A New Constructive Duality Theory for Mixed Integer Programming.

摘要

This paper extend the integer programming dual methods to mixed integer programming (MIP). Specifically, the IP dual methods are combined with Bender's partitioning method for MIP. The result is a constructive MIP duality theory with the same properties as the IP duality theory; namely, the generation of monotonically increasing dual lower bounds, integer solutions which are feasible if they are non-negative, and MIP Gomory cuts for persisting with the dual analysis until MIP problem is minimized. (Author)