Sunteți pe pagina 1din 12

QuantumEntanglementandInformation

FirstpublishedMonAug13,2001;substantiverevisionFriJan20,2006 Quantumentanglementisaphysicalresource,likeenergy,associatedwiththepeculiarnonclassical correlationsthatarepossiblebetweenseparatedquantumsystems.Entanglementcanbemeasured, transformed,andpurified.Apairofquantumsystemsinanentangledstatecanbeusedasaquantum informationchanneltoperformcomputationalandcryptographictasksthatareimpossibleforclassical systems.Thegeneralstudyoftheinformationprocessingcapabilitiesofquantumsystemsisthesubject ofquantuminformation. 1.QuantumEntanglement 2.ExploitingEntanglement:QuantumTeleportation 3.QuantumInformation 4.QuantumCryptography 5.QuantumComputation 6.InterpretativeRemarks Bibliography OtherInternetResources RelatedEntries

1.QuantumEntanglement
In1935and1936,SchrdingerpublishedatwopartarticleintheProceedingsoftheCambridge PhilosophicalSocietyinwhichhediscussedandextendedaremarkableargumentbyEinstein, Podolsky,andRosen.TheEinsteinPodolskyRosen(EPR)argumentwas,inmanyways,the culminationofEinstein'scritiqueoftheorthodoxCopenhageninterpretationofquantummechanics, andwasdesignedtoshowthatthetheoryisincomplete.(SeetheentriesontheEPRargumentandon theCopenhageninterpretation.)Inclassicalmechanicsthestateofasystemisessentiallyalistofthe system'spropertiesmoreprecisely,itisthespecificationofasetofparametersfromwhichthelistof propertiescanbereconstructed:thepositionsandmomentaofalltheparticlescomprisingthesystem (orsimilarparametersinthecaseoffields).Thedynamicsofthetheoryspecifieshowproperties changeintermsofalawofevolutionforthestate.Paulicharacterizedthismodeofdescriptionof physicalsystemsasadetachedobserveridealization.SeePauli'slettertoBorninTheBornEinstein Letters(Born,1992;p.218).OntheCopenhageninterpretation,suchadescriptionisnotpossiblefor quantumsystems.Instead,thequantumstateofasystemshouldbeunderstoodasacatalogueofwhat anobserverhasdonetothesystemandwhathasbeenobserved,andtheimportofthestatethenliesin theprobabilitiesthatcanbeinferred(intermsofthetheory)fortheoutcomesofpossiblefuture observationsonthesystem.Einsteinrejectedthisviewandproposedaseriesofargumentstoshowthat thequantumstateissimplyanincompletecharacterizationofthesystem.Themissingparametersare sometimesreferredtoashiddenparametersorhiddenvariables(althoughEinsteindidnotusethis terminology,presumablybecausehedidnotwanttoendorseanyparticularhiddenvariabletheory).

ItshouldnotbesupposedthatEinstein'sdefinitionofacompletetheoryincludedtherequirementthatit bedeterministic.Rather,herequiredcertainconditionsofseparabilityandlocalityforcomposite systemsconsistingofseparatedcomponentsystems:eachcomponentsystemseparatelyshouldbe characterizedbyitsownproperties(evenifthesepropertiesmanifestthemselvesstochastically),andit shouldbeimpossibletoalterthepropertiesofadistantsysteminstantaneously(ortheprobabilitiesof theseproperties)byactingonalocalsystem.InlateranalysesnotablyinBell'sextensionoftheEPR argumentitbecameapparentthattheseconditions,suitablyformulatedasprobabilityconstraints, areequivalenttotherequirementthatstatisticalcorrelationsbetweenseparatedsystemsshouldbe reducibletoprobabilitydistributionsovercommoncauses(deterministicorstochastic)inthesenseof Reichenbach.(SeetheentriesonBell'stheoremandonReichenbach'scommoncauseprinciple.) IntheoriginalEPRarticle,twoparticlesarepreparedfromasourceinacertainquantumstateandthen moveapart.Therearematchingcorrelationsbetweenboththepositionsofthetwoparticlesandtheir momenta:ameasurementofeitherpositionormomentumonaparticularparticlewillallowthe prediction,withcertainty,oftheoutcomeofapositionmeasurementormomentummeasurement, respectively,ontheotherparticle.Thesemeasurementsaremutuallyexclusive:eitheraposition measurementcanbeperformed,oramomentummeasurement,butnotbothsimultaneously.Either correlationcanbeobserved,butthesubsequentmeasurementofmomentum,say,afterestablishinga positioncorrelation,willnolongeryieldanycorrelationinthemomentaofthetwoparticles.Itisasif thepositionmeasurementdisturbsthecorrelationbetweenthemomentumvalues.Thepuzzleisthatthe quantumstateoftheparticlepairisinconsistentwiththeassignmentoflabelstotheparticlesseparately thatcouldbeassociatedwithappropriatelycorrelatedvaluesfortheoutcomesofpositionand momentummeasurements.Theselabelswouldbethecommoncausesofthecorrelations,andwould provideanexplanationofthecorrelationsintermsoftheinitialcorrelationsbetweenthepropertiesof thetwosystemsattheirsource.EPRconcludedthatthequantumstatewasincomplete. HereishowSchrdingerputthepuzzleinthefirstpartofhistwopartarticle(Schrdinger,1935;p. 559): YetsinceIcanpredicteitherx1orp1withoutinterferingwiththesystemNo.1andsince systemNo.1,likeascholarinanexamination,cannotpossiblyknowwhichofthetwo questionsIamgoingtoaskfirst:itsoseemsthatourscholarispreparedtogivetheright answertothefirstquestionheisasked,anyhow.Thereforehemustknowbothanswers; whichisanamazingknowledge;quiteirrespectiveofthefactthatafterhavinggivenhis firstanswerourscholarisinvariablysodisconcertedortiredout,thatallthefollowing answersarewrong. WhatSchrdingershowedwasthatiftwoparticlesarepreparedinaquantumstatesuchthatthereisa matchingcorrelationbetweentwocanonicallyconjugatedynamicalquantitiesquantitieslike positionandmomentumwhosevaluessufficetospecifyallthepropertiesofaclassicalsystemthen thereareinfinitelymanydynamicalquantitiesofthetwoparticlesforwhichthereexistsimilar matchingcorrelations:everyfunctionofthecanonicallyconjugatepairofthefirstparticlematches withthesamefunctionofthecanonicallyconjugatepairofthesecondparticle.Thus(Schrdinger,p. 559)systemNo.1doesnotonlyknowthesetwoanswersbutavastnumberofothers,andthatwithno mnemotechnicalhelpwhatsoever,atleastwithnonethatweknowof. Schrdingercoinedthetermentanglementtodescribethispeculiarconnectionbetweenquantum systems(Schrdinger,1935;p.555):

Whentwosystems,ofwhichweknowthestatesbytheirrespectiverepresentatives,enter intotemporaryphysicalinteractionduetoknownforcesbetweenthem,andwhenaftera timeofmutualinfluencethesystemsseparateagain,thentheycannolongerbedescribed inthesamewayasbefore,viz.byendowingeachofthemwitharepresentativeofitsown.I wouldnotcallthatonebutratherthecharacteristictraitofquantummechanics,theonethat enforcesitsentiredeparturefromclassicallinesofthought.Bytheinteractionthetwo representatives[thequantumstates]havebecomeentangled. Headded(Schrdinger,1935;p.555): Anotherwayofexpressingthepeculiarsituationis:thebestpossibleknowledgeofawhole doesnotnecessarilyincludethebestpossibleknowledgeofallitsparts,eventhoughthey maybeentirelyseparateandthereforevirtuallycapableofbeingbestpossiblyknown,i.e., ofpossessing,eachofthem,arepresentativeofitsown.Thelackofknowledgeisbyno meansduetotheinteractionbeinginsufficientlyknownatleastnotinthewaythatit couldpossiblybeknownmorecompletelyitisduetotheinteractionitself. Attentionhasrecentlybeencalledtotheobviousbutverydisconcertingfactthateven thoughwerestrictthedisentanglingmeasurementstoonesystem,therepresentative obtainedfortheothersystemisbynomeansindependentoftheparticularchoiceof observationswhichweselectforthatpurposeandwhichbythewayareentirelyarbitrary.It isratherdiscomfortingthatthetheoryshouldallowasystemtobesteeredorpilotedinto oneortheothertypeofstateattheexperimenter'smercyinspiteofhishavingnoaccessto it. Inthesecondpartofthepaper,Schrdingershowedthat,ingeneral,asophisticatedexperimentercan, byasuitablechoiceofoperationscarriedoutononesystem,steerthesecondsystemintoanychosen mixtureofquantumstates.Thatis,thesecondsystemcannotbesteeredintoanyparticularstateatthe whimoftheexperimenter,buttheexperimentercanconstrainthestateintowhichthesecondsystem evolvestolieinanychosensetofstates,withaprobabilitydistributionfixedbytheentangledstate.He foundthisconclusionsufficientlyunsettlingtosuggestthattheentanglementbetweentwoseparating systemswouldpersistonlyfordistancessmallenoughthatthetimetakenbylighttotravelfromone systemtotheothercouldbeneglected,comparedwiththecharacteristictimeperiodsassociatedwith otherchangesinthecompositesystem.Hespeculatedthatforlongerdistanceseachofthetwosystems mightinfactbeinastateassociatedwithacertainmixture,determinedbythepreciseformofthe entangledstate. MostphysicistsattributedthepuzzlingfeaturesofentangledquantumstatestoEinstein'sinappropriate detachedobserverviewofphysicaltheory,andregardedBohr'sreplytotheEPRargument(Bohr, 1935)asvindicatingtheCopenhageninterpretation.Thiswasunfortunate,becausethestudyof entanglementwasignoredforthirtyyearsuntilJohnBell'sreconsiderationandextensionoftheEPR argument(Bell,1964).BelllookedatentanglementinsimplersystemsthantheEPRcase:matching correlationsbetweentwovalueddynamicalquantities,suchaspolarizationorspin,oftwoseparated systemsinanentangledstate.WhatBellshowedwasthatthestatisticalcorrelationsbetweenthe measurementoutcomesofsuitablychosendifferentquantitiesonthetwosystemsareinconsistentwith aninequalityderivablefromEinstein'sseparabilityandlocalityassumptionsineffectfromthe assumptionthatthecorrelationshaveacommoncause.

Bell'sinvestigationgeneratedanongoingdebateonthefoundationsofquantummechanics.One importantfeatureofthisdebatewasconfirmationthatentanglementcanpersistoverlongdistances(see Aspectetal.),thusfalsifyingSchrdinger'ssuppositionofthespontaneousdecayofentanglementas twoentangledparticlesseparate.Butitwasnotuntilthe1980sthatphysicists,computerscientists,and cryptographersbegantoregardthenonlocalcorrelationsofentangledquantumstatesasanewkindof nonclassicalresourcethatcouldbeexploited,ratherthananembarrassmenttobeexplainedaway.For furtherdiscussionofentanglementasaphysicalresource,includingmeasuringentanglement,andthe manipulationandpurificationofentanglementbylocaloperations,see"TheJoyofEntanglement"by PopescuandRohrlichinLo,Popescu,andSpiller(1998),orNielsenandChuang(2000).

2.ExploitingEntanglement:QuantumTeleportation
ConsideragainSchrdinger'srealizationthatanentangledstatecouldbeusedtosteeradistantparticle intooneofasetofstates,withacertainprobability.Infact,thispossibilityofremotesteeringiseven moredramaticthanSchrdingerdemonstrated.SupposeAliceandBobshareanentangledstateofthe sortconsideredbyBell,saytwophotonsinanentangledstateofpolarization.Thatis,Alicehasinher possessiononeoftheentangledphotons,andBobtheother.SupposethatAlicehasanadditional photoninanunknownstateofpolarization,u.ItispossibleforAlicetoperformanoperationonthe twophotonsinherpossessionthatwilltransformBob'sphotonintooneoffourstates,dependingonthe fourpossible(random)outcomesofAlice'soperation:eitherthestateu,orastatethatisrelatedtouin adefiniteway.Alice'soperationentanglesthetwophotonsinherpossession,anddisentanglesBob's photon,steeringitintoastateu*.AfterAlicecommunicatestheoutcomeofheroperationtoBob,Bob knowseitherthatu*=u,orhowtotransformu*toubyalocaloperation.Thisphenomenonisknown asquantumteleportation. WhatisextraordinaryaboutthisphenomenonisthatAliceandBobhavemanagedtousetheirshared entangledstateasaquantumcommunicationchanneltodestroythestateuofaphotoninAlice'spartof theuniverseandrecreateitinBob'spartoftheuniverse.Sincethestateofaphotonrequiresspecifying adirectioninspace(essentiallythevalueofananglethatcanvarycontinuously),withoutashared entangledstateAlicewouldhavetoconveyaninfiniteamountofclassicalinformationtoBobforBob tobeabletoreconstructthestateuprecisely.Toseewhythisisso,considerthatthedecimalexpansion ofananglevariablerepresentedbyarealnumberisrepresentedbyapotentiallyinfinitesequenceof digitsbetween0and9.Thebinaryexpansionisrepresentedbyapotentiallyinfinitesequenceof0'sand 1's.EversinceShannonformalizedthenotionofclassicalinformation,theamountofclassical informationassociatedwithabinaryalternative(representedas0or1),whereeachalternativehas equalaprioriprobability,ismeasuredasonebinarydigitorbit.Sotospecifythevalueofanarbitrary anglevariablerequiresaninfinitenumberofbits.TospecifytheoutcomeofAlice'soperation,which hasfourpossibleoutcomes,withequalaprioriprobabilities,requirestwobitsofclassicalinformation. Remarkably,Bobcanreconstructthestateuonthebasisofjusttwobitsofclassicalinformation communicatedbyAlice,apparentlybyexploitingtheentangledstateasaquantumcommunication channeltotransfertheremaininginformation.Forfurtherdiscussionofquantumteleportation,see NielsenandChuang(2000),orRichardJosza'sarticle"QuantumInformationanditsProperties"inLo, Popescu,andSpiller(1998).

3.QuantumInformation
Formally,theamountofclassicalinformationwegain,onaverage,whenwelearnthevalueofa randomvariable(or,equivalently,theamountofuncertaintyinthevalueofarandomvariablebefore welearnitsvalue)isrepresentedbyaquantitycalledtheShannonentropy,measuredinbits(Shannon andWeaver,1949).Arandomvariableisdefinedbyaprobabilitydistributionoverasetofvalues.In thecaseofabinaryrandomvariable,withequalprobabilityforeachofthetwopossibilities,the Shannonentropyis1bit,representingmaximaluncertainty.Forallotherprobabilitiesintuitively, representingsomeinformationaboutwhichalternativeismorelikelytheShannonentropyisless than1.Forthecaseofmaximalknowledgeorzerouncertaintyaboutthealternatives,wherethe probabilitiesare0and1,theShannonentropyiszero.(Notethatthetermbitisusedtorefertothe basicunitofclassicalinformationintermsofShannonentropy,andtoanelementarytwostateclassical systemconsideredasrepresentingthepossibleoutputsofanelementaryclassicalinformationsource.) Sinceinformationisalwaysembodiedinthestateofaphysicalsystem,wecanalsothinkofthe Shannonentropyasquantifyingthephysicalresourcesrequiredtostoreclassicalinformation.Suppose AlicewishestocommunicatesomeclassicalinformationtoBoboveraclassicalcommunication channelsuchasatelephoneline,sayanemailmessage.Arelevantquestionconcernstheextentto whichthemessagecanbecompressedwithoutlossofinformation,sothatBobcanreconstructthe originalmessageaccuratelyfromthecompressedversion.AccordingtoShannon'ssourcecoding theoremornoiselesscodingtheorem(assuminganoiselesstelephonelinewithnolossofinformation), theminimalphysicalresourcerequiredtorepresentthemessage(effectively,alowerboundonthe possibilityofcompression)isgivenbytheShannonentropyofthesource. Whathappensifweusethequantumstatesofphysicalsystemstostoreinformation,ratherthan classicalstates?Itturnsoutthatquantuminformationisradicallydifferentfromclassicalinformation. Theunitofquantuminformationisthequbit,representingtheamountofquantuminformationthat canbestoredinthestateofthesimplestquantumsystem,forexample,thepolarizationstateofa photon.ThetermisduetoSchumacher(1995),whoprovedaquantumanalogueofShannon'snoiseless codingtheorem.(Byanalogywiththetermbit,thetermqubitreferstothebasicunitofquantum informationintermsofthevonNeumannentropy,andtoanelementarytwostatequantumsystem consideredasrepresentingthepossibleoutputsofanelementaryquantuminformationsource.)Aswe haveseen,anarbitrarilylargeamountofclassicalinformationcanbeencodedinaqubit.This informationcanbeprocessedandcommunicatedbut,becauseofthepeculiaritiesofquantum measurement,atmostonebitcanbeaccessed!AccordingtoatheorembyHolevo,theaccessible informationinaprobabilitydistributionoverasetofalternativequbitstatesislimitedbythevon Neumannentropy,whichisequaltotheShannonentropyonlywhenthestatesareorthogonalinthe spaceofquantumstates,andisotherwiselessthantheShannonentropy. Whileclassicalinformationcanbecopiedorcloned,thequantumnocloningtheorem(Dieks,1982; WoottersandZurek,1982)assertstheimpossibilityofcloninganunknownquantumstate.Toseewhy, considerhowwemightconstructaclassicalcopyingdevice.ANOTgateisadevicethattakesabitas inputandproducesasoutputeithera1iftheinputis0,ora0iftheinputis1.Inotherwords,aNOT gateisa1bitgatethatflipstheinputbit.AcontrolledNOTgate,orCNOTgate,takestwobitsas inputs,acontrolbitandatargetbit,andflipsthetargetbitifandonlyifthecontrolbitis1,while reproducingthecontrolbit.(Sotherearetwoinputs,thecontrolandtarget,andtwooutputs:the control,andeitherthetargetortheflippedtarget,dependingonthevalueofthecontrol.)ACNOTgate functionsasacopyingdeviceforthecontrolbitifthetargetbitissetto0,becausetheoutputofthe

targetbitisthenacopyofthecontrolbit(i.e.,theinput00producesoutput00,andtheinput10 producesoutput11).Insofaraswecanthinkofameasurementassimplyacopyingoperation,aCNOT gateistheparadigmofaclassicalmeasuringdevice.(ImagineAliceequippedwithsuchadevice,with inputandoutputcontrolandtargetwires,measuringthepropertiesofanunknownclassicalworld.The inputcontrolwireisaprobeforthepresenceofabsenceofaproperty,representedbya1ora0.The targetwirefunctionsasthepointer,whichisinitiallysetto0.Theoutputofthetargetisa1ora0, dependingonthepresenceorabsenceoftheproperty.) SupposeweattempttouseourCNOTgatetocopyanunknownqubitstate.Sincewearenowproposing toregardtheCNOTgateasadeviceforprocessingquantumstates,theevolutionfrominputstatesto outputstatesmustbeeffectedbyaphysicalquantumtransformation.Nowquantumtransformationsare linearonthelinearstatespaceofqubits.Linearityofthestatespacemeansthatforanytwoqubitstates callthem0and1thatareorthogonalinthespaceofqubitstates,therearequbitstatesthatare representedbylinearsuperpositionsorsumsof0and1,withcertaincoefficients.Suchsuperpositions e.g.,asuperpositionwithequalcoefficientsthatcouldberepresentedsymbolicallyas0+1are nonorthogonalto0andto1.Linearityofthetransformationmeansthatanytransformationmusttakea qubitstaterepresentedbythesumoftwoorthogonalqubitstoanewqubitstatethatisthesumofthe transformedorthogonalqubits.IftheCNOTgatesucceedsincopyingtwoorthogonalqubits,itcannot succeedincopyingalinearsuperpositionofthesequbits.Sincethegatefunctionslinearly,itmust insteadproduceastatethatisalinearsuperpositionoftheoutputsobtainedforthetwoorthogonal qubits.Thatistosay,theoutputofthegatewillberepresentedbyaquantumstatethatisasumoftwo terms,wherethefirsttermrepresentstheoutputofthecontrolandtargetforthefirstorthogonalqubit, andthesecondtermrepresentstheoutputofthecontrolandtargetforthesecondorthogonalqubit.This couldbewrittenas00+11.Thisisanentangledstateandnottheoutputthatwouldberequiredbya successfulcopyingoperation,wherethecontrolandtargeteachoutputsthesuperposedqubit(which couldbewrittenas(0+1)(0+1)).

4.QuantumCryptography
Linearitypreventsthepossibilityofcloningormeasuringanunknownquantumstate.Similarly,itcan beshownthatifAlicesendsBoboneoftwononorthogonalqubits,Bobcanobtaininformationabout whichofthesequbitswassentonlyattheexpenseofdisturbingthestate.Ingeneral,forquantum informationthereisnoinformationgainwithoutdisturbance.Theimpossibilityofcopyinganunknown quantumstate,orastatethatisknowntobelongtoasetofnonorthogonalstateswithacertain probability,andtheexistenceofatradeoffrelationbetweeninformationgainandstatedisturbance,is thebasisoftheapplicationofquantuminformationtocryptography.Therearequantumprotocols involvingtheexchangeofclassicalandquantuminformationthatAliceandBobcanexploittosharea secretrandomkey,whichtheycanthenusetocommunicateprivately.(SeeLo'sarticle"Quantum Cryptology"inLo,Popescu,andSpiller,1998.)Anyattemptbyaneavesdropper,Eve,tomonitorthe communicationbetweenAliceandBobwillbedetectable,inprinciple,becauseEvecannotgainany quantuminformationwithoutsomedisturbancetothequantumcommunicationchannel.Moreover,the nocloningtheoremprohibitsEvefromcopyingthequantumcommunicationsandprocessingthem offline,sotospeak,aftershemonitorstheclassicalcommunicationbetweenAliceandBob. Whilethedifferencebetweenclassicalandquantuminformationcanbeexploitedtoachievesuccessful keydistribution,thereareothercryptographicprotocolsthatarethwartedbyquantumentanglement. Bitcommitmentisakeycryptographicprotocolthatcanbeusedasasubroutineinavarietyof

importantcryptographictasks.Inabitcommitmentprotocol,AlicesuppliesanencodedbittoBob.The informationavailableintheencodingshouldbeinsufficientforBobtoascertainthevalueofthebit,but sufficient,togetherwithfurtherinformationsuppliedbyAliceatasubsequentstagewhensheis supposedtorevealthevalueofthebit,forBobtobeconvincedthattheprotocoldoesnotallowAliceto cheatbyencodingthebitinawaythatleavesherfreetorevealeither0or1atwill. Toillustratetheidea,supposeAliceclaimstheabilitytopredictadvancesordeclinesinthestock marketonadailybasis.Tosubstantiateherclaimwithoutrevealingvaluableinformation(perhapstoa potentialemployer,Bob)shesuggeststhefollowingdemonstration:Sheproposestorecordher prediction,beforethemarketopens,bywritinga0(fordecline)ora1(foradvance)onapieceof paper,whichshewilllockinasafe.ThesafewillbehandedtoBob,butAlicewillkeepthekey.Atthe endoftheday'strading,shewillannouncethebitshechoseandprovethatsheinfactmadethe commitmentattheearliertimebyhandingBobthekey.Ofcourse,thekeyandsafeprotocolisnot provablysecurefromcheatingbyBob,becausethereisnoprincipleofclassicalphysicsthatthat preventsBobfromopeningthesafeandclosingitagainwithoutleavinganytrace.Thequestionis whetherthereexistsaquantumanalogueofthisprocedurethatisunconditionallysecure:provably securebythelawsofphysicsagainstcheatingbyeitherAliceorBob.Bobcancheatifhecanobtain someinformationaboutAlice'scommitmentbeforesherevealsit(whichwouldgivehimanadvantage inrepetitionsoftheprotocolwithAlice).Alicecancheatifshecandelayactuallymakinga commitmentuntilthefinalstagewhensheisrequiredtorevealhercommitment,orifshecanchange hercommitmentatthefinalstagewithaverylowprobabilityofdetection. Itturnsoutthatunconditionallysecuretwopartybitcommitment,basedsolelyontheprinciplesof quantumorclassicalmechanics(withoutexploitingspecialrelativisticsignallingconstraints,or principlesofgeneralrelativityorthermodynamics)isimpossible.SeeMayers(1997),LoandChau (1997)andLo'sarticle"QuantumCryptology"inLo,Popescu,anSpiller(1998)forfurtherdiscussion. NotethatKent(1999)hasshownthatonecanimplementasecureclassicalbitcommitmentprotocolby exploitingrelativisticsignallingconstraintsinatimedsequenceofcommunicationsbetweenverifiably separatedsitesforbothAliceandBob.)Roughly,theimpossibilityarisesbecauseatanystepinthe protocolwhereeitherAliceorBobisrequiredtomakeadeterminatechoice(performameasurement onaparticleinthequantumchannel,chooserandomlyandperhapsconditionallybetweenasetof alternativeactionstobeimplementedontheparticleinthequantumchannel,etc.),thechoicecan delayedbyentanglingoneormoreancilla(helper)particleswiththechannelparticleinan appropriateway.Bysuitableoperationsontheancillas,thechannelparticlecanbesteeredsothatthis cheatingstrategyisundetectable.Ineffect,ifBobcanobtainnoinformationaboutthebitinthesafe, thenentanglementwillallowAlicetosteerthebittoeither0or1atwill.

5.QuantumComputation
Quantuminformationcanbeprocessed,buttheaccessibilityofthisinformationislimitedbythe Holevobound(mentionedinSection3).DavidDeutsch(1985)firstshowedhowtoexploitquantum entanglementtoperformacomputationaltaskthatisimpossibleforaclassicalcomputer.Supposewe haveablackboxthatevaluatesafunctionf.Theargumentsoff(inputs)areeither0or1.Thevalues (outputs)off(whicharealso0or1)areeitherthesameforbotharguments(inwhichcasefis constant),ordifferentforthetwoarguments(inwhichcasefissaidtobebalanced).Weareinterested indeterminingwhetherfisconstantorbalanced.Now,classically,theonlywaytodothisistorunthe blackboxtwice,forbotharguments0and1,andtopassthevalues(outputsoff)toacircuitthat

determineswhethertheyarethesame(forconstant)ordifferent(forbalanced).Deutschshowedthat ifweusequantumstatesandquantumgatestostoreandprocessinformation,thenwecandetermine whetherfisconstantorbalancedinoneevaluationofthefunctionf.Thetrickistodesignthecircuit (thesequenceofgates)toproducetheanswertoaglobalquestionaboutthefunction(constantor balanced)inanoutputqubitregisterthatcanthenbereadoutormeasured. ConsideragainthequantumCNOTgate,withtwoorthogonalqubits0and1aspossibleinputsforthe control,and0astheinputforthetarget.Onecanthinkoftheinputcontrolandoutputtargetqubits, respectively,astheargumentandassociatedvalueofafunction.ThisCNOTfunctionassociatesthe value0withtheargument0andthevalue1withtheargument1.Foralinearsuperpositionofthe orthogonalqubits,say0+1,asinputtothecontrol,andthequbitrepresenting0astheinputtothe target,theoutputistheentangledstate00+11,alinearsuperpositioninwhichthefirsttermrepresents theargument0andassociatedvalue(0)oftheCNOTfunction,andthesecondtermrepresentsthe argument1andassociatedvalue(1)oftheCNOTfunction.Theentangledstaterepresentsallpossible argumentsandcorrespondingvaluesofthefunctionasalinearsuperposition,butthisinformationis notaccessible.Whatcanbeshowntobeaccessible,byasuitablechoiceofquantumgates,is informationaboutwhetherornotthefunctionhascertainglobalproperties.Thisinformationis obtainablewithoutreadingouttheevaluationofanyindividualargumentsandvalues.(Indeed, accessinginformationintheentangledstateaboutaglobalpropertyofthefunctionwilltypically requirelosingaccesstoallinformationaboutindividualargumentsandvalues.) ThesituationisanalogousforDeutsch'sfunctionf.Heretheoutputoffcanberepresentedaseither00 +10or01+11(intheconstantcase),or00+11or01+10(inthebalancedcase).Thetwo entangledstatesintheconstantcaseareorthogonalinthe4dimensionaltwoqubitstatespaceand spanaplane.Callthistheconstantplane.Similarly,thetwoentangledstatesinthebalancedcase spanaplane,thebalancedplane.Theseplanesareorthogonalinthe4dimensionalstatespace,except foranoverlap:aline,representinga(nonentangled)twoqubitstate.Itisthereforepossibletodesigna measurementtodistinguishthetwoglobalpropertiesoff,constantorbalanced,withacertain probability(actually,1/2)offailure,whenthemeasurementyieldsanoutcomecorrespondingtothe overlapstate,whichiscommontothetwocases.Nevertheless,onlyonequeryofthefunctionis requiredwhenthemeasurementsucceedsinidentifyingtheglobalproperty.Withajudiciouschoiceof quantumgates,itisevenpossibletodesignaquantumcircuitthatalwayssucceedsindistinguishingthe twocases. Deutsch'sexampleshowshowquantuminformation,andquantumentanglement,canbeexploitedto computeaglobalpropertyofafunctioninonestepthatwouldtaketwostepsclassically.Thereare quantumalgorithmsthatachieveanexponentialspeedupoveranyknownclassicalalgorithm,andin somecasesthespeedupcanbeshowntobeexponentialoveranyclassicalalgorithm.Essentially,this isagainduetothephenomenonofentanglement.Indeed,theamountofinformationrequiredto describeageneralentangledstateofnqubitsgrowsexponentiallywithn.Thestatespace(Hilbert space)has2ndimensions,soageneralentangledstateisasuperpositionof2nnqubitstates.In classicalmechanicstherearenoentangledstates:ageneralnbitcompositesystemcanbedescribed withjustntimestheamountofinformationrequiredtodescribeasinglebitsystem.Sotheclassical simulationofaquantumprocesswouldinvolveanexponentialincreaseintheclassicalinformational resourcerequiredtorepresentthequantumstate,asthenumberofqubitsthatbecomeentangledinthe evolutiongrowslinearly,andtherewouldbeacorrespondingexponentialslowdownincalculatingthe evolution,comparedtotheactualquantumcomputationperformednaturallybythesystem.Still,there isasignificantlackofconsensusastowhatexactlyexplainsthespeedup.Foradiscussion,SeeBub

(2006,Section6.3). WhileDeutsch'sproblemisinasensetrivialandhasnointerestingapplication,therenowexistseveral quantumalgorithmsfornontrivialproblems,notablyShor'sfactorizationalgorithmforfactoringlarge compositeintegersinpolynomialtime(withdirectapplicationtopublickeycryptography,awidely usedclassicalcryptographicscheme)andGrover'sdatabasesearchalgorithm.SeeNielsenandChuang (2000),Barenco'sarticle"QuantumComputation:AnIntroduction"inLo,Popescu,andSpiller(1998) fordetails,Bub(2006,Section6),aswellastheplannedentryonquantumcomputation.

6.InterpretativeRemarks
Deutsch(1997)hasarguedthattheexponentialspeedupinquantumcomputation,andingeneralthe wayaquantumsystemprocessesinformation,canonlybeproperlyunderstoodwithintheframeworkof thesocalledmanyworldsinterpretation,originatingwithEverett.Theidea,roughly,isthatan entangledstateofthesortthatarisesinthequantumcomputationofafunction,whichrepresentsa linearsuperpositionoverallpossibleargumentsandcorrespondingvaluesofthefunction,shouldbe understoodasamanifestationofparallelcomputationsindifferentworlds.Thequantumcircuitis designedtoenablethecomputationofaglobalpropertyofthefunctionbyachievingsomesortof interferencebetweenthesedifferentworlds.Foraninsightfulcritiqueofthisideaofquantum parallelismasexplanatory,seeSteane(2003).Itshouldbenotedthatthetermmanyworldscanrefer toavarietyofinterpretationalideas,somemorerefinedthanothers.Forasophisticatedaccount,see Wallace(2003). Analternativeview,notmuchdiscussedintheliteratureinthisconnection,isthequantumlogical approach,whichemphasizesthenonBooleanstructureofpropertiesofquantumsystems.(The propertiesofaclassicalsystemformaBooleanalgebra,essentiallytheabstractcharacterizationofa settheoreticstructure.ThisisreflectedintheBooleancharacterofclassicallogic,andtheBoolengates inaclassicalcomputer.)Acrucialdifferencebetweenquantumandclassicalinformationisthe possibilityofcomputingthetruthvalueofanexclusivedisjunctionforexample,theconstant disjunctionassertingthatthevalueofthefunction(forbotharguments)iseither0or1,orthe balanceddisjunctionassertingthatthevalueofthefunction(forbotharguments)iseitherthesameas theargumentordifferentfromtheargumentwithoutcomputingthetruthvaluesofthedisjuncts. Classically,anexclusivedisjunctionistrueifandonlyifoneofthedisjunctsistrue.Ineffect,Deutsch's quantumcircuitachievesitsspeedupbyexploitingthenonBooleanstructureofquantumpropertiesto computethevalueofadisjunctiveproperty,withoutcomputingthevalueofthedisjuncts(representing theassociationofindividualargumentswithcorrespondingfunctionvalues).Forsomerecentworkby Giuntiniandothersonlogicsassociatedwithquantumgates,seeunderquantumcomputationallogics intheOtherInternetResources.(Forquantumlogicnotspecificallyinrelationtoquantum computation,seetheentryonquantumlogicandquantumprobability). Someresearchersinquantuminformationandquantumcomputationhavearguedforaninformation theoreticinterpretationofquantummechanics.Inhisreviewarticleonquantumcomputation,Andrew Steane(1998,p.119)makesthefollowingradicalsuggestion: Historically,muchoffundamentalphysicshasbeenconcernedwithdiscoveringthe fundamentalparticlesofnatureandtheequationswhichdescribetheirmotionsand interactions.Itnowappearsthatadifferentprogrammemaybeequallyimportant:to discoverthewaysthatnatureallows,andprevents,informationtobeexpressedand manipulated,ratherthanparticlestomove.

Steaneconcludeshisreviewwiththefollowingproposal(1998,p.171): Toconcludewith,Iwouldliketoproposeamorewiderangingtheoreticaltask:toarriveat asetofprincipleslikeenergyandmomentumconservation,butwhichapplytoinformation, andfromwhichmuchofquantummechanicscouldbederived.Twotestsofsuchideas wouldbewhethertheEPRBellcorrelationsthusbecametransparent,andwhetherthey renderedobvioustheproperuseoftermssuchasmeasurementandknowledge. Inlinewiththisproposal,Clifton,Bub,andHalvorson(2003)haveshownthatonecanderivethebasic kinematicfeaturesofaquantumdescriptionofphysicalsystemsfromthreefundamentalinformation theoreticconstraints: theimpossibilityofsuperluminalinformationtransferbetweentwophysicalsystemsby performingmeasurementsononeofthem theimpossibilityofperfectlybroadcastingtheinformationcontainedinanunknownphysical state(which,forpurestates,amountstonocloning) theimpossibilityofcommunicatinginformationsoastoimplementabitcommitmentprotocol withunconditionalsecurity Moreprecisely,theanalysisiscarriedoutinanalgebraicframeworkwhichallowsamathematically abstractcharacterizationofaphysicaltheorythatincludes,asspecialcases,allclassicalmechanical theoriesofbothwaveandparticlevarieties,andallvariationsonquantumtheory,includingquantum fieldtheories(plusanyhybridsofthesetheories,suchastheorieswithsuperselectionrules).Within thisframework,thethreeinformationtheoreticconstraintsareshowntojointlyentailthreephysical conditionsthataretakenasdefinitiveofwhatitmeanstobeaquantumtheoryinthemostgeneral sense,specificallythat: thealgebrasofobservablespertainingtodistinctphysicalsystemscommute(aconditionusually calledmicrocausalityorkinematicindependence) anyindividualsystem'salgebraofobservablesisnoncommutative thephysicalworldisnonlocal,inthatspacelikeseparatedsystemscanoccupyentangledstates thatpersistasthesystemsseparate ForfurtherdiscussionofthesignificanceoftheCliftonBubHalvorsonresult,seeBub(2004,2005, 2006).Forsomeinterestingbutonlyapparentcounterexamples,seeSmolin(2005)andSpekkens (2004),andthecommentarybyHalvorsonandBub(2005).SeeBruknerandZeilinger(2002)fora differentinformationtheoreticapproachtoquantummechanics,andFuchs(2002)foraradically Bayesianperspective.ForaninsightfulanalysisandcritiqueoftheBruknerZeilingerposition,see Timpson(2004).

Bibliography
Aspect,A.,Grangier,P.,Roger,G.(1982)"ExperimentalTestsofBell'sInequalitiesUsing TimeVaryingAnalyzers"PhysicalReviewLetters49:18041807 Bell,J.S.(1964)"OntheEinsteinPodolskyRosenParadox"Physics1:195200 Bennett,C.H.,DiVicenzo,B.D.(2000)"QuantumInformationandComputation"Nature404: 247255 Bohr,N.(1935)"CanQuantumMechanicalDescriptionofPhysicalRealitybeConsidered

Complete?"PhysicalReview38:696702 Born,M.(ed.)(1992)TheBornEinsteinLetters(Dordrecht:Reidel) Brukner,C.,Zeilinger,A.(2002)"InformationandFundamentalElementsoftheStructureof QuantumTheory"FestschriftforC.F.vonWeizsaeckerontheOccasionofhis90thBirthday. Availableat:arXiveprintquantph/0212084 Bub,J.(2004)"WhytheQuantum?"StudiesinHistoryandPhilosophyofModernPhysics35B: 241266 Bub,J.(2005)"QuantumMechanicsisAboutQuantumInformation"FoundationofPhysics34: 541560 Bub,J.(2006)"QuantumInformationandComputation"JohnEarmanandJeremyButterfield (eds.),HandbookofPhilosophyofPhysics(Elsevier/NorthHolland,forthcomingin2006). Availableat:arXiveprintquantph/0512125 Clifton,R.,Bub,J.,Halvorson,H.(2003)"CharacterizingQuantumTheoryinTermsof InformationTheoreticConstraints"FoundationsofPhysics33:15611591 Cover,T.M.,Thomas,J.A.(1991)ElementsofInformationTheory(NewYork:Wiley) Deutsch,D.(1985)"QuantumTheory,theChurchTuringPrincipleandtheUniversalQuantum Computer"ProceedingsoftheRoyalSociety(London)A400:97117 Deutsch,D.(1997)TheFabricofReality(London:Penguin) Dieks,D.(1982)"CommunicationbyEPRDevices"PhysicsLetersA92:271272 Einstein,A.,Podolsky,B.,Rosen,N.(1935)"CanQuantumMechanicalDescriptionofPhysical RealitybeConsideredComplete?"PhysicalReview47:777780 Everett,H.(1957)"'RelativeState'FormulationofQuantumMechanics"ReviewsofModern Physics29:454462 Feynman,R.(1996)FeynmanLecturesonComputatio,editedbyJ.G.HeyandR.W.Allen (Reading,MA:AddisonWesleyPublishingCompany) Fuchs,C.A.(2001)"QuantumFoundationsintheLightofQuantumInformation"Proceedings oftheNATOAdvancedResearchWorkshoponDecoherenceanditsImplicationsinQuantum ComputationandInformationTransfer,editedbyA.Gonis.Availableat:arXiveprintquantph/ 0106166 Fuchs,C.A.(2002)"QuantumMechanicsasQuantumInformation(andOnlyaLittleMore)." Availableat:arXiveprintquantph/0205039 Halvorson,H.,Bub,J.(2005)"CanCryptographyImplyQuantumMechanics?Replyto Smolin"QuantumInformationandComputation5(2005):170175.Availableat:arXiveprint quantph/00311065 Holevo,A.S.(1973)"StatisticalProblemsinQuantumPhysics"inG.MurayamaandJ.V. Prokhorov(eds)ProceedingsoftheSecondJapanUSSRSymposiumonProbabilityTheory,pp. 104109(Berlin:Springer) Kent,A.(1999)"UnconditionallySecureBitCommitment"PhysicalReviewLetters83:1447 1450 Lo,H.K.,Chau,H.F.(1997)"IsQuantumBitCommitmentReallyPossible?"PhysicalReview Letters78:34103413 Lo,H.K.,Popescu,S.,Spiller,T.(1998)IntroductiontoQuantumComputationand Information(Singapore:WorldScientific) Mayers,D.(1997)"UnconditionallySecureQuantumBitCommitmentisImpossible"Physical

ReviewLetters78:34143417 Nielsen,M.A.,Chuang,I.L.(2000)QuantumComputationandQuantumInformation (Cambridge:CambridgeUniversityPress) Schrdinger,E.(1935)"DiscussionofProbabilityRelationsBetweenSeparatedSystems," ProceedingsoftheCambridgePhilosophicalSociety31(1935):555563;32(1936):446451 Schumacher,B.(1995)"QuantumCoding"PhysicalReviewA51:27382747 Shannon,C.E.,Weaver,W.(1949)TheMathematicalTheoryofCommunication(Urbana: UniversityofIllinoisPress) Smolin,J.(2005)"CanQuantumCryptographyImplyQuantumMechanics?"Quantum InformationandComputation5:161169.Availableat:arXiveprintquantph/0310067 Spekkens,R.W.(2004)"InDefenseoftheEpistemicViewofQuantumStates:aToyTheory." Availableat:arXiveprintquantph/0401052 Steane,A.M.(1998)"QuantumComputing"ReportsonProgressinPhysics61:117173 Steane,A.M.(2003)"AQuantumComputerNeedsOnlyOneUniverse"StudiesinHistoryand PhilosophyofModernPhysics34B:469478.Availableat:arXiveprintquantph/0003084 Timpson,C.(2000)"QuantumInformationandtheFoundationsofQuantumMechanics." Availableat:arXiveprintquantph/0412063 vanFraassen,B.(1982)"TheCharybdisofRealism:EpistemologicalImplicationsofBell's Inequality"Synthese52:2538 Wallace,D.(2003)"EverettianRationality:DefendingDeutsch'sApproachtoProbabilityinthe EverettInterpretation"StudiesintheHistoryandPhilosophyofModernPhysics34:415439 Wootters,W.K.,Zurek,W.H.(1982)"ASingleQuantumCannotbeCloned"Nature299:802 803

OtherInternetResources
arXivEprintArchiveforQuantumPhysics PreskillLecturesonQuantumComputation PapersbyDavidMermin OxfordCentreforQuantumComputation HomePagesofResearchersonQuantumInformation

RelatedEntries
Bell'sTheorem|physics:Reichenbach'scommoncauseprinciple|quantummechanics:Copenhagen interpretationof|quantummechanics:manyworldsinterpretationof|quantumtheory:quantum computing|quantumtheory:quantumlogicandprobabilitytheory|quantumtheory:theEinstein PodolskyRosenargumentin

S-ar putea să vă placă și