English

On the Sum-of-Squares Algorithm for Bin Packing

Data Structures and Algorithms 2007-05-23 v2

Abstract

In this paper we present a theoretical analysis of the deterministic on-line {\em Sum of Squares} algorithm (SSSS) for bin packing introduced and studied experimentally in \cite{CJK99}, along with several new variants. SSSS is applicable to any instance of bin packing in which the bin capacity BB and item sizes s(a)s(a) are integral (or can be scaled to be so), and runs in time O(nB)O(nB). It performs remarkably well from an average case point of view: For any discrete distribution in which the optimal expected waste is sublinear, SSSS also has sublinear expected waste. For any discrete distribution where the optimal expected waste is bounded, SSSS has expected waste at most O(logn)O(\log n). In addition, we discuss several interesting variants on SSSS, including a randomized O(nBlogB)O(nB\log B)-time on-line algorithm SSSS^*, based on SSSS, whose expected behavior is essentially optimal for all discrete distributions. Algorithm SSSS^* also depends on a new linear-programming-based pseudopolynomial-time algorithm for solving the NP-hard problem of determining, given a discrete distribution FF, just what is the growth rate for the optimal expected waste. This article is a greatly expanded version of the conference paper \cite{sumsq2000}.

Keywords

Cite

@article{arxiv.cs/0210013,
  title  = {On the Sum-of-Squares Algorithm for Bin Packing},
  author = {Janos Csirik and David S. Johnson and Claire Kenyon and James B. Orlin and Peter W. Shor and Richard R. Weber},
  journal= {arXiv preprint arXiv:cs/0210013},
  year   = {2007}
}

Comments

72 pages