This paper furthers the recent investigation of search heuristics based on solution counting information, by proposing and evaluating algorithms to compute solution densities of variable-value pairs in knapsack constraints. Given a domain consistent constraint, our first algorithm is inspired from what was proposed for regular language membership constraints. Given a bounds consistent constraint, our second algorithm draws from discrete uniform distributions. Experiments on several problems reveal that simple search heuristics built from the information obtained by these algorithms can be very effective.
- Discrete Random Variable
- Solution Density
- Completion Problem
- Discrete Uniform Distribution
- Bound Consistency
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.