Fast Regularized Policy Mirror Descent with One-Step TD Updates MDQXTBEK0F

URI	`m2://res/MDQXTBEK0F`
Kind	preprint
Creators	Qipei Chen, Wenye Li, Yule Sun, Ke Wei
Date	2026-09-30
Subjects	arxiv cs.LG, arxiv math.OC
Scope	border

Expressions

Date	2026-09-30
Ids	arXiv 2609.39837

The mentions of the inventory that name this resource. Each name links to the card of its inventory row.

Item	Kind	Where	Source
Three-point inequality	statement	lemma 2.4	arxiv
Unregularized exact TD--PMD	statement	corollary 3.7	arxiv
Weighted value-gap reduction	statement	lemma B.1	arxiv
Five-term decomposition	statement	lemma B.2	arxiv
Bound for the true-\texorpdfstring{$Q$}{Q} drift	statement	lemma B.3	arxiv
Potential recursion	statement	lemma 3.3	arxiv
Conditional bias of the stochastic critic error	statement	lemma 4.4	arxiv
$\epsilon$-accuracy and sample complexity	statement	corollary 4.6	arxiv
Value convergence under arbitrary evaluation distributions	statement	theorem 3.4	arxiv
Bellman fixed points and contractions	statement	lemma 2.1	arxiv
Expected value-gap bound	statement	theorem 4.5	arxiv
Noise-free behavior-weighted value convergence	statement	corollary 4.1	arxiv
Bellman violation and critic comparison	statement	lemma 3.1	arxiv
Exponentially weighted operator drift	statement	lemma B.5	arxiv
Bound for the Markov-noise term	statement	lemma B.6	arxiv
Bound for the policy drift	statement	lemma B.4	arxiv
Bounded critic iterates	statement	lemma 4.2	arxiv
Performance difference lemma	statement	lemma 2.2	arxiv
Trajectory regularity along the PMD iterates	statement	lemma 4.3	arxiv
Action-value convergence	statement	corollary 3.5	arxiv
Change of distribution	statement	lemma 3.2	arxiv
Bounded regularized values	statement	lemma 2.3	arxiv
Policy convergence under strong convexity	statement	corollary 3.6	arxiv