On the density of rational lines on diagonal cubic hypersurfaces, II
Scott Parsell, Kiseok Yeon · 2026
Read On the density of rational lines on diagonal cubic hypersurfaces, II by Scott Parsell, Kiseok Yeon (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-2605-30549.txt.
Introduction
Parsell and Yeon count rational lines, straight lines with rational coordinates, on diagonal cubic hypersurfaces, curved surfaces defined by cubic equations. They prove the expected count formula for the total when the surface lives in eighteen or more variables, beating the second author's recent work. The proof sharpens a mean-value estimate for minor arcs, exploiting a shifting-variables argument in both base dimensions. The result settles the counting problem for high-dimensional diagonal cubics.
Catalog record
| CREATOR | Scott Parsell, Kiseok Yeon |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 39507 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2605.30549
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
On the density of rational lines on diagonal cubic hypersurfaces, II Scott Parsell, Kiseok Yeon Abstract . In this paper, we establish the expected asymptotic formula for the number of rational lines on a diagonal cubic hypersurface in 18 or more variables, improving on recent work of the second author. This is achieved via a refined mean value estimate for minor arcs that non-trivially exploits a shifting variables argument in both underlying dimensions. 1. Introduction The problem of establishing the existence of linear spaces on algebraic varieties has a long history, beginning with work of Brauer [ 9 ] and Birch [ 3 ] . The inductive nature of these arguments, however, does not immediately yield quantitative bounds on the number of variables required, and in fact the first general explicit bounds appeared many years later in work of Wooley [ 19 ] . Subsequent work of Brandes [ 7 ] establishes asymptotic formulae for the number of linear spaces on a variety defined by a system of homogeneous polynomial equations of the same degree, and Brandes [ 8 ] further specializes to the case of rational lines on the hypersurface defined by a single equation. The case in which the underlying variety is defined by a system of cubic forms has received significant attention, dating to work of Lewis and Schulze-Pillot [ 11 ] . Refinements due to Wooley [ 18 ] show for instance that the hypersurface defined by a cubic form in at least 37 variables contains a rational line. Brandes and Dietmann [ 6 ] later showed that 31 variables suffice in the non-singular case, and this was further improved to 29 in very recent work of Brandes, Dietmann, and Leep [ 5 ] , who also removed the non-singularity hypothesis. When attempting to harness the full power of the circle method, one often restricts attention to the case of diagonal forms. Suppose that k\geq 3 , and consider the hypersurface defined by the additive equation (1.1) \sum_{i=1}^{s}c_{i}z^{k}_{i}=0, where c_{i}\ (1\leq i\leq s) are non-zero integers. Explicit results concerning linear spaces of arbitrary dimension satisfying ( 1.1 ) have origins in work of Arkhipov, Karatsuba, and Chubarikov [ 1 ] , with later refinements occurring in [ 15 ] , [ 16 ] , and [ 12 ] . The special case of lines on a diagonal hypersurface of degree k is examined in [ 14 ] and [ 22 ] , for example, and a detailed study of the cubic case began with work of the first author [ 13 ] . To describe further progress, we set up some additional notation. Our goal is to count pairs of vectors \boldsymbol{x}=(x_{1},\ldots,x_{s})\in\mathbb{Z}^{s} and \boldsymbol{y}=(y_{1},\ldots,y_{s})\in\mathbb{Z}^{s} such that the line l:\boldsymbol{x}+t\boldsymbol{y} is contained in the hypersurface defined by (\ref{hypersurface}). It follows from the Binomial Theorem that such pairs (\boldsymbol{x},\boldsymbol{y}) are precisely the solutions of the system (1.2) \sum_{i=1}^{s}c_{i}x_{i}^{k-j}y_{i}^{j}=0\qquad(0\leq j\leq k). We therefore let N_{s}(X):=N_{s,k}(X;\boldsymbol{c}) denote the number of solutions (\boldsymbol{x},\boldsymbol{y})\in\mathbb{Z}^{2s}\cap[-X,X]^{2s} to (\ref{system}) . By making clever use of the cubic case of Vinogradov’s mean value theorem due to Wooley [ 21 ] , Zhao [ 24 ] established the expected asymptotic formula for N_{s,3}(X) whenever s\geq 21 , improving on the bound s\geq 29 due to the first author [ 17 ] . Following the work of Bourgain, Demeter, and Guo [ 4 ] on the two-dimensional cubic Parsell-Vinogradov system, one can now recover Zhao’s result in a routine manner by completing the system ( 1.2 ) to obtain the required mean value estimate. Very recently, the second author [ 23 ] was able to save two additional variables by devising a multi-dimensional version of Wooley’s shifting variables argument (see [ 20 ] ) that leverages the bounds of [ 4 ] to obtain a mean value estimate restricted to minor arcs. The surplus minor arc savings permits an interpolation with lower moments of the relevant exponential sum, thus reducing the total number of variables required. The purpose of the present paper is to refine the shifting variables argument of [ 23 ] to more fully exploit the two-dimensional nature of the problem. Our main achievement is a stronger mean value estimate that exploits non-trival summation over both integer shifts, subject to a suitable minor arc condition. It transpires that this allows us to save an extra variable, and we thus obtain the following result. Theorem 1.1 . Whenever s\geq 18 , one has N_{s,3}(X)=\sigma X^{2s-12}+O(X^{2s-12-\tau}), for some \tau>0 , where \sigma=\sigma_{\boldsymbol{c}} is a positive constant depending on the coeffcients \boldsymbol{c} . As indicated above, this sharpens [ 23 , Theorem 1.1] , where the same conclusion was obtained for s\geq 19 . Here \sigma represents a product of local densities, which is positive for s\geq 14 as a consequence of [ 13 ] , Lemma 5.1. We mention that a straightforward generalization of our methods, combined with the results of Guo and Zhang [ 10 ] on two-dimensional Parsell-Vinogradov systems, produces analogous results for larger k . In this instance, we would save roughly k/3 variables over what follows directly from [ 10 ] , in which the total number of variables required to obtain the expected asymptotic formula for N_{s,k}(X) is of order k^{3} . Since the savings is not especially compelling and does not illuminate any new ideas, we choose not to provide details in this direction. Throughout we use Vinogradov’s notation, where f\ll g means that f(x)\leq Cg(x) for some sufficiently large constant C>0. We may use f=O(g) with the same meaning. We also make liberal use of vector notation, writing a\leq|{\bf x}|\leq b to mean that each component of {\bf x} lies in [a,b] . Finally, we use \varepsilon to denote a small positive number, whose value may change from statement to statement. 2. Mean value estimates on minor arcs Our main object of interest is the exponential sum (2.1) F(\boldsymbol{\alpha}):=\sum_{1\leq x,y\leq X}e(\alpha_{1}x^{3}+\alpha_{2}x^{2}y+\alpha_{3}xy^{2}+\alpha_{4}y^{3}), with \boldsymbol{\alpha}=(\alpha_{1},\alpha_{2},\alpha_{3},\alpha_{4}). For s\in\mathbb{N} we further define J_{s}(X)=J_{s,2,3}(X) to be the number of integer solutions of the system \sum_{i=1}^{s}x_{i}^{d-j}y_{i}^{j}=\sum_{i=s+1}^{2s}x_{i}^{d-j}y_{i}^{j}\qquad(0\leq j\leq d,\ 1\leq d\leq 3) satisfying 1\leq\boldsymbol{x},\boldsymbol{y}\leq X . In this section we aim to estimate restricted mean values of the exponential sum (\ref{deg k sum}) by relating them to J_{s}(X) , for which sharp bounds are available due to Bourgain, Demeter, and Guo [ 4 ] . The particular estimate we need here, which follows from [ 4 , Theorem 1.5] , is (2.2) J_{10}(X)\ll X^{20+\varepsilon}. In the shifting variables argument of [ 23 ] one replaces x by x-z_{1} and y by y-z_{2} and averages over z_{1} and z_{2} . In analyzing the effects of these shifts on the cubic equations, it transpires that the linear forms (2.3) \displaystyle L_{1}(\boldsymbol{h},\boldsymbol{\alpha}) \displaystyle:=3h_{1}\alpha_{1}+2h_{2}\alpha_{2}+h_{3}\alpha_{3} \displaystyle L_{2}(\boldsymbol{h},\boldsymbol{\alpha}) \displaystyle:=h_{1}\alpha_{2}+2h_{2}\alpha_{3}+3h_{3}\alpha_{4} play a prominent role, as they arise naturally from the partial translation invariance inherent in a Vinogradov-type system with only the quadratic equations inhomogeneous. It is therefore useful to define the corresponding exponential sum (2.4) \displaystyle T(\boldsymbol{\alpha},\boldsymbol{\beta}):=\sum_{1\leq z_{1},z_{2}\leq X}\biggl|\sum_{|{\boldsymbol{h}}|\leq sX^{2}}e(-z_{1}L_{1}({\boldsymbol{h}},\boldsymbol{\alpha})-z_{2}L_{2}({\boldsymbol{h}},\boldsymbol{\alpha})-\boldsymbol{\beta}\cdot{\boldsymbol{h}})\biggr|, where \boldsymbol{\alpha}\in[0,1)^{4} and \boldsymbol{\beta}\in[0,1)^{3} From the argument of [ 23 ] , Lemma 3.1, we extract the following fundamental lemma, which relates mean values of F(\boldsymbol{\alpha}) to pointwise estimates for T(\boldsymbol{\alpha},\boldsymbol{\beta}) . Lemma 2.1 . Let s be a natural number, and let \mathfrak{B} denote