Duality gap in interval linear programming (Q2302757)

The duality gap in interval linear programming is studied. Concepts of a weakly and strongly zero duality gap for interval linear programming problems are introduced. The authors consider three types of interval linear programming problems: 1) problems with equality constraints and nonegative variables, 2) problems with inequality constraints without restrictions on signs of variables, 3) problems with inequality constraints and nonnegative variables. A characterization for weakly zero duality is proposed. Both sufficient and necessary conditions for a strongly zero duality gap are proved. The computational complexity of the proposed procedures is analyzed. Some open questions for further research in the area are summarized in the Conclusion.