Next: 5.8.1.5 Computing starting points Up: 5.8.1 Rational polynomial parametric/implicit Previous: 5.8.1.3 Characteristic points   Contents   Index

5.8.1.4 Analysis of singular points

Let us construct a parametric equation of a straight line through a point on the algebraic curve
    (5.93)

where and are constants and is a parameter [437,95,107]. We find the intersections between and the algebraic curve by determining the roots of . Since , Taylor expansion of the left hand side gives
    (5.94)

where partial derivatives of are evaluated at .

When and are not both zero ( ) at , (5.94) has a simple root and every line through has a single intersection with the algebraic curve at except for one case where for certain values of and . In such cases (5.94) has a double root , provided at least one of the second order partial derivatives is not zero ( ), and is tangent to the curve at .

When is a singular point ( ), and at least one of , , is not zero ( ), then is a double root and has at least two intersections at except for the values of and which satisfy

    (5.95)

In such cases, is a triple root, provided at least one of the third order partial derivatives is not zero ( ). We can solve the quadratic equation (5.95) for or which leads to the following three possibilities:
(1) Two real distinct roots: These values correspond to two distinct tangent directions at the singular point, which implies the algebraic curve has a self-intersection. The Folium of Descartes, which is shown in Fig. 1.1, has such singularity at the origin.
(2) One real double root: This value corresponds to one tangent direction at the singular point, which implies a cusp. An illustrative example, which is a semi-cubical parabola, is given in Fig. 2.3.
(3) Two complex roots: No real tangents at the singular point imply an isolated point. An example of an isolated point is given in Example 5.8.1 (see Fig. 5.21).

Example 5.8.1. Let the algebraic curve be [437], then

     
     

The -turning points can be found by finding the roots of and . We immediately deduce . Upon substitution to we obtain . Since , (0,0) is not a -turning point. Therefore (-1,0) is the only -turning point. On the other hand -turning points, which satisfy and , have no real solutions. It is apparent from the above discussion that is the only singular point. Tangents at can be obtained from , which gives , and hence no real solution. Therefore, is an isolated point. If the domain of interest is , border points are . The above algebraic curve is depicted in Fig. 5.21.

Figure 5.21: Algebraic curve with an isolated point at (0,0)

Example 5.8.2. We have studied the semi-cubical parabola in Example 2.1.1. The curve has a singular point at . Since

     
     

we have . At (0,0) we have a double root . Thus, at the singular point (0,0) we have a cusp whose tangent direction is along the axis as shown in Fig. 2.3.

Example 5.8.3. Let us consider the equation

     

within the domain , taken from Geisow [124]. This is a degree 6 algebraic curve illustrated in Fig. 5.22. On every border line segment, there are three border points. The curve has no singular points, but involves two (internal) loops and six border-to-border branches. The algebraic curve in this example has degrees in and . Consequently, using the previous formulae the number of turning points, turning points and singular points (in the entire complex plane) is bounded by , , and . However, as we can see in Fig. 5.22, these numbers overestimate the actual number of such points in the real square .
Figure 5.22: A degree six algebraic curve (adapted from [124])



Next: 5.8.1.5 Computing starting points Up: 5.8.1 Rational polynomial parametric/implicit Previous: 5.8.1.3 Characteristic points   Contents   Index
December 2009