Weighted error bound for a truncated certificate DNF
ProvedNarrowDNF.Activation.truncatedDNF_error_boundcombinatoricsnarrow-dnf-authors-importprobabilitystoc-2027
Let be finite, let be an increasing local activation family, and let be monotone and determined by its adaptive atom. Define the active union and its positive certificate . For , let be the monotone DNF whose terms are exactly the certificates with and . For every nonnegative weighting ,
This gives a width-dependent approximation bound without requiring normalized weights.
Preamble
import Mathlib.Algebra.BigOperators.Ring.Finset
import Mathlib.Algebra.Order.BigOperators.Group.Finset
import Mathlib.Data.Bool.Basic
import Mathlib.Data.Finset.Card
import Mathlib.Data.Finset.Image
import Mathlib.Data.Finset.Lattice.Fold
import Mathlib.Data.Fintype.Pi
import Mathlib.Data.Real.Basic
import Mathlib.Order.Monotone.Basic
import Mathlib.Tactic.Linarith
import Mathlib.Tactic.NormNum
import Mathlib.Tactic.Positivity
import Mathlib.Tactic.Ring
import Definitions.Def_NarrowDNF_Adaptive
import Definitions.Def_NarrowDNF_CertificateBounds
import Definitions.Def_NarrowDNF_Certificates
import Theorems.Thm_NarrowDNF_Activation_positive_certificate
import Theorems.Thm_NarrowDNF_Weighted_integer_markov
namespace NarrowDNF
end NarrowDNF
namespace NarrowDNF.Activation
end NarrowDNF.Activation
open NarrowDNF
open NarrowDNF
open NarrowDNF.Activation
open scoped BigOperators
variable {ι : Type*} [Fintype ι] [DecidableEq ι]
Formal statement
theorem NarrowDNF.Activation.truncatedDNF_error_bound {J : Activation ι} (hJ : J.Increasing)
{u : Cube ι → Bool} (hu : Monotone u) (hm : J.Measurable u) (w : ℕ)
(weights : Cube ι → ℝ) (hw : ∀ x, 0 ≤ weights x) :
(∑ x, weights x * (if u x ≠ J.truncatedDNF u w x then 1 else 0)) ≤
(∑ x, weights x * (if w < (J.certificate x).card then 1 else 0)) ∧
(∑ x, weights x * (if w < (J.certificate x).card then 1 else 0)) ≤
(∑ x, weights x * ((J.activeFinset x).card : ℝ)) / (w + 1 : ℝ) := by sorrySource