Overview

Constraint satisfaction models problems with variables, possible values, and rules that determine which combinations are allowed, helping solve tasks such as scheduling, graph coloring, and Sudoku. Classical approaches include constraint propagation and backtracking, often combined with SAT, SMT, constraint programming, or integer programming solvers. Recent work also uses graph neural networks, diffusion models, and language models to propose assignments or guide search, frequently pairing them with symbolic verification and repair. A central challenge is making learned methods reliable on genuinely hard instances and robust to changes in problem size, representation, or wording, while matching the efficiency and guarantees of classical solvers.

Latest papers

CardsList

91 papers3 papers in the last four weeks, down 75% on the four weeks before. 0.0% of all new papers.

  1. A binarized-domains arc-consistency algorithm for TCSPs: its computational analysis and its use as a filtering procedure in solution search algorithms

    Feb 22, 2020Amar IsliConstraint SatisfactionTemporal Consistency