Non-parametric estimation of stochastic differential equations from stationary time-series (Q824289)

The authors given a rigorous method to optimally select computational parameters for nonparametric estimation of the drift and diffusion coefficients in stochastic differential equations from discretely sampled stationary data. The main results are concerning on analysis of the bias and the mean-squared error of the drift and diffusion estimators. Also, some interesting numerical results are presented. These results seems to extend some good results given by \textit{P. Sura} and \textit{J. Barsugli} [Phys. Lett., A 305, No. 5, 304--311 (2002; Zbl 1001.82096)] or \textit{F. Legoll} and \textit{T. Lelièvre} [Nonlinearity 23, No. 9, 2131--2163 (2010; Zbl 1209.60036)].