Expected reward in matrix form:
ProvedTPG.integral_Y_trajLawcausal-inferencemarkov-chainnonstationaritypolicy-gradienttreatment-effect
Setting. is a finite state space, actions are , time is (the source's time ). The environment has transition matrices , initial distribution and reward laws on . For a sequence of treatment probabilities, and denote the law of the trajectory when , the actions are independent with , , and the rewards are conditionally independent with (Env.trajLaw T q); the Bernoulli policy is .
Lemma. Suppose for all and for all . Then for every time ,
where , and .
Taking gives the formula behind the value function ; other choices of give the hypothetical truncated policies and the laws conditioned on one action.
Preamble
import Mathlib import Definitions.Def_TPG_Model import Definitions.Def_TPG_Markov open MeasureTheory ProbabilityTheory open scoped BigOperators ENNReal
Formal statement
namespace TPG
theorem integral_Y_trajLaw {S : Type} [Fintype S] [DecidableEq S] [MeasurableSpace S] [MeasurableSingletonClass S]
(E : Env S) (T : ℕ) (q : ℕ → ℝ) (hq : ∀ t : ℕ, 0 ≤ q t ∧ q t ≤ 1) (M : ℝ) (hM : E.RewardBound M)
(t : Fin T) :
∫ ω, Y t ω ∂(E.trajLaw T q) = dotProduct (E.stateVec q (t : ℕ)) (E.rMix (q (t : ℕ))) := by sorry
end TPGSource
Johari, Peng, Xing, Estimation of Treatment Effects Under Nonstationarity via the Truncated Policy Gradient Estimator, revised manuscript of October 2026 (revision of arXiv:2506.05308v3, https://arxiv.org/abs/2506.05308); numbering follows the revised manuscript. Auxiliary statement, not a numbered result of the paper: it is the identity E_{L_θ^t}[Y_t] = ρ^⊤ P_θ^{1→t} r_θ of Appendix G.1, p. 42, for a time-varying sequence of treatment probabilities.