Perfil do cutoff do processo de exclusão no grafo completo

Abstract

This dissertation delves into the study of Markov chains, evolutionary processes characterized by “memory loss”, widely applied in diverse fields such as biology, statistics, and finance. The convergence of these chains to a stationary distribution is analyzed using the “total variation distance”. The times of 𝜀-mixing are introduced, representing the time required for convergence. The concept of coupling between Markov chains is presented, demonstrating its utility in de termining bounds for mixing times. The phenomenon of cutoff, an abrupt decrease in total variation distance, is explored, providing a detailed understanding of convergence. The ulti mate goal is to calculate the cutoff profile for the simple exclusion processes on complete graphs. Chapters cover the construction of chains, technical concepts, couplings, and mixing times, cul minating in the analysis of the cutoff phenomenon and its specific application to the exclusion process in the complete graph