Statistical inference for mean-field supermarket queueing models
数学专题报告
报告题目(Title)：Statistical inference for mean-field supermarket queueing models
报告人(Speaker)：Yiqiang Q. Zhao（School of Mathematics and Statistics, Carleton University）
地点(Place)：后主楼 1225
时间(Time)：2024 年 5 月 20 日，16:00—17:00
邀请人(Inviter)：洪文明
报告摘要
This is an on-going research with my collaborators. In 2022, to celebrate the publication
of the 100th volume, Queueing Systems published a special issue containing 100 Views on Queues.
I presented my suggested research in the area of statistical inference for mean-filed queueing models.
The idea of mean-field approximations originated from works by Curie (1895) and Weiss (1907) in
the area of statistical physics. Since then, it has been developed into a very important approximation
tool for describing a large number of interacting particles systems (IPS) for many important
applications in various fields, including queueing theory.
Applications of mean-field approximations to queueing systems go back to Dobrushin and Sukhov
(1976), in which the so-called supermarket models were considered . Research on statistical inference for mean-field models started only in recent years due to difficulties as stated in Della Maestra and Hoffmann (2021): (1) the required fine probabilistic tools were still in full development; and (2) the motivation for statistical inference is not obvious.
So far, according to our best knowledge, statistical inference studies on mean-field models are all
for continuous state systems, and the study on discrete-state processes , say queueing models or
jump processes, has not yet started. In this talk, we demonstrate, through the supermarket model,
the feasibility of this suggested research topic.
This talk is based on the joint work with Ahmed Sid-Ali, Ioannis Lambadaris, Wei Sun.
