Stability-based scenario generation in stochastic optimization

Werner Roemisch, R. Henrion

Scenarios are indispensable ingredients for the numerical solution of stochastic optimization problems. We review earlier stability-based approaches for scenario generation and suggest
to make use of stability estimates based on distances containing minimal information. For
linear two-stage stochastic programs the latter approach to scenario generation can be
reformulated as best approximation problem for the expected recourse function and as generalized semi-infinite program, respectively. The latter model turns out to be convex
if either right-hand sides or costs are random. Further properties and solution approaches
are also discussed..