Statistical inference versus mean field limit for Hawkes processes
We consider a population of $N$ individuals, of which we observe the number of actions as time evolves. For each couple of individuals $(i,j)$, $j$ may or not influence $i$, which we model by i.i.d. Bernoulli$(p)$-random variables, for some unknown parameter $p\in (0,1]$. Each individual acts autonomously at some unknown rate $μ>0$ and acts by mimetism at some rate depending on the number of recent actions of the individuals which influence him, the age of these actions being taken into account through an unknown function $φ$ (roughly, decreasing and with fast decay). The goal of this paper is to estimate $p$, which is the main charateristic of the graph of interactions, in the asymptotic $N\to\infty$, $t\to\infty$. The main issue is that the mean field limit (as $N \to \infty$) of this model is unidentifiable, in that it only depends on the parameters $μ$ and $pφ$. Fortunately, this mean field limit is not valid for large times. We distinguish the subcritical case, where, roughly, the mean number $m_t$ of actions per individual increases linearly and the supercritical case, where $m_t$ increases exponentially. Although the nuisance parameter $φ$ is non-parametric, we are able, in both cases, to estimate $p$ without estimating $φ$ in a nonparametric way, with a precision of order $N^{-1/2}+N^{1/2}m_t^{-1}$, up to some arbitrarily small loss. We explain, using a Gaussian toy model, the reason why this rate of convergence might be (almost) optimal.
