theorem
Determinize.Proof.RewardModel.integrable_iSup_of_bound
(μ : ℕ → MeasureTheory.Measure ℝ)
(mono : Monotone μ)
(f : ℝ → ℝ)
(hf : Measurable f)
(bound : ENNReal)
(finite : bound < ⊤)
(bounded : ∀ (n : ℕ), ∫⁻ (x : ℝ), ENNReal.ofReal ‖f x‖ ∂μ n ≤ bound)
:
MeasureTheory.Integrable f (⨆ (n : ℕ), μ n)
- first_bound (i : Fin model.size) : (match model.kind i with | Spec.FiniteModel.StateKind.returned b => |b| | Spec.FiniteModel.StateKind.rejected => 0 | Spec.FiniteModel.StateKind.transient => (List.map (fun (e : Spec.RewardModel.Edge model.size) => e.probability * (self.first e.target + |e.reward|)) (model.edges i)).sum) ≤ self.first i
- second_bound (i : Fin model.size) : (match model.kind i with | Spec.FiniteModel.StateKind.returned b => b ^ 2 | Spec.FiniteModel.StateKind.rejected => 0 | Spec.FiniteModel.StateKind.transient => (List.map (fun (e : Spec.RewardModel.Edge model.size) => e.probability * (self.second e.target + 2 * |e.reward| * self.first e.target + e.reward ^ 2)) (model.edges i)).sum) ≤ self.second i
Instances For
theorem
Determinize.Proof.RewardModel.outputWithin_mass_le_one
(model : Spec.RewardModel.Model)
(n : ℕ)
(i : Fin model.size)
:
theorem
Determinize.Proof.RewardModel.lintegral_cost_outputWithin_le
(model : Spec.RewardModel.Model)
(bounds : MomentBounds model)
(n : ℕ)
(i : Fin model.size)
:
∫⁻ (x : ℝ), Determinize.Proof.RewardModel.firstCost✝ x ∂model.outputWithin n i ≤ ENNReal.ofReal ↑(bounds.first i) ∧ ∫⁻ (x : ℝ), Determinize.Proof.RewardModel.secondCost✝ x ∂model.outputWithin n i ≤ ENNReal.ofReal ↑(bounds.second i)
theorem
Determinize.Proof.RewardModel.integrable_outputAt
(model : Spec.RewardModel.Model)
(bounds : MomentBounds model)
(i : Fin model.size)
:
MeasureTheory.Integrable (fun (x : ℝ) => x) (model.outputAt i) ∧ MeasureTheory.Integrable (fun (x : ℝ) => x ^ 2) (model.outputAt i)