Download as pdf or txt
Download as pdf or txt
You are on page 1of 36

Supergravity Notes

18th APCTP Winter Schol on Fundamental Physics, Pohang, Korea

Henning Samtleben Universit e de Lyon, Laboratoire de Physique, ENS Lyon, 46 all ee dItalie, F-69364 Lyon CEDEX 07, France henning.samtleben@ens-lyon.fr

Contents
1 Introduction 2 Remarks on symmetries and supersymmetries 2.1 Bosonic symmetries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2 Global supersymmetry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3 Local supersymmetry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Pure (N = 1) supergravity in 3.1 The vielbein formalism . . . 3.2 The Palatini action . . . . . 3.3 The supersymmetric action 3.4 Results . . . . . . . . . . . . D = 4 dimensions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 4 4 5 7 9 9 12 14 19 20 20 22 23 25 25 25 26 26 30 30 33

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

4 Matter couplings in D = 4, N = 1 supergravity 4.1 The structure of the bosonic sector . . . . . . . 4.2 N = 1 and K ahler geometry . . . . . . . . . . . 4.3 The scalar potential . . . . . . . . . . . . . . . 4.4 AdS supergravity . . . . . . . . . . . . . . . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

5 Extended supergravity in D = 4 dimensions 5.1 Multiplet structure for N > 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2 D = 4, N = 2 supergravity . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.3 Symmetric spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Supergravity in higher dimensions 6.1 Spinors in higher dimensions . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.2 Eleven-dimensional supergravity . . . . . . . . . . . . . . . . . . . . . . . . .

Introduction

There are several reasons to consider the combination of supersymmetry and gravitation. The foremost is that if supersymmetry turns out to be realized at all in nature, then it must eventually appear in the context of gravity. As is characteristic for supersymmetry, its presence is likely to improve the quantum behavior of the theory, particularly interesting in the context of gravity, a notoriously non-renormalizable theory. Indeed, in supergravity divergences are typically delayed to higher loop orders, and to date it is still not ruled out that the maximally supersymmetric extension of four-dimensional Einstein gravity might eventually be a nite theory of quantum gravity only recently very tempting indications in this direction have been unvealed. On the other hand, an underlying supersymmetric extension of general relativity provides already many interesting aspects for the theory itself, given by the intimate interplay of the two underlying symmetries. For example, the possibility of dening and satisfying BPS bounds opens a link to the treatment of solitons in general backgrounds. Last but not least of course, supergravity theories arise as the low-energy limit of string theory compactications. These lecture notes can only cover some very selected aspects of the vast eld of supergravity theories. We shall begin with pure or minimal supergravity, i.e. N = 1 supergravity in four space-time dimensions, which is minimal in the sense that it is the smallest possible supersymmetric extension of Einsteins theory of general relativity. Next, we will consider its matter couplings, and try to understand the restrictions which supersymmetry imposes on the possible construction of consistent supergravity theories. Subsequently we will study supergravity with extended supersymmetry (N > 1) and in higher dimensions (D > 4). In particular, we will discuss D = 5 supergravity which plays a distinguished role in the holographic context, providing a holographic description of aspects of particular four-dimensional gauge theories. Finally we will discuss BPS solutions of supergravity, thereby illustrating the use of the supersymmetric structure underlying matter coupled Einstein gravity for the explicit construction of solutions to the highly non-linear eld equations. There exist many introductions and reviews on the subject, of which we mention the following articles: P. van Nieuwenhuizen [1] for a general introduction, and in particular for the N = 1 case, Y. Tanii [2], especially for theories in higher dimensions and with extended supersymmetry, B. de Wit [3], in particular for the maximal supergravities, H. Samtleben [4], in particular for the gauged supergravities, and the recent book D. Freedman and A. Van Proeyen [5].

2
2.1

Remarks on symmetries and supersymmetries


Bosonic symmetries

Symmetries play a central role in our understanding of the fundamental theories. From eld theory, we are familiar with dierent types of such symmetries: Space-time symmetries are related to the geometry of space-time. E.g. relativistic eld theories are based on the Poincar e group GP = SO(1, 3) T4 , (2.1)

generated by Lorentz transformations and translations, the latter acting as (x) = (x) , (2.2)

on the elds. Elementary particles can be classied according to the irreducible representations of (2.1) which turn out to be labeled by two numbers which are the eigenvalues of the associated Casimir operators and physically correspond to the mass and the spin of the particles. Internal symmetries are related to the fact that elds may take values in some internal auxiliary space. E.g. consider a Lagrangian 1 L = i i , (2.3) 2 describing a set of scalar elds i , labelled by an index i = 1, . . . , n, such that scalars may be thought of as taking values in the vector space Rn . Obviously, this Lagrangian is invariant under orthogonal rotations i = i j j , so(n) . (2.4)

Particular interactions, such as L = V (|i i |), preserve this SO(n) symmetry. With constant symmetry parameter , these are global symmetries. More fundamental are local symmetries, in which the symmetry parameters themselves are functions of the space-time coordinates. Obviously, the Lagrangian (2.3) is no longer invariant under (2.4) if depends on the coordinates, but there is a well-known construction to render it invariant by introducing a gauge eld with minimal couplings. More precisely, with covariant derivatives dened by D i i A i j j , the covariantized Lagrangian 1 D i D i , 2 is invariant under the combined gauge transformation L = i = i j j , A i j = i j A i k k j + i k A k j = D i j , (2.6) (2.5)

(2.7)

of scalar and vector elds. Adding a standard Yang-Mills term F F to capture the dynamics of the vector elds gives rise to a consistent SO(n)-invariant theory. This is a standard construction of electrodynamics (for n = 2) or non-abelian Yang-Mills theory, based on the gauging of an underlying global symmetry group Ggauge . 4

2.2

Global supersymmetry

A natural fundamental question in view of the dierent symmetry groups above is the following: could the space-time symmetries and the internal gauge symmetries of a given model be embedded in some bigger unifying group GP Ggauge Gunifying , (2.8)

The answer is negative as stated by the famous theorem of Coleman and Mandula [6]. The notable exception to this no-go theorem has been analysed in Haag, Lopuszanski and Sohnius [7] upon inclusion of fermionic generators in the underlying algebra. The associated transformations are supersymmetries. Let us study an example. Consider the free Lagrangian of two scalar elds S , P , and a Majorana spinor L0 = 1 1 / . S S + P P i 2 2 (2.9)

Before getting into any details, this is probably a good place to pause and review our spinor conventions.

We use mostly minus signature, with = diag(+, , , ) with gamma matrices { , } = 2 . The combination 5 i 0 1 2 3 , (2.11) (2.10)

squares to the identity matrix and can be used to dene chiral (Weyl) spinors as the eigenvec1 tors of the projectors P 2 (I 5 ). Rather, we will use four-component Majorana spinors, dened by imposing the reality condition
! = T C ,

(2.12)

dened as 0 , and the charge conjugation matrix with the Dirac conjugated spinor C satisfying CC T = I = CC , and relating the gamma matrices to their transposed C 1 C = . (2.14) (2.13)

, , etc., transform covariantly under the Lorentz Spinor bilinears transform, such as group, while the reality condition (2.12) implies the relations = = = , , , 5 = 5 = 5 , 5 , etc. .

(2.15)

Coming back to the Lagrangian (2.3), we can now realize that it is invariant under a particular symmetry relating bosonic and fermionic elds, acting as S = i , P = 5 , 1 /(S i 5 P ) , = 2

(2.16)

where the symmetry parameter is a Majorana spinor itself. Invariance of (2.3) follows by a simple calculation: variation of the rst two terms after partial integration gives rise to S (i) P 5 , whereas variation of the third term yields / / S i 5 P i S i 5 P = i , (2.18) (2.17)

and precisely cancels (2.17). This is the rst and simplest example of a supersymmetry that we meet. From it, we can already infer some characteristic features of supersymmetry. Supersymmetry transforms elds of dierent spins and dierent statistics into each other. In this example, the supermultiplet is made from a complex scalar S +iP and a Majorana spinor , satisfying the Klein-Gordon and the Dirac equation, respectively. The number of bosonic degrees of freedom equals the number of fermionic degrees of freedom. This is a general property of supersymmetric theories.
3 , Mass dimensions of the scalar and spinor elds in D = 4 are [S ] = 1 and [] = 2 respectively. From (2.16) we thus read o that the supersymmetry parameter comes with mass dimension [ ] = 1 2.

The commutator of two supersymmetry transformations can easily be evaluated e.g. on the scalar eld S / [ 1 , 2 ] S = i2 S
1

S ,

(2.19)

and yields a translation with parameter = i2 1 , which indeed is among the bosonic symmetries of the Lagrangian (2.3). We can conclude in general, that supersymmetric theories are necessarily translation invariant. In terms of the associated supergenerators Q , the algebra (2.19) reads ] = ( ) P , [Q , Q with the generator P of translations. In general, interactions break the supersymmetry, there are however non-trivial interactions that can be added to (2.3) preserving invariance under supersymmetry upon simultaneous modication of the supersymmetry transformation rules (2.16). Consider Yukawa and scalar interactions Lint = ig 2 S + i 5 P + S + P2 2 8 6
2

(2.20)

(2.21)

it can be shown that the Lagrangian L0 + Lint is still supersymmetric upon proper modication of the supersymmetry transformation rules (2.16), if the two coupling constants g and are related as = g2 . (2.22)

Interestingly, at this value also the quantum behaviour of the theory changes drastically: the leading divergence in the one-loop contribution to the scalar propagator comes with a factor ( g 2 ) and thus drops out in the supersymmetric case (2.22). This is a generic feature of supersymmetric theories: divergencies are milder (or sometimes even absent!) due to cancellations between contributions from bosonic and fermionic origins.

2.3

Local supersymmetry

Let us now allow the supersymmetry parameter to depend on space-time coordinates, i.e. consider local supersymmetry instead of the global one introduced above. Upon generalising (x) , (2.23)

the commutator of two supersymmetry transformations again leads to a transformation of the form (2.19), / [ 1 , 2 ] S = i2 S
1

S ,

(2.24)

with the dierence that now the resulting translation parameter via (2.23) also depends on the space-time coordinates. The r.h.s. of (2.24) thus represents the action of an innitesimal local dieomorphisms on space-time. We conclude that a locally supersymmetric theory will necessarily be dieomorphism invariant and thus requires to treat the space-time metric as a dynamical object. The simplest theory which achieves this is Einsteins general relativity which we shall consider in the following although higher order curvature corrections may equally give rise to supersymmetric extensions. The reverse is also true: coupling of a theory with rigid supersymmetry to gravity inevitably leads to a theory with local supersymmetry. The easiest way to observe this is by considering the commutators of local Lorentz transformations and supersymmetry. Since the parameter transforms as a spinor under the Lorentz group, this commutator leads to local supersymmetry transformations. The dynamical space-time metric is described by the spin-2 state appearing in the supergravity multiplets. Consider e.g. the supergravity multiplet in N = 2 supersymmetry, which contains a spin-2 state g giving rise to the graviton, together with a vector-spinor 3 , and a spin-1 gauge eld A . If supersymmetry is unbroken, these of spin 2 denoted states will be degenerate in mass. The fact that under extended supersymmetry bosonic elds of dierent spin join in a single multiplet is quite remarkable. It shows that from a supercovariant point of view their respective interactions are truly unied: we may thus be able to recover the way gravity works from our knowledge of gauge theory by exploiting the underlying supersymmetric structure. We can now try to construct the extension of (2.3) that is invariant under a local extension of (2.16) in precise analogy to the construction of gauge theory (2.6) by introduction of covariant derivatives. Even though for supersymmetry, this construction is far more complicated and eventually leads to the introduction of an innite number of additional couplings, 7

it is quite instructive to work out in detail the rst steps. Let us rst reconsider the above supersymmetry calculation and see what goes wrong with the invariance of (2.3) under (2.16) when is coordinate-dependent. Whereas the contributions from (2.17) can still be brought into this form, the terms from (2.18) now give an extra contribution due to the fact that the derivative also hits the supersymmetry parameter : ( ) . / S i 5 P ( ) J L0 = i (2.25)

Indeed, we can identify J with the (spinor-valued) Noether current associated with the global symmetry (2.16) of the Lagrangian. Just as in (2.6), this contribution was cancelled by introduction of a gauge eld A and extension of the Lagrangian by J A , the form of (2.25) suggest to introduce a spinorial gauge eld transforming as = 1 , (2.26)

with a Noether coupling to the Lagrangian according to = i / S i 5 P . LNoether = J (2.27)

The dimensionalful coupling (with [] = 1) has been introduced in (2.26) in order to make up for the dierence of mass dimension in a fermion and the supersymmetry parameter. The eld is a space-time vector whose entries are 4-component Majorana spinors, i.e. it 3 representation of represents a 44 matrix. Counting of representation shows that it is a spin- 2 the Poincar e group. This eld is called the gravitino and naturally arises in the supergravity multiplets. However, unlike the case of gauge theory (2.6), the construction does not stop here. The -variation of (2.27) correctly cancels the unwanted contribution from (2.26), however the -variation of (2.27) leads to new unwanted terms given by LNoether i / S + i 5 P / S i 5 P . 2 (2.28)

Collecting the terms quadratic in S and in P , we identify i ( S S + P P ) = i (T (S ) + T (T )) , 2 (2.29)

where we have used the gamma-matrix identity1


( ) ( ) = 2 g ,

(2.30)

and expressed the result in terms of the energy-momentum tensor of a free scalar eld T (S ) S S 1 g S S . 2 (2.31)

In order to cancel the new variation (2.29), one is thus obliged to introduce another new eld of index structure h = h( ) with supersymmetry variation h
1

= i i ,

(2.32)

Here, and in the following, symmetrizations and antisymmetrizations are indicated by round and square brackets, respectively, and are always normalised to a total weight of 1, i.e. X A Y B = X [A Y B ] + X (A Y B ) , etc..

and coupling to the Lagrangian as LNoether,2 = h (T (S ) + T (T )) . 2 (2.33)

Putting together (2.3) and (2.33), we recognise the rst terms in the expansion of the Lagrangian for scalar elds on a curved background L = 1 2 |g | g ( S S + P P ) , (2.34)

upon expansion of the metric into g = + h + O(2 ) , = 1 |g | g = h + h + O(2 ) 2 (2.35)

Continuing this procedure to all orders would eventually reproduce the innite series corresponding to (2.34) with taking the role of the gravitational constant. In accordance with the general arguments from the supersymmetry algebra, we thus observe explicitly that local supersymmetry implies the coupling of the elds to a dynamical metric, whose dynamics in turn is described by the Einstein eld equations. Rather than going through this lengthy procedure we will in the following make use of our knowledge about Einstein gravity and its covariant matter couplings in order to construct locally supersymmetric theories. We will rst construct the minimal supergravity theory in four dimensions, i.e. the theory describing only the graviton and the associated gravitino which together form the smallest N = 1 supergravity multiplet. Our task shall be to construct 0 we recover a locally supersymmetric theory with this eld content, such that by setting Einsteins general relativity. I.e. we will work out the Lagrangian L = LEH + L[ ] , (2.36)

which consists of a part describing the graviton by the Einstein-Hilbert action, and another part describing its coupling to the gravitino.

Pure (N = 1) supergravity in D = 4 dimensions

In this section, we will construct the minimal supersymmetric extension of standard Einstein gravity in four space-time dimensions. In order to describe the coupling of spinor elds to gravity, we rst need to review the vielbein formalism of general relativity.

3.1

The vielbein formalism

We shall work in units where = = c = 1. The standard formulation of general relativity includes the metric tensor g together with the Levi-Civita connection , dening covariant derivatives of tensors, as e.g. in the case of a vector eld X = X X , (3.1)

implying that X transforms as a tensor under dieomorphisms. Further requiring that the metric is covariantly constant g = 0 , (3.2) 9

denes the Christoel symbols as


1 ( g + g g ) + K , = 2 g

(3.3)

where K is the so-called contorsion, dened in terms of a torsion tensor T = T [ ] as K = 1 2 (T + T + T ). The torsion tensor is not xed by the metricity condition (3.2) and usually set to zero, which amounts to having symmetric Christoel symbols = ( ) . Supergravity in its simplest formulation however exhibits a nontrivial torsion bilinear in the fermionic elds, as we shall see. The Riemann tensor can be expressed in terms of the Christoel symbols; it is given by
R = 2[ ] + 2[| | ] ,

(3.4)

where A[| B|] := 1 2 (A B A B ). From this one computes the Ricci tensor R = R , and the Ricci scalar R = g R . The Einstein-Hilbert action is given by the Lagrangian density LEH = 1 4 |g |R , (3.5)

where g denotes the determinant of the metric, and the eld equations are the vacuum Einstein equations 1 R 2 Rg = 0 . (3.6) In order to deal with spinors, whose transformation rules are dicult to generalise to curved backgrounds, it is helpful to reformulate this theory in the so-called vielbein formalism. The idea behind this is to consider a set of coordinates that is locally inertial, so that one can apply the usual Lorentz behaviour of spinors, and to nd a way to translate back to the original coordinate frame. To be a bit more precise, let y a (x0 ; x), a = 0, . . . , 3, denote a coordinate frame that is inertial at the space-time point x0 . We shall call these the Lorentz coordinates. Then y a (x0 ; x) ea (x0 ) := , x x=x0 compose the so-called vielbein, or tetrad in our case where a takes four dierent values. Under general coordinate transformations x x , the vielbein transforms covariantly, i.e. e a (x ) = x a e (x) , x (3.7)

while a Lorentz transformation y a y a = y b ba leads to e a (x) = eb (x)ba . In particular, the space-time metric can be expressed as g (x) = ea (x)e b (x) ab , 10 (3.8)

in terms of the Minkowski metric ab = diag(1, 1, 1, 1). The vielbein ea thus takes lower Lorentz (or at) indices a, b to lower indices in the coordinate basis (or curved) indices , . Its inverse ea performs the transformation in the other direction. As an example consider a 1-form with coordinates X , for which X = ea Xa , Xa = ea X .

On the other hand, contravariant indices are transformed as X = X a ea , X a = X ea .

We can now introduce spinors (x) in our theory, which transform as scalars under the general space-time coordinate transformations, but in a spinor representation R of the local Lorentz group: (x) = R((x)) (x) . In addition, we can convert the constant matrices of the inertial frame into matrices in the curved frame by the action of the vielbein: (x) = ea (x)a = { , } = 2g .

The choice of the locally inertial frame y a is of course only unique up to Lorentz transformations, i.e. the tetrad is only dened up to a rotation in the Lorentz indices ea eb ba . The algebra so(1, 3) of the local Lorentz transformation group SO(1, 3) is given by the generators Mab , which are antisymmetric in their indices and satisfy the commutation relations [Mab , Mcd ] = 2a[c Md]b 2b[c Md]a . The action of Mab on vectors is given by Mab X c = 2[ca Xb] , while on spinors it is Mab = 1 2 ab , where ab stands for [a b] and we have dropped the explicit spinor indices , as we will usually do from now on. On the other hand, the space-time metric g stays invariant under so(1, 3) transformations, so that these will automatically preserve the Einstein-Hilbert action. In order to couple matter to our theory, we will also need the Lorentz covariant derivatives. They are dened by
ab D + 1 2 Mab .

(3.9)

(3.10)

The ab are a set of gauge elds; there is one for each generator of so(1, 3), correspondingly they are antisymmetric in the Lorentz indices a, b. Together, they dene an object called the spin connection . It is determined by requiring that the tetrad is covariantly constant (the so-called vielbein postulate):2
a a a b a 0 = D e a e = e + b e e .

In our notation, here and in the following the covariant derivative D = D( ) always includes the spin connection, but not the Christoel symbols.

11

An equivalent way to write this is 1 a D[ e ] = T a . 2 (3.11)

This constitutes a set of algebraic equations for which has a unique solution expressing the spin connection in terms of the vielbein and the torsion. For vanishing torsion, the equations reduce to a D[ e ] = 0 , (3.12) of which the solution is given by 1 ab = ab [e] = ec abc bca cab , 2 (3.13)

where abc = ea eb ( ec ec ) are the so-called objects of anholonomity. In the presence of torsion, the solution of (3.11) is given by ab = ab [e] + K a b , where the contorsion K a b has been expressed in terms of T a after equation (3.3) above. The spin connection also denes a curvature tensor with coecients R ab [ ] = 2[ ]
ab

+ 2[ ]c .

ac

(3.14)

It can be shown that this curvature is in fact nothing but the ordinary space-time curvature where two space-time indices have been converted to Lorentz indices, i.e. R ab = R ea eb .

3.2

The Palatini action

We shall now apply the vielbein formalism to the formulation of general relativity. This is done by rewriting the Einstein-Hilbert Lagrangian (3.5) directly in terms of vielbein quantities:
1 LEH [e] = 4 ab |g |R = 1 4 |e| ea eb R .

(3.15)

Note that |e| stands for the determinant of the tetrad on the right hand side, while on the left hand side the argument e is a short hand notation to denote dependence of L on the full tetrad ea . The equations of motion for this Lagrangian are the Einstein equations (3.6). In the vielbein formalism, these second order equations can be expressed quite nicely in terms of a set of two equations of rst order. This goes under the name of Palatini (or rst order) formulation. It is achieved by considering the spin connection to be a priori independent of the tetrad, described by the Palatini Lagrangian
1 LP [e, ] = 4 |e| ea eb R ab [ ] .

(3.16)

There are now two eld equations: The rst one comes from the variation with respect to the vielbein, LP /e, and the second from the variation with respect to the spin connection, LP / . Explicitly, the variation yields 1 1 LP = |e| Ra ea R 2 2 3 ea |e| (D e a ) e[a eb ec] bc , 2 (3.17)

12

as we will show in a moment. Variation w.r.t. ab thus gives precisely (3.12) and is solved by (3.13). Upon inserting this solution = [e] into the rst equation from (3.17), we recover the Einstein equations (3.6). In the classical theory, the Palatini formulation is hence equivalent to the standard second order formulation of general relativity. It is instructive to note that the relation LEH [e] = LP [e, ]|=[e] between the EinsteinHilbert and the Palatini action provides a very simple way to also compute the variation of the former. Namely LEH [e] = LP ea
= [e]

LP bc

= [e]

bc ea

ea .

(3.18)

However, the second term vanishes identically, because LP / is proportional to (3.12) and thus zero for = [e]. I.e. the variation of LEH is simply given by the rst term of (3.17) and thus by Einsteins equations (3.6). A similar observation will greatly facilitate the computation of a supersymmetric extension of Einstein gravity in the following.

Variation of the Palatini action. Let us derive in detail the variation of the Palatini action (3.17). We rst observe that the Palatini action (3.16), can be recast in the form LP [e, ] = To see this, rst note that
abcd ef cd c d ab 1 abcd e e R [ ] 16

(3.19)

= 4

[a [e

b] f] ,

(3.20)

where one can determine the numerical factor by tracing over the index pairs (a, e) and (b, f ), on both sides. The tensor density e is obtained from eabcd by

abcd

ea eb ec ed |e| ,

(3.21)

where the determinant shows up since we are dealing with a tensor density rather than with a tensor (as a consequence is just a number). Plugging (3.21) into (3.19) and making use of (3.20) readily yields (3.16). The equivalent form (3.19) is a convenient starting point for calculating its variation with respect to the spin connection. Since only shows up in the expression for the Riemann tensor, we are interested in R ab , for which we have R ab = 2D[ ] . This can easily be seen, since according to (3.14), R ab starts with the non-covariant term ab 2[ ] , followed by a term of the form , which must complete the non-covariant rst part in order to yield a covariant quantity in total. We therefore nd LP =
c d ab 1 abcd e e 2D , 16 ab

where we dropped the antisymmetric bracket since metrisation. Integration by parts leads us to LP = 1 4
c abcd e

already takes care of the antisym-

D ed ab ,

13

where we made use of the antisymmetrisation by the tensor densities again. Using the identity abcd e c = 6|e|e[a eb ed] (which follows in complete analogy to (3.21)), we nd 3 LP = 2 |e| D e d e[a eb ed] ab . On the other hand, variation of the Palatini action w.r.t. the tetrad gives rise to
ab LP = 1 4 R [ ] (|e| ea eb ) 1 |e| R ab [ ] (2ea eb ea eb (ec ec )) = 4 b a a 1 a 1 a = 1 = 1 2 |e| R [ ] ( b 2 eb e ) ea 2 |e| (R [ ] 2 e R) ea , (3.22)

where we have used that (detA) = (detA) Tr[A1 A] for any matrix A. Together, we arrive at (3.17) for the variation of the Palatini action.

3.3

The supersymmetric action

We now come back to the reason why we introduced the vielbein formalism, which was the inclusion of spinor elds and in particular of the gravitino in our theory. While LEH (or LP ) can be considered as the kinetic term for gravity, we should now add a kinetic term for the . Such a term has in fact been known since the 1940ies, when Rarita and gauge eld Schwinger published a proposition for the kinetic term of a spin- 3 2 eld [8]: LRS =
1 5 D 2

(3.23)

Originally, Rarita and Schwinger wrote this equation in the context of quantum eld theory on at Minkowski space. With (3.23) we have already done the necessary to covariantly couple this eld to a dynamical metric. In particular, we have used the Lorentz covariant derivative D from (3.10), and introduced curved (coordinate-dependent) gamma matrices according to = e a a , such that { a , b } = ab { , } = , (3.25) (3.24)

with the at Minkowski metric ab and the curved dynamical metric g .3 Just adding LRS to the Einstein-Hilbert Lagrangian will however not yet yield a supersymmetric theory. In order to guarantee supersymmetry, we must also allow further interaction terms which contain higher powers of . Our ansatz should thus be a Lagrangian of the form L[e, ] = LEH [e] + LRS [e, ] + L4 [e, ] ,
3

(3.26)

As a note of warning, when we discussed global supersymmetry in section 2, we used coordinates x for at Minkowski space. With hindsight, we should have denoted coordinates, the constant gamma-matrices, etc. by at indices a, b, . . . . It is only now that we have coupled the theory to gravity, that the dierence becomes important.

14

where the last part only contains terms of order at least 4 in the gravitino. However, to determine the actual form of L we must rst understand how the elds transform under a supersymmetry variation. Our strategy shall be the following. We will rst present the transformation laws for the elds under a supersymmetry transformation, and verify them by checking the supersymmetry algebra to lowest order in the fermions. After that, we shall come back to the Lagrangian (3.26) and determine its precise form such that it is supersymmetric. To lowest order in the fermions, the supersymmetry transformation laws of the vielbein and the gravitino are given by ea = i a , = D . (3.27) At least heuristically, they seem of a good form, as the bosonic vielbein transforms into its presumed superpartner and as which is supposed to play the role of a gauge eld of local supersymmetry indeed transforms as D , the derivative of the local parameter. More precisely, expanding the metric around Minkowski space, these transformation laws precisely reproduce what we found in the perturbative Noether analysis of (2.26), (2.32). To establish the proposed transformations we will have to check that they satisfy the supersymmetry algebra. In particular, the commutator of two such supersymmetry transformations should give rise to a suitable combination of local symmetry transformations of the theory, in a way similar to (2.24). Consider the commutator of two supersymmetry transformations on the vielbein, [ 1 , 2 ]ea = 2[ 1 2 ] ea = 2 (i1 a ) 1 (i2 a ) = i1 a D = =
a i 2 D (2 1 1 2 ) iD (2 a 1 ) , 2 a

i2 a D

a of Majorana spinors , in the last line. where we have used the property a = In order to understand this result, dene the space-time vector eld := i2 1 , such that [ 1 , 2 ]ea = D ( e a ) = D e a + e a . (3.28) In the last equation we have used that D is the covariant derivative with respect to only the Lorentz indices, which means that it operates as a mere space-time derivative on . Taking into account (3.12), we can exchange the lower indices in the rst term of the last line in (3.28), such that by writing out the covariant derivative we are nally left with [ 1 , 2 ]ea = ea + e a + ab eb .
I II

(3.29)

We recognise part I as being the innitesimal action of a dieomorphism on ea , and part II as an innitesimal internal Lorentz transformation ea with ab = ab . We have thus shown that the transformations (3.27) close onto the vielbein as [ 1 , 2 ]ea = + , 15 (3.30)

into a combination of local symmetries: dieomorphism and Lorentz transformation. Proceeding, we should next apply the commutator of two supersymmetries to the gravitino. This however involves already fermionic terms of higher order (e.g. variation of the connection term in D 2 in 2 gives rise to terms in (1 2 ) which are of the same fermion order as the variation of possible higher order terms in 2 that we are neglecting for the moment). Furthermore, it turns out that closure of the supersymmetry algebra on the fermionic elds will require these elds to obey their rst order equations of motion meaning that the supersymmetry algebra only closes on-shell. This involves the exact form of the Lagrangian, so let us rst turn to the Lagrangian; we shall come back to this point in a short comment in section 3.4. We hence go back to the Lagrangian (3.26), and try to determine its constituents such that it becomes supersymmetric. To do this, we will make use of a trick and formally include the higher powers in the fermionic elds L4 into the rst two parts, without changing their general form. This can be achieved by properly adapting the spin connection. Our starting point is the Lagrangian L0 LEH + LRS , (3.31) which depends on the vielbein and the gravitino eld. As we have done in the case of the Palatini action, we may write this in the explicit form
ab L0 [e, , [e]] = 1 4 |e| ea eb R [ [e]] + 1 5 D 2

(3.32)

with [e] from (3.13). Computing the variation of this Lagrangian under supersymmetry, we will thus nd schematically L0 = L0 e

= [e]

e +

L0

= [e ]

L0

= [e ]

[e] e . e

(3.33)

It would greatly simplify the calculation, if we could drop the last term by the same argument used at the end of section 3.2, using the fact that L0 / was identically zero for = [e]. Unfortunately, this is no longer true. As L0 / receives additional contributions from the Rarita-Schwinger term, it follows that variation of L0 w.r.t. the spin connection gives rise to the equation a i a , (3.34) D[ e ] = 2 instead of (3.12). As we have seen for (3.11) above, the solution of (3.34) is given by the modied spin connection ab = ab [e, ] = ab [e] + K a b ,

with

[a b] + 1 a b ) , K a b = i( 2

(3.35)

which depends quadratically on the gravitino . Our ansatz for the supergravity Lagrangian will thus be
1 L = L0 [e, , [e, ]] = 4 |e| ea eb R ab [ ] + 1 5 D 2

(3.36)

= D( where is given by (3.35). We also use the notation D ) to indicate that the Lorentz connection in this covariant derivative is built by . The Lagrangian (3.36) thus diers from

16

(3.32) by quartic terms in . Correspondingly, we modify the supersymmetry transformation rules (3.27) to ea = i a , , = D

(3.37)

including additional cubic terms in the variation of the gravitino. We will now show that (3.36) already gives the full supersymmetric Lagrangian to all orders in the fermions. Schematically, the variation of (3.36) is given by L = L0 e
= [e, ]

e +

L0

= [e, ]

L0

= [e, ]

[e, ] [e, ] e + ,(3.38) e

and now the entire last term vanishes, as L0 / is proportional to (3.34) and thus identically zero for = [e, ]. We can thus in the following computation consistently neglect all contributions due to explicit variation w.r.t. the spin connection. Using (3.17), we nd that variation of the Einstein-Hilbert term then gives rise to
1 a 1 ea R ( ea ) LEH = 2 |e| R 2 i a 1 e aR a , = 2 |e| R 2

(3.39)

a = R[ where R ] a denotes the Ricci tensor obtained by contracting the curvature (3.14) of the modied spin connection . RS while again neglecting the terms from variation w.r.t. On the other hand, varying L the spin connection yields
1 LRS = ( 2

5 D ) .

1 2

5 D ( )5 D 5 D + D + D

Since the Lagrangian has to stay invariant only up to a total derivative, we can shift the in the rst term to the right by partial integration. Hence derivative D LRS =
1 2

5 D ( )5 D [ D ] 5 D [ D ] (D [ ] )5 D + (3.40)

+ (total derivatives) ,

where we have used the possibility to antisymmetrise in various indices in the bracket, as they are contracted with the totally antisymmetric tensor density. In the third term in = a D e a is quadratic the bracket, the covariant derivative of the gamma matrix D [ ] [ ] in the fermions due to (3.34). Together with the two fermionic elds and , this term is thus at least quartic in the fermions. Likewise, the last term in (3.40) is at least quartic in (4) , D ] = the fermions. Denoting these two quartic terms by LRS and making use of [D 1 ab 4 R ab , we nd
1 LRS = 16

5 ab 5 ab +

ab + (4) LRS , R

(3.41)

where we have relabeled a few of the indices in the rst terms. We observe that 5 ab = ab 5 = ab 5 + 2e [a b] 5 , 17

and since ab = ie d

dabc

and (5 )2 = 1 1, we have
dabc c

5 ab = ie d

+ 2e [a b] 5 .

c for Plugging this back into the variation (3.41) and keeping in mind that c = Majorana spinors, we are thus led to
i d e LRS = 8 dabc

ab + (4) LRS 1 c R 8

e [a b] 5 )R ab . (e [a b] 5 +

5 a for Majorana spinors, the last two terms in this expression Since we also have 5 a = mutually cancel. If we furthermore make use of the identity e d dabc = 6|e|e[a eb ec] , we nally obtain
i d LRS = 8 e dabc

ab = c R = =

6i 8 3i 4 i 2

ab + (4) LRS |e|e[a eb ec] ( c )R ] + (4) LRS |e|( [ )R a 1 ea R + (4) LRS . |e|( a ) R 2

The rst term precisely cancels the variation of the Einstein-Hilbert part (3.39). We have thus shown that up to total derivatives the variation of the full Lagrangian L from (3.26) reduces to the last two terms of (3.40) L = (4) LRS =
1 2

( )5 D [ ] ) 5 D + (D

(3.42)

quartic in the fermions. With (3.34) we nd (4) LRS =


i 2 1 a 2 ( )(a 5 D )

a 5 D )( a ) . (

(3.43)

A priori, the form of the two terms is rather dierent. In order to bring them into the same form, we will need some so-called Fierz identities: = 1 ( )I 1 ( 5 )5 1 ( a ) a + 1 ( a 5 ) a 5 + 1 ( ab ) ab , (3.44) )( 4 4 4 4 8 where the l.h.s. is to be understood as a matrix to be contracted with further spinors from the left and the right. (The easiest way to prove this identity is to note that the 16 matrices {I, 5 , a , a 5 , ab } B form a basis of the 4 4 matrices which is orthogonal with respect = ), and to Tr(AB ).) Applying (3.44) to the second term of (3.43) (with = D [ ] for and noting that due to the symmetry properties of Majorana spinors the bilinear B is non-vanishing only for = a and = ab , we nd
i 2

a 5 D )( a ) = i ( 4
i 8

a 5 b 5 a )( b 5 D ) ( ) . ( a 5 bc 5 a )( bc 5 D

Using that a b a = 2b and a bc a = 0, the second term vanishes while the rst term precisely cancels the rst term of (3.43). Together, we have thus shown that the Lagrangian (3.36) is invariant up to total derivatives under the supersymetry transformations (3.37). It constitutes the sought-after minimal supersymmetric extension of Einstein gravity in four dimensions. 18

3.4

Results

Let us collect and complete our results, and simplify the notation. The full supersymmetric Lagrangian is given by L = L0 [e, , [e, ]] from (3.36). By properly adapting the spin connection we got rid of all explicit quartic fermion interaction terms. Equation (3.34) corresponds precisely to the equation L0 / = 0 and is identically solved by given in equation (3.35) as (schematically) = [e, ] = [e] + K , depending on the tetrad and the gravitino. Equivalently, the full Lagrangian L can be brought into the form (3.26) in which the rst terms depend on the standard spin connection [e], and the quartic interaction terms are explicit. They can in fact be retrieved rather conveniently by formally expanding L0 [e, , ] around = . Since L0 is no more than quadratic in , this expansion stops after two terms; we nd (again schematically) L0 [e, , ] = L0 [e, , ]
=

L0 [e, , ]

+
=

1 2 K L0 [e, , ] 2

.
=

Now the term linear in K vanishes, because L0 / = 0 at = , as we have already used above. Moreover, the rst term on the right-hand side is precisely the full supersymmetric Lagrangian (3.36). The last term nally has only contributions from the term in the curvature R ab [ ] in the Einstein-Hilbert term. Together, this yields for the full supersymmetric Lagrangian ac b abc Kcab , (3.45) L = L0 [e, , [e]] 1 4 |e| Ka Kb c + K where Kabc given in (3.35) explicitly describes the quartic fermion interactions. One can see that the fermions are subject to an interaction of current-current type. Let us mention two more things before we conclude this chapter. The rst one is that with the full supersymmetry rules (3.37) ea = i a , , = D (3.46)

containing cubic fermion terms in the variation of the gravitino, the computation of the supersymmetry algebra leading to (3.30) needs to be revisited. Extending our previous calculation, the commutator of two supersymmetry transformations on the tetrad can be shown to be ( e a ) [ 1 , 2 ]ea = D a , = ea + e a + ab eb + i for = i2 1 as in (3.29), where the last term is due to the non-vanishing torsion (3.34). Again, the result is given by a combination of local transformations; the rst two terms being an innitesimal dieomorphism, the third term an internal Lorentz transformation (now with ab = ab ), while the last term represents an additional supersymmetry transformation with parameter = . If on the other hand one applies the commutator of supersymmetry transformations on the gravitino, one will nd that the only way to interpret the result as a combination of local transformations involves the application of the equations of motion 19

of the gravitino, i.e. the supersymmetry algebra closes only on-shell (some details for the computation can be found in [2]). This is a generic feature of supersymmetric theories. Finally, a remark concerning nomenclature: the straightforward although very tedious way to prove supersymmetry of the full Lagrangian would amount to starting from its explicit form (3.45) and check its invariance under (3.46) including lengthy terms up to order 5 due to the explicit dependence of the spin connection on the tetrad. Historically, this is how fourdimensional supergravity was rst constructed by Freedman, van Nieuwenhuizen and Ferrara in 1976 [9]. Nowadays, this is referred to as the second order formalism. Soon after, Deser and Zumino proved supersymmetry by using an analogue of the Palatini formalism, treating the spin connection as an independent eld. This rst order formalism simplies the computation but requires to nd the correct supersymmetry transformation law for the independent spin connection. The method outlined in these lectures is the most common used these days and goes under the name of the 1.5th order formalism. It combines features of both of the older approaches. Morally, we stay second order in the sense that only tetrad and gravitino are independent elds in the Lagrangian. From the rst order formalism we borrow however the observation that the equation obtained by formal variation of the Lagrangian L/ vanishes identically for = . As we have seen, this greatly simplies the computation of the higher order fermion terms.

Matter couplings in D = 4, N = 1 supergravity

In the last chapter, we have discussed how to construct minimal N = 1 supergravity in D = 4 dimensions, i.e. how to construct the supersymmetric couplings for the N = 1 supergravity ). The aim of this section is to sketch how we can implement further matter multiplet (g , in N = 1 supergravity (for details see [10, 11, 5]). Besides the supergravity multiplet, we may consider chiral multiplets (i , i ), consisting of a spin-1/2 Weyl fermion i and a complex spin-0 scalar i for each i, and vector multiplets (M , AM ) containing one spin-1 gauge eld M M A and a spin-1/2 Majorana fermion for each M . Both are representations of the N = 1 supersymmetry algebra and contain four degrees of freedom.

4.1

The structure of the bosonic sector

Let us rst discuss the general form of the couplings of the bosonic sector in four-dimensional supergravity, i.e. the interactions of the metric g , the scalar elds i and the (abelian) gauge elds AM . Restricting the dynamics to two-derivative terms, these couplings are rather constrained by dieomorphism and gauge invariance which allows for the following terms: the gravitational sector, described by the Einstein-Hilbert term 1 LEH = |e|R , 4 which was extensively discussed in the last chapter. the scalar sector 1 Lscalar = |e| g j k Gjk (i ) |e| V(i ) , 2 (4.2) (4.1)

where Gjk (i ) is the metric on the scalar target space and V(i ) is a scalar potential. 20

the vector sector 1 1 Lvector = |e| g g F F I (i ) 4 8


F F R (i ) ,

(4.3)

described by a standard Maxwell term and a topological term (which does not depend A A . The matrices on the metric) in terms of the abelian eld strength F i i I ( ) and R ( ) may depend on the scalar elds. If R is constant, the second term is a total derivative and can be considered as a generalization of the usual instanton term. Minimal couplings between scalar elds and vector elds can be introduced upon gauging part of the global symmetries of the scalar Lagrangian (4.2), i.e. introducing nontrivial gauge charges and covariant derivatives according to D = A t , (4.4)

with the generators t of the gauge algebra acting on the scalar elds via Killing vector elds of the scalar target space
i t i k () .

(4.5)

Indeed, it is a straightforward calculation to check that (4.5) is a symmetry of the i is a Killing vector, i.e. satises kinetic part of (4.2), provided k i k j + j k i = 0 Lk Gij = 0 . (4.6)

The gauge algebra is given by the algebra of Killing vectors


i i i k j j k k j j k = f k ,

(4.7)

in (4.3) are with structure constants f . Accordingly, the abelian eld strengths F then covariantized to the standard Yang-Mills form F A A f A A .

(4.8)

In supersymmetric theories, the introduction of minimal couplings in the vector/scalar sector also induces corrections in the fermionic sector, modication of the supersymmetry transformation rules, and furthermore a contribution to the scalar potential. This procedure of assigning non-trivial gauge charges to elds in the matter sector is usually referred to as gauging of the theory. The data that need to be specied in order to x the bosonic Lagrangian thus are the scalar dependent matrices Gjk (i ), I (i ), R (i ), the potential V(i ) and the minimal couplings (4.5). The fermionic couplings of the Lagrangian are then entirely determined by supersymmetry, without any freedom left. In turn, the bosonic data cannot be chosen arbitrarily, but are constrained by supersymmetry, i.e. by demanding that the bosonic Lagrangian LEH + Lscalar + Lvector can be completed by fermionic couplings to a Lagrangian invariant under local supersymmetry. The construction of matter coupled supergravities is facilitated by powerful techniques such as superspace techniques or the superconformal tensor calculus. We will not enter in 21

the details of these approaches, but rather give a brief summary of some of the important results on the resulting structures of matter couplings. As we shall describe in the next section, N = 1 supersymmetry requires the scalar target space to be a K ahler manifold, and the combination N R + i I of the vector kinetic matrices to be a holomorphic function of the complex scalar elds i . Also the scalar potential is of very particular and restricted form. Within extended supergravity, i.e. theories with N > 1 supercharges, the generic structure of the Lagrangian remains the same. However, the conditions on Gjk (i ), IM N (i ), and RM N (i ) become much more restrictive. We come back to this in section 5 below.

4.2

N = 1 and K ahler geometry

The essential restriction from N = 1 supersymmetry on the scalar geometry states that the scalar target space should be a complex K ahler manifold. A good working denition of such }, the metric is a manifold is the following: for a particular set of complex coordinates {i , ) as obtained as the second derivative of a real function, the so-called K ahler potential K (, Gij = 0 = G , Gi = i K (, ) . (4.9)

As an immediate consequence, the Christoel symbols and the Riemann tensor have only particular non-vanishing components, given by k ij ,
k ,

Rk l .

(4.10)

The same geometrical structure is required for the possible scalar couplings in globally supersymmetric theories.4 In that case, the most natural way of understanding the appearance of the K ahler potential comes from the construction of supersymmetric actions via integrals over full superspace of functions of chiral superelds S = d4 x , d4 K (, ) (4.11)

i whose integrand gives rise to terms like i K (, ) upon Taylor expansion on the fermionic coordinates . A mathematically more satisfying (coordinate-free) denition of a K ahler manifold simply 2 requires the existence of a covariantly complex structure J = I, and hermitean metric. In the coordinates of (4.9), the complex structure takes the form

Ji j

j = i i ,

J = i .

(4.12)

Let us nally observe, that the form of (4.9) shows that the target space metric is invariant under the following change of the K ahler potential ) K (, ) + f () + f ( ) , K (, for any holomorphic function f , referred to as K ahler transformations. Let us consider two examples of K ahler manifolds of complex dimension n:
Strictly speaking, local supersymmetry imposes an additional global condition on the geometry, the manifold should be K ahler-Hodge [12].
4

(4.13)

22

The homogenous K ahler potential


) = ||2 i K (,

Gi = ij ,

(4.14)

simply describes at space Cn . The K ahler potential ) = 2 ln 1 + ||2 K (, = Gi = 2 1 + ||2 ij


i 1 + ||2

, (4.15)

describes the Fubini-Study metric on complex projective space CPn = SU(n + 1)/U(n) . This is an Einstein space with Ri = 2(n + 1) Gi 2 (4.16)

Let us nally note, that N = 1 supersymmetry requires the couplings in the vector sector (4.3) to be compatible with the complex geometry of the scalar target space in the sense that the combination ) + i I (, ) , N R (, must be a holomorphic function on the K ahler manifold N = 0 . (4.18) (4.17)

4.3

The scalar potential

Apart from the kinetic part dened by the K ahler geometry we have discussed, the scalar sector (4.2) carries the scalar potential V (, ) that we shall describe now. In N = 1 supergravity, the potential has two dierent contributions ) = VF (, ) + VD (, ) , V (, (4.19)

referred to as F -term and D-term contributions, respectively. We describe them separately. The F -term contribution to the scalar potential is dened in terms of a (freely chosen) holomorphic function W (), the so-called superpotential. The potential is of the form
2 ) = eK Gi VF (, Di W D , W 3|W |

(4.20)

with the K ahler covariant derivatives Di W = i W + i K W . (4.21)

We note, that this is an indenite potential, implying that the resulting supergravity theories can in principle support AdS, dS, and Minkowski geometries. In contrast, recall that in globally supersymmetric theories, the scalar potential is always positive denite. In particular, a constant superpotential W = (in absence of scalar eld elds) denes an extension of N = 1 supergravity by a negative cosmological constant. Let us also note, that the introduction of 23

a superpotential also gives rise to additional Yukawa couplings in the fermionic sector and modied fermionic transformation rules, such as ,L = D
L

i eK/2 W 2

+ ... .

(4.22)

Here, it turns out to be convenient to dene the chiral projections


L

1 I + 5 2

1 I 5 2

etc.,

(4.23)

which map Majorana into Weyl spinors, which naturally couple to the fermions of the chiral multiplets. The D-term contribution to the scalar potential have a very dierent origin, they arise from assigning non-trivial gauge charges to the scalar elds according to the general discussion after (4.4). In order to realize the action of a non-trivial gauge group Gg on the scalar elds, this group has to be embedded into the isometry group of the scalar target space. N = 1 supergravity imposes further constraints, in order to allow for supersymmetric couplings, the generators of the gauge group must be identied with holomorphic Killing vectors on the K ahler manifold, i.e. Killing vectors that also preserve the complex structure (4.12). In explicit complex coordinates, this means that the Killing vector condition (4.6) i kj + j ki = 0 , is amended by further demanding that
i k = 0

i k + ki = 0 ,

(4.24)

i k = 0,

(4.25)

= 0 = where in the last step we have used that ik . We note, that (4.25) in fact implies ik the rst equation of (4.24), while the second equation of (4.24) implies that locally the Killing vector can be expressed as the gradient of a real function

ki = ii P ,

(4.26)

the so-called associated moment map. In term of P , the conditions (4.24), (4.25) reduce to the equation ) = 0 . i j P (, (4.27)

To realize a non-trivial gauge group, we will have to associate a holomorphic Killing vector i , i.e. a moment map P to every generator t of the gauge group. We note that the k moment maps are dened by (4.26) up to constant shifts P P + c . (4.28)

There is a canonical choice of moment maps which transform in the adjoint representation of the gauge algebra:
j i i i k j k + k k ( ) = f k ,

(4.29)

24

with the structure constants from (4.7). This xes the shift (4.28) except for those P associated to abelian factors in the gauge group. The remaining free parameters c are compatible with supersymmetry and gauge symmetry and referred to as Fayet-Iliopoulos parameters. They can be useful in order to shape the form of the scalar potential. Having introduced the minimal couplings (4.4) further implies corrections in the fermionic sector and the supersymmetry transformation rules in order to preserve supersymmetry. Eventually, supersymmetry even requires a further modication of the bosonic sector by adding the Dterm contribution VD = 1 I 1 2

P P ,

(4.30)

to the scalar potential. The moment maps appearing in the potential are those subject to (4.29). Note that unlike the F -term contribution (4.20) to the potential, the D-term contribution (4.30) is positive denite. What we have seen in this discussion is a structure typical in supergravity. Introduction of non-trivial gauge charges induces additional couplings in the fermionic sector which in turn imply a scalar potential in the bosonic sector. This is referred to as gauging supergravity, see [4] for a review. For N > 1 supergravities and in higher dimensions D > 4 this is the only known mechanism which allows the introduction of a scalar potential. In particular, there is no analogue of the F -terms (4.20) and a superpotential. In contrast to the D-term potential (4.30) however, in these cases the potential induced by gauging is generically indenite, i.e. allows for AdS and dS solutions.

4.4

AdS supergravity

This is probably the simplest extension of pure D = 4, N = 1 supergravity, by introducing a (negative!) cosmological constant to the action. See [13].

5
5.1

Extended supergravity in D = 4 dimensions


Multiplet structure for N > 1

In the last section we have discussed the couplings of chiral and vector multiplets to minimal N = 1 supergravity. In principle, there is another N = 1 multiplet that one may consider, 3 the so-called gravitino multiplet which consists of a spin- 2 eld and a spin-1 eld A . Its coupling is slightly more subtle due to a simple reason: just as the two degrees of freedom of the massless vector are described by a standard U(1) gauge theory, the additional massless gravitino must be described by introducing yet another local supersymmetry. In other words, coupling the gravitino multiplet to the N = 1 supergravity multiplet gives rise to a theory, which has two local supersymmetries and is thus referred to as N = 2 supergravity [14]. Accordingly, multiplets now organize under the N = 2 supersymmetry algebra in D = 4 dimensions. The N = 2 supergravity multiplet comprises the two N = 1 multiplets mentioned 3 above and thus includes one spin-2 eld, two spin- 2 elds and a spin-1 eld A . Analogously to the discussion of the last section, this theory may be enlarged by further coupling matter in N = 2 multiplets.

25

N =2

3 2

1 2

0 1 2 1 6 70

1 2

1 N =4 N =8 1 8 1 28 4 56

2 1 4 4 56

1 1 1 6 1 28

3 2

2 1

4 8

1 1

sugra vector hyper sugra vector sugra

Table 1: The various multiplets for N = 2, 4, 8 supersymmetry. denotes the helicity of the states, i.e. their charge under the SO(2) little group in D = 4 dimensions. Upon coupling more gravitino multiplets one arrives at theories with larger extended supersymmetry, whose structures and eld contents are more and more constrained. The particle content of the most common multiplets with extended supersymmetry (N = 2, 4, 8) are summarized in table 1. Note that while we have listed here the irreducible representations of the super-Poincar e group, we must add the CPT conjugates with ipped helicities to obtain the physical eld content. The only CPT self-conjugate multiplets are the N = 4 vector multiplet and the N = 8 supergravity multiplet. In addition there are multiplets for other values of N , involving N = 3, N = 5 or N = 6 supercharges. We restrict the discussion to N 8 multiplets since representations of supersymmetry algebras with N > 8 supercharges contain states with helicities larger than two. No consistent interacting theories involving a nite number of such elds are known. As mentioned above, the constraints on supergravity theories with extended supersymmetry are very severe. For N = 2, the scalar geometry is of special K ahler and quaternionic K ahler type for vector and hyper-multiplets, respectively. For N 3, the scalar geometry is necessarily a symmetric space, which we discuss subsequently.

5.2

D = 4, N = 2 supergravity

In N = 2 supergravity, scalar elds have their origin from nV vector multiplets and nH hypermultiplets, respectively, c.f. table 1. Accordingly, the scalar target space has the form of a direct product MSK MQK with the vector multiplet scalars describing a (2nV )-dimensional Special K ahler manifold, and the hyper-scalars describing a (4nH )-dimensional quaternionic manifold. We refer to [15, 16, 5].

5.3

Symmetric spaces

As mentioned above, for N 3, the scalar geometry in four dimensions is necessarily a symmetric space, i.e. the scalars are described as coordinates of a coset space G/K. More precisely, the scalar target space of N = 4 supergravity has to be of the form SO(6, n) SL(2) , SO(6) SO(n) SO(2) (5.1)

with some positive integer n. The dimension of this manifold is 6n +2 and thus coincides with the number of scalar elds expected for n vector multiplets and one supergravity multiplet (including the CPT conjugated multiplet) with N = 4 supercharges, c.f. table 1. 26

The (ungauged) N = 8 supergravity in D = 4 dimensions is unique and its eld content can only consist of a single supergravity multiplet. The corresponding scalar target space is given by the coset-space5 E7(7) . (5.2) SU(8) Because of dim E7(7) = 133 and dim SU(8) = 63, the target space dimension is 70 in agreement with the number of scalars in the theory, c.f. table 1. In this section, we shall discuss the structure of coset spaces G/K, which also describe the target spaces of higher-dimensional extended supergravities. In general G is the global symmetry group of the theory, and K is its maximal compact subgroup. A convenient formulation of this sigma-model has the scalar elds parametrize a G-valued matrix V (evaluated in some fundamental representation of G) and makes use of the left-invariant current J = V 1 V g Lie G . In order to accommodate the coset space structure, J is decomposed according to J = Q + P , Q k , P p , (5.4) (5.3)

where k Lie K and p denotes its complement, i.e. g = k p, orthogonal w.r.t. the CartanKilling form. The scalar Lagrangian is given by
1 e Tr (P P ) . Lscalar = 2

(5.5)

It is invariant under global G and local K transformations acting as V = V V k (x) , g , k (x) k , (5.6)

on the scalar matrix V . Under these symmetries the currents Q and P transform according to Q = k + [ k, Q ] , P = [ k, P ] , (5.7)

showing that the composite connection Q behaves as a gauge eld under K. As such it plays the role of a connection in the covariant derivatives of the fermion elds which transform linearly under the local K symmetry, e.g.
i D

1 i i k ab ab (Q )k i , 4

(5.8)

i , etc. Likewise, one denes D V V Q = P for the scalar for the gravitinos matrix V . The current P on the other hand transforms in a linear representation of K, builds the K-invariant kinetic term (5.5) and may be used to construct K-invariant fermionic interaction terms in the Lagrangian. The two symmetries (5.6) extend to the entire supergravity eld content and play a crucial role in establishing the full supersymmetric action. They will furthermore be of vital importance in organizing the construction of the gauged theories described in the next section.

By E7(7) we denote here a particular real form of the complex exceptional group E7 , whose compact part is given by SU(8).

27

The global g transformations may be expanded as = t into a basis of generators t satisfying standard Lie-algebra commutation relations [t , t ] = f t , (5.9)

with structure constants f . The local K symmetry is not a gauge symmetry associated with propagating gauge elds (the role of the gauge eld is played by the composite connection Q ), but simply takes care of the redundancy in parametrizing the coset space G/K. It is indispensable for the description of the fermionic sector, with the fermionic elds transforming in linear representations under K. In particular, the scalar matrix V transforming as (5.6) can be employed to describe couplings between bosonic and fermionic elds, transforming under G and K, respectively. To make this more explicit, it is useful to express (5.6) in indices as VM N = (t )M K VK N VM K kK N , (5.10)

with G-generators (t )M K and the underlined indices K, N referring to their transformation behavior under the subgroup K. The matrix VM N allows to construct couplings of e.g. a M transforming in the associated fundamental representation of G to bosonic eld strength F the fermionic elds according to (schematically) )N , F M VM N ( etc. , (5.11)

)N denotes the projection of the fermionic bilinear onto some K-subrepresentation where ( in the corresponding tensor product of K-representations. It is often convenient to x the local K symmetry by adopting a particular form of the matrix V , i.e. choosing a particular set of coset representatives. In this case, any global G-transformation in (5.6) needs to be accompanied by a compensating K-transformation V = V V k , (5.12)

where k depends on (and on V ) in order to restore the particular gauge choice, i.e. to preserve the chosen set of coset representatives. This denes a non-linear representation of G on the (dim G dim K) coordinates of the coset space, i.e. on the physical scalar elds. Likewise, it provides a non-linear realization of the group G on the fermion elds via the compensating transformation k . Two prominent gauge xings are the following: unitary gauge: in which the matrix V is taken of the form V = exp {a Ya } , (5.13)

where the non-compact generators Ya span the space p. In this gauge, the a transform in a linear representation of K G, thus global K-invariance of the Lagrangian remains a Y a takes the form P a = a + . . . , where dots refer to manifest. The current P = P higher order contributions. This shows that the kinetic term (5.5) is manifestly ghostfree with Gab () ab + . . . . It is here that the importance of K being the maximal compact subgroup of G shows up.

28

triangular gauge: in which the matrix V is taken of the form V = exp {m Nm } exp h , (5.14)

where = 1, . . . , rank G, labels a set of Cartan generators h of g and the Nm form a set of nilpotent generators such that the algebra spanned by {h , Nm } constitutes a Borel subalgebra of g. With suitable choice of the Borel subalgebra, it is in this gauge that a possible higher-dimensional origin of the theories becomes the most transparent. The grading associated with the chosen Borel subalgebra is related to the charges of the elds under rescaling of the volume of the internal compactication manifold. See [17] for a systematic discussion for the maximal theories and their eleven-dimensional origin. E.g. as mentioned above, the scalar sector of the maximal (N = 8) supergravity in D = 4 space-time dimensions is described by the coset space E7(7) /SU(8). This theory can be obtained by dimensionally reducing D = 11 supergravity on a torus T 7 . The elevendimensional origin of the elds can be identied in the triangular gauge associated with the GL(7) grading of E7(7) . A type IIB origin of the elds on the other hand is identied in the triangular gauge associated with a particular GL(6) SL(2) grading. For the half-maximal (N = 4) supergravity in D = 4 dimensions, the scalar sector is described by the coset space G/K = SL(2)/SO(2) SO(6, 6+ n) (SO(6) SO(6+ n)) , (5.15)

c.f. (5.1). This theory can be obtained by dimensional reduction of the ten-dimensional type-I theory with n vector multiplets, upon reduction on a torus T 6 . The ten-dimensional origin of the elds is identied in the triangular gauge associated with the GL(6) grading of SO(6, 6). In order to construct the full supersymmetric action of the theory, it is most convenient to keep the local K gauge freedom. When discussing only the bosonic sector of the theory, it is often functional to formulate the theory in terms of manifestly K-invariant objects. E.g. the scalar elds can equivalently be described in terms of the positive denite symmetric scalar matrix M dened by M V VT , (5.16)

where is a constant K-invariant positive denite matrix (e.g. for the coset space SL(N )/SO(N ), with V in the fundamental representation, is simply the identity matrix). The matrix M is manifestly K-invariant and transforms under G as M = M + M T , while the Lagrangian (5.5) takes the form Lscalar =
1 8

(5.17)

Tr ( M M1 ) .

(5.18)

To nish this section, let us evaluate the general formulas for the simplest non-trivial coset-space SL(2)/SO(2) which appears in the matter sector of several supergravity theories. With the sl(2) generators given by h= 1 0 0 1 , e= 0 1 0 0 29 , f= 0 0 1 0 , (5.19)

the matrix V in triangular gauge (5.14) is given as V = eC e e h = 1 C 0 1 e 0 0 e . (5.20)

Evaluating the non-linear realization (5.12) of SL(2) on these coset coordinates C , leads to h = 1 , h C = 2C , e C = 1 , f = C , f C = e4 C 2 . (5.21)

This shows that h acts as a scaling symmetry on the elds, whereas e acts as a shift symmetry on C , and f is realized non-linearly. This toy example exhibits already all the generic features of the global G-symmetries that we will meet in more generality in section ?? below. The matrix M for this model can be computed from (5.16) with = I2 , and is most compactly expressed in terms of a complex scalar eld = C + ie2 , giving rise to M = 1 | |2 1 , (5.22)

while the kinetic term (5.5) takes the form


1 4 e1 Lscalar = 4 e C C =

1 . 4( )2

(5.23)

It is manifestly invariant under the scaling and shift symmetries of (5.21), whereas invariance under the non-linear action of f is not obvious and sometimes referred to as a hidden symmetry. In terms of , the action of a nite SL(2) group transformation can be given in the compact form a + b , c + d for exp() = a b c d SL(2) . (5.24)

Supergravity in higher dimensions

The motivation to study supergravity theories in dimensions D > 4, and their dimensional reduction is two-fold. On the one hand supergravity appears as the low energy eective action for fundamental theories such as string theories, which generically live in higher dimensions. In this top-down approach one has to consider the dimensional reduction of these theories in order to derive from the fundamental theories a meaningful four-dimensional theories supposed to describe our universe. On the other hand from a purely four-dimensional point of view, the dimensional reduction of higher-dimensional supergravities can be considered as a powerful technique in order to construct extended D = 4 supergravity theories.

6.1

Spinors in higher dimensions

In order to discuss extended supergravities we have to know the possible types of spinors in higher space-time dimensions, see e.g. [18] for a reference. In this section we consider

30

D-dimensional Minkowski space-time with signature = diag(+1, 1, . . . , 1). The spinorial representation of the Lorentz group can be obtained from a representation of the Ddimensional Cliord algebra associated with the metric ab . The generators of this algebra, called Gamma matrices a , satisfy the anticommutation relations {a , b } = 2 ab 1 . (6.1)

The spinorial representation of the D-dimensional Lorentz algebra SO(1, D 1) can then be constructed using antisymmetric products of gamma matrices; it is straightforward to verify that the Cliord algebra (6.1) implies that the generators 1 Mab ab , 2 ab = [a b] , (6.2)

satisy the Lorentz algebra (3.9). Finite Lorentz transformations are given by exponentiation R() = exp 1 ab Mab 2 .

It can be shown that in even dimensions, the Cliord algebra has a unique (up to similarity transformations) irreducible representation CD which has 2[D/2] complex dimensions. Its elements are referred to as Dirac spinors. An explicit construction can be given recursively: denoting by j the Gamma-matrices in (D 2)-dimensional space-time, the ensemble i 0 j j 0 , i 0 1 1 0 , i

0 0 1

(6.3)

satises the algebra (6.1) in D-dimensional space-time. An important object is the following combination of Gamma matrices
+1 = iD 2 0 1 D1 ,

(6.4)

which generalizes the 5 in D = 4 dimensions. The Cliord algebra implies that it satises relations , a } = 0 , { This has two important consequences: , a ) of matrices denes a representation of the Cliord algebra in The ensemble ( D + 1 dimensions. This shows that in also odd dimensions, the Cliord algebra (6.1) has an irreducible representation CD+1 CD of 2[D/2] complex dimensions. In fact, in odd dimensions there are two inequivalent representations CD+1 , CD+1 of the Cliord , a a , which can not be achieved by a similarity algebra, related by transformation. Via (6.2) this denes a (unique and irreducible) 2[D/2] -dimensional representation of the Lorentz group in odd dimensions. commutes with all generators of the Lorentz alge In even dimensions, the operator bra: [, Mab ] = 0. This implies that as a representation of the Lorentz algebra, the 2 = 1 . (6.5)

31

D 2 3 4 5 6 7 8 9 10 11

pM

MW

dimension (real) 1 2 4 8 8 16 16 16 16 32

Table 2: Possible types of spinors in D-dimensional Minkowski space-time. Weyl, Majorana, pseudo-Majorana and Majorana-Weyl are denoted by W, M, pM and MW respectively. representation CD is reducible and may be decomposed into its irreducible parts by virtue of the projectors P 1 2 . 1 i (6.6)

The corresponding 2[D/2]1 -dimensional chiral spinors are called Weyl spinors. Let us now consider the reality properties of the spinors. It follows from (6.1) that also the complex conjugated Gamma matrices a fulll the same Cliord Algebra {a , b } = 2ab 1 . Because of the uniqueness of the representations of the Cliord algebra discussed above, this implies that a has to be related to a (up to a sign) by a similarity transformation
1 a = B a B .

This suggests to impose on the spinors on of the following reality conditions = B ; and = B+ ,

which are referred to as Majorana and pseudo-Majorana, respectively. Obviously, these conditions only make sense provided that B B = 1 , which only holds in certain space-time dimensions. A closer analysis shows that (pseudo) Majorana spinors exits in dimensions D 0, 1, 2, 3, 4 mod 8. They possess half of the degrees of freedom of a Dirac spinor. In even dimensions, we may further ask if these reality conditions are compatible with the chiral split (6.6) into Weyl spinors. It turns out that this is only the case in dimensions D 2 mod 8. All possible types of spinors in D 11 dimensional Minkowski space-time are given in table 2. Based on these structures, supergravity theories can be constructed in dimensions 32

D 11 10

spinors M MW

N 1 (1,1) (2,0) (1,0)

pM

2 1 2 1 4 2 (2,2) (2,1) (1,1) (2,0) (1,0)

pM

sM

sMW

e a , 2 + , C , 2 B , 2 , 2 e a , + , B , , A , + e a , 2 , C , 2 B , 3 A , 4 , 3 e a , , B , A , , A , , e a , 2 , C , 3 B , 6 A , 6 , 7 e a , , B , 2 A , , A , , 2 e a , 4 , 5 B , 10 A , 16 , 14 e a , 2 , B , 3 A , 2 , A , , 3 e a , 4 + , 4 , 5 B , 16 A , 20 + , 20 , 25
(+) ()

elds e a , , B e a , + , , C , B , A , + , ,
(+)

nB 128 128 128 64 8 128 56 8 128 48 8 128 40 8 128 64 32 8 24 8 12 4 4 4 128 64 24 8 8 4 4 128 64 32 16 8 8 8 4 4 4 2 2 2

e a , 4 + , 2 , 5 B , B , 8 A , 10 + , 4 , 5 e a , 2 + , 2 , B , 4 A , 2 + , 2 , A , 2 + , 2 , 4 e a , 4 + , 5 B ( ) B , 4 , 5
(+) (+)

spM

8 6 4 2

8 6 5 4 3 2

e a , 2 + , B A , 2 + ( ) B , 2 , 2 , 4 e a , 8 , 27 A , 48 , 42 e a , 6 , 15 A , 20 , 14 e a , 4 , 6 A , 4 , A , 4 , 5 e a , 2 , A A , 2 , 2 , 4 e a , 8 , 28 A , 56 , 70 e a , 6 , 16 A , 26 , 30 e a , 5 , 10 A , 11 , 10 e a , 4 , 6 A , 4 , 2 A , 4 , 6 e a , 3 , 3 A , A , 4 , 6 e a , 2 , A A , 2 , 2 , 4 e a , A , , 2

Table 1: Supermultiplets in dierent space-time dimensions.

Table 3: Supermultiplets in dierent space-time dimensions. The red background indicates the supergravity multiplets. The last column gives the number of bosonic degrees of freedom.

D 11. Similar to what we have seen in four dimensions, an analysis of the representations of the supersymmetry algebras in the various dimensions shows that apart from the supergravity multiplet for a given number of supercharges, in general there exist also matter multiplets, that can consistently be coupled to supergravity. We give a table of the various multiplets in table 3. The highest-dimensional supergravity theory is the unique eleven-dimensional theory, constructed 1978 by Cremmer, Julia and Scherk [19], which we shall discuss in some more detail now.

6.2

Eleven-dimensional supergravity

The highest-dimensional supergravity theory lives in eleven space-time dimensions and was constructed by Cremmer, Julia and Scherk [19]. Its eld content is given by the lowest massless representation of the supersymmetry algebra. For massless states which fulll P P = 0

33

we can set without loss of generality P0 = P10 and all other components to zero. The supersymmetry algebra then turns into
0 {Q , Q } = 2 P

= 2P0 1 + 10 0

(6.7)

where the Q are the 32 independent real supercharges in D = 11 dimensional Minkowski space-time. As the r.h.s. of (6.7) describes a projector of half-maximal rank in spinor space, it follows that only 16 out of the 32 supercharges act non-trivially on massless states and satisfy the Cliord algebra (6.1). From the discussion of the previous section, we then know that there is a 28 = 256-dimensional irreducible representation C16 of this algebra which then gives the eld content of the eleven-dimensional theory. Under SO(16), this representation decomposes into 128s + 128c chiral spinors (6.6). It is important to stress that in contrast to the discussion of the last section, the 16 here has no meaning as a number of space-time dimensions, but just denotes the number of supercharges; we are just exploiting the fact, that the same abstract algebra (6.1) appears in a rather dierent context again. In order to nd the space-time interpretation of the 256 physical states, we note that the 16 supercharges form the physical degrees of freedom of a space-time spinor in eleven dimensions, i.e. they build an irreducible 16-dimensional representation of the little group SO(9) under which massless states are organized in eleven dimensions. We thus consider the embedding SO(9) SO(16) of the little group into SO(16), under which the relevant representations decompose as [20] 16 16 , 128s 44 84 , 128c 128 . (6.8)

What is the meaning of these SO(9) representations? Recall, that a massless vector in eleven dimensions has 9 degrees of freedom and transforms in the fundamental representation of SO(9). The 44 corresponds to the symmetric traceless product of two vectors: these are the degrees of freedom of a massless spin-2 eld, the graviton. The 84 on the other hand describes the three-fold completely antisymmetric tensor product of vectors: 9 3 = 84 . The bosonic eld content of eleven-dimensional supergravity thus is given by the metric g and a totally anti-symmetric three-form eld A . The fermionic eld content of the theory is the irreducible 128 of SO(9) which is the Gamma-traceless vector-spinor 9 16 16 = 128. Accordingly, this describes the degrees of freedom of a massless spin-3/2 eld, the gravitino. The full eleven-dimensional supergravity multiplet thus is given by (g , , A ). A similar argument shows why supergravity theories do not exist beyond eleven dimensions: repeating the same analysis for say a twelve-dimensional space-time yields a minimal eld content that includes elds with spin larger than two. No consistent interacting theory for such elds can be constructed. The action of eleven-dimensional supergravity is given by the eleven-dimensional analogue of (3.45) 1 1 L0 [e, ] = |e| R |e| D , 4 4 (6.9)

34

supplemented by a kinetic, a topological term and a fermionic interaction term for the threeform eld A 1 2 LA [A, e, ] = |e|F F F F A 96 6912 2 + 12 F , |e| (6.10) 384 with the abelian eld strength F = 4[ A] , and quartic fermion terms. Most other supergravities can be obtained by Kaluza-Klein reduction from this theory.

References
[1] P. Van Nieuwenhuizen, Supergravity, Phys.Rept. 68 (1981) 189398. [2] Y. Tanii, Introduction to supergravities in diverse dimensions, hep-th/9802138. [3] B. de Wit, Supergravity, in Unity from Duality: Gravity, Gauge Theory and Strings (C. Bachas, A. Bilal, F. David, M. Douglas, and N. Nekrasov, eds.), Springer, 2003. hep-th/0212245. [4] H. Samtleben, Lectures on gauged supergravity and ux compactications, Class.Quant.Grav. 25 (2008) 214002, [0808.4076]. [5] D. Z. Freedman and A. Van Proeyen, Supergravity. Cambridge University Press, 2012. [6] S. R. Coleman and J. Mandula, All possible symmetries of the S matrix, Phys.Rev. 159 (1967) 12511256. [7] R. Haag, J. T. Lopuszanski, and M. Sohnius, All possible generators of supersymmetries of the S matrix, Nucl.Phys. B88 (1975) 257. [8] W. Rarita and J. Schwinger, On a theory of particles with half integral spin, Phys.Rev. 60 (1941) 61. [9] D. Z. Freedman, P. van Nieuwenhuizen, and S. Ferrara, Progress Toward a Theory of Supergravity, Phys. Rev. D13 (1976) 32143218. [10] E. Cremmer, S. Ferrara, L. Girardello, and A. Van Proeyen, Yang-Mills theories with local supersymmetry: Lagrangian, transformation laws and superHiggs eect, Nucl. Phys. B212 (1983) 413. [11] J. Wess and J. Bagger, Supersymmetry and supergravity. Princeton University Press, 1992. [12] E. Witten and J. Bagger, Quantization of Newtons constant in certain supergravity theories, Phys.Lett. B115 (1982) 202.

35

[13] P. Townsend, Cosmological constant in supergravity, Phys.Rev. D15 (1977) 28022804. [14] S. Ferrara and P. van Nieuwenhuizen, Consistent supergravity with complex spin 3/2 gauge elds, Phys.Rev.Lett. 37 (1976) 1669. [15] J. Bagger and E. Witten, Matter couplings in N = 2 supergravity, Nucl. Phys. B222 (1983) 1. [16] B. Craps, F. Roose, W. Troost, and A. Van Proeyen, What is special K ahler geometry?, Nuclear Phys. B 503 (1997), no. 3 565613. [17] E. Cremmer, B. Julia, H. Lu, and C. N. Pope, Dualisation of dualities. I, Nucl. Phys. B523 (1998) 73144, [hep-th/9710119]. [18] J. A. Strathdee, Extended Poincar e supersymmetry, Int. J. Mod. Phys. A2 (1987) 273. [19] E. Cremmer, B. Julia, and J. Scherk, Supergravity theory in 11 dimensions, Phys. Lett. B76 (1978) 409412. [20] R. Slansky, Group theory for unied model building, Phys. Rept. 79 (1981) 1128.

36

You might also like