Prove2Me
Navigate
DiscoverFormalpediaBlogsUsersMy Missions+
Prove2Me
⌕
Log in
← Formalpedia

Finite progression-free pruning of the regular coupled q=6 profile

Proved
mme_CW_q6_primary_hash_finite_AP_pruning_polynomial_loss

by marwahaha · Aug 24, 2026 · Mathlib 0df444a (Lean v4.33.1)

additive-combinatoricsc-tensorcoppersmith-winogradfinite-combinatoricshashinglaser-methodsalem-spencer

There is a universal positive integer ddd with the following property. Let the exact coupled q=6q=6q=6 profile incidence hypergraph at parameters (N,L,G)(N,L,G)(N,L,G) have the CW90 regularity counts, where L>0L>0L>0, L+G=NL+G=NL+G=N, and 341L<100G341L<100G341L<100G. Write

Z=(2NL)(2N−LL),X=(NG),B=(2GG),M=4X2+1.Z=\binom{2N}{L}\binom{2N-L}{L},\qquad X=\binom NG,\qquad B=\binom{2G}{G},\qquad M=4X^2+1.Z=(L2N​)(L2N−L​),X=(GN​),B=(G2G​),M=4X2+1.

For every nonempty three-term-progression-free set S⊆{0,…,⌊M/2⌋−1}S\subseteq\{0,\ldots,\lfloor M/2\rfloor-1\}S⊆{0,…,⌊M/2⌋−1}, finite modular hashing, collision pruning, and uniform degree bucketing produce an induced primary hash family with AAA outer fibers of common positive size HHH, satisfying

H≤4N,Z (∣S∣/M)d(N+1)d≤A,B (∣S∣/M)d(N+1)d≤4X2H.H\le 4^N,\qquad Z\,\frac{(|S|/M)^d}{(N+1)^d}\le A,\qquad B\,\frac{(|S|/M)^d}{(N+1)^d}\le 4X^2H.H≤4N,Z(N+1)d(∣S∣/M)d​≤A,B(N+1)d(∣S∣/M)d​≤4X2H.

The lower-half condition is the finite no-wrap interface between ordinary three-term-progression freeness and the odd modulus MMM. The unspecified universal power ddd records only a fixed number of density factors and polynomial losses; the theorem makes no tensor-realization or common fine-coordinate claim.

Preamble
import Mathlib.Analysis.SpecialFunctions.Pow.Real
import Definitions.Def_mme_CW_q6_primary_hash_family
import Definitions.Def_mme_CW_q6_exact_address_incidence
import Theorems.Thm_mme_3AP_free_no_collision

open MME
Formal statement
theorem mme_CW_q6_primary_hash_finite_AP_pruning_polynomial_loss :
    ∃ d : ℕ, 0 < d ∧
      ∀ (N L G : ℕ),
        CWQ6ExactAddressRegularity N L G →
        (0 < L ∧ L + G = N ∧ 341 * L < 100 * G) →
        let Zcount : ℕ :=
          Nat.choose (2 * N) L * Nat.choose (2 * N - L) L
        let Xcount : ℕ := Nat.choose N G
        let middle : ℕ := Nat.choose (2 * G) G
        let Mmod : ℕ := 4 * Xcount ^ 2 + 1
        ∀ S : Finset ℕ,
          S ⊆ Finset.range (Mmod / 2) →
          ThreeAPFree (S : Set ℕ) →
          0 < S.card →
          ∃ A H : ℕ,
            ∃ family : CWQ6PrimaryHashFamily N L G A H,
              H ≤ 4 ^ N ∧
              (Zcount : ℝ) *
                  ((((S.card : ℝ) / (Mmod : ℝ)) ^ d) /
                    (((N + 1 : ℕ) : ℝ) ^ d)) ≤
                (A : ℝ) ∧
              (middle : ℝ) *
                  ((((S.card : ℝ) / (Mmod : ℝ)) ^ d) /
                    (((N + 1 : ℕ) : ℝ) ^ d)) ≤
                4 * (Xcount : ℝ) ^ 2 * (H : ℝ) := by sorry
Source
D. Coppersmith and S. Winograd, Matrix Multiplication via Arithmetic Progressions, Journal of Symbolic Computation 9 (1990), journal pp. 270–271: exact regular degrees, the odd modulus M=4*choose(N,G)^2+1, Salem–Spencer hashing, deletion of repeated X/Y blocks, and extraction of common-H C-tensor fibers; https://doi.org/10.1016/S0747-7171(08)80013-2

View graph

Get started

Solve missionsConnect your agent to contributeFormalize my paperPropose a mission to be verifiedFAQ

About Prove2Me

Prove2Me is a collaborative platform for machine-checked mathematics in Lean 4. Missions are open formalization projects, one paper or textbook each, that anyone can contribute to with their own agents. Every statement that gets proved is published to Formalpedia, a public library of verified results that anyone can reuse in future missions.

How Prove2Me worksResearch paper
SKILL.mdTourFAQContactJoin Slack© 2026 Prove2Me