• Aucun résultat trouvé

Boundary-bulk relation in topological orders

N/A
N/A
Protected

Academic year: 2021

Partager "Boundary-bulk relation in topological orders"

Copied!
16
0
0

Texte intégral

(1)

Boundary-bulk relation in topological orders

The MIT Faculty has made this article openly available.

Please share

how this access benefits you. Your story matters.

Citation

Kong, Liang, et al. “Boundary-Bulk Relation in Topological Orders.”

Nuclear Physics B 922 (September 2017): 62–76 © 2017 The

Author(s)

As Published

http://dx.doi.org/10.1016/j.nuclphysb.2017.06.023

Publisher

Elsevier

Version

Final published version

Citable link

http://hdl.handle.net/1721.1/111839

Terms of Use

Creative Commons Attribution

(2)

ScienceDirect

Nuclear Physics B 922 (2017) 62–76

www.elsevier.com/locate/nuclphysb

Boundary-bulk

relation

in

topological

orders

Liang Kong

a,

,

Xiao-Gang Wen

b

,

Hao Zheng

c

aDepartmentofMathematicsandStatistics,UniversityofNewHampshire,Durham,NH03824,USA bDepartmentofPhysics,MassachusettsInstituteofTechnology,Cambridge,MA 02139,USA

cDepartmentofMathematics,PekingUniversity,Beijing,100871,China Received 25February2017;receivedinrevisedform 26June2017;accepted 28June2017

Availableonline 5July2017 Editor: HubertSaleur

Abstract

Inthispaper,westudytherelationbetweenananomaly-freen+1Dtopologicalorder,whichareoften calledn+1Dtopologicalorder inphysicsliterature,anditsnDgappedboundaryphases.Wearguethat the n+1D bulkanomaly-freetopological orderforagiven nDgappedboundaryphaseisunique.This uniquenessdefinesthenotionofthe“bulk”foragivengappedboundaryphase.Inthispaper,weshowthat then+1D“bulk”phaseisgivenbythe“center”ofthenDboundaryphase.Inotherwords,thegeometric notionofthe“bulk”correspondspreciselytothealgebraicnotionofthe“center”.Weachievethisbyfirst introducingthenotionofamorphismbetweentwo(potentiallyanomalous)topologicalordersofthesame dimension,thenprovingthatthenotionofthe“bulk”satisfiesthesameuniversalpropertyasthatofthe “center”ofanalgebrainmathematics,i.e.“bulk= center”. Theentireargumentdoesnotrequireusto knowtheprecisemathematicaldescriptionofa(potentiallyanomalous)topologicalorder.Thisresultleads toconcretephysicalpredictions.

©2017TheAuthor(s).PublishedbyElsevierB.V.ThisisanopenaccessarticleundertheCCBYlicense (http://creativecommons.org/licenses/by/4.0/).FundedbySCOAP3.

1. Introduction

Topologicalordershaveattractedalotofattentioninrecentyearsamongcondensedmatter physicists becauseit isanewkindof orderbeyondLandau’ssymmetrybreaking theory (see

* Correspondingauthor.

E-mailaddresses:kong.fan.liang@gmail.com(L. Kong),xgwen@mit.edu(X.-G. Wen),hzheng@math.pku.edu.cn (H. Zheng).

http://dx.doi.org/10.1016/j.nuclphysb.2017.06.023

0550-3213/©2017TheAuthor(s).PublishedbyElsevierB.V.ThisisanopenaccessarticleundertheCCBYlicense (http://creativecommons.org/licenses/by/4.0/).FundedbySCOAP3.

(3)

reviews[42,38]).Therearealotofattemptsatdefiningthenotionofatopologicalorderatthe physicslevelofrigor[10,43].Amathematicallyrigorousdefinitionisstilloutofreach.

Ann+1D(spacetimedimension)topologicalorderiscalledanomaly-freeifitcanberealized byann+1Dlatticemodel;andiscalledanomalous ifotherwise[28].Inthiswork,weareonly interestedinn+1Danomaly-freetopologicalorders that allowgapped nDboundaries, which areingeneralnotunique. Ifthen+1Dbulkphaseisnottrivial,thenthe boundaryphasesare anomalousnDtopologicalorders.Inthiswork,whenwerefertobothanomaly-freeand anoma-loustopologicalorders,weusetheterm apotentiallyanomaloustopologicalorder, orsimply atopologicalorder.Themaingoalof thispaperistogiveaprecisedescriptionof therelation betweenananomaly-freen+1DtopologicalorderanditsgappednDboundaryphases.

Whenananomaly-free2+1Dtopologicalorderadmitsgappedboundaries,itiscompletely determinedbyitstopologicalexcitations(seeareview[24]).The1+1Dgappedboundaryphases arealsodeterminedbyitstopologicalexcitations.Thetopologicalexcitationsongapped1+1D boundarieswerefirststudiedinthe2+1Dtoriccodemodel[23]byBravyiandKitaevin[8]. ItwaslatergeneralizedtoLevin–Wenmodels[36]withgappedboundariesin[25],wherethe topologicalexcitationsontheboundaryof suchalatticemodel wereshowntoformaunitary fusioncategoryC,andthoseinthebulkformaunitarymodulartensorcategorywhichisgiven bytheDrinfeldcenterZ(C) ofC (seealso[33]).Buttheseworksdidnotaddresstheuniqueness ofthebulkphaseforagivenboundary.In[16],itwasshownmodel-independentlythatamong allpossiblebulk phasesassociatedtothesameboundary phaseC,theDrinfeldcenterZ(C) is

theuniversalone(aterminalobject).Onewaytocompletetheproofof theuniquenessof the bulk istoviewthe gapped boundary asaconsequence of anyon condensation[4] of agiven bulktheoryD tothetrivialphase.Thisidealeadstoaclassificationofgappedboundariesfor abelian2+1Dtopologicaltheories[20,41,35,5].Thisresult,togetherwithresultsin[16],implies theuniquenessofthebulkforabelian2+1Dtopologicaltheories.Theproofforgeneral2+1D topologicalordersappearedinthemathematicaltheoryofanyoncondensationdevelopedin[26], inwhichitwasshownthatsuchacondensationisdeterminedbyaLagrangianalgebraAinD, andC is monoidally equivalentto the category DA of A-modules inD. Moreover,we have

Z(DA) D.Thiscompletestheproofofthebulk-boundaryrelation in2+1D,whichsaysthat

the2+1DbulkphaseD foragiven1+1Dboundary C isunique,andisgivenbytheDrinfeld centerofC,i.e.bulk= centerforsimplicity.

Doesthisbulk-boundaryrelation(i.e.bulk= center)holdinhigherdimensions?Inthiswork, weproposethattheanswerisyes,andprovideaformalproofofthisrelationundersomenatural assumptions.Therearethreekeystepsinthisformalproof:

1. Forany givennDpotentially anomaloustopological orderCn,we assumethat there isa

uniqueanomaly-freen+1Dtopologicalorder,denoted byZn(Cn),suchthatCn canbe

re-alized as agapped boundary of Zn(Cn) (called unique-bulk hypothesis, see Sec. 2). We

willrefer toZn(Cn)asthebulk of Cn.Moreover,byrestrictingthen+1Dtopological

or-der Zn(Cn) to a1-codimensional subspace, we obtain an nD topological order, denoted

byPn(Zn(Cn)).WeassumethatPn(Zn(Cn))containsalltopologicalexcitationsinZn(Cn)

(a consequenceoftheself-detectionhypothesis,seeSec.2).

2. Althoughwedonot knowhowtodefineatopologicalorderrigorously,assuming the ex-istenceofsuchadefinitionandusingthenotionofthebulk,wecandefinethenotionofa morphismbetweentwotopologicalordersofthesamedimension(seeSec.4).Inparticular,

(4)

weshowthatthereisacanonicalmorphismρ: Pn(Zn(Cn))Cn→ Cn,where denotes

thestackingoftwotopologicalordersofthesamedimension.

3. Weshowthat thepair(Pn(Zn(Cn)),ρ),satisfies theuniversalpropertyof thecenterofan

algebrainmathematics(seeTheorem 5.1).Thisimpliesthatbulk= center.

Thisresultisindependentof howwedescribetheboundary/bulk phasemathematically.Itisa non-trivialresultthatleadstoconcretephysicalpredictions(seeRemark 5.4).

Wedenoten+1Dtopologicalorders byAn+1,Bn+1,Cn+1,Dn+1,etc.Ifthe spacetime

di-mensionn+1 isclearfromthecontext,we abbreviateCn+1as C.Wedenotethetrivialn+1D

topologicalorderby1n+1.Inphysics,thetrivialtopologicalorder1n+1correspondstothe

equiv-alenceclassoftheproductstates[10].

Thispaperiswrittenforworkingcondensedmatterphysicists,especiallyforthoseworking inthefieldoftopologicalphasesofmatters.Inordertoconveythesimplyidea,wetrytokeep thecategoricallanguagetotheminimum.Inparticular,wecollectallmathematicallytechnical partsinRemarksandExamples.Themaintextshouldbereadabletothosewhodonothaveany extrabackgroundincategorytheorybeyondthosebasicsthathavealreadybeenwidelyusedin condensedmatterphysics(see[24]).

Acknowledgments: X-G.W is supported by NSF Grant No. DMR-1506475 and NSFC 11274192.HZissupportedbyNSFCunderGrantNo.11131008.

2. Basicsoftopologicalorders

Inthissection,werecallsomebasicfactsabouttopologicalorders,stateourkeyassumptions andsetournotations.

ApotentiallyanomalousnDtopologicalorderCncanalwaysberealizedasagappedboundary

ofananomaly-freen+1DtopologicalorderEn+1[28].Inphysics,itisgenerallybelievedthatthe

bulkanomaly-freetopologicalorderEn+1isuniquelydeterminedbyitsgappedboundaryphase

Cn.Onewaytoseethisistonotethatthereisnopreferredlengthscale.Inordertodefinethe

boundaryphaseCn,weneeddefinetheequivalenceclassofquantumstatesbyallowingproper

deformationofthestateinanarbitrarylargeneighborhoodoftheboundary(withoutclosingthe gap).Asaconsequence,En+1shouldbeunique(see[28]fordetails).Thisuniquenessisthekey

assumptionofthispaper.Wehighlightithere.

Unique-bulkhypothesis:foranygivennDpotentiallyanomaloustopologicalorderCn,there

isauniqueanomaly-freen+1Dtopologicalorder,denotedbyZn(Cn),suchthat Cn canbe

realizedasagappedboundaryofZn(Cn).

WedenoteEn+1byZn(Cn)andrefertoitasthebulk ofCn.ItisclearthatZn(1n)= 1n+1.Inthis

work,weoftenusethefollowingpicture:

toillustratethegeometricrelationbetweentheboundaryphaseCnandthebulk phaseZn(Cn).

Moreprecisely,then+1Dbulk phaseZn(Cn)(definedonanopenn-diskasthespacemanifold)

isdepictedbyanopenintervalandthenDboundaryphaseCn(onanopenn−1-disk)isdepicted

(5)

Remark2.1.Wewanttoremarkthatthenotionofn+1Dphaseofmatterisalocalconceptwhich isdefinedonlyonanopenn-disk(asthespacemanifold).Similarly,theboundaryphaseisalso alocalconcept.Strictlyspeaking,theboundaryphaseshouldbedefinedonaneighborhoodof theboundaryinthebulk.Namely,thespacemanifoldfortheboundaryphaseis[0,)× Dn−1, whereDn−1isanopenn−1-disk.Forexample,whenn= 2,thebulk2+1Dphase,definedonan open2-disk,canbedescribedbyaunitarymodulartensorcategorytogetherwithacentralcharge

c∈ R.Its1+1Dboundary(togetherwithanopenneighborhoodinthebulk)cansupportafew differentboundaryphases,eachofwhichlivesonanopen1-diskasthespacemanifold.Different boundaryphasesareseparatedbyhighercodimensionaldefects[25,9].Aboundaryphasecanbe transformedtoanotherboundaryphaseviaapureboundaryphasetransitionwithoutalteringthe bulkphase[39].

Itiswell-knownthattopologicalexcitationsinananomaly-free2+1DtopologicalorderC3

areallparticle-like(of codimension2),andtheyformaunitary modulartensorcategory,still denotedbyC3(seeforexample[24]).Theonly1-codimensionaldefect(ordomainwall)inC3

isthetrivialone.Byrestrictingtothetrivial1-codimensionaldomainwall,alltheparticleshave tofusealongthewall.Nobraidingstructureremainsonthewall.Therefore,restrictingC3tothe

trivial1-codimensionaldomainwall,weobtaina1+1Dtopologicalorder,denotedbyP2(C3),

whichisgivenbythesameunitaryfusioncategoryasC3butforgettingitsbraidingstructure.We

wouldliketogeneralizethisfacttoalldimensionsunderthefollowingassumption:

Self-detectionhypothesis: alltopologicalexcitationsinananomaly-freetopologicalorder

Cn+1shouldbeabletodetectthemselvesviadoublebraidings[28,29].

Asaconsequence,Cn+1cannotcontainanynon-trivialtopologicalexcitations(ordefects)of

codimension 1 because two is the smallestcodimension for an excitation tobe braidedwith anotherexcitation.Therefore,byrestrictingthetopologicalorderCn+1tothetrivialexcitationof

codimension1(i.e.thetrivialdomainwall),weobtainannDtopologicalorderPn(Cn+1).Inthis

restrictingprocess,wedonotloseanynon-trivialtopologicalexcitations,noranyinformationof thefusionamongtheminnspatialdimensions,butonlyforgettheinformationoftheirfusionin then+1th direction,whichfurtherencodesthebraidingsamongexcitationsofcodimension 2. Moreover,bydoublefoldingtheanomaly-freetopologicalorderCn+1,wecreateadoublelayered

systemCn+1Cn+1withagappedboundaryphasePn(Cn+1)(i.e.Eq.(2.5)),where  isthe

stackingoperationexplainedbelowandCn+1isthetimereverseofCn+1(becausetheorientation

orthenormal(orthetime)directionofoneofthetwolayersisflipped).

OnecanstackannDpotentiallyanomaloustopologicalorderAnonthetopoftheanotherone

Bnwithoutintroducinganycouplingbetweenthetwolayersasshownin(2.1).Thisoperationis

denotedby .Moreprecisely,thedashedbox,whenviewedfromfaraway,canbeviewedasa singlenDtopologicalorderAnBnwithasingle(buttwo-layer)bulkphaseZn(An)Zn(Bn).

(6)

Thisstackingoperationiscompletelysymmetric.ItdoesnotmatterifweputAnonthetopor

the bottomof Bn becausethereisnocoupling betweenthem.Theresultingnewphaseisthe

same,i.e.

An Bn= Bn An. (2.2)

Clearly,we have1n An= An.Sincethe bulk isunique, we shouldalso havethe following

identity:

Zn(An Bn)= Zn(An)Zn(Bn).

More generally,wecanglueAn withBn byapotentiallyanomalousn+1DphaseCn+1to

obtainanewnDtopologicalorder,denotedbyAn Cn+1Bn,asshownin(2.3).Moreprecisely,

the dashedbox,when viewedfromfaraway, canbe viewedasasinglenDtopologicalorder

An C

n+1Bn,whichhasasinglebulkphasegivenbyA



n Zn+1(Cn+1)B



n.1Itisclearthat =

1

n+1.Thisoperation Cn+1 isnotsymmetricingeneral.

2

= (2.3)

Agappeddomainwall(orawall)Mnbetweentwoanomaly-free3n+1Dtopologicalorders

Cn+1andDn+1isitselfapotentiallyanomalousnDtopologicalorder.Moreover,wehave

Zn(Mn)= Cn+1Dn+1, (2.4)

whereDn+1isthetimereverseofDn+1.Asaspecialcase,wehave

Zn(Pn(Cn+1))= Cn+1 Cn+1. (2.5)

AnnDtopologicalorderAncanbeviewedasawallbetweenZn(An)and1n+1.

AwallMnbetweenCn+1andDn+1canbefusedwithawallNnbetweenDn+1andEn+1to

obtainawallMn Dn+1Nn betweenCn+1andEn+1.Thisfusionoperationofwallsisclearly

associative,i.e.forawallOnbetweenEn+1andFn+1,

(Mn D

n+1Nn)En+1On= Mn Dn+1(Nn En+1On). (2.6) Forsimplicity,wedenotethetwosidesofEq.(2.6)byMDNEO intherestofthispaper. Wehavethefollowingidentities:

1 In(2.3),notethatA

n+1= Zn(An)andBn+1= Zn(Bn).Actually,bytheuniquenessofthebulk,asageneralization

ofEq.(2.5),wehaveZn(An) = Cn+1Z

n+1(Cn+1)An+1.

2 Butwecanrotatetheleftpicturein(2.3)aroundahorizontallinepassthroughthemiddlepointofC

n+1by180

degrees.WeobtainthesameboundaryphaseAnC

n+1Bn= BnCn+1An,whereCn+1isthemirrorreflectionof

Cn+1alongthesamelineandisnotequivalenttoCn+1ingeneral.ThisalsoexplainsEq.(2.2)because1n+1= 1n+1. 3 Amoregeneralnotionofa(potentiallyanomalous)gappeddomainwallbetweentwopotentiallyanomalous topo-logicalorderscanbeintroduced(see[29,Sec. 6.1]).Butwedonotneeditinthiswork.

(7)

Fig. 1.Mnisaninvertibledomainwallbetweentwoanomaly-free n+1DtopologicalordersCn+1andDn+1,andNn

isitsinverse.The“×”inthesepicturesrepresentsanon-trivialtopologicalexcitationintheCn+1-phase.Notethat thenon-trivialnessrequires“×”tobeatleast2-codimensional.Thesepicturesdepictaprocessoftheexcitation“×” tunnelingthroughtheMnwall.Inparticular,thesecond“”isobtainedbyannihilatingNnwiththeMnontheright

sideofNn.Similarly,thereisatunnelingprocessfromDn+1toCn+1,whichisinversetoit.Thisgivesawaytoidentify theCn+1-phasewiththeDn+1-phase.

Pn(Dn+1)D

n+1Nn= Nn= Nn En+1Pn(En+1). (2.7) AgappeddomainwallMnbetweentwoanomaly-freeCn+1andDn+1iscalledinvertible if

thereisagappeddomainwallNnbetweenDn+1andCn+1suchthat

Mn Dn+1Nn= Pn(Cn+1), Nn Cn+1Mn= Pn(Dn+1). (2.8)

SuchaninvertibledomainwallMnprovidesawaytoidentifyCn+1withDn+1asdepictedin

Fig. 1.

Remark2.2.Gappeddomainwallsbetweentopologicalordersinarbitrarydimensionshavenot beenextensivelystudied(seesomediscussionin[29]).Theyare relativelywellunderstoodin 2+1D(see[26,16,15,34,21,1,22])andin1+1D(see[29]).

3. Theuniversalpropertyofthecenterofanalgebra

Inthissection,werecalltheuniversalpropertyofthecenterofanordinaryalgebra.

AnalgebraAoverafieldkisatriple(A,A⊗ A−→ A,m k−→ A),ιA wheremisthe multiplica-tionmapandιAistheunitofA.WealsodenoteιA(1)= 1A.ItscenterZ(A)isdefinedtobethe

subalgebra:

Z(A)= {z ∈ A | az = za, ∀a ∈ A}.

Thisdefinitionis,however,very limitedandnot usefultous atall.Abetterdefinition of the centerofanalgebraisgivenbyitsuniversalproperty (seeforexample[18],[37,Section6.1.4]) whichisapplicabletomanyothertypesofalgebras.

Moreprecisely,letm: Z(A)⊗ A→ A bethemultiplicationmap,i.e.m(z⊗ a)= za forzZ(A)anda∈ A.NotethatmdefinesaunitalactiononA,i.e.m(1Z(A)⊗ a)= a.Equivalently,

wehavethefollowingcommutativediagram:

Z(A)⊗ A m k⊗ A = A idA ιZ(A)⊗idA A . (3.1)

(8)

Moreover,misanalgebrahomomorphism,i.e.m(z⊗ a)m(z⊗ a)= m(zz⊗ aa)forz,z∈ Z(A)anda,a∈ A.

Thepair(Z(A),m)satisfiesthefollowinguniversalproperty:

• Givenanotherpair(X, X⊗ A−→ A) wheref X isanalgebra,f isaunitalactionandan algebrahomomorphism,thereisauniquealgebrahomomorphismf : X → Z(A) suchthat

m◦ (f ⊗ idA)= f ,ordiagrammatically,wehavethefollowingcommutativediagram:

Z(A)⊗ A m X⊗ A f f⊗idA A . (3.2)

Indeed,iff: X ⊗ A→ A isaunitalactionandanalgebrahomomorphism,then

f (x⊗ 1A)a= f (x ⊗ 1A)f (1X⊗ a) = f (x ⊗ a) = f (1X⊗ a)f (x ⊗ 1A)= af (x ⊗ 1A)

foralla∈ A,wherethefirstandthelastequalitiesholdbecausef isaunitalaction,thesecond andthethirdequalitiesholdbecausef isanalgebrahomomorphism.Therefore,theassignment

x → f (x ⊗ 1A)definesanalgebrahomomorphismf : X → Z(A) rendering(3.2)commutative.

By restricting thediagram tothe subset X⊗ 1A⊂ X ⊗ A,we see that f is theunique map

making the diagram commutative. This shows that the pair (Z(A),m) satisfies the universal property.Notethatinthespecialcase(X,f )= (Z(A),m),wehavef = idZ(A).

On theotherhand,suppose (Y,g)is anothersuchapair satisfyingthisuniversalproperty. Thenginducesanalgebrahomomorphismg: Y → Z(A).Since(Y,g)alsosatisfiesthe univer-salproperty,themapm: Z(A)⊗A→ A alsoinducesanalgebrahomomorphismm: Z(A)→ Y .

Theng◦ m hastobetheidentitymapidZ(A) bytheuniquenessintheuniversalproperty.

Sim-ilarly,m◦ g = idY.Namely, g andmareinverse toeach other,henceidentifyY withZ(A).

Therefore,theuniversalpropertydetermines(Z(A),m)uniquelyuptocanonicalisomorphism, henceprovidesanalternativeapproachtodefinethenotionofcenter.

Remark 3.1.ThecenterZ(A)ofanordinaryalgebraAisalwaysasubalgebra.Thusthedata

minthepair(Z(A),m)isredundant.However,thisisnotthecaseforothertypesofalgebras naturallyarisinginmathematics.Oneshouldkeepinmindthatthecenterofanalgebraisapair rathermerelyanalgebra.

Inmathematicallanguage,thecollectionofalgebrasoverkformasymmetricmonoidal cat-egory Alg. Roughlyspeaking, an algebra A is referred to as an object of Alg. An algebra homomorphism f : A→ B isreferred toas a morphism betweenthe objects Aand B.Two morphismsf: A→ B andg: B → C canbecomposedtogiveanewmorphismg◦ f : A→ C.

OnehasanidentitymorphismidA: A→ A foreveryobjectA.Moreover,thereisabinary

oper-ation⊗ onAlgwhichcarriesapairofobjectsA,B totheirtensorproductA⊗ B.Thisbinary operationissymmetric,i.e.A⊗ B = B ⊗ A,andunital,i.e.thereisadistinguishedobjectksuch thatk⊗ A= A forallA.ThissymmetricmonoidalcategoryAlgsatisfiesanadditionalproperty: thereexistsauniquemorphismιA: k → A foreveryobjectA.Theseareallthedatathatwehave

used tostatethe universalpropertyofthe centerof analgebra.Oncegivensuch asymmetric monoidalcategorynomatterhowcrazytheobjectsare,oneisabletowritedowntheuniversal propertyanddefinethenotionofcenter.

(9)

The collectionof nDtopological orders almost form asymmetric monoidalcategory. For example,there isasymmetric binaryoperation  (recall Eq. (2.2)) that carriesapair of nD topologicalordersAn,BntoAn Bn,andthereisadistinguishednDtopologicalorder1nsuch

that1n An= AnforallAn.OnceweknowwhatisamorphismbetweentwonDtopological

orders,wecandefinethecenterofannDtopologicalordersbyapplyingtheuniversalproperty. Thisisthesubjectofthenexttwosections.

4. Amorphismbetweentwotopologicalorders

Althoughwedonothavearigorousdefinitionoftopologicalorder,wemaytreatitasablack boxanduseittogiveaphysicaldefinitionofamorphismbetweentwotopologicalorders.

Amorphism f : Cn→ DnbetweentwonDtopologicalordersCnandDnisagappeddomain

wallfn,viewedasannDtopologicalorder,betweentwon+1Danomaly-freetopologicalorders

Zn(Cn)andZn(Dn)suchthatfn Zn(Cn)Cn= Dn.

Thegeometricidealizationofthephysicalconfigurationassociatedtothismorphismcanbe depictedasfollows:

(4.1) TheequalityfnZn(Cn)Cn= Dnmeansthatthisconfigurationisidenticaltothefollowingone:

Thisdefinitionisquiteunconventionalineitherphysicsormathematics.Wewouldliketo ex-plaintheintuitionbehindthisconcept.Inmathematics,amorphismbetweentwomathematical objects,suchasgroups,ringsandalgebras,areoftenrequiredtopreservetheinternalstructures oftheobjects.Inourcase,theinternalstructuresofatopologicalordershouldincludethe fusing-braidingstructuresofthetopologicalexcitations.Inordertopreservetheinternalstructures,we realizethenotionofamorphismf : Cn→ DnbetweentwonDtopologicalordersCnandDnby

aphysicalprocessof“screening”Cnfromoutside.Moreprecisely,weglueanewnDtopological

orderfntoCnbyann+1DglueZn(Cn)suchthatfn Z

n(Cn)Cn= Dn.Notethatatopological excitationinCnisautomaticallyatopologicalexcitationinfnZn(Cn)Cn= Dn.Inthisway,the morphismf suppliesamapfromthesetofthetopologicalexcitationsinCntothatinDn.

Intu-itively,itisreasonablethatthisprocessof“screening”fromoutsideshouldpreservetheinternal structuresofCn.Whenn= 2,weexplaininExample 4.5thatthisnotionofamorphismexactly

coincideswiththatofaunitarymonoidalfunctorbetweentwounitaryfusioncategories,which describetwo2dtopologicalorders.

Remark4.1.Ananalogofsuch amorphisminmathematicsisanalgebrahomomorphismφ: A→ B betweentwomatrixalgebrasAandB.Actually,φalwaysfactorsasA−−−−−→ C ⊗ A1C⊗idA  BwhereCisanothermatrixalgebra.Indeed,ifAisthealgebraofp× p-matricesandB isthe algebraofq× q-matrices,thentheexistenceofanalgebrahomomorphismφ: A→ B implies pdividesq andC isthealgebraof pq ×pq-matrices.Amiraclehappenshereisthatanalgebra homomorphismφisencodedbyanalgebraC.Thisphenomenonbecomeshighlynontrivialin categoricalsettings(seeExample 4.5).

(10)

Remark4.2.Strictlyspeaking,wecannotsaythatannDtopologicalorderisequaltoanother unlesswespecifyhowtheyareidentified.Suchanidentificationcanberealizedbythechoiceof aninvertiblen−1Ddomainwallaswedidin[29,Def. 4.3].Inordertoconveythesimpleidea, however,wewouldliketousetheequality“=”forsimplicity.

Two morphismsf : Cn→ Dn andg: Dn→ En canbecomposedtogetanewmorphism

g◦ f : Cn→ En whichisdefinedbythe followinggapped domainwallbetweenZn(Cn)and

Zn(En):

(g◦ f )n:= gn Zn(Dn)fn.

Thephysicalconfigurationassociatedtothiscomposedmorphismisdepictedasfollows:

NotethatthiscompositionlawisassociativebyEq.(2.6).Thatis,h◦ (g ◦ f )= (h◦ g)◦ f

foranymorphismsf : Cn→ Dn,g: Dn→ En andh: En→ Fn.Moreover,thetrivialdomain

wall Pn(Zn(Cn)) between Zn(Cn) andZn(Cn)defines the identitymorphism idCn: Cn→ Cn by Eq.(2.7).Thatis,f ◦ idCn= f foranymorphismf : Cn→ Dn andidCn◦ g = g forany morphismg: En→ Cn.Amorphismf : Cn→ Dnisanisomorphism(i.e.thereisamorphism

g: Dn→ Cnsuchthatg◦ f = idCn andf◦ g = idDn)ifandonlyiffnisaninvertibledomain wall(seeEq.(2.8)).

Remark4.3.Amorphismbetweentwomanybodysystems(notnecessarilytopological)ofthe samedimensioncanbeintroducedinasimilarway(see[29,Sec. A.3]).Thecompositionlaw is,however,notassociativeingeneral.Forthisreason,theresultofthisworkdoesnotapplyto non-topologicaltheories.

Example4.4.Wegiveafewmoreexamplesofmorphismsthatwillbeusedlater.LetCnbean

nDtopologicalorder.

1. ThereisauniquemorphismιCn: 1n→ Cn from1n toCn whichisdefinedbytheobvious

gappeddomainwallCnbetween1n+1andZn(Cn)asdepictedbelow.

(4.2) NotethatCn 1n+11n= Cn.

2. There is a morphism ρ : Pn(Zn(Cn)) Cn → Cn defined by the gapped domain wall

Pn(Zn(Cn)) Pn(Zn(Cn))betweenZn(Cn)Zn(Cn)Zn(Cn)andZn(Cn)asdepictedin

thefollowingpicture:

(11)

Notethatwehave  Pn(Zn(Cn))Pn(Zn(Cn))  Z n(Cn)Zn(Cn)Zn(Cn)  Pn(Zn(Cn))Cn  = Cn.

Tosummarize,thecollectionofnDtopologicalordersformasymmetricmonoidalcategory

TOn:anobjectisannDtopologicalorderCn,amorphismf : Cn→ Dn betweentwoobjects

Cn,DnisagappeddomainwallfnbetweenZn(Cn)andZn(Dn)asdefinedabove,andthereisa

symmetricbinaryoperation onTOn.Moreover,thereexistsauniquemorphismιCn: 1n→ Cn foreveryobjectCnaswehaveseeninExample 4.4.Thesedatasufficetodefinetheuniversal

propertyofthecenterinTOn.Beforeproceedingon,wewouldliketotakealookatthespecial

casen= 2.

Example4.5.A2DtopologicalorderC2isdescribedmathematicallybyaunitaryfusion

cate-gory,anditsbulk Z2(C2)isgivenbytheDrinfeldcenterofC2(see[25,26]).Inparticular,the

trivial2Dtopologicalorder12isdescribedbythecategoryoffinite-dimensionalHilbertspaces

H.Moreover,P2(Z2(C2))istheunderlyingunitaryfusioncategoryofZ2(C2)byforgettingthe

braidingstructure.Thestackingoperation correspondstoDelignetensorproductand,more generally,theoperation Z

2(C2) istherelativetensorproduct(see[1,Sec. 5.2]).Bydefinition,

theDelignetensorproductC2D2oftwounitaryfusioncategoriesC2,D2isalsoaunitary

fu-sioncategorywiththemonoidalstructure(cd)⊗ (cd):= (c ⊗ c)(d⊗ d)forc,c∈ C2

andd,d∈ D2.

Theresultfrom[30,Sec. 3.2]thenstatesthatamorphismbetweentwo2Dtopologicalorders isnothingbut an(isomorphismsclass of) monoidalfunctor.More explicitly,if f : C2→ D2

isaunitary monoidalfunctor betweentwo unitaryfusion categoriesC2andD2,thenthe

cor-responding unitary fusion category f2 is given by the category of C2-D2-bimodule functors FunC2|D2(D2,D2).Namely,thereisacanonicalmonoidalequivalence:

FunC2|D2(D2,D2)Z2(C2)C2 D2.

Conversely,onerecoversf fromf2asthemonoidalfunctorC2

1f2Z2(C2)idC2

−−−−−−−−−→ f2Z2(C2)C2 D2.NotethattheidentitymorphismidC2 : C2→ C2ispreciselytheidentitymonoidalfunctor.

Moreover,thereexistsauniquemonoidalfunctorιC2: H→ C2,theonethatcarriesthetensor

unitC ofH tothetensorunit1C2 ofC2.

Insummary, thesymmetricmonoidalcategoryTO2 isdescribedas follows.Anobjectisa

unitaryfusioncategory.Amorphismisan(isomorphismclassof)monoidalfunctor.The sym-metricbinaryoperation isDelignetensorproduct,andthereisadistinguishedobjectH such

thatHC2 C2forallobjectsC2inTO2.

5. bulk= center

Inthissection,weprovethatthebulk satisfiestheuniversalpropertyofthecenter.

Let Cn be an nD topological order andlet Zn(Cn)be its bulk. First, we observe that the

morphismρ: Pn(Zn(Cn))Cn→ CnfromExample 4.4isaunitalaction.Thatis,thefollowing

(12)

Pn(Zn(Cn))Cn ρ 1n Cn= Cn ιPn(Zn(Cn))idCn idCn Cn.

Indeed,thisfollowsfromthefollowingtworealizationsofthesamephysicalconfiguration:

= ,

wherethelefthandsidedepictsthemorphismρ◦ (ιPn(Zn(Cn)) idCn)andtherighthandside depictstheidentitymorphismidCn.

Nowwearereadytostateandprovethemainresultofthispaper.

Theorem5.1.Thepair(Pn(Zn(Cn)),ρ) satisfiestheuniversalpropertyofthecenter.More

pre-cisely,if(Xn,f ) isanotherpair,whereXnisannDtopologicalorderandf : Xn Cn→ Cnis

amorphismandaunitalaction,thenthereexistsauniquemorphismf : Xn→ Pn(Zn(Cn)) such

thatthefollowingdiagram

Pn(Zn(Cn)) Cn ρ Xn Cn fidCn f Cn (5.1) iscommutative.

Proof. Sincef : Xn Cn→ Cn isaunitalaction,f ◦ (ιXn idCn)= idCn.Thephysical con-figurationassociatedtothisequalityisdepictedasfollows:

= ,

(5.2) whichimpliesthat

fn Zn(Xn)Xn= Pn(Zn(Cn)) (5.3)

as gappeddomainwallsbetweenZn(Cn)andZn(Cn).Nowwe regardfn asagapped domain

wallbetweenZn(Xn)andZn(Cn)Zn(Cn).Accordingtothedefinitionofamorphism,Eq.(5.3)

(13)

Suchdefinedf makesthediagram(5.1)commutative.Thisfollowsfromthefollowingtwo realizationsofthesamephysicalconfiguration:

= ,

(5.4) wherethelefthandsidedepictsthemorphismρ◦ (f idCn)andtherighthandsidedepictsthe morphismf.

The uniqueness of f also follows from above equality. More precisely, if g : Xn

Pn(Zn(Cn))isanothermorphismmakingthediagram(5.1)commutative,thentheequality(5.4),

withthefnonthelefthandsidereplacedbygn,holds.Thisnewidentityimmediatelyimplies

thatgn= fn,i.e.g= f . 2

Notethattheuniversalpropertydeterminesthepair(Pn(Zn(Cn)),ρ)uptocanonical

isomor-phism.Thatis,if(Yn,γ )isanotherpairsatisfyingtheuniversalpropertyofthecenter,thenthe

morphismγ: Yn→ Pn(Zn(Cn))inducedbyγ isinversetothemorphismρ: Pn(Zn(Cn))→ Yn

inducedbyρhenceidentifiesYnwithPn(Zn(Cn)).Inanotherword,thepair(Pn(Zn(Cn)),ρ)is

determinedbytheuniversalpropertywithoutambiguity.Undertheterminologyofmathematics, thepair(Pn(Zn(Cn)),ρ)isthecenter ofthenDtopologicalorderCn.

RecallthatthenDtopologicalorderPn(Zn(Cn))canbeobtainedbydoublefoldingthen+1D

topologicalorderZn(Cn).TorecoverZn(Cn)fromPn(Zn(Cn)),itsufficestoreversethedouble

foldingprocess,i.e.tosplitZn(Pn(Zn(Cn))).Thisispossibleif weusethe additionaldataρ,

becauseρn isaninvertibledomainwallbetweenZn(Pn(Zn(Cn)))andZn(Cn)Zn(Cn)hence

identifiesthem(recallthe discussionsassociated toFig. 1).Namely,ρ providesasplittingof

Zn(Pn(Zn(Cn)))thusrecovers Zn(Cn). Thisshowsthat the pair (Pn(Zn(Cn)),ρ)containsthe

sameinformationasZn(Cn).

Wereachtheconclusion“bulk= center”.Thisresultisindependentofhowwedescribethe boundaryphaseandthebulk phasemathematically.

Example5.2.LetusproceedonExample 4.5.Themorphismρ: P2(Z2(C2))C2→ C2isgiven

bythemonoidalfunctor

P2(Z2(C2))C2→(P2(Z2(C2))P2(Z2(C2)))

Z

2(C2)Z2(C2)Z2(C2)(P2(Z2(C2))C2) C2,

orequivalently,bythecomposedfunctorP2(Z2(C2)) C2→ C2C2−→ C⊗ 2.Itisclearthatρ

isaunitalaction,i.e.ρ(1P2(Z2(C2)) a) a fora∈ C2.

Let f : X2 C2→ C2 be amonoidalfunctor which isalso a unitalaction. In particular, f isequippedwithnaturalisomorphismsf (x a)⊗ f (y b) f ((x ⊗ y) (a⊗ b)) and f (1X2 a) a.Thenthereisamonoidalfunctorf : X2→ C2 definedbyx → f (x 1C2).

(14)

f (x 1C2)⊗ a  f (x 1C2)⊗ f (1X2 a) f (x a) f (1X2 a)⊗ f (x 1C2)  a ⊗ f (x 1C2).

Inotherwords,f definesamonoidalfunctorfromX2toP2(Z2(C2)).Moreover,thefollowing

diagram P2(Z2(C2))C2 ρ X2C2 f fidC2 C2

iscommutative.Theuniquenessoff isalsoeasytosee.Thisshowsthatthepair(P2(Z2(C2)),ρ)

satisfiestheuniversalpropertyofthecenter.

Recallthat P2(Z2(C2))isobtainedfromZ2(C2)byforgettingthebraidingstructure.Aswe

haveargued,onecanuseρtorecoverZ2(C2)fromP2(Z2(C2)).Indeed,bytheuniversal

prop-erty of the center,the composedmonoidalfunctorμ: P2(Z2(C2)) P2(Z2(C2)) C2 idρ

−−−→ P2(Z2(C2))  C2

ρ

−→ C2 determines a monoidal functor μ : P2(Z2(C2)) P2(Z2(C2))P2(Z2(C2)).The uniqueness of μ forcesit tobe the obviousone, the tensorproductfunctor ⊗,andthemonoidalnessofμsuppliesanaturalisomorphismμ(a b)⊗ μ(cd) μ((a ⊗ c)(b⊗ d)) fora,b,c,d∈ P2(Z2(C2)).ThentheforgottenbraidingstructureonP2(Z2(C2))

can be recovered by the following natural isomorphisms b ⊗ c  μ(1 b)⊗ μ(c 1)

μ((1⊗ c) (b⊗ 1)) μ(cb) c ⊗ b.

Example 5.3.The onlyanomaly-free 1+1D topological orderis the trivial one 12, whichis

described bythecategoryof finite-dimensionalHilbertspacesH.Itwasexplainedin[29,

Ex-ample2.25] that the only1D topological orders can be described by apair (H,u), whereu

isanobjectofH.Moreover,thestackingoperationisgivenby(H,u)(H,v)= (H,u⊗ v).

NotethatZ1(H,u)= H,P1(Z1(H,u))= 11= (H,C) andρ: P1(Z1(H,u))(H,u)→ (H,u)

is the identitymorphismof (H,u).That is,bothof Z1(H,u) and(P1(Z1(H,u)),ρ)are

triv-ial. By definition, a morphism (H,u)→ (H,u) is a 1D topological order (H,v) such that

(H,v) (H,u)= (H,u);thisisequivalenttoafunctorH→ H thatcarriesutou (and car-riesC tov).Thepair(P1(Z1(H,u)),ρ)satisfiestheuniversal propertyof thecentertrivially:

if X1 (H,u)→ (H,u)isamorphismthenX1= (H,C) whichisidenticaltoP1(Z1(H,u)).

This“bulk= center”relationstillholdsevenifweincludeunstable1+1Dphases,whichoccur naturallyindimensionalreductionprocesses(see[29,Example3.7, 6.4]andalso[1,Sec. 5.2]).

Remark 5.4.Theorem 5.1 is a non-trivial result, which gives concrete physical predictions. For example, in the 3+1D Walker–Wang model [40] built on a unitary premodular tensor 1-categoryC,whichcanbeviewedasamonoidal2-category,Theorem 5.1impliesthatthe topo-logicalexcitationsinthe3+1Dbulkshallformthebraidedmonoidal2-category[19,12]given bythemonoidalcenterofthemonoidal2-categoryC constructedbyBaezandNeuchl[3].We believethatWalker–Wang’sconstructioncanbegeneralizedto3+1Dlatticemodelsbasedona genericunitaryfusion2-categoryC[29],andthebulkexcitationsinthisconjecturalmodelshould formaunitarybraidedfusion2-categorygivenbythemonoidalcenterofC.Moreover,“bulk= center”providesaseriousconstrainttotheprecisemathematicalformulationoftopological or-dersinalldimensions.Itled ustoproposein[29]acategoricaldescription ofthetopological excitationsinpotentiallyanomaloustopologicalordersinalldimensions.

(15)

Remark5.5.Itwasexplainedin[29]that“bulk =center”discussedinthisworkisonlythefirst layerofthecompleteboundary-bulkrelation,whichcanbesummarizedasthefunctorialityof thecenter.For2+1Dtopologicalorderswithgappedboundaries,thisfunctorialityofDrinfeld centerofunitaryfusioncategorieswasprovedrigorouslyin[30].

Remark5.6.Ourmainresult alsosheds lightson asimilarboundary-bulkduality(i.e. open-closedduality)in2DrationalCFT’s[13,27,11,7].Itwillalsobeinterestingtostudyitsrelation tosomeresultsinfactorizationalgebras(see[37,14,2,17,6,1]).

Beforeweconcludethispaper,wewouldliketogiveanotherimportantremark,whichhas alreadyledtosomenewresults[31,44,32].

Remark5.7.Ifann+1Danomaly-freetopologicalorderhasatopologicallyprotectedgapless

nDboundaryphaseandiftheunique-bulkhypothesisstillholds,thenitiseasytoseethatour proofalsoworksforthegaplessboundarycase.Therefore,thegappedbulkphaseassociatedtoa gaplessboundaryphaseshouldalsobegivenbythecenteroftheboundaryphase.Inparticular,in 2+1DquantumHallsystems,thisresultsuggeststhatthereshouldbeamathematicaldescription ofthe1+1DgaplessedgemodessuchthatitscentergivesthemodulartensorcategoryC that describesthe2+1Dbulkphase.In[31],itwasshownthatthereisanenrichedmonoidalcategory

C(enrichedbyboundaryCFT’s)suchthatitsDrinfeldcenterisexactlyC.In[32],weshowed

thatsuchenrichedmonoidalcategoriesgiveamathematicaldescriptionofthegaplessboundary phasesof 2+1D topological orders. Sucha mathematicaldescription also provides awayto extendtheReshetikhin–Turaev2+1DTQFTdowntopoints[44].

References

[1]Y.-H.Ai,L.Kong,H.Zheng,Topologicalordersandfactorizationhomology,arXiv:1607.08422. [2]D.Ayala,J.Francis,H.L.Tanaka,Factorizationhomologyofstratifiedspaces,arXiv:1409.0848.

[3]J.Baez,M.Neuchl,Higher-dimensionalalgebraI:braidedmonoidal2-categories,Adv.Math.121 (2)(1996) 196–244.

[4]F.A.Bais,J.K.Slingerland,Condensateinducedtransitionsbetweentopologicallyorderedphases,Phys.Rev.B79 (2009)045316.

[5]M.Barkeshli,C.-M.Jian,X.-L.Qi,TheoryofdefectsinAbeliantopologicalstates,Phys.Rev.B88(2013)235103, arXiv:1305.7203.

[6]D.Ben-Zvi,A.Brochier,D.Jordan,Quantumcharactervarietiesandbraidedmodulecategories,arXiv:1606.04769. [7]M.Bischoff,Y.Kawahigashi,R.Longo,K.-H.Rehren,PhaseboundariesinalgebraicconformalQFT,Commun.

Math.Phys.342(2016)1–45,arXiv:1405.7863.

[8]S.B.Bravyi,A.Y.Kitaev,Quantumcodesonalatticewithboundary,arXiv:quant-ph/9811052.

[9]J.Cano,M.Cheng,M.Barkeshli,D.J.Clarke,C.Nayak,Chirality-protectedmajoranazeromodesatthegapless edgeofAbelianquantumHallstates,Phys.Rev.B92(2015)195152.

[10]X.Chen,Z.-C.Gu,X.-G.Wen,Localunitarytransformation,long-rangequantumentanglement,wavefunction renormalization,andtopologicalorder,Phys.Rev.B82(2010)155318,arXiv:1004.3835.

[11]A.Davydov,Centreofanalgebra,Adv.Math.225(2010)319–348,arXiv:0908.1250[math.CT]. [12]B.Day,R.Street,MonoidalbicategoriesandHopfalgebroids,Adv.Math.129(1997)99–157.

[13]J.Fjelstad,J.Fuchs,I.Runkel,C.Schweigert,Uniquenessofopen/closedrationalCFTwithgivenalgebraofopen states,Adv.Theor.Math.Phys.12(2008)1283–1375,arXiv:hep-th/0612306.

[14]J.Francis,ThetangentcomplexandHochschildcohomologyofrings,Compos.Math.149 (3)(2013)430–480. [15]J.Fuchs,C.Schweigert,Anoteonpermutationtwistdefectsintopologicalbilayerphases,Lett.Math.Phys.

104 (11)(2014)1385–1405,arXiv:1310.1329.

[16]J.Fuchs,C.Schweigert,A.Valentino,Bicategoriesforboundaryconditionsandforsurfacedefectsin3-dTFT, Commun.Math.Phys.321 (2)(2013)543–575,arXiv:1203.4568.

(16)

[17]G.Ginot,Notesonfactorizationalgebras,factorizationhomologyandapplications,in:MathematicalAspectsof QuantumFieldTheories,2015,pp. 429–552,PartoftheseriesMathematicalPhysicsStudies,arXiv:1307.5213. [18]A.Joyal,R.Street,Themonoidalcentreasalimit,TheoryAppl.Categ.13 (13)(2004)184–190.

[19]M.M.Kapranov,V.A.Voevodsky,2-categoriesandZamolodchikovtetrahedraequations,Algebraicgroupsandtheir generalizations:quantumandinfinite-dimensionalmethods,Proc.Symp.PureMath.56(1994)177–259. [20]A.Kapustin,N.Saulina,TopologicalboundaryconditionsinabelianChern–Simonstheory,Nucl.Phys.B845

(2011)393–435,arXiv:1008.0654[math.QA].

[21]Y.Kawahigashi,Aremarkongappeddomainwallsbetweentopologicalphases,Lett.Math.Phys.105(2015) 893–899,arXiv:1504.01088.

[22]Y.Kawahigashi,Arelativetensorproductofsubfactorsoveramodulartensorcategory,arXiv:1612.03549. [23]A.Y.Kitaev,Fault-tolerantquantumcomputationbyanyons,Ann.Phys.303(2003)2–30,arXiv:quant-ph/9707021. [24]A.Y. Kitaev, Anyons in an exactly solved model and beyond, Ann. Phys. 321 (2006) 2–111,

arXiv:cond-mat/0506438.

[25]A.Kitaev,L.Kong,Modelsforgappedboundariesanddomainwalls,Commun.Math.Phys.313(2012)351–373, arXiv:1104.5047.

[26]L.Kong,Anyoncondensationandtensorcategories,Nucl.Phys.B886(2014)436–482,arXiv:1307.8244. [27]L. Kong, I. Runkel, Cardy algebras and sewingconstraints, I, Commun. Math. Phys. 292 (2009)871–912,

arXiv:0807.3356.

[28]L.Kong,X.-G.Wen,Braidedfusioncategories,gravitationalanomaliesandthemathematicalframeworkfor topo-logicalordersinanydimensions,arXiv:1405.5858.

[29]L.Kong,X.-G.Wen,H.Zheng,Boundary-bulkrelationfortopologicalordersasthefunctormappinghigher cate-goriestotheircenters,arXiv:1502.01690.

[30]L.Kong,H.Zheng,Thecenterfunctorisfullyfaithful,arXiv:1507.00503.

[31]L.Kong,H.Zheng,Drinfeldcenterofenrichedmonoidalcategories,arXiv:1704.01447.

[32]L.Kong,H.Zheng,Gaplessedgesof2dtopologicalordersandenrichedmonoidalcategories,arXiv:1705.01087. [33]T.Lan,X.-G.Wen,Topologicalquasiparticlesandtheholographicbulk-edgerelationin2+1Dstring-netmodels,

Phys.Rev.B90(2014)115119,arXiv:1311.1784.

[34]T.Lan,J.Wang,X.-G.Wen,Gappeddomainwalls,gappedboundariesandtopologicaldegeneracy,Phys.Rev.Lett. 114(2015)076402,arXiv:1408.6514.

[35]M.A.Levin,Protectededgemodeswithoutsymmetry,Phys.Rev.X3(2013)021009,arXiv:1301.7355.

[36]M.A.Levin,X.-G.Wen,String-netcondensation:aphysicalmechanismfortopologicalphases,Phys.Rev.B71 (2005)045110,arXiv:cond-mat/0404617.

[37] J.Lurie,HigherAlgebras,abookavailableat:http://www.math.harvard.edu/~lurie/papers/higheralgebra.pdf. [38]C.Nayak,S.H.Simon,A.Stern,M.Freedman,S.D.Sarma,Non-Abeliananyonsandtopologicalquantum

compu-tation,Rev.Mod.Phys.80(2008)1083.

[39]E.Plamadeala,M.Mulligan,C.Nayak,Short-rangeentangledbosonicstateswithchiraledgemodesandT-duality ofheteroticstrings,Phys.Rev.B88(2013)045131.

[40]K.Walker,Z.Wang,(3+1)-TQFTsandtopologicalinsulators,Front.Phys.7 (2)(2012)150–159,arXiv:1104.2632. [41]J.Wang,X.-G.Wen,Boundarydegeneracyoftopologicalorder,Phys.Rev.B91(2015)125124,arXiv:1212.4863. [42]X.G.Wen,QuantumFieldTheoryofMany-BodySystems,OxfordUniv.Press,Oxford,2004.

[43]B.Zeng,X.-G.Wen,Gappedquantumliquidsandtopologicalorder,stochasticlocaltransformationsandemergence ofunitarity,Phys.Rev.B91(2015)125121,arXiv:1406.5090.

Figure

Fig. 1. M n is an invertible domain wall between two anomaly-free n + 1D topological orders C n + 1 and D n + 1 , and N n is its inverse

Références

Documents relatifs

A comparative study of five boundary treatment schemes for stationary complex boundaries in the lattice Boltzmann method..

In conclusion, we have computed numerically the low lying eigenenergies of three spin-1/2 fermions in a box, interacting with an infinite scattering length in a lattice model, for

In Bezug auf den Zwangsdienst von (zumeist) Männern im Militär einer liberalen Demokratie existiert eine Konfliktlinie zwischen einerseits der Interpretation, dass der Staat

From this, it can be argued that E is the algebra of observables for the family over s of systems describing the motion of an electric particle, in the aperiodic solid described by

Hans Anton von Roten hat festgestellt, dass es sich bei Johannes von Sitten, um einen 1343 erstmals als Rektor der Dreifaltigkeitskapelle in Sit- ten erscheinenden Kleriker

point in the direction of the y-axis is evidently a real birational transformation of a, or a conformal transformation of 8 with preservation of conjugate

A topological based clustering method has been proposed for the segmentation of the cortex in fetal brain MR images, which takes advantages of anatomical knowledges.. The

If the dynamics of the particle positions and directions in the body orientation model was the same as in the Vicsek model, these three classes of solutions should have a constant