In this talk we will mainly focus on a recent work where we consider a market consisting of many servers who behave like individual agents. Tasks (or jobs) enter this market with a certain valuation and are matched to two servers who each bid a price. If the minimum of the two prices is below the value of the task, the task accepts the service from the lowest bidder. This problem can be formulated in terms of a game with a large number of players (agents/servers) who wish to individually maximize their revenue. We examine the dynamics of this game using a mean field game framework and talk about the insights gained through this analysis. We will end the talk with a look at other examples where similar ideas can be exploited to compute distributed control strategies that are Nash-Equilibria.
Dheeraj Narasimha obtained his PhD degree from Arizona State University under Prof. Lei Ying in 2021 while working on mean-field games for density dependent continuous time markov processes. He then went on to do a post-doctoral fellowship under Prof. Srinivas Shakkottai at Texas A&M University and then completed a second postdoc at INRIA, CNRS under Dr. Nicolas Gast. His work on mean-field games has won runner-up best-paper prizes in ACM mobihoc as well as most recently in IEEE INFOCOM. The main focus of his work revolves around solving decision problems with many participants under weak interaction, be it through weak graph limits, potential games or mean field structures. Currently he is a CNI post-doctoral fellow.