Counting rational points on smooth quartic and quintic surfaces

Basic Information

• Paper ID: 2511.07060
• Title: Counting rational points on smooth quartic and quintic surfaces
• Author: Lorenzo Andreaus
• Classification: math.NT (Number Theory), math.AG (Algebraic Geometry)
• Publication Date: November 10, 2025 (arXiv preprint)
• Author Affiliation: Institut de Mathématiques de Jussieu - Paris Rive Gauche (IMJ-PRG), Université Paris Diderot
• Paper Link: https://arxiv.org/abs/2511.07060v1

Abstract

This paper studies the problem of counting rational points on smooth projective surfaces $X$ of degree $d \geq 4$ defined over a number field $K$. Let $N_{X'}(B)$ denote the number of rational points on $X$ with height at most $B$ that do not lie on any lines contained in $X$. Under appropriate assumptions on the rank of abelian varieties, the author proves that for any fixed $\varepsilon > 0$, $N_{X'}(B) \ll_X B^{4/3+\varepsilon}$

This improves the unconditional uniform bound of Salberger for $d=4$ and $d=5$. The proof is based on Heath-Brown's argument, cutting $X$ by projective planes and using the uniform Faltings theorem of Dimitrov, Gao, and Habegger to bound the number of rational points on plane sections of $X$.

Research Background and Motivation

Core Problem

This paper addresses a classical problem in algebraic geometry and number theory: counting rational points of bounded height. Specifically, for a smooth hypersurface $X \subseteq \mathbb{P}^n_K$ defined over a number field $K$, how can one bound the number $N_X(B)$ of rational points with height at most $B$?

Importance of the Problem

1. Central problem in Diophantine geometry: The distribution of rational points is a fundamental object of study in modern number theory, closely related to Diophantine equations and arithmetic geometry
2. Theoretical significance: This problem involves several deep conjectures (such as the Birch and Swinnerton-Dyer conjecture and the generalized Riemann hypothesis)
3. Intersection of geometry and arithmetic: Controlling arithmetic properties (distribution of rational points) through geometric properties (such as Kodaira dimension)

Limitations of Existing Results

1. Higher-dimensional cases: Verzobio (2019) proved that for $n \geq 4$ and $d \geq 50$, $N_X(B) \ll_{n,d,\varepsilon} B^{n-2+\varepsilon}$, but this does not apply to small degrees
2. Special nature of surfaces: When $n=3$ (surfaces), if $X$ contains $K$-rational lines, then $N_X(B) \gg B^2$, so one must exclude points on lines
3. Salberger's bound: Salberger (2023) proved $N_{X'}(B) \ll_d B^{3/\sqrt{d}}(\log B)^4 + B$, but the exponent is still large for $d=4,5$ (namely $B^{1.5}$ and $B^{1.34}$)

Motivation for This Paper

1. Improving bounds for specific degrees: For quartic and quintic surfaces ($d=4,5$), reducing the exponent from $3/\sqrt{d}$ to $4/3$
2. Generalizing Heath-Brown's method: Heath-Brown (1998) proved $N_{X'}(B) \ll_{X,\varepsilon} B^{4/3+\varepsilon}$ for cubic surfaces (under rank assumptions); this paper extends this to higher degrees
3. Unified results over arbitrary number fields: Salberger's results primarily apply to $\mathbb{Q}$; this paper establishes uniform bounds over arbitrary number fields $K$

Core Contributions

1. Main Theorem (Theorem 1.3): Under Hypothesis 1.2 (rank hypothesis), for smooth surfaces $X \subseteq \mathbb{P}^3_K$ of degree $d \geq 4$, $N_{X'}(B) \ll_{X,K,\varepsilon} B^{4/3+\varepsilon}$ This improves Salberger's bound for $d=4,5$
2. Generalization to number fields: The result holds over arbitrary number fields $K$, using Paredes-Sasyk's generalization of Salberger's bound
3. Technical contributions:
   • Systematic application of the uniform Faltings theorem (Dimitrov-Gao-Habegger)
   • Combining Grassmannian height theory with plane-cutting techniques
   • Complete treatment of contributions from curves of different genera
4. Geometric classification: Proves that smooth surfaces of degree $d \geq 4$ contain only finitely many rational plane curves of fixed degree (Lemma 2.2), using Kodaira dimension theory

Detailed Methodology

Problem Setup

• Input: Smooth projective surface $X \subseteq \mathbb{P}^3_K$ (degree $d \geq 4$), height bound $B > 0$
• Output: Upper bound on the count of rational points $N_{X'}(B)$
• Constraints:
  • $X'$ is the set obtained by removing all lines contained in $X$
  • Hypothesis 1.2 (rank hypothesis) is assumed
  • Height function $H_K$ defined as $H_K([x_0:\cdots:x_n]) = \prod_v \max\{\|x_i\|_v\}$

Overall Architecture

The proof strategy consists of three main steps:

Step 1: Plane Coverage

Using Schmidt's (1967) result, for each $x \in X(K)$ with $H_K(x) \leq B$, there exists a projective plane $\Pi \cong \mathbb{P}^2$ containing $x$ such that $H_K(\Pi) \ll_K H_K(x)^{1/3} \ll_K B^{1/3}$

Here $H_K(\Pi)$ is the height in the Grassmannian $\text{Gr}_K(3,4) \cong (\mathbb{P}^3)^\vee$, defined as: if $\Pi$ is spanned by basis $w_1, w_2, w_3$, then $H_K(\Pi) := H_K(w_1 \wedge w_2 \wedge w_3) \in \bigwedge^3 V$

Step 2: Plane Counting

By Schmidt's result or the more precise version of Thunder (1992), the number of planes with height at most $O_K(B^{1/3})$ is $\#\{\Pi : H_K(\Pi) \ll B^{1/3}\} = O_K(B^{4/3})$

Step 3: Counting Points on Plane Sections

For each plane $\Pi$, one must bound $\#(X \cap \Pi)(K)$. The key observation is:

• $\Pi \not\subseteq X$ (otherwise $X$ would be reducible, a contradiction)
• Therefore $X \cap \Pi$ is a plane curve of degree $d$

Classification by Genus

Let $C$ be an irreducible component of $X \cap \Pi$, of degree $e \leq d$, and let $\tilde{C}$ be its normalization. By the genus formula: $g(C) = \frac{(e-1)(e-2)}{2} - \sum_p \frac{m_p(m_p-1)}{2}$ where $m_p$ is the multiplicity of $C$ at singular point $p$ (including infinitely near singular points).

Case 1: $g(C) \geq 2$ (High Genus)

Key tool: Uniform Faltings theorem (Lemma 1.1, Dimitrov-Gao-Habegger 2021) $\#\tilde{C}(K) \leq c(g)^{1+\rho(\tilde{C})}$ where $\rho(\tilde{C})$ is the rank of the Jacobian $J(\tilde{C})$.

Bounding the rank:

• Since $\tilde{C}$ comes from $X \cap \Pi$ with $H_K(\Pi) \ll B^{1/3}$, the Faltings height of its Jacobian satisfies $h_F(J) \ll_X \log B$
• By $\log N_J \ll h_F(J)$ (conductor bound), combined with the rank hypothesis: $\rho(\tilde{C}) = r_J = o(\log N_J) = o(\log B)$
• Therefore there exists $B_0(\varepsilon)$ such that for $B \geq B_0(\varepsilon)$, $\rho(\tilde{C}) \leq \varepsilon \log B$

Point count estimate: $\#\tilde{C}(K) \leq c(g)^{1+\varepsilon \log B} = B^{\varepsilon c_1 \log c(g)} \ll_{X,K,\varepsilon} B^\varepsilon$ Since $g \leq \frac{(d-1)(d-2)}{2}$ is bounded, adjusting constants yields a uniform bound.

Total contribution: $O_{X,K,\varepsilon}(B^{4/3} \cdot B^\varepsilon) = O_{X,K,\varepsilon}(B^{4/3+\varepsilon})$

Case 2: $g(C) = 1$ (Elliptic Curves)

If $\tilde{C}(K) = \emptyset$, there is no contribution; otherwise $\tilde{C}$ is an elliptic curve.

Key tool: Uniform bound of Heath-Brown-Testa (2010) (Lemma 2.1) $N_C(B) \ll_K (\log B)^{1+\rho(\tilde{C})/2}$

Combined with the rank hypothesis: $N_C(B) \ll_{X,K,\varepsilon} (\log B)^{1+\varepsilon \log B/2} \ll B^\varepsilon$

Total contribution: Similarly $O_{X,K,\varepsilon}(B^{4/3+\varepsilon})$

Case 3: $g(C) = 0$ (Rational Curves)

If $\tilde{C}(K) = \emptyset$, there is no contribution; otherwise $C$ is a rational curve of degree $e$.

Degree classification:

• $e = 1$: $C$ is a line, already excluded from $X'$
• $e \geq 2$: By Paredes-Sasyk (2022) (generalizing Salberger): $N_C(B) \ll_{X,\varepsilon} B^{2/e+\varepsilon} \leq B^{1+\varepsilon}$

Finiteness: The key Lemma 2.2 proves that $X$ contains only finitely many rational plane curves of fixed degree $e$.

Proof sketch (by contradiction):

1. Assume there exist infinitely many rational plane curves of degree $e$
2. Consider the Hilbert scheme $\text{Hilb}_X^P$, where $P(m) = me + 1 - \frac{(e-1)(e-2)}{2}$
3. The set of rational curves $\text{Rat}_X$ is a quasi-projective scheme with finitely many irreducible components
4. Some irreducible component $T$ must contain infinitely many points, corresponding to a universal family $U \subseteq X \times T$
5. The projection $\pi: U \to X$ must be dominant (otherwise the image lies in a curve, contradiction)
6. Therefore $X$ is covered by a family of rational curves, i.e., $X$ is unirational with Kodaira dimension $-\infty$
7. But smooth surfaces of degree $d \geq 4$ are K3 surfaces ($d=4$, Kodaira dimension 0) or of general type ($d \geq 5$, Kodaira dimension 2), contradiction!

Total contribution: Finitely many curves $\times$ each $O(B^{1+\varepsilon})$ = $O_{X,K,\varepsilon}(B^{1+\varepsilon}) \subset O(B^{4/3+\varepsilon})$

Technical Innovations

1. Systematic application of Grassmannian heights: Using the exterior algebra structure to define subspace heights, precisely controlling the height distribution of planes
2. Key use of uniform bounds:
   • The uniform Faltings theorem avoids analyzing each curve individually
   • The rank hypothesis connects $\rho$ to the conductor $N$, which can be controlled via Faltings height
3. Geometric-arithmetic correspondence: Using Kodaira dimension to exclude pathological geometric cases, demonstrating the power of algebraic geometry in number theory
4. Technical handling of number field generalization: Using Paredes-Sasyk's generalization and Merel's torsion bound (replacing Mazur's theorem)

Rank Hypothesis (Hypothesis 1.2)

Statement of Hypothesis

For any $K$-abelian variety $A$, let $N_A$ be the conductor (the ideal norm of the conductor ideal) and $r_A$ be the rank, then $r_A = o(\log N_A) \quad \text{as} \, N_A \to \infty$

Justification

Based on the following chain of conjectures (assuming $A$ is modular):

1. Generalized Riemann Hypothesis (GRH): The non-trivial zeros of $L(A,s)$ all lie on $\text{Re}(s) = 1$
2. Functional equation: Define $\Lambda(A,s) := N_A^{s/2} ((2\pi)^{-s} \Gamma(s))^{\dim A} L(A,s)$ satisfying $\Lambda(A, 2-s) = \pm \Lambda(A,s)$
3. Zero order estimate: By GRH and classical analytic arguments: $\text{ord}_{s=1}(L(A,s)) = O\left(\frac{\log N_A}{\log \log N_A}\right) = o(\log N_A)$
4. BSD Conjecture: $\text{ord}_{s=1}(L(A,s)) = r_A$

Therefore, under these conjectures, the rank hypothesis holds.

Known Results

• Unconditional bound (Ooe-Top 1989): $r_A = O_{K,\dim A}(N_A)$ (linear growth)
• Special case of elliptic curves: If $E/\mathbb{Q}$ has a 2-torsion point, then $r_E = O\left(\frac{\log N_E}{\log \log N_E}\right)$ (Heath-Brown, known result)

Strength of the Hypothesis

If one assumes that all abelian varieties have rank bounded by a constant depending only on dimension and the number field, then one can remove $\varepsilon$, obtaining $N_{X'}(B) \ll_{X,K} B^{4/3}$

Experimental Setup

Note: This is a pure mathematics theory paper requiring no experimental verification. The following describes theoretical verification:

Theoretical Verification Framework

1. Verification on known special cases:
   • Cubic surfaces ($d=3$): Heath-Brown's original result
   • Asymptotic behavior for large degree ($d \to \infty$): Consistent with Verzobio's result
2. Boundary case checks:
   • $d=4$ (K3 surfaces): Kodaira dimension 0
   • $d=5$ (general type): Kodaira dimension 2
   • Confirming that the geometric argument of Lemma 2.2 holds in these cases
3. Constant dependence:
   • Dependence on $\varepsilon$: Realized through the $o(\cdot)$ notation in the rank hypothesis
   • Dependence on $X$: Manifested through Faltings height $h_F(J) \ll_X \log B$
   • Dependence on $K$: Through $[K:\mathbb{Q}]$ and Merel's bound

Analysis of Main Results

Quantitative Improvements

Comparing with Salberger's bound $N_{X'}(B) \ll_d B^{3/\sqrt{d}}(\log B)^4 + B$:

Conclusion: This paper's method is superior to Salberger for $d=4,5$, but Salberger's bound is better for $d \geq 6$.

Trade-off between Conditionality and Uniformity

Theoretical Significance

1. Scope of the method: Shows that Heath-Brown's plane-cutting technique can be generalized to higher degrees, but is limited by genus growth
2. Optimality question: The exponent $4/3$ is optimal for cubic surfaces (Heath-Brown); whether it is optimal for quartic and quintic surfaces remains unclear
3. Necessity of rank hypothesis: Without the rank hypothesis, the uniform Faltings theorem only gives $\#C(K) \leq c(g)^{1+r_J}$, making it impossible to effectively control $r_J$ when it could be large

Related Work

History of Rational Point Counting

1. Classical results:
   • Faltings (1983): Curves of genus $g \geq 2$ have only finitely many rational points
   • Vojta (1987): Rational points on varieties of general type are not dense
2. Development of quantitative bounds:
   • Heath-Brown (1998): Cubic surfaces $N_{X'}(B) \ll_X B^{4/3+\varepsilon}$ (under rank hypothesis)
   • Salberger (2023): Uniform bound for general degrees $N_{X'}(B) \ll_d B^{3/\sqrt{d}}(\log B)^4$
   • Verzobio (2025): Higher dimensions and high degrees $N_X(B) \ll B^{n-2+\varepsilon}$ ($d \geq 50$)

Progress on Uniform Bounds

3. Uniform Faltings theorem:
   • Dimitrov-Gao-Habegger (2021): $\#C(K) \leq c(g)^{1+\rho(C)}$
   • Kühne (2021): Removing dependence on $[K:\mathbb{Q}]$
4. Uniform bounds for elliptic curves:
   • Heath-Brown-Testa (2010): Smooth cubic curves $N_C(B) \ll (\log B)^{1+\rho/2}$

Related Work on Rank Hypothesis

5. Conditional bounds on rank:
   • Ooe-Top (1989): $r_A = O(N_A)$ (unconditional)
   • Heath-Brown: Elliptic curves with 2-torsion $r_E = O(\log N_E / \log \log N_E)$
6. Average bounds:
   • Families of elliptic curves ordered by height have bounded average rank (multiple authors)

Positioning of This Paper

This paper is at the intersection of:

• Geometric method (plane-cutting) + Arithmetic tools (uniform Faltings)
• Conditional results (rank hypothesis) + Quantitative improvements (specific exponents)
• Specific degrees ($d=4,5$) + Arbitrary number fields (generalization of $K$)

Main distinctions from related work:

• Compared to Heath-Brown: Generalization to higher degrees
• Compared to Salberger: Improved exponents for specific degrees (but conditional)
• Compared to Verzobio: Treatment of smaller degrees with better bounds

Conclusions and Discussion

Main Conclusions

1. Theorem statement: Under the rank hypothesis, smooth surfaces $X \subseteq \mathbb{P}^3_K$ of degree $d \geq 4$ satisfy $N_{X'}(B) \ll_{X,K,\varepsilon} B^{4/3+\varepsilon}$
2. Range of improvement: For $d=4$ and $d=5$, this improves Salberger's unconditional uniform bound
3. Methodological contribution: Demonstrates that combining Heath-Brown's technique with modern uniform bound tools can be generalized to higher degrees

Limitations

Theoretical Limitations

1. Conditionality:
   • Depends on the rank hypothesis (Hypothesis 1.2), which itself depends on GRH and the BSD conjecture
   • Unconditional results remain Salberger's bounds
2. Non-uniformity:
   • Implicit constants depend on $X$ (via Faltings height)
   • Cannot provide bounds uniform in $d$
3. Degree restrictions:
   • For $d \geq 6$, Salberger's bound $B^{3/\sqrt{d}}$ is superior
   • The method's effectiveness decreases as $d$ increases (due to genus growth of plane sections)

Technical Limitations

4. Genus growth problem:
   • Plane curves of degree $d$ can have genus up to $\frac{(d-1)(d-2)}{2}$
   • The constant $c(g)$ grows with $g$ (though absorbed in $B^\varepsilon$)
5. Precision of rank control:
   • The rank hypothesis gives $r = o(\log N)$, but does not specify the rate of $o(\cdot)$
   • More precise rank bounds (such as $r = O(\log N / \log \log N)$) could remove $\varepsilon$
6. Treatment of rational curves:
   • The proof of Lemma 2.2 relies on Kodaira dimension theory
   • May not generalize to more general varieties (non-hypersurfaces)

Future Directions

Theoretical Generalizations

1. Removing the rank hypothesis:
   • Seeking unconditional proofs or weaker assumptions
   • Using average rank bounds or unconditional bounds for special cases
2. Generalization to higher dimensions:
   • Similar bounds for $X \subseteq \mathbb{P}^n$ ($n \geq 4$)
   • Requires generalizing Grassmannian cutting techniques
3. Non-hypersurface cases:
   • Complete intersections, Fano varieties, and more general varieties
   • May require different geometric arguments

Technical Improvements

4. Optimizing the exponent:
   • Determining whether $4/3$ is optimal (for $d=4,5$)
   • Studying the relationship between exponent and Kodaira dimension
5. Improving constant dependence:
   • Making the implicit constant's dependence on $X$ more explicit
   • May require more refined height theory
6. Hybrid strategies:
   • Combining Salberger's and this paper's methods
   • Using different techniques for different degree ranges

Computational Aspects

7. Effectiveness:
   • Making implicit constants computable
   • Providing explicit bounds for specific surfaces (such as Fermat surfaces)
8. Numerical verification:
   • Computer verification for small degrees and small heights
   • Testing the rank hypothesis on concrete examples

In-Depth Evaluation

Strengths

Mathematical Depth

1. Interdisciplinary synthesis:
   • Skillfully combines algebraic geometry (Kodaira dimension, Hilbert schemes), number theory (height theory, rank hypothesis), and analytic number theory (L-functions)
   • Demonstrates the multidisciplinary nature of modern arithmetic geometry
2. Technical innovation:
   • Non-trivial generalization of Heath-Brown's method to higher degrees
   • Systematic use of uniform bounds (uniform Faltings, uniform Heath-Brown-Testa)
   • Elegant and powerful geometric argument in Lemma 2.2
3. Theoretical contribution:
   • Improves known best results on a concrete problem
   • Provides new perspective on understanding rational point distribution

Proof Quality

4. Clear structure:
   • Classification by genus, logically rigorous
   • Each estimate has clear source
5. Complete details:
   • Handles all possible cases (different genera, singularities, number field generalization)
   • Thorough analysis of constant dependence
6. Literature usage:
   • Cites recent results (Dimitrov-Gao-Habegger 2021, Verzobio 2025)
   • Correctly identifies and uses key tools

Weaknesses

Method Limitations

1. Conditionality issue:
   • The rank hypothesis depends on deep conjectures (GRH + BSD), far from unconditional proof
   • Limits practical applicability of the result
2. Narrow improvement range:
   • Only improves Salberger for $d=4,5$
   • For $d \geq 6$, this paper's method is inferior to existing results
3. Non-uniformity:
   • Constants depend on $X$, cannot provide uniform bounds in $d$
   • Contrasts with Salberger's advantage of uniformity in $d$

Technical Details

4. Use of rank hypothesis:
   • The form $r_A = o(\log N_A)$ is weak, does not specify the rate
   • Stronger assumptions needed to remove $\varepsilon$ in practical applications
5. Hidden constants:
   • Multiple uses of $\ll_{X,K,\varepsilon}$, implicit constants are not computable
   • Limited value for practical applications (such as determining rational points on specific surfaces)
6. Coarseness of genus estimates:
   • Uses maximum possible genus $\frac{(d-1)(d-2)}{2}$
   • Actual plane sections may have smaller genus (e.g., when containing line components)

Presentation Issues

7. Notation consistency:
   • The definition of height $H_K$ in the paper differs slightly from some references
   • While claimed to be inconsequential, may cause confusion
8. Brevity of geometric argument:
   • The proof of Lemma 2.2 relies on deep algebraic geometry (Hilbert schemes, Kodaira dimension)
   • May be difficult for non-specialists to fully understand

Impact

Academic Impact

1. Theoretical value:
   • Provides best known conditional bounds for specific degrees
   • Demonstrates the power of uniform bound tools in rational point counting
2. Methodological contribution:
   • Provides template for generalization to other cases
   • Emphasizes the central role of rank hypothesis in modern number theory
3. Problem stimulation:
   • Raises natural questions (optimal exponent, generalization to higher dimensions)
   • May inspire subsequent research

Practical Value

4. Limited direct applications:
   • Conditional result, cannot directly determine rational points on specific surfaces
   • Primary value is theoretical understanding rather than computation
5. Insights for related problems:
   • Techniques may apply to related problems (integer point counting, rational curve counting)
   • The use of uniform bounds is instructive

Reproducibility

6. Theoretical verifiability:
   • Proof logic is clear, each step has literature support
   • Mathematical community can independently verify
7. Computational infeasibility:
   • Implicit constants are not computable
   • Numerical experiments cannot verify

Applicable Scenarios

This paper's methods and results apply to:

1. Theoretical research:
   • Studying asymptotic behavior of rational point distribution
   • Exploring applicability ranges of different techniques (geometric vs. analytic)
2. Development of conditional results:
   • Studying other arithmetic problems under rank hypothesis
   • Understanding relationships between conjectures (GRH, BSD, rank hypothesis)
3. Fine analysis of specific degrees:
   • For quartic and quintic surfaces, provides best known conditional bounds
   • May guide research on specific examples
4. Method generalization:
   • Starting point for generalization to other varieties
   • Application example of uniform bound techniques

Non-applicable scenarios:

• Applications requiring unconditional results
• Practical problems requiring computable constants
• General case of degree $d \geq 6$

Key References

1. Dimitrov, V., Gao, Z., Habegger, P. (2021). "Uniformity in Mordell–Lang for curves". Annals of Mathematics 194.1, pp. 237–298.
   • Uniform Faltings theorem, core tool of this paper
2. Heath-Brown, R. (1998). "Counting rational points on cubic surfaces". Astérisque 251, pp. 13–30.
   • Original method source, result for cubic surfaces
3. Salberger, P. (2023). "Counting rational points on projective varieties". Proc. London Math. Soc. 126.4, pp. 1092–1133.
   • Target bound improved by this paper
4. Schmidt, W. M. (1967). "On heights of algebraic subspaces and diophantine approximations". Annals of Mathematics 85.3, pp. 430–472.
   • Grassmannian height theory
5. Paredes, M., Sasyk, R. (2022). "Uniform bounds for the number of rational points on varieties over global fields". Algebra & Number Theory 16.8, pp. 1941–2000.
   • Generalization to number fields

Summary

This is a high-quality number theory paper that improves the rational point counting bound for smooth quartic and quintic surfaces under the rank hypothesis. Main highlights include technical synthesis (geometry, number theory, analysis) and method generalization (Heath-Brown to higher degrees). Main limitations are conditionality (depending on deep conjectures) and narrow improvement range (only $d=4,5$). The paper has significant theoretical value for understanding rational point distribution and developing uniform bound techniques, but limited practical value. Future research could focus on removing the rank hypothesis, generalizing to higher dimensions, or optimizing the exponent.