Proceedings Abstracts of the Twenty-Fifth International Joint Conference on Artificial Intelligence

Generalizing the Edge-Finder Rule for the Cumulative Constraint / 3103
Vincent Gingras, Claude-Guy Quimper

We present two novel filtering algorithms for the Cumulative constraint based on a new energetic relaxation. We introduce a generalization of the Overload Check and Edge-Finder rules based on a function computing the earliest completion time for a set of tasks. Depending on the relaxation used to compute this function, one obtains different levels of filtering. We present two algorithms that enforce these rules. The algorithms utilize a novel data structure that we call Profile and that encodes the resource utilization over time. Experiments show that these algorithms are competitive with the state-of-the-art algorithms, by doing a greater filtering and having a faster runtime.