A0163
Title: Dirft estimation for high dimensional diffusion models
Authors: Mark Podolskij - University of Luxembourg (Luxembourg) [presenting]
Abstract: High dimensional estimation problems for continuous diffusion models are investigated. Parametric and non-parametric methods for drift and volatility estimation in fixed dimensions are nowadays well understood in the statistical and econometric literature. However, many diffusion systems appearing in practical applications are high dimensional, and thus a new estimation approach is required. We consider a LASSO type estimator of the drift function when complete paths observations are given. We derive rates of convergence for the estimator under sparsity constraints on the parameter.
