Vertices of planar curves under the action of linear transformations (Q1610990)

What happens to a vertex (or a non-vertex) of a plane curve \( \gamma \) under the action of a linear transformation? This question makes sense since the notion of curvature is not invariant under the linear group. The authors study this problem, recognizing the obvious relations to affine differential geometry. A first observation is that for every point there exists a class of transformations changing the point to a vertex. Globally, for closed smooth Jordan curves with positive curvature, it is proved that there is always a nonvoid open set of linear transformations yielding a curve with at least six vertices. The actual access to this and related results is provided by the authors' invention of the so called star curve \( \gamma^{\ast} \). The star curve, lying on a two-sphere of transformations, plays the role of a bifurcation locus for the number of vertices. Its points correspond to those linear transformations which lead to the vanishing of curvature derivatives of order at least two (multiple quasi-vertices). Using this technique another main result is gained stating in connection with the six sextatic point theorem of \textit{S. Mukhophadhyaya} [Bull. Calcutta Math. Soc. 1, 31-37 (1909; JFM 40.0624.01)] that among the six sextatic points there is at least one where the osculating conic is an ellipse. There are intimate relations to non-metric curve theory. For example, the quantities controlling the availability of multiple quasi-vertices are multiples of well known invariants of affine and projective curve theory. The results, as they stand, apply to curves with vertices of finite order. Modifications could become necessary in case of vertices with vanishing curvature derivatives of all orders.