arXiv:1307.2288v2 [cond-mat.str-el] 21 Aug 2013 - PDF document

arXiv:1307.2288v2 [cond-mat.str-el] 21 Aug 2013 QuantumDisentangledLiquidsTarunGrover1andMatthewP.A.Fisher21KavliInstituteforTheoreticalPhysics,UniversityofCalifornia,SantaBarbara,CA93106,USA2DepartmentofPhysics,UniversityofCalifornia,SantaBarbara,CA93106Weproposeandexploreanewnitetemperaturephaseoftranslationallyinvariantmulti-componentliquidswhichwecalla“QuantumDisentangledLiquid”(QDL)phase.Wecontemplatethepossibilitythatinuidsconsistingoftwo(ormore)speciesofindistinguishablequantumparticleswithalargemassratio,thelightpar-ticlesmight“localize”ontheheavyparticles.Wegiveaprecise,formaldenitionofthisQuantumDisentangledLiquidphaseintermsoftheniteenergydensitymany-particlewavefunctions.Whiletheheavyparticlesarefullythermalized,foratypicalxedcongurationoftheheavyparticles,theentanglemententropyofthelightparticlessatisesanarealaw;thisimpliesthatthelightparticleshavenotthermalized.Thus,inaQDLphase,thermalequilibrationisincomplete,andthecanonicalassumptionsofstatisticalmechanicsarenotfullyopera-tive.WeexplorethepossibilityofQDLinwater,withthelightprotondegreesoffreedombecoming“localized”ontheoxygenions.Wedonotpresentlyknowwhetheralocal,genericHamiltoniancanhaveeigenstatesoftheQDLform,andifitcannot,thenthenon-thermalbehaviordiscussedherewillexistasaninterestingcrossoverphenomenaattimescalethatdivergesastheratioofthemassoftheheavytothelightspeciesalsodivergesPACSnumbers:05.30.-d,03.65.Ud,67.10.FjContentsI.Introduction1II.QuantumEntanglementinnite“temperature”uids3A.Operatorandwavefunctionrepresentations31.Secondquantizationinthecontinuum32.Particlewavefunctionsinthecontinuum43.Quantumlatticegas4B.Entanglemententropyforspatialbipartitioning4C.Areaandvolumelaws4D.Wavefunctionsforareaandvolumelawentanglement51.Bosongroundstatewavefunctions:arealaw52.Volumelawwavefunctions53.Arealawwithdisorder:Many-BodyLocalizedstates6III.QuantumDisentangledLiquid:APossiblenewphaseintwo-componentuids?6A.DenitionofQuantumDisentangledLiquid71.Spatialandinter-componentbi-partioningdiagnostic7B.IllustrativewavefunctionfortheQDL8C.QDLwavefunctionoftheBorn-Oppenheimerform9IV.PhysicsoftheQDL10A.AnalogywithManyBodyLocalizedphysics10V.DiscussionandOpenIssues11A.StabilityoftheQDLphase?11B.ExploringthepossibilityofQDLPhysicsinWater11C.RelevancetoFinite“temperature”QuantumFieldTheory?12Acknowledgments12References12A.Volumelawentanglemententropyforrandom-signwavefunction13I.INTRODUCTIONConventionalwisdomholdsthatinordinaryliquidsat-nitetemperature,localthermalequilibrationisestablishedonscaleslongcomparedtoinelasticscatteringlengthsandtimes,andclassicalhydrodynamicsisoperative.Quantummechan-ics,whichplaysaroleonshortscalesandinsettingthemag-nitudesofthecrossoverscales,canbesafelyignoredinde-scribingqualitativebehaviorinthehydrodynamicregime.Forexample,classicalstatisticalmechanicsshouldbesufcientincapturingtheequilibriumandlinearresponsebehaviorofor-dinaryroomtemperatureliquids,e.g.water,onlonglengthandtimescales.Inthispaperwequestiontheseassumptions.Inparticu-lar,weproposeandexploreaqualitativerolethatquantumeffectsmightplayindeterminingthemacroscopicbehaviorofnitetemperatureliquids.Wefocusontwo-component(ormoregeneralmulticomponent)liquidsinwhichtherearetwo-speciesofindistinguishablequantumparticleswithalargemassratio,forexampletheionsandelectronsthatconstitutetheatomsinanatomicormolecularliquid(suchasliquidHe-lium),ortwodifferentionswithlargemassratioinamolecu-laruid,suchasoxygenandhydrogenionsinwater.Toillustratethegeneralclassofquestions,considerthepropertiesofHelium-3uidattemperaturesabovethesuper-uidtransition(ofseveralmilliKelvin).Fortemperaturesbe-lowtheFermienergy(roughly50K)He-3iswelldescribedbyFermiliquidtheory,whileathighertemperaturesadescriptionasanon-degenerate–essentiallyclassical–uidwouldbeap-propriate.Atmuchhighertemperatures,ontheelectronvoltscale,aselectronsarehighlyexcited,signicantionizationisexpected,andapictureintermsofaplasmaofelectronsand 2He-3nucleonswouldbeappropriate.Atmuchhighertem-peraturesstill,thenucleonswouldstartbreakingintoprotonsandneutrons,andwithcontinuedheatingoneshouldeventu-allyenterintoaregimedescribedasaquark-gluonandelec-tronplasma.Thecanonicalwisdommaintainsthatallsuchregimesareinthesamephase,separatedbycrossoversratherthanphasetransitions1.Butmightthisconventionalwisdombeincompleteorevenincorrect?Inthispaperweproposeandexploreapotentiallynewphaseofnitetemperaturetwo-componentquantumuidsinwhichtheconstituentshavealargemassratio.Adescriptionofthisnewphase,whichwewillcalla“QuantumDisentan-gledLiquid”(QDL),requirestreatingbothspeciesofparticlesquantummechanically.IntheQDLphasethelightparticlesare,inasensetobemademoreprecisebelow,“localized”ontheheavyparticles,andduetoquantumeffects,donotthermalize.Duetothelackoffullthermalization,acorrectdescriptionrequiresa“microcanonical”formulationwhereintheuid,connedtoaverylargebox,ispresumedtobeinapurequantumstate,perhapsaneigenstate,withaniteenergydensityabovethegroundstate.InSectionIIwerevisittheentanglementpropertiesofsingle-componentnite“temperature”uidsofindistinguish-ablequantumparticles.Wefocusonthenatureofhighlyex-citedpurequantumstates-withaniteenergydensityabovethegroundstate.Inafullythermalizedphase,forsuchstatestheentanglemententropy,foraspatialbi-partitioning,willsatisfyavolumelaw.Aswewillargue,thishighdegreeofspatialentanglementisencodedinthe“sign-structure”oftheseniteenergy-densitystates.Usingthisinsight,wecon-structavolumelawwavefunctionforbosons,whichcanbeshowntopossessamaximalentanglemententropydensitycorrespondingtoaninnite“temperature”.SectionIIIisdevotedtoadiscussionofnite“tempera-ture”two-componentquantumuids,withonespeciesmuchmoremassivethantheother.Wegiveaprecisemathemati-caldenitionoftheQDLphasewhichinvolvesadiagnostictobeperformedonthemany-particlewavefunction.Roughly,whenthepositionsoftheheavyparticlesarexed,thelightparticlesareina“disentangled”arealawphase.Foralatticegasmodeloftwobosonspecies,weconstructanexplicitQDLwavefunctionwhichcanbeshowntosatisfyallofthedeningcriterionoftheQDLstate.TheessenceoftheQDLphasecanbeconveyedbycon-sideringawavefunctionforatwo-componentuidofquan-tumparticleswhichtakesaBorn-Oppenheimer2typeform.Tobeexplicit,considerasystemofheavy,massM,bosonsandlight,massm,fermions,withmM.ABorn-Oppenheimerwavefunction,expressedintermsofthecoordi-natesoftheheavyandlightparticles,Randr,respectively,takestheform, (fRgfrg)= (fRg)det[R(r0)](1)wherethesecondtermisaSlaterdeterminantforthelightfermionsconstructedfromsinglefermionorbitals,de-noted,R(r),whichareeigenstatesoftheeffectiveBorn-OppenheimerHamiltonian,H=Xp2 2m+VR(r)](2)witheffectivepotential,VR(x)=XUMm(xR)(3)HereUMmistheinteractionpotentialbetweenthelightandheavyparticlesandtheheavyparticlecoordinatesfRgarec-numbers.Above,theindividualorbitalsRhaveal-readybeensymmetrizedinthecoordinatesfRg,sothat (fRgfrg)isavalidwavefunctionforbosonsintheco-ordinatesfRg,andforfermionsinthecoordinatesfrg.AcaricaturewavefunctionfortheconjecturedQDLphasecanbeconstructedbytakingtherstpieceofthewavefunc-tioninEq.1, (fRg),tocorrespondtoanite-energyden-sityuidphase.Fortheheavybosons,atypicalniteen-ergydensitystateforanisotropicliquidcanbeconstructedfromapermanentofplanewaves, perreik
R0,withwavevectorschosenwithrandomdirections,^k,andwithran-domamplitudestakenfromsomeappropriatedistribution,forexampleaBoltzmanndistribution,pkexp(k= )withk=~2k2=2Mand correspondingtoan“effectivetemper-ature”.AsimplewavefunctionwhichdescribestheproposedQDLphaseoftwo-componentquantumparticlesshouldthustaketheform, QDL(fRgfrg)=perreik
R0]det[R(r0)](4)Duetotherandomwavevectors,thetypicalpositionsoftheheavyparticleswillthemselvesbeinanessentiallyran-domconguration,appropriatetoaniteenergy-densityuidphase.Asaresult,theeffectivepotentialVR(x)intheBorn-OppenheimerHamiltonianforthelightfermionswillbeanessentiallyrandompotential.Andoneexpectsthatforstronginter-speciesinteractionenergyUMmthelowen-ergysingleparticleeigenstateswillbespatiallylocalized3.TheSlaterdeterminantwavefunctionthusdescribesalocal-izedphaseoffermions.OnecanimaginegeneralizingtheaboveBorn-Oppenheimerwavefunctiontoincorporate(weak)inter-fermioninteractions,replacingtheSlaterdeterminantbyaninteractingwavefunctionfora“Many-Body-Localized”(MBL)phaseoffermions,asintroducedinRefs.4,5.Ineithercase,underaspatialbi-partitioning,intoregionsAandBwithlineardimensionL,thislocalizedeigenstatesoffermionswillhaveanarealawentanglemententropy,SALd1.Incontrast,thepermanentwavefunctionfortheheavypar-ticles perreik
R0,beingbuiltfrom(random)planewavesisexpectedtohaveavolumelawentanglementen-tropy,SLd,duetotheessentiallyrandomsignstructureuponvaryingthepositionsoftheparticles(asweshalldiscussinconsiderabledetailinSectionII).Foragivenxedtypi-calcongurationofthecoordinatesofthelightfermions,sayf~rg,theSlaterdeterminantwillhaveamagnitudeandsignthatvariesrapidlywhenthepositionsoftheheavyparticlesarechanged.ButsincetheSlaterdeterminantisinanyevent 3multipliedbyaheavyparticlewavefunction, (fRg),withanessentiallyrandomsignstructure,thefullQDLwavefunc-tion QDL(fRgf~rg),isstillexpectedtohaveavolumelawentanglemententropy.ThesearethetwoprimarypropertiesofaQDLwavefunc-tion-whentheheavyparticlesarexedinatypicalcongu-rationthelightparticlesareinanarealawphase,while,withthelightparticlecoordinatesxed,theheavyparticleshaveavolumelawentanglemententropy.Inessence,theheavypar-ticleshavethermalized,butthelightparticles,“localized”bytheheavyparticles,havenot.Goingbeyondwavefunctions,thegenericstabilityoftheQDLphaseinlocalHamiltoniansofinteractingparticles,requires,attheveryleast,thegenericstabilityoftheMBL4,5state.WhileanAndersoninsulatorofnon-interactingfermionsinarandompotentialcanexistatniteenergyden-sities-providedthatallthesingleparticlestatesenteringtheSlaterdeterminantarelocalized-establishingthestabilityoftheMBL,anite-energydensitystateofinteractingelectrons,iscomparativelymuchmorenon-trivialandrecentnumericalprogress6–24hascorroboratedtheoriginalpicturepresentedinRefs.4,5.ReturningtoQDL,eventhestabilityoftheinteract-ingMBLtoarbitraryperturbationsdoesnotimplythestabilityoftheQDLphase,whichismuchmoretricky.AsillustratedintheBorn-Oppenheimerdiscussion,iftheheavyparticlemassistakentoinnity,theirpositionswillactasaquenchedrandompotentialforthelightquantumparticleswhich,ifinanarealawMBLstateatniteenergydensity,willnotther-malize.Forverylargebutnitemassratio,M=m,whentheheavyparticlesbecomequantumdynamicalandareas-sumedtothermalize,therearetwopossibilitiesforthetypicalnite-energydensityeigenstatesofthetwocomponentquan-tumsystem:(i)Thelightparticlesalsothermalize,amongthemselves,atthesame“temperature”astheheavyparticles-thisistheconventionalimplicitassumptionunderlyingthehy-pothesisofquantumstatisticalmechanics,(ii)Thelightpar-ticlesdonotthermalize-thisistheQDLphase.WhilethewavefunctionoftheQDLphasehasawelldenedandprecisedenition,asdetailedinSectionIII,itmightbethatitsimplyisnotpossiblefora(generic)localHamiltoniantohaveeigen-statesoftheQDLform.Ifthisisthecase,theM=m!1limitwouldbesingular,andthenon-thermalbehaviorofthelightparticlescouldonlyoccuratstrictlyM=1.Presum-ably,inthiscase,onecoulddeneatypicalthermalequili-brationtimeforthelightparticles,saystartingfromtheMBLstate,andthistimewoulddivergeasM!1.OntimesscalesshortcomparedtothiscrossovertimetheQDLphysicswouldbeoperative,butwouldbreakdownatlongertimes.WhetherornotthisequilibrationtimecanremaininniteatniteM-i.e.whethertheQDLphasecanexistasa“truephase”-isnotsomethingthatwecanpresentlyestablish.SectionIVAwillbedevotedtoamoregeneraldiscussionofthephysicsoftheQDLphase,drawingfurtheranalogieswiththeMany-Body-Localizedphaseofquantumparticlesinarandompotential.Inparticular,the(linearresponse)con-ductivityinanMBLphasewithaconservedparticlenumberisexpectedtovanish,evenatnite“temperature”-theparti-clescannotpropagate.Analogously,inaQDLphasewherethetwospeciescarryopposite“electric”charge,oneexpectsthatthe(linearresponse)electricalconductivityshouldvan-ishidentically.WewillalsodiscusshypotheticalexperimentsinQDLwhichcandetectlackofthermalizationofthelightparticles.InSectionVAwewillbrieydiscusstheprospectsofes-tablishingwhetheraQDLphasecanexistinalocalHamilto-nianforatwo-uidsystem,andthefeasibilityofverifyingaQDLphaseinnumerics.InSectionVBwewillexplorethepossibilitythatpurewatermightbeinaQDLphase,wheretheprotonsare“localized”bytheheavieroxygenions.Thispossibilityisinformedbytheremarkablephenomenologyofpurewater-itsincrediblyhighresistivity,andtheintricatehydrogenbondingwhichstronglycorrelatesthemotionoftheoxygenionswiththeprotons.II.QUANTUMENTANGLEMENTINFINITE“TEMPERATURE”FLUIDSConsiderasinglecomponenttranslationallyinvariantuidofindistinguishablequantumparticles(fermionsorbosons).Fornowwewillignoreanyinternal“spin”quantumnumbersoftheparticles.WeimaginethatthedynamicsoftheuidiscontrolledbyanunderlyingquantumHamiltonian.Thepre-cisenatureoftheHamiltonianwillnotbecritical,butwecanthinkofnon-relativisticparticlesmovinginthecontinuumin-teractingwithatranslationallyinvariant(say,two-body)in-teraction,seebelow.Thequantumparticlesareassumedtobeisolatedfromanythermalbath,forexampleconnedinabox.Ourprimaryfocuswillbeonthenatureofeigenstates,withaniteenergy-densityabovethegroundstate.Theeigenstatethermalizationhypothesis25–28positsthatalargesub-region(the“system”)willbehavethermally-asifatnitetemper-ature-ifonetracesoutthedegreesoffreedomoutsidethisregion(the“environment”).Wewillreviewanddiscusstheentanglementpropertiesofpurestatesinsuchasystem.A.OperatorandwavefunctionrepresentationsSincewewillbediscussingsuchnitetemperaturequan-tumuidsfromseveraldifferentangles,itwillprovecon-venienttointroduceanddeneseveraldifferentrepresenta-tions/models-employingrstandsecondquantizedoperatorsandstates/wavefunctionsinthecontinuum,aswellasalatticegasmodel.1.SecondquantizationinthecontinuumWewilldenotegeneralparticleeldoperatorsby,^dx^dyxwhichsatisfycanonical(anti-)commutationrelations,,^dx^dyx0=(xx0),withsignforfermions/bosonsre-spectively.Whenwewishtospecializetofermionsorbosonswewillreplacetheoperators,^dxbythemorefamiliarnota-tions,^cxforfermionoperatorsand^bxforbosonoperators.In 4termsoftheeldoperatorsarepresentativeuidHamiltonianisgivenby,^H=Zx^dyx(r2x=2m)^dx+Zx;x0U(xx0)^nx^nx(5)wherewehavedenedaparticledensityoperator,^nx=^dyx^dx.Heretheparticleshavemassm,andinteractviaatwo-bodyinteractionU(x).Whileweareprimarilyinterestedinthree-dimensionaluids,wedonotneedtospecifythedimen-sionalityhere.Forbosonsthegroundstateofthismodelwilllikelybeei-therabosonsuperuidorabosoncrystal.AndthefermiongroundstatecouldbeaFermiliquid,apairedsuperuidorafermioncrystal.Butwearenotinterestedingroundstates.Wewillfocusonhighlyexcitedstatesatenergydensitieswellaboveanycondensationorcrystallizationtemperatures.Asmentionedabove,weemployamicrocanonicalensem-bleinwhichtheuidisconnedwithinaboxofvolumeV,possiblywithperiodicboundaryconditions,andisolatedfromanythermalbath.Weareinterestedineigenstates,witha-niteenergydensityabovethegroundstate.Sinceourmodelistimereversalinvariantwecantaketheexacteigenstatestobereal.2.ParticlewavefunctionsinthecontinuumTodiscussrstquantizedwavefunctionsexpressedintermsofparticlecoordinates,itisconvenienttodeneacompletesetoforthogonalpositionkets,jfrgi=QNmi=1^dyrivaciwhere=12;:::;Nmlabelstheparticles,withNmde-notingthetotalnumberof(massm)particles.Forfermions,theseketsarenormalized,whilethenormalizationforthebo-sonstatesdependsonhowmanybosonsoccupythesamespa-tialpoint.Todispensewiththeseunimportantyetdistract-ingsubtleties,weassumethebosonshaveahardcore,withradiusasettingthesizeoftheparticles.Associatedwithageneralstatevector i,isapositionspacewavefunction (frg)=hfrgj iwhichissymmetric/antisymmetricun-derexchangeofanytwoparticles,forbosons/fermions.3.QuantumlatticegasAttimesbelowitwillbeconvenienttoplacetheparti-clesonalattice.Wedenealatticegasmodelintermsofcreation/annihilationoperators^dyj^dj,wherethesubscriptj=12;:::;Nlabelsthelatticesites.WeassumethelatticeHamiltonianhasa“canonical”hoppingtermandlocalinter-actions,^H=Xijtij(^dyi^dj+h:c:)+XijUij^ni^nj(6)wherebothtij0andUijarelocal.Here,^nj=^dyj^djistheon-sitedensityoperator-thehardcoreconditionimplies^njhaseigenvalues,nj=01.Themodelistakentorespectthepointandspacegroupsymmetriesofthelattice,especiallytranslations.Itishelpfultointroduceposition-spaceoccupationnumberbasisstates,jfnjgi=QNj=11^dyjnjvaci,wherenj=01de-notestheabsence/presenceofaparticleonsitej.Forfermionsonemustchoosesomeorderingforlabelingthesitesofthelat-tice-wedonotspecifythisunlessitisnecessary.Associatedwithageneralstatevector iwecandeneawavefunctionintheoccupationnumberbasis, (fnjg))=hfnjgj i.B.EntanglemententropyforspatialbipartitioningThroughoutwewillassumethesystemisinapurequantumstatewithdensitymatrix,^= ih .Thedensitymatrixcanbeeitherexpressedintermsoftheparticle-positionbasiswavefunctions,(frgfr0g)= (frg) (fr0g)(7)ortheparticle-occupationbasiswavefunctions,(fnjgfn0jg)= (fnjg) (fn0jg)(8)Considerasimplebi-partitioningintotwospatialregions,denotedAandBrespectively.InthecontinuumwedenotethevolumeofregionAasVA=LdA,the“system”,andofregionBasVB,the“environment”,withVA+VB=V.OnthelatticeNAandNBdenotethenumberofsitesinthetworegions,withNA+NB=N.Thereduceddensitymatrixforsub-regionAisobtainedasusualbytracingouttheparticlesinregionB,^A=TrB[^(9)Andthe(Renyi)entanglemententropiesfollow,SA;=1 1lnTrA[(^A)(10)withSvNA=lim!1SA;thevonNeumannentropy.C.AreaandvolumelawsNumerousstudieshaveshownthatthegroundstateofbosonsystemsdescribedbytherepresentativeHamiltoniansabovewillhavearealawentanglemententropy29–31,S(0)ALd1A.Indeed,boththebosonsuperuidandbosoncrystalgroundstatesexhibitanarealaw,andthisisbelievedtobeageneralpropertyof(unfrustrated)bosongroundstates.Apairedfermionsuperuidorfermioncrystalgroundstatewillalsohaveanarealaw,whiletheentanglemententropyofaFermiliquidgroundstatehasamultiplicativelogcorrectiontothearealaw32,33,SFLALd1AlnLA.Theentanglemententropyforeigenstateswithaniteen-ergydensityabovethegroundstateareexpectedtobequalita-tivelydifferent.Theseeigenstates,whichofferamicrocanon-icaldescriptionofanite“temperature”uid,aregenerally 5assumedtothermalize(althoughseeSectionIII,below).In-deed,theeigenstatethermalizationhypothesis25–28maintainsthatwhenVAVB,thereduceddensitymatrix^Aisthesameasathermaldensitymatrixforthesub-systemAatatemperatureforwhichthethermodynamicinternalenergyequalstheenergyperunitvolumeofthepurestate i.Thatis,thelargesub-systemBcanbeviewedasathermalbathwhicheffectivelycausesthermalizationofthemuchsmaller“system”involumeA.Inparticular,equilibriumstatisticalmechanicsemployingthereduceddensitymatrix,^Acanbeusedtocomputethevariousthermodynamicpropertiesofthesub-systemA.AconsequenceoftheeigenstatethermalizationhypothesisisthatthevonNeumannentanglemententropysatisesavol-umelaw,SvNA=sVA,andthatscanbeequatedwiththether-malentropydensityinthesub-systemA.Similarly,theRenyientropiesalsosatisfyavolumelawscaling,SA;VA.Theeigenstatethermalizationhypothesisistantamounttotheassumptionthatstatisticalmechanicsisvalidinalargequantumsystem;apurequantumstatehassub-systemswhicharecorrectlydescribedusingconventionalstatisticalmechan-icsinthecanonicalensemblewhereinthesub-systemhasthermalizedwithanappropriateheatbathatsometempera-ture.[InSectionIIIwewillintroduceanewnitetempera-turephaseofmulti-componentquantumparticleswhichonlypartiallythermalizes.]Whiletheassumedvolumelawbehaviorfornite-energydensitystatesismostplausible,itwouldbedesirabletohaveexplicitvolumelawwavefunctionsinordertodevelopaphys-icalpicturefor(quantum)thermalization.Inthenextsubsec-tionweconstructsimplewavefunctionsthatexhibitareaandvolumelaws.D.WavefunctionsforareaandvolumelawentanglementForsimplicitywefocusnowonbosonwavefunctions.Forthelatticegasmodel,themostgeneralstatevectorcanbeex-pressedas, i=Xnj (fnjg)jfnjgi(11)Sincethewavefunctionamplitudescanbetakenasreal,theycanbedecomposedas, (fnjg)= (fnjg)sgn(fnjg)(12)wheresgn(fnjg)=1.1.Bosongroundstatewavefunctions:arealawThesimplicityofbosongroundstates, 0,(oftherepre-sentativeHamiltoniansabove)isthattheirwavefunctionsarenodeless,withsgn0(fnjg)=+1forall2Ncongurations.Thewavefunctionsintermsofparticlecoordinatesarelike-wisenodeless, 0(frg)0foranyparticleconguration.Theentanglemententropyofsuchnon-negativebosongroundstateswillgenericallyfollowtheaforementionedarealaw.Toillustratethis,considerastatethatisexpressedananequalweightsuperpositionoveralloccupation-numberbasisstates, 0(fnjg)=1 p 2N(13)Thisstateencodeslargedensityuctuationsoftheparticlesandshoulddescribeabosonsuperuid.Whileithasbeenes-tablishedthatbosonsuperuidstateshaveanarearatherthanvolumelawentanglement34,35,thiscanbeexplicitlyveriedfortheaboveequalamplitudewavefunction,whichcanbere-expressedasadirectproductstate, 0(fnjg)=NYj=11 p 22nj;0+nj;1(14)Forthisproductstatetheentanglemententropyactuallyvan-ishesidentically,SA;=0.Slightmodicationsofthissu-peruidwavefunctionwouldleadtosmallcorrectionsgivinganarealaw34,35,SA;Ld1A.2.VolumelawwavefunctionsWenowwishtoconstructwavefunctionsforbosonswhichexhibitavolumelawentanglemententropy,whicharehope-fullyrepresentativeoftypicalniteenergydensityeigenstatesinsuchsystems.Whilethegroundstateofthebosonsystemisnodeless,excitedstatesnecessarilyhavenodes,withposi-tiveandnegativewavefunctionamplitudes.Andveryhighlyexcitedeigenstates-withaniteenergydensityabovethegroundstate-shouldberiddledwithnodes,likelyexhibitinganintricateandrichsignstructure.Intuitionforsuchsignstructureintricaciescanbegainedbyconsideringparticle-coordinateeigenwavefunctionsfortheidealBosegas,whichareexpressedaspermanentsofplanewavestates, IBG(frg)=perrM(15)withmatrix,M=1 p Veik
r(16)withkasetofwavevectorswith=12;:::;N.WhilethegroundstateBosecondensatehasallk=0,oneex-pectsthatatypicalniteenergydensityeigenstatewouldcor-respondtochoosingthewavevectorsktobeessentiallyran-dom,withdirections^kequallyweightedovertheunitsphere,andwithamplitudeschosenfromanappropriatedistribu-tionfunction,suchaspkexp(k= )withkineticenergyk=~2k2=2m,and anaveragekineticenergy.Suchaniteenergywavefunctionshouldcorrespondtoaneffectivetem-peratureoforderkBTeff .Wenotethatthewavefunction IBGcanbemaderealbytakingthewavevectorstocomeinkpairs. 6Itismostplausiblethatpermanentwavefunctions, IBG(frg),constructedfromplanewaveswithrandomwavevectors,shouldexhibitaverycomplicated,andessen-tiallyrandomsignstructure,astheparticlepositionsaremovedabout.Basedonthisreasoning,onewouldanticipatethatthe-niteenergydensityeigenstatesfortheBoselatticegas,haveanoccupation-numberwavefunctionrepresentationwithes-sentiallyrandomsigns.Toillustratethiswewillexploreawavefunctionforlatticebosonswhichconsistsofanequalamplitudesuperpositionofoccupationbasisstates,butwithrandomlychosensigns: (fnjg)=1 p 2Nsgn(fnjg)(17)withthesgnfunctionchosenrandomlysgn(fnjg)=1,independentlyforeachofthe2Ncongurationsfnj=01g.Ashortcalculation(seeAppendixA),showsthatthesec-ondRenyientropySA;2correspondingtothewavefunctioninEq.17foraspatialbipartitioningofthesystemintoregionsAandBisgivenbySA;2=log2NA+2NB2(NA+NB)ln(2)NA(18)whereinthesecondline,wehaveassumedthatNANB(“system”muchsmallerthan“environment”).Thisconrmsthatthemaximallyrandom-signwavefunctionexhibitsavol-umelawentanglemententropyforthevon-NeumannentropySvNAaswell(recalltheinequality36SvNA&#x-5.9;╆SA;2).Moreover,theamplitude,ln(2),isthemaximumpossiblevalue.Wheninterpretedasathermalentropythiscorrespondsto“innitetemperature”,withall2NAstatesinregionAsummedoverwithequalweightinthecanonicalpartitionfunction.Itisnaturalthatthismaximallyentropicstate,correspondstothewavefunctionwithcompletelyrandomanduncorrelatedsigns.Ifoneweretomodifythewavefunctionbyintroducingsome(spatial)correlationsintothesignstructure(stillwithhalfthesignspositiveandhalfthesignsnegative),onewouldexpectthatthecoefcientofthevolumelawwouldbereduced37.Beforeturningtothetwo-componentquantumuid,inSec-tionIII,wherewepropose,deneanddiscusstheQDLphase,itwillbehelpfultobrieyreviewtheentanglemententropyinlocalizedstatesofquantumparticlesmovinginaspatiallyde-pendentquenchedrandompotential.3.Arealawwithdisorder:Many-BodyLocalizedstatesWhileweareinterestedintranslationallyinvariantuids,itwillbeinstructivetoremindthereaderofthepropertiesofwavefunctionsforbosons(orfermions)movingthroughaspatiallyrandomquenchedpotential,described,forexample,byaddingatermtotheHamiltonianoftheform,^Hdis=XjVj^dyj^dj(19)Here,forexample,onemighttakeVjtobeasite-uncorrelatedrandompotential.Whenthedisorderstrengthgreatlyexceedstheparticleskineticandinteractionenergies,thegroundstatewavefunctionisexpectedtobelocalized,eitherinafermioninsulator3ora“Boseglass”phase38.Acaricaturewavefunc-tionforthelocalizedfermioninsulatorgroundstatemightbeconstructedintermsofadeterminantbuiltfromthelowestenergylocalizedsingleparticleorbitals,(r), MBL(frg)=det[M(20)withthematrixM=(r).Whilethiswavefunctionisaneigenstatefornon-interactingfermions,itisgenerallybelievedthatthesystemwillstillretainalocalizedinsulatinggroundstatewithweakenoughinteractions.Whilethegroundstateshouldbelocalizedwithstrongenoughdisorder,recentwork4–24hassuggestedthatsuchstronglydisorderedsystemscanhavehighlyexcitedniteen-ergydensitystatesthatarelikewiselocalized.Theseso-calledMany-Body-Localized(MBL)phases4,5whilehavinganiteenergydensityabovethegroundstate,asforathermalstate,arenotthermalized.Indeed,theentanglemententropyfortheeigenstatesinanMBLstateisbelievedtosatisfyanarealaw,SMBLA;Ld1A.Looselyspeaking,theMBLstatescanbethoughtofasreal-spacedirectproductstatesofsingleparticlelocalizedorbitals.AsweshallseeinthenextSection,forcertainnite“tem-perature”translationallyinvariantmulti-componentuids,anewkindofentanglemententropycanbedened,whichdoeshaveanarealaw,despitetheabsenceofanyquencheddisor-der.III.QUANTUMDISENTANGLEDLIQUID:APOSSIBLENEWPHASEINTWO-COMPONENTFLUIDS?Wenextconsideratranslationallyinvarianttwo-componentuidofindistinguishablequantumparticles,asetofheavyparticleswithlargemassMandasetoflightparticleswithmassmM.Initiallywewillnotspecifythequantumstatisticsofthetwosetsofparticles.TheHamiltonianwehaveinmindisageneralizationofEq.5consistingofkineticener-giesforthetwoparticletypes,alongwithshort-rangetwo-bodyinteractions,^H=^H0+^HU(21)with^H0=Zx^Dyx(r2x=2M)^Dx+Zx^dyx(r2x=2m)^dx(22)^HU=Zx;x00Umxx0^nx^nx0+UMxx0^Nx^Nx0+UmMxx0^nx^Nx0(23)where^nx^dyx^dxand^Nx^Dyx^Dx. 7WewillagainfocusonthemicrocanonicalensembleinwhichtheuidsareconnedinaboxwithvolumeV,possi-blywithperiodicboundaryconditions,andisolatedfromanythermalbath.Wewillprimarilybeinterestedinthesitua-tionwherethesystemisplacedinanexactmany-bodyeigen-stateoftheHamiltonianwithaniteenergydensityabovethegroundstate.Sinceourmodelistimereversalinvariantwecantaketheexacteigenstatestobereal.Itwillagainbeconvenienttoconsiderrstquantizedparti-clewavefunctions,whichwecandeneviathepositionkets,jfRgfrgi=NMY=1^DyRNmY=1^dyrvaci(24)whereNMandNmdenotethenumberofheavyandlightparticles,respectively.WemakenoassumptionsaboutNMNmexceptthattheratioNM=Nmremainsniteinthethermodynamiclimit.Denotingthestatevectorsas iwehavepositionspacewavefunctions, (fRgfrg)=hfRgfrgj i.Forbosons/fermionsthesewavefunctionswillbeappropriatelysymmetrized/antisymmetrizedunderex-changeofanytwoparticles.Asforthesinglecomponentcaseitwillproveconvenienttoalsodeneaquantumlatticegasmodel,intermsofop-erators^dj^Djwithj=12;::;Nlabelingthelatticesites.TheanalogouslatticetightbindingHamiltonianforthetwo-componentuidtakestheform^H=^H0+^HUwithkineticenergy,^H=Xij(tMij^Dyi^Dj+tmij^dyi^dj+h:c:)(25)wherethepositivehoppingamplitudessatisfytheinequality,tmij��tMij�0,andaninteractionenergy,^HU=XijjUmij^ni^nj+UMij^Ni^Nj+UmMij^ni^Nj(26)withthedenitions^nj=^dyj^djand^Nj=^Dyj^Dj.Weagainintroduceposition-spaceoccupationnumberba-sisstates,jfNjgfnjgi=NYj=11^DjNjNYj=11^dyjnjvaci(27)whereNj=01andnj=01denotestheabsence/presenceofaheavyparticleandlightparticleonsitej,respectively.Associatedwithageneralstatevector iwedeneawave-functionintheoccupationnumberbasis, (fNjgfnjg))=hfNjgfnjgj i.A.DenitionofQuantumDisentangledLiquidWewillassumethatthesystemisinatranslationallyinvari-antpurestate (fRgfrg),aniteenergydensityeigen-stateoftheinteractingHamiltonianinEqn.21.Wewillas-sumethatthewavefunctionisrealandhastheexchangeprop-ertiesappropriatefortheparticles'statistics.Wemakethefollowingassumptionsaboutthewavefunc-tion.Underaspatialbi-partioningintoregionAandBwithVAVB,wewillassumethattheentanglemententropyex-hibitsavolumelaw,thatisfor^A=TrRB;rB ih (28)thereduceddensitymatrixvariesas,SA=TrRA;rA[^Aln^AVA(29)Wewillalsoassumethatthelightandheavyparticleshavea“mutual”volumelawentanglement,underapartitioningbe-tweenthetwocomponents.Specically,deneareducedden-sitymatrixbyintegratingoutonlythelightparticlesthrough-outthewholevolume,V=VA+VB,^M=Trm ih (30)Thendeneanentropybytracingovertheheavyparticles,alsothroughoutthewholevolumeofthesystem,SMm=TrM^Mln^M(31)This“inter-component”mutualentropyisassumedtovarywiththewholevolumeofthesystem,SMmV.1.Spatialandinter-componentbi-partioningdiagnosticWenowintroducethefollowingdiagnosticofsuchtwo-componentwavefunctions.Itconsistsofseveralsteps.Instepone,weconstructtheprobabilitydistributionforthepositionsoftheheavyparticles,independentofthepositionofthelightparticles,P(fRg)=ZNmY=1dr (fRgfrg)2(32)Next,viewingP(fRg)asaprobabilitydistribution,wedrawsetsofcoordinatesfR(1)gfR(2)gfR(3)g;:::andsoon.Andforagivenmemberofthisensemblewereturntotheoriginalwavefunctionandnowimagineitasawavefunctionsolelyofthelightparticles,withthepositionoftheheavyparticlesf~Rgplayingtheroleofaxedparameter.Thisrequiresre-normalizingthewavefunctionas i~R1 p Z~R i(33)withZ~R=ZNmY=1dr (f~R(n)gfrg)2(34)sothatRQNm=1dr ~R(frg)2=1.Wethendeneadensitymatrix,^~R= i~R~Rh (35) 8andwiththecoordinatesoftheheavyparticlesxed,constructareducedmatrixfrom^~RbytracingoutthelightparticlesinregionB,^~RA=TrrB^~R(36)Next,weusetheabovereduceddensitymatrixtocomputethevonNeumannentanglemententropy,S~RA=TrrA[^~RAln^~RA(37)NotethatherewehaveperformedatracesolelyoverthelightparticleswithinregionA.ThevonNeumannentropythusdeneddependsonthecoordinatesofalltheheavyparticles(bothinregionsAandB),whichwearedenoting~R=f~Rg.Finally,weensembleaverageS~RAovertheprobabilitydis-tributionfunctionP(fRg)oftheheavyparticles(Eqn.32):Sm=MA=ZNMY=1d~RP(~R)S~RA(38)Herethesuperscriptm=Mistoindicatethatthisentropyreferstotheentanglemententropyobtainedbyspatiallybi-partitioningthelightparticles,withmassmandcoordinatesr,givenaxedpositionoftheheavyparticleswithmassMandcoordinatesR,onlyafterwardsaveragingoverthedis-tributionoftheheavyparticlepositions.Bysimplyswitchingthecoordinates,r$R,wecancarrythroughthisproceduretoobtaintheentropyofspatialentan-glementoftheheavyparticlesgivenwexedthepositionsofthelightparticles,whichwewilldenoteSM=mA.Inthe“conventional”descriptionofatranslationallyinvari-anttwo-componentliquidphase,onewouldtypicallyassumethatthelightandheavyparticleshaveboththermallyequili-brated,implyingavolumelawforbothentropies,Sm=MALdASM=mALdA;(Thermal)(39)Thesevolumelawbehaviorsimplythateachspeciesofparti-clecontributesapiecetotheentropyperunitvolumeinregionA.WearenallyinapositiontogiveaprecisedenitionoftheQuantumDisentangledLiquid.TheQDLphaseisde-scribedbynite-energydensitywavefunctionseachofwhichhasavolumelawunderafullspatialbi-partioning,SAVA,aninter-componententanglementvaryingasthevolumeofthewholesystem,SMmV,whileforeachwavefunc-tionthetwospatial/inter-componentbi-partioningentropiesbehavedifferently-xingthelightparticlesleavestheheavyparticlesinaspatialvolumelaw,butwiththeheavyparticlesxedthelightparticlesareinaspatialarealaw,Sm=MALd1ASM=mALdA;(QDL)(40)B.IllustrativewavefunctionfortheQDLInordertoconstructanexplicitwavefunctionthatwecanproveisinaQDLphase,itwillbeconvenienttofocusonthelatticemodelwithbothheavyandlightbosonparticles.Considerthebosonwavefunctionintheoccupationnumberbasis, QDL(fNjgfnjg)= (fNjg)NYj=11 p 22nj;0+eiNjnj;1(41)Here,thersttermontherighthandsign, (fNjg),isawavefunctionfortheheavyparticlesalone,andweassumethatthisisinthe“random-sign”volumelawstatefromSec-tionII, (fNjg)=2N2sgn(fNjg)withrandomsigns,sgn(fNjg)=1,chosenindependentlyforeachofthe2NcongurationsfNj=01g.WemustnowcheckeachoftheconditionsfortheQDLstate.Below,whenconvenient,wewillemploythesimpliednotation, QDL(N;n).Werstconsidertracingoutthelightdegreesoffreedomthroughoutthewholesystem,^M=Trm iQDLQDLh ,writingM(N;N0)=Xnj QDL(N;n) QDL(N0;n)(42)andexplicitlyperformthesummationoverfnjgtoget,M(N;N0)= (N) (N0)Yj1 2(1+ei(Nj+N0j))(43)Toassesstheinter-speciesentanglemententropyweconsider,TrM(^2M)=XNjN0jM(N;N0)M(N0;N)(44)andnd,TrM(^M)2=XN (N)4=2N22N=2N(45)givingtherequiredvolumelawfortheinter-species(2ndRenyi)entanglemententropy,SMm;2=lnTrM(^M)2=(ln2)N.Nextweconsideraspatialbipartitiontocheckthatthewavefunction QDLhasavolumelawspatialentanglement,whenboththeheavyandlightparticlesaretracedoutinaregionB.Itisconvenienttointroducethenotation,NA=fNj2AgandNB=fNj2Bg,andsimilarlyforN!n,sothatwecanwritethereduceddensitymatrix,^A=TrB[^,inashorthandnotationas,AA(NA;nA;N0A;n0A)with,A=XNB;nB QDL(NA;NB;nA;nB) QDL(N0A;NB;n0A;nB)(46) 9ThesummationoverfnBgcanbeperformedexplicitlytogive,A=f(NA;nA)f(N0A;n0A)XNB (NA;NB) (N0A;NB)(47)wherewehavedened,f(NA;nA)=1 p 2Yj2A(nA;0+eiNAnA;1)(48)WenowexamineTrA[^2A,XNA;nA;N0A;n0AA(NA;nA;N0A;n0A)A(N0A;n0A;NA;nA)(49)Wecanusetheidentity,XnA;n0AAf(NA;nA)f(N0A;n0A)]2=22NA4NA=1(50)toobtaintheexpression,TrA[^2A]=XNB;N0BXNA (NA;NB) (NA;N0B)2(51)Withthepresumedformoftheheavyparticlewavefunction,asasumoverallreal-spaceoccupationnumbercongurationswitharandomsign,weobtain,TrA[^2A]=2NA+2NB2(NA+NB)(52)Thisgivestherequiredvolumelawforthespatialbi-partitionRenyientanglemententropy,SA;2=lnTrA[^2A]=(ln(2))NAforNANB.Again,thisimpliesthatvonNeumannentropySAalsofollowsthevolumelaw.Wenextturntothewavefunctiondiagnosticwhichinvolvesfreezingthecoordinatesofonespeciesofparticle,andper-formingaspatialbi-partitioningintoregionsAandBandintegrationovertheotherparticletypeinregionB.Con-siderrstxingthepositionsoftheheavyparticles,choosingacongurationoftheoccupationnumbers,f~Njg.Oncetheheavyparticlesarexedin QDL(~N;n),thewavefunctionintermsofthelightparticlesisaproductlawinspace.Thus,thespatialentanglemententropyforthelightparticleswillactu-allyvanish,sothatSm=MA=0.Thisismorallyanarealaw,Sm=MALd1A.Nextwemustxthepositionsofthelightparticles,choos-ingsomecongurationforf~njg=01.Let~jdenotethesubsetofsiteswith~n~j=1.TheQDLwavefunctioncanthenbewrittenas, QDL(fNjgf~njg)=C(~n) (fNjg)Y~jeiN~j(53)wherethecoefcientC(~n)=Q~j(1=p 2).FollowingSec-tionIIIA1,thisrequiresnormalizingthiswavefunctionovertheheavyparticles,whichsimplysetsC(~n)=1forallcon-gurationsofthelightparticles,giving~ QDL(NA;NB~n)= (NA;NB)Q~jexp(iN~j).Wemustnowspatiallybi-partition,andintegrateouttheheavyparticlesinregionB,^~nA=TrNB~ QDLih~ QDL,~nA(NA;N0A)=Y~jA2Aei(N~jA+N0~jA)XNB (NA;NB) (N0A;NB)(54)Finally,wecancomputethetraceoverthedensitymatrixsquared,giving,TrNA[(^~nA)2]=2NA+2NB2(NA+NB)(55)oravolumelawforthesecondRenyientropy,SM=mA;2=X~nP(f~njg)lnTrNA[(^~nA)2]=(ln2)NA(56)HereP(f~njg)istheprobabilityofndingthelightparticlesinthecongurationf~njg,P(f~njg)=XNj QDL(fNjgf~njg)2(57)Again,thisimpliesvolumelawforthevonNeumannentropyaswell39.Therefore,ourillustrativewavefunction,Eqn.41,indeedsatisesallthepropertiespostulatedfortheQuantumDisentangledLiquid(QDL).C.QDLwavefunctionoftheBorn-OppenheimerformWenowconsiderawavefunctionforaQDLphaseoftwo-componentquantumparticlesmovinginthecontinuum.ThiswavefunctionwillbeconstructedbyusingaBorn-Oppenheimertypeapproach2.Tobeconcretewetaketheheavyparticlestobebosonsandthelightparticlestobefermions.Weassumethatthenumberofheavyandlightpar-ticles,NMNm,arecomparable,asaretheirmeaninterpar-ticlespacings,`dM=NM=Vand`dm=Nm=V.Weareinterestedinawavefunctionwithaverageenergyperparticleabovethegroundstate, ,whichsatises,~2 M`2M ~2 m`2m(58)Thiscorrespondstoan“effectivetemperature”wheretheheavyparticlesareina“classical”regimewhilethelightpar-ticlesareina“quantum”regime-although,asthroughoutthispaper,wearealwaystreatingbothheavyandlightparticlesasquantumparticles.Wewillalsoassumethattheheavyandlightparticleareinteractingwithoneanotherviaastrongpo-tential,UMm(x)withacharacteristicenergyscalelargecom-paredtothe“degeneracy”energyscaleforthelightparticles,UMm(0)��~2=m`2m. 10Considerawavefuctionoftheform, BOQDL(fRgfrg)=perreik
R0]det[R(r0)](59)wheretheindices;0runfrom12;:::;NMwhilethein-dices;0runfrom12;:::;Nm.TherstterminthiswavefunctiondenotesapermanentfortheheavybosonswithwavevectorskthatarechosenwitharandomdirectionandanamplitudepulledfromanappropriateBoltzmanndistribu-tion,pkexp(k= )withk=~2k2=2M.Thesecondtermisadeterminantwavefunctionforthelightfermionparticles.Here,thesingleparticleorbitals,R(r),denotethelowestenergysingleparticleeigenstatesforthelightfermionsinthepresenceofastaticcongurationoftheheavyparticles,fRg,describedbythesingleparticle“Born-Oppenheimer”Hamiltonian,^HBOc=Zx^cyx(r2x=2m)^cx+ZxVx^cyx^cx(60)wherewehavedenedapotentialenergy,Vx=NMX=1UMm(xR)(61)ItisimportanttoemphasizethatthecoordinatesoftheheavyparticlesfRjgthatenterintotheeffectivepotentialarec-numbers,notoperators.Sincetheheavyparticleshavebeenplacedintoawavefunc-tionappropriateforauid,constructedfromapermanentofrandomlyselectedsinglebosonplanewaves,atypicalcon-gurationofheavyparticlesthatcarriesappreciableweightinthefullwavefunctionwillcorrespondtoessentially“ran-dom”locationsoftheheavyparticles.Thus,intheeffectiveBorn-OppenheimerHamiltonianforthelightfermions,theef-fectivepotentialVxwillvaryrandomlyinspace.Undertheassumptionthattheinter-componentinteractionenergyscaleislargecomparedtothedegeneracyenergyofthelightpar-ticles,UMm(0)��~2=m`2m,allofthelowenergysinglefermioneigenstatesin^HBOcshouldbelocalized.Duetothislocalization,forgivenxedpositionsoftheheavyparticles,theSlaterdeterminantofthelightfermionsdescribesalocalizedphase.Whilewehaveassumedthatthelightfermionsareoccupyingthelowestenergylocal-izedeigenstates,whichisreasonableintheextremelimit, ~2=m`2m,thiscanberelaxed.Providedthehigheren-ergyoccupiedorbitalsinthefermionSlaterdeterminantareallspatiallylocalizedoneexpectsthatthepropertiesoftheQDLwavefunctionshouldnotchangequalitatively.Evenuponin-clusionofinteractionsbetweenthelightfermions,themanyfermionwavefunctioncanstilldescribealocalizedphase-aMany-Body-Localizedphase4,5.SincetheMBLwavefunctionofthelightfermionshasanarealawentanglemententropyunderspatialbi-partitioningintoregionsAandB,theentropySm=MAdenedinSectionIIIAobtainedbyxingthepositionsoftheheavyparticlesshouldsatisfyanarealaw,Sm=MALd1A,oneofthetwoimportantdeningpropertiesoftheQDLphase.Ontheotherhand,considerxingthepositionsofthelightparticlesintheQDLwavefunction,withsometypicalcon-guration,denotedf~rg.Whilethemagnitude,andmoreimportantlythesign,oftheSlaterdeterminantdet[R(~r0)],forgivenxedf~rg,willvaryappreciablyasthecoordinatesoftheheavyparticlesaremovedabout,sincethesignofthepermanentterminEq.59has,inanycase,anessentiallyran-domsignstructurealready,oneexpectsthattheentanglemententropyfortheheavyparticlesunderspatialbi-partitioningshouldbeavolumelaw,SM=mALdA.Thus,theBorn-Oppenheimertypewavefunctiondenedaboveisindeedex-pectedtodescribeaQDLphase.IV.PHYSICSOFTHEQDLThekeypropertyoftheQDLphaseasdenedaboveisthatitisnotfullythermalized.Physically,ifoneisgiventhepo-sitionsoftheheavyparticles,thelightparticlesappearasiftheywerelocalizedduetotheeffectivepotentialprovidedbytheheavyparticles.Therefore,thedynamicsandthemutualentanglementofthelightparticlesislargelydictatedbytheinstantaneouspositionsoftheheavyparticles.A.AnalogywithManyBodyLocalizedphysicsConsiderthelimitwherethemassMoftheheavyparticlesistakentoinnity.Inthislimit,theheavyparticlescanbetakenasxed“impurities”,andtheproblemreducestosolv-ingforthequantumdynamicsofthelightparticlesmovingintherandombackgroundpotentialoftheinnitelyheavypar-ticles.Let'sassumethatthisbackgroundstaticrandompo-tentialisstrongenoughthatthelightparticlesareinaMany-Body-Localized(MBL)phase4,5-thatis,generally,aniteenergydensitystatewithinwhichtheparticlescannotmove,andhavezeroconductivity(evenatniteenergydensity).IntheMBLphase,thelightparticlesareinarealaw.Inthisphaseequilibriumstatisticalmechanicsessentiallybreaksdown-theentropyperunitvolumeiszero,eventhoughthesystemisataniteenergydensity.Themicrocanonicalensemble,whenonecantakethequantumsysteminapurestate,isnotequiv-alenttothecanonicalensembleforasysteminthermalequi-libriumwithabath.Essentially,inthepresenceoftherandompotentialthequantumspectrumofthelightparticlesislocallydiscrete.Duetothisdiscreteness,itisessentiallyimpossibleforenergytomoveabout,andonepartofthesystemcannotactasathermalreservoirfortheremainingpartofthesystem.Duetothearealaw,onecanthinkoftheparticlesintheMBLphaseasessentiallybeinginadirectproductstate.Thisstateisaverylowentangledstatehavinganarealaw-indeedagenericniteenergydensitystatewillhaveavolumelawentanglemententropy.Itisimportanttoemphasizethatsuch“disentangling”requiresquantummechanics,andthereisnoclassicalanalogtotheMBLphase.OnecanthinkoftheQDLasaphasewithinwhichthelightparticlesfailtothermalizeduetothepresenceoftheheavyparticles.Whiletheheavyparticlescanmovearound 11oncetheirmassesarenite,M1=0,thelightparticlescanmovemuchmorequickly,andcanbecome“localized”bythe(nearly)staticpotentialoftheheavyparticles.Whilethispic-tureoflocalizationisnolongerstrictlyvalidoncetheheavyparticlescanmove,thedenitionintheprecedingsectionoftheQDLphaseintermsofthewavefunctionisstillprecise.Effectively,withthepositionsoftheheavyparticles“xed”inthemany-particlewavefunction,thelightparticlesareinanarealawstate.Toexposethepartialnon-thermalizationoftheQDL,con-siderahypotheticalscenariowhereonesuddenlyturnsoffallinteractionsbetweentheparticles(intra-,aswellas,inter-species).Thisallowsonetoassignacontributiontotheen-tropysolelyfromthelightparticles,`Slight'.Atthesametime,sincetheheavyparticlesintheQDLarethermalized,onecanalsoassociateawell-denedtemperatureTQDLwiththeQDL,whichcanbemeasuredpriortothemomentinter-actionsareturnedoff.Thiscanbemeasured,forexample,byconnectingtheQDLtoanon-QDLidealgas(“thermome-ter”)andmeasuringthetemperatureofthisidealgaswhenthereisnotransferofenergybetweenthetwosystems.Re-peatingtheaboveprocessatdifferenttemperatures,onecanobtainaSlight(TQDL)curve.ThemainpointisthatsincethelightparticlesintheQDLarenotthermalized,thefunc-tionSlight(TQDL)will,ingeneral,bedifferentfordifferentensemblesofthesystem,andforagivenensemble,mayevenbeindependentofTQDL,signalingabreakdownofthether-modynamicsinQDL.Suchaprocessmightberealizableinatwo-uidcoldatomicsystemwithlargemassratioforthetwocomponents,forexample,a133Cs-6Limixture40.Insuchsystems,theinteractionscanbetunedbysweepingthesys-temthroughFeshbachresonanceswhichchangestheintra-andinter-speciesscatteringlength.SubsequentlytheentropySQDLcanbemeasuredbyadiabaticallytransformingthelightparticlesintoanidealgas.V.DISCUSSIONANDOPENISSUESA.StabilityoftheQDLphase?Perhapsthemostimportantquestionpertinenttoourexplo-rationiswhetherornottheQDLphasecanexistasatyp-icalniteenergy-densityeigenstateinageneric,localtwo-componentmodelHamiltonianwithveryheavyandverylightquantumparticles?Straightforward(full)exactdiagonaliza-tionisinprinciplepossible-withthefullsetofexacteigen-statesinhandtheQDLwavefunctiondiagnosticcouldbeim-plemented.Butthiswillclearlybeverychallenging,sinceEDwillonlybepossibleinverysmallsystemswithveryfewpar-ticles-perhapsupto12sitesina1dtightbindingmodelofheavyandlightfermions.ItisunclearhoweffectiveDMRGwillbeinexploringQDLphysicsinevensimpletoymod-els.TheDMRGisbestsuitedinaccessinglowentanglementstates,usuallygroundstates,whiletheQDLstatehasavol-umelawentanglemententropy.Perhapssomevariantofdy-namicalDMRGcouldbehelpfulinusingshort-timedynam-icstoexploretheexpectedsmallarealawentanglementofthelightparticles,oncetheheavyparticlesarexed.ItappearsthatnumericalprogressontheQDLwillrequiremodicationsofexistingalgorithms/approaches,attheveryleast,andper-hapsthedevelopmentofnewalgorithmstailoredtoaccessandevolve(typical)volumelawwavefunctions.B.ExploringthepossibilityofQDLPhysicsinWaterItseemsworthwhiletoaskifaQDLphaseexistedinanactualuid,whatwouldbeitsexperimentalmanifestations?Toputourideasintoconcreteform,belowweaskcouldthepurewaterbeinaQDLphase,withthelightprotondegreesoffreedombecoming“localized”ontheoxygenions?Wethendiscusstheconsequencesifsuchwasindeedthecase,inlightofsomeofthetransportexperimentsonwater.Theoxygentoprotonmassratiois16,sowateroffers,attheveryleast,anaturalsettingtoexplorepossibleQDLphysics.Moreover,while16isnotsotremendouslylarge,thecovalentoxygen-hydrogenbonds(togetherwiththehydrogenbondsthemselves)willstronglycorrelatetherelativemotionoftheprotonsandtheoxygenions.StrongcorrelationsbetweentheheavyandlightparticlesarecentraltotheQDL.Inaddition,theprotonmustbetreatedquantummechanicallywithather-maldeBrogliewavelengthcomparabletotheinteroxygen-hydrogenseparations.Onemightbetemptedtotreattheheav-ieroxygenionssemiclassically,usingaBorn-Oppenheimerapproximation.Asdiscussedintheintroduction,theBorn-Oppenheimerpotentialproducedbyarandom(uid)congu-rationoftheoxygenions,might“localize”theprotonsontheoxygenions.ThecorrespondinglocalizedBorn-Oppenheimerwavefunction,wouldthentaketheQDLforminEq(4).Let'sthenimaginethatwateratthestandardtemperatureandpressure(STP)isinaQDLphase.Considerperfectlypurewater,withexactlytwiceasmanyhydrogenasoxygenions,andwithnoother“impurity”ionspresent.Theprotonsarethelight,chargee,spin-1/2fermions,andtheoxygenionsaretheheavy,charge2e,spin-zerobosons.IfinaQDLphase,watershouldbea“perfectinsulator”withastrictlyzero(linearresponse)electricalconductivity,despitebeingatnitetemperature.Anappliedelectricaleldwhichpushesthepro-tonsandoxygenionsinoppositedirections,couldnotsetupanelectricalcurrent(linearintheeld)sincetheprotonsare“enslaved”bytheoxygenions.Experimentally,saltwaterisafairlygoodconductor41,withaconductivityoforder101S/cm(a10\n-cmresistivity).Theconductionisdominatedbysolvatedions(salts)andtapwaterhasamuchsmallerconductivity,typically104S/cm.Dataon“pure”water,ofinteresttoushere,ishardtocomeby.Theoftenquoted“theoretical”valuefortheconductivityofpurewater42istiny,55108S/cm,andmeasuredcon-ductivitesinthe107S/cmballparkhavebeenmentionedintheliterature43.Thisconductivityisvery,verysmall,some13ordersofmagnitudesmallerthantheconductivityofametal.Itwillclearlybechallengingexperimentallytoascertainthedifferencebetween107S/cmandstrictlyzeroconduc-tivity,asinaQDLphase,especiallysinceevennominally“pure”watercontainsunwantedsolvatedionswhichwillcon- 12duct.Someprogresscanbemadebymeasuringtheelectricalconductivityof“supercriticalwater”usinghighpressures44,45,upto500kbarusing“shock”methods.Uponpressuriz-ing(nominally)purewatertheconductivityisfoundtorise.Remarkably44,45,forveryhighpressures(under“shock”)theconductivityincreasesdramaticallybysome8ordersofmag-nitude,from107S/cmatSTPto104S/cmat50kbarbeforesaturatingat20S/cmabove200kbar(estimatedtemperatureof5000K).Atthesetemperatures/pressures,purewatercon-ductsover100timesbetterthansaltwater!IfpurewateratSTPwereinfactinaninsulatingQDLphasewithzerocon-ductivity,atsuchpressuresahighlyconductingandpresum-ablyfullythermalizedphasehasbeenreached.Anotherdi-rectiontoexploreistheeffectofdeuterationontheelectricalconductivityofwater.Howwelldoesheavywaterconduct?Furtherexperimentsontheelectricalpropertiesofwatercouldbemostinteresting.C.RelevancetoFinite“temperature”QuantumFieldTheory?Viewedmorebroadly,auidofheavyandlightquantumparticlesatnite“temperatures”isaspecialcaseofatwo-componentnite“temperature”QuantumFieldTheory.OfrelevancehereisaQFTLagrangianwithtwoquantumelds,a“light”eldwithstrongquantumuctuationsinteractingstronglywitha“heavy”eldwhichisina“semi-classical”regime.Theeigenstatethermalizationhypothesis(ETH)25–28positsthatthereduceddensitymatrixinalargesub-volumeA(the“system”)obtainedbytracingoutthedegreesoffreedomintheeven-largersub-volumeB(the“environment”),equalsthethermal-densitymatrixinregionA;^ATrB ih =Z1Ae^HAQFT(62)Here^HAQFTistheQFTHamiltonianprojectedintoregionA,ZA=TrAe^HAQFTandisdeterminedbythecon-ditionthatthevonNeumannentanglemententropy,SvN=TrA^Aln(^A)satisesavolumelaw,SvN=sVAwithentropydensitysequaltothethermalentropydensity:s=T(TlnZA)=VA.Imagineatwo-componentQFT,withastronglyuctuat-ingquantumeldcoupledtoa“semi-classical”quantumeld,withniteenergy-densityeigenstatesthatareinaQDLphase.SincetheQDLwavefunctionisnotfullythermalized,itcan-notbeaccessedusingthecanonicalthermal-ensembleandtheETHisnotfullyoperative.ForaQFTintheQDLphase,ifonexesthebackground“semiclassicaleld”,theentangle-mententropyofthestronglyuctuatingquantumeldsatis-esanarealaw.Itmightbeespeciallyinterestingtoexploresuchideasfurtherinthecontextofanelementaryquantumparticleinteractingwithabackgroundsemiclassicalgravita-tionaleld.AcknowledgmentsWethankBorisShraiman,DanielFisher,FrankPollmann,JimGarrison,LeonBalents,MattHastings,MaxMetlitski,MichaelFisher,MosheGoldstein,OlexeiMotrunich,RyanMishmash,SteveShenker,T.Senthil,YvetteFisherforhelp-fuldiscussionsandespeciallyDavidHuseforhelpfuldiscus-sionsandcriticalreadingofourmanuscript.ThisresearchwassupportedinpartbytheNationalScienceFoundationun-derGrantsDMR-1101912(M.P.A.F.)andNSFPHY11-25915(T.G.),andbytheCaltechInstituteofQuantumInformationandMatter,anNSFPhysicsFrontiersCenterwithsupportoftheGordonandBettyMooreFoundation(M.P.A.F.) 1Therecould,inprinciple,alwaysexistrstordertransitionsthatseparatevariousregimesthatareotherwisesmoothlyconnected(forexample,liquid-to-gastransitioninaconventionaluid).2M.Born,J.RobertOppenheimer,AnnalenderPhysik,389(20),457(1927).3P.W.Anderson,Phys.Rev.109,1492(1958).4D.Basko,I.Aleiner,B.Altshuler,AnnalsofPhysics321,1126(2006).5I.V.Gornyi,A.D.Mirlin,D.G.Polyakov,Phys.Rev.Lett.95,206603(2005).6V.Oganesyan,D.Huse,PhysicalReviewB75,155111(2007).7M.Znidaric,T.Prosen,P.Prelovsek,Phys.Rev.B77,064426(2008).8C.Monthus,T.Garel,Phys.Rev.B81,134202(2010).9T.C.Berkelbach,D.R.Reichman,Phys.Rev.B81,224429(2010).10A.Pal,D.A.Huse,Phys.Rev.B82,174411(2010).11C.Gogolin,M.P.Muller,J.Eisert,Phys.Rev.Lett.106,040401(2011).12E.Canovi,D.Rossini,R.Fazio,G.E.Santoro,A.Silva,Phys.Rev.B83,094431(2011).13M.Rigol,V.Dunjko,V.Yurovsky,andM.Olshanii,Phys.Rev.Lett.98,050405(2007).14F.Buccheri,A.DeLuca,A.Scardicchio,Phys.Rev.B84,094203(2011).15E.Cuevas,M.Feigelman,L.Ioffe,M.Mezard,NatCommun3,1128(2012).16A.DeLuca,A.Scardicchio,Europhys.Lett.101,37003(2013).17M.Serbyn,Z.Papic,D.A.Abanin,arXiv:1305.4915(2013).18J.H.Bardarson,F.Pollmann,J.E.Moore,Phys.Rev.Lett.109,017202(2012).19R.Vosk,E.Altman,Phys.Rev.Lett.110,067204(2013).20D.A.Huse,V.Oganesyan,arXiv:1305.4915.21M.Serbyn,Z.PapicandD.A.Abanin,arXiv:1305.5554.22B.Bauer,C.Nayak,arXiv:1306.5753.23B.Swingle,arXiv:1307.0507.24Y.Bahri,R.Vosk,E.Altman,A.Vishwanath,arXiv:1307.4092.25J.M.Deutsch,Phys.Rev.A43,2046(1991).26M.Srednicki,Phys.Rev.E50,888(1994);J.Phys.A29,L75(1996);J.Phys.A32,1163(1999).27H.Tasaki,Phys.Rev.Lett.80,1373(1998).28M.Rigol,V.Dunjko,andM.Olshanii,Nature452,854(2008). 1329J.Eisert,M.Cramer,M.B.Plenio,Rev.Mod.Phys.82,277(2010).30L.Bombelli,R.K.Koul,J.Lee,R.D.Sorkin,Phys.Rev.D34,373(1986).31M.Srednicki,Phys.Rev.Lett.71,66(1993).32M.Wolf,Phys.Rev.Lett.96,010404(2006).33D.GioevandI.Klich,Phys.Rev.Lett.96,100503(2006).34T.Barthel,M.-C.Chung,U.Schollwock,Phys.Rev.A74,022329(2006).35MaxA.Metlitski,TarunGrover,arXiv:1112.5166.36M.A.Nielsen,I.L.Chuang,QuantumComputationandQuantumInformation,CambridgeUniversityPress(2011).37Notealsothatifonewereto“bias”thesignssothatthepositivesignsoccurredwithsomeprobabilityp�1=2thenonewouldndthatSA;2O(1).38M.P.A.Fisher,P.B.Weichman,G.Grinstein,andD.S.Fisher,Phys.Rev.B40,546(1989).39ThevonNeumannentropySsatisesS�S236.Therefore,ipiSi�ipiS2;iforasetofpositivenumbersfpig.40M.Repp,R.Pires,J.Ulmanis,R.Heck,E.D.Kuhnle,M.Wei-demuller,E.Tiemann,Phys.Rev.A87,010701(2013).41See,forexample,M.Hill,TheSea,Vol.1(1963)42G.Aylward,T.Findlay,SIChemicalData,3rded.,J.Wiley,NewYork(1994).43R.M.Pashley,M.Rzechowicz,L.R.Pashley,M.J.Francis,J.Phys.Chem.B,109(3),1231(2005).44A.C.MitchellandW.J.Nellis,J.Chem.Phys.76(12),6273(1982).45G.A.Lyzengab,T.J.Ahrens,W.J.NellisandA.C.Mitchell,J.Chem.Phys.76(12),6282,(1982).AppendixA:Volumelawentanglemententropyforrandom-signwavefunctionThe“random-sign”wavefunctionisgivenby (fnjg)=1 p 2Nsgn(fnjg)(A1)withthesgnfunctionchosenrandomlysgn(fnjg)=1,independentlyforeachofthe2Ncongurationsfnj=01g.Todemonstratethatthismaximallyrandom-signwavefunc-tionhasavolumelawentanglemententropy,itwillbeconve-nienttointroduceashort-handnotation, (nA;nB) (fnj2Agfnj2Bg)(A2)Withthisnotationthedensitymatrixissimply,(nA;nB;n0A;n0B)= (nA;nB) (n0A;n0B)(A3)ThereduceddensitymatrixcanbeobtainedbytracingoverthesitesinregionB:A(nA;n0A)=XnB(nA;nB;n0A;nB)(A4)wherePnBdenotesasummationovernj=01forj2B.ThevolumelawiseasiesttoestablishfromthesecondRenyientropy,whichrequirescomputing,TrA[^2A]=XnA;n0AA(nA;n0A)A(n0A;nA)(A5)Re-expressingintermsofthewavefunctionsgives,TrA[^2A]=XnA;n0AjMA(nA;n0A)2=XnB;n0BjMB(nB;n0B)2(A6)withthedenitions,MA(nA;n0A)=XnB (nA;nB) (n0A;nB)(A7)MB(nB;n0B)=XnA (nA;nB) (nA;n0B)(A8)Fortherandom-signwavefunction,MBcanbeexpressedas,MB(nB;n0B)=1 2NXnAsgn(nA;nB)sgn(nA;n0B)(A9)Considertwocases,nB=n0BandnB=n0B.Intheformercasewehave,MB(nB;nB)=2NA=2N.Forthelattercasewecanusethecentrallimittheorem,hjMB(nB=n0B)2i=2NA=22N(A10)Here,thebracketshiindicateanaverageoveranensem-bleofrandom-signwavefunctions,butforagiventypicalrandom-signwavefunctionthisformshouldholdinthelargeNAlimit.Combiningtheaboveexpressionsgives,TrA[^2A]=2NA+2NB2(NA+NB)(A11)whichimpliesthatthesecondRenyientropy,SA;2ln(2)NA,whenNANB.