The Hasse principle for homogeneous polynomials with random coefficients over thin sets II

Daniel Flores and Kiseok Yeon · 2025

Read The Hasse principle for homogeneous polynomials with random coefficients over thin sets II by Daniel Flores and Kiseok Yeon (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-2506-01291.txt.

Introduction

Daniel Flores and Kiseok Yeon study the Hasse principle for homogeneous polynomials with random coefficients over thin sets in a 2025 paper. The work asks when a form that has solutions in each local field must also have a whole-number solution, for forms picked at random from a thin set.

The forms are built through the Veronese embedding, which lists all monomials of a set degree in order. By taking on a new lattice problem with the geometry of numbers, the authors confirm the claim for large cases. Their bounds cover many variables against the degree of the form.

Catalog record

CREATORDaniel Flores and Kiseok Yeon
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE55 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2506-01291

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

The Hasse principle for homogeneous polynomials with random coefficients over thin sets II Daniel Flores and Kiseok Yeon Abstract . Let d d and n n be natural numbers. Let ν d , n : ℝ n → ℝ N \nu_{d,n}:\mathbb{R}^{n}\rightarrow\mathbb{R}^{N} denote the Veronese embedding with N = N n , d := ( n + d − 1 d ) N=N_{n,d}:=\binom{n+d-1}{d} , defined by listing all the monomials of degree d d in n n variables using the lexicographical ordering. Let ⟨ 𝒂 , ν d , n ​ ( 𝒙 ) ⟩ ∈ ℤ ⁡ [ 𝒙 ] \langle{\boldsymbol{a}},\nu_{d,n}(\boldsymbol{x})\rangle\in{\mathbb{Z}}[\boldsymbol{x}] be a homogeneous polynomial in n n variables of degree d d with integer coefficients 𝒂 \boldsymbol{a} , where ⟨ ⋅ , ⋅ ⟩ \langle\cdot,\cdot\rangle denotes the inner product. For a non-singular form P ∈ ℤ ⁡ [ 𝒙 ] P\in\mathbb{Z}[\boldsymbol{x}] of degree k ( ≤ d ) k\ (\leq d) in N N variables, consider a set of integer vectors 𝒂 ∈ ℤ N \boldsymbol{a}\in\mathbb{Z}^{N} , defined by 𝔄 ( A ; P ) = { 𝒂 ∈ ℤ N : P ( 𝒂 ) = 0 , ∥ 𝒂 ∥ ∞ ≤ A } . \mathfrak{A}(A;P)=\{\boldsymbol{a}\in{\mathbb{Z}}^{N}:\ P(\boldsymbol{a})=0,\ \|\boldsymbol{a}\|_{\infty}\leq A\}. By handling a new lattice problem via the geometry of numbers, we confirm that whenever n > 24 ​ d n>24d and d ≥ 17 , d\geq 17, the proportion of integer coefficients 𝒂 ∈ 𝔄 ⁡ ( A , P ) \boldsymbol{a}\in\mathfrak{A}(A;P) , whose associated equation f 𝒂 ​ ( 𝒙 ) = 0 f_{\boldsymbol{a}}(\boldsymbol{x})=0 satisfies the Hasse principle, converges to 1 1 as A → ∞ A\rightarrow\infty . This improves on the recent work of the second author. 2020 Mathematics Subject Classification 11E76,14G12 1. Introduction and statement of the results This paper is concerned with the solubility for random diophantine equations. By [ References ] and [ References ], we see that the positive portion of homogeneous polynomials in a sufficiently large number of variables in terms of degree is soluble in {\mathbb{Q}} . To explain more precisely, we let \nu_{d,n}:{\mathbb{R}}^{n}\rightarrow{\mathbb{R}}^{N} denote the Veronese embedding with N=N_{d,n}:=\binom{n+d-1}{d} , defined by listing all the monomials of degree d in n variables using the lexicographical ordering. Let \langle{\boldsymbol{a}},\nu_{d,n}({\boldsymbol{x}})\rangle\in{\mathbb{Z}}[{\boldsymbol{x}}] be a homogeneous polynomial in n variables of degree d with integer coefficients {\boldsymbol{a}} , where \langle\cdot,\cdot\rangle denotes the inner product. Here and throughout this paper, we write f_{{\boldsymbol{a}}}({\boldsymbol{x}})=\langle{\boldsymbol{a}},\nu_{d,n}({\boldsymbol{x}})\rangle . Define \mathfrak{A}(A):=\{{\boldsymbol{a}}\in{\mathbb{Z}}^{N}:\ \|{\boldsymbol{a}}\|_{\infty}\leq A\}. Browning, Sawin and Le Boudec [ References ] proved that the proportion of {\boldsymbol{a}}\in{\mathbb{Z}}^{N} in \mathfrak{A}(A) , whose associated equation f_{{\boldsymbol{a}}}({\boldsymbol{x}})=0 satisfies the Hasse principle, converges to 1 as A\rightarrow\infty, provided that n\geq d+1 and d\geq 4. Furthermore, Poonen and Voloch [ References ] proved that the portion of {\boldsymbol{a}}\in{\mathbb{Z}}^{N} in \mathfrak{A}(A) , whose associated equation f_{{\boldsymbol{a}}}({\boldsymbol{x}})=0 is everywhere locally soluble, converges to a positive constant c as A\rightarrow\infty, provided that n\geq 2 and d\geq 2. Hence, combining these two results, we conclude that the positive portion of homogeneous polynomial equations in n variables of degree d is soluble in {\mathbb{Q}} , provided that n\geq d+1 and d\geq 4. The recent works ([ References ], [ References ]) of Lee, Lee and the second author showed that such a conclusion on global solubility remains true even when the coefficients {\boldsymbol{a}}\in{\mathbb{Z}}^{N} are constrained by a polynomial condition under a modest condition on the number of variables. More precisely, let P be a non-singular form in n variables of degree k\geq 2 . Define \mathfrak{A}(A;P):=\{{\boldsymbol{a}}\in{\mathbb{Z}}^{N}:\ P({\boldsymbol{a}})=0,\ \|{\boldsymbol{a}}\|_{\infty}\leq A\}. The second author [ References ] proved that the proportion of {\boldsymbol{a}}\in{\mathbb{Z}}^{N} in \mathfrak{A}(A;P) , whose associated equation f_{{\boldsymbol{a}}}({\boldsymbol{x}})=0 satisfies the Hasse principle, converges to 1 as A\rightarrow\infty, provided that n\geq 32d+17 , d\geq 14 and k\leq d. Furthermore, Lee, Lee and the second author [ References ] showed that the portion of {\boldsymbol{a}}\in{\mathbb{Z}}^{N} in \mathfrak{A}(A;P) , whose associated equation f_{{\boldsymbol{a}}}({\boldsymbol{x}})=0 is everywhere locally soluble, converges to a constant c_{P} as A\rightarrow\infty, provided that n\geq 2 , d\geq 2 and k\leq C(n,d) for some constant C(n,d) . This constant c_{P} is positive if there exists {\boldsymbol{a}}\in\mathfrak{A}(A;P) such that f_{{\boldsymbol{a}}}({\boldsymbol{x}})=0 has an integer solution {\boldsymbol{x}}\in{\mathbb{Z}}^{n} with \nabla f_{{\boldsymbol{a}}}({\boldsymbol{x}})\neq{\bf 0}. The main purpose of this paper is to relax on the bound for the number of variables n\geq 32d+17 for such a conclusion. The crucial ingredient for this improvement is to handle a new lattice counting problem (see Lemma 3.1 below) which naturally arises in the argument. To do so, we use tools from the geometry of numbers. Our hope is that this argument described in this paper may be helpful to make further improvement. In order to describe our main theorems, we temporarily pause here and provide some definitions. Recall that f_{{\boldsymbol{a}}}({\boldsymbol{x}}) is a homogeneous polynomial in n variables of degree d. Furthermore, for {\boldsymbol{a}}\in{\mathbb{Z}}^{N} and X>0 , we define \mathcal{I}_{{\boldsymbol{a}}}(X):=\{{\boldsymbol{x}}\in[1,X]^{n}\cap{\mathbb{Z}}^{n}:\ f_{{\boldsymbol{a}}}({\boldsymbol{x}})=0\}. We note here that our argument proceeds for fixed X>0 , and thus for simplicity, we write (1.1) w=\log X and (1.2) W=\displaystyle\prod_{p\leq w}p^{\lfloor\log w/\log p\rfloor}. Observe here that an application of the prime number theorem reveals that \log W\leq 2w, which implies (1.3) W\leq X^{2}. For L>0 and {\boldsymbol{a}}\in{\mathbb{Z}}^{N}, we define (1.4) \sigma({\boldsymbol{a}};L)=L^{-(n-1)}\#\{{\boldsymbol{g}}\in[1,L]^{n}:\ f_{{\boldsymbol{a}}}({\boldsymbol{g}})\equiv 0\ \text{mod}\ L\}. We notice that by the Chinese remainder theorem one has (1.5) \sigma({\boldsymbol{a}};L)=\displaystyle\prod_{p^{r}\|L}\sigma({\boldsymbol{a}};p^{r}). Then, on recalling the definition (\ref{def2.3}) of W , we write (1.6) \mathfrak{S}_{{\boldsymbol{a}}}^{*}=\sigma({\boldsymbol{a}};W)=\displaystyle\prod_{p^{r}\|W}\sigma({\boldsymbol{a}};p^{r}). Recall the definition (\ref{def2.2}) of w. Put \zeta=w^{-4-1/(8d)}, and we introduce an auxiliary function \mathfrak{w}_{\zeta}(\beta)=\zeta\cdot\left(\frac{\text{sin}(\pi\zeta\beta)}{\pi\zeta\beta}\right)^{2}. Note here that we chose \zeta differently from [ References ] in order to optimize the result. This function has the Fourier transform \widehat{\mathfrak{w}}_{\zeta}(\xi)=\int_{-\infty}^{\infty}\mathfrak{w}_{\zeta}(\beta)e(-\beta\xi)d\beta=\text{max}\{0,1-|\xi|/\zeta\}. For {\boldsymbol{a}}\in{\mathbb{Z}}^{N} and A,X>0, we define (1.7) \mathfrak{J}_{{\boldsymbol{a}}}^{*}:=\mathfrak{J}_{{\boldsymbol{a}}}^{*}(A,X)=A^{-1}X^{n-d}\displaystyle\int_{[0,1]^{n}}\zeta^{-1}\widehat{\mathfrak{w}}_{\zeta}(A^{-1}f_{{\boldsymbol{a}}}(\boldsymbol{\gamma}))d\boldsymbol{\gamma}. Definition 1.1 . Let n and d be natural numbers with d\geq 2. Consider the monomials of degree d in n variables x_{1},\ldots,x_{n} . In particular, the number of these monomials is N=\binom{n+d-1}{d}. Then, define v_{d}({\boldsymbol{x}})\in{\mathbb{R}}^{n} and w_{d}({\boldsymbol{x}})\in{\mathbb{R}}^{N-n} to be vectors associated with those monomials such that (v_{d}({\boldsymbol{x}}))_{i} is x_{i}^{d} with i=1,\ldots,n and (w_{d}({\boldsymbol{x}}))_{j} ’s are remaining monomials in lexicographical order with j=1,\ldots,N-n , respectively. For example, we find that v_{3}(x_{1},x_{2})=(x_{1}^{3},x_{2}^{3})\ \text{and}\ w_{3}(x_{1},x_{2})=(x_{1}^{2}x_{2},x_{1}x_{2}^{2}). Recall the definition of f_{{\boldsymbol{a}}}({\boldsymbol{x}}) , that is f_{{\boldsymbol{a}}}({\boldsymbol{x}})=\langle{\boldsymbol{a}},\nu_{d,n}({\boldsymbol{x}})\rangle . Let us define a permutation [\ \cdot\ ]:\ {\mathbb{Z}}^{N}\rightarrow{\mathbb{Z}}^{N} in the following way: for a given ({\boldsymbol{b}},{\boldsymbol{c}})\in{\mathbb{Z}}^{N} with {\boldsymbol{b}}\in{\mathbb{Z}}^{n} and {\boldsymbol{c}}\in{\mathbb{Z}}^{N-n} , the permutation [\ \cdot\ ] is mapping ({\boldsymbol{b}},{\boldsymbol{c}})\in{\mathbb{Z}}^{N} to {\boldsymbol{a}}\in{\mathbb{Z}}^{N} such that (1.8) f_{{\boldsymbol{a}}}({\boldsymbol{x}})=\langle{\boldsymbol{b}},v_{d}({\boldsymbol{x}})\rangle+\langle{\boldsymbol{c}},w_{d}({\boldsymbol{x}})\rangle. Theorem 1.2 . Let A and X be positive numbers and let n and d be natural numbers with d\geq 4. Let n=8s+r with s\in{\mathbb{N}} and 1\leq r\leq 8. Suppose that s\geq 3d and X^{2d}\leq A\leq X^{s-d}. Suppose that P\in{\mathbb{Z}}[{\boldsymbol{x}}] is a non-singular form of degree k in N_{d,n} variables. Then, whenever N_{d,n}\geq 200k(k-1)2^{k-1} , there is a positive number \delta<1 such that \displaystyle\sum_{\begin{subarray}{c}\|{\boldsymbol{a}}\|_{\infty}\leq A\\ P({\boldsymbol{a}})=0\end{subarray}}\left|\mathcal{I}_{{\boldsymbol{a}}}(X)-\mathfrak{S}_{{\boldsymbol{a}}}^{*}\mathfrak{J}_{{\boldsymbol{a}}}^{*}\right|^{2}\ll A^{N-4}X^{2n-2d}(\log A)^{-\delta}. Recall the definition of N:=N_{d,n} . In advance of the statement of the following theorem, define a set \mathcal{A}^{\text{loc}}_{d,n}(A;P) of integer vectors {\boldsymbol{a}}\in{\mathbb{Z}}^{N} in \mathfrak{A}(A;P) having the property that the associated equation f_{{\boldsymbol{a}}}({\boldsymbol{x}})=0 is everywhere locally soluble. Theorem 1.3 . Let A and X be positive numbers with X^{3}\leq A. Suppose that n and d are natural numbers with n>d+1 and d\geq 2 . Suppose that P\in{\mathbb{Z}}[{\boldsymbol{x}}] is a non-singular form of degree k in N_{d,n} variables. Then, whenever N_{d,n}\geq 1000n^{2}8^{k} ,

Read the complete text