Qiaomin Xie

Research profile

Research summary

Qiaomin Xie is a researcher at the University of Wisconsin–Madison working in applied probability and reinforcement learning theory, with an OpenAlex-indexed record showing an h-index of 15 and an i10-index of 17. The publication record is strongly front-loaded in recent years: annual output rose from 10 works in 2020 to 15 in 2023, 17 in 2024, and 16 recorded in 2026 to date, giving 38 works over the most recent three-year window. Topic attribution spans reinforcement learning in robotics (42 works), bandit algorithms (24), stochastic gradient optimization (16), cloud computing and resource management (15), and queuing theory (15), while the field distribution splits evenly between probability theory and applied probability / reinforcement learning theory. Two 2026 arXiv preprints reviewed in the Evolve catalog — on steady-state convergence of stochastic approximation and on Gaussian approximation for martingale sums from uniformly ergodic Markov chains — both place Xie in the last author position. Evolve's assessments credit the first with the first steady-state convergence theory for contractive stochastic approximation under Markovian noise, with optimal O(√α) Gaussian rates, and note the second claims first optimal O(n^-1/2) Wasserstein bounds; both carry medium reproducibility readiness. The recent collaboration footprint is narrow — 2 distinct coauthors, with Yixuan Zhang the recurring partner — and the reproducibility dimension scores 55. Output and impact dimension scores are not yet available in this snapshot.

Research keywords

Highlights

• First steady-state convergence theory for contractive stochastic approximation under Markovian noise

  Evolve's verdict on this 2026 arXiv preprint credits it with the first steady-state convergence (SSC) theory for contractive stochastic approximation under Markovian noise, establishing optimal O(√α) Gaussian rates; assessed as rigorous, though key lemmas rest on companion preprints, and reproducibility readiness is medium.

• Optimal Wasserstein Gaussian approximation for Markov-chain martingale sums

  For this 2026 arXiv preprint, Evolve notes it claims the first optimal O(n^-1/2) Wasserstein (W_p) bounds for multivariate martingale sums from uniformly ergodic Markov chains, drawing on Stein's method; the proof reads coherent though dense, with no numerical verification included, and reproducibility readiness is medium.

• A sharply accelerating publication record

  OpenAlex counts show annual output rising from 1–3 works per year in 2011–2019 to 10 (2020), 15 (2023), and a peak of 17 (2024), with 16 works already recorded in 2026 and 38 works over the most recent three-year window; the record carries an h-index of 15 and an i10-index of 17.

Timeline

1. 2011Earliest recorded output in the OpenAlex record: 1 work.
2. 2020Annual output jumps to 10 works, up from 3 in 2019 — the start of a sustained climb.
3. 2023Output reaches 15 works, more than doubling the 5 recorded in each of 2021 and 2022.
4. 2024Peak annual count in the record: 17 works.
5. 202616 works recorded to date; two arXiv preprints reviewed by Evolve — on steady-state convergence of stochastic approximation and on Gaussian approximation for Markov-chain martingale sums — list Xie in the last author position.

Papers in the Evolve catalog

• 2026-09-14

  Steady-State Convergence of Stochastic Approximation

• 2026-09-08

  Gaussian Approximation for Multivariate Martingale Sums from Uniformly Ergodic Markov Chains

Frequent co-authors