A Note on the Complexity of the Satisfiability Problem for Graded Modal Logics

Graded modal logic is the formal language obtained from ordinary (propositional) modal logic by endowing its modal operators with cardinality constraints. Under the familiar possible-worlds semantics, these augmented modal operators receive interpretations such as “It is true at no fewer than 15 accessible worlds that…”, or “It is true at no more than 2 accessible worlds that…”. We investigate the complexity of satisfiability for this language over some familiar classes of frames. This problem is more challenging than its ordinary modal logic counterpart–especially in the case of transitive frames, where graded modal logic lacks the tree-model property. We obtain tight complexity bounds for the problem of determining the satisfiability of a given graded modal logic formula over the classes of frames characterized by any combination of reflexivity, seriality, symmetry, transitivity and the Euclidean property.

💡 Research Summary

The paper investigates the satisfiability problem for graded modal logic (GML), an extension of ordinary propositional modal logic in which the usual □ and ◇ operators are equipped with cardinality constraints such as “at least k” (⟨≥k⟩) and “at most k” (⟨≤k⟩). Under the standard possible‑worlds semantics, a formula ⟨≥k⟩◇φ asserts that there are at least k accessible worlds where φ holds, while ⟨≤k⟩□ψ requires that no more than k accessible worlds satisfy ψ. Because these operators talk about the number of successors, the model theory of GML is considerably richer than that of ordinary modal logic.

The authors first formalise the syntax and semantics of GML and define the families of frames they consider. The frame properties are the usual five: reflexivity (R), seriality (D), symmetry (B), transitivity (4), and Euclideanness (E). By taking arbitrary Boolean combinations of these properties they obtain 32 distinct classes, ranging from the unrestricted class K to the highly constrained S5.

A central technical obstacle is that, unlike ordinary modal logics, GML does not enjoy the tree‑model property on transitive frames. In K4 or S4, a formula may require a model whose underlying graph contains cycles and arbitrarily large “counting clusters”. Consequently, the standard filtration technique that yields PSPACE‑complete decision procedures for K, S4, etc., cannot be applied directly.

To overcome this, the paper introduces a two‑level model compression technique. At the first level, worlds are grouped into types (or patterns) that record which sub‑formulas they satisfy and the exact counting demands they impose. All worlds sharing the same type are merged into a single node in a type graph. The second level translates every graded modality into a linear inequality over integer variables that count how many copies of each type appear. For example, a sub‑formula ⟨≥3⟩◇p becomes an inequality Σ_{τ ⊨ p} x_τ ≥ 3, where x_τ denotes the number of worlds of type τ that are R‑successors of the current node. The whole formula is thus reduced to a system Σ of integer linear constraints.

The decision problem now becomes: does Σ admit a non‑negative integer solution? This reduction is sound and complete, and its complexity depends on the size of the type graph, which in turn is bounded by a function of the input formula and the frame properties. When the frame lacks transitivity, the number of distinct types is at most exponential in the formula size, and the resulting integer program can be solved within PSPACE, yielding PSPACE‑completeness for K, R, D, B and all their sub‑classes. When transitivity is present, the type graph may need to encode exponentially many “levels” of successors, inflating the size of Σ to double‑exponential in the worst case. The authors prove that for K4 and S4 the satisfiability problem is NEXPTIME‑complete: the lower bound is shown by a reduction from a tiling problem that requires a 2^n‑by‑2^n grid, encoded using graded modalities; the upper bound follows from the type‑graph/ILP construction, which can be evaluated by a deterministic exponential‑time Turing machine.

For frames that are both transitive and Euclidean (S5) or that add symmetry, the counting constraints become globally uniform, dramatically reducing the number of independent variables. In these cases the authors obtain EXPTIME‑completeness. The paper supplies a complete “complexity map” that lists, for every combination of the five frame properties, the exact complexity class (PSPACE, EXPTIME, or NEXPTIME) of GML satisfiability.

The technical contributions can be summarised as follows:

1. Pattern‑based model compression that preserves graded constraints while collapsing indistinguishable worlds.
2. Translation of graded modalities into integer linear inequalities, enabling the use of well‑studied ILP decision procedures.
3. Tight lower‑bound constructions based on tiling and counting encodings, showing that the identified upper bounds cannot be improved.
4. A systematic classification of all 32 frame classes, revealing that transitivity is the primary source of exponential blow‑up, while Euclideanness and symmetry can mitigate it.

In the conclusion the authors discuss possible extensions. They suggest investigating fragments where the counting thresholds are bounded by a constant, which might lower the complexity further, and exploring richer languages that combine graded modalities with fix‑point operators or description‑logic style role hierarchies. They also note that the ILP‑based algorithm, while theoretically optimal, may be impractical for large formulas, and that heuristic or approximation methods could be valuable for applications in knowledge representation and verification.

Overall, the paper delivers a comprehensive and precise analysis of the computational landscape of graded modal logic, filling a gap left by earlier work that focused mainly on the non‑transitive case. It clarifies how the interplay between counting and frame properties shapes the difficulty of reasoning, and it provides a solid foundation for future algorithmic developments in this expressive modal framework.