Title: Bayesian parameter inference estimation for partially observed fractional Brownian motion
Authors: Mohamed Maama - KAUST University (Saudi Arabia) [presenting]
Abstract: State space models are widely used in several branches of science, including statistics, applied mathematics, biology, and economics. We consider static bayesian parameter estimation for partially observed diffusions with fractional Brownian motion (fBm). We elaborate on adaptive Markov chain Monte Carlo algorithms that permit us to infer static parameters of stochastic processes based on the Euler-Maruyama approximation. We simulate our algorithms on two models, the first is an Ornstein Uhlenbeck process driven by fBm, and the second is the CoxIngersollRoss (CIR) model with real data. By using numerical results, we compare the efficiency of our algorithms by using the mean square error versus cost.