Documentation

LeanPool.Feige.InsertionAnalyticAssembly

Assembly of the analytic insertion edges #

This file performs the last finite bookkeeping step in the chain-insertion argument for Theorem 2.1. Once each genuine edge of the old chain is represented by the two signed-exponential endpoint laws from the local transfer step, and the terminal edge is represented by its one-sided common law, all positivity, monotonicity, and transfer hypotheses required by exists_insertChainPerm_dominates_reveal follow automatically.

def Feige.insertionEdgeScale {n : } (γ : Fin (n + 1)) (σ : Equiv.Perm (Fin n)) (j : ) :

The low-side scale of the old coordinate changed at edge j; it equals 1 at the terminal edge, which changes the distinguished exponential.

Equations
Instances For
    theorem Feige.insertionTheta_eq_zero_of_ge {n : } (γ β : Fin (n + 1)) (σ : Equiv.Perm (Fin n)) {r : } (hr : n + 1 r) :
    insertionTheta γ β σ r = 0

    The zero-extended interpolation sequence is identically zero from its sentinel onward.

    theorem Feige.exists_insertChainPerm_dominates_reveal_of_realizedEdges {n : } (γ β : Fin (n + 1)) (hγpos : ∀ (i : Fin (n + 1)), 0 < γ i) (hγle : ∀ (i : Fin (n + 1)), γ i 1) (hβpos : ∀ (i : Fin (n + 1)), 0 < β i) (hγord : Antitone γ) (σ : Equiv.Perm (Fin n)) {g : Finset (Fin (n + 1))} (hg : Monotone g) (Fs : Fin nList LikelihoodRatio.SignedExpFactor) (hrealInterior : ∀ (j : Fin n), Lemma43.RealizesInsertionEdge (insertionUpperK γ β σ) (insertionOldK γ β σ) (insertionWidth γ β σ) (insertionTheta γ β σ) (↑j) (TransferStein.zPlusLaw (MeasureTheory.volume.withDensity (LikelihoodRatio.finiteSignedExpSumDensity (Fs j))) (γ (σ j).castSucc)) (TransferStein.zMinusLaw (MeasureTheory.volume.withDensity (LikelihoodRatio.finiteSignedExpSumDensity (Fs j))) (β (σ j).castSucc)) (γ (Fin.last n)) (β (Fin.last n))) (μterminal : MeasureTheory.Measure ) [MeasureTheory.IsProbabilityMeasure μterminal] (hterminalSupport : μterminal (Set.Ioi 0) = 0) (hrealTerminal : Lemma43.RealizesInsertionEdge (insertionUpperK γ β σ) (insertionOldK γ β σ) (insertionWidth γ β σ) (insertionTheta γ β σ) n (TransferStein.zPlusLaw μterminal 1) (TransferStein.zMinusLaw μterminal 1) (γ (Fin.last n)) (β (Fin.last n))) :
    ∃ (J : Fin (n + 1)), booleanChainExpectation (fun (i : Fin n) => γ i.castSucc) (fun (i : Fin n) => β i.castSucc) σ (revealedLastPayoff (twoPointHighProbability γ β (Fin.last n)) g) booleanChainExpectation γ β (insertChainPerm σ J) g

    The chain-insertion conclusion after all measure-theoretic endpoint identifications have been exposed as RealizesInsertionEdge hypotheses.

    There is one genuine signed-exponential edge for every old coordinate. The final edge changes the distinguished E₀ from positive to negative; its common law need only be supported on the nonpositive half-line.