Nearly circular domains which are integrable close to the boundary are ellipses (Q1752923)

This paper considers the standard billiard map on a billiard table that has a strictly convex boundary. It is known that the billiard system is globally integrable if the boundary of the table is a circle, and integrable if the boundary is an ellipse. Birkhoff (or perhaps Birkhoff and Poritsky) conjectured that the boundary of a strictly convex integrable billiard table must be an ellipse. The conjecture remains unresolved. The authors consider the question using a stronger notion of integrability, rational integrability (a local integrability near the boundary). The critical definition is as follows: if \(\Omega\) is a strictly convex domain in \(\mathbb{R}^2\), a curve is called an integrable rational caustic for the billiard map in \(\Omega\) if the corresponding non-contractible invariant curve of periodic points, so the corresponding rotation number is rational. If \(q_0\geq 2\) and if the billiard map in \(\Omega\) admits rational caustics of rotation number \(p/q\) for all \(p/q\) with \(0<p/4<{1\over q_0}\), then \(\Omega\) is said to be \(q_0\)-rationally integrable. The authors then conjecture the following: for any \(q_0\geq 3\) there exist \(e_0= e_0(q_0)\in (0,1)\), \(m_0= m_0(q_0)\), \(n_0= n_0(q_0)\in\mathbb{N}\), the following holds: for each \(0<e\leq e_0\) and \(c>0\) there exists \(\varepsilon= \varepsilon(e,c,q_0)\) such that any \(q_0\)-rationally integrable \(C^{m_0}\)-smooth domain \(\Omega\) whose boundary is \(C^{n_0}\) \(\varepsilon\)-close to the ellipse \({\mathcal E}_{e,c}\) (with eccentricity \(e\) and semi-focal distance \(c\)) is itself an ellipse. Their main theorem says that the conjecture is true for \(q_0= 2,3,4,5\) with \(m_0= 40q_0\) and \(n_0= 3q_0\). The bulk of the paper contains a proof of the conjecture in same cases, and a proof of the other cases based on some additional assumptions of non-degeneracy.