Universal duality in conic convex optimization (Q868444)

For a given pair of dual convex problems in a conic form, the authors introduce the concept of universal duality. By this they mean a situation where a zero duality gap occurs for every linear objective function and right-hand side of constraint functions. Among others, they provide necessary and sufficient conditions for universal duality and give a relationship between universal duality for conic optimization and boundedness of the primal and dual feasible sets. They also illustrate the results on a class of semindefinite programs in control theory.