Premium
An Approximate Innovation Method For The Estimation Of Diffusion Processes From Discrete Data
Author(s) -
Jimenez J. C.,
Ozaki T.
Publication year - 2006
Publication title -
journal of time series analysis
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.576
H-Index - 54
eISSN - 1467-9892
pISSN - 0143-9782
DOI - 10.1111/j.1467-9892.2005.00454.x
Subject(s) - estimator , mathematics , linearization , multiplicative function , diffusion , estimation , multiplicative noise , noise (video) , diffusion process , filter (signal processing) , extension (predicate logic) , component (thermodynamics) , mathematical optimization , algorithm , computer science , innovation diffusion , statistics , mathematical analysis , nonlinear system , signal transfer function , artificial intelligence , analog signal , image (mathematics) , knowledge management , management , quantum mechanics , computer vision , thermodynamics , programming language , physics , digital signal processing , computer hardware , economics
. In this paper, an approximate innovation method is introduced for the estimation of diffusion processes, given a set of discrete and noisy observations of some of their components. The method is based on a recent extension of local linearization filters to the general case of continuous–discrete state–space models with multiplicative noise. This filtering method provides adequate approximations for the prediction and filter estimates that are required by the innovation method in the estimation of the unknown parameters and the unobserved component of the diffusion process. The performance of approximate innovation estimators is illustrated by means of numerical simulations.