On absolutely focusing mirrors (Q4026039)

[For the entire collection see Zbl 0744.00035.] A plane billiard is a Hamiltonian system whose potential is zero inside some closed domain \(Q\) and infinity in its boundary \(\partial Q\), which we suppose to consist of a finite number of curved mirrors, called components, of class \({\mathcal C}^ 3\). Assume also that for each component either the curvature is always of the same sign or the curvature of the component is identically zero. A component is called dispersing, focusing or neutral if these signs are respectively \(> 0\), \(< 0\), \(= 0\). Billiard trajectories are straight lines inside \(Q\) and we assume these trajectories obey the law of reflection. The difficulty arising from hitting a singular point of \(\partial Q\) is taken care by proving that the invariant phase volume of the set of such trajectories is zero. A billiard is a Sinai billiard if all components of the boundary are dispersing. These billiards are mixing systems with positive entropy if all regular components of \(\partial Q\) intersect transversally. If \(\partial Q\) has only neutral components, the billiard has zero entropy. But if there are focusing components, a number of questions arise. The author calls a focusing component absolutely focusing (notation \(\Gamma^ f\)) if for any infinitesimal beam of parallel rays which falls upon \(\Gamma^ f\) after the series of consecutive reflections from \(\Gamma^ f\) this beam will leave \(\Gamma^ f\) as convergent (focusing). The author has the following conjecture: A focusing component \(\Gamma^ f\) can be a regular component of a boundary of mixing billiard if and only if it is absolutely focusing. This paper is part of a series of papers trying to prove the conjecture. In the paper the author also shows a relation between billiards and continued fractions.