Hearin Center for Enterprise Science

 

About the Hearin Center for Enterprise Science

Publications

Resources

Consortiums

Home

Consortiums


PROJECT 4

DISTRIBUTION AND ASSIGNMENT

Mohammad Amini,
University of Memphis

Fred GloverGary Kochenberger
School of Business Administration, 
 University of Mississippi

Proposed Grant $68,282

Abstract:
Consider a decision-making environment involving a number of stakeholders with a variety of requirements or needs. The decision-maker’s objective is to offer a set of actions or options to satisfy the complete set of the stakeholder’s requirements or 

needs, simultaneously. This problem, called the Constraint Satisfiability Problem (CSP),arises in a large number of real-world settings. Important examples include the "Triad of Detailing" and its interconnected allocation, placement, and assignment or detailing functions within the Navy, as well as related applications other military organizations. In addition, a large number of other decision problems can be reframed as CSPs. These problems are known to be NP-complete, implying their computational intractability if optimal solutions are required. Hence, success of current methods that seek the "best" solution is limited to "small-scale" CSPs. For some large-scale problem instances, current methods are often unable even to find feasible solutions. 
The objectives of this proposed research are as follows. (1) Investigate new approaches to modeling CSP problems in the form of a network optimization problem, using special representations referred to as NETFORMS. (2) Develop and implement new and effective generalized network based branch-and-bound approaches to the NETFORM models. The solution generation process will take advantage of the best existing generalized network solution approaches that have been established to be two orders of magnitude faster in obtaining optimal solutions than alternative approaches. The new approach will allow decision-makers, within military or civilian organizations, to approach real-world, "large-scale" CSP instances more effectively. A major by product of this research will be the creation of a general solution platform for solving a large variety of combinatorial problems that have been recast into the NETFORM framework.