“Optimal precommitment strategies in some time-inconsistent control problems”
Christopher Miller, UC Berkeley
Wednesday, November 9, 2016 2:00 - 3:00PM EEB 248
Abstract: Time-inconsistency is a feature of a dynamic optimization problem which causes the Dynamic Programming Principle to fail. This arises in many applications including dynamic optimization of certain risk measures (e.g., conditional value-at-risk, mean-variance, etc.) and problems whose objective function depends non-linearly on an expected value. There are various interpretations and notions of solution to a time-inconsistent problem. In this talk, we focus on optimal precommitment strategies in continuous-time and demonstrate cases where the time-inconsistent problem may be re-written as an optimization problem over the value function of an auxiliary time-consistent optimization problem. While the auxiliary value function typically has a higher dimensional state space, we discuss instances where structure of the problem allows a dimensionality reduction.
Bio: Christopher W. Miller is a Ph.D. candidate in Applied Mathematics at the University of California, Berkeley. Previously, he received a B.S. in Mathematics and Biomedical Engineering with a minor in Economics from the University of Texas at Austin. His research focuses on applications of partial differential equations and optimal stochastic control, particularly in mathematical finance.