Table Of Content

arXiv:gr-qc/0601095v1 23 Jan 2006 E d i t e d b y 1 The spin-foam-representation of LQG Alejandro Perez Centre de Physique Théorique, Campus de Luminy, 13288 Marseille, France. Unité Mixte de Recherche (UMR 6207) du CNRS et des Universités Aix-Marseille I, Aix-Marseille II, et du Sud Toulon-Var; laboratoire afilié à la FRUMAM (FR 2291). Abstract The problem of background independent quantum gravity is the problem of defining a quantum field theory of matter and gravity in the absence of an underlying background geometry. Loop quantum gravity (LQG)isapromisingproposalforaddressingthisdifficulttask. Despite the steady progress of the field, dynamics remains to a large extend an open issue in LQG. Here we present the main ideas behind a series of proposals for addressing the issue of dynamics. We refer to these con- structions as the spin foam representation of LQG. This set of ideas can be viewed as a systematic attempt at the construction of the path integral representation of LQG. The spin foam representation is mathematically precise in 2+1 dimensions, so we will start this chapter by showing how it arises in the canonicalquantizationofthissimpletheory. Thistoymodelwillbeused topreciselydescribethetruegeometricmeaningofthehistoriesthatare summed over in the path integral of generally covarianttheories. In four dimensions similar structures appear. We call these construc- tionsspin foam modelsastheirdefinitionisincompleteinthesencethat at least one of the following issues remains unclear: 1) the connection to a canonical formulation, and 2) regularization independence (renor- malizability). In the second part of this chapter we will describe the definition of these models emphasizing the importance of these open issues. 1 2 Alejandro Perez 1.1 The path integral for generally covariant systems LQG is based on the canonical (Hamiltonian) quantization of general relativity whose gauge symmetry is diffeomorphism invariance. In the Hamiltonianformulationthepresenceofgaugesymmetries(DiracP.M.) givesrisetorelationshipsamongthephasespacevariables—schematically C(p,q)=0for(p,q)∈Γ—whicharereferredtoasconstraints. Thecon- straintsrestrictthesetofpossiblestatesofthetheorybyrequiringthem to layonthe constrainthyper-surface. Inaddition, throughthe Poisson bracket,theconstraintsgeneratemotionassociatedtogaugetransforma- tionsontheconstraintsurface(seeFig. (1.1)). Thesetofphysicalstates (the so called reduced phase space Γ ) is isomorphic to the space of red orbits,i.e., two points onthe same gaugeorbitrepresentthe same state in Γ described in different gauges (Fig. 1.1). red In general relativity the absence of a preferred notion of time implies that the Hamiltonian of gravity is a linear combination of constraints. This means thatHamilton equationscannotbe interpretedastime evolution and rather correspond to motion along gauge orbits of general relativity. In generally covariant systems conventional time evolution is pure gauge: from an initial data satisfying the constraints one re- covers a spacetime by selecting a particular one-parameter family of gauge-transformations (in the standard ADM context this amounts for choosing a particular lapse function N(t) and shift Na(t)). ConstraintHamiltonian Γ vector field Γ quantizing H kin gauge orbit CONSTRAINT reducing reducing pin foam rep) SURFACE (s Γ quantizing H red phys Γ red Fig. 1.1. On the left: the geometry of phase space in gauge theories. On the right: thequantization path of LQG(continuous arrows). Fromthisperspectivethenotionofspacetimebecomessecondaryand the dynamical interpretation of the the theory seems problematic (in the quantum theory this is refered to as the “problem of time”). A The spin-foam-representation of LQG 3 possible reason for this apparent problem is the central role played by the spacetime representation of classical gravity solutions. However, the reason for this is to a large part due to the applicability of the conceptoftestobservers(ormoregenerallytestfields)inclassicalgeneral relativity†. Duetothefactthatthisidealizationisagoodapproximation tothe(classical)processofobservationthe notionofspacetimeisuseful in classical gravity. As emphasized by Einstein with his hole argument (see Rovelli C. (2005) for a modern explanation) only the information in relational statements (independent of any spacetime representation) have physical meaning. In classical gravity it remains useful to have a spacetime representation when dealing with idealized test observers. For instance tosolvethegeodesicequationandthenaskdiff-invariant-questionssuch as: what is the proper time elapsed on particle 1 between two succesive crossings with particle 2? However, already in the classical theory the advantageofthe spacetime picture becomes,by far,less clearif the test particles are replaced by real objects coupling to the gravitational field †. However, this possibility is no longer available in quantum gravity where at the Planck scale (ℓ ≈ 10−33cm) the quantum fluctuations of p the gravitational field become so important that there is no way (not even in principle‡) to make observations without affecting the gravita- tionalfield. Inthis contextthere cannotbe any,a priori, notionoftime andhencenonotionofspacetimeispossibleatthefundamentallevel. A spacetime picture wouldonly arise in the semi-classicalregime with the identification of some subsystems that approximate the notion of test observers. What is the meaning of the path integral in the background independent context? The previous discussion rules out the conventional † Most (if not all) of the textbook applications of general relativity make use of this concept together with the knowledge of certain exact solutions. In special situationsthereareevenpreferredcoordinatesystemsbasedonthisnotionwhich greatlysimplifyinterpretation(e.g. co-movingobserversincosmology,orobservers atinfinityforisolatedsystems). † In this case one would need first to solve the constraints of general relativity in order to find the initial data representing the self-gravitating objects. Then one wouldhave essentiallytwochoices: 1) Fixalapse N(t) andashiftNa(t), evolve withtheconstraints,obtainaspacetime(outofthedata)inaparticulargauge,and finallyaskthediff-invariant-question;or2)trytoanswerthequestionbysimply studyingthe data itself(without t-evolution). It isfarfromobvious whether the firstoption(theconventional one)isanyeasierthanthesecond. ‡ InordertomakeaPlanckscaleobservationweneedaPlanckenergyprobe(think ofaPlanckenergyphoton). Itwouldbeabsurdtosupposethatonecandisregard theinteractionofsuchphotonwiththegravitationalfieldtreatingitastestphoton. 4 Alejandro Perez interpretation of the path integral. There is no meaningful notion of transition amplitude between states at different times t >t or equiv- 1 0 alently a notion of “unitary time evolution” represented by an operator U(t −t ). Nevertheless, a path integral representation of generally co- 1 0 variant systems arises as a tool for implementing the constraints in the quantum theory as we argue below. Due to the difficulty associated with the explicit description of the reducedphasespace Γ , in LQGone followsDirac’s prescription. One red startsbyquantizingunconstrainedphasespaceΓ,representingthecanon- ical variables as self-adjoint operators in a kinematical Hilbert space H . Poisson brackets are replaced by commutators in the standard kin way, and the constraints are promoted to self-adjoint operators (see Fig. 1.1). If there are no anomalies the Poisson algebra of classical constraints is represented by the commutator algebra of the associated quantum constraints. In this way the quantum constraints become the infinitesimalgeneratorsof gaugetransformationsin H . The physical kin HilbertspaceH isdefinedasthekerneloftheconstraints,andhence phys associatedto gauge invariant states. Assuming for simplicity that there is only one constraint we have ψ ∈H iff exp[iNCˆ]|ψi=|ψi ∀ N ∈R, phys whereU(N)=exp[iNCˆ]istheunitaryoperatorassociatedtothegauge transformation generated by the constraint C with parameter N. One can characterize the set of gauge invariant states, and hence construct H ,byappropriatelydefininganotionof‘averaging’alongtheorbits phys generatedbytheconstraintsinH . Forinstanceifonecanmakesense kin of the projector P :H →H where P := dN U(N). (1.8) kin phys Z It is apparent from the definition that for any ψ ∈ H then Pψ ∈ kin H . The path integral representation arises in the representation of phys theunitaryoperatorU(N)asasumovergauge-historiesinawaywhich istechnicallyanalogoustostandardpathintegralinquantummechanics. The physicalinterpretationis howeverquite differentas wewill show in Sec. 1.2.4. The spin foam representation arises naturally as the path integralrepresentationofthe fieldtheoreticalanalogofP inthecontext of LQG. Needles is to say that many mathematical subtleties appear when one applies the above formal construction to concrete examples (Giulini D. & Marolf D., (1999)). The spin-foam-representation of LQG 5 1.2 Spin foams in 3d quantum gravity Herewederivethespin foam representationofLQGinasimplesolvable example: 2+1 gravity. For the definition of spin foam models directly in the covariant picture see Freidel (2005), and other approaches to 3d quantum gravity see Carlip S. (1998). 1.2.1 The classical theory Riemanniangravity in 3 dimensions is a theory with no local degrees of freedom, i.e., a topological theory. Its action (in the first order formal- ism) is given by S[e,ω]= Tr(e∧F(ω)), (1.9) Z M whereM =Σ×R (forΣ anarbitraryRiemannsurface),ω is anSU(2)- connection and the triad e is an su(2)-valued 1-form. The gauge symmetries of the action are the local SU(2) gauge transformations δe=[e,α], δω =d α, (1.10) ω where α is a su(2)-valued 0-form, and the ‘topological’ gauge transformation δe=d η, δω =0, (1.11) ω whered denotesthecovariantexteriorderivativeandηisasu(2)-valued ω 0-form. Thefirstinvarianceismanifestfromtheformoftheaction,while the second is a consequence of the Bianchi identity, d F(ω) = 0. The ω gauge symmetries are so large that all the solutions to the equations of motionarelocallypuregauge. Thetheoryhasonlyglobalortopological degrees of freedom. Uponthe standard2+1decomposition,the phase spacein these variables is parametrized by the pull back to Σ of ω and e. In local coor- dinates one can express them in terms of the 2-dimensional connection Ai andthe triadfieldEb =ǫbcekδ wherea=1,2arespacecoordinate a j c jk indices and i,j =1,2,3 are su(2) indices. The Poisson bracket is given by {Ai(x),Eb(y)}=δb δi δ(2)(x,y). (1.12) a j a j Local symmetries of the theory are generated by the first class constraints D Eb =0, Fi (A)=0, (1.13) b j ab 6 Alejandro Perez which are referred to as the Gauss law and the curvature constraint re- spectively. Thissimpletheoryhasbeenquantizedinvariouswaysinthe literature,here we will use it to introduce the spin foam representation. 1.2.2 Spin foams from the Hamiltonian formulation The physical Hilbert space, H , is defined by those ‘states in H ’ phys kin thatareannihilatedbytheconstraints. AsdiscussedinThiemann(2005) (see also Rovelli C. (2005) and Thiemann (2005)), spin network states \ solvetheGaussconstraint—D Ea|si=0—astheyaremanifestlySU(2) a i gaugeinvariant. Tocompletethequantizationoneneedstocharacterize the space of solutions of the quantum curvature constraintsFi , and to ab provide it with the physical inner product. As discussed in Sec. 1.1 we b can achieve this if we can make sense of the following formalexpression for the generalized projection operator P: [ P = D[N] exp(i Tr[NF(A)])= δ[F(A)], (1.14) Z Z xY⊂Σ Σ b where N(x)∈su(2). Notice that this is just the field theoretical analog ofequation(1.8). P willbedefinedbelowbyitsactiononadensesubset of test-states called the cylindrical functions Cyl ⊂ H (see Ashtekar kin & Lewandowski(2004)). If P exists then we have hsPU[N],s′i=hsP,s′i ∀ s,s′ ∈Cyl, N(x)∈su(2) (1.15) whereU[N]=exp(i Tr[NFˆ(A)]). P canbeviewedasamapP :Cyl→ K ⊂ Cyl⋆ (the spaRce of linear functionals of Cyl) where K denotes F F the kernel of the curvature constraint. The physical inner product is defined as hs′,si :=hs′P,si, (1.16) p where h,i is the inner product in H , and the physical Hilbert space kin as H :=Cyl/J for J :={s∈Cyl s.t. hs,si =0}, (1.17) phys p where the bar denotes the standard Cauchy completion of the quotient space in the physical norm. One can make (1.14) a rigorousdefinition if one introduces a regularization. Aregularizationis necessarytoavoidthe naiveUV divergences that appear in QFT when one quantizes non-linear expressions of the The spin-foam-representation of LQG 7 canonicalfields suchas F(A) inthis case (orthose representinginterac- tions in standard particle physics). A rigorous quantization is achieved iftheregulatorcanberemovedwithouttheappearanceofinfinities,and if the number of ambiguities appearing in this process is under control (more about this in Sec. 1.3.1). We shall see that all this can be done in the simple toy example of this section. W p ε Σ Fig. 1.2. Cellular decomposition of the space manifold Σ (a square lattice of size ǫ in this example), and the infinitesimal plaquette holonomy Wp[A]. We now introduce the regularization. Givena partitionof Σ in terms of2-dimensionalplaquettesofcoordinateareaǫ2 (Fig.1.2)onecanwrite the integral F[N]:= Tr[NF(A)]= lim ǫ2Tr[N F ] (1.18) p p Z ǫ→0 Xp Σ as a limit of a Riemann sum, where N and F are values of the smear- p p ing field N and the curvature ǫabFi [A] at some interior point of the ab plaquette p and ǫab is the Levi-Civita tensor. Similarly the holonomy W [A] around the boundary of the plaquette p (see Figure 1.2) is given p by W [A]= +ǫ2F (A)+O(ǫ2). (1.19) p p 1 The previous two equations imply that F[N] = lim Tr[N W ], ǫ→0 p p p and lead to the following definition: given s,s′ ∈ Cyl P(think of spin network states) the physical inner product (1.16) is given by hs′P,si:= lim hs dN exp(iTr[N W ]),si. (1.20) p p p ǫ→0 Z Yp The partition is chosen so that the links of the underlying spin network graphs border the plaquettes. One can easily perform the integration over the N using the identity (Peter-Weyl theorem) p j dN exp(iTr[NW])= (2j+1) Tr[Π(W)], (1.21) Z Xj 8 Alejandro Perez j where Π(W) is the spin j unitary irreducible representation of SU(2). Using the previous equation hs′P,si:= lim np(ǫ) (2jp+1) hs′ Tr[jΠp(Wp)]),si, (1.22) ǫ→0 Yp Xjp where the spin j is associated to the p-th plaquette, and n (ǫ) is the p p numberofplaquettes. SincetheelementsofthesetofWilsonloopopera- tors{W } commute,the orderingofplaquette-operatorsinthe previous p product does not matter. The limit ǫ → 0 exists and one can give a closedexpressionforthe physicalinnerproduct. Thatthe regulatorcan be removedfollows from the orthonormalityof SU(2) irreducible repre- sentationswhichimpliesthatthetwospinsumsassociatedtotheaction oftwoneighboringplaquettescollapsesintoasinglesumovertheaction ofthe fusionofthecorrespondingplaquettes(seeFig1.3). Onecanalso showthatitisfinite†,andsatisfiesallthepropertiesofaninnerproduct (Noui K. & Perez A. (2005)). (2j+1)(2k+1) = (2k+1) Pjk j k Pk k Fig. 1.3. In two dimensions the action of two neighboring plaquette-sums on the vacuum is equivalent to the action of a single larger plaquette action obtainedfromthefusionoftheoriginalones. Thisimpliesthetrivialscalingof thephysicalinnerproductunderrefinementoftheregulatorandtheexistence of a well definedlimit ǫ→0. 1.2.3 The spin foam representation Each Tr[jΠp(Wp)] in (1.22) acts in Hkin by creating a closed loop in the j representationat the boundary of the correspondingplaquette (Figs. p 1.4 and 1.6). Now, in order to obtain the spin foam representation we introduceanon-physical(coordinatetime)asfollows: Insteadofworking † The physical inner product between spin network states satisfies the following inequality hs,s′ip ≤C (2j+1)2−2g, (cid:12) (cid:12) Xj (cid:12) (cid:12) forsomepositiveconstantC. Theconvergenceofthesumforgenusg≥2follows