 |
Center for Nonlinear Studies |
Chance constraints are a valuable tool for the design of safe and robust decisions in uncertain environments. In this talk, I will discuss some intrinsic convexity properties of chance constraints, which we can sometimes guarantee theoretically, and often exploit numerically. I will first show that for a large class of chance constrained sets, convexity can be guaranteed for safety probability values greater than a computable threshold. Secondly, I’ll present an exact reformulation of general chance constrained problems as convex bilevel programs. I will then derive a tractable penalty approach to solve such problems and finally present an open-source python toolbox with fast computational procedures.
Host: Yury Maximov