Id: arxiv/2601.21131 Url: https://arxiv.org/abs/2601.21131 Tags: math.ST, cs.IT, cs.LG, stat.ML Score: 0 Title: Thompson sampling: Precise arm-pull dynamics and adaptive inference Domain: math.ST Source: arxiv Authors: Qiyang Han Reviews: 0 Subject: math.ST Category: research Comments: 0 Disputed: false Division: math Mentions: 0 Discussed: 0 Published: 2026-01-29 Page: https://delphi.science/r/arxiv/2601.21131 Adaptive sampling schemes are well known to create complex dependence that may invalidate conventional inference methods. A recent line of work shows that this need not be the case for UCB-type algorithms in multi-armed bandits. A central emerging theme is a `stability' property with asymptotically deterministic arm-pull counts in these algorithms, making inference as easy as in the i.i.d. setting. In this paper, we study the precise arm-pull dynamics in another canonical class of Thompson-sampling type algorithms. We show that the phenomenology is qualitatively different: the arm-pull count…