Abstract: Sequential decision making is often based on mathematical models, hich may or may not adequately portray reality. To have confidence in the proposed action it is crucial to understand the sensitivity of these decisions with respect to the underlying modeling assumptions. The leading theme of my PhD is to investigate topologies for stochastic processes, which are suitable for sequential decision making in discrete time. Several mathematicians with vastly different backgrounds came up with topologies to tackle this question and astonishingly, in discrete time on the space of stochastic processes equipped with their canonical filtration, they all coincide. We also discuss the extension of this topology to stochastic processes with arbitrary filtrations.
The Wasserstein space of Filtered Processes
19.11.2020 17:30 - 18:30
Location:
live stream via Zoom