While numerous methods have been proposed for computing distances between
probability distributions in Euclidean space, relatively little attention has
been given to computing such distances for distributions on graphs. However,
there has been a marked increase in data that either lies