Rational Verification: From Model Checking to Equilibrium Checking

Authors: Michael Wooldridge, Julian Gutierrez, Paul Harrenstein, Enrico Marchioni, Giuseppe Perelli, Alexis Toumi

AAAI 2016 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We give an overview of a prototype software tool for rational verification: the E AG L E system takes as input a R E AC T I V E M O D U L E S specification of a set of agents (Alur and Henzinger 1999), a goal for each agent, specified as a formula of the temporal logic CTL (Emerson 1990), and a collection of strategies, one for each agent in the system; the system then checks whether the strategies form a Nash equilibrium. We now describe a prototype tool we have developed as a proof of concept for equilibrium checking. We now present a case study based on the system presented in (Fisman, Kupferman, and Lustig 2010). One can show and formally verify using E A G L E that this game has a Nash equilibrium where no player gets its goal achieved, and another that is Pareto optimal, where both players get their goal achieved.
Researcher Affiliation Academia Michael Wooldridge, Julian Gutierrez, Paul Harrenstein, Enrico Marchioni, Giuseppe Perelli, Alexis Toumi Department of Computer Science, University of Oxford
Pseudocode No The paper presents RML module definitions (e.g., "module m0 controls u0, d0 init [] -> u 0 := , d 0 := update [] -> u 0 := , d 0 :=") which are concrete examples in a specific language for modeling agents, rather than general pseudocode or algorithm blocks describing a computational procedure.
Open Source Code No The paper mentions a prototype tool called E AG L E but does not provide any link or explicit statement about its public availability or open-source status.
Open Datasets No The paper uses a "case study based on the system presented in (Fisman, Kupferman, and Lustig 2010)" to demonstrate its prototype tool. This case study involves formal system specifications and agent goals, not a traditional dataset used for training, validation, or testing in an empirical study context.
Dataset Splits No The paper uses a case study involving a formal system model rather than a dataset, so there is no mention of dataset splits (training, validation, test) relevant for reproducibility.
Hardware Specification No The paper does not provide any specific hardware details (e.g., GPU/CPU models, memory) used for running the described case study or prototype tool.
Software Dependencies No The paper mentions "R E AC T I V E M O D U L E S Language (R M L)" and the prototype tool "E AG L E" but does not specify version numbers for these or any other software dependencies.
Experiment Setup No The paper describes the formal setup of the multi-agent system (RML modules, CTL formulas for goals) for the case study, but it does not include details on hyperparameters, optimizer settings, or other system-level training configurations typically found in experimental setups.