Greedy Approximation via Duality for Packing, Combinatorial Auctions and Routing

We study simple greedy approximation algorithms for general class of integer packing problems. We provide a novel analysis based on the duality theory of linear programming. This enables to significantly improve on the approximation ratios of these greedy methods, and gives a unified analysis of greedy for many packing problems. We show matching lower bounds on the ratios of such greedy methods. Applications to some specific problems, including mechanism design for combinatorial auctions, are also shown.

• Type: Conference
• Year: 2005
• Where: MFCS
• Authors: Piotr Krysta
• Keywords: Greedy, Greedy Methods, MFCS 2005, Simple Greedy Approximation