Stochastic programming, despite its immense modeling capabilities, is well known to be computationally excruciating. In this paper, we introduce a unified framework of approximating multiperiod stochastic programming from the perspective of robust optimization. Specifically, we propose a framework that integrates multistage modeling with safeguarding constraints. The framework is computationally tractable in the form of second order cone programming (SOCP) and scalable across periods. We compare the computational performance of our proposal with classical stochastic programming approach using sampling approximations and report very encouraging results for a class of project management problems.
Working Paper, NUS Business School