An improved approximation algorithm for combinatorial auctions with submodular bidders

Shahar Dobzinski*, Michael Schapira

*Corresponding author for this work

Research output: Contribution to conferencePaperpeer-review

127 Scopus citations

Abstract

We explore the allocation problem in combinatorial auctions with submodular bidders. We provide an e/e-1-approximation algorithm for this problem, Moreover, our algorithm applies to the more general class of XOS bidders. By presenting a matching unconditional lower bound in the communication model, we prove that the upper bound is tight for the XOS class. Our algorithm improves upon the previously known 2-approximation algorithm. In fact, we also exhibit another algorithm which obtains an approximation ratio better than 2 for submodular bidders, even in the value queries model. Throughout the paper we highlight interesting connections between combinatorial auctions with XOS and submodular bidders and various other combinatorial optimization problems. In particular, we discuss coverage problems and online problems.

Original languageAmerican English
Pages1064-1073
Number of pages10
DOIs
StatePublished - 2006
EventSeventeenth Annual ACM-SIAM Symposium on Discrete Algorithms - Miami, FL, United States
Duration: 22 Jan 200624 Jan 2006

Conference

ConferenceSeventeenth Annual ACM-SIAM Symposium on Discrete Algorithms
Country/TerritoryUnited States
CityMiami, FL
Period22/01/0624/01/06

Fingerprint

Dive into the research topics of 'An improved approximation algorithm for combinatorial auctions with submodular bidders'. Together they form a unique fingerprint.

Cite this