Arnak DALALYAN - Abstract

ASYMPTOTIC STATISTICAL EQUIVALENCE FOR MULTIDIMENSIONAL ERGODIC DIFFUSIONS

Arnak DALALYAN

Univerité Pierre et Marie Curie
Paris, France

Markuss REISS
Humboldt University
Berlin, Germany

ABSTRACT

We consider the statistical model of ergodic $d$-dimensional diffusion with unknown drift $b$ assumed to be the gradient of a potential $V$. The observation is either high frequency discrete sample or a continuous record of a sample path. We show that under mild conditions this model is locally asymptotically equivalent in the sense of Le Cam's distance to a signal in Gaussian white noise model.