报告题目：A merge-sort-based adaptive optimization method for order-of-addition experiments

报告人： 王亚平

报告时间：2026年10月25日 10:00-11:00

报告地点：嘟嘟成人网 106会议室

报告摘要：The primary goal of an order-of-addition (OofA) experiment is to identify an ordering of a set of components that optimizes the response. Existing model-based approaches are effective when the assumed order-effect model is adequate, but the estimated optimum can be sensitive to model misspecification and estimation error. We propose Merge-OofA, an adaptive methodology based on merge sort, to search for the optimal order without fitting a global response model. The procedure starts with singleton blocks and sequentially merges adjacent ordered blocks through relocation and local adjacent-swap comparisons. Under a strict pairwise-consistency condition in the noiseless setting, Merge-OofA recovers the unique optimal order of all m components using Θ(m log m) distinct objective evaluations. We also develop Merge-ECDF, which calibrates adjacent-swap decisions using an empirical distribution of noise differences, and extend the sorting procedure to screening OofA experiments in which each run contains q < m components. Numerical studies compare the merge-based procedures with the quick-sort-based OofA method (Quick-OofA) and component-position screening benchmarks in full-order and screening scheduling problems, and demonstrate the utility of the proposed methods for direct sequence optimization.

报告人简介：王亚平，华东师范大学嘟嘟成人网 教授，博士生导师，主要研究方向为试验设计与分析，计算机试验与机器学习，相关成果发表在AoS、JASA、Biometrika等期刊上，入选教育部国家高层次青年人才计划、中国科协青年人才托举工程等。