Documentation

LeanPool.ErdosGinzburgZiv.Solution

Checked proofs of the independent Palomar statements #

This module deliberately does not import Challenge. The definitions below are the same independent definitions, and the bridge lemmas identify their constants with the ones used by the full EGZ proof.

Every length-n sequence in (ZMod p)^d has p distinct positions whose vectors sum to zero. Repeated vector values are allowed.

Equations
Instances For
    def PalomarEGZ.isHollow (p : ℕ) {d s : ℕ} (v : Fin s → Fin d → ZMod p) :

    A family is p-hollow if a nonnegative integer combination of total weight p sums to zero exactly when one member receives all the weight. At primes this condition forces the family to have no repeated vectors.

    Equations
    Instances For
      noncomputable def PalomarEGZ.egzConstant (p d : ℕ) :

      The Erdős–Ginzburg–Ziv constant: the least length at which every sequence has a p-term zero sum. For prime p this defining set is nonempty, by pigeonhole with the bound (p-1) * p^d + 1.

      Equations
      Instances For
        noncomputable def PalomarEGZ.hollowConstant (p d : ℕ) :

        The maximum size of a p-hollow family. For prime p, admitted lengths are nonempty and bounded by p^d, so this natural supremum is attained. Only prime values are used below; Mathlib's total natural supremum also assigns a value at degenerate moduli where these lengths may be unbounded.

        Equations
        Instances For

          Tending to infinity through prime natural numbers.

          Equations
          Instances For
            theorem PalomarEGZ.theorem_1_2 (d : ℕ) (hd : 0 < d) :
            (fun (p : ℕ) => ↑(egzConstant p d) - ↑p * ↑(hollowConstant p d)) =o[atTopAlongPrimes] fun (p : ℕ) => ↑p

            Theorem 1.2: for each fixed positive dimension, s((F_p)^d) = p * w((F_p)^d) + o(p) as p tends to infinity through primes.

            theorem PalomarEGZ.main_upper_bound (d : ℕ) (hd : 0 < d) (ε : ℝ) :
            0 < ε → ∀ᶠ (p : ℕ) in atTopAlongPrimes, ↑(egzConstant p d) ≤ (↑(hollowConstant p d) + ε) * ↑p

            The corresponding upper estimate: for each positive error ε, every sufficiently large prime satisfies s((F_p)^d) ≤ (w((F_p)^d) + ε) p.