97. Nucleolus-type allocations in hierarchies when cooperation is costly
Invited abstract in session TC-2: Allocations, stream Game theory.
Thursday, 12:00 - 13:30Room: C 103
Authors (first author is the speaker)
| 1. | Tamás Solymosi
|
| Operations Research and Actuarial Sciences, Corvinus University of Budapest |
Abstract
We consider a multi-agent decision situation when cooperation is possible, but constrained by a hierarchy of the agents. Moreover, having access to a "crucial resource" is necessary for being able to materialize the potential profit-making capabilities of the agents. Making this resource, capable of serving any subgroup of the agents, available has a fixed investment cost. Utilization of the resource is hierarchical, represented by a rooted tree graph. The root represents the resource, each other node of the tree represents one of the agents with a given profit-making potential that can only materialize if all agents on the path to the root are also participating.
We study fair and stable allocations in such ”hierarchical joint venture” situations determined by solutions of associated cooperative games. We define the value of a coalition of agents as the sum of the individual potential profits of those members who are connected to the root via other members within the coalition minus the fixed (independent of the coalition to be served) investment cost of the "crucial resource". We consider the standard, the per-capita, and the disruption nucleoli, and investigate whether and how these nucleolus-type allocations can be computed directly from the parameters modeling the ”hierarchical joint venture” situation, so there is no need to explicitly generate the exponential-size associated cooperative game and compute its nucleoli with general-purpose algorithms.
Keywords
- Linear and nonlinear optimization
Status: accepted
Back to the list of papers