A Unifying Framework for Structural Properties of CSPs: Definitions, Complexity, Tractability

Literature on Constraint Satisfaction exhibits the definition of several structural properties that can be possessed by CSPs, like (in)consistency, substitutability or interchangeability. Current tools for constraint solving typically detect such properties efficiently by means of incomplete yet effective algorithms, and use them to reduce the search space and boost search. In this paper, we provide a unifying framework encompassing most of the properties known so far, both in CSP and other fields literature, and shed light on the semantical relationships among them. This gives a unified and comprehensive view of the topic, allows new, unknown, properties to emerge, and clarifies the computational complexity of the various detection problems. In particular, among the others, two new concepts, fixability and removability emerge, that come out to be the ideal characterisations of values that may be safely assigned or removed from a variables domain, while preserving problem satisfiability. These two notions subsume a large number of known properties, including inconsistency, substitutability and others. Because of the computational intractability of all the property-detection problems, by following the CSP approach we then determine a number of relaxations which provide sufficient conditions for their tractability. In particular, we exploit forms of language restrictions and local reasoning.

💡 Research Summary

The paper presents a comprehensive unifying framework for the myriad structural properties that have been studied in the Constraint Satisfaction Problem (CSP) literature. After recalling the basic CSP model—variables, finite domains, and a set of constraints—the authors systematically catalogue existing notions such as various levels of consistency, inconsistency, substitutability, interchangeability, and forced values. They show that each of these can be expressed as a logical predicate over variable‑value pairs, and that the predicates form a lattice of inclusion relationships.

The core contribution is the introduction of two new concepts: fixability and removability. A value v for variable x is fixable if assigning v to x never destroys the existence of a solution; this subsumes both forced values (which must appear in every solution) and substitutable values (which can be replaced without loss). A value v is removable if deleting v from x’s domain never eliminates all solutions; this captures inconsistency (values that can never appear) and interchangeability (values that can be swapped freely). The authors prove formal theorems establishing that fixability implies both substitutability and forcedness, while removability implies inconsistency and interchangeability, and that the two notions are independent yet together provide a complete safety criterion for domain manipulation.

On the computational side, the paper proves that detecting any of these structural properties is NP‑complete in the general case. By reducing from 3‑SAT to the fixability/removability decision problem and vice‑versa, the authors demonstrate NP‑hardness; membership in NP follows from straightforward nondeterministic verification. They then explore tractable subclasses obtained by language restrictions (e.g., binary constraints, Horn clauses, 2‑SAT, global linear constraints) where polynomial‑time algorithms exist.

To make the framework useful in practice, two sufficient‑condition approaches are proposed. The first exploits language restrictions: when constraints belong to a tractable class, local checks on each variable’s incident constraints are enough to decide fixability or removability. The second is local reasoning: by examining only the sub‑graph induced by a variable and its neighboring constraints, one can derive sufficient conditions that guarantee the global property without exhaustive search. The paper shows that for many common CSP topologies—such as tree‑structured binary networks—these local tests are both sound and complete.

Finally, the authors discuss how the unified view enables the discovery of previously unknown properties and suggests new algorithmic strategies. For instance, a “safe value reduction” technique can be built by first testing fixability and removability; values satisfying either condition can be respectively fixed or pruned, shrinking the search space while preserving satisfiability. The framework is modular, allowing existing solvers to incorporate the new tests as preprocessing or during search. The paper concludes with open research directions: extending tractability results to richer constraint languages, building automated tools that infer structural properties from problem instances, and conducting empirical evaluations to quantify performance gains in real‑world applications. In sum, the work bridges theoretical complexity analysis with practical algorithm design, offering a solid foundation for future advances in CSP solving.