Page 91 - Geometric Modeling and Algebraic Geometry
P. 91

5 Canal Surfaces Defined by Quadratic Families of Spheres  89

                                                                                      2
                              F = ω + 6+2x − 4 y   2  ω + 13 − 28 y +4 x + x +2 y   2   ω 2
                                                                              2
                                                                        2
                                   4
                                             2
                                                                   2
                                                      3

                                +6 x +2 − 4 y  2  −x +1 − 2 y 2  ω
                                     2
                                                    2
                                                                                           3
                                + 4 − 12 x +2 y  2  − 3 x − 4 y 2  4 y +5 x 2  − 4 x +2 y  2  ,
                                                         2
                                                                    2
                                                                                  2
                                           2
                           where ω = x + y + z − r .
                                              2
                                                   2
                                      2
                                          2
                              Moreover, F is an irreducible polynomial. Indeed, a plane curve Γ(E) ∩{r =
                           const}∩{z =0} is defined by an irreducible polynomial of degree 8 as it is an
                           offset of the ellipse E (see, e.g. [8]). Hence deg Γ(E)=8 as well.
                           Theorem 11. Let E be an ellipse of type E ++ . Then the hypersurface Γ(E) is a real
                           3-dimensional variety of degree 8, and its set of finite real double points has four
                           parts E ∪ H 1 ∪ H 2 ∪ Ω where:
                                      E :        x /a + y /b =1,           z = r =0,
                                                  2
                                                         2
                                                     2
                                                            2
                                     H 1 : x /(a − b ) − z /b + r /b =1,   y =0,
                                            2
                                                                2
                                                   2
                                                                  2
                                               2
                                                           2
                                                         2
                                     H 2 : −y /(a − b ) − z /a + r /a =1,  x =0,
                                            2
                                                                2
                                                    2
                                                         2
                                                            2
                                                2
                                                                   2
                           H 1 and H 2 are hyperboloids of 1-sheet and 2-sheet, resp., (the only non-isotropic
                           lines on Γ(E) are two rulings of the hyperboloid H 1 ), Ω is the absolute quadric.
                           Proof. It remains to determine double points. We start from the following geometric
                           description of Γ(E). The hypersurface Γ(E) consists of points in R that define
                                                                                    4
                                                                                    1
                           spheres tangent to the ellipse. All spheres touching in one point define an isotropic
                           line in R . If two isotropic lines have a common point then this point is a double
                                  4
                                  1
                           point on Γ(E). Therefore, a common point of two isotropic lines corresponds to a
                           sphere that touches the ellipse E in two points. There are two families of circles on
                           the (x, y)-plane that touch the ellipse E in two points. One family consists of inside
                           circles with centers on the x-axis, another one of outside circles with centers on the
                           y-axis. A pencil of spheres that contains this circle goes through each of such circle,
                           i.e. spheres of the pencil are tangent to the ellipse E in two points. We notice that
                           these two families of spheres are defined by two equations y =0 and x =0 in
                           R . Therefore, the equations of the hiperboloids H 1 and H 2 are obtained from the
                             4
                             1
                           equation F =0 as the hyperplane sections.
                           Remark 12. For the ellipse E in Theorem 11 hypersurface Γ(E) has also other sin-
                           gularities (not only double points). Consider a projection of Γ(E) to the (x, y)-plane.
                           Singularities of the ellipse offsets define the evolute curve (see Fig. 5.8), which is an
                           envelope of normal lines to the ellipse E in the plane. Let a surface K ⊂ Γ(E) be
                           an envelope of all isotropic lines in Γ(E). Since the isotropic lines are projected to
                           normals, K is projected to the evolute, and all points of K are singular in Γ(E). The
                           well-known parametrization of the evolute enables us to parametrize K:
                                                       (b + a )sin t    1 − s 2   1+ s 2
                                                                  3
                                                             2
                                                         2
                                K(t, s)=((b − a )cos t,             ,h(t)     ,h(t)     ),
                                           2
                                                    3
                                               2
                                                             a            2s        2s
                           where h(t) =(b sin (t)+ a cos (t)) /a . Note that a general point on K be-
                                              2
                                          2
                                    2
                                                    2
                                                               2
                                                            3
                                                        2
                           longs to a segment of an isotropic line on Γ(E) bounded by hyperplanes x =0 and
   86   87   88   89   90   91   92   93   94   95   96