Localized charge is paid by the kernel Gini increment
ProvedDerivativeChainContinuation.near_collision_kernel_chargecontinuation-20261026derivative-chainsgeometric-budgets
Let be finite weights and a nonnegative row-stochastic barycentric kernel from complex points to . Write for its mean edge cost and for its conditional pair distance. Then
The weights need not be normalized.
Preamble
import Mathlib import Definitions.Def_DerivativeChainContinuation_FiniteNormMean import Definitions.Def_DerivativeChainContinuation_NearCollision import Definitions.Def_DerivativeChainPrivate_foundations open DerivativeChainPrivate set_option autoImplicit false set_option backward.isDefEq.respectTransparency false open Polynomial Matrix open scoped BigOperators set_option maxHeartbeats 1000000
Formal statement
theorem DerivativeChainContinuation.near_collision_kernel_charge {ι κ : Type*} [Fintype ι] [Fintype κ]
(w : ι → ℝ) (K : ι → κ → ℝ) (x : ι → ℂ) (y : κ → ℂ)
(hw : ∀ i, 0 ≤ w i) (hK : ∀ i a, 0 ≤ K i a)
(hKs : ∀ i, ∑ a, K i a=1) (hKm : ∀ i, ∑ a, (K i a:ℂ)*y a=x i) :
nearCollisionCharge w x (kernelCost K x y) ≤
(∑ i, ∑ j, w i*w j*kernelPairDistance K y i j)-finiteGini w x := by sorrySource