Supergravity in d = 11 and d = 10

Felix Br ummer

The rst equality shows that the integral does not depend on the metric. The second one keeps general covariance manifest (here the metric and Hodge are those of M, so is a scalar).

A familiar example where p-form notation can sometimes be useful is electromagnetism in d = 4. These are geometric animals which will frequently There is a 1-form A with components A (the gauge appear, so we start by giving a (somewhat sloppy) potential), which is locally determined only up to overview of p-form notation. A p-form in d p a U(1) gauge transformation dimensions is an antisymmetric tensor eld with p A A + d (6) indices. In particular, 0-forms are scalars and 1forms are vectors. All operations on a p-form with any 0-form. It gives rise to a 2-form F = can be dened in terms of its components 1 ...p d A with components F (the eld strength). F is in some local coordinates. What we need is invariant under gauge transformations of Eq. (6) by Eq. (2). The kinetic action in some 4d space-time Addition: Two p-forms are added by adding S is their components. 1 1 F F = d4 x g F F . (7) Exterior dierentiation: The d operator turns S = 2 S 4 a p-form into a (p + 1)-form. It acts as One may couple an external particle of charge q (d )1 ...p = (p + 1) [ 1 ...p ] . (1) (an electron) to the gauge eld. This is done Here (as always) [ ] denotes weighted antisym- by integrating the 1-form A over the particles 1dimensional worldline C S : metrization. Note that d d (anything) = 0 by symmetry . (2) S= 1 2 F F + q
S C

p-forms

A + ...

(8)

The wedge product: Given a p-form and a Alternatively one could group all charged particles q -form , one may form a (p + q )-form into a current density 1-form J : whose components are 1 (p + q )! S= F F + A J + . . . (9) [1 ...p 1 ...q ] . ( )1 ...p 1 ...q = 2 S S p! q ! In this formulation Maxwells equations are very Hodge dualization: The operator turns a p- short: d F = 0, d F = J . (10) form into a (dp)-form in d-dimensional space: The homogeneous equation, or Bianchi identity g 1 ...p ()1 ...dp = 1 ...p . 1 ...dp is as usual a direct consequence of working with a p! (4) gauge potential. In p-form language it follows from Here g = det(g ), and the symbol with all F = d A and Eq. (2). lower indices is dened to be 1 or 0 as usual, with indices raised by the inverse metric g . Integration: A p-form can be integrated over a p-dimensional manifold. In local coordinates, =
M

(3)

Spinors in d dimensions

dp x 0 1 (p1) d x
p

(5) g () . 1

The d-dimensional Cliord algebra can be dened in analogy to the four-dimensional case. It is represented by d Dirac matrices 0 , . . . , d1 such that { , } = 2 1 . (11)

A representation of the Lorentz algebra so(d 1, 1) It turns out that one may consistently impose this condition on Dirac spinors if d 0, 1, 2, 3, 4 is then furnished by mod 8. One may impose it on Weyl spinors only if i = [ , ] . (12) d 2 mod 8. 4 From these considerations we obtain the numIn d = 2, a possible choice of Dirac matrices is ber of real degrees of freedom f in the minimal spinor representation in d dimensions: 0 1 0 1 8 9 10 11 12 0 = , 1 = . (13) d 2 3 4 5 6 7 1 0 1 0 f 1 2 4 8 8 16 16 16 16 32 64 From this one can iteratively construct the for even d: = d2 = 1 1 0 0 1 0 1 1 0 , , ( = 0, . . . , d 3), 0 i i 0 , (14) where are the Dirac matrices in d2 dimensions, and 1 is the 2d/21 2d/21 unit matrix. The dimension of the Dirac representation is thus 2d/2 , and there are 2d/2+1 real degrees of freedom contained in a 2d/2 -dimensional complex Dirac spinor. The analogue of 5 in d = 4 is = id/21 0 1 d1 . has the properties ()2 = 1, {, } = 0, [, ] = 0 . (16) (15)

Why d = 11 is special

d1 = 1

Using the above table we can deduce that the maximal dimension for supersymmetry is d = 11. The argument goes as follows: For arbitrary d we can always compactify on a torus T d4 to d = 4, which preserves all supersymmetries. Supersymmetry transformations are generated by spinorial supercharges. A d = 4 supercharge has f = 4 degrees of freedom; a d = 11 supercharge has f = 8 4 degrees of freedom, i.e. it corresponds to 8 distinct supercharges {QA }, or N = 8 SUSY, after torus compactication to d = 4. A hypothetical d 12 supercharge would correspond to at least N = 16 SUSY in d = 4. We will show that such a large number of d = 4 supercharges would lead to 4d elds with spin > 2, and comment on why this is unacceptable.

In d = 4 the supercharges are written as twocomponent complex Weyl spinors which satisfy the The eigenvalues of are 1. Dirac spinors in even algebra (in the absence of central charges) d are reducible since they admit a decomposition A into two Weyl spinors of opposite eigenvalues, or {QA } = 2 , QB P B , chiralities: The Dirac representation splits into two {Q, Q} = {Q, Q} = [P, Q] = [P, Q] = [P, P ] = 0 Weyl representations of dimension 2d/21 . (19) If d is odd, one obtains the Cliord algebra by are Weyl taking the Dirac matrices of the (d 1)-dimensional Here P is the four-momentum, and Cliord algebra and adding d = . The Dirac rep- spinor indices, A and B label the supercharges, and resentation of the Lorentz algebra is then 2(d1)/2 - we have suppressed all indices in the second line. dimensional and irreducible. Consider now a massless supermultiplet, P 2 = In certain dimensions it is possible to fur- 0. In a reference frame where P = (E, 0, 0, E ), ther halve the independent degrees of freedom of the rst line of Eq. (19) becomes a spinor by imposing the Majorana condition 4E 0 A {QA B . (20) }= , QB = B (17) 0 0 where B is some product of Dirac matrices satisfy- The operators QA A are represented by zero 2 and Q2 ing in this frame since they anticommute with everyB B 1 = . (18) thing. The operators QA 1 lower the helicity of any 2

state by 1 2 . A N = 8 massless multiplet contains a spin-2 eld (a graviton); its two polarisation states |h+2 and |h2 with helicities 2 are related by repeatedly acting with the QA 1:
8 |h2 Q1 1 Q1 |h+2 .

Here F4 = d A3 is the U(1) eld strength 4-form. The Chern-Simons term is gauge invariant up to a total derivative, since under a gauge transformation Eq. (22) it transforms as A3 F 4 F 4 = A3 F4 F4 + A3 F4 F4 + d 2 F4 F4 d (2 F4 F4 ) . (24)

(21)

A cticious massless multiplet of N = 16 SUSY in d = 4 would have to contain a spin4 eld. It is often stated that interacting theories with elds of spin > 2 cannot be consistently coupled to gravity, although recently some exceptions may have been discovered (involving an innite number of elds, and on curved backgrounds). Barring such exotic constructions, the maximal supersymmetry in d = 4 would be N = 8, and the maximal dimension for supersymmetry d = 11. The above arguments can also be applied, with slight modications, to massive multiplets, and generalized to the case non-vanishing central charges.

The last equality follows from the product rule and from d F4 = 0 The terms in the action involving are somewhat more complicated. They are however also fully xed by local supersymmetry and gauge invariance.

Type IIA in d = 10

When compactifying d = 11 SUGRA on S 1 and truncating to the massless spectrum, one obtains 4 d = 11 supergravity type IIA SUGRA in d = 10. The number of real supercharges is still 32, contained in two MajoranaIn d = 11 there is a unique supergravity theory; it Weyl spinors of opposite chiralities. The eld concontains 32 real supercharges, or equivalently a d = tent is 11 Majorana spinor supercharge. The gravitational multiplet is the unique supersymmetry multiplet. the graviton and its gravitino superpartner, It contains the graviton, a gravitino , and a 3-form A3 . There is a U(1) gauge symmetry under which the gravitino is charged and for which A3 is the gauge eld. It transforms as A3 A3 + d 2 with 2 any 2-form, analogous to Eq. (6). The bosonic action up to two derivatives contains the usual Einstein-Hilbert term along with a kinetic term and a Chern-Simons term for A3 , S 1 2 2 + 1 2 2 d11 x g R F4 F4 1 12 2 A3 F 4 F 4 . (23) 3 (22) a real scalar dilaton and its dilatino superpartner , a 2-form gauge potential B2 with eld strength H3 = d B2 , a 1-form gauge potential C1 and a 3-form gauge potential C3 with eld strengths F2 = d C1 and F4 = d C3 . Alternatively, one may dualize the eld strengths to obtain F6 = F4 and F8 = F2 , with magnetic gauge potentials C5 and C7 , such that d C5 = F6 and d C7 = F8 . The formulations in terms of either C1 and C7 are equivalent, and similarly for C3 and C5 . There is a version, so-called massive type IIA, in which also a 0-form eld strength G0 is allowed. G0 does not descend from any gauge potential and does not appear in dimensional reduction (but it does appear in string theory). We will not discuss this theory any further.

The origin of these elds can be understood from torus compactication (cf. the very rst seminar on Kaluza-Klein theory). C1 originates from the gravitational multiplet in d = 11, just as the original KK theory had a massless photon 1-form. The dilaton we are also familiar with; essentially, it parameterizes the radius of the circle. B2 and C3 descend from the A3 of the d = 11 SUGRA: B2 is the zero mode of A3 with one leg along the S 1 direction, while C3 is the zero mode of A3 with all legs in the uncompactied dimensions.

C3 , C5 , and C7 respectively. For instance, the action in the presence of a 2-brane of tension T and with world-volume W contains a term S=T
W

C3 + . . . ,

(29)

the analogue of Eq. (8).

Type IIB in d = 10

The gauge variation for C3 takes a slightly non- There is a second maximal supergravity in ten distandard form: mensions which cannot be (directly) obtained from d = 11. It contains 32 real supercharges in two C1 C1 + d0 , Majorana-Weyl spinors of the same chirality. The B2 B2 + d1 , (25) eld content of the gravitational supermultiplet is C3 C3 + d2 + 0 H3 the graviton and the gravitino, with 0,1,2 arbitrary. It is convenient to dene the a real scalar dilaton and its dilatino superinvariant eld strength partner , F4 = F4 C1 H3 . (26) a 2-form B2 with eld strength H3 = d B2 , F4 is not a closed form, but satises a non-standard Bianchi identity: dF4 = F2 H3 . With this denition the bosonic action is S= 1 22 10 1 2 410 1 2 410 d10 x g e2 R + 4()2 (27) a 0-form C0 , a 2-form C2 and a 4-form C4 with eld strengths F1 = d C0 , F3 = d C2 and F5 = d C4 . Motivated again by non-standard gauge transformation laws, it is convenient to dene the combinations F3 = F3 C0 H3 , 1 1 F5 = F5 C2 H3 + B2 F3 . 2 2 (30)

e2 H3 H3 + F2 F2 + F4 F4 B2 F4 F4 .

The eld strength combination F5 is restricted to (28) be self-dual: F5 = F5 (31) Again the Chern-Simons term in the last line is (a restriction which cannot be derived from an acgauge invariant up to a total derivative. tion, but must be imposed on the equations of moType IIA SUGRA is the low-energy limit of tion as a further constraint). The bosonic action type IIA string theory. Type IIA string theory reads contains several higher-dimensional objects which 1 couple to, and provide sources for, the higher pd10 x g e2 R + 4()2 S= 2 form gauge elds. Classically, this is analogous 210 to electrons coupling to the electromagnetic gauge 1 2 e2 H3 H3 + F1 F1 eld, which we described by integrating the gauge 410 (32) potential over the electron worldline. Type IIA + F F + F F 3 3 5 5 string theory contains D-branes with D = 0, 2, 4, 6. These are objects with D spatial dimensions and 1 2 C4 H3 F3 . D + 1-dimensional world-volumes, coupling to C1 , 410 4

Type IIB supergravity is the low-energy limit 7 Type I in d = 10 of type IIB string theory, which contains a fundamental string and various D-branes of odd D Type I is not a maximal supergravity since it only coupling to the gauge elds, similar to the even contains 16 supercharges in a single MajoranaD-branes of type IIA . Weyl spinor. It is obtained by compactifying d = 1 The action exhibits an interesting symmetry 11 supergravity on S /Z2 and adding gauge deunder modular transformations. We perform a grees of freedom. There are two multiplets. The Weyl rescaling to the Einstein frame metric G = gravitational multiplet contains e/2 g and dene the graviton and the gravitino, = C0 + ie (the axio-dilaton) and G 3 = F3 H 3 . (34) The vector multiplet contains The action Eq. (32) becomes (with the Einstein frame metric used everywhere) S= 1 22 10 1 2 2 10 d10 x G R 2(Im )2 a 1-form non-abelian gauge potential A1 in the adjoint representation of some gauge group and its gaugino superpartner. (33) the dilaton and the dilatino, a 2-form B2 with eld strength H3 .

(35) While it would seem at rst sight that this leaves much freedom to engineer theories with various 1 C4 G3 G3 vector multiplets and large gauge symmetries, it . 2 Im turns out that the theory is in fact tightly re8 i 10 stricted by anomaly cancellation. The classic result of Green and Schwarz in the 1980s shows that Recall that SL (2, R) is the group of real for gravitational anomalies to cancel, the dimen2 2 matrices with determinant one. A matrix sion of the gauge group should be 496. Cancellaa b SL (2, R) acts on the elds as tion of gauge anomalies then restricts the possible c d gauge groups to either SO(32) or E8 E8 (the cases E8 U(1)248 and U(1)496 , while not covered by this a + b , argument, are believed not to admit any UV comc + d pletion). G3 G3 , (36) c + d Both SO(32) and E8 E8 gauged supergravities have UV completions in terms of string theoF5 F5 , ries. The SO(32) can arise as the low-energy limit G G . of either type I string theory or SO(32) heterotic The action Eq. (35) is invariant under this trans- string theory. The E8 E8 case is the low-energy limit of E8 E8 heterotic string theory. formation. This is because a + b c + d Im Im . |c + d|2 = , (c + d)2

1 G3 G3 1 + F5 F5 6 Im 2

(37)

Compactication

The invariance under the subgroup SL(2, Z), where a, b, c, d are integers, lifts to S-duality in For various reasons (among others, we know type IIB string theory, which might play some role only of four non-compact dimensions), it is interin the upcoming talks. esting to look at solutions of the form MC , where 5

A trivial exact solutions to the eld equations is always given by at d = 10 or d = 11 Minkowski space with all other elds vanishing.

M is an m-dimensional non-compact manifold and C is a (10 m)- or (11 m)-dimensional compact manifold. We recapitulate some of the statements made in talk on Kaluza-Klein theory: One obtains a eld theory on M.

Together with the diculties of obtaining chiral fermions which was also mentioned in the KK theory lecture, it seems that the KK idea cannot be straightforwardly applied to 10-dimensional or 11-dimensional supergravity to produce a realistic model of the real world. String theory does however provide ways to avoid these obstacles.

The degrees of freedom are the KK modes, Because they are of some interest for the upi.e. (for bosonic elds) the eigenfunctions of coming talks (especially on moduli stabilization the covariant Laplacians 2 on C . and on AdS/CFT), we close with an interesting class of exact solutions of type IIB SUGRA on Even if the metric on C (and thus the Laplace homogeneous spaces. Take M and C to be soand Dirac operators) is not known explicitly, lutions to the d = 5 vacuum Einstein equations information about the massless spectrum can in Lorentzian and Riemannian signatures respecbe gained from topology. tively; M negatively and C positively curved. The most important examples have M =AdS5 , the If C has isometry group G, there will be a unique d = 5 Lorentzian manifold of topology R5 gauge symmetry with gauge group G. and constant negative curvature. One obtains a solution to the equations of motion provided that The last point motivates a search for a solution there is also some 5-form ux in the background. with some C whose isometry group is large enough Explicitly, with m, n . . . indices on M and , . . . to contain the Standard Model. Note however that, indices on C : upon compactication on a product space, the cur Fmnopq = r gM mnopq , vature scalar splits as follows: F = r gC , R = RM + RC + . . . (38) (40) Rmn = 4 r2 gmn , so the Einstein-Hilbert term in d dimensions becomes, after carrying out the integral over C , 1 2 2 d dd x g R R = 4 r2 g , all other elds = 0 .

In the simplest example, C is C = S 5 = (39) SO(6)/SO(5) a 5-sphere of radius r, and M =AdS5 1 with scale 1/r. This is the near-horizon geometry dm x gM (RM + ) + . . . = 2 2 m for a stack of D3 branes (recall that these indeed source the 5-form gauge eld). Another important where C RC is a cosmological constant. One example has C = T 1,1 = (SU(2) SU(2))/U(1). faces two options: While these solutions have no immediate relevance for 4d physics, they are extremely important in the Calabi-Yau compactication on manifolds C AdS/CFT correspondence. with RC = 0, but whose Riemann tensor is still non-vanishing otherwise one has just a torus compactication with too much SUSY. References These manifolds have no isometries and therefore no gauge symmetry. Dierential forms: Nakahara, Geometry, Compactications with large isometry groups, e.g. on homogeneous spaces C = G/H with G and H compact Lie groups. C then has the isometry group G. However, for small compactication radii the cosmological constant will be unacceptably large. 6 Topology and Physics SUSY in d = 4: Wess & Bagger, Supersymmetry and Supergravity, Chapter 2 Everything and more: Polchinski, String Theory Vol. II, Appendix B

Disclaimer
If you think youve found a sign error, a missing prefactor, or a gaping hole in the reasoning, youre probably correct.