Inactive-fiber kernels and off-diagonal residual correlations
DefinitionNarrowDNF_HatamiCrossTermscombinatoricsnarrow-dnf-authors-importprobabilitystoc-2027
Let be a local activation family on a finite Boolean cube, and let denote the Bernoulli mass of the labels restricted to . The inactive-fiber kernel is
The code writes this as zero on firing inputs and the reciprocal square on inactive inputs. For , all local masses are positive.
For a real-valued function , let be its Fourier component conditioned on the adaptive partition formed from activations indexed inside . Given a finite retained family , define
The sum uses ordered pairs and discards negative correlations. It is the nonnegative off-diagonal quantity controlled by the corrected residual estimate; the estimate itself is not part of these definitions.
Definition code
import Mathlib.Algebra.BigOperators.Group.Finset.Basic
import Mathlib.Algebra.BigOperators.GroupWithZero.Finset
import Mathlib.Algebra.BigOperators.Ring.Finset
import Mathlib.Algebra.Order.BigOperators.Group.Finset
import Mathlib.Algebra.Order.BigOperators.Ring.Finset
import Mathlib.Analysis.SpecialFunctions.Log.Basic
import Mathlib.Analysis.SpecialFunctions.Pow.Real
import Mathlib.Analysis.SpecialFunctions.Sqrt
import Mathlib.Data.Bool.Basic
import Mathlib.Data.Finset.Card
import Mathlib.Data.Finset.Image
import Mathlib.Data.Finset.Insert
import Mathlib.Data.Finset.Lattice.Fold
import Mathlib.Data.Fintype.BigOperators
import Mathlib.Data.Fintype.Pi
import Mathlib.Data.Real.Basic
import Mathlib.Logic.Equiv.Prod
import Mathlib.Order.Monotone.Basic
import Mathlib.Tactic
import Mathlib.Tactic.FieldSimp
import Mathlib.Tactic.GCongr
import Mathlib.Tactic.Linarith
import Mathlib.Tactic.Positivity
import Mathlib.Tactic.Ring
import Definitions.Def_NarrowDNF_Adaptive
import Definitions.Def_NarrowDNF_HatamiActivation
import Definitions.Def_NarrowDNF_HatamiResidual
import Definitions.Def_NarrowDNF_ProductMeasure
namespace NarrowDNF.HatamiActivation
end NarrowDNF.HatamiActivation
namespace NarrowDNF.HatamiCrossTerms
end NarrowDNF.HatamiCrossTerms
namespace NarrowDNF.HatamiResidual
end NarrowDNF.HatamiResidual
namespace NarrowDNF.ProductMeasure
end NarrowDNF.ProductMeasure
/-!
# Hatami's corrected cross-term estimate
Finite cross-term estimates for the exact restricted-activation conditional
components. The repaired coefficient factor is retained explicitly.
-/
noncomputable section
open scoped BigOperators
namespace NarrowDNF.HatamiCrossTerms
open ProductMeasure HatamiResidual HatamiActivation
attribute [local instance] Classical.propDecidable
variable {ι : Type*} [Fintype ι] [DecidableEq ι]
/-- The inactive overlap-fiber kernel used in Hatami's cross-term sum. -/
def inactiveKernel (p : ℝ) (J : Activation ι) (T : Finset ι) : ℝ :=
expectation p (fun x => if J.fires T x then 0 else 1 / localMass p T x ^ 2)
/-- Nonnegative envelope of the off-diagonal conditional-component correlations. -/
def crossEnvelope (p : ℝ) (J : Activation ι) (f : Cube ι → ℝ)
(A : Finset (Finset ι)) : ℝ :=
∑ S ∈ A, ∑ R ∈ A, if S = R then 0 else
max 0 (expectation p (fun x => approxComponent p J f S x * approxComponent p J f R x))
end NarrowDNF.HatamiCrossTerms
Source