Last modified: 2017-03-16
A Note on Optimization under Uncertainty: Comparing Probabilistically Constrained and Robust Optimization Methodology
Dealing with optimization problems, data entering the optimization process are usually of uncertain nature. There are several approaches to deal with data uncertainty, starting with classical sensitivity analysis and/or parametric programming. In this paper, we concentrate on two specific approaches, namely on chance constrained (stochastic) and robust optimization. Chance (probabilistically) constrained optimization is based on the assumption that underlying uncertainty is driven by a probability distribution, that is, considered as a random vector. On the other hand, robust optimization deals with the situation in which uncertainty is given only by a membership of the uncertain factor to an explicitly defined set. We compare these two approaches with respect to the behavior of optimal values and optimal solution sets and discuss the drawbacks of each of two approaches on an illustrative example.
Stochastic Optimization; Chance Constrained Optimization; Robust Optimization
Full Text: PDF