The complexity of hierarchical systems

We introduce a procedure for coarse graining a given hierarchical structure and show how it leads to an effective saturation of the complexity value with increasing number of lower levels. Secondly, we verify that this coarse grained measure has the property of isolating the most diverse trees as the ones with maximal complexity. As a corollary, we cast the dynamical measure of complexity of Bachas and Huberman in terms of purely static properties of trees representing ultradiffusion. We also discuss the differences between the coarse-grained measure of complexity and that provided by relaxation processes.