ProbXiv
sign in
Problem archiveProblem record

Statement

How well separated can a family of point-line pairs in the unit square be? For every ε>0\varepsilon > 0 there are arbitrarily large families (x1,ℓ1),…,(xn,ℓn)(x_1,\ell_1),\ldots,(x_n,\ell_n) in [0,1]2[0,1]^2 with xi∈ℓix_i \in \ell_i and dist(xi,ℓj)≥n−2/3−ε\mathrm{dist}(x_i,\ell_j) \ge n^{-2/3-\varepsilon} for all i≠ji \ne j. Combined with earlier work of Cohen, Pohoata and Zakharov this settles the problem at the sharp exponent 2/32/3. The same construction disproves a conjecture of Hunter, Pohoata, Verstraete and Zhang about induced point-line matchings over finite fields.

Record

Comments

No person has examined this. Nothing here has been checked at all. say whether it holds →

  1. construction · #1

    Cosmin Pohoata, using GPT-5.6 Pro

    That credit came with the record as it was imported. No ProbXiv account is credited for this work, and nobody has answered for it here.

    AI involvement
    ai co developed
    — a person and a model developed the result together.

    The acknowledgement draws the line precisely. The author's own plan was to use a high-degree number field analogue of the Hunter-Pohoata-Verstraete-Zhang construction to reach the Ruzsa endpoint. In his words, the decisive new idea of using the codimension-one, square-difference-free, trace-zero lattice in place of a Ruzsa-like set, which is what upgrades the exponent to the sharp one, is entirely due to GPT-5.6 Pro.

Sign in with an institutional address to take part in the discussion. Reading every thread stays open to everyone.

Sign in

Solve with an agent

Open the statement in a chat, with the problem and the ground rules already written into the prompt.

This opens a third-party site. Nothing is posted back to ProbXiv and nothing you write there is recorded here — what a model gives you is an attempt, which a person still has to check.