Вы находитесь на странице: 1из 5

String Theory, Quantum Gravity and Locality Cyrus C.

Taylor Case WesternReserve University In this talk, I'd like to explain a little bit about what stringtheory is, why theoretical physicists are so excited about it, and why I think that it will have a ratherprofound of impact on some of our ideas about the structure the physical world. Let me begin by reviewing the way we think aboutquantummechanicalparticlesin a relativistic setting. The classical dynamics of the theory is specified by assuming that the action for a given particle trajectoryis proportionalto the relativisticintervaltraversed the by the particle. If one sets up a canonical formalismwith the variableparametrizing of particletrajectoryplaying the role of time, then one finds that the four-momentum the particleis constrained,with the constraintjust being the relativisticrelationbetween the of mass, energy, and three-momentum a particle. The particlethus lives on a subspaceof the full phase space, and one finds thaton this subspace, the Hamiltonianfor the system vanishes. The quantizationof this system is staightforward.Let us adopt the Schrodingerpicture.Then the Schrodingerequationjust tells us that the wave function doesn't depend on how we parametrizethe particlestrajectory. (This is a consequence of the fact that the Hamiltonianvanishes). However, we have to implementour constraint. Demanding that it annihilatethe wave function when we treatit as a quantummechanical operator,we find that we have derived the Klein-Gordanequation. For a review of the general procedure,see for example (Scherk 1975). Quantizationof the stringproceeds in an analogousfashion. One postulatesthat the action for a given stringtrajectoryis proportionalto the areaof the world sheet, sets up a canonical formalism, and again finds thatthe system is constrained. In this case, however, one finds that there are an infinite numberof constraints,with the Hamiltonianagain vanishing on the constraintsurface.These constraints,which have non-vanishingPoisson Because bracketswith each other,can be thoughtof as generatinggauge transformations. of normalorderingambiguitiesin the expressions for the constraints,the quantization requiresmore care than in the case of the particle. (This is perhapsnot terriblysurprising, since we can thinkof the phase space of the string as being infinite dimensional, correspondingto the Fouriermodes of the position of the string). For details of the procedure, see, for example, (Green, Schwarzand Witten 1987). The upshot is that in twentysix dimensions, (or ten in the case of the superstring), one can constructa unitaryquantum theory.The theory now has an infinite numberof states, including states thatcan be identified with gauge bosons (open string)and with the graviton(closed string). Further, the theory is again invariantunderreparametrizations the stringworld sheet. This is of quite importantonce we consider interactingstrings.

PSA 1988, Volume 2, pp. 107-111 Copyright? 1989 by the Philosophy of Science Association

108 In the case of a theoryof interactingparticles,the pictureof interactionsis not terribly nice from a geometric viewpoint: the process in which, say, threeparticlesinteractat a local vertex prohibitsus from thinkingof the worldlinesof the particlesas a manifold. interactions,and so on. In the Three-particleinteractionsare distinct from four-particle context of a second-quantizedtheory,this becomes quite important,and is one way of understanding problemsassociatedwith renormalizability.Briefly, one specifies the theory by assuming some finite set of kinds of particle interactions.For example, in quantum electrodynamics,one has a single kind of three-particle interaction,correspondingto the interactionof a photon with, say, an incoming and outgoing electron. When one calculates amplitudesfor the scatteringof variousparticles,one finds typically thatthey diverge. If these infinities can be removed by a systematicscheme for redefiningthe coupling constantsassociated with the originalinteractionvertices, then the theory is renormalizable.(See Itzykson and Zuber 1980 for a textbook accountof this procedure.) If not, then new vertices must be introducedto cancel the infinities. These typically requirethe introductionof still more vertices. Typically,these will in turnresult in new infinities, and one is left with a theoryin which one would need to specify an infinite amountof informationin orderto calculate anything. Such theories are said to be nonrenormalizable,and since they have essentially no predictivepower, are avoided like the plague. Einstein's theoryof generalrelativity,as presentlyunderstood,is non-renormalizable evident when matteris in four space-timedimensions. These problemsare particularly included. See (Goroff and Sagnotti 1986), and referencestherein. In the context of stringtheory,the pictureis much rosier. Stringsdon't join at a point, but rather(in the case of closed strings,for example) througha process which can be depicted as a plumber'sT-junction. In twenty six dimensions, the theoryis invariant,so thatwe can thinkaboutFeynmandiagramsas being built reparametrization out of T-junctionsand lengths of pipe, all of which are made out of rubber. That is, we identify diagramsif they can be continuouslydeformedinto one another. The upshotis that therereally is only one kind of interaction,and so there appearsto be no mechanism This doesn't constitutea proof of renorby which the theorycan be non-renormalizable. there appearto be some technical malizability,and indeed, in the case of superstrings, obstructionsto carryingout an actualproof; nevertheless,the basic pictureis very appealingand widely acceptedas being (provisionally)true. For a review of the current statusof stringperturbation theory,see (D'Hoker and Phong 1988). Thus far, we have discussed stringpropagationin a flat twenty-six (or ten) dimensional spacetime. The first question one asks as a physicist is: where is the real world? Candelas,et al., (1985) proposedthat,as a first step, we assume that the ten dimensional of spacetime has the structure a productof flat four-dimensionalMinkowksi space with a six-dimensionalcompact manifold. The rough idea is thatif the size of the internalspace is small comparedwith the distances which we can currentlyprobe, then we are only seeing the low energy effective stringtheory,which will be four dimensional. Candelas,et. al, (1985) showed that the heteroticstringconsistentlypropagateson such a manifold if it of has the structure a Calabi-Yaumanifold. Further, they showed that such compactifications can lead to realistic low energy theories. The numberof such Calabi-Yaucompactifications is quite large, but believed to be finite (Yau 1985). Among these is a manifold yielding a low energy theorywith only three generationsof fermions.This may be an essentially unique three generationmodel (Rusjanand Senjanovic 1988; Mohapatra 1988). Constructionsof this sort are in the realm of model building, and seem ratherfar from deductionsfrom some "fundamental theoryof everything".Nevertheless, there are a numberof beautifulaspect to the Calabi-Yauapproach. The numberof generationsof

109

fermions is determinedby the Euler numberof the internalmanifold. The role of the exponentialsof the gauge Higgs field is played by expectationvalues of path-ordered fields about non-contractible loops. Yukawacouplings are, in principle,calculable. (For a general review, see Green, Schwarzand Witten, 1987). It is fascinating to note that it appearsthat some of the Calbi-Yaumodels can be constructedin a radicallydifferentfashion, directly in four dimensions, by adding additional degrees of freedom to the string world sheet which can be used to define conformalfield theories. (See, for example, Gepner 1988/89). This raises questions about the meaning of the "dimensionality" space in a particularlyacute form. In the same context, it is of interestingto note that it is possible to embed 10-dimensionalsuperstringin 26 dimensional bosonic string theories (Casher,et al., 1985). At this point, it should be clear that there are some ratherintriguing questions of prinit. ciple associated with stringtheory as we currently understand In the first place, it is apparentlya theory of generalrelativity,in thatit has a dynamical graviton,yet we have to specify some backgroundmetric in orderto formulatethe theory. The origin of general covariance seems terriblyobscure, and the process of constructingthe theory seems rathercircular.Second, ratherthan having a "fundamental theoryof everything",we appearto have at least several thousandconsistent stringtheories, with ratherobscure interconnectionsbetween them. A serious studentof string theory can't help but suspect that there is some deeper theory lurkingbehindwhat we currentlyknow of string theory. This hidden theory presumablystandsin much the same relationto the stringperturbation theory that we know as a full quantumfield theory standsto its perturbation theory about some set of extremaof the classical action. Perhapsthe most pedestrianapproachto constructingsuch a non-perturbative string theory is string field theory. Here, the basic idea is to define a theory which standsin the same relation to strings as an interactingscalarfield theorydoes to particles. A variety of proposals have been made, of which the most successful thus far has been an open string field theory proposed by Witten (1986). Space does not permita detailed discussion of the theory,but I would like to comment on a few aspects of it. The theory is formulated in terms of a field defined on the infinite dimensionalconfigurationspace of the open string, the BRS operatorof the open string, and a definition of what it means to multiply two strings together. The theory looks very much like a gauge field theory defined on an infinite dimensional space, and bearsmany formal similaritiesto gauge theories in 2+1 dimensions in which the action is just a Chern-Simonsterm.The operatorsin the theory have been explicitly constructed(Gross and Jevicki 1987), and the theory apparently reproduceswhat is known of string theory when treatedperturbatively (Giddings, Martinecand Witten 1986). Further,it has an intriguingnon-perturbative structure,in which it appearsthat the dependanceon any particularbackgroundmetric can be thought of as the expansion of a more fundamentalaction about a particularsolution of the classical equationsof motion (Horowitz, et al. 1986). However, because of the essential nonof locality of the string interaction,the canonical structure the theory is not well-understood outside of the context of perturbation theory,and there are reasons to suspect that a correctcanonical quantizationmay not reproducethe known S-matrix(Eliezer and Woodard1989). Finally, I would like to comment briefly on some very recent developments.Witten has been pursuinga programof study of topological quantumfield theories;that is, quantum field theories in which the physical states are associated with a non-trivialtopological structure. Because there is no dependenceon the metric, there is no notion of locality. Particularlyinterestingexamples are providedby the 2+1 dimensional Cher-Simons theories I mentioned above (Witten 1988a). It is particularlyintriguingto note that the Einstein-Hilbertaction in 2+1 dimensions can be interpretedas such a theory: general covariance is apparentlyrealized in what Wittenterms an "unbroken" phase since the

110

expectationvalue of the metric vanishes and thereis no notion of locality (Witten 1988b). That these theories are relatedto stringtheoriescomes from the observationthat the be whole of currentalgebratheory can apparently reconstructedfrom such three dimensional theories. In conclusion, it is apparentthat a varietyof intriguingideas about the structureof space-time are being rapidlyinjected into theoreticalphysics. It would seem that these ideas should be of some interestto philosophersof science.

References Candelas,P., Horowitz, G., Strominger,A. and Witten,E. (1985), "Vacuum Nucl. Phys. B258, 46. Configurationsfor Superstrings", Casher,A., et al., "ConsistentSuperstringsas Solutions of the D=26 Bosonic String Theory",Phys. Lett. 162B, 121. Theory", D'Hoker, E. and Phong, D. H. (1988), "The Geometryof StringPerturbation Rev. Mod. Phys. 60, 917. Eliezer, D. A., and Woodard,R. P. (1989), "TheProblemof Nonlocality in String Theory",Brown HET-693. Gepner,D. (1988/89), "YukawaCouplingsfor Calabi-YauString Compactifications", Nucl. Phys. B311, 191. Giddings, S., Martinec,E. and Witten,E. (1986), Phys. Lett. 176B, 362. Goroff, M. and Sagnotti, A. (1986), "The UltravioletBehavior of Einstein Gravity", Nucl. Phys. B266, 709. Green, M. B., Schwarz, J. S., and Witten,E. (1987), SuperstringTheory. (Two volumes). New York: CambridgeUniversity Press. of Formulation InteractingStringField Gross, D. and Jevicki, A. (1987), "Operator Theory",Nucl. Phys. B283, 1. A. Horowitz, G., Lykken,J., Rohm, R. and Strominger, (1986), "A Purely Cubic Action for StringField Theory",Phys. Rev.Lett. 57, 283. Itzykson, C. and Zuber,J.-B. (1980), QuantumField Theory. New York:McGrawHill. P. Mohapatra, (1988), "Commentson Possible New Three GenerationCalabi Yau Manifolds",Phys. Lett. 214B, 199. of Rusjan,E. and Senjanovic,G. (1988), "HonestSymmetriesand Complex Structures the Three-Generation SuperstringModel",Phys. Lett. 214B, 193. to Scherk, J. (1975), "An Introduction the Theory of Dual Models and Strings",Rev. Mod. Phys. 47, 123. Witten,E. (1986), "Non-Commutative Geometryand StringField Theory",Nucl. Phys. B268, 253.

111

___ -___

Field Theory and the Jones Polynomial", to appearin _ . (1988a), "Quantum Comm.Math. Phys. ?. (1988b), "2+1 Dimensional Gravityas an Exactly Soluble System", to appear in Nucl. Phys. B.

Yau, S.-T. (1985), "CompactThree Dimensional KahlerManifolds with Zero Ricci in Curvature", Symposiumon Anomalies, Geometry,Topology, W. Bardeenand A. White (eds.). Singapore: WorldScientific, p. 395.

Вам также может понравиться