Random conic bundle surfaces satisfy the Hasse principle

Abstract: We establish the Hasse principle for $100\%$ of conic bundles over $\mathbb{P}<sup>1_{\mathbb{Q}}$.

• The paper establishes that nearly 100% of conic bundle surfaces satisfy the Hasse principle when organized by the heights of their coefficients.
• It introduces a novel Hilbert symbol detector and refined Fourier analysis to control error terms and minor arc contributions in the proof.
• Quantitative results reveal that the failure probability decays superlogarithmically, dramatically surpassing previous positive proportion benchmarks.

Random Conic Bundle Surfaces and the Typical Validity of the Hasse Principle

Introduction and Context

The Hasse principle governs the solvability of Diophantine equations over global fields: a variety $X$ defined over a number field $k$ satisfies the Hasse principle if the existence of local points in all completions of $k$ guarantees the existence of a $k$-rational point. Failures of the Hasse principle reflect subtle arithmetic phenomena, often encapsulated by the Brauer–Manin obstruction for rationally connected varieties. Explicit counterexamples are rare and typically rely on fine-tuned parameter choices or geometric configurations.

Conic bundle surfaces, i.e., smooth projective surfaces over $\mathbb{Q}$ admitting $\pi: X \to \mathbb{P}^1$ with fibers conics, are foundational in the study of the Hasse principle and the Brauer–Manin obstruction. Canonically described by $f_1(t)x^2 + f_2(t)y^2 = f_3(t)z^2$ with separable $f_i \in \mathbb{Z}[t]$, they include special subcases such as degree 4 del Pezzo surfaces and Châtelet surfaces. While much is known in particular cases—often via descent and explicit arithmetic—comprehensive probabilistic or statistical results for general conic bundle surfaces have remained incomplete.

Main Results

This paper establishes a strong statistical version of the Hasse principle for conic bundle surfaces:

Primary Theorem: For any fixed bounding degrees on $f_i$, when conic bundle surfaces are ordered by the heights of the coefficients of the $f_i$, 100% of such surfaces defined over $k$0 satisfy the Hasse principle [(2604.07047), Theorem 1.1 & 1.2].

In particular, for $k$1 of prescribed degrees, almost all surfaces in the family

$k$2

have a rational point as soon as they are everywhere locally soluble. This assertion remains true even when allowing arbitrary prescribed factorization patterns of degree-bounded $k$3, with only a vanishing proportion of counterexamples.

Quantitative refinement: For all large $k$4, the proportion of conic bundle surfaces of height at most $k$5 that satisfy the Hasse principle tends to $k$6 for any $k$7.

These are unconditional, not relying on Schinzel's hypothesis or related unproved assumptions, and markedly stronger than previous results which established only a positive proportion of Hasse-principle-satisfying conic bundles [MR452704].

Technical Innovations

The analytic backbone of the argument combines several significant advances:

• Summability Kernels and Fourier-Analytic Decomposition:

Heat kernel weights are introduced on the space of polynomial coefficients, yielding a Fourier-analytic model where the Jacobi theta function’s transformation law produces super-exponential decay off the major arcs. This enables control of minor arc contributions in the associated circle method averages for general arithmetic functions $k$8, requiring only equidistribution in arithmetic progressions to small moduli.

• Hilbert Symbol Detector with Zero Average:

The existence of rational points on conics fibered over $k$9 is detected using an analytic version of the Hilbert symbol that is constructed to have mean zero over local fields off a thin exceptional set. This leads to crucial cancellation in character sums and dispersion estimates, reducing the level of distribution required for the underlying arithmetic functions. The construction refines previous approaches reliant on combinatorial decompositions and implements a precise random/deterministic split.

• Second Moment Estimates and Level Lowering:

The analysis yields nontrivial $k$0 mean bounds for sums of arithmetic functions over values of random polynomials via explicit major/minor arc decomposition, Diophantine lattice point counting, and a new level lowering mechanism in the dispersion argument tied directly to the Hilbert symbol’s average properties.

• Reduction to Local Densities and Probabilistic Model:

For everywhere locally soluble bundles, explicit lower bounds are obtained for the number of rational (not just local) fibers, refinable to an asymptotic formula for the count of fibers with rational points in terms of local densities, with uniform error bounds.

Relation to Prior Work

Previous positive proportion results for conic bundles were limited either to special factorization types or relied (at least conditionally) on strong conjectures such as Schinzel's hypothesis [MR667708, MR3292295, MR3194818]. The positive proportion claimed in [MR452704] for general bundles is strictly superseded by the "typical = 100%" result here.

Counterexamples to the Hasse principle are known for particular degenerate configurations, e.g., when quartic $k$1 splits as a product of two irreducibles [MR286743], or among pairs of quadratic polynomials for Châtelet surfaces [MR3976470, MR3198755]. However, the density of such counterexamples is shown in this paper to be negligible, vanishing compared to the full parameter space.

This work also places conic bundles within the context of recent statistical advancements for other families—random Fano hypersurfaces [MR4564262], norm varieties (Diao, 22 Jun 2025), random Diophantine equations [MR3177289], and more—further cementing a formal probabilistic description of local-global principles as the generic expectation in arithmetic geometry.

Numerical Strength and Contradictory Claims

Contradictory to the failure loci:

The core numerical claim, which is bold relative to prior knowledge, is that the probability of Hasse principle failure among conic bundle surfaces (even when allowing for all possible prescribed factorizations and arbitrary degrees) is $k$2 in the limit. For any positive $k$3, the fraction of surfaces violating the Hasse principle does not just fail to be dense or Zariski-dense, but is quantitatively controlled to decay considerably faster than any negative power of $k$4 in terms of the height parameter $k$5.

Implications and Outlook

Theoretical Implications

This result substantiates, in the full measure-theoretic sense, the expectation that the Brauer–Manin obstruction, as postulated in Colliot-Thélène's conjecture, captures all failures of the Hasse principle for random conic bundles under very mild constraints. From the perspective of arithmetic statistics, it provides a classification template: for large families of rationally connected surfaces, local-global principles are not only "often" valid but are overwhelmingly so.

The machinery developed—in particular the analytic Hilbert symbol with controlled averages, the random/deterministic decomposition, and the technique of controlling averages via carefully weighted Fourier analysis—extend directly to several other families of arithmetic significance, including norm form varieties, complete intersections, and possibly beyond.

Practical and Future Directions

While conic bundle surfaces are not of direct algorithmic relevance for mainstream AI, the general approach provides tools for quantifying randomness and typicality in high-dimensional parameter spaces subject to arithmetic constraints. This is of interest wherever one needs rigorous guarantees on the density of counterexamples to certain arithmetic or geometric properties—key, for instance, in cryptographic protocol design, probabilistic model selection, or rigorous sampling over solution spaces.

From an algorithmic and computational number theory viewpoint, the explicit detector construction for rational points could eventually inform efficient heuristics and probabilistic verification procedures in practical instances. Moreover, the Fourier-analytic summability kernel technique may find uses in other analytical statistics or signal-processing contexts in AI where minor arc analysis and control of low-probability exceptional sets are necessary.

The logical next steps involve extending such 100% Hasse principle results to higher-dimensional or more general classes of varieties (e.g., arbitrary rationally connected varieties, higher-degree norm forms), possibly benefiting from further harmonic or probabilistic refinements.

Conclusion

This work establishes that the Hasse principle holds for $k$6 of conic bundle surfaces over $k$7 when ordered by the maximal coefficient, even allowing for general prescribed factorization patterns and arbitrary degrees. The statistical methodology—anchored in Fourier analysis, analytic number theory, and a tailored random/deterministic split—refines the arithmetic understanding of local-global principles and sets a structural prototype for future work in arithmetic statistics of algebraic varieties (2604.07047).