Entropy-driven cutoff phenomena (Q690635)

The authors provide sufficient conditions for a family of finite ergodic Markov chains to exhibit cutoff, abrupt convergence to equilibrium. These conditions are in terms of certain concentration of the distribution of hitting time of the ``relevant part'' of the state space. The main result is used to prove cutoff in several models, including the coupon collector, top-in-at-random, Ehrenfest urn, lazy random walk on the hypercube, Glauber dynamics for the mean-field Ising model, and partially-diffusive random walk. The result about cutoff in the partially-diffusive random walk is new.