Skip to Main content Skip to Navigation

Méthodes de quantification optimale pour le filtrage et applications à la finance

Abstract : We develop a grid based numerical approach to solve a filtering problem, using results on optimal quantization of random variables. We construct two filtering algorithms using zero order and first order approximation techniques. We suggest implementable versions of these algorithms and study the approximation error behavior by considering the stationnary property of optimal quantizers. The grid approach is then compared to the particle one based on Monte Carlo methods. The study is done over a set of different state models. In a second part, we have been interested in the advantadge given by quantization methods to preprocess offline the information. This permitted to develop a filtering algorithm based on observation (and signal) quantization. Here also the error convergence rate to zero as the quantizer size goes to infinity is studied. Finally, the quantization of the filter as a random variable is studied in order to solve a problem of pricing an American option in an unobserved stochastic volatility market. All results are illustrated by numerical experiments
Document type :
Complete list of metadatas
Contributor : Afef Sellami <>
Submitted on : Friday, February 10, 2006 - 5:36:13 PM
Last modification on : Wednesday, December 9, 2020 - 3:08:58 PM
Long-term archiving on: : Saturday, April 3, 2010 - 10:21:02 PM


  • HAL Id : tel-00011586, version 1


Afef Sellami. Méthodes de quantification optimale pour le filtrage et applications à la finance. Mathématiques [math]. Université Paris Dauphine - Paris IX, 2005. Français. ⟨tel-00011586⟩



Record views


Files downloads