Integrating Constraint Propagation into Dynamic Programming
A new research paper bridges dynamic programming (DP) and constraint programming (CP) by integrating constraint propagation into DP. The method uses a general-purpose CP solver within the Domain-Independent Dynamic Programming framework to prune states and transitions. Experiments on three combinatorial optimization problems—Single Machine Scheduling with Time Windows, Resource Constrained Project Scheduling Problem (RCPSP), and Travelling Salesperson Problem with Time Windows (TSPTW)—show that the hybrid approach improves performance.
Key facts
- Paper arXiv:2603.16648v2 integrates constraint propagation into DP
- Uses a general-purpose CP solver in the Domain-Independent Dynamic Programming framework
- Evaluated on Single Machine Scheduling with Time Windows, RCPSP, and TSPTW
- Constraint propagation prunes states and transitions in DP
- Bridges two paradigms: state-based (DP) and constraint-based (CP)
Entities
—