1: \begin{abstract}

2: %We study the classic problem of how a group of self-interested agents should divide the value that they collectively create. The traditional approach in Cooperative Game Theory is to study \emph{Core stability} with the implicit assumption that are infinite copies of a single project, and agents can partition themselves as much as they like. In contrast, our setting models a highly general combinatorial setting with a finite set of non-identical projects each having a subadditive valuation function $(v_k(S))_{k=1}^m$. Our main result in this paper is a black-box mechanism that reduces the problem of computing a near-optimal core stable solution to the purely algorithmic problem of welfare maximization; we apply this to compute an approximately core stable solution that extracts one-fourth of the optimal social welfare for the class of subadditive valuations. In addition we show stronger results for several popular sub-classes: anonymous, fractionally subadditive, and submodular functions as well as provide new, fast approximation algorithms for welfare maximization with anonymous functions. Finally, we establish a connection between our utility sharing game, and the well-studied simultaneous auctions with item bidding; we adapt our results to compute approximate pure Nash equilibria for these auctions. Our results indicate that while designing approximation algorithms for selfish agents, one can often compute high quality stable solutions without giving up too much welfare.

3: %\end{abstract}

4: