D. Levanony et al., RECURSIVE-IDENTIFICATION IN CONTINUOUS-TIME STOCHASTIC-PROCESSES, Stochastic processes and their applications, 49(2), 1994, pp. 245-275
Recursive parameter estimation in diffusion processes is considered. F
irst, stability and asymptotic properties of the global, off-line MLE
(maximum likelihood estimator) are obtained under explicit conditions.
The MLE evolution equation is then derived by employing a generalized
Ito differentiation rule. This equation, which is highly sensitive to
initial conditions, is then modified to yield an algorithm (infinite
dimensional in general) which results in an estimator that, irrespecti
ve of initial conditions, is consistent and asymptotically efficient a
nd in addition, converges rapidly to the MLE. The structure of the alg
orithm indicates that well known gradient and Newton type algorithms a
re first-order approximations. The results cover a wide class of proce
sses, including nonstationary or even divergent ones.