BanditRLlib
Lean gate passed before this site build; local proof declarations are shown as compiled.

Lean module · EXP3

BanditRLProof.Exp3RealizedDeviationTail

# Finite-horizon realized EXP3 concentration This module shifts the generated predictable realized-deviation process by one time step so that its deterministic zero initial value and every actual round fit Mathlib's conditional sub-Gaussian sum theorem. The resulting public theorem bounds the full finite-horizon realized-minus-exploration-mixed loss sum under a probability environment prior.

Module map

Declarations
11
Placeholders
0

Imports

BanditRLProof.Exp3RealizedConcentration

Imported by

BanditRLProof, BanditRLProof.Exp3ComparatorConfidence, BanditRLProof.Exp3RealizedConfidence

Declarations

Open an item to read its exact compact statement and source link. Detailed teaching notes are linked when registered.

theorem BanditRLProof.Exp3.sampledPredictableSelectedDeviation_zero_hasCondSubgaussianMGF Compiled

No declaration docstring is present; use the chapter context and exact statement below.

theorem sampledPredictableSelectedDeviation_zero_hasCondSubgaussianMGF {Env : Type u} {Action : Type v} [MeasurableSpace Env] [StandardBorelSpace Env] [Nonempty Env] [MeasurableSpace Action] [MeasurableSingletonClass Action] [StandardBorelSpace Action] [Nonempty Action] [DecidableEq Action] (prior : Measure Env) [IsFiniteMeasure prior] (arms : Finset Action) (harms : arms.Nonempty) (eta gamma : Real) (hgamma_nonneg : 0 <= gamma) (hgamma_le_one : gamma <= 1) (loss : PredictableLossVector Env Action) : let mu := prior ⊗ₘ sampledImportanceWeightedTrajectoryKernel arms harms eta gamma hgamma_nonneg hgamma_le_one loss.environment ProbabilityTheory.HasCondSubgaussianMGF ((inferInstance : MeasurableSpace Env).comap (fun sample : Env × ((k : Nat) -> Action × Real) => sample.1)) measurable_fst.comap_le (sampledTrajectorySelectedDeviationAt arms eta gamma loss 0) (Concentration.intervalVarianceProxy 0 1) mu
theorem BanditRLProof.Exp3.sampledPredictableRealizedDeviation_zero_hasCondSubgaussianMGF Compiled

No declaration docstring is present; use the chapter context and exact statement below.

theorem sampledPredictableRealizedDeviation_zero_hasCondSubgaussianMGF {Env : Type u} {Action : Type v} [MeasurableSpace Env] [StandardBorelSpace Env] [Nonempty Env] [MeasurableSpace Action] [MeasurableSingletonClass Action] [StandardBorelSpace Action] [Nonempty Action] [DecidableEq Action] (prior : Measure Env) [IsFiniteMeasure prior] (arms : Finset Action) (harms : arms.Nonempty) (eta gamma : Real) (hgamma_nonneg : 0 <= gamma) (hgamma_le_one : gamma <= 1) (loss : PredictableLossVector Env Action) : let mu := prior ⊗ₘ sampledImportanceWeightedTrajectoryKernel arms harms eta gamma hgamma_nonneg hgamma_le_one loss.environment ProbabilityTheory.HasCondSubgaussianMGF ((inferInstance : MeasurableSpace Env).comap (fun sample : Env × ((k : Nat) -> Action × Real) => sample.1)) measurable_fst.comap_le (sampledTrajectoryRealizedDeviationAt arms eta gamma loss 0) (Concentration.intervalVarianceProxy 0 1) mu
def BanditRLProof.Exp3.sampledPredictableDeviationFiltration Compiled

No declaration docstring is present; use the chapter context and exact statement below.

def sampledPredictableDeviationFiltration (Env : Type u) (Action : Type v) [MeasurableSpace Env] [MeasurableSpace Action] : Filtration Nat (inferInstance : MeasurableSpace (Env × ((k : Nat) -> Action × Real))) where
theorem BanditRLProof.Exp3.sampledPredictableDeviationFiltration_zero Compiled

No declaration docstring is present; use the chapter context and exact statement below.

theorem sampledPredictableDeviationFiltration_zero (Env : Type u) (Action : Type v) [MeasurableSpace Env] [MeasurableSpace Action] : sampledPredictableDeviationFiltration Env Action 0 = (inferInstance : MeasurableSpace Env).comap (fun sample : Env × ((k : Nat) -> Action × Real) => sample.1)
theorem BanditRLProof.Exp3.sampledPredictableDeviationFiltration_succ Compiled

No declaration docstring is present; use the chapter context and exact statement below.

theorem sampledPredictableDeviationFiltration_succ (Env : Type u) (Action : Type v) [MeasurableSpace Env] [MeasurableSpace Action] (i : Nat) : sampledPredictableDeviationFiltration Env Action (i + 1) = (inferInstance : MeasurableSpace (Env × History.FinitePairHistory Action Real i)).comap (fun sample : Env × ((k : Nat) -> Action × Real) => (sample.1, Preorder.frestrictLe i sample.2))
def BanditRLProof.Exp3.sampledPredictableRealizedDeviationProcess Compiled

No declaration docstring is present; use the chapter context and exact statement below.

noncomputable def sampledPredictableRealizedDeviationProcess {Env : Type u} {Action : Type v} [MeasurableSpace Env] [MeasurableSpace Action] [DecidableEq Action] (arms : Finset Action) (eta gamma : Real) (loss : PredictableLossVector Env Action) : Nat -> Env × ((k : Nat) -> Action × Real) -> Real | 0, _sample => 0 | i + 1, sample => sampledTrajectoryRealizedDeviationAt arms eta gamma loss i sample noncomputable def sampledPredictableRealizedDeviationProxy : Nat -> NNReal | 0 => 0 | _i + 1 => Concentration.intervalVarianceProxy 0 1 theorem sampledPredictableRealizedDeviationProcess_stronglyAdapted {Env : Type u} {Action : Type v} [MeasurableSpace Env] [MeasurableSpace Action] [MeasurableSingletonClass Action] [DecidableEq Action] (arms : Finset Action) (harms : arms.Nonempty) (eta gamma : Real) (hgamma_nonneg : 0 <= gamma) (hgamma_le_one : gamma <= 1) (loss : PredictableLossVector Env Action) : StronglyAdapted (sampledPredictableDeviationFiltration Env Action) (sampledPredictableRealizedDeviationProcess arms eta gamma loss)
def BanditRLProof.Exp3.sampledPredictableRealizedDeviationProxy Compiled

No declaration docstring is present; use the chapter context and exact statement below.

noncomputable def sampledPredictableRealizedDeviationProxy : Nat -> NNReal | 0 => 0 | _i + 1 => Concentration.intervalVarianceProxy 0 1 theorem sampledPredictableRealizedDeviationProcess_stronglyAdapted {Env : Type u} {Action : Type v} [MeasurableSpace Env] [MeasurableSpace Action] [MeasurableSingletonClass Action] [DecidableEq Action] (arms : Finset Action) (harms : arms.Nonempty) (eta gamma : Real) (hgamma_nonneg : 0 <= gamma) (hgamma_le_one : gamma <= 1) (loss : PredictableLossVector Env Action) : StronglyAdapted (sampledPredictableDeviationFiltration Env Action) (sampledPredictableRealizedDeviationProcess arms eta gamma loss)
theorem BanditRLProof.Exp3.sampledPredictableRealizedDeviationProcess_stronglyAdapted Compiled

No declaration docstring is present; use the chapter context and exact statement below.

theorem sampledPredictableRealizedDeviationProcess_stronglyAdapted {Env : Type u} {Action : Type v} [MeasurableSpace Env] [MeasurableSpace Action] [MeasurableSingletonClass Action] [DecidableEq Action] (arms : Finset Action) (harms : arms.Nonempty) (eta gamma : Real) (hgamma_nonneg : 0 <= gamma) (hgamma_le_one : gamma <= 1) (loss : PredictableLossVector Env Action) : StronglyAdapted (sampledPredictableDeviationFiltration Env Action) (sampledPredictableRealizedDeviationProcess arms eta gamma loss)
theorem BanditRLProof.Exp3.sampledPredictableRealizedDeviationProcess_sum_range_succ Compiled

No declaration docstring is present; use the chapter context and exact statement below.

theorem sampledPredictableRealizedDeviationProcess_sum_range_succ {Env : Type u} {Action : Type v} [MeasurableSpace Env] [MeasurableSpace Action] [DecidableEq Action] (arms : Finset Action) (eta gamma : Real) (loss : PredictableLossVector Env Action) (horizon : Nat) (sample : Env × ((k : Nat) -> Action × Real)) : (Finset.range (horizon + 1)).sum (fun i => sampledPredictableRealizedDeviationProcess arms eta gamma loss i sample) = (Finset.range horizon).sum (fun i => sampledTrajectoryRealizedDeviationAt arms eta gamma loss i sample)
theorem BanditRLProof.Exp3.sampledPredictableRealizedDeviationProxy_sum_range_succ Compiled

No declaration docstring is present; use the chapter context and exact statement below.

theorem sampledPredictableRealizedDeviationProxy_sum_range_succ (horizon : Nat) : (Finset.range (horizon + 1)).sum sampledPredictableRealizedDeviationProxy = (horizon : NNReal) * Concentration.intervalVarianceProxy 0 1
theorem BanditRLProof.Exp3.sampledPredictableRealizedDeviation_sum_tail_ennreal Compiled

Finite-horizon one-sided concentration for the realized predictable EXP3 deviation from its exploration-mixed conditional mean.

theorem sampledPredictableRealizedDeviation_sum_tail_ennreal {Env : Type u} {Action : Type v} [MeasurableSpace Env] [StandardBorelSpace Env] [Nonempty Env] [MeasurableSpace Action] [MeasurableSingletonClass Action] [StandardBorelSpace Action] [Nonempty Action] [DecidableEq Action] (prior : Measure Env) [IsProbabilityMeasure prior] (arms : Finset Action) (harms : arms.Nonempty) (eta gamma : Real) (hgamma_nonneg : 0 <= gamma) (hgamma_le_one : gamma <= 1) (loss : PredictableLossVector Env Action) (horizon : Nat) {eps : Real} (heps : 0 <= eps) : let mu := prior ⊗ₘ sampledImportanceWeightedTrajectoryKernel arms harms eta gamma hgamma_nonneg hgamma_le_one loss.environment mu {sample | eps <= (Finset.range horizon).sum (fun i => sampledTrajectoryRealizedDeviationAt arms eta gamma loss i sample)} <= ENNReal.ofReal (Real.exp (-eps ^ 2 / (2 * ((((horizon : NNReal) * Concentration.intervalVarianceProxy 0 1 : NNReal)) : Real))))