本科毕业外文资料翻译.docx

上传人:b****3 文档编号:10487038 上传时间:2023-05-26 格式:DOCX 页数:13 大小:27.97KB
下载 相关 举报
本科毕业外文资料翻译.docx_第1页
第1页 / 共13页
本科毕业外文资料翻译.docx_第2页
第2页 / 共13页
本科毕业外文资料翻译.docx_第3页
第3页 / 共13页
本科毕业外文资料翻译.docx_第4页
第4页 / 共13页
本科毕业外文资料翻译.docx_第5页
第5页 / 共13页
本科毕业外文资料翻译.docx_第6页
第6页 / 共13页
本科毕业外文资料翻译.docx_第7页
第7页 / 共13页
本科毕业外文资料翻译.docx_第8页
第8页 / 共13页
本科毕业外文资料翻译.docx_第9页
第9页 / 共13页
本科毕业外文资料翻译.docx_第10页
第10页 / 共13页
本科毕业外文资料翻译.docx_第11页
第11页 / 共13页
本科毕业外文资料翻译.docx_第12页
第12页 / 共13页
本科毕业外文资料翻译.docx_第13页
第13页 / 共13页
亲,该文档总共13页,全部预览完了,如果喜欢就下载吧!
下载资源
资源描述

本科毕业外文资料翻译.docx

《本科毕业外文资料翻译.docx》由会员分享,可在线阅读,更多相关《本科毕业外文资料翻译.docx(13页珍藏版)》请在冰点文库上搜索。

本科毕业外文资料翻译.docx

本科毕业外文资料翻译

外文资料翻译—原文部分

BiddingClubs:

InstitutionalizedCollusioninAuctions

(FromKevinLeyton.BiddingClubs:

InstitutionalizedCollusioninAuctions,ACM[J].

2000,(11):

272—27)

Weintroduceaclassofmechanisms,calledbiddingclubs,foragentstocoordinatetheirbiddinginauctions.Inabiddingclubagentsfirstconducta“pre-auction”withintheclub;dependingontheoutcomeofthepre-auctionsomesubsetofthemembersoftheclubbidintheprimaryauctioninaprescribedway;and,insomecases,certainmonetarytransferstakeplaceaftertheauction.Biddingclubshaveself-enforcingcollusionpropertiesinthecontextofsecond-priceauctions.Weshowthatthisisstilltruewhenmultipleauctionstakeplaceforsubstitutablegoods,aswellasforcomplementarygoods.Wealsopresentabiddingclubprotocolforfirst-priceauctions.Finally,weshowcaseswherebiddingclubshaveself-enforcingcooperationprotocolsinarbitrarymechanisms.

WiththeexplodingpopularityofauctionsontheInter-netandelsewherehascomeincreasedinterestinsystemstoassist(softwareorhuman)agentsbiddinginsuchauctions.Mostofthesesystemshavetodatedonelittlemorethanaggregateinformationfrommultipleauctionsandpresentittotheuserinaconvenientfashion(e.g.,).Thereisnowbeginningtoemergeasecondgenerationofsystemswhichactuallyprovidebiddingadviceandautomationservicestobidders,goingbeyondthefamiliarproxybiddingfeatureprevalentinonlineauctionstotherealmofbona-fidedecisionsupport.

Thispaperlooksevenbeyondsuchsystems,whicharegearedtowardsassistingasinglebidder,andpresentsaclassofsystemstoassistacollectionofbidders,“biddingclubs”.Theideaissimilartotheideabehind“buyerclubs”ontheInternet(e.g.,and),namelytoaggregatethemarketpowerofindividualbidders.Thenewtwististhatwhereasinabuyerclubthereisaperfectalignmentofthevariousbuyers’interests(sincetherethemorebuyersjoininapurchasethelowerthepriceforeveryone),inabiddingclubthereisamorecomplexstrategicrelationshipamongthem,andthebiddingclubrulesmustbedesignedaccordingly.

Here’sasimpleexample.Consideranauctionwithasingleseller,andsixpotentialbuyers.Assumethatthreeofthepotentialbuyers–A,BandC,withcorresponding(secret)valuationsv1>v2>v3–attempttocoordinatetheirbidding.Assumetheauctionisafirst-priceauction.Underwellknownassumptionsfromtheauctionliterature,itwouldbetheinterestofeachbiddertobidexactly5/6ofhistruevalueintheauction.ThusAwouldendupwithasurplusofv1/6(ifhewinstheauction)or0(ifhedoesn’t),andBandCwithasurplusof0.Istheresomepre-agreementA,BandCcanmakethatwillcauseallofthemtocomeoutoftheauctionatleastaswelloff,andsomeofthemstrictlybetteroff?

Onecouldnaivelysaythattheywouldeachrevealtheirvaluationstooneanotheragreeingthatonlythehighestwouldgoontotheauction;Awouldthere-forebetheonegoingon,andwhenhebidsintheauctionhewouldbidlowerthan5v1/6(abidof3v1/4willwork,giventheabove-mentionedassumptions),andthusincreasehisexpectedsurplus.TheobviousflawinthismechanismisthatA,BandCwillhaveincentivetolieinthisinitialphase;thiscouldstillbetrueifAwereobligedtopayBandCacertainamountiftheysatitoutandhewontheauction.

Theaboveprotocolisasimpleinstanceoftheclassbiddingclubs.Ingeneral,givensomeprimarymechanism(typically,anauction),abiddingclubprotocolisasfollows:

1.Somesetofbiddersareinvitedtojointhebiddingclub,andinformedofitsrules.Theotherbiddersarenotmadeawareoftheexistenceofthebiddingclub;weassumeherethattheyarenotevenawareofthepossibilityofitsexistence.

2.Thebiddershavethefreedomtojointheclubornot.Iftheydoitisassumedthattheyareguaranteedtofollowitsrules.

3.Thebidding-clubcoordinator(orsimply‘coordinator’)asksthemembersforcertainprivateinformation,suchastheirvaluationsforthegoodthatisbeingsold.Noticethatingeneralbiddersmaycheatabouttheirvaluations.

4.Thecoordinatordetermines,accordingtopre-specifiedrules,howthemembersshouldbehaveintheprimarymechanismbasedontheinformationtheyallsupply.

5.Thecoordinatormayalsodetermine(andenforce)additionalmonetarytransfersoftheclubmembers,basedontheresultsofthemainmechanism.

6.Thecoordinatoractsonlyasarepresentativeofbidders.

Itmayseemnaturaltoaskwhyacoordinatorshouldbewillingand/orabletofunctionasatrustedthirdparty,withoutattentionhavingbeenpaidtoitsownincentives.Webelievethatitisbestnottoseethecoordinatorasaparty(withinterestsofitsown)atall;rather,weconceiveofacoordinatorasasoftwareagentwhichisabletoactonlyaccordingtoits(commonly-known)programming.Itisthereforepossibleforthecoordinatortoactreliably—andforagentstobeconfidentthatthecoordinatorwillactreliably—evenincaseswherethecoordinatorstandstogainnothingthroughitsefforts.Wedoassumethatcoordinatorsshouldnotcostmoneytooperate—allofourcoordinatorsarebudget-balancedexceptforonethat(unavoidably!

)makesmoney.Finally,wehaveoftenbeenaskedaboutthelegalissuessurroundingtheuseofbiddingclubs.Whilethisisaninterestingandpertinentquestion,itexceedsbothourexpertiseandthescopeofthispaper.

Itturnsoutthat,whilethesimplemechanismoutlinedearlierfails,amoresophisticatedonewillensurethatBandCdonotparticipateintheprimaryauction,andthatAisthereforeassuredhigherexpectedpayoffintheauction.Moregenerally,thecontributionsofthispaperareasfollows:

1.Wepresentaprotocolforself-enforcingcooperationinsecond-priceauctionsforsubstitutegoods.

2.Wepresentaprotocolforself-enforcingcooperationinsecond-priceauctionsforcomplementarygoods.

3.Wepresentaprotocolforself-enforcingcollusioninfirst-price(aswellasDutch)auctions,inwhichonlysomeoftheagentscoordinatetheiractivities,andwhichdoesnotmakeanyuseofmonetarytransfers.

4.Wepresentaprotocolforself-enforcingcooperationingeneralauctionsandeconomicmechanisms,whentheagents’types(e.g.valuationsforgoods)aretakenfromafiniteset.

Thestrategicinteractionamongself-interestedagentsisaprimarytopicofstudyinmicroeconomics[4]andgametheory[1].Inparticular,thedesignofprotocolsforstrategicinteractionsisthesubjectofthefieldtermedmechanismdesign[1].Theroleofamechanism(inparticular,auction)designeristodefineagamewhoseequilibriumstrategiesaredesirableinsomerespectoranother.Thus,thedesignofabiddingclubconsistsoftakingagivenmechanism–theprimaryauction–andturningitintoamoreelaborateone,namelyonewithanaddedfirststageinwhichasubsetoftheplayersplayinsomenewly-designedgame(aswellassomeadditionalrulesregardingbehaviorintheprimaryauctionandpossiblesidepaymentsaftertheauction).

Researchonstrategicaspectsofmulti-agentactivityinArtificialIntelligencehasgrownrapidlyintherecentyears.Thisworkhasconcentratedonthedesignofprotocolsforagents’interaction[7,3,9],andsharesmuchincommonwithworkonmechanismdesignineconomics.Manyprinciplesandideasgrewupfromthemechanismdesignliterature,andhavebeenadaptedtotheAIcontext.

AlthoughthestudyofdealsamongagentshasreceivedmuchattentionintheAIliterature(seee.g.[7]),andalthoughthestudyanddesignofcontractsiscentraltoinformationeconomics[4](andreceivedmuchattentionintherecentAIliterature[8]),theliteratureoncooperationunderincompleteinformationinauctionsandtradesisquitelimited.Inparticular,theliteratureoncollusioninauctionsissomewhatspotty.Itisstilltoobroadtogiveacompleteoverviewofit,andthebulkofitisinformal.Intheformalliteratureonthetopic,theresultsarequitespecific,andcertainlydonotapplyinsettingsofparallelauctions(witheithersubstitutabilityorcomplementarityamonggoods),first-priceauctionswithoutside-payments,andgeneralmechanisms,whicharethefocusofourtechnicalresults.TheclosestresultfromtheliteratureofwhichweareawareisbyGrahamandMarshall[2],whopresentaprotocolforself-enforcingcollusionbyasubsetoftheparticipantsofa(single-good)second-priceauction.Wediscussthisresultbelow.Additionalrelatedstudyofcollusioninauctionscanbefoundin[5].

Wenowpresentsomepreliminariesofauctiontheory,aswellasadescriptionoftheclassicalauctionmodeldiscussedinthepaperandourparallelauctionmodel.

Inordertoanalyzeauctionswehavetodiscusstheinformationavailabletotheparticipants.Weassumetheindependentprivatevaluesmodel,withnoexternalities.Eachagentiisassumedtohaveavaluationviselectedfromtheintervalofrealnumbers[0;1]orfromafinitedomain,whichcapturesitsmaximalwillingnesstopayforthegood.Wefurtherassumethatthisvaluationisselectedfromtheuniformdistributionontheinterval[0;1]oronafinitedomain.Foreaseofpresentationwewillassumethecontinuouscase,excludingthesectionongeneralmechanisms,wheretheassumptionthatthesetofpossiblevaluationsisfiniteisrequiredforourresult.Ifagentiobtainsthegoodandisaskedtopayp,aswellasaparticipationfeed,thenitsutility,ui,isgivenbyvi.p.d;otherwise,ifitisnotassignedanygoodthenitsutilityis.d;iftheagentdoesnotparticipateintheauctionthenits

展开阅读全文
相关资源
猜你喜欢
相关搜索
资源标签

当前位置:首页 > 经管营销 > 经济市场

copyright@ 2008-2023 冰点文库 网站版权所有

经营许可证编号:鄂ICP备19020893号-2