Classical Field Theory Pierre Deligne and Daniel S. Freed Introduction ‘A physical system is usually described in terms of states and observables. (See the discussion in (LFaddeev, §1.1).) In the Hamiltonian framework of classical mechanic the states form a symplectic manifold (JM, 2) end the observables are functions on M. The dynamics of a (time-invariant) system is described by a ‘one-parameter group of symplectic diffeomorphisms; the genersting function is the ‘energy ot hamiltonian, The system is sai tobe fre if (ML) is an affine symplectic space and the motion is by a one-parameter group of affine symplectic transforma- tons. This general description applies to systems which include classical particles, strings, felds, and other types of objects. Often the dynamics of the theory embedded in a larger symmetry group. For example, in relativistic old thoorios fone astumnes that (M, 0) carries a representation of the Poincaré group. Many'clasial systems admit a lagrangian description in which (M,0) is de- rived from a relatively simple expression, called the lagrangian density. One of the sain features of a lagrangian description is that the conserved quantity—called the Noether charye—corresponding to a symmetry is computed directly from the lagrangian. Furthermore, in field theories there is a local Nozther current which ntegrates to the global charge and which gives rise to local conservation laws. This reflects the physical fact that, for example, we can measure energy in any r ‘of space, not just the total energy over all of space. In this text we develop the basic ideas of classical lagrangian field theory. The examples we have in mind are the ones which arise in relativistic quantum feld theory as treated in other parts ‘of the book. One should be careful in trying to apply the formalism developed here to consirained systems (eg, nonholonomie constraints in classical mochanical systems, or constraints imposed on superfilds in the superspace descriptions of su- persymmetrc fields theories). In most of the exposition, we will assume that fields are arbitrary sections of some fixed fiber bundle B ever spacetime. Also, we develop the formalism in purely local way from the lagrangian density, hardly mentioning its integral, the action. If one is interested in calculus of variations problems, then the emphasis is different: the action i of primary interest and boundary conditions play a cruclal roe. A last warning is that not all classical field theories admit a la ‘grangian description. ‘Theres a free example whichis important in two-dimensional ‘conformal field theory: a fre chiral scalar field in two dimensions.138 P DELIGNE, D. FREED, CLASSICAL FIELD THEORY Jn Chapter 1 we review some classical mechanics. In a few standard examples ‘we describe the classical equations of motion and the construction of e symplectic structare onthe space of elasscal solutions. Noether'stheorem—the construction of fa conserved quantity from a one-parameter group of symmetries—is also discussed, We treat both nonrelativistic and relativistic examples, and in §1.4 we show bow to obtain nonrelativistic Galilean spacetime as a limit of relativistic Minkowski spacetime, The general theory of clasical fields is laid out in Chapter 2. A classical lngrangisn Bld theory’ consists of a spacetime M, a space of fields , and a lav srangian density L. We are mostly interested in’ Minkowski spacetime, but the theory is quite general and applies to spacetimes which are curved Lorentzian or Riemannian manifolds as well. The fields ate some sort of functions on M, more precisely sections of a Sber bundle B over M. The lagrangian density is a density ‘on MT for each point of F. It is assumed! to be of a local nature on M. More precisely, for some & the value of L(@) at a point m of M should depend only on the jet of ¢ at m. Usually k= 1. The lagrangian density E determines Buler- Lagrange equations DL = 0, also called equations of motion, which cut out the space of extremals MC F. It isa field, the tangent space of F at ¢ is the space ‘of sections of the vector bundle @*7(E/M) over Mf. The Euler-Lagrange equation [DL is a morphism of vector bundles from $"7(E/3M) to the bundle of densities on 1M, Its characteristic property is that for a deformation with compact support elu of #, one has [ae Zeta @ Seon] = ‘The formalism naturally takes place in the double complex of differential forms con F x M. This is qualified in two ways: () as we want L(8) to be a density, rather than a differential form of maximal degree, this double complex should be twisted by the orientation bundle of M; and (li) we want to consider only (p,g)-forms a Which are focal on IM: to a field @ and tangent vectors...) € d*7(E/M). te form a actacies & gofori al PsEin=.-y69) on M, aud for sone & the value of ((6:61,---§) at a point m of Mf should depend only on the ket at: m of 6 and the &. The cohomology of the double complax of local forms has been investigated by F. Tokeus?, Write the exterior differential as D = 6 + d, with 6 of degree (1.0) and d of degree (0,1). Takens* main result is Theorem 2.15: For p > 0 the complex (2d) of local dferenial forma is exact except in top degree. One can view fas an inductive limit, in &, of spaces of global sections of soft sheaves fon E, and this makes the exactness of (2,4) {p > 0, « # top) a local question on E. Let J*(E) be the buncle over M of Kets of sections of B. Takens also ‘observes that the associated simple complex Mis the inductive limit over k of the de Rham complexes of the J*(B). As the projections JM(E) — B ate fibrations, with fibers affine spaces, they induce isomorphisms in cohomology. It follows that the cohomology of B maps isomorphically to that of We include a proof of “Takens’ results in an appendix to Chapter 2 Tatehough we sett to lngrangiane which depend loely on eld, In quantum Beld theory coe mess efeciv lagangians which are oot loa. Some of th formal aepets carryover to wena ISgragian, but we have not talon the trouble ta ditingich them See References atthe ond cf the maniprropucTion 139 If the lagrangian density (8) depends only on the frst Jot of 4, it defines (Lym —I}form + (where n = dim’), which we call the variational J-form, characterized by the following properties: (i) the value of the (n ~ 1}-form (65) ‘at m MT depends only on the I-et of @ at m, and on the value of§ atm; and (i) DL La. ‘The variational (1,n — 1)-form + encodes the usual integration by parts arse tment which oceuts in computing Bulet-Lagrange equations. For more genera la- ‘rangians, the choice of a local (1,n — 1)-form + with DL = 51 + dy should be ‘considered part of the definition ofthe theory. Such a form » always exists and, by ‘Takens’ theorem, i is unique up to the addition of d@, for @ a local (1,n~2)-form. Classical mechanics corresponds to AM = R (time). In clasical mechanics, one is use to the following package: (i) the space of extremals M isa symplectic man- fod; (i) syrametris give symplectic automorphisms of (Mw), and infinitesimal symmetries & are given by generating functions Q, with dQ = ~i(€)w. A particular ‘ase is the one-parameter group of time translations, whose generating function is minus the hamiltonian. In (3), Q ie ambiguous up to an additive constant (and its existence can be obstructed by H*(M,R)). This ambiguity ean often be re- ‘moved by refining (i) t0: (i!) the space of extremals carries a canonial principal Rebundle 7 with connection V whore curvature is. Infinitesimal automorphisms of (M.T,1) can then be identified with pais (€,@) consisting of a vector field € fon M and a function Q such that dQ = «é)w. Infield theory one has a similar Hazniltonian interpretation if M i given as time > space, orf one as a suitable notion of spacelike hypersurface. A new problom is ‘to obtain counterparts local on M of those constructions. These local counterparts do not require a notion of space-ike hypersurface, o 8 apliting into time x space, ‘but rather exist for any spacotime M7. Tn particular, they make sense for Euclidean analogs, obtained by a Wick rotation {+ i. For the 2-form on M, which we now ‘denote ‘1, this is done in a paper of G. Zuckerman if w is the local (2,n ~ 1)- form by on F x Mf, then the 2-form on M is deduced from w by integration ‘on any spacelike hypersurface. Similarly, one ean expreso the generating functions of infinitesimal symmetries as integrals over space-like hypersurfaces of conserved currents, called Nocther currents. Let € be a vector field on F which is local: & at me M depends only on some jet of $ at m. Suppose itis a generalized symmetry fof L, in the sense that for some given (0,n ~ Ilocal form ag, one has Lig) L = dag ‘Thon the Noether current je = 6} ~ ag is conserved: dje(6) = 0 for all & in M,. The vector Beld£ is tangent to M, and the integral Qe of jg on a spacelike hypersurface is a corresponding generating function Tt can be more natural to consider symmetries which act on F and M simulta: neously. We will ny that symmetry is manifest fit preserves L and ‘exactly. Por example, time translation is usually a manifest symmetry when it acts on both land M, If we lot it act only on Belds, not on spacetime, itis only a generalized symmetry (with a= ~«(0,)L). 55ee Rteences10 P. DELIGNE, D, FREED, CLASSICAL FIELD THEORY ‘Theories whose field content includes a connection or metric posses an infinite dimensional group of local “gauge” symmetries. (For metrics the gauge symmetry {group isthe group of diffeomorphisms) We ean freeze the metzie or connection at fixed values gp, Ao; then local symmetries which fx gp or Ao act as global syzmme- tries in the theory of the remaining fields. The Noether current for these global symmetries may be computed by differentiating the total lagrangian with respect to the metric or connection, as explained in §28 and §2.9. ‘The derivative with respect to the metric is called the energy-momentum tensor ‘In §2.10 we discuss time-invariant feld configurations of finite energy on space- limes which are time x space, Among these we find classical vacua and solitons We also explain briefly perturbations around a classical vacuum and the Higgs rmechanisi, Chapter & summarizes the haste fre Iagrangian Geld theories om Minkowski spacetime. We treat scalar fields, spinor feds, and abelian gauge Selds (connee- tions). Our goa isto illustrate the general theory inthe simplest case and to record useful formulas, In clasical physics one of the main applications of field theory isto eletro- ‘magnetism. Chapter 4, which is a discussion of gauge theory in general, begins with a brief treatment of Maxwell's equations in the lagrangian framework. Some farnliarity with this material is necessary to understand the intuition behind more complicated models with gauge fields, which aze encountered in many lectures in these volumes, In §4.2 we review the basic geometry of connections in principal bundles, paying special attention to the universal connection (in terms of which ‘we write gauge theory lagrangians). ‘Then in {4.3 we write the lagrangian for ‘Yang-Mills theory and describe come additional "S-erms” which may appear in low dimensions. Pinaly, we define electric and magnetic charge in §4.4 and diseuse to relationship to global gauge transformations, ‘The general bosonic lagrangians without gravity usually include only scalar fields and gauge fields, We discuss a general lagrangian for these fields in Chapter 5; it includes many important bosonic theories as special cases. ‘There are special topological terms in lagrangians which are luvariant under (orientation-preserving) diffeomorphisms. Some, like the &-terms mentioned above, are related to primary topological invariants. More subtle are the ones associated to secondary invariants, like the Wess-Zumino-Witten term in a o-model or the CChert-Simons term in three-dimensional gauge theory. In Chapter 6 we briefly introduce the main examples and explain how the action acquires a more subtle geometric meaning. The geometric home for the lagrangian i a “T-caleuus” which extends the usual calculus of diflerental forma, as we indicate in §6.3 Finally, in Chapter 7 we discuss the “Wick rotation” of a lagrangian from Minkowski spacetime to Euclidean spacetime. For reference we collect the signs 1nd factors of =I which occur inthis analytic continustion, ‘To a large exient this text presents a preliminary version of this material; ‘we are not satisfied at all with our understanding in many places. Still we felt it important to include some mathematical framework for the computations in lagrangian feld theory. Our treatment is guided by what is neoded to follow the lectures and problems recorded in these volumes, and we hope at least t0 have provided sufficient background for that. Other mathematical accounts appest in the referencesnerropuerion a ‘The definition ofa classical lagrangian that we use (Definition 2:99) is adopted from a lecture of Joseph Bernstein. The basic formalism appears in the eforemer~ toned paper of Gregg Zuckerman. In prepating this text we also benefited from ‘many discussions with David Kazhdan, John Morgen, Nati Seiberg, and Ed Witten, among others.CHAPTER 1 Classical Mechanics ‘The equation of motion of classical mechanics is the Euler-Lagrange equation for extremizing the ection integral. From this extremal description we construct 2 canonical closed 2form w on the space JM of classical evolutions. Tt turns out that w is nondegenerate, and so M is a symplectic manifold. More precisely, we ‘construct on Ma canonical R-torsor* with connection whose curvature sw As ‘usual in symplectic grometry, to an infinitesimal symmetry we asociate a function svhich is called the Nocther charge, (It is often called the momentum or momentum map.) Tn quantum theory we apply exp(f-) to turn this R-torsor into a unitary line bundle with connection whose enrvatute is iu. §1.1. The nonrelativistic particle “We treat three cases: the free particle, a eystem of particles with potential, and the electromagnetic field Free Particle Let X = R” be Euclidean space with its standard inner product (-,). The evolution ‘of a classical fee parile of mass m is described by & map 2 from B (time) to X. The lagrengian densiy® isthe density on aay b= Fiat ae, where Ja]? is the inner product (dz/d,dz/dt). ‘The intogral of the lagrangian ew aa sefor from time f to time é I the action integral, or simply action. If we deform x we have SL = m2, 62) a ra 6) dt ~ af mz 62)} ia this content “Ror menos ‘prnspal Rebun ahold une the deat at im plce of the form A (1), bat i tis section we simply viet R ands ientify (ferme and densaaa P. DELIGNE, D, FREED, CLASSICAL FELD THEORY Here 6 is the differential on the space F of trajectries 2 of the particle, d is the differential on R, and the second minus sign® in the second line of (1.3) arises since Sand d anticommute on ¥ xR, Integrating we find iy sane f na [amy oy tot] eesti lt ath nm 8 on at on natin ste hcampe spat Go) Th dal nen an an 20 ‘hove solutions are uniform motion. ‘The boundary term leads one to consider, for ‘ech time f, the Lefora as) (0) = mlz{t), 62(0) on the space F ofall paths 2. On the subspace M of extremals (solutions to (1.7)) the setion $ is a function whose differential is a9) 55 = ltr) — alto) on M. It follows that the 2-form on M defined by rt w(t) += Salt) m(6z(¢)2(0) Js independent of t. It is even independent of ¢ for a specific reason, namely (1.9), with a compatibility among the reasons if thee times fo < ty < ty are considered: (ayy five [oe [oo ‘This can be rephrased as deBining an R-torsor with connection (7, V) on M whose curvature is for each Axed fo It the trivial Retorsor T(a) with the con- nection V(te) given by 7(fo)- By (1.8) adition of ~ fi L gives an isomorphism fom (Plo), Vo) to (Pes) Wla}), and by (1.1) these isomorphisms form a compatible system of isomorphisms. ‘The desired (7, ) isthe “common value {projective init) ofthe (7(0), 90). I we Bx fo, the map 2 + (a(t), #(to) from the space M of extremals to the tangent bundle TX of X is an isomorphism. If we use (1.8) to map the tangent ‘bundle to the cotangent bundle, chen (exp 2) i the pullback of the canonical [efor (fesp. canonteal 2form) on the cotangent bundle 7 move oval oot Sx denote a tangent vector oF -shich cave there it no minus sgn. We ‘raphe that our computations take place on FR. Por example, “in (01) te eaaton faap Fx RR and iit time derivate, "To convert (28) io the ore usual formula, het € Sen setae (eld) om Fan se os Genser x ‘Apnly the contraction (Q) to (9). Note thin comming G) pact din the lst term we pik ‘pe nina sign Thus we nd as Ast = i= mts, x a afte}CHAPTER 1. CLASSICAL MECHANICS sas ‘System of Classical Particles with Potential More generally, if we consider a system of classical particles with rigid constraints, the configuration space is a Riemannisn manifold X with Riemannian structure sgiven by twie the kizetic energy. Evolution is described by a map from (sime) to X. The lagrangian density is (12) L here the masses are included in the metric. Ifin addition we have an external field ‘of forces depending on a potential, oF interaction between the particles described by potentials, then the potentials are encoded by a real-valued function V on X, and the lagrangian density is cus be {Sur -ve)}ae ‘The free story can be sepeated with the following changes. ‘The Euler-Lagrange ‘equation (1.7) is now Newton's law ag) Veet grad ‘The L-form on the space of « is (1.8) fe) = (a(t), 62(0) and the symplectic 2form is (116) wit (B8(0) A 62(0). ‘The construction of the R-torsor is as before. ‘The identification of the space of solutions with the initial data at a fixed time depends on suitable completeness assumptions. Electromagnetic Field ‘An electromagnetic field ean be deseribed as an R-torsor with connection (P,V) on spacetime Rx X. For now choose a trivializaton of P (a "gauge"}, and 90 write V 5 & [form a, We separate the time and space componants by an a ‘Then V is the scalar potential and A is the vector potential, The evolution of single charged particle is described as before by a map: R— X, and fora particle fof mass m and charge q the lagrangian density is dt A. (1g) le?’ at ~ ex". ‘The action S = {it Lis the sum of «kinetic energy term and of ¢ times the parallel transport along 2 fram 2(lo) to 2(t,), computed in the choven gauge. In BL the boundary term is now 19) 4) = mia(t),dz(¢) - oA (6e().8 Py DELIGNE, D. FREED, CLASSICAL FIELD THEORY Fix fo then under suitable completeness assumptions the map + (z((o) (to)) denies the space §M of extremals withthe tangent bundle TX. Notice that the map fom the tangeat bundle to the cotangent bundle given by (1.18 the previous zap shifted by 94. IE we change the gauge, the space ofextremals doesnot change. This becomes cleat if we do not interpret fi L as a number, but rather 8s an isomorphism from gPig.xty)) © Gea) the Sut of a Kinetic eneray term and of parallel transport. “This isomorpinsm is manifestly gauge independent. On the space of cextremals we contitue (0 have an R-torsr with connection whose curvature is, independent of the gauge. For to fixed it is naturally ('o,2(t)) (@P) with ‘connection given by (4,2(¢a))"(@0)-+m2(e), 6(a)).Iaomorphie botweon thi Retossor with connection for diferent choices off ae given by he action parallel transport pls kinetic energy taro Jn this example we see that the action need not be a number, but rather ean be an ehment of an R-torsor. We discuss such topological terms further in Chapter 6 §1.2. The relativistic particle Let X be n-dimensional Minkowski spacetime, This is standard n-dimensional affine space with a Lorenta metric. Fix affine coordinates 23, 2%,..."~! so that the metric takes the form (1.20) lay? — (de? —---— (aa " ‘The corresponding bass of the underlying voctor space of transitions is called an ‘inertial frame. Here c is the speed of light. I s often convenient to set 2° = ‘The worldline of a relativistic particle is represented by a map ROX 2(7) = (U7),2(0),...,2 1) with (dz/dr,dz/dr) 2 0 and dt/dr > 0. The lagrangian density of a fee particle of ret mase rg is ay ee eel)" a For « physical particle d/dr lies in the positive light cone; in particular, we have ipo" as npngan's ara Se Ponca eg oye Fata as ty the gp DMB) eater ee An aon gel {fc oe pce Hagens ota! espace hypenrns He ttf, Onan fo nance aber Riberebon a et cn re korres Strona aie ei aged mesons des engin een Shoham eotant The cena rts con asi ef $i the atin ie the neta ned ty pera oe fo dtr then SE sum of two terms: (i) an integral J", where 2(7)) € He; and (il) boundary terms. They to Ht 1.2) ge HE S2(m) 0.23) ih] = ~moe STCHAPTER 1, CLASSICAL MECHANICS usr Note that this expeession is Dif(RJ-invariant. As previously, := 6H] is inde Jendent of Hy on the space of extremals and turns it into a syapletic manifold howe precisely, the space of extremals caries an R-torsor T with connection whose aMivatase isa For esch choice of spacetke hypersurface Hit can be identified ith the trivial Retorsor, with connection given by fff]. Action integrals give Transitive system of isomorphisms between the descriptions of (TV) given by fiferent H. "To go to the nonrelativistic limit we proceed as follows. We work in our chosen coordinate aystem and consider velocities which are smal relative to the speed of Tight ¢- If we take for coordinate 7 the time f, then the path is (129 2 = (420.120), ‘The lagrangian is then amy (T= FTE at (125) { where 24 Linge? +O(25) rod +2? +008 )} a ‘dat \* a2) Z(H) is he winity squared. The cond tn of (1.28) shows ha th nonrltiiie rae Joost nthe knee egy tog mins the potenti ergy mc! of reese test. The eater can a he hat the onrlabiten(1.23) eRe iGparrace (t= cna (18) (orm — m9) Te cvltviiceting toe not maken ontrode ii ont or aspen afte, an to node oti On the oer ad bakgroune {1SeiShaguae dcr be nteaded om Minkowk spcetine angry tothe §1.3. Noether’s theorem ‘We have seen a aumber of examples of how, because equations of motion are Elen: Lagrange equations, the space of classical evolutions M catries a symplectic struc. tune w. More precisely, it carries « eanonical R-torsor T with connection V whose Curvature is wt Any automorphism of the data used to define the extremality com. Gition induce: an sutomorphism of (!M,T,). The same holds for infinitesimal fsutomorphiema. Infinitesimal antomorphisms of (M,7,) can be identified with functions Q on M. Namely, if a vector field & on T is such an infinitesimal av tomorphistn let € denote the projection to M and & the horizontal Lift of € to T. ‘Then 27) é ‘where ¢ is the infinitesimal action of R which defines the torsor structure, The infinitesimal automorphism é of T respects the connection if and only if (228) 4Q = -e(6)u.vas P. DBLIGNE, D. FREED, CLASSICAL FIELD THEORY 1 the curvature wis nondegenerate, ie, turns M into a symplectic manifold, then (1.2) shows that vector field ¢ is determined by the generating function Q. For functions Q;,03 corresponding to vector elds €o,,€a,, me compute the vector fed which corresponds to the Poisson bracket (Qy,@:} 10 be (1.29) £a..00) (Note: Brackets of infinitesimal automorphisms are given by the opposite of Polson brackets, inthe same way thatthe Lie algebra of diffeomorphisms isthe set of vector fields with bracket opposite te the usual brackets of vector fields.) If we have a trvialization of (T,) which is preserved by an infinitesimal ue tomorphism then Ika. 60a (2.89) Q= Hern, where is the connection fore on M induced from the trivialization. Ifthe triv- {alization changes, then there is an additional term in (1.90). In §2 we develop a more systematic formalism in which to make computations, s in thie section when the trivilization changes we simply report the rest We apply these ideas to the free particle in both the nonrelativistic and rel ativstic settings. Consider fist a free nonrelativistic particle of mass mo moving in RY, Let 2!,....2” be standard coordinates on R” and &,, the standard metric. ‘The isometries of R* induce automorphisms of (M,T,) which Sx the trivialize. tion ata fixed ttn fg. Thus we can use (1.30) in conjunction with (1.8) to compute the charges. ‘The charge corresponding to the infinitesimal translation € — 6/22" is the linear momentum (si) = mobi sé! ‘The charge corresponding to a1 infinitesimal rotation € = 29/02 ~2%8/x"is the ‘angular momentum (132) 1M; = mobj( at ~ 234) Now the hamiltonian or energy Bis the charge associated to minus time translation in the domain R, and this does not preserve the trivialization of (T,V) induced at a fixed time. Rather, the derivative ofthe trivialization at to is Loto), where we write the lagrangian density at Lodt. (The change in trivialization comes because infitesimal time translation aly preserves the lagrangisn up to a total derivative.) Accounting for this we compute the nontelativstie energy (sve Example 2.105). nals (1.33) B= bale ‘There isan additional symmetry called a Galilean boost. We write the infinitesimal version a8 time-varying vector field € = 19/0x". Again there is an additional term in the formula for the charge Nj since the lagrangian is only preserved up to total derivative (34) My mobiy(ti? — 2!)(CHAPTER 1. CLASSICAL MECHANICS 140 ‘This is simply minus the intial postion times the mass. ‘Consider now a relativistic particle, The lagrangian (1.22) is invariant under the isometries of X, the Poincaré group, and we would like to compute the corre- sponding charges. Fix a splitting of Minkowski spacetime into time x space and. use the hypersurface ¢ = to to trivialize (T,V). We perametrize the worldline as in (1.24). The subgroup of Poincaré which fixes this hypersurface also fixes this trivialization and we use (1.30) and (1.28) o compute the associated charges. Thus the momentum p, isthe charge corresponding to € = 2/8r", which we compute to te - Fae = ano) Here we have introduced the relativistic mass mo (1.36) mo) = etre [Note that the nonrelativistic limit v/e ~+ 0 ofthe relativistic mass m(v) is Ube rest ‘mass mp. Similarly, an infinitesimal spatial rotation § = 2'0/Ox! ~ 278/8x" has charge the angular momentum (1.35) han ys = adataizt— 2429) ca 7 A= Ie Now the change corresponding to € = ~2/0t ithe relativistic eneroy B. As inthe nonrelativistic case the trivialization changes according to the lagrangian (1.25) (divided by the standard density dé), and go we compute Binstein’s famous formula (oS (ate* — 22") (1.38) Wo) Note that (B,~p:) transforms in the dusl to the standard representation of the Lorentz group. In n= 4 dimensions this quantity is called the 4-momentum. Finally, the infinitesimal Lorentz boost (1.39) has corresponding charge mods (ti? — x7) _ _ (1.40) Mea OTe = m(w)8is(te? — 2°), Here again the symmetry does not preserve the trivalization, so there is an extra ‘term to compte. ‘The enarge (18 simply minus the mtial position of the particle ‘times the relativistic mass, “The nontelaivistic limits of (1.95), (1.37), (1.38), (140) give (181), (182), (1.33), (134). (Brom B in (1.88) we must subtract the rest energy moc Wo obtain the nonrelativistic energy (.3).)180 P. DELIGNE, D. FREED, CLASSICAL FIELD THEORY §1.4. Synthesis Let Mz denote n-dimensional Minkowski spacetime with metric g, given in (1.20) Let, Mog denote the limiting space as ¢ + oo. Of course, the affine spaces which underlie Mz and Moo are identical. Let W be the vector space of translations of Mec. The limit Jim g=' = gz! of the inverse metric is a degenerate quadratic form on W*, It has e one-dimensional kernel $* C W* whose aunihilator $c W’ determines codimension one foliation of Mas by alfine subspaces. Then gc on W/S" is inverse to a negative definite mesric on $. We think of this foliation (with the metric on the leaves) as defining the simultaneous spatial events in Mee We also fix a scale di © S® for time; then there is « one-dimensional space of distinguished afin time functions which differ by @ constant. The group Goo of affine transformations of Mze which preserve gz! and dt is the Galilean group. Let Ge denote the group of affine transformations of M, which preserve ge; the double cover ofits identity component isthe Poincaré group. Note that Ge, uy act on the ‘same affine space, so there is well-defined sense in which G+ Guo a8 €— 00. ‘The translations sit inside G, and Ga. In Gay the spatial translations lie in 4 subgroup # of dimension 2(n ~ 1) generated by vector Relds (3 + 0, where ¢ is any afine time function and 8,8 are spatial translations. Vector fields 10 generate affine transformations called Galilean boosts; they are the e — 00 limit of Lorent2 Doosts (1.39). ‘There isa nontrivial central extension of Gus which restricts nontrivilly on H. For a geometric picture we begin with a fixed (n-}1}-dimensional Minkowski space time NV with vector space of translations V, Far each spacelike vector v € V we consider the subgroup G, of the Poinearé group which fixes v, Tt acts on the quo- tient afine space My = N/R. v. This quotient space M, inherits a Minkowski metric we describe in two equivalent ways: it is the metric induced from the or- thogonal space (R-u)+ ¢ V, or equivalently its inverse is the subspace tnetric on the annihilator (V/R v}* = (R-v)® CV". The eubgroup Gy isthe trivial central ‘extension ofthe Poincaré group of Mf, by translations along v. Now consider ¢ € V a lightlike vector. Set My = N/R, and let Ge be the subgroup of the Poincaré group of NV which fixed €. Now the inverse metric on V* restricts to a degenerate form on (V/R- 6)" (R-€)° CV". Also, we take the functional (,-) to be the dé above. The group of afine transformations of Afe which preserves the degenerate form and Cis the Galilean group. Thus Ge is the central extension of this group. by translations along v. Note that the subspace of (V/R: £)° & (R-€)° previously called S* is here (€4)", the annihilator ofthe orthogonal subspace to &. (Since & is lightlke, we have R-€C C*, which is why it makes sense to define dt ~ (€,»,) Also, Sis simply C2 /R For an explicit description introduce coordinates ¢*,1>,21,...2°-! on N so that the metric is? aay) de de — (da = (da eed ol ight nese, 69 we tt Yo one.(CHAPTER 1, CLASSICAL MECHANICS 151 ‘Take ¢ = 9, = a/2¢*. ‘Then (E4)° is the span of dt~, and she possible time functions are (at~ +b: a,b € R), The Lie algebra of Gis spanted by (central element) oy (transations) (spatial rotation) (Galilean boosts ‘The nontrivial brake 0.0) 12,83) = 626 reflects the nontrivial central extension. Note from (1.31) and (1.34) that for the free towelativisti partele the Pesson brackr ofthe Noether charges corresponding (0 8, and B, (1s) tras Comparing (1.43) and (1.48) we see that the conserved quantity corresponding to 2, in a theory should be identified with the mass. Thus in the Galilean theory there is an additional conserved quantity over the relativistic theory—the mass. (The bracket of elenents in the centrally extended Galilean algebra and the Pois- son bracket of the corresponding Noether charges are opposite. This is the usual situation for lft group ations on symplectic manifolds: see §5 of [I-Signs}.) Quite generally, suppase we have a classical theory whose slate space (Mts) carries a symplectic action of either the Galilean group or the Poincaré group. Ifinitesimally we have an entihomomorphism from the Lie algebra g of that group ‘nto the Lie algebra of vectar felds on Ad, We assume that each veotor field € in the image satisfies ~1(¢)w is exact, ic, isthe symplectic gradient of some function. For a general group G, the existence and uniqueness of a life g > C°°(M) wich Js an antihornomorphism of Lie algebras is measured by #"(g) and H"(g). For the Poincaré group there is a unique lift; for the Galilean group there isa lft of the central extension we constructed above, unique up to a shift of the total energy. (OF spacetime has cimension $, then we can also shift angular momentum.) The central element then maps to a locally constant function which isthe total mass of the system. In the Poincaré case the vector space V of translations maps toa vector space of functions which—after choosing an inertial frame—is (~E,p1,..-,Pa—1)s ‘where B is energy and p, are linear momenta. Under Lorenta transformations this transforms as the éoeficients of an element in® V", and out of we construct an invariant using the metric a5) EX/e? ~ |p? = mie. ‘This defines the “rest mass" mo ofthe system. Tt is « Poincaré invariant, though in general not locally constant, function on M. In the Galilean ease there is a Galilean invariant codimension one subspace $C V of translations in spatial direction, and ron atthe veto tal 10 (E,~p) (BYP.asa P. DELIGNE, D, FREED, CLASSICAL FIELD THEORY S carvies a metric. Ifthe Galilean algebra, and not a central extension, were to Jif to functions, then we would conclude that the norm square ipl? of the total ‘momentum is Galilean-invariant, But clearly pj? typically changes under boosts, and this shows why boasts and spatial translations have a nonzero commutator it the central extension ‘We can formulate the theory ofthe free nonrelativistic particle ky considering its worldline in the spice May. ‘The difference from the relativitie situation is that the time functions provide distinguished parametrizations of the worldline vp to translation, This allows us to “couple” the free nonrelativistic particle to fn arbitrary potential faction V, as in (1.13); the integrated lagrangian (action) does not change if we t'anslate time and so ie Galilean invariant. By contrast, as stated earlier we cannot introduce a Poincaré invariant coupling of the relativistic Particle to a potential function. Rather, we can couple it to fields, specifically to fan electromagnetic fetd (abelian connection) and to a gravitational feld (variable Lorentz metric). In thenext chapter we take up the general theory of fieldsCHAPTER 2 Lagrangian Theory of Classical Fields §2.1, Dimensional analysis ‘Tis sien wool to follow the advive we give to beginning students: Check units! ‘Every physical quantity has unite attached to it. The basic units are units of mass length, and time, ‘The number measuring « physical quantity Q depends ‘on the choice of units, and Q i said to be of dimension M*L°TS if the aumber q measuring Q is multiplied by 1~*,"*y~* when the units are multiplied by A, Notation: [@] = M°LYT*, For example, the action integral $ of (1.2), where the inegral is taken between prescribed instants, has the dimension of an actin: en (s)=auer- “The dimension of a p-form is defined to be the dimension of its value on s fixed -pevector, The lagrangian density (1-1) (a I-form on the time line), the variational ‘Lforn 7 of (1.8) (a Iform on the space of trajectories), and the closed 2-form ws all have the dimension of an action ‘The dimension of the conserved quantity Q corresponding to an infinitesimal Gabiean transformation 7 is given by @2 [} = factonlin In the relativistic setting we will tually impose ¢ = 1, s0 that a unit of length gives ‘one-time. Thon faction) becomes ML, and the conserved quantities corresponding, to infinitesimal generators of the Poincaré group once again have 3) {Q) = fection) ‘We can consider electric charge to have units C and take the constant qin (1.18) to hive [q] = C; then the I-form «in (1.17) which represents the electromagnetic field has units of action divided by charge: 4) lo =aeetrtc~ ‘Altematively, we can work in a aystem of units in which charge is expressed in terms of mass, length, and time by es) C= MT.1s P. DELIGNE, D, FREED, CLASSICAL FIELD THEORY ‘This comes from declaring the constant k in Coulomb's law *F = kgyaa/r®” to be dimensionless ‘Universal physical constants allow us to convert units. In relativistic theories the speed of light (20) aur identifies time with length, In quattum theories Planck's constant en W=Mer has units of action, which eliminates one of M,L,T- In theories of gravity the [Newton constant G in the Newton fore law *F = Grmyma/r® has units (28) [=2Mor, Which again allows us to eliminate one of M,L,. Ina relativistic quantum fekd Uteuty we cat uae 6/2 to express all uiits i tris of mass, and 20 ench physical quantity has a mass dimension. If the theory includes gravity (eg string theory), then we can use oh, G to express everything in terms of dimensionless quantities. §2.2, Densities and twisted differential forms Tn this section M i an ordinary (not super) manifold TI M is oriented of dimenton n, then n-forms with compact support can be integrated on M. Changing the orientation mute the integral by —L This leads to the consideration of densities, defined to be sections of ea) Dens M := A"T'M @ 04, where oy isthe orientation line Bundle. On any Af, oriented (and orientable) ot tot, the Integral of density ws with compact support is unambiguously defined: ‘one writes asa sum of densities, with suppor in orientable local charts U, one orients wach U, and doing s0 ientfie with an nform on Uy and one defines Ju Su, More generally, we wll have to consider the complex of tuisted forms, the tensor product of iy withthe local system oy. We teat eu a8 being in cobomologial degree —n, and define (210) Ng? a 957? oom ‘A compectly supported element of £7 canbe integrated on a normally oriented submanifold of M of codimension p. Elements of 057 can be viewed as sections oF en) ATM @ DeneM; the isomorphism with the description (2.10) comes by contracting a p-vector field wrth a density. Tn the sequel we often sy simply ‘orm’ for ‘twisted for In the case of supermafels, densities ae related not to sn exterior power of the cotangent bundle, but to its Berernian, and (2.11) defines the eomponeats of the comple of integral denies. Se [-Superaymmetry, 339-12 for deta (On a Riemannian manifold M the Hodge s-operetor Is an isomorphism (222) #05, — 9591 ‘Zhen fora pps quantity @ inated ef wxitng (Q}= AI we usally ite(CHAPTER 2 LAGRANGIAN THEORY OF CLASSICAL FIELDS 138 §2.3. Fields and lagrangians Let Mf be a smooth manifold of dimension . We formulate field theory of MG. In the standard physical extup AM = M” is affine Minkowsli spacetime. For n = 1 this i 1", which is (afine) time; i is the appropriate “spacetime” MM for classical mechanics. In usual exemples of feld theory we cen analytically continue to imagi- nary time and s0 obtain « field theory on Euclidean space. Often field theories ean also be formulated for curved metrcs—of Lorentz or Euclidean signature. We also allow the spacetime M to be a upermanifold, As we cautioned inthe introduction, four framework is not adequate for many types of constrained mechanical systems Fields on Af ate (emooth) sections of a given fiber bundle E> M. Let F denote the space of all sections. For example, the besic field in a o-moadet is ‘a map @: M — X for some auxiliary manifold X. In this ease we simply have B= MX. We can ao study a twisted version in which H is not a product. An Important case is when E is a vector bundle. The basic field in @ gauge theory is ‘connection A with guage group some specified Lie group G. If we fx a principal Gbundie P+ M, then A is a section of a certain associated bundle of affine spaces, For many’ purpests itis best not to fix P and rather to view the collection ofall connections as a category. However, in this section we simply view P as fixe. ‘There is an evaluation map Qs) a FxM—E, Agein we allow £—andzo also F—to be a supermanifold, even if Mis an ordinary manifold ‘Of course, afield theory typically contains several fields 4, and correspondingly B= XE, isafibor product. Roughly speaking, each Ey decompases into an intrinsic part times an extrinsic part. The intrinsic partis associated to the principal frame bundle of Mf via a representation of GL (or Spin), and the representation determines the type of field. Thus sclar field is asaciated to the trivial representation. The basic field gM — X in a o-model isa typical example. A scalar field may take values in a nonlinear epsce, but the extzinsic values of other types of felds are linear A p-form field is a p-form on Mit may take values in some vector bundle over M. Physicists often use the word “vector field’ to refer to a I-form field, They also use “vector field’ or ‘gauge field’ to refer to a connection, which isa type of field which was discussed above. There are also “connectionlike” versions of p-forms (se $6.3) ‘but only for abelian structure groups. If M is a spin manifold then we can also consider spinor fields, which are sections of a spin bundle possibly tensored ith another extrinsic vector bundle. The procise choice of spinor bundle varies with the example. Theories of gravity also include a metric on X, sometimes called @ ‘gravitational field’. Ir theories of supergravity there is also » Rarita-Schwinger field, which Is «section of an irreducible subbundle of $@T"M, where S is a spin bundle “The spin of a field depends on the representation of Spin (or GL) which defines it, Namely, we decompose the compleified representation under the Spin(2) sub- {roup which double covers the group of rotations in some 2-plane, We obtain a sum ‘of one-dimensional representations which we label by ll integers 0, +1/2, +1, Replace these numbers by their absolute values; then the largest number which ‘cers is the spin ofthe field. Thus a scalar field has spin 0, a spinor field spin 1/2, fa p-lorm field or connection spin 1, a Rarite-Schwinger feld spin 3/2, and a met spin 2158, P. DELIGNE, D, FREED, CLASSICAL FIELD THEORY ‘There are pysical veagons to vestict to these values ofthe spin in quantum field theories (see [Il-Dynamics of QF, §2.4). Tn unitary quantum fell theories thetef connection between spin and sla- tistics, the statistics being whether the fd is even of od (in the sense of super: geometry): Held of iategral spin ate even and feds of hllntegral spin are odd ‘The sual physical terminology fs that even elds are bosons ad oad Belds are fermions. This spintatsics connection Is violated in some nomutitaty topologe cal eld theories In sone theories there are local gauge symmettas (ese Definition 2.98) which sot onthe felds. This cursor elds whith are pforms (p > 1) and fr connections in gnuge theory. (A p-form avis gauge equivalent to.a-+d3 for 6 any (p~1)form) “The gauge ayinmetie of a connection areas usual. Local symmetries are sections cof bundle of groups over M Lat F dentte the quotient ofthe space of Belds F hy the action of local symmetries. In ovr netation we treat F as n manifold, though this may not be true and we may seed to work equivariantly on some space which projects onto F. (For example in gauge theory we often Bx a basepoint on each ‘Component of M. ‘Then the group of geuge transformations which equal the identity 1 the basepoints acts fea on the space of connoctions, and the finite dimensional {r0up of sutomorphism atthe baspoints acts on the quotient) ‘Our formulation is not meant to inclide theories of gravity, where one ofthe fields isa mecrc and the group of dffeororphims of M act as aloes symmetry. We wil, however, tse the metric as « bacground field when we discus the energy- momentum tensor. ‘Then in some cases leg, with spinor feds) the fiber bundle E — M vaties with the metric. ‘The appropriate modification of this setup i deserted later when we discuss the energy-inomentum tensor ‘We will woe a complex which is, basally, the de Rham complex of F x M. Reflecting the product structure of F M, iti a double complex. Suppose frst. that Mis oriented, of dimension n. ‘The complex "11 we wil use is then the de Ram complex shifted by niin other words, 1"*1(F x M) ithe space of pforms ‘on F with values in the epace of (n ~ ghforms on M. We let 6 be the exterior derivative of F,d the exterior derivative” of forms on M, and D = 6 +d the total ‘xtorior derivative. For general M we use twisted (n-~q)forms: instead ofthe de Rhain complex of M, we uso 9) = ty @ ou Mote televant is the subcomplex 2!" of local elements of 1+"), where a € {2-4 isa to be local if for some k a any 4 € F the value ofthe twisted (n—q)- form a(dsé,,..-s§) at m € M depends only on the bet at m of §,6),....é9. Here sare tangents to F at 4. IF i the space of sections of a bundle E over M, aud if w*): JFE —+ M is the bundle of k-jets of sections of m: B+ M, such an a is & global section over J*E of 8, 5,4, OF" "My, Note that k cannot be kept fixed, a5 d maps Wy pjyy O20 NG" 0 Mrrergyy ORE DTT ‘As mentioned inthe Introduction, the double complex gated by Takens, who proved the fllowirg has been investi= Jur dian convention is that for a € QM(F) and Fe AIPM), am Kans) = (-aPanas.CHAPTER 2, LAGRANGIAN THEORY OF CLASSICAL FIELDS ast ‘hoorem 2.16 (Takens). For 9 > 0 the complex (O5I4(F x M)vd) of lot differential forms is exact except in the top degree |e “This theorem plays a crucial role in the discussion of generalized symmetries (92.6), and also plays a part in the general discusion of this section. We also make use of the following generalization, proved by the sane method. Theorem 2.16 (Takens). Let Vi + B, i = 1,...,p, be vector bundles, where tue require p 2 1. Let V = xeVj be the fiber product. For} € F a section of E, let Vp te the space of sections of $*V = M. Consider in 2!*(Y, x M) the subcomples of forms a($;Ciy-.-y 9) which are Remulttinear in G- Then this subcomples is tact except in the Lop degre |el = 0. In the appendix to Chapter 2 we give a proof of these theorems. ‘We wil ue the following picture to depict elements inthe double complex u We remark that certain topological (terms in) lagrangans do not Bt into this formalism, We discuss the necessary modifications in Chapter 6. "The basic ingredient ina classical Med theory is lagrangian (density) a7) Le nQr xan: it has unite of ation: as) ‘The notation in (2.16) indicates that for each field ¢ in F we have adensity L(4) ‘on M, while @16) indicates that the integral of L(g) on a fixed compact region ‘of M has the dimension of an action. Typically, £(@) i «local funetional of and. for most fundamental lagrangians, the value of L(d) ata point m depends only or the Let of g at m (gravity is an exception}. ‘Typically, the integral J, L(@) is divergent. Even if, as often, one can restsict one’s attention to fields decaying rapidly at spatial infinity, so that the integral be- teen two spacelike hypersurfaces converges, the integration over time will diverge Computations given below show that otherwise the expected symplectic structure ‘on the space cf extremals would vanish. However, for a deformation with compact support of @, and if we assume Z local, then 6£(¢) is a density with compact sup- port, which ean be integrated. One says that dp is extremal if f6L = 0 for any Such deformation of do, We now make this more precise. One considers families of fields fu, with of0] = do, and with du) independent of w in the complement158 >. DELIGNE, D. PREBD, CLASSICAL FIELD THEORY ‘of a compact repion of M. The density ££ L(dfu)|,_. then has compact support. ‘The extremality condition is that its integral shold vanish, for any glu). Ie is suggestive, if abusive, to write the condition fy L(6fu) = 0 ‘Let us suppose that the fields ¢ are sections of a bundle E over M. For a deformation glu] of do, the density L(dfu)) at w = 0 will depend only on the feld of vertical vectors £¢ on B, along dy, which is a section € of #57(E/M). The Retinear fore (219) ef aw on the sections with compact support of é§7(E/) can be uniquely written as (220) ez) for DL a (@3T(E/M))*-valued density. The uniity of DL. is cleas, and its ex tence isa local question. It is proved by the usual integration by parts argument leading to Buler Lagrange equations. Bxtremality of gy means that DL = 0 at gy (Buler-Lagrenge equations). We note that DL is in 1", with a value at m€ Mf depending only on some jet of @ atm, and on the value at m of 6 ‘This formalism does not hold if, as Is typical of superspace formulations of supersymmetric theories, the fields are sections of a bundle E subjected to con- straints. The problem is that € is no longer arbitrary in @§T(E/M). Let M be the space of extremals, We now explain that if spacetime Mis time x space, or ab least if there isa suitable notion of spece-like hypersurface, then the construction fon M of a closed 2formm—on the model of Chapter 1—uses only conditions on L whiel hold for such constrained superspace formulations. The crucial condition is that the constraints allow é to be deformed independently in disjoint regions of M. (Formally: the sheaf of fields, a subsheaf of the sheaf of C*.sections of B/M, is soft). In practice, for superspace formulations, this means that superfelds can be ‘described in components. Let Hy and Hz be spacectke hypersurfaces with, to simplify the picture, Hy before Ha. In superspace formulations, Hy and Hz should be codimension 1/0 submanifolds of M. We consider em [6 which we at frst assume to converge. If extremal, and if 6] deforms 4 (elo 60, shen en) EP vot will in general not vanish at u = 0, except ifthe support of the deformation is be- tween H, and Hs. For deformations with compact suppor, it will be the difference ‘of boundary terms attached to Hf; and Hy a jh (228) Bf, Mon =r -r(CHAPTER 2, LAGRANGIAN THEORY OF CLASSICAL FIELDS 180 In this formula, Fis a Horm on F, defined only on M, whose value at a tangent vector & at an extremal é depends only on some jet of along Hj. We now restrict ‘those Iforms to (the tangent bunele of) JM. Equation (2.23) means that, with 5 the de Rham diferential on M1, (228) sft aT 0 that the exterior derivative 6D, is independent of i. One defines the canon closed 2-form ou M by (225) amo, Remark 2.26. If the integral of L(¢) on the whole of M did make sense, this construction would collapee: T's(@) would be 6 of the integration of L(@) from the Infinite past to Ha, giving 2 = OF: = 0 Remark 227. Instead of assuming {to exist, one may consider oly Sas Mich cooede vith axed extremal de spatial abut. The given contruction thon dfn, for tangent vector 9'M whose support terete region of Spacetime in bees any fo space ike hyperrace ino compoct st Remark 2.28. As in Chapter 1, the construction gives an R-torsor T with con- nection V on Mf, whose curvature is 1. The choice ofa space-like hypersurface 1 trivilizes 7, ard the trvializations given by Hy and Ha dlfer by fy L(9). Remark 2.29. Adding to the lagrangian density an exact term da, for in Qe does not change the space of extremels. For H a spacelike hypersurface, it changes the corresponding I-form I’ on M by ax) gf doteny ot eno ‘The torsor with connection (7,7) for L, and the one (T°, ¥') for L+ da, can be identified, by sending the rivialization Oy of T corresponding to a spacelike Inypersurface H1to the tivialization Oj, ~ Jy a of T*- In Chapter 1 (particle im an electromagnetic fc), we saw one example where a change of gauge changes L(6) by an exact tem, and where the torsor T of Remark 2.28 is defined, but where & teiviazation of T is given by the choice of H and of a gauge along H Remark 2.81, We now give a variant ofthe dofiition of Q, to make cleaver on ‘what it dopends. Let gle) be a 2parameter family of extromals. At dy = 6[0,0} in M, we have the two tangent vectors & = Z¢lu,2] at (0,0) and & = Fou] _at (0,0). Whatis (é,, G2)? Let be a space-like hypersurface, It has a past and ‘future side, and we identity « neighborhood of H with x (—1,1), with h € ‘corresponding 0 (hy 0) and the future side of H corresponding to Hx (0,1). Let Giltse] bea family of fields, deforming fu 0),egrecing with oe] in the Future, fnd with glu in the past: for some a, & with -1<0 6 duluse] = au), at (1,0) for to.10 P.DBLIGNE, D, FREED, CLASSICAL FIELD THEORY et dal] be a sins fry, withthe cles of wand v interchanged. Then 23) 6.8) = [0.0.1661)- 8,0,U60) at Cw) The integrand vanshes for ¢ > b (where L(61) = L(ga)) and for t < a (where BL(Gn} = BuL(da} = 0). IF dure] = dy at (yt) fr h ouside a compact of H fone can take p= ¢1 = ds outside of such a compact, and the integrand has then compact suppor. ‘We now relate the definition of to the previous one. We fist consider one-parameter family glu of extremal, with ff] = go. Let ¢i[a] agree with fu) fort b, and with o for ' 1) are exact. The -lasscal moduli space is then the space of constan: values ofthe scalar fields. Now the energy density O(gq) of a constant scalar ficld go: Mf — X satisfies (2.200) tt.» © (60) = V (0) lat a=" foc a potential energy function V: X —+R. Assume’ that V > 0. The only Enite energy constant scalar fields have V(x) = 0, and so em) Mune = V0) isthe catsical modi apace. If there is also connection il in the theory with fuuze group G, then G acts on X and Vi an invariant f\netin. In that case the “ft of dividing by glebal gotige transformations i that the clsscal moduli space isthe quotient 29) Moye = VMONG. Returning fora moment to. general field theory £ = L++yen a manifold M, fo afd configuration dy € F. Then there is a perturbation theory for the fuctuations farcend dy in which the space of fields io Ty, and the N™ order perturbative Tagrangian isthe N order jt of Lat do. ‘On Minkowski spacetime M we often perturb around a vacvum solution." Re- calla at ava al gage BeldeAy are trivial and we can wo thom to trivialize fll bundles. Then any scalar yi « constant in a manifold X. In the perturbation theary the fuctuations of « tevial connection Ap on a prinoal G-bundle P le in Mad P) & My (9), where we use the erivializtion. The Muccuations ¢ ofthe constant scala gy le in TyyX_ Spinoe felds, RaritaSchwinger field, and pform fel all vanish at a vacuum configuration and we consider the feds in the orginal lagangian to be fluctuations about zero, Now since we areata vacuum solution the pertrbative lagrangian starts out with quadeatic terms. In the quadratic ap- provimation 3 is typically & fee scalar field in the Bucidean space TjgX with tras Hear, V; there may be higher derivative terms as wel. There fs mass tmairic forthe spinor feds which also depends on gy. The massless uctuations ofthe gouge Beld le in (gy), where pgp is the Lie algchea of the stabilizer subgroup Gyy at do ofthe action on X. Often Gg, is called she wnbroten gauge frotp, though this terminology is confusing. Other componentsof the gauge Ruta Sica are masive due tothe Higs mechani, which we lua inthe following example "Wfo may probs the enrey density nonnegative for al lds: @,(@) 0 for ald, where eam An 9(d) = Cx o)|a\er~'2 er 3Pfie ld and thew fae’ ae acated in 58 bs the fest stop in m perturbative construction of qunatom they aroued the chosen186 P. DBLIGNE, D, FREED, CLASSICAL FIELD THEORY Example 2.204. We work on n-dimensional Minkowski spacetime M". The fields ace e complex scalar field ga real spinor field? py and aT xT connection A ‘As in any gauge theory, we work on the quotient ofthe space offs by gauge transformations. Let TT act on € by (21,44): = 2A; and suppose that ¢ is ‘ szction of the associated hermitian Iie bundle. The spinor fed Vis not coupled toa. Let (2.205) b= (Ilda? — H1Fal? + F(0PW) — WIC - [aIPy? = 660) lena "The potential energy is (2.206) V(¢o) = Heol? — Iléoll?)? ‘and V~1(0) consists of the origin and the unit circle. The clasieal moduli space is the quotient of V~(0) by T x T and so consists of two isolated points. They fre the two clasical vacua. At the origi the perturbative lagrangian is the same ‘as (2.206), except that A should now be viewed as perturbation A= Ap +aof a trivial connection; a isan (2R #iR)-valued I-form on M. We use Ap to construct 4 trivalizingseetion of the hermaitian Tine bundle, and 50 view ¢ as a map Mf —» C. “Then the perturbative lagrangian is (2.207) Lo = (ap-+a- $1? — Fda? + 0) — 19IP (= H6IPY 49) a2 From the quadratic part of Lo we read off that ¢ is massive (with mass 1), # is rmacslest, and a is massless. The entire gouge group T x T is unbroken, ‘Now we expand around gp = 1, which is most easly accomplished by subst tating ¢= 14 ¢ into Lo (N48 +0144) ~ Fda? + 316 DY) = WS = + BI) ~ 0+ dey) = (dP = 0+ [ou — onl?) [a2] + higher oder terms, (2.208) Lead ; gidal? +5 (BY) al(Re In the last line we wrote only the quadratic part; it contains the information about. masses, Alko, we wrote a = (aj,a); the last term is the nonn square of a 1 = ay ~ a9, ‘The imaginary part of the complex scalar field is gauged awey by constant gange transformations (which preserve the trivial connestion y)- Thus: the only scalar field is the rel scalar field Reg with mass square 4. The spinor y 's now massive with mass 2 (and so mass square 4). Set? 28 = ay ~ az and iy =a; +02. Then the quadratic part of the lagrangian involving age 2 Ps (2.208) — 314s — Flea + 206 “Te proses enor prsamtation $ of Spin —1) ke ot eri bat we de anu tat there lean Invariant shew form A'S ~ Roo thas we can write mast term "Hee the macro seal, tthe massive Fld B= somewhat sriteayCHAPTER 2. LAGRANGIAN THEORY OF CLASSICAL FIELDS sr So we see that fi has mass square d and 7 is massless. Note that 7 les in the Lie algebra of the stabilizer of 1 € C, which isthe diagonal T CT x T. ‘This is the Imbroken gauge group. The appearance of a mass for is an example of the Higgs smechanist In general, if His group of globel symmetries of C, then H acts on the space of static classical solutions My. For @ € My we have the subgroup Stabé CH which fixes @. We say that ¢ spontancously breaks #7 down to Stab¢, and thal Stab ¢ is the group of unbroken symmetries. (For example, we asserted above that in MinkowsKi spacetime Poincaré symmetry is unbroken ata vacuum solution.) Td i ‘a vecuium solution, the homogeneous space H/Stabdy is embedded in V~"(0) by the Hf action. ‘Then the perturbative scalar fields with values in /stab dy C Tyo X are massless. These fields are called (classical) Goldtone bosons. They are massiess scalar fields guaranteed by the symmetry. Of course, there may be other roasless scalar fields which are not related to symmetry Example 2.210, Consider a theory of a complex scalar #: M —» C with poten- tial (2.206). Now there is a global T symmetry and the moduli space consists of the origin and the unitcrcle. We do not divide by the global symmetry; rether, st ‘classical vacuum on the unit citele there is a single (real) Goldstone boson field due to the symmetry. At the origin F acts trivially and the entire complex sealar [A soliton is stati classical solution whose energy is not s global minimum. In many examples the space FE, of finite energy static fields is not connected, ‘and a soliton is a minimum energy configuration in a component where the global rninimam is not achieved, For example, consider a scalar Beld ¢: M? — Ron two-dimensional Minkowski spacetime with lagrangian density eau) b= (Jlaal? ~ V¢6)) la Suppose V > 0 and V-'0} = {a,8}. Then the space of inte energy static fields fas components: A stati eld #¢,2) ~ 2) depends only on the spatil vila tir the Rive energy condition means Jim (2) and _im_ (2) le in V0) {a There are 2 vacuum solutions d{z) = a and gz) = b. ‘There are solitons wth fim p(2) = ay lim oz) = b and ako (antieolitons with tim oz) = 4, im 62) §2.11. Dimensional reduction Suppose £ = L++7 is & Poincaré invariant field theory on n-dimensional Minkowski spacetime MI. We obtain a theory on M™-! as follows. As usual lt 2°,...,2" 1 denote cordizates on M*, and 2,,».,2"~? coordinates on M™:!, viewed as the ‘qocient of My tconsations in the 2? direction, Let 1 denote the vector field on F induced by the action of n, and define (2212) Fant = BEF :Lilénoi)6 = 9} ‘These are the fields which are constant in the 2"~" diection, We identify F-1 with space offs on M°"2. Then the dimensionally reduced theory is ems) Linn £0 Maer € MeFi x MO),sa P. DBLIGNE, D, FREED, CLASSICAL FIELD THEORY A scalar field on 2M” reduces to a scalar field on M"™!, but for fields of higher spin the identification of Fy with flelds on M"-' is more complicated, |For ‘example, a I-form on M" reduces to a I-form plus a scalar field on M™=" ‘More generally, if M —+ A isa fiber bundle with a section of Det 7(M/M"), then we ean reduce field theory on M to afield theory on Mf. Tnstend of restricting to fields constant along the bers (dimensional reduction) we can abo include Relds which fluctuate along the fibers (compactifiction); from the point of view of Af” ‘here is an infinite nuaber of euch fields Appendix: Takens? acy. ‘This appendix gives the proof of Theorem 2.16. Our setting is sightly diferent fom that of Takers, who works on the infinite jet bundle J§%,, of E/M. To go ftom local to global results, he wes that if 0 — Fy ++. "Sq — Faqs is a0 eect serene of sheaves, with F; sft for ¢ 0, we ate $0 prove that the corresponding complex (Rd) is acyelic, except in top degree. It will be more ‘convenient to prove a more general statement. Fix vector bundles F,.-.,¥y on B. Let V be their product. The complex (9%, d) for V — M isthe eemplex of formas 4 on M depending (locally) on a setion'§ of E/M and on sections Ey. y6 of 9°V;, For some k, the value of a(fs6iy.-os6) at me M should cepeta only on the hejet of @ and the Ket of, --r,fp atm We denote Qe yu the subcomplen consisting of the a depending Rlincaly on each of €,-...6. The following 1s restatement of Theorem 216. y theorem ‘Theorem 2.214. Ifp>0, the comples Seman is acyclic except in top degre. 1 we take all V to be the relative tangent bundle Ta, then (QR, d) i a dnect factor of Srcous: i i the antisymmetric part for the action of the symmetric ‘soup Sp. The acyclcity of (2,4) for p > 0 (except in top degres) hence fellows from the theorem. Proof. For eq in E, with image mp in M, one can choose (a) a neighborhood U fof mo, and a local coordinate system x: U <+ RS, (b) a neighborioed W’ of en, decomposition W = F x U and a local coordinate system F — R’, and (¢) trivializations of the V; on W. In such «local coordinate system, ain cesar CAM uniquely be written as a finite sun (2.215) (85 £y 0-6) = YD On, ng SOME «BPE In the formula, each nis @ multindex (niy...,M4) and 8% = OP" ...0} isthe corresponding iterated derivative in the coordinate system (13,..-y24). We define 1s usual [rg] = Song. For some k each dn,,..ng depends only on the jet of @ and takes values in @Y" @M%,: i sa section on J*(E/M) of the tensor product ofthe inverse images of the Vj" and of fi, We write F for the increasing filtration byCHAPTER 2, LAGRANGIAN THEORY OF CLASSICAL FIELDS 19 Slr. Ie does not depend on the local coordinate systems used. The diferentia Trnaps Fy to Figs Replacing in (2215) each 2” by its ymbol in Sym!™"(Ty), we obtain the following description of Gr (uy) Te 8 the inductive iit i of the space of sections over JB of the vetor bundle (2216) @ackSym" (Ty) @V™ OM, vith each factor to be replace by its pulk-back to Jf ye The degree Vis the sum ofthe symmetric algebra degrees, The difrentiald inden Grid): Ge (On mat) > Gees WE which islinear over the functions. Iti deduced from a morphism of vector buries on A 27) (sym ctuy) 20m — (69mg) os" by pullback and tensorization by @V™. Locally on M, for {e;} a basis of Th, @2i7) is (2218) soo ((Er0 © (eat passe 01) Joere “The crx ofthe mater snow the flowing Lemma 2.219. On M, for each integer N, the eompler of vector bundles with components (part of degree (N +4) of 8 Syis"(Tar)) © My ‘and differential the morphisms (2.218) of veetor bundles is acyclic, except in top degree Proof. This isto be checked point by point, One then recognizes in the complex fa variant of the Kossul complex, and one can write an explicit homotopy. In more detail: for T the tangent space ata point of M, one has to consider a complex with, components (22% (gsm-em) ease One hes 6 Sym"(7) the diagonal embedding, and if $ fs any supplement, we have omy am (§2) 0 lag? — 9} rat > 1 my 7) = pdb A od — > # |a | meh len {For 00.6 Te is instructive to check the units in the lagrangian. Comparing the two terms we see that-m must have vite L? (to match the units of d). In a relativistic quantum theory we can replace m by meh; then m has wnits of mass, Of course, in that ‘context we usually work in a natural system of units with c= = 1. In that case has mass dimension (n~ 2)/2. Te differential of along the space of fields @ is 86 h +d — m2 662] OA {d+ dg-+-m2g|dal} - {69 Ad8), 67 8) ‘So from (3.5) the classical feld equation is a9) (04mg and the variational I-frm is (3.10) ‘The local symplectic frm w (11) Equation (39) ie most easily analyzed through the Pourier transform oe 40 = ayn [© pix) = Lf grvaTmnigy 6) - oun [. ta a" Th) (x) [deh 2) ‘Then equation (8.9) transforms to (3.13) (Clk? + m?)6(4) ‘Thus the Fourier transform d is supported on the mass shell (aay Om = {hE Mal =m} Note that ince is real we have (33 a) =H).(CHAPTER 5. PREE FIELD THEORIES 08 White k = (lo,--.sky-t; then ky has units £1 and is called the wave murber. The frequency w m kye has units T-* ‘Werean formulate the theory for an arbitrary metric 9 on M as (10) y= (39° 2400.6" 6") ula, ‘where uy isthe canonical density associated to the metric 9. Now a computation shows (37) where (338) ‘So the energy-momentum tensor, 38 defined in (2.176), is (319) T, = db dy — oy In local coordinates we write (se (2.177) 20) Taw = 8b 046+ (~Bldgl + Specialize to Minkowski spacetime. The energy density To |d"~!2| is given by *). nonnegative and only vanishes for @ = 0. Minus the momentum (aan) Tn = (F(00? +3 SOO" + Tr Note that this i density Ta ja" (3.22) By Proposition 2.190 this agrees with the energy-momentum tensor (2.160) given by the canonical Noether currents of translations, as we now verify directly. The vector field g, induced by By is ea) Gale" = 8 Gs)86 = -2,6- ‘Then minus the astocited Noster current is os A HGy)(L 4) = =L ada +948 06 = (Spek + 9ub0.0) + dx”, where gus is the Lorente metric ‘The “trace” of Tis (25) PT yy = (Fa 9.88.94 8, which vanishes ifn = 2 and m = 0, Thus the theory of @ masses scalar feld is conformally invariant in 2 dimensions.194 P. DBLIGNB, b. FREED, CLASSICAL FIELD THEORY §8.3. Complex scalar fields Our convention is that if)" denotes a real bilinear form, then it also denotes the extension to a bilinear form over the complex numbers. If (1) is areal inner product on a real veetor space W, then the associated hermitiea norm on the ‘complexifcation We Is (326) ws (0) (Over the reals or complexes we always use the notation (327) hel? = (w,0)- A complex scalar field is a map ®: Mf — C. The free lagranglan for afield of mass m 20s (8.28) b= { (dB, a8) ~ m2 @,9)} a ‘This theory is equivalent toa theory of two uncoupled free real scalars d,s; simply set got vale vi ‘Then (3.28) reduces to the sum of two copies of (3.7). The variation of (8.28) is (330) (3:29) Bm (6B A dO) + (4B 1 db) — m26B & — mB 69} [2] = 8B a db + mB |e z|) — 68 (d+ dB 4 mB [al] ay for the vatiational 1-form (ss!) SB Aad + 80 A od, So the equation of motion is (332) (O+ma ‘The aralysis of (8.82) proceeds asin the real case. Note the significant difference ‘hat there is no reality condition (3.15) in the Fourier transform. ‘The citcle group T ¢ C operates on # by scalar multipkication, and this syra- rmecry manifestly preserves the lagrangian. (In the language of §2 iti an “internal symmetry.) The corresponding infinitesimal symmetry € i (333) 1(¢)58 = VT. {Of course, we also have the conjugate equation «(6}65 = -/=I ) So the associated Noether current (see (2.88) is (6)(L +9) = VAl(@ + dB - 5 +48), 434) *(CHAPTER 3, FREE FISLD THEORIES 108, ‘The corresponding giobal charge is (3.38) % Va |, eB Boa) e's) a) ‘The enerny-momentum tensor is (30) 1, = dB -db yy ~ 9 Ly or (see (2.177) (a7) Tuy = 0, 0,0 (~Z 8,40) +72 (6,4) ow z We evaluate the energy-momentum tensor for a plane wave (338) (2) = ast, where & = (oy.--yheot) i in the dual of V, the vector space underlying M. We restore the constants necessary to trest masa mass, 6 replace m by mc/. Compo- rents of the wave number (ki,-.-,4y-1) have units L~! and the frequency w = Boe thas units T~! The complex constant @ has units V/MJTL® *. By the equation of motion (8:32), we have (639) Uap = mPet ‘Then (8.37) reduces to (40) Tyy = [alk §8.4. Spinor fields Let V be the vector space underlying Minkowski spacetime M, and let Spin(V) denote the Lorentz group. Suppose S is real spin representation of Spin(V). This means that there ate symmetric pairings ses ov ae fses —v which satisly a Uuitord relation. Let {¢,) be a basis of V and (f*) a basis of 8. ‘We use the dual bases {eH}. {f4} for V",S". Write Tad) = Tee Gy FAP) = Pegs then the Chiford relation is G43) Fete + Boetrg, = aghog108 P. DELIGNE, D. FREED, CLASSICAL FIELD THEORY let TIS denote the odd vector space which is $ with odd parity. A spinor field (pin 1/2 fermion) is a map y: M —» TLS. With respect to the basis we write Gas) le) = vale A dual spinor field is @ map d: M+ T1S*. The analyst of spinor felds below ‘apts easily to dual spinor fields, ‘A'mass pairing for spinor fields is @ skew-symmetric pairing 6.45) M: NS SR. Nonzero pairings need not exist** If thee is a nonzero mass pairing M, then there exists a normalized skew-symmettie pairing 6.40) Ses ‘the normalization condition is Gan Feb arto For simplicity assume that ¢is irreducible. Then we can write (43) Mam, mer We write (49) YM p= Meda, where the name of a bilinear form is written between the arguments. ‘The kinetic term in the fre lagrangian is the Diree form 50) ob = 1(9. 8,9) Equation (3.50) defines a symmetric (in the graded sense) bilinear form up to ‘an exact term, s0 itis exactly symmetric after integrating over Mf, assuming No contribution at infinity. The fre lagrangian is Fy, 3,0 es) b= {5 ve - Fee} iar “The variation of Lis 88 DY-+ Fo BOG - 66M V} [2 PDS) = TY, bv 380m Syn + FD, (a Sf) Tor ample, or m = 2 take $ to be halspinorrepretntaton Iti avedimensionl, so NS=o.(CHAPTER a. FREE FIELD THEORIES wr and s0 (354 61= 54(Dv— 64) Waar for the variational forma y= FP Sy) Hapa} lb bv) a4) (355) Now Af determines a mep 8 — 8° and P a map 8 —+ V.@S*, so that both 1M and D define operators from spinor fields to dual spinor fields. We use the samme lewers M, P to denote these operators. Then the equation of motion is (336) Da My. ‘We analyze (3.56) via the Fourier transform (3.12): (357) VATE isl) = MOVE) for alle Malply both sides of 8.57) by V=IPz,ty. After some simplification using (3.43), (B47), and (8.57) we find (358) PGR) = MPG) for all. ‘Thus the Fourier transform of a solution is supported on the mass shell On. Fur- thermore, equation (9.57) defines a subbundle TIS’ ofthe trivial bundle Om x TIS, and is a section of TIS’. The rank of 5" is dim $/2. Again we have a reality condition (3.15). For m > 0 the subgroup of Spin(V/) which stabilizes k € On is {isomorphic to Spin(ni—1); the Ser of Sat kis a spin representation of Spin(n~ 1). For m = 0 the stabilizer has reductive part isomorphic to Spin(n —2), and the fiber of S" at k isa spin ropresentstion of Spin(n ~ 2) Tn some cases the minimal real spin representation $ admits an action of © for H commuting with Spin(V). Then if P is seaquilinear we have an action of T for Sp which preserves the kinetic term wPY. (The mass pairing (3.45) may break this symmetry.) If there are multiple copies ofthe spin representation, then there is a larger compact group of manifest symmetries which rotate the various copies. ach such symmetry has a Noether current which may be computed from (3.55). As for the energy-momentum tensor we first compute minus the canonical [Nother current associated to translations, as in (2.160). Translation 2, induces a vector Reld &. with HEa)da” = 8 HG,)60 = 0,0 ‘Using (8.51) and (9.56) we compute ~u(G,)(L-+7) (359) yr #2", where (3.60) Ou = range ~ (GVDV — 50M Be ‘This is not symmetric inp,» (a8 expected) Fhe oon encetina case: For chil pir fed in 2 dimensons equation (3.57) aay Ut tn Burne atti rather ceo righemewing, depending oe the chiral See ff Supersolur ‘ons, 26) forthe pre formals.108 , DELIGNE, D, FREED, CLASSICAL FIELD THEORY §3.5. Abelian gauge fields We restrict ourselves to the massless case, In $ dimensions the Chern-Simons functional serves as @ mass term (soe Problem FP4 of (I-Homework)). In any dimension the Higgs mechanism can be used to introduce a mass. For another discussion of thie material, sv the solution to Problem FPS of [I-Homework]. ‘The standard physical interpretation (Maxwel’s equations) is discussed in §4.1 ‘The basics of principal bundles and connections are reviewed in §4.2. ‘We choose az gauge group the circle group TT of unit norm complex numbers. ‘On the Lie algebra Lio(E) 3 Y=1X we have the postive definite inner product (a1) (ab) =~0b, abe V=TR. ‘The field in a gauge theory is a connection A on M with structure group , and ‘we work with fields up to gauge equivalence. The space of equivalence classes is a real affine space. The lagrangian is 1 baa hoa) Lipa (3.62) =~ 3IFaP le] 1 qe [Foun Fire) PH 9h Ie, “where the curvature is (663) Fiat da” Note that since (364) Mdot nde = 9M 9 Be )B_NKE A, we bave (3.5) 9 = Photo lA ile) Now &4 is an imaginary (2.68) the sign comes from commuting § past d. Thus the variation of the lagrangian is SL = (454 +P) (367) AGA +Fa) ~ (6A Nd Fa Hence the canonical Lform is (6A 4) =~ (OA Foo)" 9"* 8p) (3.68)‘CHAPTER 3. PREE FIELD THEORIES 190 and the classical equation of motion is (269) aeFa (a7) oP" 8pF for all v ‘This isan affine equation for A, and since gauge transformations act by an affine action, the moduli space 34 of clasical solutions modulo gaug> equivalence is an tiffine subepece ofthe space of fields modulo equivalence. ‘The local symplectic form (5A A sd) UQubAL ~ 854g) \5AQ) 0% 9" B,)1Ee) (81) Of course, the detailed analysis involves choosing an origin, 90 we may as well from the beginning work with the vector space of translations associated to con nections. Hence in this paragraph only we take the field to bea real Iform a up to addition of an exact real function. The lagrangian is (a2) and the equation ef motion is (3.73) dsda=0 [As usual, we consiler the Fourier transform d= dda on V*, The vector space of classical solutions isthe fist cohomology space of the complex (a4) muy 40m) “#4 001) (On the Fourie: transforms we have ai) =A (15) 10) = VT} de da(k) = IPa(k) ~ (618), An easy argumen; shows that the fist cohomology is isomorphic to the colee- tion of Real functions (4: lightcone — V* @ C} on the lightcone which satisfy {lk}, 8) = 0 modo the ot of Functions {a(#) = F(U)k), where f ranges over the complex fanetions on the lightcone. (The Reality condition is tha of the Fouri tcansform,) The sabgroup of Spin(V) which stabilizes k has reductive par isomor- Dhic to Spin(n ~2), and d isa section ofa bundle whose fiber at isthe vector epresentation of Spin(n — 2). Return now tothe abelian gauge Geld and lagrangan (8.52) To compute the energy-momentum tensor, we fst compute the Noether current J, of minus trans- Iation by 0, (ee (2.160). Note that 3, induces the veto ld Gy onthe space Fa of fis defined by (see (440)) (376) ~H0,)Fa200 P. DELIGNE, D. FREED, CLASSICAL FIELD THEORY Let &y = 9, + &. Then HO, +E NE+Y) 2 O.\Fa hPa ~ Fa ns@,)+ Fa) je (a7) We find after some computation that (a7) yu = (Fan Foe) 0 + SFA Bans where @. i defined in (2160) Notice that i i symmetric. ‘We can compute the energy-momentum tensor more directly using the alter native definition (2.176). Couple the lagrangian (3.82) to an arbitrary metric g 67) bg = Paes Fond 00° nl) Using (8.17) and diferentiting with respect to a~t we find the energy-momentum tensor to be (see (2.177) (80) To =~ (Eo Food? + HEAP Qs which agres with (8.78), a8 it mast by Propesition 2.190. In (4.8) we remit (8.80) in terms of the electric and magnetic components. Fora Lform field o the energy-momentum tensor computed by diferetiating the metre is again (3.80) (where we interpret Fy. = One 2,0,). However, the canonical Noether current, dirs from (2.77) since the indeed vector Beld fcts with an aditonal tent over (2.70) (ast) G10 =~ Lie(O, Ja = =(8,}da~ a,x ‘Then jp picks up an extra term (382) dB, rn eda: = —d{e(9,)a-*da) + (8, ads da. This is written as the sum of an exact term and a term which vanishes om shell "The trace" of T, (383) PT = - DIFP, vanishes when n = 4 This ays thatthe lgrangian (3.62) i conformally Sneriant dimensionsCHAPTER 4 Gauge Theory §4.1. Classical electromagnetism [As in §3.5 we consider an abelian gauge field on Minkowski space M. To make contact with the usual formulas of electromagnetism we take the gauge group to be the multiplicative group R>® rather than the circle group T. (In classical physics the electric charges not quantized, so this makes mote physical sense.) Thus the ‘ghuge field A sa connection on a prineipal R”?-bundle The curvature Fis real 2sform, The formulas of §8.5 hold, only now (8.61) is replaced by ay (2,0) 0b, abER, We work in any dimension , chough of course the electromagnetism of our world isn = 4. Choose 2 splitting M = M' x of Minkowski spacetime into time x space. For now NV can be any Riemannian manifold. Let 2° = ct be the standard coordinate on M'. Define the electric and magnetic fields Ba 6 0'(N) «a , Ba 6 9(N) by (as) Fa = Ba-dt Es Eq end B depend on time t, but neither contains dt (a) (8)Ea = (@)Ba Let ay be the Hodge star operator for the positive definite metric on N. Then if sis the Hodge + operator on A/V (with it usual metric of signature +—~---), we have ( So the lagrangian (2.62) i= ) tFg = Lew Based A+nBy (o) {Bal _ta,P) Wa, 2022 P. DELIGNE, b. PREED, CLASSICAL FIELD THEORY ‘where the norms of the forms E,,B, are computed in the positive definite met- ie on N. A multiplicative constant is usually inerted in L, depending on the ‘units used we omit it? The equation of motion (3.69) together with the Bianchi ‘identity dF = 0 comprise Maxwells law in empty space: dB, =0 Ey Fi a an & dey Baa 0 wae ‘The initial components of the energy-momentum tensor (8.80) reduce to 1/lBal? D T= 5(“Z- + Ba) as Bs ata = Faults the clasical expressions for the energy density and the Poynting vector, while the space components Tj; comprise the Maxwell sess tensor. (Recall that ~T, Integrates to the field momentum.) We can couple the electromagnetic field to a current J € A-1(M). The lagrangian density is 1 (as) Lan b ea naPat dna “The second term is not well-dafined in goneral since Aisa connection, not a I-frm, In fact, if J isa conserved current (4.10) y=, then we can write J = dK and ao by Stokes’ theorem aay) [onane [are is welldefined, assuming suitable decay at infinity. (We discuss such “topological terms” in Chapter 6.) A typical conserved current che Poincaré dual of a close? ‘curve in M, in which case (4.11) is the holonony of A around the curve. In coordinates we write (42) J (dy )Iea, ‘where J? = opis ctimes the chorge density p and minus the spatial components ~J* comprise the current density j (4.3) I= ep|d* | edt ng “mam of units, the constant ie writen ene? 10/4(CHAPTER 4. GAUGE THEORY 209 We can treat J as an external field, which is fixed, or as a dynamical field which varies, For example, we can take J to be Poincaré dual to the path of « dynamical moving point charge of rest mass map and charge . In that case wo add the kinetic term (1.22) for the point paticleto the lagrangian. Thea the fels ae the particle cB) M and the connection A on M. The classical action is ay 5= f-meltflinia [2a [Bearer In thepe formulas we compute the action forthe proper time interval (Ty, Ta] C R. ‘The second term should be interpreted os a multiple ofthe logarithm of the parallel transport of A along a piece of the path 2; see §6.1. It is not a real number, but rather lives in a real torsor (depending on the endpoints). Note that the path is oriented. "The equation of motion for the lagrangian (4.9) is (429) deFyad. Rewriting in terms of B4,Ba,p,j we have dey Ba = cp id's] Ea , 2 Bary Bg = nGt 4h, which are two of Maxwell's equations. (The top two equations in (4.7) remain unchanged.) ‘The action (4.14) is invariant under reparametrizations of the path 2. We ‘compute the equation of motion ignoring boundary terms. The equation for Ais & special ease of (4.16) where the current J is Poincaré dual to g times the path 2. ‘The veriation of the fist term with respect to the path 2—after integration by parts—leads to a contribution to the equation of motion of (see (1.25)) Ce) a 7a) ‘where vis the velocity measured in our fixed system of coordinates. To vary the second vert with respect o x, we use the formula thatthe variation of the holonomy. fs the integral of the curvature evaluated on the variation of the path, Thus the contribution of thie term to the equation of motion isthe vector associated vis the metric ta the Lform (4.16) (ass) (8/01-+ ¥) Fa = GE a — aun) (We omit the de component of the Lom.) So the equation or # aa) 2 (Bag) < es —aonea% “" denoting the dual vector. In four dimensions this is the Lorentz force law,208 , DELIGNE, D, FREED, CLASSICAL FIELD THEORY §4.2. Principal bundles and connections Jn this section we establish some notation and review standard notions. Let M be ‘an arbitrary (super)manifold, All vector fields are even Let P~ Mf be a principal bundle with structure group » Lie group G, often called the gauge group. If M is a supermanifold, then eo is P. We take G to act fn the right. Blements ¢ ¢ gin the Lie algebra of G induce vertical vector fields ¢ on P, and (6.2) = (GG Suppose Vis a space with a left G action, In many interesting cases V is @ linear space and the action is linear, but it need not be. There is an associated bundle V — Mf whose sections are equivariant maps f: P= V. So f satisfies (4.20) f09)=97!-f@), PEP, 966, where \ denotes the action of G on V. Write g = exp(é() and differentiate with respect tot to find (aan Sfa-oh Sem where now "denotes the infinitesimal action of ¢ on V. ‘The bundle associated to the adjoint action of G on g is the adjoint bundle a” = ad P= P xc 9. Asection ofthe adjoint. bundle is then a map ¢: P + g with (pp) = Ad,-» ep). Using the infinitesimal g action on P we can identity ¢ with a invariant vertical vector field é on P. Namely, define & to be the vertical vector ‘corresponding to the infinitesimal action by e(p}. ‘Then by (4.22) (4.22) ate) tae ‘and the corresponding vertical vector feld i Zo]. The fiber of ad P at me Mf is the Lie algebra of infinitesimal antomorphisms of the fiber of P at m. Tt acts on the lft, which explains the minus in (4.22). Equation (422) is a special case ofthe following: If f: P ~ V isa section of an asociated bundle, then (423) hanes. (Compare (421)) "There is also en adjoint bundle of groups P xc — M. Sections of this bundle ‘ct as automorphisms of P, often called gauge transformations ‘A connection on P is a G-invariant distsibution on P which projects isomor- phically onto TM. Thus a vector field 7) on M has a horizontal lift which is a Gcinvatiant vector field on P. Equivalently, a connection is encoded in a g-valued ‘form A € (a) which satises, QAR e cen ae RA=Ad-(A}, 9 €6. Here Ry: P+ P is the (cight) action of gon P. The curvature Fy € 4(9) is (425) B as dianay(CHAPTER 4. GAUGE THEORY 208 Ft follows from (4.24) that F € O3,(ad P) is a 2form on the base Mf with values in the adjoint bundle. If m,n are vector fields on M, then (429) ~ (tinsel - frm) is 4 G-invariant vertical vector field on P, and by (4.25) it corresponds to the section azn rede) Fa = (GRD ofthe adjoint busle Tet e: P ~» ghea section ofthe adjoint bundie and the associated G-invaiant vertical vector Geld, Then for any vector field n on M, (428) A connection A on P induces a covariant derivative V on any associated bun dle V" = P xq ¥. For a vector field 9 on M, and a section f: P—-V of VP, (429) Vad =H. ‘The covariant derivative may be viewed as an operator dq: 0"(VP) = Q'(VP), defined by (430) daf = (d+ AV If G isa linear algebraic group, then we have the Tannakian statement: A connec: tion on P is equiralent to system of connections on all associated vector bundles which is compatisle with the tensor product of representations ‘Automorphisms of P—gauge transformations—act on connections by pullback, ‘An automnotphismn : P+ P may be represented by an equivariant map 9: PG, defined by (430) lp) = stp) ‘Then if A € Mh(6 sw connection, we have (4.32) PAs Ads Ato td, whore gag i the pullback by 9 of the Maurer-Cartan form on (2 ‘The Bianchi identity is simply the Jacobi identity for horizontal vector elds on P: (433) [Vs Veal Val + Ira» Yrals Pal + ans Pale Poa Equivalently, (434) nF a where da: Qy(ad P) —+ My (ad P) is the extension of the differential d using the connection A206 P. DBLIGNE, D, FREED, CLASSICAL FIELD THBORY ‘The set of all connections on P is an alfine space Ap. One sees this easily from (4.24), which ae afine equations. The associated vector space of translations Is My (ad P), The group Gp of gauge transformations acts on Ap on the left using pullback by the inverse (or pushforward of the associated horizontal distribution), Let AULP denote the group of all diffeomorphisms e: P — P which commute with the right G action. Such a y induces a diféomorphism of M, and there is an exact sequence (438) 1+ Gp — Aut P — Dia ‘The last map may not be onto; the image isthe subgroup of Diff Mf which preserves the topological type of P+ M. Also, the sequence (4.35) is not usually split. Infinitesimallyat the level of Lie algebras—we do have a short exact sequence (430) 0 + Lie(Gp) = Lie(Aut P) — (M1) + 0. Furthermore, s connection A gives a epliting as vector spaces (but not as Lie algebras): to a vector field 9 € (AM) we attach its horizontal lift. It soften more geometric (and more physical) not to fix perticular bundle P. ‘Then instead of a single affine space of connections, we study the category Cy4(G) ‘ofall connections on all principal G-bundles over M. A morphism in Cy(G) is an ‘somorphism of principal bundles which preserves the given connections. The set of ‘equivalence clases Cyy(G) may be identified as a disjoint union of spaces Ap/Sp, where P runs over a set of representatives of topological types of G-bundles an MY. In a lagrangian field theory including connections, the category Cyy(@) is part of the "space" of feds Fix a bundle P. On the product bundle (sn, PaApxP—ApxM there is a universal connection A. Its restriction to {4} x P is, and its resteetion to Ap x {p} is zero, It is straightforward to compute the curvature of A at (A,m): evaluated ot mista € TM and a1,09 © 9, (ad P) we obtain mdm )Fa = elma)lm Ea (4.38) un)e(aFa = oe). Hoe) ‘Theve are equations for elements of the fiber (atl P)a ‘The group Aut P acts on P (on the leit) by the product ofits actions on Ap ‘and P. The wniversal connection A is invariant under this action. Lagrangians in gauge theory are gauge invariant functions (partial densities) on Ap x M which are computed from A, and the (Aut P)-invariance of A implies that suck lograngians are invariant under subgroups of Diff M. From (4.95) we see that Diff M acts on the space Ciy(GF of equivalence classes ‘of connections. Also, the discussion after (4.98) shows that there is canonical lit ‘of X (IM) to vector files on the universal bundle P. Piet, iA © Ap is a connection and 7a vector field on M, then n acts on A using the horizontal lit (4.39) Lie(i)A = oli} = on) FaCHAPTER «. GAUGE THEORY 207 Since difeomorpisns of act on fee by pullback, the induced vector field co Anis (440) “nF, at Ae Ap “Thus (4a) Ee Fa tn is the vector feld on Ap x M induced by m. Its horizontal lit to P (using the universal connection A) is the desired induced vector Reld on B. Por vector fields m1,72 on M, we compute (442) (Er oGra} = Gran) ~ dal Falnor2)} (aa) (uso = fone ~ Fans) Equation (442) asserts that the map n> é, iss homomorphism** up to infinivesi- ral gauge transformations, as is evident from the sequence (4.38). Equation (4.43) implies that ifm and mp comme, then the symmetry &y, commutes with m up to fa gauge transformation. ‘This is used in superspace formulations of supersymmettic ‘gauge theories, where the symmetry generated by the vector fields Qu on supe ‘pace commute up to a gauge transformation with the action of vector fields D used to build lagrangians. §4.3, Pure Yang-Mi ‘The data which define » pure gauge theory are: theory Lie group with Lie algetra 6 ee) 6.) invariant inner product on We work on Minkowski nspace M, The field in the theory is a connection A ‘on # principal G-bundle over MM. As explained in §4.2, the collection of feds Is best regarded as a category. We write lagranglans whieh are invariant under gauge transformations Recal (4.25) that for a nonabelian group the curvature F is a nonlinear func- tion of the connection A, So nonabelian gauge theories are not free—the lagrangian is not quadratic n the feld A. For perturbation theory one introduces @ coupling constant (or several if G isa product of simpler groups). We simply absorb these constants in the inner product," “The pure Yang-Mil lagrangian is given by the same formulas as (8.02} (a4 1 “, = HMPA ns) When e Le algebra cts on s manifold on the lft we expect an oattomomorpham. Buti ‘wo viow te space of votes fede ae the Law alate of the group of difoomerphzme, then the Induced brace f minus the uel Lie Brack of vector Bes208 P, DELIGNE, D, FREED, CLASSICAL FIELD THEORY "The manipulations following (5.62) are valid for arbitrary gauge groups, except that swe must replase d by d in the nonabelian case. Thus the equation of motion is (446) 46a ‘and the variational 1-form and local symplectic form are y= (ANF A), w= (GAA edybA). (447) Equation (4:45) is called the Yang-Mille equation. Of course, the connection A always satisfies the Bianchi identity (424), ‘The discussion of the energy-momentum tensor carries over without change, ‘and formula (380) holds: (44s) ~ Bae Food? + SIE Pe Pure Yang-Mills theory is conformally invariant in 4 dimensions (se (3:83), ‘The energy density is ees (449) Teo = ED IRal + FDP, where i,j = 1,...y2—Leun over spatial indices. Therefore, the fed configurations of minimal energy are lat, and the moduli space of vacua on Minkowski space is & point—the equivalence clase ofthe trivial connection ‘Fundamental lngrangians in physical theories are constrained by renormaliz- bitty. With that criterion there ate a few terms one can add to the Yang-Mills Tagrangian (4.5) in dimensions 2, 3, and 4. These terms are topological in nature ‘We often refer to them as “terms In n=? dimensions suppose (450) G5 Rk fs trace on g, ie, a linear map for which ((a,8}}) = 0 for all 0 € g, Then an ‘additional possible term in the lagrangian is (4.51) h; o Be hFal ‘We include the constant @ € R since this s the form in which this @-term is usually written, Thece is a constant @ for each independent trace on g. On s compact 2manifold M (or on the space of fields with finite action on Minkowski space) the invegeal of (4.51) is locally constant on the space of fields, Thus Ly does not affect the equstionscof motion. Tt does, however, enter into the formula for the variational Iform. Namely, the variational Iform for pure Yang-Mills plus Le is 432) T= HUAN oF) + £64).CHAPTER 4 GAUGE THEORY 200 ‘The local symplectic for and energy-momentum tensor are unchanged, Inn = dimensions there is a Chern-Simons term; we discuss it in §6.2. Tbs liffrent than the #-ermo (451) and (4.54): the Chera-Simons term is not locally ‘constant on the space of fields, and so does have a dynamical effect. Tn n= dimensions there is a O-term associated to an invariant bilinear forme (453) (:9e0—Rk (The form need not be an inner product; in particular, it need not be identical to (44d).) The é-torm is ° (454) a= spa Pa A Fad [As with La, the intagrl ef Li locally constant on the space of fields. The modified vatiatlonal -form is (455) ~eansta-+ By eean tan §4.4. Blectric and magnetic charge Te begin, ht pctne be prods Mw BN. far caste teic Bi emt te ese Gt eon yd meson Sine [owes tnd the magnetic charge enclosed by a closed 2-dimensional submanifold EC N is a ats) [P4 (as) Be= f Fa= f B TIneach case E need not be a boundary, hence the quotes around ‘enclosed’. We we the same terminology fora general spacetime M, which may also be Euclidean then only the frst equations in (4.56) and (4.57) make sense. Fora gauge field with ‘general gxuge group G, electric and magnetic charge are associated to a trace (48) re—oR on the Lie algebra 650) = fort), (459) Be) = [1 Magnetic charge is an example of a topological charge; 8(E) is a topological variant which depends enly on the topological type ofthe principal bundle under Tying A. Hence itis locally constant function on the space of felds—in particular,210 . DBLIGNE, D, FREED, CLASSICAL FIELD THEORY ‘the space of clasicl solutions. I s central in the sense that it Poison commutes with any other charge Blectric charge is not generally conserved in a finite region. For example, in classical electromagnetism it follows from (4.16) that the time derivative of the electric charge enclosed by 5 is proportional to the integral ofthe spatial eurrent through ©. Als, if = 42 then the electric charge enclosed by © is proportional to the integral of the charge density p over 2. This is the usual statement of Gauss’ law. We generalize to field theory, where the curtent J i computed from matter fields, as follows. We work with a gauge field A fora general gauge group @. Elements cin the center of g give infinitesimal global gauge transformations ¢ which act trivially on the space of connections: dye = 0 forall A. From §238 it follows that if Li, is a gauge invariant “matter lagrangian” in which A is 2 fixed background field'and which only depends on the I-jet of the matter fields ¢, then the Noether ‘current j. for ¢ in the theory £', is computed by differentiating L', with respect to A (cee Proposition 2140). We now consider A as a dynainical variable by adding the Yang-Mills lagrangian to L',. We write the signs for a Lorenta manifold. Proposition 4.60. Let is(6) be the matter lagrangion described above, and set (481) 14,0) = Ha naa) + Ual@) ‘Suppose 0 is spacelike (n—~1)-dimensional region with emeoth boundary E. Then 1 ¢ i the global gauge trensformation corresponding to a central element ¢ € §, fon-shell we have (452) axa) = f 46 here re) € 9° i the trace ¢m (.6) omg andj 8 the canonical Noethercurzent orc inthe theory Ci Remark 4.63. In Minkowski space we often take 1 to be a spacelike hyperplane— the limit of lage ballin which case isthe sphere at “sata infinity”—the init ‘of large spheres. Then the limit of £7°(5) isthe total electric charge in the system. Proof. ‘There is a caroncal variational 1-form snc the agrangian only depends on the Let ofthe eld. Let C= Ly = Cyt be the total lgraglan. The ‘ustion of motion is (a4) (Deru) "1+ (DEA) Evaluate this on ac, where a € hy is an arbitrary I-form and ae is viewed a5 & tangent to the space of connections: By manipulations as in §3.5 we have (465) a\(DEyy)M9l = =a A fede Fa). By Proposition 2.149 the second tarm is j(a). Since o i arbi (468) (a+ Fa) = dle aFa) Equation (462) i the integral version of (4.60, Formula (4.62) is not modified if we add» #-torm (4.51) of (4.54) to (4.61). (This has the effect of adding a multiple of F4 to +Fa in (4.65), but that extra piece vanishes by Bianchi.) However, formule (4.62) is modified inthe theory of monopoles; se (I Dynamics of QFT, 9.6]CHAPTER 5 o-Models and Coupled Gauge Theories $9.1. Nonlinear o-models “We begin with some preliminary remacks, First, if M is a manifold and E— At ‘a real or conplex vector bundle with connection V, then we form the twisted de ham complex a) 3,(E) 2%, O4,(B) 2 03,2) Tt s not complex in general, but rather 62) B= Ro, where Re isthe curvature of . Next, if 6: M —+ X is a map between manifolds, then dg € Q4(@°TX). Ie now PX has a connection V, then 63) dod) = 6°Te, where Ty is che torsion of the connection ‘J. We will apply this to the Levi-Civita connection o' a Riemannian manifold; its torsion vanishes, "The dats which define a nonlinear o-model are 5 x Riemannian manifold ed V:X—R potential energy function ‘The field in the theory is a map ¢: M —» X; the space of fields is F = Map(M, X), Diffeomorphisms of X act on F by composition, and diffeomorphisms of M a:t ‘on F using composition by the inverse. In each case the subgroup of isometriss preserves the lagrangian (5.6) below. There is canonical evaluation map 63) eFxM XxX which is imvatiant under the product action of Diff M on ¥ x M. Infinitesial symmetries act on derivatives of the field ¢ via the Levi-Civita covariant deriva- tive V. Thisintroduces curvature terms into the bracket of infinitesimal syrametriss in general, Since e i Invariant under Diff M, and since the lagrangian (6.6) belewaa . DELIGNE, D, FREED, CLASSICAL FIELD THEORY ‘can be writen intern of Sis easy to check that isometies of M are manifest symmetries (zs are isometries of X). “Te o-mnodellgrangian with potentials agi? — 9" b= {de ev est 6) = Hap nde) ov (ta Note that |d@) is computed using both the metric on AM and the metric on X. We cary out the femiiar analysis SL» (bod A140) ~(66,0" ead V) eho —d{ (66 Add) } ~ (60. dry + d9) — (60,0" grad) [a2 Hence the eqeation of motion is (68) Dy6=-$ erdV, ‘where Dy is the covariant extension of the wave operator (3.2). Despite this no- tation one shouldn't lose sight of Newton's law, which is the special case n= 1 of ‘equation (6.8). The variational 1-form and local symplectic ‘orm are 1 (6 add), w= (rd 6$.N 66). {es easy to couple this theory to an arbitrary metric gon Mf: wn 69) 610) Ly = (Sor 0b 0.0~6°V) u4(2) ‘The computation leading to (3.19) is essentially unchanged, and so the energy- ‘momentum tensor is 1) ‘or more explisity (se (2.177)) (db d6)ny— 9, lds? + 8°V) oe ‘On Minkowsk’ space the moduli space of vacua—feld configurations which minimize the energy deasity Thais (13) Mage = VH0) 12) Tue ~ 8,8 818+ (— ‘assuming that 0 isthe minimum valve of the potential energy V. If € is a Killing vector feld(infinivesimal isometry) on , then the associated [Noether current is (614) J (Goad). ‘There ate also topological terms one can add to a pure o-model, usually called Wess-Zumino terms. We describe them in Chapter 6,CHAPTER 5. o-MODELS AND COUPLED GAUGE THEORIES as §5.2. Gauge theory with bosonic matter ‘The theory in this seetion is the most general bosonic theory without gravity”? (though we do no: include all possible terme—eg. topological terms—in the la- trangian). We can also describe it as @ gauged nonlinear o-model, or as & gauge theory with boson'e matter. The data which defines the theory are e Lie group with Lie alsa ¢ was enunin mand on wh es by mes Vex —R poet oan earn ander The els we exw czaestinn on ome pial Chane P—> & section of the associated bundle P xo X —» M It is often convenient to view # as an equivariant map 4: P— X. The lagrangian combines (4.45) and (56): Lines 4 2 ge (5.17) b= (-FIFaP + 3ldagl? - 9°V) laa ‘There is a new term (18) (6A. ddd) in SL from the coupling of A and 4,50 a new term in the equations of motion. The variational I-form is (619) 1 = (66 0 odd) ~ GAN +P a) ‘The energy-momentum tensor is (6.20) Tu = (On)u8 (On)o8— (Eves Fo) 9°? + (ZIAD ~ Sad? + 6°V) our ‘The moduli space of vacua is (621) Meee = V0) /G assuring that O i the minimum value of the potential energy V. In §28 we described a simple exemple of “gauging 2 syimmetty” for p> 2, fr vample The lagangian bere cover most fendarentat (ve fete) ageanglansCHAPTER 6 ‘Topological Terms We have already introduced “@-terms" in gauge theory in §4.3. These are wort of topological term related to primary topological invariants, this ease characteristic clases of principal bundles. Corresponding tert oovur in ¢-medel fs well, Namely, if @: Af — X is a field in a o-madel defined on an oriented spacetime M", and w € 2"(X) ie a clooed differential form, then we can insert & tem (en) edu into a lagrangian. (The minus sign indicates that (6.1) is a contribution to the ppotentil energy.) Here c ¢ R is a constant; if w has periods which are 2xh times intogers, then only ¢ (mod Z) enters into the quantum theory. In this chapter we ‘consider a different type of topological term which is related to secondary topo- logical invariants. Por example, in a gauge theory the holonomy of a connection is'a secondary invariant associated to a frst Cher class; in three dimensions the Chern-Simons invariant is associated to # four-dimensional characteristic class, In covmode's we meet Wese-Zumino-Witlen terms which are secondary invariants as- sociated to cohomology classes in the target space X. ‘Tost such terms ito the general theory of Chapter 2, we need local difeential- ‘geometic objects which integrate to these secondary invariants. We indicate an ‘extensicn of ordinary calculus to include these objects briefly, though a systematic development of foundations for this extension is lacking. In the first few sections ve focus instead on examples. orboth #-terins and topologiel terms we need to work over an oriented space time. The orientation allows us to pass from differential forms to densities. Note that neither 8.terms nor topological terms depend on a metric on spacetime. §6.1. Gauge theory ‘The simplest example of a topologicl term was already discussed at the end of §1-1 and in 41. Namely, consider a theory on a spacetime M” which includes a con- nection A for the group R°°. Suppose 2: R — M is a parametrized path which ‘may be fixed or variable. ‘Then we can introduce a term (62) -9a"A asane P. DELIGNE, D, FREED, CLASSICAL FIELD THEORY into the lagrangian ofthe theory, where q € Risa constant. Note that (6.2) is meant to be integrated over R, not over spacetime A. It is unchanged by orientation: preserving difeomorphisis of Rin other words, it oniy depencs onthe image of 2 Viewed as an oriented submanifold of Now -a2°4 is diferent from other terms in lagrangians we have sen so far. AA connection A on an R>°.bundle P — M is an element of £(P), and 2°A a ‘form on the pullback x*P. It is not a I-form on the base &. To compute the action we can fix a tivalization of =P. Taking into account the dependence on this trivialization, the action over an interval (71,73 CR is a homomorphism of Retorsors, the R-torsrs being q times the fibers of "Pat Ty and Ts, In the lagrangian approach to the quantum theory itis not the ation S, but rather the exponential exp(V=1'/h) which enters. ‘Then this term in the action i parallel transport in the circle bundle associated to P via the homomorphism wot ars (vane For usual kinetic and potential terms, the exponentiated action is an element of ‘This parallel transport is instead 8 homomorphism of circle torsors ‘The equation of mation (Lorentz force law) arising from (6.2 in classical elec tromagnetisn is discussed at the end of 4.1 In the quantum theory of electromagnetism the electric charge gis quantized in ouitable units. Tt makes sense fom the beginning to regard the gauge field A 5 a connection in a T-bundle (as opposed to an R>*-bundle). More generally, in quantum Yang-Mills theories the gauge feld is a connection for a compact gauge group G. ‘Then it makes sense to write a term 2°7(A) in the lagrangian, where the trace r= gq» Ris 1/V/=T times the differential of an abelian character G'—> T. This term in the exponentiated action is again interpreted as a homomorphism of circle torsrs In three-dimensional gauge theory theres topological term due to Chern and Simons. Let M® be a 3-dimensionel oriented spacstime and considera theory with gauge field A. 1° is the gauge group, and () an invariant inner product on the Tie algebra 6, then the Chern-Simons term is (63) os tos = (AnFs)~ HARAAAD If A is a connection on P, then this term is a S-form on the total space of P. Assume for convenience that M3 = Af! x 52 ig the product of time M? and a space 3? which is a closed oriented surface. Then once again the exponentiated action on (7), Ta] x Bis best interpreted as a homomorphism between T-torsos Tj and Tp defined from (6.4) at {T,} x ¥ and (T2} x E. (For this we need () 22 Ie in a distinguished lettice of inner products.) For trivializable bundles P the torsors are trivialized by a choice of trviaization of P, aud may be constructed by such trivilizations. We omit the details, (For nontrvializable bundles we need refinement of the inner product ¢,-} to an element of H*(BG;2) in ender to define ‘he torsors and the exponentiated action) Locally we ean break gauge invariance and fix « trivilization of the bundle which carries the conection A. Then (6.4) pulls down to local Sform on Mf, We compute 65) bles +dy= AAI 4),CHAPTER 6. TOPOLOGICAL TERMS ar where 6s) AN BA), So for pure Chert- Simons theory the equation of motion is on Fa For the particular spacetime M? = M x 5, the space of solutions 71 up vo gauge equivalence isthe moduli space of fat connections on ©, The local symplectic fort (68) ” (6AN8AY integrates on E to (twice) the usual syrapletic form: on 7A, ‘The equations of motion (6.7) and the local symplectic form (6.8) are gauge invariant, whereas the variational Iform (6.6) isnot. This suggests that « Noether charge is po longer well-defined real nurmber. Indeed, if «is a central element of g, interpreted as & constant gauge transformation, then the associated Noether ceurent from (6.6) is formally (69) den Ay "The integral over e region $€ & is minus the logarithm of the holonomy around 6D of the bundle associated ta the exponential of the trace (¢,») on g. This logarithm is welkdefined only up to integer shifts. (Ifthe Chern-Simons term is added to the lagrangian of a standard 3-dimensional theory, as in (4.61), then this term contributes to the electric charge (4.62).) §6.2. Wess-Zumino-Witten terms Consider a -model with field g Mf —+ X, whore the spacetime M isan oriented n- ‘manifold. Under the simplifying hypothesis Hy i(X) = Ha(X) = 0, we construct 2 term in an action associated to a closed form © © 9°*¥(X) with periods in 2n7Z, ‘To remove the simplifying topological assumption we need a more precise form of this data, a8 we explain in §6:3. For simplicity, eonsider a product spacetime M= MN" where we assume space N to be & compact oriented n-manifold. We first construct a T-torsor 7; = 7;(®) for eech map f: NV —+ X as follows Since Hy-i(X) =O there exist extensions F: ¥" + X to oriented n-chains ¥" wwith OF = f. Bach such F trivializes Tj. If F,, Fy are two such extensions, then since Hy(X) = O we can find an oriented (n+ I}-chain H: 2°" =» X with OF Fr~F). Then the isomorphism fom the F}-trivialization to the F-trivialiaation Le Because of the assumption on the periods of ©, this is independent of the choice of H. ‘We can allow Hy_(X) ta be torsion in this construction if we specify a coho- mology class © © A"(X;2nK2) such that the de Rhamn class of isthe imoge op (6:10)ns P. DBLIGNE, D. FREED, CLASSICAL FIELD THEORY of cin real cohomology. Then a multiple of f bounds, and we essentially use the choice of to divide by that multiple in the abelian group 7. For another perspective, consider the usual diagram, Map(W,X) «Nv x (eu) |r Map(N, X) ‘Then ae is &2form on Map(W,X) with paiods in 2nk2, and 1,e%c i a refine rent in H2(Map(N,X);2RR2). So there exist principal T-bundles with onnes tion over Map(N, X) whose curvature is —(V/—1/A)rye"® and whose Chern class is r.e*e/2uh. Tho construction above gives particular such T-bandle i can be extended to produce a connection as wal. ‘Now if: [Ta] x N —» X ts ld, the exponentiated action exp(V=T 5/2) of gis naturally an element ofthe crletosor (612) Tog = Toray Ten yen Namely, i Fie a tivaliation of Tarynn then Fe+ 4 ~ F; bounds an (n-+3)- chain H: 2° ~ X, and we use formula (6.10) to compute the action inthis tevialietion ‘We remark thatthe geuge theory constructions of §6.1 may be understood—at least heuristeally—as special capes in which X = BG isthe clasiyng space ofthe gauge group C. §6.3. Smooth Deligne cohomology Fix a smooth manifold X and an integer? n, Let us consider on X the constant sheaf Z, the constant sheaf Rand the following complex of sheaves, denoted F*(0) (or simply F(D)): the subcomplex of the de Rham complex given by cine af pen om w= { op, tpsn Both the cohomology of X with coefficients in Z and the hypercchomology of X ‘with coeficionts in F() map to the cohomology of X with coeficent in R. The first because Z maps to R. The second is just the cohomology of the complex A(X) + W(X) starting in degree n, which is a subcomplex of the de Rham complex computing the real cohomology. Suppose we have somehow lifted those maps at the cochain level. Iu other words, suppase we have found natural complexes (Cj. de). (Cze.den) (or simply (C3,dr)), and (Cf, da) with HGR) = H(X,2), (Cp) = W(X, FU), and HG) = HAR), and found morphisms of complexes ory vaser: Ob, Cpx > OF Our cic of indening males the Secrition of products blow move nature. However, thee i ‘sift the application olagranginn Bld thor” Ii coeyeles forte cohomology group D**™ ‘which enter lagrangins for neiensional spacetimes(CHAPTER 6. TOPOLOGICAL TERMS 20 inducing the maps we described in echomology. This can be done in many ways— tone way will be described later—but the philosophy of cohomological algebra tells that they are essentially equivalent. ‘One can then form a mapping cone (Kd) with Kr = Chach@ce" 6.15) (ou) dom dy +dp ~ dn + 0 ve. [A peyele © = (ca, cp cx) of Kis the data of preycles ox and cr of Cz and land of a homology between their images in Cg. ‘The mapping cone K behaves as if tne had a short exact sequence of complexes + K — Cz@Cp — Ca — 0. For ‘stance, one has a long exact sequence of cohomology groups. The smooth Deligne Cohomolgy group DP ix HP(K), For p en ie ie HPM*(X,R/2). Por p > n itis HP(X,2). We will be mainly interestd in D” =D", Tesits in an exact sequence (616) HX, Z) + HN(X,R) + D® = H(X,2) @ (closed neforms) — HX.) and it is an extension of the group of closed n-forms with integral periods by BMX R/2). ‘This description of D* isnot the most economical, but it suggests the func- torial properties to be expected. Products: One has product maps on Z and fs well a5 product maps F" @ F" — F°. If the corresponding cup-product in Cohomology is expressed at the cochain level, giving products in Cay Ce) sod Cr compatible with py and ge, one obtains products D>" @ D¥™ — Dem". In: feed, hornoiogies c and c between the images of cz and cp (resp. ce and cp) give f homology between the images of exe and erp. In fact, there are two naturally chomologous homologies: plez)¢ + or(c) and apc) + plce)e- Integration: IE f proper submersion J: X — Y of relative dimension d has oriented fibers, inte- gration along the Ribers / has meaning in integral and real cohomology, a8 well ax from F" on X to F*-¢ on Y. Expressed compatibly at the cochain level, it should provide yyy: D>*(X) —~ D?-4*-€Y), and in particular D(X) — D*-4(Y), One way to find complexes Cj, Cp, and Cj is to use the Cech method for computing cohomology. If(U}er ivan open covering of X such thatthe nonempty intersections Uy, = Uigl“AU,, ate contractible, one can use for Gj the Cech ‘complex (617) = [TP .44s2s and for Cp. and Cj the simple comp asociated to the double complex CH= TIP Case PO) oa TIP Gein) ce For n > 0 the map Z. F ~ of complexes of sheaves on X is injective. If 80 8 ya ~ oe: C ® Cp — Ch, the mapping cone Khas the same cohomology, up20 P, DELIGNE, D, FREED, CLASSICAL FIELD THEORY to a shift of index by 1, asthe eokemnel of ya — gr, a complex whose cohomology 's the hypercohomelogy of the complex of sheaves (6.19) (C(values in B/Z) 0! + If. we denote by {2° this complex, we have in the Cechist computation Thea, nd D> is the cobomalogy group H?-Y ofthe complex For n = 1 we find that D! is the group of C™ maps from X to the circle R/Z. For n=2 elements of D? are represented by cocycles (6:20) coker va ~ vr) 7 oa I %y —0 See ae 94) +e = Gu 6.22) ee a1 ay = doy The exp(2n/=Tg,) are the transition fonctions for a F-bundle, trivialized on the Uj, and the 2r/—Ta, viewed as local connection 1-frms, provide this T. bundle with a concecton. ‘The group D? is the group of isomorphism classes of “T-ounles with conection. More precisely, «cocycle defines a bundle with connec- tion and a homology defines an isomorphism between bundles. In 98.2, what we needed to avoid assumptions onthe homology of X isnot just «closed (n+ 1)-frm with integral periods, but rather a class in D1, or rather a cocyete giving sucha clas, If H*(X, 8/2) = 0 (equivelenty: Hy(X,Z) = 0) one has (6:23) D**9 + (clased (n + 1)-forms with integral periods). If H°-"(X,R/Z) =0 (equivalently: Fy ), Use ambiguity in the choice ‘ofa cocycle becomes irelevant: if ¢; and cz are two coeycles representing the same ‘clas in D™", not aly are they homologous, ¢; ~ ¢2 = de, but any two choices c and e” for ¢ are homologous: ¢ ~ e = dé. Ia cocycle ¢ chosen and if is a field, ie, a map from M to X, then cis cocycle giving a class in D°*1(M). For NV in M a compact subvariety of dimension n= 1, ntagration of #"c on N’ should produce @ T-principal homogeneous ‘space—a one-dimensional complex vector space £(N,%e) with metric attached to N and g*e. Indeed, integration on N maps D"**(M) to D? of a point. If 1V is the boundary of @ singular chain , integration of ¢"¢ on S should produce unit vector in CIN, g%6). For instance, in a Hamiltonian picture, a spacelike hypersurface WV, supposed here compact, would give L(N,¢*c) and the slice $ between two such hypersurfaces Ny and N; would provide an isomorphism (the action integral) irom £(N;, °c) to L(Na,¢"e). ‘A systematic treatment of these expectations has yet to be given,CHAPTER 7 ‘Wick Rotation: From Minkowski Space to Euclidean Space ‘A besie constraint on a Minkowski space action is that it be real. An action Sy isthe integral of a Ingrangian density Lye over Minkowski space [os CChovee a time ton M. Then we (Wick) rotate to Euclidean space E ty introducing imaginary time aay Su (72) revolt By convention the Euclidean action is 1/\/=T times the rotated Minkowski action: (73) ry Su=Se~ { be Note that eV=T5w = e-S#. Also, Se is not real in general ‘We describe the continuation to Euclidean space more precisely for a a-model ‘The field is a map 6: Af —+ X into some Riemannian manifold. The complexifi- cation ofthe space of maps M —+ X is the space of holomorphic maps Me — Xe between the complexifed spaces. ‘The lagrangian extends to a holomorphic fune- tion on this space, and the Buclidean action isthe restriction of this continuation to maps E —» X, (Note that Be = Mc so B.C Me.) There is a similar picture for other types of feds ‘We consider four types of terms which typically occur in an action: kinetic terms for bosons, potential terms, topological terms (also Oterms), and kinetic terms for fermions, Tn thie chapter set the speed of Light Minkowski space M. So the metric is ‘We use the conventions in §8.1 for co) gue = dl — (ds! ~~ (ae ‘On Euclidean space E we use the positive definite metric (5) ge = de? + (da!) oo + (YP, $0 as to avoid confusion, we fic the standard orientations ,2',...,2%-! on M and r,2',...,2"" on E, aud we write lagrangians as forms rather than densities, ‘hough we often omit the "A" sign. Let 78) aga da! Ae nde amama . DELIGNE, D. FREED, CLASSICAL FIELD THEORY 87.1. Kinetic terms for bosons Consider a particle of mass m moving in seme Riemannian manifold X. Te ia described by a map £:M! — X. Then the kinetic anergy density is (ae (1.13)) m |e)? an tn = |S at ‘The continuation to imaginary time—after dividing by yT—is m |e ar (78) te= 5 [FI In higher dimensions, we considera real scalar feld on Minkowski space, which is 1 real function #: M+. The kinetic lagrangian is os) Lag = 5 lily dts, ‘where |- [ais the norm (7.4) on Mf. The continuation to B ie (7.10) Le 1 wel drat 2 il dra where |: isthe Euclidean norm (7.5). For a gauge field? A the kinetic term is the Yang-Mlslagrangian (72) tus Fal eter ‘The continuation to Euclidean space ie 7.12) Lem ZIFalb deat §7.2. Potential terms For the particle x + R— X, the potential energy is described by a function V XR, The corresponding term in the lagrangian is 713) Ly = -Ver(e)) a "The continuation to imaginary time is (4) Le = V(x(s)) dr. ‘The extension to higher dlimensions is the same: Potentst terms appear with a — sign in Minkowshi actions and with a + sign in Bucideas actions Fer» emaete example of rotating gouge Fal v9 Euclidean spce, se the sltin to Prob lem 1 of fl Dynamics of QPT, Bnercles,(CHAPTER 7. WICK ROTATION 0 EUCLIDEAN SPACE as §7.3. Topological terms and 6-terms Leta be a real Mform on X and consider the legrangian (Tor 2: — X) (735) Ly = ~2"a, ‘The continuation to imaginary time is innocuous except for the conventional divi= sion by V=E (726) be=vaire. Hence in the Euclidean (imaginary time) lagrangian the topologial term is imag inary. A similar factor is introduced into the continuation of any topological term {in the sense of Chapter 6). "The O-terms in gauge theory (§4.3) behave similarly. For example, the coutin- ation of| a cay ty £ue to Buctidean space is (738) Le = VA Z(o) §7.4. Kinetic torms for fermions ‘We begin with the particle 2: R -» X and add an odd field W which isa section of 2 IITX, the parity-eversed pullback of the tangent bundle. ‘Thon wis real and in real time its kinetic term in the lagrangian is my (7.19) Lu = FO, Goat Rotating to imaginary time and dividing by VT, we obtain -vaz Consider now a real spinor felé yin n-dimensional Minkowski spacetime. We retain the notation of §3.4, but we set the mass M = 0. (The mass term is @. potential energy term, 60 is covered by the discussion of §7:2.) The lagrangian in Minkowski space is ea (7.20) Le coy ar. Lu = 5oupaea te ay 2 = et Vdadn tarts ‘The fist observation when rotating to Buclidean space is that a real spin represen {ation $ of Spin(1,n ~ 1) extends to & compler representation Se ofthe complex spin group Sping(n), but the restriction to Spin(n) C Sping(n} is not necessarily teal. ‘Thus inthe Euclidean theory we take the field (722) eB MSe24 P. DELIGNE, D, FREED, CLASSICAL FIELD THEORY to bea complex spinor Feld. Again: hare are no renity condition en spinor its sn Buclidean space. Define psitings Pe: S8@S¢— Ve me Pe: S208 Ve yy 2, = Va wi 0 TE, cay ele —V Aad, he = -v=T Ea eden = ~ Cae eye = - Eu), ‘where {= 1,...,n — 1 runs over the spatial indices. Then I'p, T's satisly a Clifford relation, (7.25) Ceyr(Pelie + Fayed = als 8 ‘The Buclideen Dirac form ie (7:26) wey ‘and the factors in (7.24) are chosen so that the lagrangian (7.21) rotates to = FaPvadshe (ra) Le = 50Devdr ad te Jn Euclidean space,REFERENCES ‘There are many mathematical accounts of classical mechanics, for example V.1 Arnold, Mathematical methods of elassial mechanics, Translated from ‘the 1974 Russian original by K, Vogtmann and A. Weinstein, Springer, New York, 1989, Many elements of the formalisin we adopt here for classical lagrangian field cheory say be found in G. 3. Zuckerman, Action principles and global geometry, Mathematical As- pects of String Theory, ed. S. T. Yau, World Scientific Publishing, 1987, pp. 259-284 ‘The cotomology of local forms was investigated in F.Takens, A globel version of the inverse problem of the caleulus of varia tions, J. Dif, Geom. 14 (1979), 543-562, ‘The original reference for Nosther's theorem is E. Noether, Invariante Variotionsprobleme, Nachr. Kanig. Gesell. Gottingen, Math-Phys, Kl.(1918), 235 257; Bnglish translation in: Transport. Theory and Stat. Phys. 1 (1971), 186-207.