Conference abstracts
Session C3 - Continuous Optimization
July 17, 19:00 ~ 19:30 - Room 111
Solving chance-constrained problems using a kernel VaR estimator
Andreas Waechter
Northwestern University, United States - andreasw.waechter@northwestern.edu
We present a reformulation of nonlinear optimization problems with chance constraints that replaces a probabilistic constraint by a nonparametric value-at-risk estimator. The estimator smooths the historical VaR via a kernel, resulting in a better approximation of the quantile and reducing the nonconvexity of the feasible region. An optimization algorithm is proposed that permits the efficient treatment of joint chance constraints. Theoretical convergence properties and numerical experiments are presented.
Joint work with Alejandra Pena-Ordieres (Northwestern University, USA).