PARAMETER ESTIMATION IN STOCHASTIC PARTIAL DIFFERENTIAL EQUATIONS


Sergey LOTOTSKY

University of South California, Los Angeles
United States


ABSTRACT

Many natural processes can be modelled using a second-order parabolic equation driven by a space-time white noise. The general form of the operators in the equation is dictated by the underlying deterministic model, and the noise is added to compensate for the model errors and short-time fluctuations. Typically, a realization of the process can be observed and the observations used to further specify the model, namely, to estimate the coefficients of the operators in the equation. In this talk, I will show that a consistent estimator in SPDE models can exist even if both the observation time interval and the amplitude of noise are fixed.



Back to the Workshop's Program