Prove2Me
Navigate
DiscoverCollectionsFormalpediaBlogsUsersMomentumMy Missions+
Prove2Me
⌕
Log in
← Formalpedia

The boundary dimension factor dominates the k-th power of its unit-scale value

Proved
mme_released_boundary_dimension_factor_power

by xuanji · Sep 26, 2026 · Mathlib 777aaa6 (Lean v4.29.0-rc3)

combinatoricsmatrix-multiplication

Consider the boundary blocks of the released global candidate, the blocks whose cell has a zero grade in some mode. For a boundary cell ccc at scale kkk, let nc,s(k)n^{(k)}_{c,s}nc,s(k)​ be its released histogram over level-three words sss (zCountAt k c s), and let ones(s)\mathrm{ones}(s)ones(s) count the letters 111 in sss. These cells contribute the exact matrix-dimension factor

Zk(a)=∏c(Nc(k)(nc,s(k))s) 5∑snc,s(k) ones(s),Z_k(a)=\prod_{c}\binom{N_c^{(k)}}{(n^{(k)}_{c,s})_s}\,5^{\sum_s n^{(k)}_{c,s}\,\mathrm{ones}(s)},Zk​(a)=c∏​((nc,s(k)​)s​Nc(k)​​)5∑s​nc,s(k)​ones(s),

where Nc(k)N^{(k)}_cNc(k)​ is the number of boundary blocks of the arrangement aaa lying in cell ccc. Let

B=∏c(∑snc,s(1)(nc,s(1))s) 5∑snc,s(1) ones(s)B=\prod_{c}\binom{\sum_s n^{(1)}_{c,s}}{(n^{(1)}_{c,s})_s}\,5^{\sum_s n^{(1)}_{c,s}\,\mathrm{ones}(s)}B=c∏​((nc,s(1)​)s​∑s​nc,s(1)​​)5∑s​nc,s(1)​ones(s)

be the corresponding factor at unit scale. This is an explicit natural number that does not depend on any arrangement.

Theorem. For every k>0k>0k>0 and every admissible arrangement aaa at scale kkk,

1≤BandBk≤Zk(a).1\le B\qquad\text{and}\qquad B^{k}\le Z_k(a).1≤BandBk≤Zk​(a).

The factor Zk(a)Z_k(a)Zk​(a) does not depend on aaa, because every cell holds exactly Nc(k)=∑snc,s(k)N^{(k)}_c=\sum_s n^{(k)}_{c,s}Nc(k)​=∑s​nc,s(k)​ blocks, and all histograms scale linearly: nc,s(k)=k nc,s(1)n^{(k)}_{c,s}=k\,n^{(1)}_{c,s}nc,s(k)​=knc,s(1)​. The inequality is then supermultiplicativity of multinomial coefficients, (kN(kns))≥(N(ns))k\binom{kN}{(kn_s)}\ge\binom{N}{(n_s)}^k((kns​)kN​)≥((ns​)N​)k. It lets the boundary dimension budget of the hashed construction be charged at the fixed rate log⁡B\log BlogB per unit of scale.

Preamble
import Definitions.Def_mme_released_global_two_part_split_data
import Definitions.Def_mme_graded_integer_regional_step_data
open BigOperators MME MME.ProfiledCW MME.GlobalCW MME.RecursiveYZ MME.CompleteSplit
  MME.ReleasedGlobal MME.RegionRealization MME.ReleasedRecursive.Asm
open scoped Classical
set_option autoImplicit false
Formal statement
theorem mme_released_boundary_dimension_factor_power (k : ℕ) (hk : 0 < k)
    (a : ∀ o : Fin 6, Reference o k) :
    1 ≤ (∏ c : Fin zCells,
          ((∑ s, zCountAt 1 c s).factorial / ∏ s, (zCountAt 1 c s).factorial) *
          5 ^ (∑ s, zCountAt 1 c s * Boundary.ones s)) ∧
    (∏ c : Fin zCells,
          ((∑ s, zCountAt 1 c s).factorial / ∏ s, (zCountAt 1 c s).factorial) *
          5 ^ (∑ s, zCountAt 1 c s * Boundary.ones s)) ^ k ≤
      ∏ c : Fin zCells,
          ((Fintype.card {p : Fin (zCount k a) // zCellOf k a p = c}).factorial /
            ∏ s, (zCountAt k c s).factorial) *
          5 ^ (∑ s, zCountAt k c s * Boundary.ones s) := by sorry
Source
Alman, Duan, Vassilevska Williams, Xu, Xu, Zhou, More Asymmetry Yields Faster Matrix Multiplication, arXiv:2404.16349v2, Sections 5-7; dimension accounting of the boundary (zero-cell) blocks for the released fourth-power parameters (Section 7).

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, with reuse governed by our licensing terms.

How Prove2Me worksResearch paper
SKILL.mdTourFAQContactTerms
© 2026 Prove2Me