Estimation.DTR.MeanZero
Roll-up for the stagewise conditioning lemmas and the headline mean-zero theorem for the sequential doubly robust score.
Headline 1 core · 0 supporting Assembles the two stagewise cancellation lemmas into the headline theorem seqDR_mean_zero. ★ seqDR_mean_zero
Assembles the two stagewise cancellation lemmas into the headline theorem
seqDR_mean_zero. The proof first centers the score on the source probability
space and then transports the integral to the observed-data law.
Mean zero of the sequential doubly robust score. Under the two-stage DTR backdoor assumptions — sequential exchangeability, consistency, stagewise positivity, and integrability of every counterfactual outcome, uniform two-stage strict overlap: the target-regime propensity at each stage lies almost surely in [ε, 1-ε] for some ε in (0, 1/2], and a finite second moment for the observed factual outcome, then the sequential doubly robust influence function ψ_seqDR has expectation zero under the observed two-stage data law.
Formal statement
Proof (Lean source)
StageOne 1 core · 1 supporting Completes the stagewise ingredients for sequential-score centering. ★ cond_exp_residual_zero_stage1
Completes the stagewise ingredients for sequential-score centering. It proves conditional mean-zero of the stage-one outcome residual and identifies the integral of the stage-zero truth regression with the target value.
For a potential-outcome system, treatment space, and two-stage covariate spaces, a dynamic-treatment-regime estimation system, an overlap level, strict overlap, the sequential back-door assumptions, and square-integrability of the factual outcome, the stage-one treatment-weighted outcome residual has conditional mean zero almost surely given the stage-one history.
Formal statement
Proof (Lean source)
1 supporting declaration (lemmas, instances)
-
theta_zero_factualS₀_integrallemma — The DTR estimand θ₀ = E[Y(dbar)] lifts to an integral against P.μ: under DTR backdoor assumptions, θ₀ = ∫ ω, μ₀_val(factualS 0 ω) ∂P.μ, since μ₀_val ∘ factualS 0 is the σ(historyBundle 0)-conditional expectation of Y_of dbar and P.μ is a probability measure.hypothesesconclusionS.θ₀ = ∫ ω, S.μ₀_val (S.toPOLongitudinalPathSystem.factualS ⟨0, by decide⟩ ω) ∂P.μProof (Lean source)
lemma theta_zero_factualS₀_integral (S : DTREstimationSystem P δ γ) (hA : S.toPOLongitudinalPathSystem.Assumptions) : S.θ₀ = ∫ ω, S.μ₀_val (S.toPOLongitudinalPathSystem.factualS ⟨0, by decide⟩ ω) ∂P.μ := by unfold DTREstimationSystem.θ₀ treatmentPathMean calc ∫ ω, S.toPOLongitudinalPathSystem.Y_of S.dbar ω ∂P.μ = ∫ ω, (S.toPOLongitudinalPathSystem.historyBundle 0 (by decide)).condExpGiven (S.toPOLongitudinalPathSystem.Y_of S.dbar) P.μ ω ∂P.μ := by exact (MeasureTheory.integral_condExp (S.toPOLongitudinalPathSystem.historyBundle 0 (by decide)).sigma_le).symm _ = ∫ ω, S.μ₀_val (S.toPOLongitudinalPathSystem.factualS ⟨0, by decide⟩ ω) ∂P.μ := MeasureTheory.integral_congr_ae (S.μ₀_compat hA)
StageZero 1 core · 3 supporting Develops the stage-zero half of the sequential-score cancellation argument. ★ cond_exp_residual_zero_stage0
Develops the stage-zero half of the sequential-score cancellation argument. It
proves measurability of ψ_seqDR, nondegeneracy of both stagewise propensity
scores, and the conditional mean-zero identity for the stage-zero regression
increment.
For a potential-outcome system, treatment space, and two-stage covariate spaces, a dynamic-treatment-regime estimation system, an overlap level, strict overlap, the sequential back-door assumptions, and square-integrability of the factual outcome, the stage-zero treatment-weighted regression increment has conditional mean zero almost surely given the stage-zero history.
Formal statement
Proof (Lean source)
3 supporting declarations (lemmas, instances)
-
measurable_ψ_seqDRlemma — Measurability of the sequential DR influence function on the data tuple (s₀, d₀, s₁, d₁, y) : γ 0 × δ × γ 1 × δ × ℝ. Decomposes into Measurable.add/Measurable.mul/Measurable.div chained against the projections, the indicator functions indEq, and the value-space nuisance functions stored in S.η₀.hypothesesconclusionMeasurable S.ψ_seqDRProof (Lean source)
@[fun_prop] lemma measurable_ψ_seqDR (S : DTREstimationSystem P δ γ) : Measurable S.ψ_seqDR := by exact S.measurable_seqDRMomentFunctional S.η₀ S.θ₀ -
propScore_ne_zero_stage0lemma — Stage-0 propensity is a.e. nonzero under the DTR backdoor assumptions. The conditional indicator μ[1{D₀ = dbar 0} | σ(historyBundle 0)] is identified via e₀_compat with e₀_val ∘ factualS 0, and e₀_val > 0 pointwise on γ 0.hypothesesconclusion∀ᵐ ω ∂P.μ, (S.toPOLongitudinalPathSystem.historyBundle 0 (by decide)).condExpGiven ((S.toPOLongitudinalPathSystem.dVar ⟨0, by decide⟩).indicator (S.dbar ⟨0, by decide⟩)) P.μ ω≠ 0Proof (Lean source)
lemma propScore_ne_zero_stage0 (S : DTREstimationSystem P δ γ) (hA : S.toPOLongitudinalPathSystem.Assumptions) : ∀ᵐ ω ∂P.μ, (S.toPOLongitudinalPathSystem.historyBundle 0 (by decide)).condExpGiven ((S.toPOLongitudinalPathSystem.dVar ⟨0, by decide⟩).indicator (S.dbar ⟨0, by decide⟩)) P.μ ω ≠ 0 := by filter_upwards [hA.overlap S.dbar ⟨0, by decide⟩] with ω hω exact ne_of_gt hω -
propScore_ne_zero_stage1lemma — Stage-1 propensity is a.e. nonzero under the DTR backdoor assumptions. Analogous to propScore_ne_zero_stage0 via e₁_compat and e₁_pos.hypothesesconclusion∀ᵐ ω ∂P.μ, (S.toPOLongitudinalPathSystem.historyBundle 1 (by decide)).condExpGiven ((S.toPOLongitudinalPathSystem.dVar ⟨1, by decide⟩).indicator (S.dbar ⟨1, by decide⟩)) P.μ ω≠ 0Proof (Lean source)
lemma propScore_ne_zero_stage1 (S : DTREstimationSystem P δ γ) (hA : S.toPOLongitudinalPathSystem.Assumptions) : ∀ᵐ ω ∂P.μ, (S.toPOLongitudinalPathSystem.historyBundle 1 (by decide)).condExpGiven ((S.toPOLongitudinalPathSystem.dVar ⟨1, by decide⟩).indicator (S.dbar ⟨1, by decide⟩)) P.μ ω ≠ 0 := by filter_upwards [hA.overlap S.dbar ⟨1, by decide⟩] with ω hω exact ne_of_gt hω