94   95