On the Expressiveness and Complexity of ATL

ATL is a temporal logic geared towards the specification and verification of properties in multi-agents systems. It allows to reason on the existence of strategies for coalitions of agents in order to enforce a given property. In this paper, we first precisely characterize the complexity of ATL model-checking over Alternating Transition Systems and Concurrent Game Structures when the number of agents is not fixed. We prove that it is \Delta^P_2 - and \Delta^P_?_3-complete, depending on the underlying multi-agent model (ATS and CGS resp.). We also consider the same problems for some extensions of ATL. We then consider expressiveness issues. We show how ATS and CGS are related and provide translations between these models w.r.t. alternating bisimulation. We also prove that the standard definition of ATL (built on modalities “Next”, “Always” and “Until”) cannot express the duals of its modalities: it is necessary to explicitely add the modality “Release”.

💡 Research Summary

This paper conducts a thorough investigation of Alternating‑time Temporal Logic (ATL), focusing on two fundamental aspects: the exact computational complexity of model‑checking and the expressive power of the logic when interpreted over two standard multi‑agent formalisms—Alternating Transition Systems (ATS) and Concurrent Game Structures (CGS).

Complexity Results
The authors first distinguish between explicit CGS (where the transition table is given directly) and implicit CGS (where transitions are described by Boolean formulas). ATS are naturally equivalent to explicit CGS. The central operation in ATL model‑checking is the computation of the controllable predecessor set CPre(A, S), which determines whether a coalition A can force the system into a set of states S in one step.

• For explicit CGS (and thus ATS) the authors prove that CPre can be decided in AC⁰, i.e., by constant‑depth circuits, and consequently that ATL model‑checking is ΔP₂‑complete (P with an NP oracle). This improves on earlier claims of PTIME‑completeness, showing that when the number of agents is unbounded the problem lies one level higher in the polynomial hierarchy.
• For implicit CGS the transition relation is compactly encoded; the authors show that deciding CPre becomes ΔP₃‑complete (P with an NP⁽ᴺᴾ⁾ oracle). As a direct consequence, ATL model‑checking over implicit CGS is ΔP₃‑complete.

These results are extended to the richer logics ATL⁺ and ATL*; the same hierarchy separation holds, depending on whether the underlying model is an ATS/explicit CGS or an implicit CGS.

Expressiveness and Model Translation
The paper then addresses the relationship between ATS and CGS. It provides constructive translations in both directions that preserve alternating bisimulation, proving that the two formalisms have identical expressive power with respect to ATL. Consequently, any property expressible in ATL on a CGS can be expressed on an equivalent ATS and vice versa, allowing practitioners to choose the more convenient representation without loss of logical fidelity.

Missing Dual Operators
A key insight concerns the standard ATL syntax, which includes only the modalities “Next” (X), “Always” (G), and “Until” (U). The authors demonstrate that, unlike CTL/LTL, ATL cannot define the dual of “Until”, namely the “Release” (R) operator, using only these three primitives. The inability stems from the strategic nature of ATL’s quantifiers: the negation of a coalition’s ability to enforce a property does not correspond to the opponent coalition’s ability to enforce its complement. They prove that adding an explicit Release modality is necessary for a fully expressive ATL.

Implications
The findings have practical implications for the design of ATL‑based verification tools. The choice between explicit and implicit encodings of the transition relation directly impacts the algorithmic complexity of model‑checking. Moreover, the necessity of the Release operator suggests that future extensions of ATL should incorporate it to achieve completeness.

In summary, the paper precisely characterizes ATL model‑checking as ΔP₂‑complete for ATS/explicit CGS and ΔP₃‑complete for implicit CGS, establishes the equivalence of ATS and CGS under alternating bisimulation, and proves that the standard ATL language lacks the expressive power to capture Release without an explicit addition. These contributions deepen our theoretical understanding of ATL and guide the development of more efficient and expressive verification frameworks for multi‑agent systems.