A class of volumetric barrier decomposition algorithms for stochastic quadratic programming (Q884639)

A class of volumetric barrier decomposition algorithms for solving two-stage stochastic quadratic programs (SQP) with recourse is derived. It is shown that certain members of the class of algorithms have polynomial complexity. Initially, a volumetric barrier for SQP is introduced. Then, it is shown that the set of barrier functions for positive values of the barrier parameter comprises a self-concordant family. Based on this property a class of volumetric barrier decomposition algorithms is presented. Finally, a convergence and complexity analysis of certain members of this class of algorithms is presented.