Pith. sign in

REVIEW 4 minor 5 references

A full-rank lattice construction answers Gowers's 2009 point-spacing question affirmatively in every dimension, and shows that the existence of such point sets cannot alone prove Littlewood's conjecture.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

T0 review · deepseek-v4-flash

2026-08-01 01:31 UTC pith:L5TMY7HF

load-bearing objection A one-page lattice construction that answers Gowers's 2009 question and closes a proposed route to Littlewood; the math is correct, the footnote needs a rewrite.

arxiv 2607.27780 v1 pith:L5TMY7HF submitted 2026-07-30 math.NT

On a question of Gowers related to Littlewood's conjecture

Frederik Broucke , M\'at\'e Matolcsi , Szil\'ard Gy. R\'ev\'esz This is my paper
classification math.NT MSC 11H0611R0411R29
keywords Littlewood's conjectureGowers problemhyperbolic regionlattice constructionMinkowski embeddingtotally real number fieldsdiscriminantsDelsarte problem
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper proves that for every dimension d there is a full-rank lattice whose only intersection with the hyperbolic region {|x1...xd| < 1} is the origin. Scaling this lattice yields, for arbitrarily large n, n points in the unit cube such that the product of coordinate differences between any two distinct points is at least c/n for any c below the reciprocal of the lattice covolume. This unconditionally and explicitly answers a 2009 question of Gowers, which was posed as a potential route toward Littlewood's conjecture in Diophantine approximation. The authors conclude that the mere existence of such point sets is compatible with both truth and falsity of Littlewood's conjecture, so any proof of the conjecture must use additional structure beyond this kind of spacing.

Core claim

Theorem 2 states that for any dimension d there exists a full-rank lattice Λ in R^d with Λ ∩ {x : |x1 · · · xd| < 1} = {0}. The proof embeds the ring of integers of a real Galois number field K of degree d into R^d via its d embeddings; for each lattice point, the product of its coordinates equals the algebraic norm of the corresponding integer, which is a rational integer. Thus a nonzero lattice point's coordinate product has absolute value at least 1, placing it outside the open hyperbolic region. Scaling the lattice gives point sets Y_R = {λ/R^{1/d} : λ ∈ Λ ∩ [0,R^{1/d}]^d}, whose pairwise products of coordinate differences exceed 1/R, yielding the asserted answer to Gowers's Problem 1.

What carries the argument

The central object is the Minkowski embedding of the ring of integers O_K of a totally real (or real Galois) number field K of degree d, sending α to (σ_1(α),...,σ_d(α)). The crucial identity is ∏ σ_i(α) = Norm_{K/Q}(α) ∈ Z for α ∈ O_K, which forces any nonzero lattice point to have coordinate product with absolute value at least 1. This single integer-norm identity carries the whole construction: it is what lets the lattice miss the hyperbolic region completely.

Load-bearing premise

The 'any dimension' part of the theorem depends on the existence, for every degree d, of a totally real (or real Galois) number field of degree d; the paper sketches such fields via a cyclotomic construction but does not give a full proof, so this existence is the load-bearing step.

What would settle it

For the explicit dimension-2 lattice Λ = (1,1)Z + (√2, -√2)Z, compute the coordinate-product of any nonzero lattice point, which equals m^2 - 2n^2 with m,n ∈ Z; this is never a nonzero integer of absolute value less than 1. A direct finite search over coefficients in, say, the range [-1000,1000] would confirm the identity, and any nonzero combination with |m^2 - 2n^2| < 1 would overturn the construction.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

If this is right

  • Gowers's Problem 1 has an unconditional affirmative answer in every dimension d, with point sets constructed explicitly from algebraic number fields.
  • The existence of such point sets is compatible with both truth and falsity of Littlewood's conjecture, so this geometric approach alone cannot settle the conjecture.
  • The asymptotic constant c_d is tied to the minimal discriminant of totally real number fields of degree d; in dimension 2 it is 1/√5 and in dimension 3 it is 1/7.
  • In the Delsarte-problem reformulation, the maximal size N_d(ε) of ε-separated sets in the torus satisfies N_d(ε) ≍ 1/ε, with explicit bounds on the Delsarte constants.
  • The constructed point sets in dimension 3 can be chosen, up to scaling, to lie in a fixed real cubic field, the same setting where Littlewood's conjecture is known to hold.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same Minkowski-embedding argument works for any finite-index additive subgroup of O_K, so one can freely increase the lattice covolume; decreasing it below the square root of the field discriminant would require a non-field-theoretic construction, and a computational search for such lattices in small dimensions would test whether the constant is truly optimal.
  • The dimensional uniformity of the construction suggests a closer analogy between hyperbolic-region lattices and simultaneous Diophantine approximation; one could try to generalize the construction to regions defined by products of powers of coordinates, where norms of algebraic integers still give integer obstructions.
  • Because the proof identifies the best constant with minimal discriminants, the natural next question is whether the asymptotic relation N_d(ε) ~ C_d/ε can be made exact, perhaps with C_d equal to a function of the minimal discriminant, a conjecture that goes beyond the bounds proved in the paper.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

0 major / 4 minor

Summary. The paper answers Gowers's Problem 1 affirmatively in every dimension d by constructing a full-rank lattice Λ in R^d whose only intersection with the hyperbolic region H_d = {x : |x_1⋯x_d| < 1} is the origin. The lattice is the Minkowski embedding of the ring of integers of a real Galois (or totally real) number field K of degree d; for a nonzero algebraic integer α, the field norm N_{K/Q}(α) is a nonzero integer, so the product of the conjugates has absolute value at least 1. Scaling the lattice gives, for every large n, point sets in [0,1]^d with pairwise 'distance' at least c/n for any c < 1/vol(R^d/Λ). The paper also relates this to the Delsarte problem, establishing that the maximal separation number N_d(ε) satisfies N_d(ε) ≍ 1/ε, and it proves the needed upper bound directly rather than relying on the cited strong duality theorem.

Significance. The result is a clean and significant resolution of a question posed by Gowers: it shows that the existence of such point sets alone cannot prove Littlewood's conjecture without additional structure. The construction is explicit, parameter-free, and yields optimal-order constants in all dimensions; the norm argument is elegant and reduces the problem to a classical fact about number fields. The paper is largely self-contained: the lower bound for N_d(ε) is supplied by the lattice construction, and the upper bound is proved directly, so the citation to the authors' preprint [1] for strong duality is not load-bearing. The only substantive presentation gap is a garbled footnote on the existence of the required fields; since the existence is a standard theorem, this is a local, easily fixable issue and does not affect the central argument.

minor comments (4)
  1. [Section 2, footnote 1] The cyclotomic construction of a degree-d real Galois extension is garbled. The automorphism is written as 'ζ_p ↦ ζ_p^d' and the sentence 'This field has order p−1/(p−1)/d = d' is unintelligible; taken literally, it does not demonstrate the existence of a degree-d totally real subfield. Since Theorem 2 asserts 'for any dimension d', please replace the footnote with the standard construction (or cite a textbook): for a prime p ≡ 1 mod 2d, take the fixed field of the unique subgroup of Gal(Q(ζ_p)/Q) of order (p−1)/d; the fixed field has degree d and is totally real because the subgroup contains complex conjugation. This is a local fix and does not affect the rest of the proof.
  2. [Section 3, last display] In the final inequality chain, '2dε' appears instead of '2^d ε' in both occurrences (ε ≤ D(H_d^ε) ≤ 2^d ε/c_d (1+o(1)) and c_d/(2^d ε)(1+o(1)) ≤ D'(H_d^ε)). The preceding estimates make the intended exponent clear, but the display should be corrected.
  3. [Section 2, covolume paragraph] Typo: 'covloume' should be 'covolume'.
  4. [Section 3, Delsarte definition] Minor notational issue: the paper defines F(H_d^ε) as functions 'non-positive' on T^d \ H_d^ε; the later test function φ = (1/ε)χ∗χ is nonnegative. The use of 'non-positive' here means '≤ 0' (zero outside H_d^ε), which is consistent, but the wording could be clarified to avoid confusion.

Circularity Check

0 steps flagged

No circularity in the main derivation: the Gowers construction follows directly from the integer norm. The self-citation for strong duality is not load-bearing, and the garbled cyclotomic footnote is a presentation defect, not a circular step.

full rationale

The central claim is self-contained. Theorem 2 follows from the Minkowski embedding of the ring of integers of a real Galois field: for every nonzero algebraic integer, the field norm is a nonzero integer, so no nonzero lattice point can lie in the hyperbolic region |x_1...x_d|<1. The subsequent scaling argument is an elementary volume comparison and does not presume the conclusion. The constants c_d are read off from known discriminant tables, not fitted to the target. Reference [1] is a self-citation for strong duality in the Delsarte section, but the key inequality N_d(ε)≤1/ε is proved directly in the text, and strong duality is a parameter-free general theorem, so it does not make the derivation circular. The footnote's cyclotomic construction of real Galois fields is garbled as typeset, and the displayed generator is not generally the intended one, but the needed existence is classical and explicit examples cover dimensions 2 and 3; this is a minor presentation defect, not an instance of a prediction reducing to its assumptions. Overall, no part of the chain defines, fits, or renames its output as an input.

Axiom & Free-Parameter Ledger

0 free parameters · 5 axioms · 0 invented entities

The central construction relies only on standard algebraic number theory and elementary lattice point counting. No parameter is fitted to data, and no new entities are introduced. The exact constants depend on an external discriminant table, but the main existence result does not.

axioms (5)
  • standard math Minkowski embedding of the ring of integers of a totally real number field of degree d is a full-rank lattice in R^d with covolume sqrt(Delta_K).
    Invoked in the proof of Theorem 2 and in the covolume computation after the proof.
  • standard math For a nonzero algebraic integer alpha, the norm N_{K/Q}(alpha) is a rational integer with |N| >= 1, and N(alpha)=0 implies alpha=0.
    This is the core fact that yields Lambda ∩ H_d = {0}.
  • standard math There exist totally real Galois extensions of every degree d, e.g., via subfields of real cyclotomic fields.
    Needed for the 'any dimension d' form of Theorem 2; footnote 1 sketches a cyclotomic construction.
  • domain assumption The minimal discriminants of totally real quadratic and cubic fields are 5 and 49 (Odlyzko's table).
    Used only for the exact numerical constants c_2 = 1/sqrt(5) and c_3 = 1/7; the affirmative answer to Gowers's question does not depend on these values.
  • domain assumption Strong duality D(H)D'(H) = 1 for the Delsarte problem, cited from [1].
    Used in Section 3 for the Delsarte inequalities; the paper gives a direct proof of the one-sided bound actually needed, so this axiom is auxiliary.

pith-pipeline@v1.3.0-daily-deepseek · 4037 in / 16714 out tokens · 161107 ms · 2026-08-01T01:31:13.411416+00:00 · methodology

0 comments
read the original abstract

In a blogpost in 2009, Gowers raised a possible approach to Littlewood's conjecture in Diophantine approximation, leading to a question about the existence of sufficiently many points in the unit cube such that "hyperbolic distance" between any two of them is large. In this note we answer this question by an explicit construction. This shows that this approach to prove Littlewood's conjecture cannot work, unless some further refinements are added.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

5 extracted references · 1 linked inside Pith

  1. [1]

    E. E. Berdysheva, B. Farkas, M Ga´ al, M. D. Ramabulana, Sz. Gy. R´ ev´ esz,Duality for Delsarte’s extremal problem on locally compact Abelian groups, preprint available at arXiv:2603.18287 (2026)

    Pith/arXiv arXiv 2026

  2. [2]

    J. W. S. Cassels, H. P. F. Swinnerton-Dyer,On the product of three homogeneous linear forms and the indefinite ternary quadratic forms, Philos. Trans. Roy. Soc. London Ser. A248(1955), 73–96

    1955

  3. [3]

    Einsiedler, A

    M. Einsiedler, A. Katok, E. Lindenstrauss,Invariant measures and the set of exceptions to Littlewood’s conjecture, Ann. of Math.164(2006), no. 2, 513–560

    2006

  4. [4]

    Gowers,Problems related to Littlewood’s conjecture, blogpost available athttps://gowers

    T. Gowers,Problems related to Littlewood’s conjecture, blogpost available athttps://gowers. wordpress.com/2009/11/17/problems-related-to-littlewoods-conjecture-2/, 2009. Accessed on 19/06/2026

    2009

  5. [5]

    A. M. Odlyzko,Bounds for discriminants and related estimates for class numbers, regulators and zeros of zeta functions: a survey of recent results, J. Th´ eor. Nombres Bordeaux2(1990), no. 1, 119–141. ON A QUESTION OF GOWERS RELATED TO LITTLEWOOD’S CONJECTURE 5 Alfr´ed R´enyi Institute of Mathematics, Re´altanoda utca 13–15, 1053 Budapest, Hungary Email a...

    1990