Random Diophantine equations of large degree
Tim Browning, Will Sawin · 2025
Read Random Diophantine equations of large degree by Tim Browning, Will Sawin (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2510-26191.txt.
Introduction
Tim Browning and Will Sawin study how often a random large Diophantine equation has a rational solution. A Diophantine equation asks for solutions in whole or rational numbers. They consider high surfaces, higher dimensional surfaces cut out by one uniform formula in powers whose weights are each plus or minus one. They ask for the chance that such a surface contains a rational point.
Browning and Sawin let both the degree of the formula in powers and the dimension of the surface tend to infinity, and study how the chance behaves in that limit. The problem sits at the meeting point of number theory and chance, since the equation is random but the being of a rational point is a rigid number based property. Their work addresses the large degree regime for these sign coefficient high surfaces.
Catalog record
| CREATOR | Tim Browning, Will Sawin |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 46 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2510-26191
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Random Diophantine equations of large degree Tim Browning, Will Sawin Abstract . Among the set of hypersurfaces of degree d d and dimension ℓ \ell defined by the vanishing of a homogeneous polynomial with coefficients in the set { − 1 , 1 } \{-1,1\} , we investigate the probability that a hypersurface contains a rational point as d d and ℓ \ell tend to ∞ \infty . 2010 Mathematics Subject Classification 11D45 (11G50, 11P21, 14G05) 1. Introduction Questions about the arithmetic properties of randomly chosen polynomials with integer coefficients are of central concern in number theory. In the simplest model, the large box model , one fixes the degree of the polynomial and allows the coefficients to vary uniformly in \{-H,\dots,H\} , as H\to\infty . In the large degree model , on the other hand, one constrains the coefficients to be in \{-1,+1\} , say, and allows the degree of the polynomial to grow. It has been known for a long time that random polynomials f\in\mathbb{Z}[x] are irreducible over \mathbb{Q} , when adhering to the large box model. The best known error term is due to Kuba [ 5 ] , who shows that the probability of f being reducible is O_{d}(H^{-1}) , for fixed degree d=\deg(f)\geqslant 3 . The large degree model is significantly harder to analyse, and we only have a conditional treatment. Thus, under the assumption of the Riemann Hypothesis for Dedekind zeta functions, the question of irreducibility in the large degree model has been dealt with by Breuillard and Varjú [ 2 ] . In this paper we shall be concerned with the Hasse principle for hypersurfaces V\subset\mathbb{P}^{n} defined by randomly chosen homogeneous polynomials f\in\mathbb{Z}[x_{0},\dots,x_{n}] of degree d . In the large box model, this topic was investigated by Poonen and Voloch [ 6 ] . In [ 6 , Conjecture 2.2.(ii)] it is asserted that in the Fano regime n\geqslant d , 100\% of the hypersurfaces should satisfy the Hasse principle. For the range n<d-1 , corresponding to hypersurfaces of general type, it is predicted in [ 6 , Conjecture 2.2.(i)] that 0\% of the hypersurfaces should have a rational point. The first of these conjectures is settled for hypersurfaces of dimension at least 3 in joint work of the authors with Le Boudec [ 3 ] . We shall address the existence of rational points in the large degree model. Let d,n\geqslant 2 and let N_{d,n}=\binom{n+d}{d} be the number of monomials of degree d in n+1 variables. Let \mathbb{B}_{d,n} be the set of hypersurfaces in \mathbb{P}^{n} defined by the vanishing of a homogeneous polynomial of degree d with coefficients in the set \{-1,1\} . We thus have (1.1) \#\mathbb{B}_{d,n}=2^{N_{d,n}-1}. We shall be interested in the ratio r_{d,n}=\frac{\#\{V\in\mathbb{B}_{d,n}:V(\mathbb{Q})\neq\emptyset\}}{\#\mathbb{B}_{d,n}}, which measures the density of hypersurfaces V\in\mathbb{B}_{d,n} with rational points. The following result describes the size of r_{d,n} , as d\to\infty , and is rather close to covering the entire Fano range. Theorem 1.1 . For n\geqslant d+\log d , we have r_{d,n}=1+O\left(\frac{1}{d^{1/4}}\right), where the implied constant is absolute. Our next result shows that if we are ready to allow the integer d to avoid an appropriate density 0 set of integers, then the condition on n can be relaxed dramatically. Theorem 1.2 . There exists a set \mathcal{D} of density 1 in the positive integers, such that for d\in\mathcal{D} and n\geqslant d^{1/2}\log d , we have r_{d,n}=1+O\left(\frac{1}{\sqrt{d}}\right), where the implied constant is absolute. It follows that in the range n\geqslant d^{1/2}\log d , the proportion of hypersurfaces in \mathbb{B}_{d,n} which have rational points approaches 100\% , as d\to\infty while running over an appropriate set of density 1 . It is rather striking to see that so much of the general type range is handled in this result. The most troublesome degrees d to handle are those whose 2 -adic valuations are large and the set \mathcal{D} is chosen to avoid these. Returning to the question of irreducibility, it follows from work of Bary-Soroker and Kozma [ 1 ] that the proportion of ternary degree d forms that are irreducible over \mathbb{Q} approaches 100\% as d\to\infty . A form f\in\mathbb{Z}[x_{0},\dots,x_{n}] is irreducible over \mathbb{Q} if f(x_{0},x_{1},x_{2},0,\dots,0) has this property. It therefore follows that the proportion of hypersurfaces in \mathbb{B}_{d,n} that are irreducible over \mathbb{Q} approaches 100\% as d\to\infty , provided that n\geqslant 2 . Unlike the case of univariate polynomials, this result is unconditional. On the other hand, while the large box model version of the following conjecture is straightforward, it would be interesting to determine its validity in the large degree model. Conjecture 1.3 . 100% of hypersurfaces V\in\mathbb{B}_{d,n} are smooth, as d\to\infty . The proof of Theorems 1.1 and 1.2 is rather combinatorial and will be completed in Section 5 . The key idea is to study the moments M_{k}(U)=\frac{1}{\#\mathbb{B}_{d,n}}\sum_{V\in\mathbb{B}_{d,n}}\left(\sum_{x\in U\cap V(\mathbb{Q})}1\right)^{k}, for k\in\{1,2\} , where U is an appropriately chosen set of points in \mathbb{P}^{n}(\mathbb{Q}) , always with coordinates drawn from the set \{-1,0,+1\} . One accesses r_{d,n} through the Cauchy–Schwarz inequality, which yields M_{1}(U)^{2}\leqslant r_{d,n}M_{2}(U) . Providing estimates for the moments M_{k}(U) is the focus of Section 4 . Along the way we will need information about the density of locally soluble hypersurfaces in the large degree model, which is the focus of Section 2 , in addition to various estimates for binomial coefficients and sums involving them, for which the relevant facts are collected together in Section 3 . Acknowledgements . The authors are very grateful to Pierre Le Boudec, whose contributions would have merited co-authorship. At his request, however, he is not listed as a co-author, since he has chosen to pursue a career outside academia. T.B. was supported by a grant from the IAS School of Mathematics and a FWF grant (DOI 10.55776/P36278). W.S. was supported by NSF grant DMS-2502029 and a Sloan Research Fellowship. 2. Local solubility In this section we investigate the ratio r_{d,n}^{\mathrm{loc}}=\frac{\#\{V\in\mathbb{B}_{d,n}:V(\mathbf{A}_{\mathbb{Q}})\neq\emptyset\}}{\#\mathbb{B}_{d,n}}, where \mathbf{A}_{\mathbb{Q}} is the ring of adèles. This is the probability that a hypersurface in \mathbb{B}_{d,n} is everywhere locally soluble. We begin with the following elementary fact about the parity of N_{e,\ell}=\binom{e+\ell}{e} . Lemma 2.1 . N_{e,\ell} is even if and only if e and \ell share a common digit 1 in their binary expansions. Proof. Given any integer e\geqslant 1 , we let s_{2}(e) denote the sum of the digits in the binary expansion of e . Let v_{2}(m) denote the 2 -adic valuation of any integer m\geqslant 1 . Legendre’s formula states that v_{2}(e!)=e-s_{2}(e). It follows that for any e,\ell\geqslant 1 , we have (2.1) v_{2}(N_{e,\ell})=s_{2}(e)+s_{2}(\ell)-s_{2}(e+\ell). As a result, we see that N_{e,\ell} is even if and only if s_{2}(e)+s_{2}(\ell)>s_{2}(e+\ell), which is equivalent to demanding that e and \ell share a common digit 1 in their binary expansions. ∎ It will be convenient to have an alternative formulation of N_{d,\ell} , for any d,\ell\geqslant 1 . Let j\in\{1,\dots,\ell\} . Then the number of monomials of degree d in \ell+1 variables is equal to the sum over k\in\{0,\dots,d\} of the number of monomials of degree k in j given variables multiplied by the number of monomials of degree d-k in the remaining \ell+1-j variables. In other words, we have (2.2) N_{d,\ell}=\sum_{k=0}^{d}\binom{k+j-1}{k}\binom{d-k+\ell-j}{d-k}. We now let c_{d,\ell}(j) denote the number of monomials of degree