Prof. Dr. Sören Christensen
Decision making is risky business and choosing the best time for a decision is fundamental. Therefore, it is not surprising that one key problem in stochastic control theory is the following: When should a decision maker control a system with a stochastic component to maximize the expected reward? In this talk, I will explain fundamental ideas and recent developments in this field by considering an easy toy model for harvesting and natural resource management. In more mathematical terms, we use as a basis an impulse-control model for underlying diffusion processes. In variants of this model, we discuss mean-field interactions and obtain mathematical results for reinforcement learning.
Datum: 16.05.2019, Raum: G03-106, Zeit: 17:00