Séminaire de Probabilités

Spatiotemporal Hawkes processes with a graphon-induced connectivity structure

par Justin Baars

Europe/Paris
Description

We introduce a spatiotemporal self-exciting point process (Nt(x)), boundedly finite both over time [0,) and space X, with excitation structure determined by a graphon W on X2. This graphon Hawkes process generalizes both the multivariate Hawkes process and the Hawkes process on a countable network, and despite being infinite-dimensional, it is surprisingly tractable. After proving existence, uniqueness and stability results, we show, both in the annealed and in the quenched case, that for compact, Euclidean XRm, any graphon Hawkes process can be obtained as the suitable limit of d-dimensional Hawkes processes N~d, as d. Furthermore, in the stable regime, we establish an FLLN and an FCLT for our infinite-dimensional process on compact XRm, while in the unstable regime we prove divergence of NT(X)/T, as T. Finally, we exploit a cluster representation to derive fixed-point equations for the Laplace functional of N, for which we set up a recursive approximation procedure. We apply these results to show that, starting with multivariate Hawkes processes N~td converging to stable graphon Hawkes processes, the limits d and t commute.