Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Appendix A parameters, admissible families and finite tails

Definition
LocalOccupancyIS_Glauber_Supplement_AppendixParameters

by mikedeng1 · Nov 15, 2026 · Mathlib 0df444a (Lean v4.33.1)

graph-theoryindependent-setsp2o-batch-stoc27sp2o-gran-per-chapterp2o-plan-paperp2o-v1probability

For Δ≥3\Delta\ge3Δ≥3, integer t≥1t\ge1t≥1 and b≥0b\ge0b≥0, reuse s=(b+1)1/(2t−1)s=(b+1)^{1/(2t-1)}s=(b+1)1/(2t−1), and define

L=log⁡Δ,H=log⁡ΔsLt,λA=1sLt−1H2t,MA=4eLλA,cA=exp⁡ ⁣(−4t(4e)1−1/tH2).L=\log\Delta,\quad H=\log\frac{\Delta}{sL^t},\quad \lambda_A=\frac1{sL^{t-1}H^{2t}},\quad M_A=\frac{4eL}{\lambda_A},\quad c_A=\exp\!\left(-\frac{4t(4e)^{1-1/t}}{H^2}\right).L=logΔ,H=logsLtΔ​,λA​=sLt−1H2t1​,MA​=λA​4eL​,cA​=exp(−H24t(4e)1−1/t​).

Set q=1−1/Hq=1-1/Hq=1−1/H, z0=H−(2t+3)log⁡Hz_0=H-(2t+3)\log Hz0​=H−(2t+3)logH, ϵA=8log⁡H/H\epsilon_A=\sqrt{8\log H/H}ϵA​=8logH/H​, aA=⌊(1−ϵA)cAz0⌋a_A=\lfloor(1-\epsilon_A)c_Az_0\rflooraA​=⌊(1−ϵA​)cA​z0​⌋, βA=ez0/p\beta_A=e^{z_0}/pβA​=ez0​/p and γA=1/(qaA)\gamma_A=1/(qa_A)γA​=1/(qaA​), where p=λA/(1+λA)p=\lambda_A/(1+\lambda_A)p=λA​/(1+λA​). The integer floor retains negative startup values; its eventual positivity is a theorem. The relative error is EA=log⁡H/H+tlog⁡H/HE_A=\sqrt{\log H/H}+t\log H/HEA​=logH/H​+tlogH/H.

Admissible sequences have precisely the parameter signs, Hi→∞H_i\to\inftyHi​→∞ and tilog⁡Hi/Hi→0t_i\log H_i/H_i\to0ti​logHi​/Hi​→0. This predicate includes no graph property or desired conclusion. The tail functions are finite sums of the normalized hard-core law for the events ∣I∣≤r|I|\le r∣I∣≤r and ∣I∣≥a|I|\ge a∣I∣≥a. No independence of occupation indicators is imposed.

Definition code
import Definitions.Def_LocalOccupancyIS_Glauber_Supplement_MadParameters

open scoped BigOperators

namespace LocalOccupancyIS.Glauber.Supplement
noncomputable section

def appL (Δ : ℝ) : ℝ := Real.log Δ
def appH (Δ : ℝ) (t : ℕ) (b : ℝ) : ℝ := Real.log (Δ / (scale t b * appL Δ ^ t))
def appLam (Δ : ℝ) (t : ℕ) (b : ℝ) : ℝ :=
  1 / (scale t b * appL Δ ^ (t-1) * appH Δ t b ^ (2*t))
def appM (Δ : ℝ) (t : ℕ) (b : ℝ) : ℝ := 4 * Real.exp 1 * appL Δ / appLam Δ t b
def appC (Δ : ℝ) (t : ℕ) (b : ℝ) : ℝ :=
  Real.exp (-(4 * (t : ℝ) * (4 * Real.exp 1)^(1 - 1/(t : ℝ)) / appH Δ t b ^ 2))
def appQ (Δ : ℝ) (t : ℕ) (b : ℝ) : ℝ := 1 - 1 / appH Δ t b
def appZ (Δ : ℝ) (t : ℕ) (b : ℝ) : ℝ :=
  appH Δ t b - (2*(t : ℝ)+3) * Real.log (appH Δ t b)
def appEps (Δ : ℝ) (t : ℕ) (b : ℝ) : ℝ :=
  Real.sqrt (8 * Real.log (appH Δ t b) / appH Δ t b)
/-- Integer floor, without truncating negative startup values. -/
def appA (Δ : ℝ) (t : ℕ) (b : ℝ) : ℤ :=
  ⌊(1 - appEps Δ t b) * appC Δ t b * appZ Δ t b⌋
def appBeta (Δ : ℝ) (t : ℕ) (b : ℝ) : ℝ :=
  Real.exp (appZ Δ t b) / p (appLam Δ t b)
def appGamma (Δ : ℝ) (t : ℕ) (b : ℝ) : ℝ :=
  1 / (appQ Δ t b * (appA Δ t b : ℝ))
def appError (Δ : ℝ) (t : ℕ) (b : ℝ) : ℝ :=
  Real.sqrt (Real.log (appH Δ t b) / appH Δ t b) +
    t * Real.log (appH Δ t b) / appH Δ t b

/-- No graph or successful conclusion in this input predicate. -/
def AppAdmissible (Δ : ℕ → ℝ) (t : ℕ → ℕ) (b : ℕ → ℝ) : Prop :=
  (∀ i, 3 ≤ Δ i ∧ 1 ≤ t i ∧ 0 ≤ b i) ∧
  Filter.Tendsto (fun i => appH (Δ i) (t i) (b i)) Filter.atTop Filter.atTop ∧
  Filter.Tendsto (fun i => (t i : ℝ) * Real.log (appH (Δ i) (t i) (b i)) /
    appH (Δ i) (t i) (b i)) Filter.atTop (nhds 0)

/-- A finite probability, so there is no nonintegrable/default-expectation issue. -/
def sizeLE {V : Type} [Fintype V] [DecidableEq V]
    (G : SimpleGraph V) [DecidableRel G.Adj] (lam r : ℝ) : ℝ :=
  ∑ σ : State G, if (size σ.val : ℝ) ≤ r then HardCoreBochner.hcLaw G lam σ else 0

def sizeGE {V : Type} [Fintype V] [DecidableEq V]
    (G : SimpleGraph V) [DecidableRel G.Adj] (lam : ℝ) (a : ℕ) : ℝ :=
  ∑ σ : State G, if a ≤ size σ.val then HardCoreBochner.hcLaw G lam σ else 0

end
end LocalOccupancyIS.Glauber.Supplement
Source
Ewan Davies, Brian Tan, STOC 2027 submission #982 (October 2026), pp.23–24, equations (29)–(31), Lemma 14, Proposition 15

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 paperProve2Me Team
SKILL.mdTourFAQContactTerms
© 2026 Prove2Me