Avg Over Le Avg Over Add Total Variation Distance
ProvedLIDT.MIPStarRE.LDT.avgOver_le_avgOver_add_totalVariationDistanceBounded expectations change by at most total variation. Let μ and ν be probability distributions and let f : α → Error take values in [0,1]. Then replacing μ by ν can increase the average by no more than the total variation distance between the two distributions.
This is the distribution-replacement estimate used in the pasting bounds comparing distinct-tuple sampling with uniform sampling.
Formalization credit: Sirui Lu, Ruixuan Deng, Yanqiao Zhu, and Zhengfeng Ji (arXiv:2609.19814). Mathematical source: Zhengfeng Ji, Anand Natarajan, Thomas Vidick, John Wright, and Henry Yuen (arXiv:2009.12982).
Formalization Note. The source declaration is MIPStarRE.LDT.avgOver_le_avgOver_add_totalVariationDistance; its Lean statement gives the exact quantifier order and all boundary hypotheses.
import Definitions.Def_LIDT_MIPStarRE_LDT_Basic_Distribution import Definitions.Def_LIDT_MIPStarRE_LDT_Basic_DistributionUniformSums import Definitions.Def_LIDT_MIPStarRE_LDT_Basic_PMFAverages import Definitions.Def_LIDT_MIPStarRE_LDT_Basic_PMFUniformAverages import Definitions.Def_LIDT_MIPStarRE_LDT_Basic_ParametersBase import Definitions.Def_LIDT_MIPStarRE_Quantum_FiniteMatrix_Basic import Definitions.Def_LIDT_MIPStarRE_Quantum_FiniteMatrix_BlockDiagonal import Definitions.Def_LIDT_MIPStarRE_Quantum_FiniteMatrix_NormalizedTrace import Definitions.Def_LIDT_MIPStarRE_Quantum_FiniteMatrix_Order import Definitions.Def_LIDT_MIPStarRE_Quantum_FiniteMatrix_TracePairing import Mathlib import Mathlib.Analysis.MeanInequalitiesPow import Mathlib.Probability.ProbabilityMassFunction.Integrals import Mathlib.Probability.ProbabilityMassFunction.Monad open LIDT /-! # Average lemmas for finite-support distributions This module contains the finite-sum and uniform-average lemmas for the project `Distribution` type. The primitive distribution definitions, uniform distributions, and total variation distance live in `MIPStarRE.LDT.Basic.Distribution`. ## Main declarations The main results are the algebraic rules for `avgOver`, scalar and operator-valued forms of the module-valued finite-sum rules for uniform distributions, and the comparison between finite uniform averages and Mathlib uniform probability mass functions. ## References These are formalization-internal finite probability lemmas for the low individual degree test development. -/ @[expose] public section open scoped BigOperators MatrixOrder _root_.Matrix ComplexOrder open LIDT.MIPStarRE.LDT /-! ### Averaging infrastructure Linearity proofs use `Distribution.weightedSumLinearMap`; order and support estimates use Mathlib's `Finset.sum` API. -/
theorem LIDT.MIPStarRE.LDT.avgOver_le_avgOver_add_totalVariationDistance {α : Type*} [DecidableEq α]
(μ ν : _root_.LIDT.MIPStarRE.LDT.Distribution α) (hμ : μ.IsProbability) (hν : ν.IsProbability)
(f : α → _root_.LIDT.MIPStarRE.LDT.Error)
(hf_nonneg : ∀ a, 0 ≤ f a)
(hf_le_one : ∀ a, f a ≤ 1) :
_root_.LIDT.MIPStarRE.LDT.avgOver ν f ≤ _root_.LIDT.MIPStarRE.LDT.avgOver μ f + _root_.LIDT.MIPStarRE.LDT.totalVariationDistance μ ν := by sorry