On the dual billiard problem (Q1908486)

The paper begins with a set up of the dual billiard problem with a smooth dual billiard curve \(\gamma\) with positive curvature. Let \(x\) be a point outside of the billiard curve. There are two tangent lines to \(\gamma\) through \(x\), choose one of them and reflect \(x\) in the point of tangency, we obtain a dual billiard map \(T\). Does the billiard map have periodic orbits? Can its orbits have some points (or one point) of \(\gamma\) as limit point? What is the relation between the classical billiard problem and the dual billiard problem? What is the situation in higher dimension? and so on. The present article addresses some of these questions and contains some answers to them. Some of the basic observations in the paper are the following: 1. \(T\) has periodic orbits of all periods \(n \geq 3\). 2. The dual billiard map is area-preserving. 3. The dual billiard map has invariant curves arbitrarily close to \(\gamma\) and arbitrarily far from it. 4. The higher dimensional dual billiard map \(T\) is a symplectomorphism. 5. Existence of periodic orbits for higher dimensional dual billiard maps.