On Defining Design Patterns to Generalize and Leverage Automated Constraint Solving
Abstract: This position paper reflects on the generalization of adaptive methods for Constraint Programming (CP) solving mechanisms, and suggests the use of application-oriented descriptions as a means to broaden CP adoption in the industry. We regard as an adaptive method any procedure that modifies the behavior of the solving process according to previous experience gathered from similar cases. Despite its design being much of a creativity matter, many patterns emerge from the comparison of existing methods. As a starting point, we propose a framework with design patterns for learning time, search modification, and case discrimination. Those patterns provide a glimpse of the circumstances in which certain design choices are better suited than others, and thus define a language to handle and address application domains with diverse needs.
Keywords: Algorithm Selection, Hardness Models, Autonomous Search
Category 1: Optimization Software and Modeling Systems (Optimization Software Design Principles )
Category 2: Combinatorial Optimization (Other )
Category 3: Integer Programming (Other )
Citation: Position paper at the Future of CP track in the CP 2012 conference.
Entry Submitted: 10/11/2012
Modify/Update this entry
|Visitors||Authors||More about us||Links|
Search, Browse the Repository
Give us feedback
|Optimization Journals, Sites, Societies|