An axiomatic approach to the cut-off phenomenon for random walks on large distance-regular graphs (Q1588317)

The cut-off phenomenon of Diaconis is a critical phenomenon for the rate of convergence to equilibrium observed for many classes of random walks. It occurs for many relevant examples like card shuffling problems and particle systems. The author extends the well-known approach of Diaconis, Shashahani, and many others for random walks associated with finite Gelfand pairs to the setting of finite association schemes and, in particular, distance-regular graphs. Using the spectral data of the associated adjacency matrices (which can be expressed in terms of finite orthogonal polynomials), he presents general upper and lower bounds for the total variation norm distance from the equilibrium. The results are illustrated by several examples. The results of this paper are closely related to those of a recent paper of E. D. Belsey, and it must be mentioned that for some examples clearly much better results are known.