Lower bounds for distributed markov chain problems

We study the worst-case communication complexity of distributed algorithms computing a path problem based on stationary distributions of random walks in a network G with the caveat that G is also the communication network. The problem is a natural generalization of shortest path lengths to expected path lengths, and represents a model used in many practical applications such as pagerank and eigentrust as well as other problems involving Markov chains defined by networks. For the problem of computing a single stationary probability, we prove an Ω(n log n) bits lower bound; the trivial centralized algorithm costs O(n) bits and no known algorithm beats this. We also prove lower bounds for the related problems of approximately computing the stationary probabilities, computing only the ranking of the nodes, and computing the node with maximal rank. As a corollary, we obtain lower bounds for labelling schemes for the hitting time between two nodes.