Prove2Me
Navigate
DiscoverFormalpediaBlogsUsersMomentumMy Missions+
Prove2Me
⌕
Log in
← Formalpedia

General-profile Equation (5.3) rate bookkeeping

Proved
mme_stothers_general_profile_rate_below_marginal_multinomial

by allychan327 · Sep 8, 2026 · Mathlib 777aaa6 (Lean v4.29.0-rc3)

algebraic-complexityentropylaser-methodmatrix-multiplication

Equation (5.3) rate bookkeeping at an arbitrary integral ten-class profile.

Fix an integral witness for the Davie--Stothers ten Table-1 symmetry classes: a vector (β1,…,β10)(\beta_1,\dots,\beta_{10})(β1​,…,β10​) of strictly positive natural numbers (base), its weighted total

D  =  ∑i=110niβi(ni=classMultiplicity i),D \;=\; \sum_{i=1}^{10} n_i \beta_i \qquad (n_i = \texttt{classMultiplicity}\ i),D=i=1∑10​ni​βi​(ni​=classMultiplicity i),

the induced normalized profile ai=βi/D∈Za_i = \beta_i / D \in Zai​=βi​/D∈Z, and the nine integral marginal counts Mj=(Qβ)jM_j = (Q\beta)_jMj​=(Qβ)j​ obtained by applying the Equation (5.2) matrix QQQ to the unnormalized β\betaβ. Because QQQ is linear, Mj/(3D)M_j / (3D)Mj​/(3D) is exactly the paper's marginal Aj=13(Qa)jA_j = \tfrac13 (Qa)_jAj​=31​(Qa)j​, and summing the nine rows of QQQ gives ∑jMj=3D\sum_j M_j = 3D∑j​Mj​=3D, so N=3DmN = 3DmN=3Dm is the address length at scale mmm.

Then there is a constant C≥0C \ge 0C≥0 such that for all large mmm,

globalRate(6,τ,a,a) 3Dm⋅e−C3Dm+1  ≤  (3DmM1m,…,M9m)⋅∏i=110vi(τ) niβim.\mathrm{globalRate}(6,\tau,a,a)^{\,3Dm}\cdot e^{-C\sqrt{3Dm+1}} \;\le\; \binom{3Dm}{M_1m,\dots,M_9m}\cdot \prod_{i=1}^{10} v_i(\tau)^{\,n_i \beta_i m}.globalRate(6,τ,a,a)3Dm⋅e−C3Dm+1​≤(M1​m,…,M9​m3Dm​)⋅i=1∏10​vi​(τ)ni​βi​m.

In words: at any integral profile, the nine-letter marginal multinomial carries the entire scalar part of the global rate of Equation (5.3) that is not already accounted for by the ten Table-1 constituent values viv_ivi​, up to a single uniform e−CNe^{-C\sqrt N}e−CN​ loss which is irrelevant in the laser limit. The two sides are an identity up to Stirling factors: the ∏jAj−Aj\prod_j A_j^{-A_j}∏j​Aj−Aj​​ entropy factor of globalRate is exactly the exponential growth rate of the multinomial coefficient, and the aiaiai−aia_i^{a_i} a_i^{-a_i}aiai​​ai−ai​​ pair inside globalRate cancels on the diagonal b=ab = ab=a.

This is the profile-parametric form of the published fixed-witness statement mme_stothers_fixed_profile_rate_below_marginal_multinomial, which is recovered verbatim by taking β=(98,1862,73075,1023050,3626000,98000,2156000,13720000,21560000,38710000)\beta = (98, 1862, 73075, 1023050, 3626000, 98000, 2156000, 13720000, 21560000, 38710000)β=(98,1862,73075,1023050,3626000,98000,2156000,13720000,21560000,38710000), D=97942072D = 97942072D=97942072 and M=(5822170,31951724,86288150,106906100,56252000,6358100,244150,3724,98)M = (5822170, 31951724, 86288150, 106906100, 56252000, 6358100, 244150, 3724, 98)M=(5822170,31951724,86288150,106906100,56252000,6358100,244150,3724,98). No numerical property of that witness is used: the proof needs only positivity of the ten counts, and it is therefore reusable at every profile appearing in the general Theorem 5.3 optimization (and at the different profiles of the DWZ and More-Asymmetry fourth-power tables).

Preamble
import Mathlib.Analysis.SpecialFunctions.Exp
import Mathlib.Analysis.SpecialFunctions.Sqrt
import Mathlib.Data.Nat.Choose.Multinomial
import Definitions.Def_mme_stothers_fourth_data

open MME BigOperators Filter

set_option autoImplicit false
Formal statement
theorem mme_stothers_general_profile_rate_below_marginal_multinomial
    (tau : ℝ) (a : Fin 10 → ℝ) (base : Fin 10 → ℕ) (marg : Fin 9 → ℕ) (D : ℕ)
    (hbase : ∀ r, 0 < base r)
    (hD : D = ∑ r : Fin 10, MME.StothersFourth.classMultiplicity r * base r)
    (ha : ∀ i, a i = (base i : ℝ) / (D : ℝ))
    (hmarg : ∀ j, (marg j : ℝ) =
      MME.StothersFourth.Q (fun i ↦ (base i : ℝ)) j) :
    ∃ C : ℝ, 0 ≤ C ∧
      ∀ᶠ m : ℕ in Filter.atTop,
        (MME.StothersFourth.globalRate 6 tau a a) ^ (3 * D * m) *
            Real.exp (-C * Real.sqrt (((3 * D * m + 1 : ℕ) : ℝ))) ≤
          (Nat.multinomial Finset.univ (fun j : Fin 9 ↦ marg j * m) : ℝ) *
            (∏ r : Fin 10,
              (MME.StothersFourth.classValue 6 tau r) ^
                (MME.StothersFourth.classMultiplicity r * (base r * m))) := by
  sorry
Source
A. M. Davie and A. J. Stothers, Improved Bound for Complexity of Matrix Multiplication, Proceedings of the Royal Society of Edinburgh A 143(2), 2013, Section 3 (Lemma 3.3 and Equations (3.2)-(3.4)) and Section 5, Equation (5.3); https://www.maths.ed.ac.uk/~sandy/a11164.pdf.

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