Chaotic motion in the breathing circle billiard

The paper rigorously proves that, for a class R̃ of radius functions R(t), there exists c0 > 0 such that the reduced billiard map P has positive topological entropy and supports many invariant probability measures with positive metric entropy, via an Aubry–Mather “converse KAM” argument that shows the absence of invariant circles on an interval of rotation numbers and then applies a metric-entropy criterion for twist maps (see the definition of P via a generating function in Proposition 4.4 and the main chaotic-dynamics theorem in Theorem 6.1) . By contrast, the model’s proposed proof attempts a different route: find a hyperbolic m-resonant fixed point for the c=0 integrable map and then use Melnikov splitting to produce a transverse homoclinic and a Smale horseshoe. The key step asserted by the model—that the m-resonant fixed point is hyperbolic under only the assumptions defining R̃—requires the inequality m^2(−R''(t̄))/R(t̄) > 4, which is not implied by the paper’s hypotheses (Definition 2.2 and the bounds used in Lemma 6.2) and thus is unproven. Without establishing this hyperbolicity, the Melnikov/horseshoe part is unsupported. The rest of the model’s computations (e.g., the small-c expansion of h and the formal Melnikov potential derivative) are consistent algebraically with the paper’s generating function h(t0,t1) in (4.5) but do not fill the missing hypothesis. Hence the paper’s argument stands, while the model’s proof is incomplete for the stated class R̃.