billiards.ams

# Although it was illustrated here with a billiard ow

This preview shows page 1. Sign up to view the full content.

This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: t lead naturally to the Math. Intelligencer, vol.17 no.1, (1995), 8{16. 14 J.L. King generalization for ows on an arbitrary pinched cusp. Although it was illustrated here with a billiard ow, the conservativity result holds mutatis mutandis for any measure-preserving continuous ow whose induced measure is pinched. The illustrations in this article were drawn with the excellent computer facilities at the Mathematical Sciences Research Institute, whom I thank for its hospitality. (A1) In the setting of Lemma 9, under a continuous ow transformation the set of recurrent points {in addition to being a full-measure set{ must be residual (must include a dense G set), once one adds the natural assumption that gives positive measure to every non-empty open set. In contrast, if there is no such conservative invariant measure , then the set of recurrent points need not be residual. Nonetheless, Birkho established that there is at least one recurrent point under any continuous ow or transformation on a compact space. The transformation x x + 1 on the topological circle R shows that no more can be guaranteed. (A2) An unexplained coincidence occurs for ux measure in the special case where our billiard table is bounded by an elliptical cushion C := @ . It turns out that the set C of inward pointing cueballs breaks up into TC-invariant subsets; one for each ellipse E which is inside of, and has the same foci as, C . The invariant set consists of those cueballs v C whose ow trajectory will pass tangent to E before it again hits C . This invariant decomposition of C implies that ux( ), on C, breaks up into measures parameterized by confocal ellipses E . When suitably normalized, each of these measures turns out to the the \Poncelet CE -measure&quot; of 1], which arises from what appears to be an entirely unrelated construction. (A3) Billiard ows are a kind of geodesic ow on surfaces of only zero and in nite curvature. A stronger result (see, for example, Donnay 1988) is known for the geodesic ow on the surface-of-revolution around the x-axis generated by a di erentiable f : 0; ) R+. If the surface is \pinched&quot;, liminfx!1 f (x) = 0, then every geodesic orbit is bounded {except for the obvious ones which ow directly out the cusp. (A4) An open and probably di cult research question is suggested by Sullivan's 1982 result on the geodesic ow on a cusp of constant negative curvature. Letting dist(v) denote the distance of the footpoint of v to some chosen point on the surface, Sullivan gives an explicit speed function D(t) such that 7! f1g 2 1 ! limsup dist((t) v) = 1 ; D t !1 t for a.e. v. Paul Shields raised the tantalizing question of characterizing the nite-area cuspidal billiard tables which have such a speed function. References V.J. Donnay, Geodesic ow on the two-sphere, Part I: Positive measure entropy, Ergodic Theory and Dynamical Systems 8 (1988), 531{553. J.L. King, Three Problems in search of a Measure, Amer. Math. Monthly 101 # 7 (1994), 609{628. D. Sullivan, Disjoint spheres, approximation by imaginary quadratic numbers, and the logarithm law for geodesics, Acta Math. 149 (1982), 215{237. Printed: February 14, 1995 Filename is Article/18Billiards/billiards.ams.tex...
View Full Document

Ask a homework question - tutors are online