Nonparametric Bayesian estimation for multivariate Hawkes processes (Q2215756)

The authors of the paper under review study the properties of Bayesian nonparametric procedures in the context of multivariate Hawkes processes. Their aim is to give some general results on posterior concentration rates for such models and to analyze some families of nonparametric priors. Namely, two types of posterior concentration rates are provided: the first one is in terms of the L1-distance on the stochastic intensities of the Hawkes process and the second one is in terms of the L1-distance on the estimated parameter. The authors propose an MCMC algorithm to simulate from the posterior distribution for the priors constructed from piecewise constant functions. Finally, a simulation study is conducted to illustrate the results and to study empirically the inference on functional connectivity graphs of neurons.