Number of infinite clusters

TheoremShift-invariant countably valued functions are a.s. constant under ergodicity#✓fully proved

A measurable function with countable values that is invariant under all shifts is almost surely constant under an ergodic probability measure. See also Mathlib's ErgodicSMul and aeconst_of_forall_preimage_smul_ae_eq, which state ergodicity of a group action through a.e.-invariant sets.

theorem ergodic_ae_const_of_shiftInvariant [IsProbabilityMeasure μ]
    {β : Type*} [Countable β] [MeasurableSpace β] [MeasurableSingletonClass β]
    (herg : IsErgodic (G := G) μ)
    (f : ConfigSpace E → β) (hf : Measurable f)
    (hinv : ∀ (g : G) (ω : ConfigSpace E), f (shift g ω) = f ω) :
    ∃ k : β, μ {ω | f ω = k} = 1
Uses 4 · Used by 1

DefinitionTranslation of the open subgraph as a graph homomorphism#✓fully proved

Translation by g as a graph homomorphism between the open subgraphs of ω and of its shift.

def translateHom (g : Multiplicative (Site d)) (ω : ConfigSpace (Sym2 (Site d))) :
    openSubgraph d ω →g openSubgraph d (shift g ω) where
  toFun x := g • x
  map_rel' {x y} h := (openSubgraph_adj_shift g ω x y).mpr h
Uses 5 · Used by 2

DefinitionInverse translation of the open subgraph#✓fully proved

The inverse of translateHom.

def translateHomInv (g : Multiplicative (Site d)) (ω : ConfigSpace (Sym2 (Site d))) :
    openSubgraph d (shift g ω) →g openSubgraph d ω where
  toFun a := g⁻¹ • a
  map_rel' {a b} h := by
    have h' := (openSubgraph_adj_shift g ω (g⁻¹ • a) (g⁻¹ • b)).mp
    simp only [smul_inv_smul] at h'
    exact h' h
Uses 5 · Used by 2

DefinitionSet of infinite open clusters#✓fully proved

The set of infinite open clusters of ω.

def infiniteClusters (d : ℕ) (ω : ConfigSpace (Sym2 (Site d))) : Set (Set (Site d)) :=
  {C | C.Infinite ∧ ∃ x, C = cluster d ω x}
Uses 3 · Used by 3

DefinitionNumber of infinite open clusters#✓fully proved

The number N(ω) ∈ ℕ∞ of infinite open clusters of ω.

noncomputable def numInfiniteClusters (d : ℕ) (ω : ConfigSpace (Sym2 (Site d))) : ℕ∞ :=
  (infiniteClusters d ω).encard
Uses 3 · Used by 6

TheoremMeasurability of the one-arm event#✓fully proved

The one-arm event crossingEvent d n (the origin is connected to a site outside box d (n - 1)) is measurable.

theorem measurableSet_crossingEvent (n : ℕ) : MeasurableSet (crossingEvent d n)
Uses 8 · Used by 7

DefinitionRepresentative sites of the infinite clusters#✓fully proved

The set of sites x in an infinite cluster that are the Encodable-least site of their cluster (one representative per infinite cluster).

def repSet (d : ℕ) (ω : ConfigSpace (Sym2 (Site d))) : Set (Site d) :=
  {x | (cluster d ω x).Infinite ∧
        ∀ y, Connected d ω x y → Encodable.encode x ≤ Encodable.encode y}
Uses 4 · Used by 1

TheoremNumber of infinite clusters is almost surely constant#✓fully proved

The number of infinite clusters is almost surely constant (Newman–Schulman 1981) for any translation-ergodic probability measure on bond configurations of ℤ^d.

theorem numInfiniteClusters_ae_const
    (μ : Measure (ConfigSpace (Sym2 (Site d)))) [IsProbabilityMeasure μ]
    (herg : IsErgodic (G := Multiplicative (Site d)) μ) :
    ∃ k : ℕ∞, μ {ω | numInfiniteClusters d ω = k} = 1
Uses 16 · Used by 1

DefinitionEvent of at least two infinite open clusters#✓fully proved

The event that there are at least two infinite open clusters.

def atLeastTwoInfinite (d : ℕ) : Set (ConfigSpace (Sym2 (Site d))) :=
  {ω | 2 ≤ numInfiniteClusters d ω}
Uses 3 · Used by 4

DefinitionConfiguration with a finite edge set forced open#✓fully proved

The configuration obtained from ω by opening every edge of F.

noncomputable def forceOpenFinset (F : Finset (Sym2 (Site d)))
    (ω : ConfigSpace (Sym2 (Site d))) : ConfigSpace (Sym2 (Site d)) :=
  fun e => if e ∈ F then true else ω e
Uses 2 · Used by 2

DefinitionEndpoints of a finite edge set#✓fully proved

The endpoints of the edges of F.

def anchorSites (F : Finset (Sym2 (Site d))) : Set (Site d) :=
  {a | ∃ e ∈ F, a ∈ e}
Uses 1

DefinitionConfigurations with \(k\) infinite clusters, two merged by opening \(F\)#✓fully proved

Configurations with exactly k infinite clusters in which opening F connects two distinct infinite clusters.

def MergeWitness (d : ℕ) (F : Finset (Sym2 (Site d))) (k : ℕ∞) :
    Set (ConfigSpace (Sym2 (Site d))) :=
  {ω | numInfiniteClusters d ω = k ∧
    ∃ x y : Site d, (cluster d ω x).Infinite ∧ (cluster d ω y).Infinite ∧
      cluster d ω x ≠ cluster d ω y ∧ Connected d (forceOpenFinset F ω) x y}
Uses 6

DefinitionFinite energy for opening finite edge sets#✓fully proved

Finite energy for opening edges: for every finite edge set F, the law of forceOpenFinset F ω is absolutely continuous with respect to μ.

def HasFiniteEnergyMerge (μ : Measure (ConfigSpace (Sym2 (Site d)))) : Prop :=
  ∀ F : Finset (Sym2 (Site d)), (μ.map (fun ω => forceOpenFinset F ω)) ≪ μ
Uses 3

LemmaZero or one infinite cluster, given ergodicity and \(\mu(N \ge 2) = 0\)#✓fully proved

For an ergodic measure with μ(N ≥ 2) = 0, either a.s. N = 0 or a.s. N = 1.

lemma numInfiniteClusters_zero_or_one
    (μ : Measure (ConfigSpace (Sym2 (Site d)))) [IsProbabilityMeasure μ]
    (herg : IsErgodic (G := Multiplicative (Site d)) μ)
    (hmerge : μ (atLeastTwoInfinite d) = 0) :
    μ {ω | numInfiniteClusters d ω = 0} = 1 ∨ μ {ω | numInfiniteClusters d ω = 1} = 1
Uses 6
14 fully proved

The dependency explorer needs JavaScript and its graph data.

Download the complete graph data.