Algebraic billiards in the Fermat hyperbola

Basic Information

• Paper ID: 2406.16172
• Title: Algebraic billiards in the Fermat hyperbola
• Author: Max Weinreich (Harvard University)
• Classification: math.DS (Dynamical Systems)
• Publication Date: November 4, 2025 (v3)
• Paper Link: https://arxiv.org/abs/2406.16172v3

Abstract

This paper proves two results concerning billiard dynamics in general algebraic curves of degree $d \geq 2$: First, the dynamical degree grows quadratically in $d$; second, the set of complex periodic points has measure zero, which implies Ivrii's conjecture for the classical billiard map in general algebraic regions. To prove these results, the author studies a new billiard table—the Fermat hyperbola—whose indeterminacy points satisfy exceptional properties. Over the complex numbers $\mathbb{C}$, an algebraically stable model is constructed via iterated blowups; over more general fields, essential stability is established for certain large and nef divisors.

Research Background and Motivation

Problem Background

Classical billiard systems are discrete-time dynamical systems describing a point particle bouncing inside a planar region $\Omega \subset \mathbb{R}^2$, and represent a classical problem in dynamical systems theory. The core objective of billiard research is to prove that general billiards exhibit chaotic properties.

Core Problems

This paper focuses on two fundamental questions:

1. Dynamical Degree Problem: The dynamical degree $\lambda_1$ is an algebraic analogue of entropy, controlling the degree growth of classical curve sequences (such as wavefronts and reflected caustics) in billiard systems. How does the dynamical degree grow for billiards in algebraic curves of degree $d$?
2. Ivrii's Conjecture: This conjecture asserts that the set of periodic points in billiards within smooth boundary planar regions has measure zero. It is one of the central open problems in billiard theory, closely related to Weyl's conjecture on the asymptotic behavior of the spectrum of the Laplace operator.

Limitations of Existing Methods

• Birkhoff's Conjecture: Although it was recently proven that $C^2$-generic convex billiards have positive topological entropy, proving $h_{top} > 0$ for specific non-elliptic billiards remains difficult.
• Progress on Ivrii's Conjecture: Known cases include strictly convex analytic boundaries, concave piecewise analytic regions, and $C^{\infty}$-generic regions. However, for periods 5 and higher, the problem remains open.
• Challenges in Algebraic Billiards: Due to the multivalued nature of the correspondence, describing the dynamics of any specific algebraic billiard is a difficult problem. A general construction method for algebraically stable models is lacking.

Research Motivation

The author employs a specialization strategy: by carefully selecting a special algebraic curve (the Fermat hyperbola), the author leverages its special geometric properties to prove properties of general curves. The key property of the Fermat hyperbola is that its indeterminacy points satisfy exceptional conditions, which greatly simplifies the analysis of indeterminate orbits.

Core Contributions

1. Quadratic Lower Bound on Dynamical Degree (Theorem 1.2): For general algebraic curves of degree $d$, the dynamical degree of the billiard correspondence satisfies $\lambda_1(b_{gen}) \geq \frac{2d^2 - 3d + \sqrt{(2d^2-3d)^2-4(d-1)}}{2} \geq 2d^2 - 3d - 1$ This is an algebraic analogue of chaos in general billiards.
2. Proof of Ivrii's Conjecture (Theorem 1.5):
   • Proves that billiards in general complex algebraic curves of degree $d \geq 2$ are not $n$-reflective for all periods $n$
   • Derives that the set of periodic points of the classical billiard map has measure zero for real algebraic plane curves with algebraically independent coefficients
   • Therefore, Weyl's conjecture holds for these real regions
3. Exact Computation for Fermat Hyperbola Billiards (Theorem 6.5): Computes the exact dynamical degree of the Fermat hyperbola billiard: $\lambda_1(b) = \frac{2d^2 - 3d + \sqrt{(2d^2-3d)^2-4(d-1)}}{2}$
4. Construction of Improved Birational Models (Theorem 1.7):
   • Constructs an essentially stable model for Fermat hyperbola billiards for all degrees $d \geq 2$
   • Over the complex numbers, constructs an algebraically stable model when $d$ is odd
   • Proves that the billiard is completely integrable for $d=2$ and admits a regular model for $d=3$

Methodology Details

Task Definition

Algebraic Billiard Correspondence: Given a smooth algebraic curve $C \subset \mathbb{P}^2_k$ of degree $d \geq 2$ and the unit tangent space $D \cong \mathbb{P}^1$ of a non-degenerate quadratic form $\Theta$, the billiard correspondence is defined as $b_{C,D} : C \times D \, \, \dashv \!\!\!\!\!\! \vdash \, \, C \times D$ It is the composition of the secant correspondence $s$ and the reflection correspondence $r$: $b = r \circ s$.

• Secant Correspondence $s$: $(x,v) \mapsto \{(x', v) : x' \in C \cap \ell(x,v) \setminus \{x\}\}$, where $\ell(x,v)$ is the line through $x$ in direction $v$
• Reflection Correspondence $r$: $(x,v) \mapsto (x, r_x(v))$, where $r_x$ is reflection across the tangent line to $C$ at $x$

Geometric Structure of the Fermat Hyperbola

Definition: The Fermat hyperbola is defined as $C: (X_0 - iX_1)^d + (X_0 + iX_1)^d = X_2^d$ paired with the standard quadratic form $D: Q_0^2 + Q_1^2 = Q_2^2$.

Key Properties (Lemmas 3.7, 3.8):

1. $C$ is a smooth curve in $\mathbb{P}^2$
2. There are $d$ points at infinity, none of which are isotropic points $[1:\pm i:0]$
3. Points with tangent slopes $\pm i$ are exactly $\{[\frac{\zeta}{2} : \mp i\frac{\zeta}{2} : 1] : \zeta^d = 1\}$
4. At these points, the tangent lines have maximal contact order with $C$ and no other intersection points

Exceptional Property (Lemma 3.8): This is the core property of the Fermat hyperbola

• $|\text{Ind}\, s| = 2d$, $|\text{Ind}\, r| = 2d$
• $\text{Ind}\, s \cap \text{Ind}\, r = \emptyset$
• If $p \in \text{Ind}\, s$, then $r^{-1}(p) = \{p\}$
• If $p \in \text{Ind}\, r$, then $s^{-1}(p) = \{p\}$

This property greatly simplifies the structure of indeterminate orbits.

Model Construction: Iterated Blowups

Local Formula (Proposition 3.11): Near $p \in \text{Ind}\, r$, choose formal coordinates $(y,z)$ such that:

• Reflection: $r(y,z) = (\frac{z^{d-1}(1+O(z^d))}{y}, z)$
• Secant: given by the system $y' = y, \quad -dy^2 = z^{d-1} + z^{d-2}z' + \cdots + (z')^{d-1} + A(z,z')$ where $A \in \langle z, z' \rangle^d$

Standard Iterated Blowup (Definition 3.12): For the formal neighborhood $(A^2, 0)$, define the $(d-1)$-fold iterated blowup $\pi_k: V_{\pi_k} \to (A^2, 0)$:

• Step 1: Blowup at the origin, obtaining exceptional divisor $E_1$
• Step $k$: Blowup at the specific point $(0,0)$ on $E_{k-1}$, obtaining $E_k$
• In local coordinates $(u_k, v_k)$, $\pi_k(u_k, v_k) = (u_k v_k^k, v_k)$

Construction of Model $P$ (Definition 3.13): $\pi: P \to C \times D$ is the composition of the following blowups:

1. Simple blowup at each $p \in \text{Ind}\, s$
2. $(d-1)$-fold standard iterated blowup at each $p \in \text{Ind}\, r$

Exceptional divisor structure: $E = \sum_{p \in \text{Ind}\, r} \sum_{k=1}^{d-1} E_k^{(p)} + \sum_{p \in \text{Ind}\, s} E^{(p)}$ with a total of $2d^2$ irreducible components.

Define the lifted correspondences: $\hat{b} = \pi^{-1} \circ b \circ \pi, \quad \hat{r} = \pi^{-1} \circ r \circ \pi, \quad \hat{s} = \pi^{-1} \circ s \circ \pi$

Technical Innovations

1. Non-standard Blowup Strategy: Rather than repeatedly blowing up to stabilize orbits (which may introduce new destabilizing orbits), the author designs the correct iterated blowup all at once based on local formulas.
2. Essential Stability Concept: Introduces a weakened stability concept—the existence of a large and nef divisor $\Delta$ such that $(f^n)_* \Delta \cdot \Delta = (f_*)^n \Delta \cdot \Delta$ This suffices to compute the dynamical degree and remains applicable when algebraic stability cannot be constructed.
3. Dynamical Analysis of Midpoint Divisors: When $d$ is odd, $E_{mid}^{(p)} = E_{(d-1)/2}^{(p)}$ plays a special role in billiard dynamics. By studying the complex dynamics of the 1-dimensional correspondence $\beta$ on it, algebraic stability is proven.
4. Double Cover Technique: When $d$ is even, by constructing a local double cover $\delta: \tilde{G} \to G$, $(y,z) \mapsto (\tilde{y}, \tilde{z}^2)$, an "artificial midpoint divisor" is created in the lifted space.

Experimental Setup

This is pure mathematical theoretical research with no numerical experiments or datasets. All results are obtained through rigorous mathematical proof.

Computational Verification

The author provides specific matrix calculations to verify the dynamical degree:

Proposition 6.4: The quotient space $N_0 = N/N_1$ is a free abelian group of rank 4, generated by $\hat{C}_0, \hat{D}_0, E_{\infty}, F_{d-1}$. On this basis:

Pushforward of reflection: $r = \begin{pmatrix} 1 & 0 & 0 & 0 \\ d(d-1) & 1 & 0 & 2d \\ 0 & 0 & 1 & 0 \\ -(d-1) & 0 & 0 & -1 \end{pmatrix}$

Pushforward of secant: $s = \begin{pmatrix} d-1 & 2 & d-1 & 0 \\ 0 & d-1 & 0 & 0 \\ 0 & -1 & -1 & 0 \\ 0 & 0 & 0 & d-1 \end{pmatrix}$

Pushforward of billiard: $b = \begin{pmatrix} d-1 & 2 & 2d & 0 \\ d^3-2d^2+d & 2d^2-d-1 & 2d^3 & 2d^2-2d \\ 0 & -1 & -1 & 0 \\ -(d-1)^2 & -2d+2 & -(2d-2)d & -(d-1) \end{pmatrix}$

Experimental Results

Main Results

Theorem 6.5 (Exact Dynamical Degree for Fermat Hyperbola): $\lambda_1(b) = \frac{2d^2 - 3d + \sqrt{(2d^2-3d)^2-4(d-1)}}{2}$

This value comes from the largest root of the characteristic polynomial $\det(\lambda I - b) = (\lambda - (d-1))^2(\lambda^2 - (2d^3-2d)\lambda + d - 1)$

Specific Numerical Values:

• $d=2$: $\lambda_1 = 1$ (completely integrable case)
• $d=3$: $\lambda_1 = \frac{15 + \sqrt{221}}{2} \approx 14.93$
• $d=4$: $\lambda_1 = \frac{28 + \sqrt{780}}{2} \approx 27.93$
• Asymptotically: $\lambda_1 \sim 2d^2$ (quadratic growth)

Theorem 1.2 (Lower Bound for General Curves): Through specialization arguments, the dynamical degree of general degree $d$ curves is at least the above value.

Algebraic Stability Results

Proposition 6.3 (Essential Stability): For all $d \geq 2$ and all fields $k$ (under characteristic conditions), $\hat{b}: P \, \, \dashv \!\!\!\!\!\! \vdash \, \, P$ is essentially stable, hence $\lambda_1(b) = \text{rad}\, \hat{b}_*$.

Theorem 7.5 (Odd Degree Case): When $k = \mathbb{C}$ and $d$ is odd, $\hat{b}$ is algebraically stable.

Theorem 7.13 (Even Degree Case): When $k = \mathbb{C}$ and $d$ is even, the modified model $\hat{b}_+: P_+ \, \, \dashv \!\!\!\!\!\! \vdash \, \, P_+$ is algebraically stable.

Proof Key: Uses the invariant subset $U = \mathbb{C} \setminus D$ (exterior of the unit disk) on the midpoint divisor $E_{mid}$. Through Lemma 7.4 (odd case) or Lemma 7.12 (even case), one proves $\hat{b}(U) \subset U$, hence $\hat{b}^n(\text{Exc}\, \hat{b}) \cap \text{Ind}\, \hat{b} = \emptyset, \quad \forall n \geq 0$ which is a sufficient condition for algebraic stability.

Proof of Ivrii's Conjecture

Theorem 7.14 (= Theorem 1.5):

1. Fermat Hyperbola: For all $n \in \mathbb{N}$, the Fermat hyperbola billiard is not $n$-reflective.
   Proof Sketch: Assume $b$ is $n$-reflective, so the diagonal is contained in $\Gamma_{b^n}$. Take $p \in \text{Ind}\, r$, and the point $u_0 = \infty \in E_{mid}^{(p)}$ satisfies $\hat{b}(u_0) \subset U$ but $u_0 \notin U$, contradiction.
2. General Curves: Billiards in general complex algebraic curves of degree $d \geq 2$ are not $n$-reflective.
   Proof Sketch: By specialization arguments, if general curves are $n$-reflective, then the Fermat hyperbola should also be $n$-reflective, contradicting (1).
3. Ivrii's Conjecture for Real Billiards: Let $T \subset \mathbb{R}^2$ be a real algebraic curve defined by a degree $d$ polynomial with algebraically independent coefficients, and $\Omega$ be a bounded component of $\mathbb{R}^2 \setminus T$. Then the set of periodic points of the classical billiard map has measure zero.
   Proof Sketch: The set of $n$-periodic points $R_n$ is a proper subvariety of $C \times D$ with dimension at most 1. Hence $R_n(\mathbb{R})$ has measure zero in $W = T_C \times D$. Taking the union over all $n$ yields the conclusion.

Case Studies

Specific Example (end of Section 7.3): The curve $T: 0 = e^{\sqrt{2}}x^4 + 0.3e^{\sqrt{3}}x^3y + e^{\sqrt{5}}x^2y^2 + \cdots + 0.3e^{\sqrt{23}}$ has a bounded non-convex component whose coefficients are algebraically independent by the Lindemann-Weierstrass theorem. This is a new example satisfying Ivrii's conjecture.

Related Work

Classical Billiard Theory

1. Birkhoff's Conjecture: Ellipses are the only strictly convex smooth billiards with zero topological entropy. Recently BDMLD+24 proved that $C^2$-generic convex billiards have positive topological entropy.
2. Known Cases of Ivrii's Conjecture:
   • Strictly convex globally regular analytic boundaries Vas84
   • Concave piecewise analytic regions Vas86
   • $C^{\infty}$-generic regions PS88
   • Period-3 and period-4 points Ryc89, GK12

Algebraic Billiards

1. Glutsyuk's Work Glu14a, Glu14b, Glu21:
   • Generalizes billiards to complex algebraic curves
   • Classifies 3-reflective and 4-reflective billiards
   • Proves the case of odd periodic orbits (when there are no isotropic points at infinity)
   • Proves the polynomial Birkhoff conjecture
2. Author's Previous Work Wei25:
   • Provides an upper bound on the dynamical degree of general billiards: $\lambda_1(b_{gen}) \leq \rho_d < 2d^2 - d - 3$
   • Identifies and resolves some destabilizing orbits

Algebraic Dynamics

1. Dynamical Degree Theory DS08, Tru20:
   • Dinh-Sibony introduce the dynamical degree for complex correspondences
   • Truong generalizes to arbitrary algebraically closed fields
   • The dynamical degree of correspondences may not be log-concave Tru20
2. Algebraically Stable Models DF01, FJ11, Bir25:
   • Birational surface maps always have algebraically stable models
   • However, some rational maps on $\mathbb{P}^2$ lack algebraically stable models Fav03, BDJ20
   • For surface correspondences, a general construction method is unknown
3. Known Computations of Correspondence Dynamical Degrees:
   • Monomial correspondences DR21
   • Hurwitz correspondences Ram20
   • This paper is the first complete computation of the dynamical degree for a non-trivial algebraic billiard

Advantages of This Work

1. Methodological Innovation: Non-standard blowup strategy, designing iterated blowups based on local formulas
2. Completeness of Results: Exact computation for specific billiards combined with results for general cases
3. Technical Breakthrough: First construction of an algebraically stable model for smooth algebraic billiards ($d>2$)
4. Broad Applicability: Proof of Ivrii's conjecture applies to non-convex algebraic regions

Conclusions and Discussion

Main Conclusions

1. Dynamical Degree: The dynamical degree of billiards in general degree $d$ algebraic curves is at least $2d^2 - 3d - 1$, exhibiting quadratic growth. This is an algebraic analogue of billiard chaos.
2. Ivrii's Conjecture: Proves that the set of periodic points of the classical billiard map has measure zero for real algebraic curves with algebraically independent coefficients, hence Weyl's conjecture holds for these regions.
3. Model Construction: For Fermat hyperbola billiards, constructs an algebraically stable model over the complex numbers (first example for $d>2$) and an essentially stable model over general fields.

Limitations

1. Specialization Gap:
   • Conjecture 1.3: The exact dynamical degree of general billiards should be $\rho_d$ (the largest root of a certain cubic polynomial)
   • Current lower and upper bounds still have a small gap
   • Excluding other destabilizing orbits is difficult due to exponential growth of correspondence orbits
2. Restrictions on Algebraic Stability:
   • Essential stability suffices to compute the dynamical degree but is weaker than algebraic stability
   • Only essential stability can be proven in positive characteristic
   • Construction of algebraically stable models for general curves remains an open problem
3. Scope of Ivrii's Conjecture:
   • The proof requires the strong assumption of algebraically independent coefficients
   • For specific non-generic curves, the conclusion may not hold
   • Periods 5 and higher remain open in the classical setting
4. Technical Limitations:
   • The dynamical degree of correspondences lacks semicontinuity (unlike for maps)
   • Computability is unknown
   • Case-by-case analysis is required

Future Directions

1. Prove Conjecture 1.3: Determine the exact dynamical degree of general billiards $\lambda_1(b_{gen}) = \rho_d$
2. Generalization to Other Curves:
   • Construct algebraically stable models for other special curves (e.g., general conics)
   • Study more curve families with exceptional properties
3. General Theory of Correspondences:
   • Develop general construction methods for improved models of surface correspondences
   • Study computability and semicontinuity of dynamical degrees
4. Extension of Ivrii's Conjecture:
   • Prove complex Ivrii's conjecture for more specific billiards
   • Establish real Ivrii's conjecture for larger families of algebraic curves
5. Connections with Physics:
   • Study spectral properties of quantum billiards
   • Verify higher-order terms of Weyl's conjecture

In-Depth Evaluation

Strengths

1. Methodological Innovation:
   • The non-standard blowup strategy is original, avoiding the difficulties of traditional methods where repeated blowups may introduce new problems
   • The introduction of the essential stability concept cleverly circumvents the difficulty that complete algebraic stability is hard to achieve over general fields
   • The dynamical analysis of midpoint divisors reduces high-dimensional problems to 1-dimensional ones, demonstrating profound geometric insight
2. Technical Depth:
   • The derivation of local formulas (Proposition 3.11) requires delicate formal power series calculations
   • The construction of iterated blowups is based on deep understanding of Puiseux series balls
   • Matrix calculations (Proposition 6.4) are exact and verifiable
   • The discovery of the invariant subset $U$ (Lemma 7.4) exemplifies the combination of complex analysis and algebraic geometry
3. Completeness of Results:
   • Combines exact computations for special cases with qualitative results for general cases
   • Covers all degrees $d \geq 2$ and fields of different characteristics
   • The proof of Ivrii's conjecture extends from the complex case to the real case with complete logical chain
4. Clarity of Exposition:
   • Clear structure: preliminaries → special curves → model construction → computations → applications
   • Rigorous definitions with clear hierarchy of lemmas, propositions, and theorems
   • Provides intuitive geometric diagrams (Figures 1-6)
   • Systematic introduction to formal correspondence theory (Section 2.2) fills a gap in the literature
5. Academic Value:
   • First complete computation of the dynamical degree for a non-trivial algebraic billiard
   • First construction of an algebraically stable model for smooth algebraic billiards with $d > 2$
   • Substantial progress in proving Ivrii's conjecture in the non-convex case
   • Provides important examples for correspondence dynamics

Weaknesses

1. Specificity of Method:
   • The exceptional property of the Fermat hyperbola (Lemma 3.8) is crucial but very special
   • Unclear how many other curve families possess similar properties
   • Limited generalizability of the method
2. Gaps in Results:
   • Gaps remain between upper and lower bounds on the dynamical degree (though small)
   • Conjecture 1.3 remains unproven, indicating that some destabilizing orbits are not yet understood
   • The even degree case requires additional satellite blowups, less elegant than the odd case
3. Technical Complexity:
   • The construction of iterated blowups is quite technical, requiring extensive symbolic and local coordinate calculations
   • The double cover technique in Section 7 (even degree case) increases difficulty of understanding
   • Formal correspondence theory (Section 2.2), while rigorous, may be overly abstract for non-specialists
4. Limited Scope of Applications:
   • How to verify algebraic independence of coefficients in practice for Ivrii's conjecture?
   • The specific example given (end of Section 7.3) is correct but somewhat artificial
   • For low degrees ($d=2,3$), results are complete but not particularly surprising
5. Connection with Physics:
   • While Weyl's conjecture is mentioned, physical implications are not deeply discussed
   • What does quadratic growth of dynamical degree mean physically?
   • Lacks numerical simulations or visualizations of billiard trajectory complexity

Impact

1. Contribution to the Field:
   • Billiard Theory: Provides quantitative characterization of chaos in algebraic billiards, complementing classical billiard theory
   • Algebraic Dynamics: Enriches examples of correspondence dynamical degrees and demonstrates new computational techniques
   • Algebraic Geometry: The iterated blowup construction method may inspire other applications
   • Spectral Theory: Provides evidence for Weyl's conjecture on new classes of regions
2. Practical Value:
   • Strong theoretical focus with limited immediate applications
   • May have long-term impact on understanding quantum billiards and wave equations
   • Computational methods (essential stability + linear algebra) may apply to other correspondences
3. Reproducibility:
   • All proofs are constructive and verifiable in principle
   • Matrix calculations can be verified with computer algebra systems (e.g., SageMath)
   • Lack of code or computational notebooks is a minor shortcoming
4. Inspirational Value:
   • Specialization strategy: studying general objects through carefully chosen special examples
   • Importance of exceptional properties: special geometric properties can greatly simplify dynamical analysis
   • Essential stability concept may be useful in other contexts where complete stability is unattainable

Applicable Scenarios

1. Theoretical Research:
   • Researchers in algebraic dynamical systems studying dynamical degrees of correspondences
   • Billiard theorists studying chaos and periodic orbits
   • Algebraic geometers studying birational geometry and blowup techniques
2. Related Problems:
   • Computing dynamical degrees for other algebraic dynamical systems (e.g., Hénon maps, rational maps)
   • Classification of completely integrable systems
   • Inverse problems in spectral theory
3. Directions for Generalization:
   • Higher-dimensional billiards (billiards in surfaces)
   • Billiards under other group actions
   • Billiard analogues in noncommutative geometry

References

The paper cites 43 references spanning billiard theory, algebraic dynamics, and algebraic geometry. Key references include:

1. Gut12 Gutkin - Billiard dynamics: survey of billiard theory
2. Glu14a, Glu14b, Glu21 Glutsyuk - foundational work on complex algebraic billiards
3. Wei25 Author's previous work - upper bounds on dynamical degree of general billiards
4. DF01 Diller-Favre - algebraic stability for birational surface maps
5. Tru20 Truong - relative dynamical degree for correspondences
6. Ivr80 Ivrii - statement of Ivrii's conjecture
7. BDMLD+24 Bessa et al. - positive topological entropy for generic convex billiards
8. FJ04, FJ07 Favre-Jonsson - valuation tree theory, inspiring blowup constructions

Overall Assessment: This is a high-quality pure mathematics paper making substantial progress in algebraic billiard theory. By carefully selecting the Fermat hyperbola as a special object, the author successfully proves two important properties of general algebraic curve billiards: quadratic growth of dynamical degree and Ivrii's conjecture. Technically, the introduction of non-standard blowup strategies and the essential stability concept demonstrate originality. While the specificity of the method limits direct generalization, it provides important examples for correspondence dynamics and opens new research directions in billiard theory. The paper is rigorously and clearly written, representing a significant contribution to algebraic dynamical systems.