Model checking coalitional games in shortage resource scenarios

Verification of multi-agents systems (MAS) has been recently studied taking into account the need of expressing resource bounds. Several logics for specifying properties of MAS have been presented in quite a variety of scenarios with bounded resources. In this paper, we study a different formalism, called Priced Resource-Bounded Alternating-time Temporal Logic (PRBATL), whose main novelty consists in moving the notion of resources from a syntactic level (part of the formula) to a semantic one (part of the model). This allows us to track the evolution of the resource availability along the computations and provides us with a formalisms capable to model a number of real-world scenarios. Two relevant aspects are the notion of global availability of the resources on the market, that are shared by the agents, and the notion of price of resources, depending on their availability. In a previous work of ours, an initial step towards this new formalism was introduced, along with an EXPTIME algorithm for the model checking problem. In this paper we better analyze the features of the proposed formalism, also in comparison with previous approaches. The main technical contribution is the proof of the EXPTIME-hardness of the the model checking problem for PRBATL, based on a reduction from the acceptance problem for Linearly-Bounded Alternating Turing Machines. In particular, since the problem has multiple parameters, we show two fixed-parameter reductions.

💡 Research Summary

The paper introduces a novel logical framework called Priced Resource‑Bounded Alternating‑time Temporal Logic (PRB‑ATL) for the verification of multi‑agent systems (MAS) where resources are scarce and their availability changes over time. Unlike earlier resource‑bounded logics such as RB‑CL, RB‑ATL, or RAL, which embed resource bounds syntactically inside formulas and treat resources as private to a coalition, PRB‑ATL lifts resources to the semantic level: a global market vector records the current quantity of each resource type, and a price function maps the current market state to a cost for each resource. This design enables the logic to capture competition for shared resources, dynamic price fluctuations, and the possibility of resource production (subject to a fixed upper bound) while preserving decidability.

The syntax of PRB‑ATL extends ATL with team operators ⟨⟨A,$⟩⟩ that are parameterised by a private “money” vector $ for the coalition A, as well as comparison atoms that test the current market resource vector M. Formulas can express reachability, safety, until, and release properties under the constraint that the coalition never exceeds its monetary budget and that actions respect the current market availability.

A priced game structure G = (Q,π,d,D,qty,δ,ρ,M₀) is defined. Q is a finite set of locations, π labels each location with atomic propositions, d(q,a) gives the number of actions available to agent a at q, and D(q) collects all joint action profiles. qty(q,a,α) specifies the amount of each resource consumed (or produced) by action α of agent a; the “do‑nothing” action consumes nothing. The transition function δ maps a location and a joint action profile to the next location. The price function ρ(M,q,a) returns a vector of prices for each resource type, depending on the current market vector M, the location, and the acting agent. The initial market vector M₀ encodes the total amount of each resource present at the start of the system.

The authors recall an EXPTIME upper‑bound algorithm from their earlier work: model checking proceeds by exploring an extended state space that includes the current location, the market vector, and each agent’s private money vector. The algorithm is exponential in the number of agents n, the number of resource types r, and the binary size of the largest component of M₀ (denoted log M). This yields an EXPTIME upper bound for PRB‑ATL model checking.

The main technical contribution of the current paper is to prove that this upper bound is tight. The authors construct two fixed‑parameter reductions from the acceptance problem for Linearly‑Bounded Alternating Turing Machines (LB‑ATM), which is known to be EXPTIME‑complete. The first reduction treats the binary size of the initial market vector M₀ as the only parameter, fixing n and r; the second treats the number of resource types r as the only parameter, fixing n and the magnitude of M₀. Both reductions encode the computation of an LB‑ATM into a PRB‑ATL game structure such that the ATM accepts iff a certain PRB‑ATL formula holds. Consequently, the model checking problem for PRB‑ATL is EXPTIME‑hard, establishing EXPTIME‑completeness.

A detailed comparative analysis follows. In RB‑CL and RB‑ATL, resource bounds appear only in formulas and apply solely to the proponent coalition; the opponent’s actions are unrestricted with respect to resources, and there is no notion of a shared market or dynamic pricing. Consequently, certain unrealistic behaviours (e.g., infinite consumption loops) can satisfy formulas that would be rejected under PRB‑ATL. Moreover, RB‑ATL allows only consumption, not production, which limits its ability to model infinite‑horizon games where resources can be replenished. RAL and RAL* explore a broader landscape of syntactic/semantic variants, showing that even minor changes can shift the model checking problem from decidable to undecidable. PRB‑ATL, by contrast, deliberately incorporates shared resources, price dynamics, and bounded production to retain decidability while offering richer expressive power.

The paper also sketches several realistic scenarios that fit the PRB‑ATL framework: a program that can request additional memory from a shared pool, a car‑racing game where each car has private fuel but may refuel at a public station whose price rises when the overall fuel supply is low, or a leasing system where agents lease cars and must pay market‑dependent fees. In each case, the global resource vector captures the total amount of the commodity, while the price function models market scarcity.

Finally, the authors outline future research directions: extending the price function to non‑linear or stochastic models, allowing multiple private resources per agent, integrating richer economic mechanisms (e.g., auctions), and developing more efficient, possibly approximate, model checking techniques for large‑scale systems.

In summary, the paper presents PRB‑ATL as a powerful, expressive logic for reasoning about coalitional games under resource scarcity, proves that its model checking problem is EXPTIME‑complete via two parameterised reductions, and situates the work within the broader landscape of resource‑aware temporal logics, highlighting both its theoretical significance and practical applicability.