Generalised Fermat equations in dense variables over finite fields and rings

Sam Chow and Zi Li Lim and Akshat Mudgal · 2026

Read Generalised Fermat equations in dense variables over finite fields and rings by Sam Chow and Zi Li Lim and Akshat Mudgal (2026) 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-2601-00135.txt.

Introduction

Sam Chow, Zi Li Lim, and Akshat Mudgal count solutions to Fermat style equations over dense sets. Their 2026 study works in finite fields and in cyclic rings. The set must be dense enough. The field and ring sizes must avoid small prime divisors. Under those limits, the equations have the expected number of solutions inside the set.

The team also shows its density threshold cannot be improved. The proof mixes several tools from additive number theory. These include Fourier decay for Bohr sets and mixed character sum bounds. It also uses equidistribution of polynomial sequences. Popular Cauchy Davenport lemmas and a regularity lemma play a part. For number theorists, the result says dense sets behave like the whole space. That holds even for these rigid polynomial equations.

Catalog record

CREATORSam Chow and Zi Li Lim and Akshat Mudgal
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE52 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2601-00135

Browse the full library · About the Tower of Babel

Full text

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

Generalised Fermat equations in dense variables over finite fields and rings Sam Chow and Zi Li Lim and Akshat Mudgal Abstract . Let A A be a sufficiently dense subset of a finite field 𝔽 q \mathbb{F}_{q} or a finite, cyclic ring ℤ / N ​ ℤ \mathbb{Z}/N\mathbb{Z} . Assuming that q q and N N have no small prime divisors, we show that generalised Fermat equations have the expected number of solutions over A A . We further show that our density threshold is optimal. Our proofs involve average Fourier decay for Bohr sets, mixed character sum bounds, equidistribution of polynomial sequences, popular Cauchy–Davenport lemmas, and a regularity-type lemma due to Semchankau. 2020 Mathematics Subject Classification 11B30 (primary); 11D41, 11D45, 11P05, 11T23 (secondary) Dedicated to Trevor Wooley on the occasion of his (5^{3}-4^{3}-1^{3})^{\mathrm{th}} birthday 1. Introduction This paper studies solutions to Fermat equations over dense subsets of finite fields and cyclic rings. Such a problem seems to have been first studied by Csikvári–Gyarmati–Sárközy [ 6 ] , who considered solutions to the equation (1.1) x+y=z^{2} with the variables lying in dense subsets of finite fields. They observed that the set (1.2) A=\{x\in\mathbb{F}_{p}:x\in(0,p/4)\ \text{and}\ x^{2}\in(p/2,p)\}, while satisfying |A|\gg p , does not exhibit any solutions to ( 1.1 ). They further asked whether a colouring version of this result might exist, that is, for any partition \mathbb{F}_{p}=C_{1}\cup\dots\cup C_{r} with r\in\mathbb{N} , does there always exist some 1\leqslant i\leqslant r such that C_{i} has solutions to ( 1.1 )? The latter question, along with its generalisation to more broader equations of the shape (1.3) x^{\alpha}+y^{\beta}=z^{\gamma}, was resolved by Lindqvist [ 15 ] . In fact, Lindqvist proved that for any \alpha,\beta,\gamma\in\mathbb{N} and any partition \mathbb{F}_{p}=C_{1}\cup\dots\cup C_{r} , with r\in\mathbb{N} , there must exist some 1\leqslant i\leqslant r such that C_{i} has \gg_{r,\alpha,\beta,\gamma}p^{2} many solutions to ( 1.3 ). This matches the total number of solutions to ( 1.3 ) in \mathbb{F}_{p} , up to multiplicative constants. Such a phenomenon has been referred to as supersaturation , by analogy with the extremal combinatorics parlance. We refer the reader to the pioneering works of Varnavides [ 21 ] and Frankl–Graham–Rödl [ 8 ] , who established these types of results for linear equations over integers. See also [ 3 , 4 , 16 ] and [ 14 ] for supersaturation results for non-linear equations over the integers and finite, cyclic rings, respectively. Returning to the set A in ( 1.2 ), one can prove using standard Fourier-analytic methods that |A|\geqslant p/8-o(p) . A natural question then is whether there is any larger set that does not contain solutions to ( 1.1 ). A straightforward consequence of the results in this paper is that this is roughly the best possible construction. Corollary 1.1 . For every \kappa>0 , every prime p sufficiently large in terms of \kappa , and every set A\subseteq\mathbb{F}_{p} satisfying |A|\geqslant p(1/8+\kappa) , one has \sum_{x,y,z\in A}\mathds{1}_{x+y=z^{2}}\gg_{\kappa}p^{2}. In fact, our results in this paper provide optimal density thresholds along with supersaturation for a wider family of equations in broader algebraic settings. Writing [n]=\{1,2,\dots,n\} for every n\in\mathbb{N} , our first main result over general finite fields is as follows. Theorem 1.2 . Let r and i_{1}<\dots<i_{r} be positive integers, and let s\geqslant 3 be an integer. Let [s]=I_{1}\cup\dots\cup I_{r} be a partition of [s] into discrete intervals I_{1},\dots,I_{r} such that |I_{j}|=k_{j} for all j . Let {\kappa}>0 , and let {\varepsilon}>0 be sufficiently small in terms of s,i_{r},\kappa . Let p be a prime sufficiently large in terms of s,i_{r},\kappa , and let q=p^{m} for some m\in\mathbb{N} . Let u\in\mathbb{F}_{q} , let c_{1},\ldots,c_{s}\in\mathbb{F}_{q}\setminus\{0\} , and let A\subseteq\mathbb{F}_{q} with (1.4) |A|\geqslant\left(\prod_{1\leqslant j\leqslant r}\frac{1}{rk_{j}}+\kappa\right)q. Then there are at least {\varepsilon}q^{s-1} many solutions \mathbf{x}\in A^{s} to (1.5) \sum_{1\leqslant j\leqslant r}\sum_{n\in I_{j}}c_{n}x_{n}^{i_{j}}=u. Corollary 1.1 can be deduced directly from Theorem 1.2 by setting r=2,\quad(i_{1},i_{2})=(1,2),\quad[3]=\{1,2\}\cup\{3\},\quad u=0. Another setting of interest is the case when r=1 and i_{1}=k . This can be interpreted as a dense Waring-type problem in \mathbb{F}_{p} . Corollary 1.3 . Let s\geqslant 3 and k\geqslant 1 be integers. Let {\kappa}>0 , and let {\varepsilon}>0 be sufficiently small in terms of s,k,{\kappa} . Let p be a prime sufficiently large in terms of s,k,\kappa , and let q=p^{m} for some m\in\mathbb{N} . Let u\in\mathbb{F}_{q} , let c_{1},\ldots,c_{s}\in\mathbb{F}_{q}\setminus\{0\} , and let A\subseteq\mathbb{F}_{q} with \frac{|A|}{q}\geqslant\frac{1}{s}+\kappa. Then there are at least {\varepsilon}q^{s-1} many solutions \mathbf{x}\in A^{s} to (1.6) c_{1}x_{1}^{k}+\cdots+c_{s}x_{s}^{k}=u. Waring’s problem over \mathbb{N} concerns representing all positive integers — or all but finitely many — as sums of at most s many k^{\mathrm{th}} powers of positive integers. The majority of interest has revolved around the search for the best possible dependence of s on k , see [ 1 ] . More recently, there has been work which has focused on proving dense versions of Waring’s problem, that is, given a sufficiently dense subset \mathcal{A} of the k^{\mathrm{th}} powers, one wants to represent all sufficiently large positive integers as a sum of at most s many elements from \mathcal{A} , see [ 17 ] . One can also replace the k^{\textrm{th}} powers with other arithmetically interesting sets, see the very nice work of Shao [ 19 ] which considers the analogous problem over dense subsets of primes, see also [ 13 ] . All the aforementioned works proceed via applying the transference principle from additive combinatorics [ 9 ] to switch this problem to the \mathbb{Z}/N\mathbb{Z} setting, for some appropriately chosen N\in\mathbb{N} . Our second main result is an optimal version of Theorem 1.2 over finite, cyclic rings \mathbb{Z}/N\mathbb{Z} , for suitably rough N . Theorem 1.4 . Let r and i_{1}<\dots<i_{r} be positive integers, and let s\geqslant 3 be an integer. Let [s]=I_{1}\cup\dots\cup I_{r} be a partition of [s] into discrete intervals I_{1},\dots,I_{r} such that |I_{j}|=k_{j} for all j . Let {\kappa}>0 , let \Omega\geqslant 1 , and let {\varepsilon}>0 be sufficiently small in terms of s,i_{r},\kappa,\Omega . Then there exists a large constant R depending only on s,i_{r},\kappa,\Omega such that the following is true. Let N=p_{1}^{m_{1}}\cdots p_{t}^{m_{t}} with (1.7) t,m_{1},\dots,m_{t}\leqslant\Omega\quad\text{and}\quad p_{1},\dots,p_{m}\geqslant R. Let u\in\mathbb{Z}/N\mathbb{Z} , let c_{1},\ldots,c_{s}\in(\mathbb{Z}/N\mathbb{Z})^{\times} , and let A\subseteq\mathbb{Z}/N\mathbb{Z} with (1.8) |A|\geqslant\left(\prod_{1\leqslant j\leqslant r}\frac{1}{rk_{j}}+\kappa\right)N. Then there are at least {\varepsilon}N^{s-1} many solutions \mathbf{x}\in A^{s} to ( 1.5 ). Above and in the sequel, the p_{j} are understood to be pairwise distinct primes. As before, Theorem 1.4 dispenses the following optimal dense Waring result in \mathbb{Z}/N\mathbb{Z} for suitably rough numbers N . Corollary 1.5 . Let s\geqslant 3 and k\geqslant 1 be integers. Let {\kappa}>0, let \Omega\geqslant 1 , and let {\varepsilon}>0 be sufficiently small in terms of s,k,{\kappa},\Omega . Then, there exists a large constant R depending only on s,k,\kappa,\Omega such that the following is true. Let N=p_{1}^{m_{1}}\cdots p_{t}^{m_{t}}\in\mathbb{N} with ( 1.7 ). Let c_{1},\ldots,c_{s}\in(\mathbb{Z}/N\mathbb{Z})^{\times} , let u\in\mathbb{Z}/N\mathbb{Z} , and let A\subseteq\mathbb{Z}/N\mathbb{Z} with \frac{|A|}{N}\geqslant\frac{1}{s}+\kappa. Then there are at least {\varepsilon}N^{s-1} many solutions \mathbf{x}\in A^{s} to ( 1.6 ). The only previously known result in this direction is due to the second author [ 14 ] , who considered the setting of Corollary 1.5 when k=2 and s\geqslant 5 , but with more general choices of N . Hence, in comparison, our method gives optimal density results for higher powers with potentially distinct exponents in

Read the complete text