Convergence of the approximate cores to the aspiration core in partitioning games (Q2355013)

Cooperative games study the redistribution of collective gains or costs contemplating cooperation among individuals or organizations. This paper considers a family of cooperative games with transferable utility, the so-called partitioning games (cf. [\textit{M. Kaneko} and \textit{M. H. Wooders}, Math. Soc. Sci. 3, 313--327 (1982; Zbl 0493.90089)]). A partitioning game is characterized by a finite set \(N\) of players together with a set of coalitions, \(\pi\), or subsets of \(N\), and a payoff function defined on the elements of \(\pi\). In these games, only coalitions of \(\pi\) are relevant and players must be arranged through partitions taken from \(\pi\). The motivation for this class of models is that in many situations it is not possible for certain players to be integrated into a coalition for family, economic, political or other reasons. It can also be difficult to form large coalitions, or the game may have a special structure, so that only small coalitions are relevant. An example is found in the assignment games studied in [\textit{L. S. Shapley} and \textit{M. Shubik}, Int. J. Game Theory 1, 111--130 (1971; Zbl 0236.90078)]. A very important concept for cooperative games solutions is the well-known core, but it has the disadvantage that it can be empty and thus, various new non-empty solutions have been proposed. Among them, we have the approximate core (cf. [\textit{M. H. Wooders}, J. Math. Econ. 11, 277--300 (1983; Zbl 0518.90101)]) and the aspiration core (cf. [\textit{E. Bennett}, Int. J. Game Theory 12, 1--28 (1983; Zbl 0504.90093)]). The main objective of the current paper is to provide a connection between these two solutions in the context of partitioning games. It is noteworthy that \textit{M. Kaneko} and \textit{M. H. Wooders} [Math. Soc. Sci. 3, 313--327 (1982; Zbl 0493.90089)] gave necessary and sufficient conditions under which the core of a partitioning game is not empty. However, these conditions are very restrictive. The approximate core is based on replicas of games to get non-empty approximate cores provided that the number of replications is large enough. As for the aspiration core (or balanced aspiration set), it supposes that the cooperation of the players can be supported by overlapping coalitions structures called balanced families, which suggest the coalitions that can be formed. When the core is not empty, the core and the aspiration core agree. Another important concept in this setup is the \(\varepsilon\)-approximate core (for \(\varepsilon> 0\)) of a cooperative game with transferable utility, which is the set of payments that verify a certain property of equal-treatment in the \(\varepsilon\)-core (cf. [\textit{L. S. Shapley} and \textit{M. Shubik}, Econometrica 34, 805--827 (1966; Zbl 0154.45303)]) of any replica of the game. As for the main results, it is shown that the cores of the replicated games, in a subsequence of the replica games where the non-emptiness of the core is ensured, are equal to the aspiration core of the original game. Furthermore, the author proves that the collection of \(\varepsilon\)-approximate cores converges to the aspiration core as \(\varepsilon\) goes to zero. The conclusion of the results is that the underlying notions to the approximate core and to the aspiration core are fully in accordance.