报告人:徐礼虎(澳门大学珠海研究院)
报告时间:2026年11月20日(星期五)16:00-17:30
报告地点:科技楼南楼711会议室
报告摘要:We study Markovian many-server queues with heavy-tailed batch arrivals and reneging under the first-come first-served discipline. Batches arrive at Poisson times, with the arrival rates scaling in a power law of parameter 𝛼∈(0, 2) as batch size increases. We investigate the finite-horizon and steady-state approximations of appropriately scaled queueing processes 𝑋𝑛 in a modified Halfin-Whitt regime. The asymptotic behavior differs sharply across three regimes: 𝛼∈(0, 1), 𝛼=1 and 𝛼∈(1, 2). The stochastic process limit Y of 𝑋𝑛 is governed by SDEs with a piecewise linear drift and driven by a stable-type Lévy process, which is a truncated 𝛼-stable subordinator, a compensated Lévy process with jumps bounded by one, and a truncated stable-type or 𝛼-stable Lévy process, respectively.We further prove that the stationary distribution of the scaled queue converges to the stationary distribution of the limiting process and establish explicit Wasserstein-1 convergence rates. The steady-state estimates are obtained through the Duhamel principle, generator comparison, and semigroup gradient bounds.
报告人简介:徐礼虎,教授,2008年博士毕业于英国帝国理工大学,2007-2011在Bonn University做博士后;2026年至今在澳门大学工作,研究方向主要包括随机分析和应用概率及随机动力系统等,在AoS, AAP, PTRF等杂志共计发表50余篇论文。
邀请人:吴付科