Show simple item record

EXPRESSING PREFERENCES WITH PRICE-VECTOR AGENTS IN COMBINATORIAL AUCTIONS

dc.contributor.advisorRaghavan, Subramanianen_US
dc.contributor.authorDay, Robert Warrenen_US
dc.date.accessioned2004-08-27T05:41:11Z
dc.date.available2004-08-27T05:41:11Z
dc.date.issued2004-08-12en_US
dc.identifier.urihttp://hdl.handle.net/1903/1849
dc.description.abstractIn this work, we investigate two combinatorial auction formats in which each bidding bidder may be represented by a collection of unit-demand or price-vector agents. In the first model, bidder preferences are aggregated in a Bid Table, through which a bidder in a combinatorial auction may express several forms of subadditive preferences. We show that the gross substitutes property holds for this model, and design a large-scale combinatorial auction using bid tables as a demand revelation stage, determining linear price signals for later stages. The constraints of this model coincide naturally with the restrictions of recently proposed FAA landing-slot auctions, and we provide a slot auction design based on this model. In a second model, we explore the more complex behavior possible when each bidder's collection of price-vector agents coordinate based on a bidder-specified ordering of the auction items. With this coordination each bidder is able to convey a rich set of preferences, including the ability to express both superadditive and subadditive bundle synergies (i.e., substitutes and complements). The instructions for this collection of agents are tabulated in a lower-triangular Matrix Bid, and we compare the use of matrix bids to other compact techniques for writing down a wide variety of bidding information. We show that the winner determination problem for this Matrix Bid Auction is NP-hard, provide results from a series of computational experiments, and develop IP techniques for improving run time. In addition to the results on price-vector agents, bid tables, and matrix bidding, we present a new technique for achieving bidder-Pareto-optimal core outcomes in a sealed-bid combinatorial auction. The key idea of this iterative procedure is the formulation of the separation problem for core constraints at an arbitrary point in winner payment space.en_US
dc.format.extent980634 bytes
dc.format.mimetypeapplication/pdf
dc.language.isoen_US
dc.titleEXPRESSING PREFERENCES WITH PRICE-VECTOR AGENTS IN COMBINATORIAL AUCTIONSen_US
dc.typeDissertationen_US
dc.contributor.publisherDigital Repository at the University of Marylanden_US
dc.contributor.publisherUniversity of Maryland (College Park, Md.)en_US
dc.contributor.departmentApplied Mathematics and Scientific Computationen_US
dc.subject.pqcontrolledApplied Mechanicsen_US
dc.subject.pqcontrolledBusiness Administration, Managementen_US
dc.subject.pqcontrolledEconomics, Theoryen_US
dc.subject.pquncontrolledcombinatorial auctionsen_US
dc.subject.pquncontrolledmatrix bidsen_US
dc.subject.pquncontrolledbid tablesen_US
dc.subject.pquncontrolledcore constraint generationen_US
dc.subject.pquncontrolledairport auctionsen_US
dc.subject.pquncontrolledbidding languagesen_US


Files in this item

Thumbnail

This item appears in the following Collection(s)

Show simple item record