Table of Contents
Fetching ...

Lagrangian split tori in $S^2 \times S^2$ and billiards

Joé Brendel, Joontae Kim

TL;DR

This work classifies split Lagrangian tori in $X_{\alpha}=(S^2\times S^2,\omega_{\alpha})$ with a non-monotone split form by reducing the problem to good billiard trajectories in rectangles, yielding an alpha-independent classification. The central machinery uses symmetric probes and billiard unfoldings to realize fibre equivalences and to derive Chekanov-type obstructions that constrain Hamiltonian monodromy, producing a detailed monodromy theorem across regions of the Delzant square. The results have wide-ranging applications, including a complete ball-embedding dichotomy for split tori, implications for Lagrangian packing (including infinite packings in irrational cases), and a late-Poincaré recurrences framework, together with new insights into the topology of the space of Lagrangian tori. The work also confirms that certain toric-fibration fibres in non-monotone $S^2\times S^2$ are not exotic and establishes a robust connection between billiard dynamics and symplectic topology, offering a versatile toolkit for future extensions to other toric manifolds and Hirzebruch surfaces.

Abstract

In this paper, we classify up to Hamiltonian isotopy Lagrangian tori that split as a product of circles in $S^2 \times S^2$, when the latter is equipped with a non-monotone split symplectic form. We show that this classification is equivalent to a problem of mathematical billiards in rectangles. We give many applications, among others: (1) answering a question on Lagrangian packing numbers raised by Polterovich--Shelukhin, (2) studying the topology of the space of Lagrangian tori, and (3) determining which split tori are images under symplectic ball embeddings of Chekanov or product tori in $\mathbb{R}^4$.

Lagrangian split tori in $S^2 \times S^2$ and billiards

TL;DR

This work classifies split Lagrangian tori in with a non-monotone split form by reducing the problem to good billiard trajectories in rectangles, yielding an alpha-independent classification. The central machinery uses symmetric probes and billiard unfoldings to realize fibre equivalences and to derive Chekanov-type obstructions that constrain Hamiltonian monodromy, producing a detailed monodromy theorem across regions of the Delzant square. The results have wide-ranging applications, including a complete ball-embedding dichotomy for split tori, implications for Lagrangian packing (including infinite packings in irrational cases), and a late-Poincaré recurrences framework, together with new insights into the topology of the space of Lagrangian tori. The work also confirms that certain toric-fibration fibres in non-monotone are not exotic and establishes a robust connection between billiard dynamics and symplectic topology, offering a versatile toolkit for future extensions to other toric manifolds and Hirzebruch surfaces.

Abstract

In this paper, we classify up to Hamiltonian isotopy Lagrangian tori that split as a product of circles in , when the latter is equipped with a non-monotone split symplectic form. We show that this classification is equivalent to a problem of mathematical billiards in rectangles. We give many applications, among others: (1) answering a question on Lagrangian packing numbers raised by Polterovich--Shelukhin, (2) studying the topology of the space of Lagrangian tori, and (3) determining which split tori are images under symplectic ball embeddings of Chekanov or product tori in .

Paper Structure

This paper contains 32 sections, 30 theorems, 90 equations, 7 figures.

Key Result

Theorem 1.3

Let $T(x,y),T(x',y') \subset X_{\alpha}$ be split tori. Then $T(x,y) \cong T(x',y')$ if and only if the good billiard trajectory emanating from $(x,y)$ has any of the points $(\pm x', \pm y')$ as admissible bouncing point.

Figures (7)

  • Figure 1: A split torus $T(x,y) \subset X_{\alpha}$ on the right-hand side and its base point in the rectangle $\square_{\alpha} = [-\alpha-1,\alpha+1] \times [-\alpha,\alpha]$ on the left-hand side. The parameters $x$ and $y$ correspond to the signed areas between the respective circles and equators.
  • Figure 2: A point $(x,y) \in \square_{\alpha}$, its billiard table $\square_{r(x,y)}$, part of its good billiard trajectory (dotted line), and corresponding bouncing points (black dots). The set $\Sigma$ is indicated by dashed lines and the set $Q$ as defined in \ref{['eq:Qdefset']} in grey.
  • Figure 3: Illustration of \ref{['thm:ballcases']}. The set of ball-Clifford split tori is indicated by dashed lines. The dots represent the second set in the union \ref{['eq:nonmonotonecomplement']} i.e. the tori which are ball-Chekanov but not ball-nonmonotone. The central segment drawn by a solid line consists of the points which are of neither of the three types in \ref{['def:types']}. The set of ball-nonmonotone tori is the complement of the union of dashed lines, the central segment and the dots. The set of ball-Chekanov tori is dense in $\square_{\alpha}$ and hence cannot be meaningfully represented.
  • Figure 4: Some examples and one non-example of symmetric probes in $\square_{\alpha}$. The segments $\sigma_k$ have slope $(k,-1)$ for $k \in \{-1,0,1,2,3\}$ and $\sigma$ has slope $(1,0)$. The dashed segment $\sigma_3$ is not a symmetric probe, since its intersection with the the vertical edge is not integrally transverse.
  • Figure 5: Sketch of the folding map $F_r \colon \mathbb{R}^2 \rightarrow \square_r$. It maps the line $l_{(x,y)}$ of slope $(1,1)$ on the left-hand side to the good billiard trajectory of $(x,y)$ on the right hand side and intersection points with the grid to bouncing points.
  • ...and 2 more figures

Theorems & Definitions (48)

  • Definition 1.1
  • Definition 1.2
  • Theorem 1.3
  • Remark 1.4
  • Remark 1.5
  • Remark 1.6
  • Remark 1.7
  • Definition 1.8
  • Corollary 1.9
  • Corollary 1.10
  • ...and 38 more