In September 2026, Dr. Peiwen Yu from the Department of Intelligent Business and Management at the Business School of Soochow University, together with Qidong Lai, Yuan Tian, and Guangwu Liu, had their paper, “Solving Markov Decision Processes via Largest-size Average Estimator,” accepted by INFORMS Journal on Computing. The journal is a leading international journal in operations research and management science and is included in the UTD24 journal list.

This study focuses on simulation optimization for Markov decision processes (MDPs), where decision makers often need to estimate the values of alternative decisions using a limited simulation budget and select the best decision accordingly. The paper proposes a new Largest-Size Average (LSA) estimator that exploits sample-allocation information generated during adaptive simulation and uses the decision receiving the largest number of samples as the basis for value estimation. The paper establishes theoretical properties of the LSA estimator and incorporates it into an Adaptive Multistage Sampling (AMS) framework for solving finite-horizon MDPs. The results show that LSA can effectively use adaptive sample-allocation information to improve value-estimation accuracy and support better decision and policy selection. This study provides a new approach to solving complex stochastic dynamic decision problems with limited simulation resources and contributes to research on simulation optimization, reinforcement learning, and sequential decision-making.

Peiwen Yu holds a Ph.D. in Management Sciences and is currently a Lecturer in the Department of Intelligent Business and Management at the Business School of Soochow University. She received her Ph.D. from City University of Hong Kong in 2023, after earning her master’s degree from the Hong Kong University of Science and Technology and bachelor’s degree from Nanjing University. Her research interests include stochastic simulation and optimization, machine learning, and their applications in financial engineering and risk management.