IISER MOHALI

Abstract: Many networked systems in practice exhibit symmetric interaction among the agents, and their performance is governed by the collective behavior of all the agents. For instance, in a load-balancing network consisting of many servers, the instantaneous arrival rate of jobs to a server depends only on the network's load profile, i.e., the number of servers at each possible load. In such networks, the system’s dynamics are said to be driven by “mean-field interaction.” The performance of these networks is governed by the collective behavior (i.e., the mean field) of all the agents; e.g, the average delay of a job is determined by the equilibrium behavior of the load profile of the system. In this talk, we first review some basics of continuous-time Markov chains. We provide a description of some networked systems that can be modeled using a mean field interacting particle system. We then describe a generic mean field model and provide a qualitative description of its scaling limit when the number of particles becomes large.

Tea: 05:00 PM