The strong cylinder conjecture and Rédei's factorization conjecture in three dimensions
The strong cylinder conjecture asks whether every set of p^2 points in F_p^3 meeting each affine plane in a multiple of p points is a union of p parallel affine lines. Rédei's 1970 conjecture asks whether, in a normalized factorization A⊕B=F_p^3, one factor must lie in a plane through zero. Rédei's conjecture had previously been verified only for p ≤ 13.
The proof assumes a counterexample for p ≥ 11, since small primes were settled earlier. It studies the projective dual configuration of white and black points and shows that all plane intersections have size 0, p, or 2p. Projection moments are interpolated through linear and quadratic forms, using complete intersections and duality for evaluation relations. A classification of generalized hyperfocused arcs then rules out a quadric containing the white point set, which yields a contradiction. Rédei's conjecture follows from p-divisibility of the larger factor, power sums of linear forms, and Rédei's planar factorization theorem. The authors used OpenAI GPT-6 Astra to help develop the proofs and write Lean 4 formalizations, which are available on GitHub.
The strong cylinder conjecture is proved for every prime p. As a consequence, Rédei's factorization conjecture holds in three dimensions: in every normalized factorization of F_p^3, one factor is contained in a plane through zero.
