TL;DR
This paper investigates the distribution of preperiodic points and points with small canonical heights for quadratic polynomials and rational functions over the rational numbers. The authors provide computational evidence supporting conjectures about the uniform boundedness of preperiodic points and the lower bounds of canonical heights for non-preperiodic points, particularly for quadratic rational functions, a less-studied area.
What Problem Does This Paper Address?
The paper tackles questions at the intersection of number theory and dynamical systems, specifically concerning the behavior of iterated rational functions over the rational numbers. A central theme is the Dynamical Uniform Boundedness Conjecture, which posits that for any degree $d \ge 2$, there's a uniform upper bound on the number of rational preperiodic points for any rational function of degree $d$. A related conjecture by Silverman suggests that non-preperiodic rational points must have a canonical height that grows at least linearly with the height of the function itself. While these conjectures are well-posed, proving them, even for the simplest case of quadratic polynomials, has been challenging. The paper aims to provide computational support for these conjectures by exploring a vast parameter space of quadratic polynomials and rational functions, going beyond previous computational efforts.
Methods
The core of the paper's contribution lies in its extensive computational search. The authors employ a strategy of starting with a point $x$ and searching for maps $\phi$ that yield interesting orbits for $x$, rather than the reverse. This approach is more efficient for finding specific types of points.
For quadratic polynomials of the form $\phi_c(z) = z^2 + c$, they leverage properties of preperiodic points derived from Lemma 2.1 and 2.2, which constrain the relationship between $x$ and $c$. An algorithm (Algorithm 2.4) is developed to systematically search for pairs $(x, c)$ within specific ranges, focusing on those with small canonical height ratios $h_c(x)/h(c)$.
For quadratic rational functions, the approach is more complex. They utilize a parametrization of the moduli space of quadratic rational maps and their orbits (PM2). An algorithm (Algorithm 3.5) is designed to search for pairs $(x, \phi)$ where $x$ is a point at infinity and $\phi$ is a rational function, focusing on orbits with specific lengths and small height ratios $h(\phi(\infty))/h(\phi)$.
Datasets
The paper does not use pre-existing datasets for training or evaluation. Instead, it generates its own data through extensive computational searches. The primary "datasets" are the collections of pairs $(x, \phi)$ identified by their algorithms that satisfy certain criteria (e.g., small height ratio, specific preperiodic orbit structures). These findings are made available online at http://www3.amherst.edu/~rlbenedetto/quadratdata/.
Key Insights and Ideas
- Challenging Manes' Conjecture: The paper provides counterexamples to a conjecture by Manes regarding the bound on preperiodic points for quadratic rational functions, finding maps with 14 preperiodic points, exceeding her proposed bound of 12.
- New Preperiodic Orbit Structures: For rational functions, they discover new preperiodic orbit structures, including a rational 7-cycle and maps with 14 rational preperiodic points arising from 2-periodic cycles with a tail of length 6, and 3-periodic cycles with a tail of length 5.
- Strong Support for Conjectures: The extensive computations offer significant numerical evidence supporting Morton-Silverman's Uniform Boundedness Conjecture (Conjecture 1) and Silverman's conjecture on the lower bound of canonical heights (Conjecture 3), especially for quadratic rational functions.
- The "Record Holder" for Polynomials: The pair $(x, c) = (7/12, -181/144)$ for the quadratic polynomial $\phi_c(z) = z^2 + c$ continues to hold the record for the smallest canonical height ratio $h_c(x)/h(c) \approx 0.00660$, suggesting that finding significantly smaller ratios for polynomials is difficult.
- Structure of Rational Function Moduli Space: The paper connects the moduli space of certain quadratic rational maps with preperiodic orbits to elliptic surfaces, providing a deeper geometric understanding of these dynamical systems.
Key Results
- Counterexamples to Manes' Conjecture: The paper presents quadratic rational functions with 14 Q-rational preperiodic points, disproving Manes' conjecture that this number is at most 12.
- New Preperiodic Orbit Structures: Discovered maps with rational 7-cycles and maps with 14 preperiodic points arising from 2- and 3-periodic cycles.
- Computational Evidence for Conjectures: Extensive searches provide strong numerical support for Conjectures 1 (Uniform Boundedness) and 3 (Small Canonical Heights) for quadratic polynomials and rational functions.
- Conjecture 5 for Rational Functions: Based on their findings, the authors propose Conjecture 5, which provides specific bounds on the number of preperiodic points and the minimum canonical height for quadratic rational functions.
- Birational Equivalence to Elliptic Surface: Theorem 3.6 establishes a birational equivalence between a moduli space of quadratic rational maps with specific preperiodic orbit structures and an elliptic surface with positive rank over $Q(t)$.
Limitations and Future Work
- Search Range Limitations: While the search space for both polynomials and rational functions is significantly larger than previous studies, it is still finite. The authors acknowledge that they might have missed some points with very large denominators or height ratios outside their search parameters.
- Finiteness Questions: The paper notes that the finiteness of maps with 7-cycles or 3-periodic cycles with a tail of length 5 remains an open question, though they found infinitely many maps with a 2-periodic cycle having a tail of length 6.
- Conjecture 5: The paper proposes Conjecture 5 for rational functions, which is based on computational evidence and awaits formal proof.
- Geometric Understanding of Moduli Spaces: The paper hints at the interest in finding descriptions of other moduli surfaces corresponding to preperiodic orbits of different lengths (e.g., X6,1, X4,3, X3,4, X2,5) and determining if they contain infinitely many rational points.
- Understanding the "Record Holder": The continued dominance of the $(7/12, -181/144)$ pair for polynomials suggests further investigation into why such small height ratios are rare.
Summary updated