Filtering Bounded Knapsack Constraints in Expected Sublinear Time

We present a highly efficient incremental algorithm for propagating bounded knapsack constraints. Our algorithm is based on the sublinear filtering algorithm for binary knapsack constraints by Katriel et al. and achieves similar speed-ups of one to two orders of magnitude when compared with its lineartime counterpart. We also show that the representation of bounded knapsacks as binary knapsacks leads to ineffective filtering behavior. Experiments on standard knapsack benchmarks show that the new algorithm significantly outperforms existing methods for handling bounded knapsack constraints.

• Type: Conference
• Year: 2010
• Where: AAAI
• Authors: Yuri Malitsky, Meinolf Sellmann, Radoslaw Szymanek
• Keywords: AAAI 2010, Algorithm, Binary Knapsack Constraints, Intelligent Agents, Knapsack Constraints