Robust optimization with integer variables and its applications to flood protection



In this talk I present my research on robust optimization - a paradigm for problems affected by uncertainty, with a focus on problems with integer variables, often encountered in applications. 

First, I present a brief introduction to robust optimization, highlighting my research up to now. Then, I illustrate the challenges related to multi-stage optimization with integer variables, in particular, the difficulty with formulating integers as computationally tractable functions of the problem parameters. For these challenges, I propose a methodology of iterative partitions of the uncertainty set. In the end, I present an application of robust integer optimization to a large-scale industrial problem: flood protection in the Rhine Estuary area, analyzed in cooperation with the Deltares institute. I conclude by discussing unresolved challenges and ideas for future research.