Bootstrapping exchangeable random graphs

We introduce two new bootstraps for exchangeable random graphs. One, the “empirical graphon bootstrap”, is based purely on resampling, while other, “histogram a model-based “sieve” bootstrap. show that both of them accurately approximate sampling distributions motif densities, i.e., normalized counts number times fixed subgraphs appear in network. These densities characterize distribution (infinite) networks. Our therefore give valid quantification uncertainty inferences about fundamental network statistics, and so parameters identifiable from them.