On Conditional Distributions for Stochastic Processes,
Abstract
The author asks THE FOLLOWING QUESTION: What meaning can and should be given to conditional probability or conditional distribution, given an event of probability zero. This important question is challenging and vexing, but no fully satisfactory answer is known. One difficulty with the conventional countably additive approach is that there exists no everywhere proper regular conditional distribution given the tail sigma-field, nor given the field of events in the past of a wide sense stopping time. It is suggested that this difficulty can sometimes, and perhaps always, be overcome, if finitely additive probability measures are accepted. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Nov 01, 1972
- Accession Number
- AD0751966
Entities
People
- Lester E. Dubins
Organizations
- University of California, Berkeley