Nonparametric estimation for irregularly sampled Lévy processes (Q1744225)

The present paper proposes a new statistical methodology for Lévy processes that are sampled at low frequency where the observation times are deterministic but allowed to be irregularly spaced. The only requirement for the asymptotics is that the maximal distance between any two observation times is bounded and that the maximal observation time diverges to infinity. The paper proposes a methodology to estimate the jump dynamics under some additional assumptions, for which minimax optimality is proved. Secondly, an estimator for the distributional density is proposed, which is no longer straightforward in this situation of irregularly spaced low frequency sampling. For both estimators nonasymptotic risk bounds are obtained and the performance is evaluated by means of a simulation study.