A limit theorem for the prediction process under absolute continuity
Consider a stochastic process with two probability laws, one of which is absolutely continuous with respect to the other. Under each law, we look at a process consisting of the conditional distributions of the future given the past. Blackwell and Dubins showed in discrete case that those conditional distributions merge as we observe more and more; more precisely, the total variation distance between them converges to 0 a.s. In this paper we prove its extension to continuous time case using the prediction process of F. B. Knight.
KeywordsConditional Distribution Absolute Continuity Prediction Process Optional Projection Total Variation Distance
Unable to display preview. Download preview PDF.
- Meyer, P.-A. (1976). La théorie de la prédiction de F. Knight, Séminaire de Probabilités X. Lect. Notes in Math. 511. Springer-Verlag, Berlin Heidelberg New York, 86–103.Google Scholar
- Yor, M. and Meyer, P.-A. (1976). Sur la théorie de la prédiction. et le problème de décomposition des tribus F t+o, Séminaire de Probabilités X, Lect. Notes in Math. 511, Springer-Verlag, Berlin Heidelberg New York, 104–117.Google Scholar