Bootstrap estimators for the tail-index and for the count statistics of graphex processes (Q2219229)

For modeling empirical graphs, graphex processes are used, hence generalizations of vertex-exchangeable random graphs, also called graphon models. In the present article, graphex processes are considered from the statistical point of view, where especially the following items are taken into account: a) The power-law behavior of the degree distribution, as the number of vertices, and b) the sparsity, a property of a sequence of graphs, i.e, the growth of the number of edges in terms of the number of vertices. It is shown that sparse graphex samples have a simply identifiable signature which can be uncovered by using the subsampling invariance. Moreover, estimators are proposed for a selected choice of functionals of the graphon, where estimators originate from the subsampling theory for graphex processes, hence, can be interpreted as a form of bootstrap procedure. These estimators are then used to construct statistical tests for certain counts of motifs.