A new lower bound on the price of anarchy of selfish bin packing

Abstract

In standard bin packing, the goal is to partition or pack items with positive sizes of at most 1 into a minimum number of subsets, called bins, each of a total size no larger than 1. We study bin packing games. In these games, given a valid partition of the items, each item has a cost associated with it, based on the partition and on its size. Specifically, the cost of an item is the ratio between its size and the total size of items packed into its bin, that is, the cost sharing is in proportion to item sizes. We study pure Nash equilibria, which exist for all such games, and prove a new lower bound on the price of anarchy, which is the asymptotic worst-case ratio between the cost of the worst Nash equilibrium and a socially optimal packing.