Simple Supergravity

Gianguido Dall’Agata and Marco Zagermann

arXiv:2212.10044v2 [hep-th] 15 Feb 2023

Abstract We present a short overview of the structure and couplings of supergrav-

ity theories at the component level. We do so with as little technical machinery
as possible, working directly with the physical on-shell fields and using explicit
computations and geometrical reasoning to arrive at the result, highlighting the new
properties of supersymmetry in the context of a gravitational theory.

1 Introduction

Supergravity is soon going to turn 50 [2, 3]. During this half of a century it lived
several lives and it has been used and studied from various different vantage points,
including the analysis of quantum gravity and black hole physics, string theory, par-
ticle physics phenomenology, cosmology and mathematics. Each of these different
approaches advanced our understanding of the features and structure of supergravity
theories and, in turn, supergravity brought new ideas and fertilized each of these
fields of study.
One of the main problems of the uninitiated who is interested in supergravity is
that it is a rather technical subject and most of the introductory books and reviews deal
with it by first emphasizing some specific mathematical formalism (like superspace
and superfields or the group manifold approach, or the superconformal approach) and
only after some significant effort by the reader they enter into the discussion of the
physical properties of the theory. While each of these approaches has its advantages
and can be at some point necessary to obtain significant progress, we felt that a
more simple hands on introduction, where every aspect is dealt with directly at the
component level, emphasizing the physical features and their mathematical origin,

Gianguido Dall’Agata
Dip. di Fisica e Astronomia “Galileo Galilei”, Università di Padova and INFN, Sezione di Padova,
Via Marzolo 8, 35131 Padova, Italy,
Marco Zagermann

1
2 Gianguido Dall’Agata and Marco Zagermann

was missing. For this reason, while there are already several great reviews that use
one of the aforementioned techniques (see for instance [4, 5, 6, 7, 8, 9, 10, 11, 12, 13]),
we worked on a new physics first introduction to the subject, which took shape in
the lecture notes [1].
The current chapter is a short redacted excerpt of the more detailed and complete
presentation given in [1]. Here we mainly focus on the very basic ingredients that
are needed for a first introduction to the subject, which we hope will work as an
invitation for the reader to deepen their knowledge of the subject.
We also stress that in this collection the reader is also going to find short intro-
ductions to some of the alternative approaches mentioned above, as well as various
applications.

2 What is supergravity?

Depending on which aspect one wants to emphasize, one could define supergravity
theories in three different ways:
1. Supergravity theories are supersymmetric field theories with gravity where the
supersymmetry transformations act nontrivially also on the gravitational field.
2. Supergravity theories are supersymmetric field theories in which supersymmetry
is realized not only as a global (rigid) symmetry but as a local (gauge) symmetry.
3. Supergravity theories are field theories 1 with consistently interacting spin-3/2
fields 2. In supergravity, these spin-3/2 fields are called gravitino fields (or gravi-
tini).
In this section, we illustrate why these three apparently different characterizations
describe essentially the same class of theories. We do so in the assumption that gravity
is described by Einstein’s general theory of relativity, so that supergravity actions
consist of the Einstein–Hilbert term plus a restricted class of matter actions coupled
to gravity. To this end, we revisit the simplest globally supersymmetric field theory
in four dimensions, the free massless Wess–Zumino model for one chiral multiplet,
and discuss how this theory has to be changed when supersymmetry is turned into
a local symmetry, following Definition 2. As we will see, making supersymmetry
local by a simple iterative procedure (the “Noether method”) directly exhibits the
need for the gravitino field (cf. Definition 3) and its superpartner, the graviton (cf.
Definition 1), and suggests the supersymmetry transformation laws of these fields.
We end this section with a discussion of some basic properties of the gravitino field.

1 We assume finitely many fields and couplings as well as consistency with unitarity and diverse
spacetime backgrounds.
2 For the sake of readability, we do not distinguish carefully here between spin and helicity, i.e. “spin
𝑠” should be understood as “helicity ±𝑠” in the massless case.
Simple Supergravity 3

2.1 Promoting supersymmetry to a local symmetry3

Consider the free massless Wess–Zumino model for one chiral multiplet (𝜙, 𝜒),
where 𝜙(𝑥) is a complex scalar and 𝜒(𝑥) a Majorana spinor field4 with Lagrangian


L = −𝜕𝜇 𝜙𝜕 𝜇 𝜙∗ − 𝜒 𝑅 𝜕/ 𝜒𝐿 + 𝜒 𝐿 𝜕/ 𝜒𝑅 . (1)

We recall that the mass dimensions of these fields are 𝐷 [𝜙] = 1, and 𝐷 [ 𝜒] = 3/2.
The Lagrangian (1) is invariant up to a total derivative under supersymmetric
under the variations5

𝛿 𝜖 𝜙 = 𝜖 𝐿 𝜒𝐿 ⇐⇒ 𝛿 𝜖 𝜙∗ = 𝜖 𝑅 𝜒𝑅 (2)
1 1
𝛿 𝜖 𝜒𝐿 = 𝜕/ 𝜙𝜖 𝑅 ⇐⇒ 𝛿 𝜖 𝜒𝑅 = 𝜕/ 𝜙∗ 𝜖 𝐿 . (3)
2 2
Note that 𝐷 [𝜖] = −1/2. In our conventions, (3) is equivalent to
1 1
𝛿 𝜖 𝜒 𝐿 = − 𝜖 𝑅 𝜕/𝜙 ⇐⇒ 𝛿 𝜖 𝜒 𝑅 = − 𝜖 𝐿 𝜕/ 𝜙∗ . (4)
2 2
To check this explicitly, we write the fermionic term of the Lagrangian (1) as
Lfer = −𝜒 𝑅 𝜕/ 𝜒𝐿 + 𝜕𝜇 ( 𝜒 𝑅 )𝛾 𝜇 𝜒𝐿 and trace the terms involving 𝜖 𝐿 , which come only
from the variation of 𝜙 and 𝜒 𝑅 , because those proportional to 𝜖 𝑅 follow by hermitean
conjugation:

𝛿L = −𝜕𝜇 (𝛿𝜙)𝜕 𝜇 𝜙∗ − 𝛿 𝜒 𝑅 𝜕/ 𝜒𝐿 + 𝜕𝜇 (𝛿 𝜒 𝑅 )𝛾 𝜇 𝜒𝐿 + h.c. (5)

Integrating by parts the first and the second term gives, using (2), (3) and (186),
 
𝛿L = 𝛿𝜙𝜙∗ + 2𝜕𝜇 (𝛿 𝜒 𝑅 )𝛾 𝜇 𝜒𝐿 + 𝜕𝜇 − 𝛿𝜙𝜕 𝜇 𝜙∗ − 𝛿 𝜒 𝑅 𝛾 𝜇 𝜒𝐿 +h.c.
| {z }
≡K 𝜇
= −𝜕𝜇 (𝜖 𝐿 ) 𝜕/𝜙∗ 𝛾 𝜇 𝜒𝐿 + 𝜕𝜇 K 𝜇 + h.c. (6)

3 This section is reprinted from [1] © 2021 Springer-Verlag GmbH Germany, part of Springer
Nature. Reproduced with permissions. All rights reserved.
4 Throughout this article we use anti-commuting four-component Majorana spinors to describe
fermionic degrees of freedom. Our conventions are summarized in Appendix 5 and follow the
textbook [1], where many further details can be found.
5 Note that, since we are not using auxiliary fields, the supersymmetry algebra closes only on-shell:
1
[ 𝛿 𝜖2 , 𝛿 𝜖1 ] 𝜙 = 2 ( 𝜖 1 𝛾 𝜇 𝜖2 )𝜕𝜇 𝜙
  1
𝛿 𝜖2 , 𝛿 𝜖1 𝜒𝐿 = 2 ( 𝜖 1 𝛾 𝜇 𝜖2 )𝜕𝜇 𝜒𝐿 + [. . . ] 𝜕𝜒
/ 𝐿,

where [. . . ] denotes a non-vanishing expression of the fields and supersymmetry parameters. The
last
 term then  vanishes due to the field equation 𝜕𝜒
/ 𝐿 = 0, and one obtains the usual susy algebra
𝛿 𝜖2 , 𝛿 𝜖1 = 12 ( 𝜖 1 𝛾 𝜇 𝜖2 )𝜕𝜇 on all fields.
4 Gianguido Dall’Agata and Marco Zagermann

As promised, the result is that under global supersymmetry, where the supersymme-
try parameter is constant, 𝜕𝜇 𝜖 = 0, the Lagrangian transforms into a total derivative:

𝛿 𝜖 L = 𝜕𝜇 (K 𝜇 + K 𝜇 ∗ ) ≡ 𝜕𝜇 𝐾 𝜇 . (7)

When dealing with local supersymmetry, however, the parameter 𝜖 becomes a

local function of the coordinates, 𝜖 = 𝜖 (𝑥), and the Lagrangian is no longer invariant
up to a total derivative. The new non-invariant part of the Lagrangian is
𝜇 𝜇
𝛿 𝜖 Lnew = (𝜕𝜇 𝜖) 𝑗 𝜇 = (𝜕𝜇 𝜖 𝐿 ) 𝑗 𝐿 + (𝜕𝜇 𝜖 𝑅 ) 𝑗 𝑅 , (8)

where
𝜇 𝜇
𝑗 𝐿 ≡ −𝜕/𝜙∗ 𝛾 𝜇 𝜒𝐿 , 𝑗 𝑅 ≡ −𝜕/𝜙𝛾 𝜇 𝜒𝑅 (9)
𝜇 𝜇
give the super-Noether current 𝑗𝜇
= 𝑗 𝐿 + 𝑗 𝑅 . In fact, it can be easily checked that this
supercurrent is a conserved current, namely that 𝜕𝜇 𝑗 𝜇 = 0, upon using the equations
of motion for the fields 𝜙 and 𝜒. It should also be noted that the dimension of these
𝜇
currents is 𝐷 [ 𝑗 𝐿,𝑅 ] = 7/2.
We can now apply Noether’s method and associate to the supercurrent (9) a
gauge field that compensates the non-invariance of the Lagrangian (8). This gauge
field, 𝜓 𝜇 𝛼 , has to have a spinorial index (i.e. the index 𝛼 = 1, 2, 3, 4, which we will
suppress again in the following) and a spacetime index (𝜇 = 0, 1, 2, 3), such that

𝛿 𝜖 𝜓 𝜇 𝐿,𝑅 = 𝑀 𝑃 𝜕𝜇 𝜖 𝐿,𝑅 , 𝛿 𝜖 𝜓 𝜇 𝐿,𝑅 = 𝑀 𝑃 𝜕𝜇 𝜖 𝐿,𝑅 , (10)

where 𝑀 𝑃 is a mass parameter that is needed to relate the mass dimension 3/2 of the
fermionic field 𝜓 𝜇 and the dimension of the supersymmetry parameter 𝐷 [𝜖] = −1/2.
As suggested by the notation, 𝑀 𝑃 will later be identified with the (reduced) Planck
mass.
The Noether procedure tells us that we need to add a new piece to the Lagrangian:

′ 1  𝜇 𝜇

LWZ =− 𝜓 𝜇𝐿 𝑗 𝐿 + 𝜓 𝜇𝑅 𝑗 𝑅 . (11)
𝑀𝑃
Again 𝑀 𝑃 is needed to get a Lagrangian density whose total mass dimension is 4,
and this dimensionful coupling in the action can be viewed as a first sign that we
eventually need gravity in local supersymmetry.
Using (10) in the variation of (11), we now precisely compensate the variation
of the original Wess–Zumino multiplet, but now there is a new piece to compensate
𝜇
in the variation of (11) from 𝛿 𝜖 𝑗 𝑅,𝐿 , which is in general non-vanishing. To see this,
𝜇
it suffices to consider the variation of the term 𝜓 𝜇𝐿 𝑗 𝐿 that is quadratic in the scalar
𝜇
fields. This term comes from the variation of 𝜒𝐿 inside 𝑗 𝐿 :

1
𝜓 𝜇𝐿 𝜕/ 𝜙∗ 𝛾 𝜇 𝛿 𝜖 𝜒𝐿 = 𝜓 𝛾 𝜈 𝛾 𝜇 𝛾 𝜌 𝜖 𝑅 𝜕𝜈 𝜙∗ 𝜕𝜌 𝜙 = 𝜓 𝜇𝐿 𝛾 𝜈 𝜖 𝑅 𝑇 𝜇𝜈 + . . . (12)
2 𝜇𝐿
where, using some gamma matrix algebra,
Simple Supergravity 5

1
𝑇 𝜇𝜈 = 𝜕 ( 𝜇 𝜙𝜕 𝜈) 𝜙∗ − 𝜂 𝜇𝜈 (𝜕𝜎 𝜙𝜕 𝜎 𝜙∗ ), (13)
2
and the dots stand for terms involving 𝛾 𝜈𝜇𝜌 . One can show that variations bilinear
in 𝜒 likewise give the energy momentum tensor for the field 𝜒. So,

′ 1
𝛿LWZ ∼ 𝜖 𝛾 𝜇 𝜓 𝜈 𝑇 𝜇𝜈 + . . . (14)
𝑀𝑃
In order to cancel this term, we now introduce a new current which is a symmetric
tensor 𝑔 𝜇𝜈 with transformation rule

1
𝛿𝑔 𝜇𝜈 ∼ 𝜖 𝛾 ( 𝜇 𝜓 𝜈) , (15)
𝑀𝑃
and add a new piece to the Lagrangian with a coupling between the tensor field 𝑔 𝜇𝜈
and the energy momentum tensor:
′′
LWZ ∼ −𝑔 𝜇𝜈 𝑇 𝜇𝜈 . (16)

As only the spacetime metric can couple to the energy momentum tensor, local
supersymmetry requires the coupling of the Wess–Zumino multiplet to gravity de-
scribed by a dynamical spacetime metric, 𝑔 𝜇𝜈 , and 𝜓 𝜇 must be its superpartner, the
gravitino, as follows from the transformation law (15). As in ordinary gauge theories,
one also adds kinetic terms for these new “gauge” fields, and we thus expect a final
result of the form
L= Lkin (𝜙) + Lkin ( 𝜒) + Lint (𝜙, 𝜒, 𝑔 𝜇𝜈 , 𝜓 𝜇 )
| {z } | {z }
LWZ ′
LWZ ′′
+ LWZ +...
(17)
+ Lkin (𝑔 𝜇𝜈 ) + Lkin (𝜓 𝜇 )

where the dots indicate possible further interaction terms.

We used the chiral multiplet to guess the supersymmetry transformation rules of
the supergravity multiplet. These rules, however, should hold also in the absence of
the chiral multiplet, and we thus arrive at a motivated guess for the Lagrangian and
transformation laws of pure N = 1 supergravity:

Lpure sugra = Lkin (𝑔 𝜇𝜈 ) + Lkin (𝜓 𝜇 ) +Lint (𝑔 𝜇𝜈 , 𝜓 𝜇 ),

| {z } | {z } (18)
𝑀𝑃 2 √
2 −𝑔𝑅 − 21 𝜓 𝜇 𝛾 𝜇𝜈𝜌 𝜕𝜈 𝜓𝜌 | cov

using
1
𝛿𝑔 𝜇𝜈 ≃ 𝜖 𝛾 ( 𝜇 𝜓 𝜈) | cov , (19)
𝑀𝑃
and
𝛿𝜓 𝜇 ≃ 𝑀 𝑃 𝜕𝜇 𝜖 | cov , (20)
6 Gianguido Dall’Agata and Marco Zagermann

where cov stands for a proper spacetime covariantization, and Lint denotes possible
interaction terms that are not contained in the covariantizations of the kinetic terms
(e.g., four-Fermion terms). This spacetime covariantization is done in the vierbein
formalism (cf. Appendix 5.2 and 5.5) and leads to the expressions (23), (26) and
(36). Moreover, as we will see in Section 3, also the additional interaction terms not
related to spacetime covariantization can elegantly be absorbed into the covariantized
kinetic terms by working with covariant derivatives with non-trivial torsion. Before
we come to this, however, let us briefly pause and take a quick look at some basic
aspects of the gravitino field.

2.2 Some remarks on the gravitino field and gravitino multiplets

A vector-spinor field, 𝜓 𝜇 𝛼 (𝑥), a priori has 16 degrees of freedom. The action for a
free vector-spinor field in Minkowski spacetime is the Rarita–Schwinger action [14]
1 1
L3/2 = − 𝜓 𝜇 𝛾 𝜇𝜈𝜌 𝜕𝜈 𝜓 𝜌 + 𝑚 3/2 𝜓 𝜇 𝛾 𝜇𝜈 𝜓 𝜈 , (21)
2 2
where 𝑚 3/2 is the physical mass of the corresponding particle in Minkowski space-
time. A careful analysis of the equations of motion shows that 𝜓 𝜇 𝛼 only propagates
two physical degrees of freedom in the massless case, corresponding to states of he-
licity ±3/2, and four physical degrees of freedom in the massive case, corresponding
to the polarization states of a massive spin-3/2 particle. The other degrees of freedom
are either off-shell or auxiliary, as required to write a Lorentz-invariant action [15].
In the massless case, the Rarita–Schwinger action only consists of the kinetic
term and is thus invariant under the gauge symmetry

𝛿𝜓 𝜇 = 𝜕𝜇 Λ, (22)

where Λ(𝑥) is an arbitrary Majorana spinor. Just as for vector gauge bosons, this
gauge invariance is necessary to eliminate longitudinal polarization states and has
to be preserved by interactions so as to respect unitarity. In the context of super-
gravity, where 𝜓 𝜇 𝛼 (𝑥) is the gravitino, the gauge symmetry (22) is simply local
supersymmetry.
While we encountered the gravitino as the superpartner of the helicity ±2 graviton,
one might also wonder whether it would be possible to write down sensible field
theories where the superpartner of a helicity ±3/2 particle is instead a helicity ±1
particle described by a vector field, 𝐴 𝜇 (𝑥). Such a multiplet is referred to as a
gravitino multiplet, and a globally supersymmetric free field theory for this multiplet
indeed exists. As soon as one tries to introduce ineractions, however, the gauge
Simple Supergravity 7

invariance (22) must be promoted to an additional local supersymmetry, and one

arrives at a theory with several local supersymmetries, i.e. extended supergravity6.

3 Minimal supergravity in four dimensions

Having clarified that local supersymmetry requires the coupling to gravity, we now
want to show how to write down the minimal supersymmetric model, which involves
only the gravity multiplet. This represents the minimal supersymmetric extension
of General Relativity, and provides a sufficient setup to investigate and illustrate
a number of general features of supergravity theories, which remain valid in the
presence of additional matter mutiplets, additional supersymmetries or more general
spacetime dimensions.
We will also show explicitly the calculations needed to prove supersymmetry
invariance as they clarify the origin and meaning of a series of structures common to
all supergravity theories. Once more, additional details and extensions can be found
in [1].

3.1 The minimal action7

In the following we will often use the language of differential forms in order to
simplify calculations. For instance, the action we want to supersymmetrize must
contain the Einstein–Hilbert 𝑆 𝐸 𝐻 and Rarita–Schwinger 𝑆 𝑅𝑆 actions,
∫ ∫ !
4 4
𝑀 𝑃2 1 𝜇𝜈𝜌
𝑆= 𝑑 𝑥 (L 𝐸 𝐻 + L 𝑅𝑆 ) = 𝑑 𝑥𝑒 𝑅 − 𝜓𝜇𝛾 𝐷 𝜈 𝜓𝜌 , (23)
2 2

which, in the language of differential forms become

∫ ∫
𝑀 𝑃2 𝑖
𝑆 = 𝑆 𝐸 𝐻 + 𝑆 𝑅𝑆 = 𝑅 𝑎𝑏 ∧ 𝑒 𝑐 ∧ 𝑒 𝑑 𝜖 𝑎𝑏𝑐𝑑 + 𝑒 𝑎 ∧ 𝜓 ∧ 𝛾5 𝛾 𝑎 𝐷𝜓. (24)
4 2
Here, the indices 𝑎, 𝑏, 𝑐, . . . are local Lorentz indices referring to orthonormal
frames, 𝑒 𝑎 = 𝑒 𝑎𝜇 𝑑𝑥 𝜇 , with constant epsilon tensor, 𝜖0123 = 1, and 𝑅 𝑎𝑏 and 𝜓 denote,
respectively, the curvature two-form (175) of the spin connection and the gravitino

6 Once more, consistent interacting theories including the gravitino may be obtained by allowing
for higher spin gields and couplings between an infinite number of fields [16], but this is not of
interest for our discussion.
7 This section is reprinted from [1] © 2021 Springer-Verlag GmbH Germany, part of Springer
Nature. Reproduced with permissions. All rights reserved.
8 Gianguido Dall’Agata and Marco Zagermann

one-form 𝜓 ≡ 𝜓 𝜇 𝑑𝑥 𝜇 . 𝐷 denotes the Lorentz covariant derivative (210). For further

details on the formalism and our conventions, the reader is referred to the Appendix.
Since we are coupling the spin 3/2 field 𝜓 𝜇 to gravity, the covariant derivative
in its kinetic term should a priori be the full covariant derivative, ∇, and not just
the Lorentz-covariant derivative, 𝐷, we have used in the above expressions. The full
covariant derivative ∇ contains both the Levi–Civita connection, Γ, coupling to the
vector index 𝜇 of the gravitino, as well as the spin connection, 𝜔, coupling to the
(suppressed) spinor index. However, even if we had used ∇, it would appear in the
action only in anti-symmetrized form,8
1
∇ [𝜈 𝜓 𝜌] = 𝜕 [𝜈 𝜓 𝜌] + 𝜔 𝑎𝑏 𝛾 𝑎𝑏 𝜓 𝜌] − Γ 𝜎
[𝜈𝜌] 𝜓 𝜎 , (25)
4 [𝜈
and the last term is identically zero so that ∇ [𝜈 𝜓 𝜌] = 𝐷 [𝜈 𝜓 𝜌] , and we can indeed
use the Lorentz-covariant derivative 𝐷 in the kinetic term of the gravitino. In fact,
the Levi–Civita connection in terms of Christoffel symbols will never really appear
in the following.
We now discuss the invariance under supersymmetry of (23). We start by making
one simple assumption that is motivated by our previous discussion on the gravitino
being the gauge field of supersymmetry. This means that the gravitino transforma-
tion rule should be proportional to the (covariant) derivative of the supersymmetry
parameter  
1 𝑎𝑏
𝛿 𝜖 𝜓 𝜇 = 𝑀 𝑃 𝐷 𝜇 𝜖 ≡ 𝑀 𝑃 𝜕𝜇 + 𝜔 𝜇 𝛾 𝑎𝑏 𝜖, (26)
4
with the conjugate field satisfying
 
1 𝑎𝑏
𝛿 𝜖 𝜓 𝜇 = 𝑀 𝑃 𝜕𝜇 𝜖 − 𝜖 𝛾 𝑎𝑏 𝜔 𝜇 ≡ 𝑀 𝑃 𝐷 𝜇 𝜖 . (27)
4

Having specified only the gravitino supersymmetry transformation so far, the next
thing we would like to obtain is the transformation rule of the vierbein. We could
simply make an educated guess in line with our considerations leading to eq. (15),
but let us try to actually derive the vierbein transformation law from what we already
have. From the variation of 𝑆 𝐸 𝐻 we see that the only contribution with 𝛿𝑒 𝑎 comes
multiplied by the curvature 𝑅 𝑎𝑏 . We therefore try to single out from 𝛿𝑆 𝑅𝑆 all possible
terms that give the same type of contributions proportional to the curvature of the
spin connection. Supersymmetry invariance will then determine 𝛿𝑒 𝑎 , and we will
then check the invariance of the full action.
The variation of the gravitini in the Rarita–Schwinger Lagrangian gives

8 It is important here, that Γ really denotes the torsion-free Levi–Civita connection. As we will see
later, it is useful to include a torsion piece bilinear in the gravitini in the spin connection (but not
in the connection Γ, which should stay torsion-free). The connections defined by Γ and 𝜔 are then
no longer equivalent connections.
Simple Supergravity 9
𝑒 𝑒
𝛿L 𝑅𝑆 = − 𝜓 𝜇 𝛾 𝜇𝜈𝜌 𝐷 𝜈 𝛿𝜓 𝜌 − 𝛿𝜓 𝜇 𝛾 𝜇𝜈𝜌 𝐷 𝜈 𝜓 𝜌 + . . .
2 2
𝑒 𝑒 (28)
= − 𝜓 𝜇 𝛾 𝜇𝜈𝜌 𝐷 𝜈 𝛿𝜓 𝜌 − 𝐷 𝜈 𝜓 𝜌 𝛾 𝜇𝜈𝜌 𝛿𝜓 𝜇 + . . . ,
2 2

where we used the identity 𝜒𝛾 𝜇𝜈𝜌 𝜆 = 𝜆𝛾 𝜇𝜈𝜌 𝜒 for anticommuting Majorana spinors,
and the dots refer to the variations of the vierbein, 𝛿𝑒 𝑎𝜇 , and the spin connection,
𝜔 𝑎𝑏
𝜇 , which we do not consider for now because they give terms that are not of the
form we need. Inserting (26) in (28) we obtain
𝑒 𝑒
𝛿L 𝑅𝑆 = −𝑀 𝑃 𝜓 𝛾 𝜇𝜈𝜌 𝐷 𝜈 𝐷 𝜌 𝜖 − 𝑀 𝑃 𝐷 𝜈 𝜓 𝜌 𝛾 𝜇𝜈𝜌 𝐷 𝜇 𝜖 + . . . (29)
2 𝜇 2
Integrating the last term by parts we can replace it by
𝑒  𝑒
− 𝜕𝜇 𝑀 𝑃 𝐷 𝜈 𝜓 𝜌 𝛾 𝜇𝜈𝜌 𝜖 + 𝑀 𝑃 𝐷 𝜇 𝐷 𝜈 𝜓 𝜌 𝛾 𝜇𝜈𝜌 𝜖 (30)
2 2
plus terms involving derivatives of the vielbeine, 𝐷 𝜇 𝑒 𝑎𝜈 , which we also neglect in
this first step, because they will not give contributions proportional to the curvature
𝑅 𝑎𝑏 . The equivalence of (30) to the last term in (29) can easily be checked either by
recalling that the 𝛾-matrices are covariantly constant in the sense that
1
𝐷 𝜇 𝛾 𝑎 = 𝜕𝜇 𝛾 𝑎 + 𝜔 𝜇 𝑎 𝑏 𝛾 𝑏 + 𝜔 𝑏𝑐 [𝛾𝑏𝑐 , 𝛾 𝑎 ] = 0, (31)
4 𝜇
or that 𝐷 𝜇 (scalar) = 𝜕𝜇 (scalar). From the definition of the covariant derivative
acting on fermions, we find
1
[𝐷 𝜇 , 𝐷 𝜈 ] = 𝑅 𝜇𝜈 𝑎𝑏 𝛾 𝑎𝑏 , (32)
4

and therefore, using 𝛾 𝑎𝑏 𝜓 𝜌 = −𝜓 𝜌 𝛾 𝑎𝑏 ,

1
𝐷 [𝜇 𝐷 𝜈 𝜓 𝜌] = − 𝑅 [𝜇𝜈 𝑎𝑏 𝜓 𝜌] 𝛾 𝑎𝑏 . (33)
8
Hence the variation of the Rarita–Schwinger term becomes
𝑒 𝑒
𝛿L 𝑅𝑆 = − 𝑀 𝑃 𝜓 𝜇 𝛾 𝜇𝜈𝜌 𝛾 𝑎𝑏 𝜖 𝑅 𝜈𝜌 𝑎𝑏 − 𝑀 𝑃 𝜓 𝜌 𝛾 𝑎𝑏 𝛾 𝜇𝜈𝜌 𝜖 𝑅 𝜇𝜈 𝑎𝑏 + . . .
16 16
𝑒
= − 𝑀 𝑃 𝜓 𝜇 {𝛾 𝜇𝜈𝜌 , 𝛾 𝑎𝑏 } 𝜖 𝑅 𝜈𝜌 𝑎𝑏 + . . .
16
 
𝑒 𝜈 𝜇 1 𝜇
= − 𝑀𝑃 𝜓 𝜇 𝛾 𝜖 𝑅𝜈 − 𝛿 𝜈 𝑅 + . . . (34)
2 2
𝑒
= − 𝑀 𝑃 𝜓 𝜇 𝛾 𝜈 𝜖𝐺 𝜈 𝜇 + . . . ,
2
10 Gianguido Dall’Agata and Marco Zagermann

where we introduced the Einstein tensor 𝐺 𝜇𝜈 and used the identity {𝛾 𝜇𝜈𝜌 , 𝛾 𝑎𝑏 } =
−12 𝛾 [𝜇 𝑒 𝑎𝜈 𝑒 𝑏𝜌] . The dots contain all terms that do not multiply the curvature of the
spin connection. As expected, this can be compensated by
𝛿L 𝐸 𝐻 𝑎
𝛿𝑒 𝜇 = −𝑀 𝑃2 𝑒 𝑒 𝜈𝑎 𝐺 𝜈 𝜇 𝛿𝑒 𝑎𝜇 , (35)
𝛿𝑒 𝑎𝜇

which is also proportional to the same combination of the curvature, provided we

define the variation of the vierbein as
1
𝛿𝑒 𝑎𝜇 = 𝜖 𝛾𝑎𝜓𝜇 (36)
2𝑀 𝑃

(recall that 𝜖 𝛾 𝑎 𝜓 𝜇 = −𝜓 𝜇 𝛾 𝑎 𝜖). Note that, proceeding in this way, we did not simply
guess the variation of the vielbein from our considerations in the previous section,
but instead really derived it. On the other hand, we see immediately that (36) is
indeed consistent with (15).
In order to complete the proof of the invariance of the action (23), we still need
to discuss the following variations:
𝛿L 𝐸 𝐻
i) 𝛿 𝜖 𝜔 𝑎𝑏
𝜇 ;
𝛿𝜔 𝑎𝑏
𝜇

𝛿L 𝑅𝑆
ii) 𝛿 𝜖 𝜔 𝑎𝑏
𝜇 ;
𝛿𝜔 𝑎𝑏
𝜇

𝛿L 𝑅𝑆
iii) 𝛿 𝜖 𝑒 𝑎𝜇 ;
𝛿𝑒 𝑎𝜇

𝛿L 𝑅𝑆
iv)Terms involving 𝐷𝑒 𝑎 from the partial integration in 𝛿𝜖 𝜓𝜇.
𝛿𝜓 𝜇
We also need to understand and specify 𝛿 𝜖 𝜔 𝑎𝑏 . As we will see, the variation of the
spin connection will depend on the formalism (first, second or 1.5 order) used to
prove the invariance of the action.
To do this calculation, we go back to the form expression (24). The variation of
the action is then
Simple Supergravity 11

𝑀 𝑃2 𝑀2
𝛿𝑆 = 𝐷𝛿𝜔 𝑎𝑏 ∧ 𝑒 𝑐 ∧ 𝑒 𝑑 𝜖 𝑎𝑏𝑐𝑑 + 𝑃 𝑅 𝑎𝑏 ∧ 𝛿𝑒 𝑐 ∧ 𝑒 𝑑 𝜖 𝑎𝑏𝑐𝑑
4 2
| {z } | {z }
𝐵1 𝐴1
𝑖 𝑖
+ 𝛿𝑒 𝑎 ∧ 𝜓 ∧ 𝛾5 𝛾 𝑎 𝐷𝜓 + 𝑒 𝑎 ∧ 𝛿𝜓 ∧ 𝛾5 𝛾 𝑎 𝐷𝜓
2 2 (37)
| {z } | {z }
𝐵2 𝐴2
𝑖 𝑖
− 𝑒 𝑎 ∧ 𝜓 ∧ 𝛾5 𝛾 𝑎 𝛾𝑐𝑑 𝜓 ∧ 𝛿𝜔 𝑐𝑑 + 𝑒 𝑎 ∧ 𝜓 ∧ 𝛾5 𝛾 𝑎 𝐷𝛿𝜓 ,
8 2
| {z } | {z }
𝐵3 𝐴3

where the first line is the variation of 𝑆 𝐸 𝐻 , and the last two lines come from varying
𝑆 𝑅𝑆 , with 𝐵3 being due to the variation of the spin connection inside 𝐷.
We know from previous computations that the term 𝐴1, coming from 𝛿L 𝐸 𝐻 /𝛿𝑒 𝑎 ,
and the terms involving 𝐷 2 𝜖, 𝐷 2 𝜓 coming from 𝛿L 𝑅𝑆 /𝛿𝜓 cancel. In detail, the 𝐷 2 𝜖-
term is 𝐴3, where one uses the explicit expression for 𝛿𝜓, and the 𝐷 2 𝜓-term can be
extracted from 𝐴2 using the same steps that also led to (36). To do so, we switch the
2-form 𝐷𝜓 and the 1-form 𝛿𝜓, using (206), so that
𝑖 𝑎 𝑖 𝑖
𝐴2 ≡ 𝑒 ∧𝛿𝜓 ∧𝛾5 𝛾 𝑎 𝐷𝜓 = 𝑒 𝑎 ∧𝐷𝜓 ∧𝛾5 𝛾 𝑎 𝛿𝜓 = 𝑀 𝑃 𝑒 𝑎 ∧𝐷𝜓 ∧𝛾5 𝛾 𝑎 𝐷𝜖 (38)
2 2 2
and, integrating again by parts,
 
𝑖 𝑖
𝐴2 = −𝑀 𝑃 𝑑 𝑒 𝑎 ∧ 𝐷𝜓𝛾5 𝛾 𝑎 𝜖 + 𝑀 𝑃 𝐷𝑒 𝑎 ∧ 𝐷𝜓𝛾5 𝛾 𝑎 𝜖
2 2
| {z }
𝐴2′′ (39)
𝑖
− 𝑀 𝑃 𝑒 𝑎 ∧ 𝐷𝐷𝜓𝛾5 𝛾 𝑎 𝜖 .
2
| {z }
𝐴2′

The term 𝐴2′ then cancels 𝐴1 and 𝐴3 as before, and we are left with 𝛿𝑆 = 𝐵1 + 𝐵2 +
𝐵3 + 𝐴2 ′′ plus boundary terms.
To proceed further, we integrate by parts the term 𝐵1 and get
!
𝑀 𝑃2 𝑎𝑏 𝑐 𝑑
𝑀 𝑃2 𝑎𝑏 𝑐 𝑑
𝐵1 = 𝛿𝜔 ∧ 𝐷𝑒 ∧ 𝑒 𝜖 𝑎𝑏𝑐𝑑 + 𝑑 𝛿𝜔 ∧ 𝑒 ∧ 𝑒 𝜖 𝑎𝑏𝑐𝑑 .
2 4

In order to write 𝐵3 in a very similar form, we can use properties of the 𝛾-matrices

𝜓 ∧ 𝛾5 𝛾 𝑎 𝛾𝑐𝑑 𝜓 = 𝜓 ∧ 𝛾5 (𝛾 𝑎𝑐𝑑 + 𝜂 𝑎𝑐 𝛾 𝑑 − 𝜂 𝑎𝑑 𝛾𝑐 )𝜓 = −𝑖 𝜓 ∧ 𝛾 𝑒 𝜓𝜖 𝑎𝑐𝑑𝑒 (40)

so that, after some relabelling and reordering,

1
𝐵3 = − 𝛿𝜔 𝑎𝑏 ∧ 𝜓 ∧ 𝛾 𝑐 𝜓 ∧ 𝑒 𝑑 𝜖 𝑎𝑏𝑐𝑑 . (41)
8
12 Gianguido Dall’Agata and Marco Zagermann

Discarding boundary terms and inserting also (36) in B2, we then have
!
𝑀 𝑃2 1
𝛿𝑆 = 𝛿𝜔 𝑎𝑏 ∧ 𝐷𝑒 𝑐 − 𝜓 ∧ 𝛾 𝑐 𝜓 ∧ 𝑒 𝑑 𝜖 𝑎𝑏𝑐𝑑
2 4𝑀 𝑃2 (42)
𝑖 𝑖
+ (𝜖 𝛾 𝑎 𝜓) ∧ (𝜓 ∧ 𝛾5 𝛾 𝑎 𝐷𝜓) + 𝑀 𝑃 𝐷𝑒 𝑎 ∧ 𝐷𝜓𝛾5 𝛾 𝑎 𝜖.
4𝑀 𝑃 2
This expression
 can be simplified
 by rewriting the second line so that the torsion
𝑎 1 𝑎
piece 𝐷𝑒 − 4𝑀 2 𝜓 ∧ 𝛾 𝜓 can also be factored out. While the last term of (42)
𝑃
obviously contains a derivative of the vielbein, the other term needs a reshuffling
of the gravitini in order to produce the right bilinear without derivatives. We can
achieve this by using the Fierz identity
1   1  
𝜓 ∧𝜓= 𝜓 ∧ 𝛾 𝑎 𝜓 𝛾𝑎 − 𝜓 ∧ 𝛾 𝑎𝑏 𝜓 𝛾 𝑎𝑏 (43)
4 8
and the fact that 𝛾 𝑐 𝛾 𝑎𝑏 𝛾𝑐 = 0 and 𝛾 𝑐 𝛾 𝑎 𝛾𝑐 = −2𝛾 𝑎 :
𝑖 𝑖
(𝜖 𝛾 𝑎 𝜓) ∧ (𝜓 ∧ 𝛾5 𝛾 𝑎 𝐷𝜓) = − (𝜓 ∧ 𝛾 𝑎 𝜓) ∧ (𝐷𝜓𝛾5 𝛾 𝑎 𝜖). (44)
4𝑀 𝑃 8𝑀 𝑃
Altogether, 𝛿𝑆 can then be written as
!
𝑀𝑃 1 h i
𝑎 𝑎 𝑏𝑐 𝑑
𝛿𝑆 = 𝐷𝑒 − 𝜓 ∧ 𝛾 𝜓 ∧ 𝑖𝐷𝜓𝛾 5 𝛾 𝑎 𝜖 + 𝑀 𝑃 𝛿𝜔 ∧ 𝑒 𝜖 𝑎𝑏𝑐𝑑
2 4𝑀 𝑃2

!  
𝑀𝑃 𝑎 1 𝑎 1 𝑏𝑐𝑑 𝑏𝑐 𝑑
= 𝐷𝑒 − 𝜓 ∧ 𝛾 𝜓 ∧ − 𝐷𝜓𝛾 𝜖 + 𝑀 𝑃 𝛿𝜔 ∧ 𝑒 𝜖 𝑎𝑏𝑐𝑑 , (45)
2 4𝑀 𝑃2 6

where we have used 𝛾5 𝛾 𝑎 = (𝑖/6)𝜖 𝑎𝑏𝑐𝑑 𝛾 𝑏𝑐𝑑 . At this point the variation of the spin
connection assumes a primary role and we can try to set (45) to zero in various
different ways.

3.1.1 Second order formalism

In this case, one imposes the so-called conventional constraint,

1
𝐷𝑒 𝑎 = 𝜓 ∧ 𝛾 𝑎 𝜓, (46)
4𝑀 𝑃2

which determines the spin connection, 𝜔 𝑎𝑏 𝜇 = 𝜔 ˆ 𝑎𝑏

𝜇 (𝑒, 𝜓), as the solution to this
equation. The spin connection is thus treated from the very beginning as a dependent
field, whose supersymmetry variation follows from the supersymmetry variations of
𝑒 𝑎𝜇 and 𝜓 𝜇 via the chain rule and the explicit functional dependence of 𝜔ˆ 𝑎𝑏
𝜇 (𝑒, 𝜓).
Simple Supergravity 13

By simple inspection of (45), however, we see that (46) already implies that 𝛿 𝜖 𝑆 = 0,
and we don’t really need to know 𝛿 𝜖 𝜔 𝑎𝑏
𝜇 .
Let us nevertheless use the torsion
1
𝑇𝑎 = 𝜓 ∧ 𝛾𝑎𝜓 (47)
4𝑀 𝑃2

to solve 𝐷𝑒 𝑎 = 𝑇 𝑎 for the spin connection, which yields

1  [𝑎 𝑏] [𝑎 𝑏 ] [𝑎 𝑏 ]

𝜔ˆ 𝑎𝑏 𝑎𝑏
𝜇 (𝑒, 𝜓) = 𝜔 𝜇 (𝑒) − 𝜓 𝛾 𝜇 𝜓 − 𝜓 𝜇 𝛾 𝜓 − 𝜓 𝛾 𝜓 𝜇 , (48)
4𝑀 𝑃2

where 𝜔 𝑎𝑏
𝜇 (𝑒) is the torsion-free spin connection (171), and the remaining piece is the
contorsion tensor (183). This is used in the original approach of Ferrara, Freedman
and Van Nieuwenhuizen in [2]. It is interesting to point out that the supersymmetry
variation of this connection, inherited from the variations of 𝑒 𝑎𝜇 and 𝜓 𝜇 , does not
contain derivative terms, 𝜕𝜖, of the supersymmetry parameter:
1  
𝛿 𝜔ˆ 𝑎𝑏
𝜇 = 𝜖 𝛾 𝜌 (𝐷 𝜎 𝜓 𝜏 ) 2𝑒 𝜌[𝑎 𝑒 𝑏𝜏 ] 𝑔 𝜇 𝜎 − 𝑒 𝜏[𝑎 𝑒 𝑏𝜎] 𝑔 𝜇𝜌 .
𝑀𝑃

This is the reason why 𝜔ˆ 𝑎𝑏

𝜇 is often called supercovariant.

3.1.2 First order formalism

This is the approach followed by Deser and Zumino in their original paper [3].
Asking for the invariance of the action, 𝛿𝑆 = 0, via the vanishing of the term in
square brackets in (45) fixes the variation of the spin connection to
1 𝑐 𝑏𝑒 1 𝑏 𝑐𝑒
𝛿𝜔 𝑏𝑐 𝑏𝑐
𝜇 = 𝐵𝜇 − 𝑒 𝐵 + 𝑒 𝜇 𝐵𝑒 , (49)
2 𝜇 𝑒 2
with
𝑖
𝐵𝑏𝑐
𝜇 = 𝜖 𝛾 𝜇 𝛾5 𝐷 𝜌 𝜓 𝜎 𝜖 𝜌𝜎𝑏𝑐 . (50)
2𝑀 𝑃
This can be extracted using the same trick that is used in Appendix 5.2 to derive
the form (171) of the torsion-free spin connection in terms of the vierbein from
the torsion constraint. It should be noted that (49) is not the same as the variation
derived using second order approach in the previous subsection. However, they
become equivalent upon using the gravitino equations of motion.

3.1.3 1.5-order formalism

In the 1.5 order formalism, one uses the fact that (46) can be obtained as a field
equation from varying the action with respect to 𝜔 𝑎𝑏
𝜇 (as is obvious from the terms
14 Gianguido Dall’Agata and Marco Zagermann

proportional to 𝛿𝜔 𝑏𝑐 in (45)). Thus, when we determine the supersymmetry variation

of the action and require 𝜔 𝑎𝑏
𝜇 to be determined by

1 𝛿𝑆
𝐷𝑒 𝑎 = 2
𝜓 ∧ 𝛾𝑎𝜓 ⇔ = 0,
4𝑀 𝑃 𝛿𝜔 𝑎𝑏
𝜇

we can immediately drop all terms proportional to 𝛿𝜔 𝑎𝑏 𝜇 , as these are proportional to

𝛿𝑆
𝛿 𝜔 𝜇𝑎𝑏
, which vanishes on-shell. Obviously, for the simple action we consider here,
the only advantage over the second order formalism is that we would not have to
keep track of the 𝛿𝜔 𝑎𝑏 𝜇 terms in (45). Just as in the second order formalism, the
vanishing of the supersymmetry variation (45) is thus obtained by using (46), with
the difference that (46) is now not imposed by hand, but arises as a field equation for
the independent field 𝜔 𝑎𝑏𝜇 . It is in this sense that the 1.5 order formalism combines
elements from the first order formalism (the a priori independence of the field 𝜔 𝑎𝑏 𝜇 )
and from the second order formalism (the use of (46) for the cancellation of (45)).
It should be stressed that the 1.5 order trick of using on-shell field equations in
the supersymmetry variation can only be used for auxiliary fields such as 𝜔 𝑎𝑏 𝜇 .

3.2 Gauging the Poincaré algebra9

When introducing general relativity as well as supergravity, we discussed the possi-

bility of considering the vierbein and the spin connection as independent quantities.
Do we have any conceptual reason behind this, in addition to the simplification of
some computations? We will now see that an interesting perspective on gravity,
which can help when dealing with supergravity, is that of considering gravity itself
as a sort of a gauge theory where the gauge group is the Poincaré group [17, 18, 20].
This analogy will work only to a certain extent, but it will be very useful for un-
derstanding many specific new features that have to be introduced when one wants
to promote supersymmetry to a local symmetry of nature. In fact, supergravity is
the gauge theory of supersymmetry and therefore there must be a way to describe
it as a theory where the gauge group is the Poincaré supergroup (or some other
supergroup). For the sake of simplicity in this subsection we set 𝑀 𝑃 = 1.
Consider an ordinary gauge transformation 𝛿 𝜖 = 𝜖 𝐴𝑇𝐴 , where 𝑇𝐴 are the gauge
generators satisfying
[𝑇𝐴 , 𝑇𝐵 ] = 𝑓 𝐴𝐵 𝐶 𝑇𝐶 ,
with structure constants 𝑓 𝐴𝐵 𝐶 . If this is a global symmetry of an action, it can be
made local by introducing vector fields, 𝐴 𝜇𝐴 , for each symmetry so that the algebra
 
[𝛿(𝜖1𝐴 ), 𝛿(𝜖2𝐵 )] = 𝛿 𝜖2𝐵 𝜖1𝐴 𝑓 𝐴𝐵 𝐶 , (51)

9 This section is reprinted from [1] © 2021 Springer-Verlag GmbH Germany, part of Springer
Nature. Reproduced with permissions. All rights reserved.
Simple Supergravity 15

with symmetry parameters 𝜖𝑖𝐴 has a faithful realization on them,

𝛿 𝜖 𝐴 𝜇𝐴 = 𝜕𝜇 𝜖 𝐴 + 𝜖 𝐶 𝐴 𝐵 𝐴
𝜇 𝑓 𝐵𝐶 , (52)

and we can introduce covariant derivatives

𝐷 𝜇 = 𝜕𝜇 − 𝐴 𝜇𝐴𝑇𝐴 (53)

acting non-trivially on fields which transform in non-trivial representations of the

gauge group. The curvature, defined as
𝐴 𝐴 𝐴 𝐴
[𝐷 𝜇 , 𝐷 𝜈 ] = −𝐹𝜇𝜈 𝑇 ⇔ 𝐹𝜇𝜈 ≡ 2𝜕 [𝜇 𝐴𝜈] + 𝐴𝐵 𝐶 𝐴
𝜇 𝐴 𝜈 𝑓 𝐵𝐶 , (54)

transforms covariantly:
𝐴
𝛿 𝜖 𝐹𝜇𝜈 = 𝜖 𝐶 𝐹𝜇𝜈
𝐵
𝑓 𝐵𝐶 𝐴 . (55)
Let us now imagine that we want to make local the symmetries of the Poincaré
group. The usual procedure is to introduce gauge fields in correspondence with the
generators of the algebra. For the Poincaré algebra with generators 𝑃 𝑎 and 𝑀𝑎𝑏 , this
means introducing two gauge fields, 𝑒 𝑎𝜇 and 𝜔 𝑎𝑏
𝜇 , so as to match the gauge generators,

1 𝑎𝑏
𝐴 𝜇𝐴𝑇𝐴 = 𝑒 𝑎𝜇 𝑃 𝑎 + 𝜔 𝑀𝑎𝑏 . (56)
2 𝜇
Given the particular form of the Poincaré algebra, the gauge curvatures of these
vectors are precisely
𝑇 𝑎 = 𝑑𝑒 𝑎 + 𝜔 𝑎 𝑏 ∧ 𝑒 𝑏 (57)
and
𝑅 𝑎 𝑏 = 𝑑𝜔 𝑎 𝑏 + 𝜔 𝑎 𝑐 ∧ 𝜔 𝑐 𝑏 , (58)
where the spin connection and the vierbein are so far independent fields.
This construction is perfectly legitimate. However, it clearly leads to an ordinary
gauge theory10 and not to a gravity theory as we would like. From the 𝑇 𝑎 and 𝑅 𝑎𝑏
curvatures we could construct kinetic terms giving the propagation of independent
degrees of freedom and discuss the resulting gauge theory, where the Poincaré group
is realised on the vector fields as

𝛿 𝑃 𝜔 𝑎𝑏 = 0, (59)

𝛿 𝑃 𝑒 𝑎 = 𝐷𝜖 𝑎 , (60)

𝛿 𝑀 𝜔 𝑎𝑏 = 𝑑Λ𝑎𝑏 − 𝜔 𝑎 𝑐 Λ𝑐𝑏 − Λ𝑎𝑐 𝜔 𝑐 𝑏 , (61)

𝛿 𝑀 𝑒 𝑎 = Λ𝑎 𝑐 𝑒 𝑐 , (62)

10 Due to the Poincaré algebra being non-semisimple and non-compact, the standard kinetic terms
of the gauge fields would not be positive definite so that this would actually not be a unitary theory.
16 Gianguido Dall’Agata and Marco Zagermann

with gauge parameters 𝜖 𝑎 and Λ𝑎𝑏 . If, on the other hand, we want to get only the
metric degrees of freedom we have to impose a constraint between 𝜔 𝑎𝑏 and 𝑒 𝑎 . The
constraint that does this job is the conventional constraint or torsion constraint (see
[19, 20] for details)÷,
𝑇 𝑎 = 0. (63)
This constraint, however, is not invariant under (59)–(60):

𝛿 𝑃𝑇 𝑎 = 𝛿 𝑃 𝐷𝑒 𝑎 = 𝐷𝛿 𝑃 𝑒 𝑎 = 𝐷𝐷𝜖 𝑎 = −𝑅 𝑎 𝑏 𝜖 𝑏 ≠ 0. (64)

This means that if we impose the conventional constraint, translation symmetry is

broken. Moreover, it is also clear that now the spin connection 𝜔 𝑎𝑏 cannot be treated
as independent of the vielbein anymore and hence the transformation (59) will no
longer be valid. Indeed, since 𝜔 𝑎𝑏 = 𝜔 𝑎𝑏 (𝑒), the spin connection is not invariant
under translations ∫
𝛿𝜔 𝑎𝑏
𝛿 𝑃 𝜔 𝑎𝑏 = 𝑑4𝑥 𝛿 𝑃 𝑒 𝑐 ≠ 0.
𝛿𝑒 𝑐
The final outcome of this discussion is that, when the conventional constraint is
imposed, the Poincaré gauge algebra is deformed and translational symmetry is
replaced by a new invariance under diffeomorphisms. This can be seen by considering
the commutator of two translation generators on the vierbein:

[𝛿 𝑃2 , 𝛿 𝑃1 ]𝑒 𝑎 = −𝛿 𝑃 [2 (𝐷𝜖1]
𝑎
) = −𝛿 𝑃 [2 𝜔 𝑎 𝑐 𝜖1]
𝑐
, (65)

and now 𝛿 𝑃 𝜔 𝑎𝑏 ≠ 0. The resulting algebra then has a non-vanishing commutator

[𝑃, 𝑃] ≠ 0,

as is appropriate for general coordinate transformations, which do not commute.

Actually one can check that using the constraint (63) the translation generators on
the vielbein take the form of general coordinate transformations.
The action constructed from the curvatures and the vierbein then is invariant with
respect to local Lorentz transformations and diffeomorphisms. The infinitesimal
change of a function under a diffeomorphism is given by the Lie derivative 𝐿 𝜖 , and
therefore the action is going to be invariant if (cf. Appendix 5.7)
∫ ∫
𝐿𝜖 𝑆 = 𝑑 (𝚤 𝜖 L) + 𝚤 𝜖 𝑑L = 0,

but the first term is a total derivative that can be discarded while the second is zero
because 𝑑L has one degree more than the top-form. Finally, in the construction of an
action, we will not make use of a kinetic term of the form 𝑅 𝑎𝑏 ∧★𝑅 𝑎𝑏 because of the
conventional constraint which makes it quartic in the derivatives. The appropriate
quadratic term is the Einstein–Hilbert action above.
The method we have outlined in this section can be easily extended to generic
supergravity theories by extending the Poincaré algebra to the super Poincaré al-
gebra by including fermionic generators and possibly other bosonic generators for
Simple Supergravity 17

the internal symmetries. The power of this approach lies in the ease of guessing
the transformation laws under the various symmetries, including supersymmetry.
This means that this approach can be used as a guide to derive and construct the
Lagrangian and/or the equations of motion of systems respecting any symmetry
group we would like to realize. Once again, we stress that one has to be careful with
its application because of the constraints that will be needed to obtain a consistent
gravity theory (invariant under diffeomorphisms). Imposing these constraints will
break the transformation rules that do not preserve them.
We end this discussion with a few remarks on the gauging of the super Poincaré
group. We could gauge this algebra by adding new vector fields 𝜓 𝜇 𝐴 for the fermionic
generators 𝑄 𝐴. From the algebra we then have
1 𝑎𝑏 𝐴
𝐴 𝜇𝐴𝑇𝐴 = 𝑒 𝑎𝜇 𝑃 𝑎 + 𝜔 𝑀𝑎𝑏 + 𝜓 𝜇 𝐴𝑄 𝐴 + 𝜓 𝜇 𝑄 𝐴 , (66)
2 𝜇
and we can read the supersymmetry transformations by applying

𝛿 𝜖 𝐴 𝜇𝐴 = 𝜕𝜇 𝜖 𝐴 + 𝜖 𝐶 𝐴 𝐵 𝐴
𝜇 𝑓 𝐵𝐶 .

For instance, for N = 1 supergravity, we would get that the spin connection is
invariant,
𝛿 𝜖 𝜔 𝑎𝑏 = 0, (67)
because the Lorentz generator never appears on the right hand side of any commutator
involving the supersymmetry generator. However, just like for the bosonic case, we
should impose a torsional constraint in order for the vierbein and spin connection
not to be independent. Doing so, we would fix the form of the spin connection as
𝜔 𝑎𝑏 = 𝜔 𝑎𝑏 (𝑒, 𝜓) and could check the new realization of the algebra on the fields.
One last interesting remark involves the definition of the gauge curvatures for the
super-Poincaré algebra. From the structure constants of the supersymmetry algebra,
one can deduce a new definition for the curvatures, including the one of the trans-
lation generators, which is usually denoted as the 2-form T 𝑎 . Since the translation
generators 𝑃 𝑎 appear on the right hand side of the commutator of two supercharges
the corresponding curvature definition is now
1 𝐴 𝑎
T 𝑎 = 𝐷𝑒 𝑎 − 𝜓 𝛾 𝜓𝐴 (68)
4
and involves a fermion bilinear. This means that imposing the constraint T 𝑎 = 0
results in a spin connection depending on the gravitino fields. Hence, supergravity is
often referred to as a theory with non-trivial torsion for the spin connection, because
T 𝑎 = 0 implies 𝐷𝑒 𝑎 ≠ 0.
18 Gianguido Dall’Agata and Marco Zagermann

3.3 Adding a cosmological constant11

So far we considered the construction of a supergravity action around a Minkowski

background, whose non-linear completion led to Einstein gravity coupled to a grav-
itino field without a cosmological constant. In ordinary Einstein gravity, however,
we can always add a cosmological constant Λ to obtain the action
∫  
2 4 √ 1
𝑆 = 𝑀𝑃 𝑑 𝑥 −𝑔 𝑅 − Λ (69)
2

and find a maximally symmetric vacuum with

Λ > 0 ↔ de Sitter (dS),

𝑅 𝜇𝜈 = Λ𝑔 𝜇𝜈 (70)
Λ < 0 ↔ Anti-de Sitter (AdS).

It is natural to ask whether these solutions and the corresponding actions can be
supersymmetrized in a natural way. In this section, we will focus on pure supergravity
theories (without matter multiplets). If we want to construct a supergravity action
generalizing (69), we should be able to find a supergroup that contains the symmetry
group of AdS and/or dS spacetime. We will now see that minimal supersymmetry
constrains the closure of the algebra in a way that only one of the two options is
consistent12.
Before discussing the corresponding superalgebras, we note that the symmetry
groups of both AdS and dS in 𝑑 dimensions, SO(1,𝑑) and SO(2,𝑑 − 1), respectively,
can be embedded in SO(2, 𝑑). For 𝑑 = 4, the (A)dS algebra is described by 10
anti-Hermitian generators 𝑀 𝐴𝐵 satisfying the commutator relations

[𝑀 𝐴𝐵 , 𝑀𝐶𝐷 ] = −2 𝜂𝐶 [ 𝐴 𝑀 𝐵]𝐷 + 2 𝜂 𝐷 [ 𝐴 𝑀 𝐵]𝐶 , (71)

with 𝜂 𝐴𝐵 = diag{− + + + −} for AdS, and 𝜂 𝐴𝐵 = diag{− + + + +} for dS space. The

explicit (A)dS algebra follows by identifying 𝑀5𝑎 = ℓ 𝑃 𝑎 , where we split 𝐴 = {𝑎, 5},
with 𝑎, 𝑏, . . . = 0, 1, 2, 3, and ℓ is the radius of curvature of (A)dS:

[𝑀𝑎𝑏 , 𝑀𝑐𝑑 ] = −2 𝜂 𝑐 [𝑎 𝑀𝑏 ]𝑑 + 2 𝜂 𝑑 [𝑎 𝑀𝑏 ]𝑐 ,
[𝑃 𝑎 , 𝑀𝑏𝑐 ] = 2 𝜂 𝑎 [𝑏 𝑃𝑐] , (72)
1
[𝑃 𝑎 , 𝑃𝑏 ] = ± 2 𝑀𝑎𝑏 ,
ℓ

11 This section is reprinted from [1] © 2021 Springer-Verlag GmbH Germany, part of Springer
Nature. Reproduced with permissions. All rights reserved.
12 There are consistent de Sitter superalgebras with extended supersymmetries, but they do not
allow for positive weight representations and hence their realizations have the wrong sign in front
of the kinetic terms of some of their fields [21, 22].
Simple Supergravity 19

where the last commutator is equivalent to [𝑀5𝑎 , 𝑀5𝑏 ] = −𝜂55 𝑀𝑎𝑏 and 𝜂55 = −1
for AdS space and 𝜂55 = +1 for dS. Hence, the upper sign is for AdS and the lower
one for dS spacetime. Clearly, when ℓ → ∞ the (A)dS curvature goes to zero and
one gets back the Poincaré algebra.
To construct the full superalgebra, one needs to specify also the commutators
with the supercharges. In particular, [𝑃 𝑎 , 𝑄] cannot be zero anymore, as it used to
be in the super-Poincaré case, because we would no longer close the super Jacobi
identities, as

[[𝑃 𝑎 , 𝑃𝑏 ] , 𝑄] + [[𝑃 𝑎 , 𝑄] , 𝑃𝑏 ] − [[𝑃𝑏 , 𝑄] , 𝑃 𝑎 ] = 0. (73)

| {z } | {z } | {z }
∼𝑀𝑎𝑏 0 0

We therefore need to impose new commutator relations such that the momenta
do not commute with the supercharges. To respect Lorentz covariance and the
graded algebra structure, the result of the commutator should be proportional to
the supercharges and come with some gamma matrices. In principle there are two
possibilities that respect the Majorana condition on the supercharges

[𝑃 𝑎 , 𝑄] ∼ 𝛾 𝑎 𝑄, or [𝑃 𝑎 , 𝑄] ∼ 𝛾 𝑎 𝛾5 𝑄. (74)

In the first case the coefficient multiplying the right hand side should be real, while
in the second it should be imaginary. If we use a chiral notation, the sign and the
ambiguities can be reabsorbed in a single dimensionful complex coefficient 𝑔. ˜ We
stress this fact, because there are sometimes wrong statements in the literature about
this. Once we introduce the chiral notation, the new commutators are
𝑔˜ 𝑔˜ ∗
[𝑃 𝑎 , 𝑄 𝑅 ] = − 𝛾 𝑎 𝑄 𝐿 , [𝑃 𝑎 , 𝑄 𝐿 ] = − 𝛾𝑎 𝑄 𝑅 . (75)
2 2
Once we introduce these new non-trivial commutators, the super Jacobi identity
can be satisfied, though only for the AdS case. This is readily seen by explicitly
computing the results of the various commutators:
!
0 = [𝑃 𝑎 , [𝑃𝑏 , 𝑄 𝐿 ]] + [𝑄 𝐿 , [𝑃 𝑎 , 𝑃𝑏 ]] − [𝑃𝑏 , [𝑃 𝑎 , 𝑄 𝐿 ]]
|𝑔| 2 1
= (𝛾𝑏 𝛾 𝑎 − 𝛾 𝑎 𝛾𝑏 ) 𝑄 𝐿 ± 2 [𝑄 𝐿 , 𝑀𝑎𝑏 ] (76)
4 ℓ
|𝑔| 2 1
=− 𝛾 𝑎𝑏 𝑄 𝐿 ± 2 𝛾 𝑎𝑏 𝑄 𝐿 ,
2 2ℓ
where we used the commutation relations in (217), but with the anti-Hermitian
generators 𝑃 𝑎 = 𝑖 P𝑎 and 𝑀𝑎𝑏 = 𝑖 M 𝑎𝑏 . It is now clear that only for the plus sign
we can get a solution:
1
|𝑔| 2 = 2 . (77)
ℓ
Hence, only for AdS we can write a consistent supersymmetric completion with
a single supercharge. We finally note that the closure of the super-Jacobi identites
20 Gianguido Dall’Agata and Marco Zagermann

requires that another commutator gets modified, namely

{𝑄 𝐿 , 𝑄 𝐿 } = 𝑔˜ ∗ 𝛾 𝑎𝑏 𝑀𝑎𝑏 . (78)

The constraint imposing that the superalgebra can be defined only for the AdS
supergroup and not for dS implies a very important fact: a positive cosmological
constant will always break supersymmetry, while a negative cosmological constant
may be compatible with supersymmetry.
Although matter couplings or extended supersymmetries may allow for de Sitter
vacua in a supersymmetric theory, the vacuum itself will always break supersymme-
try.
Once supersymmetry is broken, one could describe this phase of the theory by
using non-linear realizations, as it is customary for any other symmetry whose linear
action is broken. This has been the subject of intense scrutiny (see for instance
[23, 24, 25, 26]) and one can indeed write actions for theories with dS vacua where
supersymmetry is non-linearly realized. Since an effective discussion of this topic
requires some additional technical introduction to superfields in supergravity, we
will not deal with it here, but refer the reader to the literature on the subject, such as
[9].

3.3.1 Construction of the action

Now that we established that the anti de Sitter group can be consistently extended
to a supergroup, we would like to realize it in terms of a supersymmetric action that
includes a negative cosmological constant. We will proceed in a fashion similar to
what has been done in the flat case, starting from the supersymmetry transformation
of the gravitino and then trying to close the action of the supersymmetry transforma-
tion on the free Lagrangian for the gravity multiplet, possibly introducing interaction
terms. We therefore need to fix first the supersymmetry transformation rule of the
gravitino. Since the algebra has been modified with respect to the case without
cosmological constant, we expect that also the supersymmetry transformations get
modified accordingly.
As sketched in Section 3.2, we can generically deduce the supersymmetry transfor-
mation properties of the various fields from the structure constants of the underlying
superalgebra. For supergravity without cosmological constant, one would do this by
viewing supergravity as a gauge theory of the super-Poincaré group, but with some
constraints needed to relate the vierbein and the spin-connection degrees of freedom.
By using this trick, we can now deduce the supersymmetry transformation of
the gravitino in the presence of a negative cosmological constant by looking at
the structure constants of the 𝐴𝑑𝑆 superalgebra coming from commutators which
have a supersymmetry generator on the right hand side. This inspection shows that
a new term in the supersymmetry transformation of the gravitino should appear
because of the non-zero commutator (75) between the translation generators and
supersymmetry generators. If one interprets the spin connection term in the Lorentz
Simple Supergravity 21

covariant derivative in the original gravitino transformation (26) as due to the non-
vanishing commutator of 𝑀𝑎𝑏 with 𝑄, the new non-vanishing commutator (75)
between 𝑃 𝑎 and 𝑄 should then analogously lead to an additional contribution to
the gravitino transformation so as to make the transformation covariant with respect
to the full AdS isometry group. In 𝛿𝜓 𝜇𝐿 , this additional contribution should be of
the form given in the first equation of (75) contracted with the gauge field of the
translation generator 𝑃 𝑎 , i.e. with the vierbein 𝑒 𝑎𝜇 . We therefore should have
𝑔 2
𝛿𝜓 𝜇𝐿 = 𝑀 𝑃 𝐷 𝜇 𝜖 𝐿 − 𝑀 𝛾 𝜇 𝜖𝑅 , (79)
2 𝑃
where now we have a dimensionless constant 𝑔 ∈ C, because of the introduction
of the dimensionful 𝑀 𝑃 and 𝑀 𝑃2 factors. Clearly, this has to go together with the
conjugate relation:
𝑔∗ 2
𝛿𝜓 𝜇𝑅 = 𝑀 𝑃 𝐷 𝜇 𝜖 𝑅 − 𝑀 𝛾𝜇𝜖𝐿 . (80)
2 𝑃
Once again we stress that 𝑔 here can be any complex number, because there are
sometimes wrong statements in the literature about this.
To construct the action, we start from the action (23) with the vierbein transfor-
mation rule
1
𝛿𝑒 𝑎𝜇 = 𝜖 𝐿 𝛾 𝑎 𝜓 𝜇𝑅 + ℎ.𝑐.,
2𝑀 𝑃
and (79) for the gravitino
𝑔 2
𝛿𝜓 𝜇𝐿 = 𝑀 𝑃 𝐷 𝜇 𝜖 𝐿 − 𝑀 𝛾 𝜇 𝜖𝑅 .
2 𝑃
The reason we start from the action without the cosmological constant, rather than
adding explicitly the cosmological constant among the bosonic terms right from the
beginning, is that it will automatically be enforced by supersymmetry in an iterative
procedure at higher order in 𝑔, as will become clear momentarily.
The gravitino relation differs from the one in (26) by a shift term proportional
to the constant 𝑔. Clearly this shift breaks the supersymmetry of the original action
(23) and we need to restore it by adding additional terms to it. In the following, we
will establish again the invariance under supersymmetry of a modified action. To
our knowledge this was first done in [27].
The first supersymmetry-breaking effect of the shift term (proportional to 𝑔) is
that of generating new terms in the variation of the Rarita–Schwinger part of the
Lagrangian. To compute these terms, we use the supersymmetry variation of the
conjugate gravitino, which, in form notation, reads

𝑔∗ 2
𝛿𝜓 𝑅 = 𝑀 𝑃 𝐷𝜖 𝑅 + 𝑀 𝜖 𝐿 𝛾𝑎 𝑒 𝑎 , (81)
2 𝑃
as one may easily verify. Denoting by 𝛿 𝑔 the variations due to the O(𝑔) shift
term in the gravitino transformation law, the uncancelled variation of L 𝑅𝑆 under
supersymmetry is then
22 Gianguido Dall’Agata and Marco Zagermann

𝑖 𝑎 𝑖
𝛿 𝑔 L 𝑅𝑆 = 𝑒 ∧ 𝛿 𝑔 𝜓 𝑅 ∧ 𝛾5 𝛾 𝑎 𝐷𝜓 𝐿 + 𝑒 𝑎 ∧ 𝜓 𝐿 ∧ 𝛾5 𝛾 𝑎 𝐷𝛿 𝑔 𝜓 𝑅 + ℎ.𝑐.
2 2
𝑖
= 𝑔 ∗ 𝑀 𝑃2 𝑒 𝑎 ∧ 𝑒 𝑏 ∧ 𝜖 𝐿 𝛾𝑏 𝛾5 𝛾 𝑎 𝐷𝜓 𝐿
4


− 4𝑖 𝑔 ∗ 𝑀 𝑃2 𝑒 𝑎 ∧ 𝜓 𝐿 ∧ 𝛾5 𝛾 𝑎 𝐷 𝑒 𝑏 𝛾𝑏 𝜖 𝐿 + ℎ.𝑐.
𝑖 ∗ 2 𝑎 𝑖
= 𝑔 𝑀 𝑃 𝑒 ∧ 𝑒 𝑏 ∧ 𝜖 𝐿 𝛾5 𝛾 𝑎𝑏 𝐷𝜓 𝐿 − 𝑔 ∗ 𝑀 𝑃2 𝑒 𝑎 ∧ 𝐷𝑒 𝑏 ∧ 𝜓 𝐿 𝛾5 𝛾 𝑎 𝛾𝑏 𝜖 𝐿
4 4
𝑖
− 𝑔 ∗ 𝑀 𝑃2 𝑒 𝑎 ∧ 𝑒 𝑏 ∧ 𝜓 𝐿 ∧ 𝛾5 𝛾 𝑎𝑏 𝐷𝜖 𝐿 + ℎ.𝑐.
4
(82)
Integrating by parts the first term in the last equality, we get
 
𝑖 1
𝛿 𝑔 L 𝑅𝑆 = 𝑔 ∗ 𝑀 𝑃2 𝑒 𝑎 ∧ 𝐷𝑒 𝑏 ∧ 𝜖 𝐿 𝛾5 𝛾 𝑎𝑏 𝜓 𝐿 − 𝜓 𝐿 𝛾5 𝛾 𝑎 𝛾𝑏 𝜖 𝐿
2 2
𝑖 ∗ 2 𝑎
− 𝑔 𝑀 𝑃 𝑒 ∧ 𝑒 𝑏 ∧ 𝜓 𝐿 ∧ 𝛾5 𝛾 𝑎𝑏 𝐷𝜖 𝐿 + ℎ.𝑐. (83)
2
The first term, proportional to 𝐷𝑒 𝑎 , plays a similar role as in the case without a
cosmological constant and will be discussed later after equation (90). Since the
remaining terms are proportional to the derivative of the supersymmetry parameter,
we can try and use the supersymmetry transformation rule of the gravitini, 𝛿𝜓 𝐿 =
𝑀 𝑃 𝐷𝜖 𝐿 + O(𝑔), to cancel them. For this reason we add a mass-like term to the
Lagrangian:
𝑖 ∗ 𝑖
LM𝜓 = 𝑔 𝑀 𝑃 𝑒 𝑎 ∧ 𝑒 𝑏 ∧ 𝜓 𝐿 ∧ 𝛾5 𝛾 𝑎𝑏 𝜓 𝐿 + 𝑔 𝑀 𝑃 𝑒 𝑎 ∧ 𝑒 𝑏 ∧ 𝜓 𝑅 ∧ 𝛾5 𝛾 𝑎𝑏 𝜓 𝑅 , (84)
4 4
so that the variation of 𝜓 in (84) compensates for (83) at order 𝑔. We point out here
the reduced Planck mass factors, so that the whole coefficient has mass dimension
1, as well as the overall factor 1/4, which is due to the double variation required to
match (83).
While the introduction of (84) allows the cancellation of the 𝐷𝜀 terms in (83), it
also gives rise to two further new variations we have to take care of. One variation
comes at order 𝑔 from the variation of the vierbeine in (84). We will discuss its
cancellation together with the cancellation of the 𝐷𝑒 𝑎 terms in (83) further below.
The other new variation of (84) is of order 𝑔 2 and arises when the order 𝑔 shift term
in the gravitino transformation is used in the variation of the gravitini in (84). Indeed,
the order 𝑔 2 variation of the gravitino mass term produces (suppressing the wedges)
𝑖  
𝛿 𝑔 L M 𝜓 = − |𝑔| 2 𝑀 𝑃3 𝑒 𝑎 𝑒 𝑏 𝜓 𝐿 𝛾5 𝛾 𝑎𝑏 𝑒 𝑐 𝛾𝑐 𝜖 𝑅 − 𝜖 𝑅 𝑒 𝑐 𝛾𝑐 𝛾5 𝛾 𝑎𝑏 𝜓 𝐿 + ℎ.𝑐. (85)
8
Putting together the gamma matrices and using the duality relation 𝛾5 𝛾 𝑎𝑏𝑐 =
−𝑖 𝜖 𝑎𝑏𝑐𝑑 𝛾 𝑑 we obtain
Simple Supergravity 23

|𝑔| 2 𝑎 𝑏 𝑐  
𝛿 𝑔 L M 𝜓 = 𝑀 𝑃3 𝑒 𝑒 𝑒 𝜖 𝑎𝑏𝑐𝑑 𝜓 𝐿 𝛾 𝑑 𝜖 𝑅 + 𝜓 𝑅 𝛾 𝑑 𝜖 𝐿
4
2 𝜖 𝑅 𝛾 𝑑 𝜓𝐿 + 𝜖 𝐿 𝛾 𝑑 𝜓𝑅
4 |𝑔|
= −𝑀 𝑃 𝑒 𝑎 𝑒 𝑏 𝑒 𝑐 𝜖 𝑎𝑏𝑐𝑑
2 2𝑀 𝑃
2 (86)
|𝑔|
= −𝑀 𝑃4 𝑒 𝑎 𝑒 𝑏 𝑒 𝑐 𝜖 𝑎𝑏𝑐𝑑 𝛿𝑒 𝑑
2
𝑀4  
= −|𝑔| 2 𝑃 𝛿 𝑒 𝑎 𝑒 𝑏 𝑒 𝑐 𝑒 𝑑 𝜖 𝑎𝑏𝑐𝑑 .
8

We further recall from Appendix 5.2 that 𝑒 𝑎 𝑒 𝑏 𝑒 𝑐 𝑒 𝑑 𝜖 𝑎𝑏𝑐𝑑 = +4! 𝑑 4 𝑥 𝑒, and then
realize that we need to add a single term of order |𝑔| 2 to the Lagrangian to cancel
(86): ∫ ∫
3 𝑑 4 𝑥 𝑒 𝑀 𝑃4 |𝑔| 2 = −𝑀 𝑃2 𝑑 4 𝑥 𝑒 Λ. (87)

This is a cosmological constant term. Notice that there is no choice of the sign of
this cosmological constant
3
Λ = −3 𝑀 𝑃2 |𝑔| 2 = − < 0. (88)
ℓ2
This agrees with the discussion following from the supersymmetry algebra.
It is also extremely important to note that the variation of (87) does not generate
terms of order 𝑔 3 and that therefore supersymmetry closes at order 𝑔 2 .
The only variation left is the vierbein variation in the gravitino mass term together
with the already mentioned first term in (83). Using

4(𝜓 𝑅 ∧ 𝛾 𝑎𝑏 𝜓 𝑅 ) ∧ (𝜀𝛾 𝑏 𝜓) = 3(𝜓 𝑅 𝛾 𝑎𝑏 𝜀 𝑅 ) ∧ (𝜓 ∧ 𝛾 𝑏 𝜓) + (𝜓 𝑅 𝜀 𝑅 ) ∧ (𝜓 ∧ 𝛾 𝑎 𝜓), (89)

which one can derive from the Fierz identities (212) - (216), one obtains for all
remaining uncancelled variations

𝛿L =
!   
𝑀𝑃 𝑎 1 𝑎 ∗ 𝑏 1
𝐷𝑒 − 𝜓 ∧ 𝛾 𝜓 ∧ 𝑖 𝑔 𝑀 𝑃 𝑒 ∧ 𝜖 𝐿 𝛾 𝑏𝑎 𝜓 𝐿 − 𝜓 𝛾 𝑏 𝛾 𝑎 𝜖 𝐿
2 4𝑀 𝑃2 2 𝐿
 
1
−𝑖 𝑔 𝑀 𝑃 𝑒 𝑏 ∧ 𝜖 𝑅 𝛾𝑏𝑎 𝜓 𝑅 − 𝜓 𝑅 𝛾𝑏 𝛾 𝑎 𝜖 𝑅 (90)
2
 
1 𝑏𝑐𝑑 𝑏𝑐 𝑑
+𝜖 𝑎𝑏𝑐𝑑 − 𝐷𝜓𝛾 𝜖 + 𝑀 𝑃 𝛿𝜔 ∧ 𝑒 ,
6

where the last term proportional to 𝜀 𝑎𝑏𝑐𝑑 is the same as in the case without cosmo-
logical constant.
This completes the proof of the invariance of the action in any of the formalisms
described above. In the second order formalism this variation vanishes because of
24 Gianguido Dall’Agata and Marco Zagermann

the torsion constraint. In the first order formalism we deduce from this variation the
expression for 𝛿𝜔 𝑎𝑏 that makes it vanish. Finally, in the 1.5 formalism, the equations
of motion for the spin connection do not change and hence once again the full
Lagrangian is invariant under supersymmetry.
Bringing the mass term (84) to the standard form without differential forms, the
final Lagrangian is therefore the following

𝑀 𝑃2 𝑒 𝑒
L= 𝑒𝑅 − 𝜓 𝜇𝑅 𝛾 𝜇𝜈𝜌 𝐷 𝜈 𝜓 𝜌𝐿 − 𝜓 𝜇𝐿 𝛾 𝜇𝜈𝜌 𝐷 𝜈 𝜓 𝜌𝑅
2 2 2
𝑔 𝑔∗
−𝑒 𝑀 𝑃 𝜓 𝜇𝑅 𝛾 𝜇𝜈 𝜓 𝜈𝑅 − 𝑒 𝑀 𝑃 𝜓 𝛾 𝜇𝜈 𝜓 𝜈𝐿 (91)
2 2 𝜇𝐿
+3 𝑒 𝑀 𝑃4 |𝑔| 2 ,

where we should remember that in the second order formalism 𝜔 𝑎𝑏 = 𝜔 𝑎𝑏 (𝑒, 𝜓).
The final supersymmetry transformations are
1
𝛿𝑒 𝑎𝜇 = 𝜖 𝐿 𝛾 𝑎 𝜓 𝜇𝑅 + ℎ.𝑐., (92)
2𝑀 𝑃
𝑔 2
𝛿𝜓 𝜇𝐿 = 𝑀 𝑃 𝐷 𝜇 𝜖 𝐿 − 𝑀 𝛾 𝜇 𝜖𝑅 . (93)
2 𝑃
Now that we completed the construction of a supersymmetric action for super-
gravity with a (negative) cosmological constant, we can make some comments.
First of all, we have seen from the construction that we have performed only mini-
mal modifications. After shifting the supersymmetry transformation of the gravitino
field, we introduced the smallest set of terms needed to cancel O(𝑔) and O(𝑔 2 )
terms in the supersymmetry variations. As already pointed out above, supersym-
metry closes at order |𝑔| 2 . There is no need to introduce any term of order 𝑔 3 or
more.
All the modifications can be summarized in three main pieces:
• a shift in the fermionic supersymmetry rules at O(𝑔);

• a mass-like term of O(𝑔) for the fermions;

• a potential term at order O(𝑔 2 ).

Although these modifications have been forced by the presence of the cosmological
constant in a pure gravity theory, one finds the pattern outlined above also in all
gauged supergravity theories for models with extended supersymmetries. In fact, in
extended supergravities the appearance of non-abelian gauge groups is tied to the
presence of a non-trivial scalar potential, which may act as an effective cosmological
constant. The result is that the gauging procedure introduces the same three main
modifications listed above, where the mass-like term for the fermions in the gen-
eral case with scalar fields becomes a Yukawa-like coupling, and the cosmological
constant term becomes a scalar potential.
Simple Supergravity 25

It can also be seen that this scalar potential (in this case a pure cosmological
constant) can be expressed as the square of the shifts of the supersymmetry variations
of the fermionic fields:
3
𝑉 𝜖 𝐿 𝛾 𝑎 𝜖 𝑅 = −3𝑀 𝑃4 |𝑔| 2 𝜖 𝐿 𝛾 𝑎 𝜖 𝑅 = − 𝛿 𝑔 𝜓 𝜇𝑅 𝛾 𝑎 𝛿 𝑔 𝜓 𝐿 .
𝜇
(94)
2
Note the minus sign in front of the squared gravitino shifts. This identity is called
the supersymmetric Ward identity [28].

4 Matter couplings in supergravity13

Once the construction of the pure supergravity theory has been completed, we can
try and analyze what needs to be done to couple matter multiplets in a consistent way.
This is what we are going to discuss in this section. In our presentation, we will limit
as much as possible the details of the derivations and focus instead on the new features
of matter couplings in supergravity as compared to what is already required by rigid
supersymmetry. In fact, already relaxing the requirement of studying renormalizable
interactions generalizes quite a lot the possible couplings of globally supersymmetric
theories, without the need of resorting to (super)gravity. We would therefore like in
the following to pinpoint the signatures that are unique to the supersymmetrization
of the gravitational interaction.
The discussion on the matter couplings clearly depends heavily on what kind of
matter we allow in these couplings and if we allow more than 2-derivative terms.
More general Lagrangians could be obtained by introducing additional matter multi-
plets, such as tensor multiplets. The tensor fields in such tensor multiplets, however,
can in general be dualized to either massless scalar or massive vector fields so that the
theory will be eventually of the standard form we will write down. On the other hand,
these dualities are often non-perturbative or may require complicated field redefini-
tions. It may therefore be interesting to study these models directly with tensor fields,
also because such tensor fields naturally arise from string theory compactifications,
but we will not discuss this here.
Other generalizations may include higher degree form fields, higher spin fields
and/or higher derivative terms. While for each of these generalizations one can work
out the construction that successfully leads to supersymmetric theories, we will stick
to the simplest minimal theory coupled to vector and chiral multiplets, which are
enough to provide rich theories with matter and gauge couplings.

In the following, we denote the complex scalars of 𝑛𝐶 chiral multiplets by 𝜙 𝑚

(𝑚, 𝑛, . . . = 1, . . . , 𝑛𝐶 ) and their complex conjugates by 𝜙 𝑚 and use 𝜒𝐿𝑚 and 𝜒𝑅𝑚 for
the chiral projections of their fermionic superpartners. The component fields of 𝑛𝑉
vector multiplets are the gaugini, 𝜆 𝐼 , and the vector fields, 𝐴 𝐼𝜇 (𝐼, 𝐽, . . . = 1, . . . , 𝑛𝑉 ).

13 This section is reprinted from [1] © 2021 Springer-Verlag GmbH Germany, part of Springer
Nature. Reproduced with permissions. All rights reserved.
26 Gianguido Dall’Agata and Marco Zagermann

General N = 1 globally supersymmetric theories of chiral and vector multiplets

are completely specified by the following data
• The numbers, 𝑛𝐶 and 𝑛𝑉 , of the chiral and vector multiplets;
• The Kähler potential 𝐾 (𝜙 𝑚 , 𝜙 𝑚 ) that determines the geometry of the scalar man-
ifold, Mscalar ;
• The holomorphic superpotential 𝑊 (𝜙 𝑚 ) that encodes the self-interactions of the
chiral multiplets;
• The holomorphic gauge kinetic function 𝑓 𝐼 𝐽 (𝜙 𝑚 ) related to the kinetic terms of
the vector multiplets;
• The action of the gauge group on Mscalar , as specified by the holomorphic Killing
vectors, 𝜉 𝐼𝑚 (𝜙 𝑛 ), and the corresponding Killing prepotentials, P𝐼 (𝜙 𝑚 , 𝜙 𝑚 );
• The real Fayet–Iliopoulos terms, 𝜂 𝐼 , which might be non-zero for Abelian gauge
group factors.
When we couple such a theory to supergravity, making it locally supersymmet-
ric, there will be additional couplings of the matter multiplets to the supergravity
multiplet, but also new and modified couplings among the fields of the matter mul-
tiplets themselves [29, 30, 31]. All these additional or modified couplings are still
completely specified by the above-mentioned data that already specified a theory in
global supersymmetry. As we will now explain, they will appear in the Lagrangian
with inverse powers of 𝑀 𝑃 .

4.1 Coupling chiral multiplets to supergravity

Let us start with the modifications that are necessary in order to make a globally
supersymmetric Wess–Zumino model also invariant under local supersymmetry. The
globally supersymmetric Lagrangian in Minkowski space is
h i
L𝑊 𝑍 = −𝑔𝑚𝑛 (𝜕𝜇 𝜙 𝑚 ) (𝜕 𝜇 𝜙 𝑛 ) + 𝜒 𝑚
𝐿 /
D 𝜒𝑅
𝑛
+ 𝜒 𝑛 / 𝑚
𝑅 D 𝜒 𝐿

∗
−(D𝑚 𝜕𝑛 𝑊) 𝜒 𝑚 𝑛 𝑚 𝑛
𝐿 𝜒𝐿 − (D𝑚 𝜕𝑛 𝑊 ) 𝜒 𝑅 𝜒𝑅 (95)

−𝑔 𝑚𝑛 (𝜕𝑚𝑊) (𝜕𝑛 𝑊 ∗ ) + O( 𝜒4 ),

where 𝑔𝑚𝑛 = 𝜕𝑚 𝜕𝑛 𝐾 is the Kähler metric. The derivative D is covariant with

respect to arbitrary holomorphic scalar field reparameterizations and hence contains
𝑝
the Christoffel symbols, Γ𝑚𝑛 , on the scalar manifold, e.g.,

D 𝜇 𝜒𝐿𝑚 ≡ 𝜕𝜇 𝜒𝐿𝑚 + (𝜕𝜇 𝜙 𝑛 )Γ𝑚 𝑙

𝑛𝑙 𝜒𝐿 . (96)
Simple Supergravity 27

If we now allow for a spacetime dependent supersymmetry parameter, 𝜖 = 𝜖 (𝑥), the

derivative in the kinetic terms of the fermions 𝜒 𝑚 will produce new terms when it
acts on 𝜖 (𝑥) coming from the supersymmetry transformations of the chiral fermions
1 𝑚 1 𝑚𝑛
𝛿 𝜒𝐿𝑚 = 𝜕/𝜙 𝜖 𝑅 − 𝑔 (𝜕𝑛𝑊 ∗ )𝜖 𝐿 + O( 𝜒 𝜒𝜖), (97)
2 2
1 1 𝑚𝑛
𝛿 𝜒𝑅𝑚 = 𝜕/𝜙 𝑚 𝜖 𝐿 − 𝑔 (𝜕𝑛 𝑊)𝜖 𝑅 + O( 𝜒 𝜒𝜖). (98)
2 2
The result is an uncancelled variation of the form
𝜇 𝜇
𝛿LWZ = 𝐽 𝑅 𝜕𝜇 𝜖 𝑅 + 𝐽 𝐿 𝜕𝜇 𝜖 𝐿 , (99)

where the supercurrents are

𝜇 𝜇 𝜇
𝐽 𝐿 = −𝑔𝑚𝑛 𝜒 𝑚
𝐿𝛾 𝜕 𝜙 + 𝜒 𝑛𝑅 𝛾 𝜇 𝜕𝑛 𝑊 ∗ ,
𝜇/ 𝑛
𝐽𝑅 = (𝐽 𝐿 ) 𝑐 . (100)

As we have already shown in section 2.1 for the special case of a free Wess–Zumino
model, the cancellation of these terms is achieved by adding the Noether couplings
to the gravitino
1 h 𝜇 𝜇
i
LNoether = − 𝐽 𝑅 𝜓 𝜇𝑅 + 𝐽 𝐿 𝜓 𝜇𝐿 . (101)
𝑀𝑃
Using 𝛿𝜓 𝜇 = 𝑀 𝑃 𝜕𝜇 𝜖, one then finds that everything cancels modulo terms that
come from the variation of the supercurrents themselves:
1 h 𝜇
i
𝛿(L𝑊 𝑍 + LNoether ) = − (𝛿𝐽 𝑅 )𝜓 𝜇𝑅 + h.c. (102)
𝑀𝑃
These terms are of the form

𝛿(L𝑊 𝑍 + LNoether ) = −𝛿𝑔 𝜇𝜈 𝑇𝜇𝜈 + 𝑍1 + 𝑍2 . (103)

Just as discussed in chapter 2.1, 𝑇𝜇𝜈 is the energy momentum tensor of L𝑊 𝑍 , and
the new field 𝑔 𝜇𝜈 is identified with the spacetime metric, signalling the necessity
for a coupling to gravity. The minimal coupling to a dynamical metric is achieved
by covariantizing everything with respect to general spacetime coordinate and local
Lorentz-transformations and by adding the pure supergravity Lagrangian. The metric
variation of this covariantized Lagrangian then precisely cancels the first term in
(103), and the theory would be supersymmetric if there weren’t also the two additional
terms 𝑍1 and 𝑍2 in eq. (103) that we have neglected so far. As we will now show, these
two terms are actually quite important, as they lead to additional 𝑀 𝑃−2 -suppressed
interactions between the fields of the chiral multiplets themselves that have some
far-reaching consequences.
In order to make this more precise, let us first state what 𝑍1 and 𝑍2 are:
28 Gianguido Dall’Agata and Marco Zagermann
𝑒
𝑍1 = − 𝑔𝑚𝑛 𝜓 𝜇 𝛾 𝜇𝜈𝜌 𝛾5 𝜖 (𝜕𝜈 𝜙 𝑚 ) (𝜕𝜌 𝜙 𝑛 ), (104)
2𝑀 𝑃
𝑒  
𝑍2 = 𝜓 𝜇𝐿 𝛾 𝜇𝜈 𝜖 𝐿 (𝜕𝜈 𝑊 ∗ ) + 𝜓 𝜇𝑅 𝛾 𝜇𝜈 𝜖 𝑅 (𝜕𝜈 𝑊) . (105)
𝑀𝑃

The first term 𝑍1 comes from the variations of the form 𝛿 𝜒𝐿𝑚 ∼ 12 𝜕/ 𝜙 𝑚 𝜖 𝑅 in 𝐽 𝐿 and
𝜇

its conjugate, which give rise to terms with three antisymmetrized gamma matrices
as well as terms with one gamma matrix. The latter are part of the energy momentum
tensor terms in (103) (because 𝛿𝑔 𝜇𝜈 involves only one gamma matrix), whereas the
terms with three antisymmetrized gamma matrices are precisely given by 𝑍1 . The
first term of 𝑍2 is due to the variations 𝛿 𝜒𝐿𝑚 ∼ − 12 𝑔 𝑚𝑛 (𝜕𝑛𝑊 ∗ )𝜖 𝐿 in the first term in
(100), and due to the variation 𝛿 𝜒𝑅𝑚 ∼ 12 𝜕/𝜙 𝑚 𝜖 𝐿 in the second term in (100). The
𝜇
second term in 𝑍2 arises from the analogous variations of 𝐽𝑅 .
We will now see that the cancellation of 𝑍1 and 𝑍2 requires the introduction of
new terms with important consequences.

4.2 The Kähler covariant derivative

In order to cancel 𝑍1 , we first rewrite it by using the relation between the metric of
the scalar manifold and the Kähler potential:
 
𝑍1 ∼𝜓 𝜇 𝛾 𝜇𝜈𝜌 𝛾5 𝜖 (𝜕𝜈 𝜙 𝑚 ) 𝜕𝜌 𝜙 𝑛 𝜕𝑚 𝜕𝑛¯ 𝐾
(106)
1

=𝜓 𝜇 𝛾 𝜇𝜈𝜌 𝛾5 𝜖 𝜕𝜌 𝜙 𝑛¯ 𝜕𝜈 𝜕𝑛¯ 𝐾 − 𝜕𝜌 𝜙 𝑚 𝜕𝜈 𝜕𝑚 𝐾 .
2
Using the last expression, we then integrate by parts the spacetime derivative that
acts on the Kähler potential. This produces in particular terms where the derivative
acts on 𝜖 and terms where it acts on 𝜓 𝜇 . The former term is

𝑒 1

𝜓 𝛾 𝜇𝜈𝜌 𝛾5 (𝐷 𝜈 𝜖) 𝜕𝜌 𝜙 𝑛¯ 𝜕𝑛¯ 𝐾 − 𝜕𝜌 𝜙 𝑚 𝜕𝑚 𝐾 . (107)
2𝑀 𝑃 𝜇 2
We now repeat our old trick and simply add the negative of this term (times a factor
1/2) to the Lagrangian, but with 𝐷 𝜈 𝜖 replaced by 𝜓 𝜈 ,
!
𝑒 𝑖
LKähler cov = − 𝜓 𝜇 𝛾 𝜇𝜈𝜌 𝑄 𝜈 (𝜙)𝛾5 𝜓 𝜌 , (108)
2 2𝑀 𝑃2

where 𝑄 𝜈 is a composite vector field,

𝑖 h i
𝑄 𝜈 (𝜙) ≡ (𝜕𝑛 𝐾)𝜕𝜈 𝜙 𝑛 − (𝜕𝑚 𝐾)𝜕𝜈 𝜙 𝑚 . (109)
2
Varying the two gravitini in this expression would then precisely cancel (107).
Simple Supergravity 29

The cancellation of the remaining term in 𝑍1 , where the derivative acts on 𝜓 𝜇 ,

will be discussed later (see footnote 14 in this section).
The new interaction term LKähler cov , however, now poses another problem: as
one easily verifies, it is not invariant under Kähler transformations 𝐾 → 𝐾 + ℎ + ℎ ∗ .
LKähler cov would thus seem to single out a particular Kähler potential, even though a
specific Kähler potential is not an intrinsic geometrical object on a Kähler manifold.
In general, the Kähler potential is in fact only locally defined and requires Kähler
transformations on the overlaps of local coordinate patches. So if the Lagrangian
was not Kähler invariant, the physics would in general also be different for different
coordinate patches of the scalar manifold.
To understand the resolution of this problem, we observe that the term LKähler cov
can be absorbed into the Rarita–Schwinger action by modifying the covariant deriva-
tive with a new term,
𝑒
LRS + LKähler cov = − 𝜓 𝜇 𝛾 𝜇𝜈𝜌 D𝜈 (𝜔, 𝑄)𝜓 𝜌 , (110)
2
where
𝑖
D [𝜈 (𝜔, 𝑄)𝜓 𝜌] ≡ 𝐷 [𝜈 (𝜔)𝜓 𝜌] + 𝑄 [𝜈 𝛾5 𝜓 𝜌] , (111)
2𝑀 𝑃2
with 𝐷 𝜈 being the Lorentz covariant derivative. To understand the significance of
this modification, one notes that 𝑄 𝜇 transforms under Kähler transformations like a
U(1) connection:
𝑄 𝜇 → 𝑄 𝜇 + 𝜕𝜇 Im(ℎ). (112)
More precisely, 𝑄 𝜇 is a composite U(1) connection, i.e., it is not an elementary
vector field, but rather a function of the scalar fields and their derivatives.
We now see that we can render the Lagrangian invariant if we require that Kähler
transformations, 𝐾 → 𝐾 + ℎ + ℎ ∗ , be accompanied by chiral rotations of the gravitino:
" #
𝑖
𝜓 𝜇 → exp − Im(ℎ(𝜙))𝛾5 𝜓 𝜇 . (113)
2𝑀 𝑃2

Indeed, the derivative (111) then transforms covariantly,

" #
𝑖
D [𝜇 𝜓 𝜌] → exp − Im(ℎ(𝜙))𝛾5 D [𝜇 𝜓 𝜌] , (114)
2𝑀 𝑃2

and the combination (110) is Kähler-(and obviously also locally Lorentz-) invariant.
These geometric arguments thus suggest that, in supergravity, Kähler transformations
on the scalar manifold also act on the gravitino as a chiral U(1) symmetry, with 𝑄 𝜇
being the corresponding (composite) U(1) connection. If this is to make sense, this
nontrivial action of Kähler transformations on the gravitini should also be compatible
with supersymmetry. As we will now show, this requirement will lead to further
interesting differences with respect to global supersymmetry and provides further
consistency checks.
30 Gianguido Dall’Agata and Marco Zagermann

First we note that if the gravitino transforms under Kähler transformations, the
consistency with the supersymmetry transformation law 𝛿𝜓 𝜇 ∼ 𝑀 𝑃 𝐷 𝜇 𝜖 also re-
quires that 𝜖 transforms under Kähler transformations,
" #
𝑖
𝜖 → exp − Im(ℎ(𝜙))𝛾5 𝜖 (115)
2𝑀 𝑃2

and that its derivative (as it appears in 𝛿𝜓 𝜇 ) should also be covariantized, 14

𝑖
D 𝜇 (𝜔, 𝑄)𝜖 ≡ 𝐷 𝜇 (𝜔)𝜖 + 𝑄 𝜇 𝛾5 𝜖. (116)
2𝑀 𝑃2

This in turn implies, because of 𝛿 𝜒𝐿𝑚 = 21 𝜕/ 𝜙 𝑚 𝜖 𝑅 + . . ., that also the chiral fermions
transform under Kähler transformations,
" #
𝑚 𝑖
𝜒 → exp + Im(ℎ(𝜙))𝛾5 𝜒 𝑚 , (117)
2𝑀 𝑃2

and that their derivatives have to be Lorentz-, Mscalar -reparameterization- and

Kähler-covariant, e.g., 15
𝑖
D 𝜇 𝜒𝐿𝑚 ≡ 𝐷 𝜇 𝜒𝐿𝑚 + (𝜕𝜇 𝜙 𝑛 )Γ𝑚 𝑙
𝑛𝑙 𝜒𝐿 − 𝑄 𝜇 𝜒𝐿𝑚 . (118)
2𝑀 𝑃2

Note that there is a different sign in (117) (and hence also in (118)) compared
to the corresponding terms of the gravitino or the supersymmetry transformation
parameter (cf. (113) and (115) as well as (111) and (116)). This sign difference
arises because one has to move the 𝛾5 matrix in (115) through one gamma matrix in
the supersymmetry transformation 𝛿 𝜒𝐿𝑚 = 12 𝜕/𝜙 𝑚 𝜖 𝑅 + . . ..
Although we will discuss gauge multiplets later, we already mention here that
𝛿𝜆 𝐼 ∼ 14 𝛾 𝜇𝜈 F𝜇𝜈
𝐼 𝜖 + . . . implies that also the gaugini transform nontrivially under

Kähler transformations (with the same sign as 𝜓 𝜇 and 𝜖)

" #
𝐼 𝑖
𝛿𝜆 → exp − Im(ℎ(𝜙))𝛾5 𝜆 𝐼 , (119)
2𝑀 𝑃2

and that likewise all their derivatives have to be properly covariantized with respect
to Kähler transformations (again with the same sign as for 𝜓 𝜇 and 𝜖).
To conclude, all fermion fields and not just the gravitino are charged with respect
to a composite chiral U(1) symmetry that is related to Kähler transformations and that

14 This is indeed confirmed by computing the gravitino variations of L 𝑅𝑆 with the new Kähler
covariant transformation law, 𝛿 𝜓𝜇 ∼ 𝑀 𝑃 D 𝜇 𝜖 , which leads to a new term that precisely cancels
the remaining uncancelled part of 𝑍1 (i.e. the part of 𝑍1 with a derivative acting on 𝜓 𝜇 ).
15 For the sake of simplicity, we do not introduce a new symbol for the Kähler covariantized
derivative and still call it D 𝜇 .
Simple Supergravity 31

is not present in the global case. It should be emphasized that in the limit of global
supersymmetry, 𝑀 𝑃 → ∞, these chiral rotations become trivial, as is signalled by the
inverse powers of 𝑀 𝑃 . This is consistent with the rigid supersymmetry Lagrangian
(95), where this chiral composite U(1) is not encountered.
Interestingly, the above non-trivial transformations of the fermions under Käh-
ler transformations also imply that the superpotential and its derivatives have to
transform as we will show in section 4.3 below. The result is that
" #
1
𝑊 → exp − 2 ℎ(𝜙 ) 𝑊 (𝜙 𝑚 )
𝑚
(120)
𝑀𝑃

and its derivatives have to be Kähler covariantized as follows

" #
𝐾 𝐾
2𝑀 2 2𝑀 2
(𝜕𝑛 𝐾)
𝜕𝑛𝑊 → 𝑒 𝑃 D𝑛 𝑊 ≡ 𝑒 𝑃 𝜕𝑛 + 𝑊. (121)
𝑀 𝑃2

To summarize: The cancellation of 𝑍1 by adding LKähler cov gives rise to the

interpretation that the fermions and the superpotential should transform non-trivially
under Kähler transformations. In order to ensure this, all derivatives of the fermions
and the superpotential have to be Kähler covariantized, and the superpotential terms
have to be dressed with an exponential of the Kähler potential. One can show that all
these modifications are indeed also necessary for the cancellation of various other
variations we have not discussed here in detail. In general, we define the Kähler
covariant derivatives in field space as
!
𝑝
𝐷 𝑚 Φ = 𝜕𝑚 + 2 𝜕𝑚 𝐾 Φ,
𝑀𝑃
! (122)
𝑝
𝐷 𝑚 Φ = 𝜕 𝑚 − 2 𝜕 𝑚 𝐾 Φ,
𝑀𝑃

where 𝑝 is the Kähler “charge” of the field Φ.

4.3 Additional bare superpotential terms

In global supersymmetry, all superpotential terms always appear with at least one
derivative with respect to the scalar fields. As we saw in the previous subsection,
the coupling to supergravity (in particular the cancellation of the term 𝑍1 ) requires
a Kähler covariantization of these derivatives of 𝑊, which then introduces “bare”
𝑊-terms inside these Kähler covariant derivatives, i.e., 𝑊-terms that are not differ-
entiated with respect to any scalar field. In this subsection, we show that there are
additional “bare” superpotential terms in the Lagrangian and the supersymmetry
32 Gianguido Dall’Agata and Marco Zagermann

transformation laws. Their necessity follows from the cancellation of the term 𝑍2 to
which we now turn.
In order to cancel the 𝑍2 -term,
𝑒  
𝑍2 ≡ 𝜓 𝜇𝐿 𝛾 𝜇𝜈 𝜖 𝐿 (𝜕𝜈 𝑊 ∗ ) + 𝜓 𝜇𝑅 𝛾 𝜇𝜈 𝜖 𝑅 (𝜕𝜈 𝑊) , (123)
𝑀𝑃
we proceed as we did for 𝑍1 and first perform an integration by parts. This will then
give again terms with a derivative acting on the supersymmetry parameter 𝜖 and
terms where the derivative acts on on the gravitini 𝜓 𝜇 . To cancel the former, we
then again add to the Lagrangian a term where the derivatives of 𝜖 are replaced by
gravitini, or, more precisely,
𝑒  ∗ 
2
𝑊 𝜓 𝜇𝐿 𝛾 𝜇𝜈 𝜓 𝜈𝐿 + 𝑊 𝜓 𝜇𝑅 𝛾 𝜇𝜈 𝜓 𝜈𝑅 . (124)
2𝑀 𝑃

This term is an obvious mass-like term for the gravitino and therefore, following
the rules we have learned in the case of pure supergravity in the presence of a
cosmological constant, we have to further modify the variation of the gravitino field
by adding a new term of the form
1
𝛿new 𝜓 𝜇𝐿 ∼ 𝑊 𝛾 𝜇 𝜖𝑅 . (125)
2𝑀 𝑃
This new variation applied to the Rarita–Schwinger action also gives the term re-
quired to cancel the second piece coming from the partial integration of 𝑍2 , namely
the term with the derivative acting on the gravitino.
Before proceeding further, let us come back to the Kähler covariantization of the
superpotential terms and prove (120). Subjecting (125) to Kähler transformations
tells us that the left hand side transforms as
" #
1 ∗
exp − (ℎ(𝜙) − ℎ (𝜙)) , (126)
4𝑀 𝑃2

while the epsilon parameter on the right hand side transforms with the opposite sign
due to the opposite chirality:
" #
1 ∗
exp + (ℎ(𝜙) − ℎ (𝜙)) . (127)
4𝑀 𝑃2

At this point it is obvious that in order for the Kähler transformation to be compatible
with supersymmetry, we need to transform also the superpotential, as we already
mentioned earlier. The superpotential, on the other hand, is a holomorphic function
by construction and hence can transform only with a holomorphic factor,
" #
𝛼
𝑊 → exp − 2 (ℎ(𝜙)) 𝑊, (128)
𝑀𝑃
Simple Supergravity 33

where 𝛼 is a real constant. In order to get the same rotation on the left and on
the right hand side of (125) we still need something that transforms under Kähler
transformations with the exponential of ℎ + ℎ∗ , like the exponential of the Kähler
2
potential itself, e𝛽𝐾 /𝑀𝑃 . The right coefficients follow then by equating the two sides:
1 1 𝛼 𝛽
− 2
(ℎ(𝜙) − ℎ∗ (𝜙)) = + 2
(ℎ(𝜙) − ℎ∗ (𝜙)) − 2 (ℎ(𝜙)) + 2 (ℎ(𝜙) + ℎ∗ (𝜙)).
4𝑀 𝑃 4𝑀 𝑃 𝑀𝑃 𝑀𝑃
(129)
This fixes 𝛼 = 1, 𝛽 = 1/2 and tells us that we have to replace the superpotential with
the combination
2
e𝐾 /(2𝑀𝑃 ) 𝑊 (130)
and that indeed 𝑊 transforms under Kähler transformations as in (120).
Coming back to the check of supersymmetry invariance, we now see that the
new transformation law for the gravitino (125) applied to the new bilinear term
(124), gives a new variation of the form |𝑊 | 2 𝜓𝛾𝜖. Not too surprisingly, this can
then finally be cancelled by adding a new contribution ∼ −𝑒|𝑊 | 2 to the scalar
potential and varying the vierbein determinant 𝑒. This is the generalization of the
procedure derived in section 3.3 for the case of a constant superpotential, i.e., for
pure supergravity with a cosmological constant.
Although it may be hard to believe, it turns out that, after proper Kähler covari-
antizations, the above modifications are sufficient to ensure also the cancellations of
all the other variations we have not considered explicitly here.
The end result is the Lagrangian

𝑀 𝑃2 1
𝑒 −1 L = 𝑅(𝑒, 𝜔(𝑒)) − 𝜓 𝜇 𝛾 𝜇𝜈𝜌 D𝜈 (𝜔(𝑒), 𝑄)𝜓 𝜌
2 2
h i
−𝑔𝑚𝑛 (𝜕𝜇 𝜙 𝑚 ) (𝜕 𝜇 𝜙 𝑛 ) + 𝜒 𝑚
𝐿D/ 𝜒𝑅𝑛 + 𝜒 𝑛𝑅 D
/ 𝜒𝐿𝑚
n 2
o
− 𝑒 𝐾 /2𝑀𝑃 (D𝑚 D𝑛 𝑊) 𝜒 𝑚 𝑛
𝐿 𝜒𝐿 + h.c. (131)
1 n 2
o
+ 𝑔𝑚𝑛 𝜓 𝜇𝐿 𝛾 𝜈 𝛾 𝜇 𝜒𝐿𝑚 (𝜕𝜈 𝜙 𝑛 ) + 𝜓 𝜇𝑅 𝛾 𝜇 𝜒𝐿𝑚 𝑒 𝐾 /2𝑀𝑃 D𝑚 𝑊 + h.c.
𝑀𝑃
1 n 𝐾 /2𝑀 2 𝜇𝜈
o
+ 𝑒 𝑃 𝑊𝜓
𝜇𝑅 𝛾 𝜓 𝜈𝑅 + ℎ.𝑐 − 𝑉 (𝜙 𝑚 , 𝜙 𝑛 ),
2𝑀 𝑃2

with the scalar potential given by the sum of two contributions

" #
2
𝐾 /𝑀𝑃 𝑚𝑛 ∗ 3|𝑊 | 2
𝑉=𝑒 𝑔 (D𝑚𝑊) (D𝑛 𝑊 ) − , (132)
𝑀 𝑃2

where the first term is the Kähler covariantization of the F-terms from global super-
symmetry and the second is a genuine contribution from gravitational couplings, in
the sense that it is a variation of the vierbein determinant that leads to a cancellation
34 Gianguido Dall’Agata and Marco Zagermann

of the |𝑊 | 2 terms mentioned after (130). In the Lagrangian (131), the first line is the
Kähler covariantization of the pure supergravity action. The second and third line
correspond to the Kähler and spacetime covariant Wess–Zumino action (without the
potential). Note that now
!" ! # !
𝜕𝑚 𝐾 𝜕𝑛 𝐾 𝜕 𝑝 𝐾
D𝑚 D𝑛 𝑊 = 𝜕𝑚 + 𝜕𝑛 + 2 𝑊 − Γ𝑚𝑛 𝑝 𝜕 𝑝 + 2 𝑊. (133)
𝑀 𝑃2 𝑀𝑃 𝑀𝑃

The fourth line is the Kähler and spacetime covariant

h 𝜇 Noethericoupling of the su-
1 𝜇
percurrents to the gravitino, LNoether = − 𝑀𝑃 𝐽 𝑅 𝜓 𝜇𝑅 + 𝐽 𝐿 𝜓 𝜇𝐿 (with the fermions
moved into a different order). The fifth line, finally, contains the 𝑊-dependent extra
terms as well as the (Kähler covariantized) scalar potential of the Wess–Zumino
model.
The supersymmetry transformation rules, up to 3-fermion terms, are
1
𝛿𝑒 𝑎𝜇 = 𝜖 𝛾𝑎𝜓𝜇,
2𝑀 𝑃
1 𝐾 /2𝑀 2
𝛿𝜓 𝜇𝐿 = 𝑀 𝑃 D 𝜇 (𝜔(𝑒), 𝑄 𝜈 )𝜖 𝐿 + 𝑒 𝑃 𝑊𝛾 𝜖 ,
𝜇 𝑅
2𝑀 𝑃 (134)
𝛿𝜙 𝑚 = 𝜖 𝐿 𝜒𝐿𝑚 ,
1 𝑚 1 2
𝛿 𝜒𝐿𝑚 = 𝜕/𝜙 𝜖 𝑅 − 𝑔 𝑚𝑛 𝑒 𝐾 /2𝑀𝑃 (D𝑛 𝑊 ∗ )𝜖 𝐿 .
2 2
Obviously, in the 𝑀 𝑃 → ∞ limit, these equations reduce to the globally supersym-
metric theory. One also notices that truncating out the chiral multiplets and keeping
𝐾
2
a constant superpotential e 2𝑀𝑃 𝑊 = −𝑔 𝑀 𝑃3 gives back the pure supergravity La-
grangian with cosmological constant, eq. (91).
Note further that in supergravity the Kähler potential and the superpotential are no
longer independent, as one can shift terms back and forth via Kähler transformations.
In fact, as long as 𝑊 is not equal to zero, one can even make the superpotential
equal to 𝑀 𝑃3 by performing a Kähler transformation with ℎ(𝜙) = 𝑀 𝑃2 log(𝑊/𝑀 𝑃3 )
(cf. eq. (120)). More generally, instead of using the two functions 𝐾 and 𝑊, one can
express the entire Lagrangian in terms of the function

|𝑊 | 2
G = 𝐾 + 𝑀 𝑃2 log , (135)
𝑀 𝑃6

which is manifestly Kähler invariant. For instance, the part of the scalar potential
coming from the superpotential becomes
2
 
𝑉 = 𝑒 G/𝑀𝑃 𝑀 𝑃2 𝑔 𝑚𝑛¯ G𝑚 G𝑛¯ − 3𝑀 𝑃4 . (136)

Note, however, that by doing so one cannot recover the 𝑊 = 0 case, which has to be
discussed separately. Hence the usefulness of leaving explicit both 𝐾 and 𝑊 in our
approach.
Simple Supergravity 35

4.4 Inclusion of vector multiplets

The inclusion of vector multiplets requires the following changes:

1. All terms that were already present in global supersymmetry are also present in
supergravity, but they all have to be made spacetime and Kähler covariant.
2. A new Noether coupling of the vector multiplet supercurrent
 
𝜇 𝐽 1 𝐼 𝜇 𝜈𝜌 𝑖 𝜇
𝐽 𝑉 𝑀 ≡ 𝑒𝜆 − (Re 𝑓 𝐼 𝐽 )F𝜈𝜌 𝛾 𝛾 − P𝐽 𝛾 𝛾5 (137)
4 2

to the gravitino has to be introduced:

′ 1 𝜇
LNoether =− 𝐽 𝜓𝜇 (138)
𝑀𝑃 𝑉 𝑀
in order to cancel terms of the form
𝜇
𝛿L = 𝐽 𝑉 𝑀 𝜕𝜇 𝜖 (139)
𝐼
that arise due to the derivative in − 12 𝑒 (Re 𝑓 𝐼 𝐽 )𝜆 b
𝜕/𝜆 𝐽 when it acts on the 𝜖 in 𝛿𝜆.
In the above equations, F𝜇𝜈𝐼 denotes the usual gauge covariant field strengths, and

a hat on a derivative denotes a gauge covariant derivative.

3. The composite Kähler connection 𝑄 𝜇 receives an additional contribution propor-
tional to 𝐴 𝐼𝜇 P𝐼 for each of the gauged isometries:

𝑖  
𝑄 𝜇 = 𝑄 𝜇 (𝜙 𝑚 , 𝜙 𝑛 , 𝐴 𝐼𝜇 ) = (𝜕𝑛 𝐾)𝜕𝜇 𝜙 𝑛 − (𝜕𝑚 𝐾)𝜕𝜇 𝜙 𝑚 + 𝐴 𝐼𝜇 P𝐼 (140)
2
𝑖  
= (𝜕𝑛 𝐾) 𝜕b𝜇 𝜙 𝑛 − (𝜕𝑚 𝐾) 𝜕b𝜇 𝜙 𝑚 + 𝐴 𝐼𝜇 Im(𝑟 𝐼 ), (141)
2
where the last equality follows from the form of the prepotentials (we will see
more on this in section 4.5). This additional term is needed, e.g., in order to
cancel a variation proportional to F𝜇𝜈𝐽 P 𝜖 𝛾 𝜇𝜈𝜌 𝛾 𝜓 that occurs in the variation
𝐽 5 𝜌
1 𝜇
− 𝑀𝑃 𝛿(𝐽𝑉 𝑀 )𝜓 𝜇 and is not of the form −𝛿𝑔 𝜇𝜈 𝑇𝜇𝜈 .
It should be noted that this additional contribution to 𝑄 𝜇 has another important
consequence. Namely, if one shifts the Killing prepotential P𝐼 of an Abelian
factor by a Fayet–Iliopoulos constant, P𝐼 → P𝐼 + 𝜂 𝐼 , one introduces new chiral
gauge interactions for all fermions, including, e.g., the gravitino,
𝑖
D𝜇 𝜓𝜈 → D𝜇 𝜓𝜈 + 𝐴 𝐼𝜇 𝜂 𝐼 𝛾5 𝜓 𝜈 , (142)
2𝑀 𝑃2

which can easily lead to quantum anomalies [32]. Thus, the introduction of Fayet–
Iliopoulos constants in N = 1 supergravity requires some care. We will actually
36 Gianguido Dall’Agata and Marco Zagermann

see later on that in supergravity the Fayet–Iliopoulos terms are related to the
non-invariance of the superpotential under gauge transformations.
Ignoring 4-fermion terms, the end result of all these modifications is the following
general matter coupled Lagrangian 16

𝑀 𝑃2 1
𝑒 −1 L = 𝑅(𝑒, 𝜔(𝑒)) − 𝜓 𝜇 𝛾 𝜇𝜈𝜌 D𝜈 (𝜔(𝑒), 𝑄)𝜓 𝜌
2 h 2 i
−𝑔𝑚𝑛 ( 𝜕b𝜇 𝜙 𝑚 ) ( 𝜕b𝜇 𝜙 𝑛 ) + 𝜒 𝑚 6 b 𝜒𝑅𝑛 + 𝜒 𝑛𝑅 D
𝐿 D 6 b 𝜒𝐿𝑚
 
1 𝐼 𝜇𝜈 𝐽 1 𝐼 b 𝐽
+(Re 𝑓 𝐼 𝐽 ) − F𝜇𝜈 F − 𝜆 D𝜆 6
4 2
1 h i
+ (Im 𝑓 𝐼 𝐽 ) F𝜇𝜈 𝐼
F𝜌𝜎𝐽 𝜇𝜈𝜌𝜎
𝜖 − 2𝑖 D b𝜇 (𝑒𝜆 𝐼 𝛾5 𝛾 𝜇 𝜆 𝐽 )
8
1 𝑖 𝐼
+ − 𝑓 𝐼 𝐽 ,𝑚 F𝜇𝜈 𝐼
𝜒𝑚 𝜇𝜈 𝐽
𝐿 𝛾 𝜆 𝐿 + 𝐷 𝑓 𝐼 𝐽 ,𝑚 𝜒 𝐿 𝜆
𝑚 𝐽
4 2
1 2 𝐼
+ 𝑒 𝐾 /2𝑀𝑃 (D𝑚 𝑊)𝑔 𝑚𝑛 𝑓 𝐼∗𝐽 ,𝑛 𝜆 𝑅 𝜆 𝑅𝐽
4 o
2 𝐼
−𝑒 𝐾 /2𝑀𝑃 (D𝑚 D𝑛 𝑊) 𝜒 𝑚 𝑛 𝑛 𝑚
𝐿 𝜒𝐿 − 2𝜉 𝐼 𝑔 𝑚𝑛 𝜆 𝜒𝐿 + h.c.
1
+ (Re 𝑓 𝐼 𝐽 )𝜓 𝜇 𝛾 𝜈𝜌 𝛾 𝜇 𝜆 𝐽 F𝜈𝜌
𝐼
4𝑀 𝑃
 
1 𝜈 𝜇 𝑚 b 𝑛
+ 𝑔𝑚𝑛 𝜓 𝜇𝐿 𝛾 𝛾 𝜒𝐿 ( 𝜕𝜈 𝜙 ) + ℎ.𝑐.
𝑀𝑃
   
1 𝜇 𝑖 𝐼 𝑚 𝐾 /2𝑀𝑃2
+ 𝜓 𝜇𝑅 𝛾 𝜆 P𝐼 + 𝜒 𝐿 𝑒 D𝑚𝑊 + ℎ.𝑐.
𝑀𝑃 2 𝐿
1 n 𝐾 /2𝑀 2 𝜇𝜈
o
+ 𝑒 𝑃 𝑊𝜓
𝜇𝑅 𝛾 𝜓 𝜈𝑅 + ℎ.𝑐 − 𝑉 (𝜙 𝑚 , 𝜙 𝑛 ), (143)
2𝑀 𝑃2

with the scalar potential

" #
2
𝐾 /𝑀𝑃 𝑚𝑛 3|𝑊 | 2 1
𝑉 =𝑒 𝑔 (D𝑚 𝑊) (D𝑛 𝑊) − + (Re 𝑓 𝐼 𝐽 )𝐷 𝐼 𝐷 𝐽 . (144)
𝑀 𝑃2 2

The supersymmetry transformation rules, up to 3-fermion terms, are

16 When the gauge kinetic function is not gauge invariant, so-called generalized Chern-Simons
terms of the form 𝐴𝐼 ∧ 𝐴𝐽 ∧ 𝑑 𝐴𝐾 and 𝐴𝐼 ∧ 𝐴𝐽 ∧ 𝐴𝐾 ∧ 𝐴𝐿 may be possible. Their form, however,
is the same as in global supersymmetry [33].
Simple Supergravity 37

1
𝛿𝑒 𝑎𝜇 = 𝜖 𝛾𝑎𝜓𝜇,
2𝑀 𝑃
1 𝐾 /2𝑀 2
𝛿𝜓 𝜇𝐿 = 𝑀 𝑃 D 𝜇 (𝜔(𝑒), 𝑄)𝜖 𝐿 + 𝑒 𝑃 𝑊𝛾 𝜖 ,
𝜇 𝑅
2𝑀 𝑃
𝛿𝜙 𝑚 = 𝜖 𝐿 𝜒𝐿𝑚 ,
1 b 𝑚 1 2
𝛿 𝜒𝐿𝑚 = 6 𝜕𝜙 𝜖 𝑅 − 𝑔 𝑚𝑛 𝑒 𝐾 /(2𝑀𝑃 ) (D𝑛 𝑊 ∗ )𝜖 𝐿 ,
2 2
1
𝛿 𝐴 𝐼𝜇 = − 𝜖 𝛾 𝜇 𝜆 𝐼 ,
2
1 𝑖
𝛿𝜆 𝐼 = 𝛾 𝜇𝜈 F𝜇𝜈 𝐼
𝜖 + 𝛾5 𝐷 𝐼 𝜖. (145)
4 2
It is again easy to see that, in the global limit, 𝑀 𝑃 → ∞, the above equations reduce
to the globally supersymmetric theory.
For completeness, we display the full (i.e., local Lorentz-, scalar coordinate-,
Kähler-, and gauge- covariant) derivative of 𝜆 𝐼 and 𝜒 𝑚 :

b𝜇 𝜒 𝑚 = 𝐷 𝜇 𝜒 𝑚 + ( 𝜕b𝜇 𝜙 𝑛 )Γ𝑚 𝜒𝑙 − 𝐴 𝐼𝜇 (𝜕𝑛 𝜉 𝑚 ) 𝜒 𝑛 − 𝑖

D 𝐿 𝐿 𝑛𝑙 𝐿 𝐼 𝐿 𝑄 𝜇 𝜒𝐿𝑚 ,
2𝑀 𝑃2
b𝜇 𝜆 𝐼 = 𝐷 𝜇 𝜆 𝐼 + 𝐴 𝜇𝐽 𝑓 𝐼 𝜆 𝐾 + 𝑖
D 𝐽𝐾 𝑄 𝜇 𝛾5 𝜆 𝐼 , (146)
2𝑀 𝑃2

where, as usual, 𝐷 𝜇 denotes the Lorentz covariant derivative. The full covariant
derivatives of 𝜓 𝜇 and 𝜖 are just as for 𝜆 𝐼 , except for the gauge covariantization term
𝐴 𝜇𝐽 𝑓 𝐽𝐼 𝐾 𝜆 𝐾 , which is absent for these fermions (hence, we can omit the hat on their
derivatives).
More details on the action and the 4-Fermi terms can be found in [34, 33].

4.5 More on D-terms

Although the D-terms and the D-term potential take the same form as in global super-
symmetry, local supersymmetry does have some interesting implications also for the
D-terms. To understand this, we recall that the general matter-coupled supergravity
Lagrangian is invariant under Kähler transformations that act at the same time on
the Kähler potential, the superpotential and the fermions. As in global supersymme-
try, a gauge transformation therefore does not necessarily have to leave the Kähler
potential invariant, but may in general transform it with a Kähler transformation,

𝛿gauge 𝐾 ≡ 𝜉 𝐼𝑚 𝜕𝑚 𝐾 + 𝜉 𝐼𝑚 𝜕𝑚 𝐾 = 𝑟 𝐼 + 𝑟 𝐼∗ . (147)

However, in supergravity theories this requires a non-trivial action also on the su-
perpotential (if 𝑊 ≠ 0)
38 Gianguido Dall’Agata and Marco Zagermann
𝑟𝐼
𝛿gauge𝑊 ≡ 𝜉 𝐼𝑚 𝜕𝑚𝑊 = − 𝑊, (148)
𝑀 𝑃2

so that the combination G in (135) remains invariant:

𝛿gauge G = 𝜉 𝐼𝑚 𝜕𝑚 G + 𝜉 𝐼𝑚¯ 𝜕𝑚¯ G = 0 (149)

For all the points in field space where 𝑊 ≠ 0 we can then also rewrite 𝑟 𝐼 as

𝜕𝑚𝑊
𝑟 𝐼 = −𝑀 𝑃2 𝜉 𝐼𝑚 , (150)
𝑊
so that the Killing prepotentials can be also expressed in terms of the gauge-invariant
quantity G:
" #
𝑀 𝑃2 𝜕𝑚𝑊 𝑀 2 D𝑚 𝑊
𝑚 𝑚
P𝐼 = 𝑖𝜉 𝐼 𝜕𝑚 𝐾 − 𝑖𝑟 𝐼 = 𝑖𝜉 𝐼 𝜕𝑚 𝐾 + = 𝑖𝜉 𝐼𝑚 𝑃 = 𝑖𝜉 𝐼𝑚 𝜕𝑚 G. (151)
𝑊 𝑊

The D-terms are thus

𝐷 𝐼 = 𝑖 (Re 𝑓 ) −1𝐼 𝐽 𝜉 𝐽𝑚 𝜕𝑚 G, (152)
and the total scalar potential with F-terms and D-terms can be written in a very
compact and suggestive form
2
h i
𝑉 = 𝑒 G/𝑀𝑃 ℎ 𝑚𝑛 G𝑚 G𝑛 − 3 𝑀 𝑃2 , (153)

where 2
𝑒 −G/𝑀𝑃
𝑚𝑛
ℎ ≡𝑔 + 𝑚𝑛
(Re 𝑓 ) −1𝐼 𝐽 𝜉 𝐼𝑚 𝜉 𝐽𝑛 , (154)
2
so that the new metric contains the Kähler metric giving the F-term potential (136)
and the additional term coming from the D-term potential.
We can now comment on some of the differences between global and local
supersymmetry, which may lead to relevant physical differences.
First of all, we see that in supergravity D-terms and F-terms are not independent
of one another, rather the D-terms are (for 𝑊 ≠ 0) a particular combination of the
F-terms (152).
The next important difference concerns the Fayet–Iliopoulos constants. Just as in
global supersymmetry, the gauge transformation (147) of the Kähler potential fixes
𝑟 𝐼 only up to an additive imaginary constant, 𝑖𝜂 𝐼 and hence P𝐼 up to an additive real
constant 𝜂 𝐼 . However, consistency of the P𝐼 moment maps with the gauge symmetry
requires the equivariance condition
1
{P𝐼 , P𝐽 } = (𝛿 𝐼 P𝐽 − 𝛿 𝐽 P𝐼 ) = 𝑓 𝐼 𝐽 𝐾 P𝐾 (155)
2
and this restricts the possible values of the 𝜂 𝐼 constants except for U(1) factors. The
difference to global supersymmetry now is that the superpotential 𝑊 also transforms
Simple Supergravity 39

under gauge transformations as in (148) so that a shift of 𝑟 𝐼 by an additive constant 𝑖𝜂 𝐼

implies that 𝑊 transforms with an additional phase factor under the corresponding
U(1) transformation. In other words, changing 𝜂 𝐼 changes the U(1) charge of 𝑊.
Note that such U(1) transformations due to FI constants may even occur when the
Kähler potential is invariant under this U(1) factor, because, according to (147), this
only implies 𝑟 𝐼 (𝜙) = 𝑖 𝜂 𝐼 .
Another important effect of a FI constant is that it leads to a chiral U(1) trans-
formation of the fermions, as follows from their non-trivial transformations under
Kähler transformations described in section 4.2. As explained around (142), this
may then easily lead to anomalous gauge couplings and requires some care.

Acknowledgements This review is based on the book [1], which grew out of the many interactions
we had in our careers. We would only like here to recall those who played a pivotal role in shaping
our understanding of supergravity: A. Ceresole, R. D’Auria, S. Ferrara, M. Günaydin, A. Van
Proeyen and F. Zwirner. We also thank once more L. Álvarez Gaumé, who first suggested to turn
our lecture notes into a book.

5 Appendix: Conventions, spinors and useful relations

5.1 General Relativity and spacetime conventions

We use 𝑥 𝜇 (𝜇, 𝜈, . . . = 0, 1, 2, 3) to denote local 4D spacetime coordinates and 𝜕𝜇 as

the corresponding coordinate basis vectors of the tangent spaces.
The metric tensor has components 𝑔 𝜇𝜈 and signature (− + ++), and the Christoffel
𝜌 𝜌
symbols, Γ 𝜇𝜈 = Γ𝜈𝜇 , always refer to the usual torsion-free Levi-Civita connection,

𝜌 𝜌 1 𝜌𝜎

Γ 𝜇𝜈 = Γ 𝜇𝜈 (𝑔) = 𝑔 𝜕𝜇 𝑔 𝜈 𝜎 + 𝜕𝜈 𝑔 𝜇 𝜎 − 𝜕𝜎 𝑔 𝜇𝜈 . (156)
2
The curvature tensor is defined as
𝜇 𝜇
𝑅𝜌𝜎 𝜇 𝜈 ≡ 2 𝜕[𝜌 Γ 𝜎 ]𝜈 + 2 Γ 𝜏 [𝜌 Γ 𝜏𝜎 ]𝜈 , (157)

where here and throughout the text we use symmetrization () and antisymmetrization
[] with weight one, i.e. (𝜇𝜈) = 1/2 (𝜇𝜈 + 𝜈𝜇), [𝜇𝜈] = 1/2 (𝜇𝜈 − 𝜈𝜇) etc.
The Ricci tensor, the Ricci scalar and the Einstein-Hilbert action are given by

𝑅 𝜇𝜈 ≡ 𝑅𝜌𝜇 𝜌 𝜈 , (158)
𝑅 ≡ 𝑅𝜌𝜇 𝜌 𝜈 𝑔 𝜇𝜈 = 𝑅 𝜇𝜈 𝑔 𝜇𝜈 , (159)
∫ ∫
𝑀2 √ 1 √
𝑆𝐸 𝐻 = 𝑃 𝑑 4 𝑥 −𝑔 𝑅 = 𝑑 4 𝑥 −𝑔 𝑅, (160)
2 16𝜋𝐺 𝑁
40 Gianguido Dall’Agata and Marco Zagermann

where 𝑔 ≡ det(𝑔 𝜇𝜈 ), 𝐺 𝑁 denotes Newton’s constant, and 𝑀 𝑃 is the reduced Planck

mass (with 𝑐 = 1, ℏ = 1),
1
𝑀𝑃 = √ = 2.44 · 1018 GeV. (161)
8𝜋𝐺 𝑁

5.2 Vierbein and Cartan’s formalism

In order to describe the coupling to fermions, an orthonormal basis of tangent vectors,

𝑒 𝑎 (𝑎 = 0, 1, 2, 3), is introduced at each point of the spacetime manifold,

𝑔(𝑒 𝑎 , 𝑒 𝑏 ) = 𝜂 𝑎𝑏 , (162)

where
𝜂 𝑎𝑏 = diag(−1, +1, +1, +1) (163)
is the flat Minkowski metric. Orthonormality fixes the 𝑒 𝑎 only up to local Lorentz
transformations, 𝑒 𝑎 → 𝑒 𝑏 Λ𝑏 𝑎 (𝑥), with Λ𝑏 𝑎 (𝑥) ∈ 𝑆𝑂 (1, 3).
The basis change from 𝑒 𝑎 to the coordinate basis vectors 𝜕𝜇 and vice versa is
written as
𝜇
𝑒 𝑎 = 𝑒 𝑎 (𝑥)𝜕𝜇 , 𝜕𝜇 = 𝑒 𝑎𝜇 (𝑥)𝑒 𝑎 , (164)
𝜇
where 𝑒 𝑎𝜇 𝑒 𝜈𝑎 = 𝛿 𝜈𝜇 , 𝑒 𝑎 𝑒 𝑏𝜇 = 𝛿 𝑏𝑎 . By a common abuse of terminology, we refer to both
the vectors 𝑒 𝑎 as well as the matrix of conversion coefficients, 𝑒 𝑎𝜇 (𝑥), as the vierbein.
The vierbein 𝑒 𝑎𝜇 and its inverse can be used to convert the “curved” (or “world”)
indices 𝜇, 𝜈, . . . of any tensor field to “flat” (or “local Lorentz”) indices 𝑎, 𝑏, . . ., in
particular,
𝑔 𝜇𝜈 (𝑥) = 𝑒 𝑎𝜇 (𝑥)𝑒 𝑏𝜈 (𝑥)𝜂 𝑎𝑏 , (165)
which also implies q
√
−𝑔 ≡ − det(𝑔 𝜇𝜈 ) = det 𝑒 𝑎𝜇 ≡ 𝑒 (166)
and shows that the vierbein 𝑒 𝑎𝜇 contains the same information as the metric.
When tensor fields are expressed in terms of the flat indices 𝑎, 𝑏, . . ., the con-
nection coefficients of the covariant derivative are denoted as 𝜔 𝜇 𝑎 𝑏 , so that, e.g.,
∇ 𝜇𝑉 𝑎 = 𝜕𝜇𝑉 𝑎 +𝜔 𝜇 𝑎 𝑏 𝑉 𝑏 . This covariant derivative is equivalent to the one for curved
indices as defined by the Christoffel symbols (in the sense that ∇ 𝜇 𝑉 𝑎 = 𝑒 𝑎𝜈 ∇ 𝜇 𝑉 𝜈
etc.) if
𝜌
∇ 𝜇 𝑒 𝑎𝜈 ≡ 𝜕𝜇 𝑒 𝑎𝜈 + 𝜔 𝜇 𝑎 𝑏 𝑒 𝑏𝜈 − Γ 𝜇𝜈 𝑒 𝜌𝑎 = 0, (167)
where, as suggested by the notation, the left hand side can be viewed as a total
covariant derivative that acts on curved and flat indices of the vierbein, implementing
covariance with respect to general coordinate and local Lorentz transformations.
Eq. (167) is sometimes referred to as the “vierbein postulate”.
In supergravity, it is useful to define also a derivative operator, 𝐷 𝜇, that is covariant
only with respect to local Lorentz transformations, but not necessarily with respect to
Simple Supergravity 41

general coordinate transformations, i.e., all local Lorentz indices will be contracted
with spin connections, but there are no Christoffel symbols that contract any world
index. An important example is the Lorentz covariant derivative of the vierbein 𝑒 𝑎𝜈 ,

𝐷 𝜇 𝑒 𝑎𝜈 ≡ 𝜕𝜇 𝑒 𝑎𝜈 + 𝜔 𝜇 𝑎 𝑏 𝑒 𝑏𝜈 . (168)
𝜌
In view of (167), this expression does not vanish, but is equal to Γ 𝜇𝜈 𝑒 𝜌𝑎 . Recalling that
the antisymmetrization Γ [𝜇𝜈] = 12 𝑇𝜇𝜈 𝜌 is just the torsion tensor of this connection,
𝜌

we can thus write

1 1
𝐷 [𝜇 𝑒 𝑎𝜈] = 𝑇𝜇𝜈 𝑎 = 𝑇𝜇𝜈 𝜌 𝑒 𝜌𝑎 , (169)
2 2
which is now a proper tensor field that vanishes for the Levi-Civita connection.
Just as for the Levi–Civita connection in curved indices, the vanishing of the
torsion tensor can be used to deduce an explicit expression for the spin connection
in terms of the vierbein, 𝜔 𝜇 𝑎𝑏 = 𝜔 𝜇 𝑎𝑏 (𝑒). To this end, one considers
 
𝑡 𝑑𝑐,𝑎 ≡ 𝑒 𝜈 [𝑑 𝑒 𝜇𝑐] 𝜕𝜇 𝑒 𝑎𝜈 + 𝜔 𝜇 𝑎 𝑏 𝑒 𝑏𝜈 ,

which is is zero because of the assumed vanishing torsion. This then also implies the
vanishing of the following combination,

𝑡 𝑑𝑐,𝑎 − 𝑡 𝑐𝑎,𝑑 − 𝑡 𝑎𝑑,𝑐 = 0, (170)

in which only one term in the spin connection, 𝑒 𝜌𝑎 𝜔 𝜌 𝑐𝑑 , survives so that, after
multiplying by 𝑒 𝜇𝑎 , one obtains

𝜔 𝜇 𝑐𝑑 (𝑒) = 2𝑒 𝜈 [𝑐 𝜕 [𝜇 𝑒 𝑑]
𝜈]
− 𝑒 𝑎 𝜇 𝑒 𝜈 [𝑐 𝑒 𝜎𝑑] 𝜕𝜈 𝑒 𝑎𝜎 . (171)

This connection is called the torsion-free spin connection, whose equivalence to the
Levi Civita connection (156) can also be verified directly.
As mentioned above, even though the Lorentz covariant derivative, 𝐷 𝜇 , is not
covariant with respect to general coordinate transformations, the antisymmetrized
derivative (169) still forms a proper tensor field. In supergravity, all equations can
similarly be expressed in terms of antisymmetrized Lorentz covariant derivatives
𝜌
only so that the Christoffel symbols Γ 𝜇𝜈 never occur. This allows one to use com-
pact differential form notation, which can substantially reduce the index clutter in
computations. To this end, we introduce the co-frame of one-forms, 𝑒 𝑎 , dual to the
vectors 𝑒 𝑏 ,
𝑒 𝑎 ≡ 𝑒 𝑎𝜇 𝑑𝑥 𝜇 , (172)
as well as the connection one-forms,

𝜔 𝑎 𝑏 ≡ 𝜔 𝜇 𝑎 𝑏 𝑑𝑥 𝜇 . (173)

In terms of 𝑒 𝑎 , the expression (169) for the torsion tensor reads

42 Gianguido Dall’Agata and Marco Zagermann

1 𝜇
𝐷𝑒 𝑎 ≡ 𝑑𝑒 𝑎 + 𝜔 𝑎 𝑏 ∧ 𝑒 𝑏 = 𝑇 𝑎 = 𝑑𝑥 ∧ 𝑑𝑥 𝜈 𝑇𝜇𝜈 𝑎 , (174)
2
whereas the curvature tensor of the connection 𝜔 𝜇 𝑏 𝑎 is given by the two-form

𝑅 𝑎 𝑏 = 𝑑𝜔 𝑎 𝑏 + 𝜔 𝑎 𝑐 ∧ 𝜔 𝑐 𝑏 . (175)

The conditions for the connection 𝜔 𝜇 𝑏 𝑎 to be equivalent to the Levi–Civita

connection can now be written as
• 𝐷𝜂 𝑎𝑏 ≡ 𝑑𝜂 𝑎𝑏 + 𝜔 𝑎 𝑐 𝜂 𝑐𝑏 + 𝜔 𝑏 𝑐 𝜂 𝑎𝑐 = 0 (metric compatibility);
• 𝑇 𝑎 = 0 (vanishing torsion).
The first condition implies that the spin connection is antisymmetric in its indices,
𝜔 (𝑎𝑏) = 0, when both are raised or lowered with 𝜂, and the second equation (in
combination with (174)) is equivalent to

𝐷𝑒 𝑎 = 0. (176)

In Cartan’s formalism the Einstein–Hilbert action reads

∫
𝑀 𝑃2
𝑆𝐸 𝐻 = 𝑅 𝑎𝑏 ∧ 𝑒 𝑐 ∧ 𝑒 𝑑 𝜖 𝑎𝑏𝑐𝑑 , (177)
4
where 𝜖0123 = 1, and the standard form (160) is recovered via the identification of
the four-dimensional measure,

𝑑𝑥 𝜇 ∧ 𝑑𝑥 𝜈 ∧ 𝑑𝑥 𝜌 ∧ 𝑑𝑥 𝜎 = −𝑑 4 𝑥 𝑒 𝜖 𝜇𝜈𝜌𝜎 , (178)

with 𝜖 𝜇𝜈𝜌𝜎 = 𝑒 𝑎𝜇 𝑒 𝑏𝜈 𝑒 𝑐𝜌 𝑒 𝑑𝜎 𝜖 𝑎𝑏𝑐𝑑 .

Considering the vierbein and spin connection as independent fields, yields the
field equations

𝛿𝑆
= 0 ⇔ 𝐺 𝜇𝜈 = 0, (179)
𝛿𝑒 𝑎𝜇
𝛿𝑆
= 0 ⇔ 𝑇 𝑎 = 0, (180)
𝛿𝜔 𝑎𝑏
𝜇

where
1
𝐺 𝜇𝜈 ≡ 𝑅 𝜇𝜈 − 𝑔 𝜇𝜈 𝑅 (181)
2
is the usual Einstein-tensor. Hence, in this formalism 𝜔 𝜇 𝑎𝑏 is dynamically fixed to be
the torsion-free spin connection, and one recovers the standard Einstein equations.
In supergravity, on the other hand, the spin connection also appears in the kinetic
term of the gravitino so that Eq. (180) receives a correction term bilinear in the
gravitino fields. Thus, in supergravity, the spin connection as determined from the
Simple Supergravity 43

field equation has torsion and is no longer equivalent to the torsion-free Levi-Civita
connection. This is not a problem, because the Christoffel symbols never appear in
the theory, and one is always free to redefine the spin connection as

𝜔 𝜇 𝑎𝑏 = 𝜔 𝜇 𝑎𝑏 (𝑒) + 𝜅 𝜇 𝑎𝑏 , (182)

where 𝜔 𝜇 𝑎𝑏 (𝑒) is the torsion-free spin connection (171), and 𝜅 𝜇 𝑎𝑏 is the contorsion
tensor,  
𝜅 𝜇 𝑎𝑏 = 𝑒 𝑐𝜇 𝑇 𝑎𝑏 𝑐 − 𝑇𝑐 𝑎 𝑏 − 𝑇 𝑏 𝑐 𝑎 , (183)

bilinear in the fermions. Espressing the theory in terms of 𝜔 𝜇 𝑎𝑏 (𝑒) then introduces
explicit four-Fermion interactions in the action.

5.3 Gamma matrices

We first introduce gamma matrices in flat Minkowski space with orthonormal basis
vectors labelled by the indices (𝑎, 𝑏, . . . = 0, 1, 2, 3) and the standard Minkowski
metric
𝜂 𝑎𝑏 = diag(−1, +1, +1, +1). (184)
The Lorentz algebra generators are denoted by 𝑀𝑎𝑏 = −𝑀𝑏𝑎 and satisfy

[𝑀𝑎𝑏 , 𝑀𝑐𝑑 ] = −2 𝜂 𝑐 [𝑎 𝑀𝑏 ]𝑑 + 2 𝜂 𝑑 [𝑎 𝑀𝑏 ]𝑐 . (185)

The gamma matrices, 𝛾 𝑎 , obey following basic relations

{𝛾 𝑎 , 𝛾𝑏 } = 2 𝜂 𝑎𝑏 14 , (186)
𝛾0† = −𝛾0 , 𝛾𝑖† = +𝛾𝑖 , 𝛾𝑇𝑎 = ±𝛾 𝑎 , (187)
1
Σ𝑎𝑏 := [𝛾 𝑎 , 𝛾𝑏 ], (188)
4
𝛾 𝑎1 ...𝑎 𝑝 ≡ 𝛾 [𝑎1 𝛾 𝑎2 . . . 𝛾 𝑎 𝑝 ] , (189)
5 0 1 2 3
𝛾5 ≡ 𝛾 ≡ −𝑖𝛾 𝛾 𝛾 𝛾 = +𝑖𝛾0 𝛾1 𝛾2 𝛾3 , (190)
(𝛾5 ) 2 = 14 , {𝛾5 , 𝛾 𝑎 } = 0, [𝛾5 , Σ𝑎𝑏 ] = 0. (191)

We also introduce the charge conjugation matrix 𝐶, satisfying

𝐶 𝑇 = −𝐶 = 𝐶 −1 = 𝐶 † , (192)
𝛾𝑇𝑎 = −𝐶𝛾 𝑎 𝐶 −1 , (193)

implying the following symmetry properties

𝐶 𝑇 = −𝐶, (𝐶𝛾 𝑎 )𝑇 = (𝐶𝛾 𝑎 ), (𝐶𝛾 𝑎𝑏 )𝑇 = (𝐶𝛾 𝑎𝑏 ),

(194)
(𝐶𝛾 𝑎𝑏𝑐 )𝑇 = −(𝐶𝛾 𝑎𝑏𝑐 ), (𝐶𝛾 𝑎𝑏𝑐𝑑 )𝑇 = −(𝐶𝛾 𝑎𝑏𝑐𝑑 ).
44 Gianguido Dall’Agata and Marco Zagermann

The matrix 𝛾5 also enters a number of useful duality relations between the antisym-
metrized products of gamma matrices,
1
𝛾 𝑎𝑏𝑐 = 𝑖 𝜖 𝑎𝑏𝑐𝑑 𝛾 𝑑 𝛾5 , 𝑖 𝛾 𝑎 𝛾5 = 𝜖 𝑎𝑏𝑐𝑑 𝛾 𝑏𝑐𝑑 ,
3!
1
𝛾 𝑎𝑏𝑐𝑑 = −𝑖 𝜖 𝑎𝑏𝑐𝑑 𝛾5 , 𝑖 𝛾5 = 𝜖 𝑎𝑏𝑐𝑑 𝛾 𝑎𝑏𝑐𝑑 , (195)
4!
𝑖 𝑎𝑏𝑐𝑑
𝛾 𝑎𝑏 = 𝜖 𝛾𝑐𝑑 𝛾5 ,
2
where 𝜖 𝑎𝑏𝑐𝑑 is the antisymmetric epsilon tensor with

𝜖0123 = 1. (196)

5.4 Spinors

Throughout this text, we use the four-component spinor notation. A translation to

the two-component spinor notation in our conventions can be found in [1].

5.4.1 Dirac spinors

Dirac spinors are elements of the representation space of the Clifford algebra (186).
As the matrices Σ𝑎𝑏 provide an explicit linear representation of the Lorentz algebra
generators 𝑀𝑎𝑏 , Dirac spinors inherit also a Lorentz algebra representation.
For a generic Dirac spinor, 𝜓, we define the Dirac conjugate as

𝜓 ≡ 𝑖𝜓 † 𝛾 0 = −𝑖𝜓 † 𝛾0 (197)

so that bilinears such as 𝜓 𝜒 are Lorentz invariant, because of Σ†𝑎𝑏 𝛾0 = −𝛾0 Σ𝑎𝑏 .
On 4D Dirac spinors, however, the Lorentz algebra representation provided by
Σ𝑎𝑏 is reducible. This reducibility can be resolved either by a chirality condition,
leading to Weyl spinors, or by a reality condition, leading to Majorana spinors.

5.4.2 The Weyl condition

The Weyl condition projects out the part of a spinor that has a particular handedness.
In order to impose it, one uses the chirality projectors
1  1 
𝑃𝐿 ≡ 1 + 𝛾5 , 𝑃𝑅 ≡ 1 − 𝛾5 , (198)
2 2
and defines left- and right-handed spinors,
Simple Supergravity 45

𝜓 𝐿 ≡ 𝑃 𝐿 𝜓, 𝜓 𝑅 ≡ 𝑃 𝑅 𝜓, (Weyl condition) (199)

satisfying 𝛾 5 𝜓 𝐿 = 𝜓 𝐿 and 𝛾 5 𝜓 𝑅 = −𝜓 𝑅 . This projection is consistent with Lorentz

covariance, and left- and right-handed spinors form separate representations of the
Lorentz group.
Using 𝛾5† = 𝛾5 as well as 𝑃 𝐿 𝛾0 = 𝛾0 𝑃 𝑅 and 𝑃 𝑅 𝛾0 = 𝛾0 𝑃 𝐿 , one finds

𝜓 𝑅 = 𝜓𝑃 𝐿 , 𝜓 𝐿 = 𝜓𝑃 𝑅 , (200)

where
𝜓 𝑅 ≡ (𝑃 𝑅 𝜓) = −𝑖(𝑃 𝑅 𝜓) † 𝛾0 . (201)
Because of this we will often write

𝜓 𝐿 ≡ 𝜓𝑃 𝐿 = 𝜓 𝑅 , 𝜓 𝑅 ≡ 𝜓𝑃 𝑅 = 𝜓 𝐿 . (202)

5.4.3 The Majorana condition

The Majorana condition is a reality condition that is usually expressed in terms of

the charge conjugation matrix, 𝐶. In terms of 𝐶, the charge conjugate spinor of a
four-component spinor, 𝜓, is defined as
𝑇
𝜓 𝑐 = 𝐶𝜓 = 𝑖𝐶𝛾 0𝑇 𝜓 ∗ . (203)

A Majorana spinor is then a spinor that equals its own charge conjugate,

𝜓 𝑐 = 𝜓. (Majorana condition) (204)

For a Majorana spinor, the Dirac conjugate (197) takes on the simple form

𝜓 = 𝜓𝑇 𝐶. (205)

For anti-commuting Majorana spinors, this then implies



 +𝜓 2 𝑀𝜓1
 for 𝑀 = 14 , 𝛾 𝑎𝑏𝑐 , 𝛾 𝑎𝑏𝑐𝑑
𝜓 1 𝑀𝜓2 = (206)

 −𝜓 𝑀𝜓1
 2 for 𝑀 = 𝛾 𝑎 , 𝛾 𝑎𝑏

Unless stated otherwise, we will always use anti-commuting Majorana spinors,

but often also take in addition the chiral projections 𝜓 𝐿 and 𝜓 𝑅 of these Majorana
spinors, which therefore are not independent.
More specifically, the chiral projections, 𝜓 𝐿 and 𝜓 𝑅 , of a Majorana spinor 𝜓
satisfy
(𝜓 𝐿 ) 𝑐 = 𝜓 𝑅 , (𝜓 𝑅 ) 𝑐 = 𝜓 𝐿 , (207)
showing that 𝜓 𝐿 and 𝜓 𝑅 are no longer Majorana spinors themselves. In other words,
a spinor in 4D cannot be simultaneously chiral and Majorana, but it nevertheless
46 Gianguido Dall’Agata and Marco Zagermann

does make sense to talk about the chiral projections 𝜓 𝐿 or 𝜓 𝑅 of a given Majorana
spinor 𝜓.
In fact, several identities that hold for Majorana spinors still have a close analogue
for their chiral projections. In particular, the chiral projection 𝜓 𝐿 of a Majorana spinor
𝜓 satisfies
𝜓 𝐿 ≡ 𝜓𝑃 𝐿 = (𝜓 𝐿 )𝑇 𝐶 (208)
and similarly for 𝜓 𝑅 , as follows from (205) and 𝐶𝛾5 = 𝛾5𝑇 𝐶. From this, one can
obtain more symmetry properties for the chiral projections that are very similar to
eqs. (206) for the Majorana spinors themselves,

𝜒 𝐿 𝜓 𝐿 = 𝜒𝑇𝐿 𝐶𝜓 𝐿 = −𝜓𝑇𝐿 𝐶 𝑇 𝜒𝐿 = 𝜓 𝐿 𝜒𝐿 ,
𝜒 𝐿 𝛾 𝑎 𝜓 𝑅 = −𝜓 𝑅 𝛾 𝑎 𝜒𝐿 , 𝜒 𝐿 𝛾 𝑎𝑏 𝜓 𝐿 = −𝜓 𝐿 𝛾 𝑎𝑏 𝜒𝐿 (209)
𝜒 𝐿 𝛾 𝑎𝑏𝑐 𝜓 𝑅 = 𝜓 𝑅 𝛾 𝑎𝑏𝑐 𝜒𝐿 .

5.5 Spinors in curved spacetime

Spinors are double-valued representations of the Lorentz group. On a curved man-

ifold, the relevant Lorentz transformations are the local Lorentz transformations of
the vierbein so that
1
𝐷 𝜇 𝜓 = 𝜕𝜇 𝜓 + 𝜔 𝜇 𝑎𝑏 𝛾 𝑎𝑏 𝜓, (210)
4
is the proper Lorentz covariant derivative of a fermion field 𝜓.
The gamma matrices with a curved index 𝜇 are obtained from the constant 𝛾 𝑎 via
contaction with a vierbein:
𝛾 𝜇 ≡ 𝑒 𝑎𝜇 𝛾 𝑎 . (211)
The 𝛾 𝜇 are then in general no longer constant, and they transform non-trivially under
a variation of the vierbein.

5.6 Fierz identities

We will often need to rewrite 3 or 4-Fermi terms to complete our analysis of the
supersymmetry properties of an action, and hence Fierz identities will be extremely
useful. We list here the main ones for two spinors:
1 1
𝜓 𝑅 𝜒 𝑅 = − 𝜒 𝑅 𝜓 𝑅 𝑃 𝑅 + 𝜒 𝑅 𝛾 𝑎𝑏 𝜓 𝑅 𝛾 𝑎𝑏 𝑃 𝑅 , (212)
2 8
1
𝜓𝑅 𝜒 𝐿 = − 𝜒 𝐿 𝛾 𝑎 𝜓𝑅 𝛾𝑎 𝑃 𝐿 , (213)
2
Simple Supergravity 47

where for the sake of clarity we explicitly left the projectors on the right hand side.
We often make use of spinor 1-forms 𝜓 = 𝑑𝑥 𝜇 𝜓 𝜇 . Exchanging such spinor one-
forms then leads to an additional minus sign from the anti-commutativity of the
wedge product, and hence one has
1
𝜓 𝑅 ∧ 𝜓 𝑅 = − 𝜓 𝑅 ∧ 𝛾 𝑎𝑏 𝜓 𝑅 𝛾 𝑎𝑏 𝑃 𝑅 , (214)
8
1
𝜓𝑅 ∧ 𝜓 𝐿 = 𝜓 ∧ 𝛾 𝑎 𝜓𝑅 𝛾𝑎 𝑃 𝐿 , (215)
2 𝐿

where now 𝜓 𝐿 ∧ 𝜓 𝐿 = 0 because of (209) and the wedge product. A crucial conse-
quence is the so-called cyclic identity:

𝛾 𝑎 𝜓 𝐿 ∧ 𝜓 𝐿 ∧ 𝛾 𝑎 𝜓 𝑅 = 0. (216)

5.7 Lie derivative on p-forms.

The Lie derivative of a p-form 𝐴 𝑝 along the flow of a vector field 𝑉 is defined as

1 ∗

𝐿 𝑉 𝐴 𝑝 = lim 𝜎𝑡 𝐴 𝑝 (𝜎𝑡 (𝑥)) − 𝐴 𝑝 (𝑥) ,
𝑡→0 𝑡

where 𝜎𝑡∗ is the pull back of the differential form along the flow generated by the
vector field 𝑉. When applied to a scalar valued 𝑝-form this reduces to

𝐿 𝑉 𝐴 𝑝 = (𝚤𝑉 𝑑 + 𝑑𝚤𝑉 ) 𝐴 𝑝 .

5.8 The supersymmetry algebra

The N = 1 supersymmetry algebra in 4 dimensions can be represented as follows

{𝑄, 𝑄} = −2𝑖𝛾 𝑎 P𝑎 ,
[P𝑎 , 𝑄] = 0,
𝑖
[M 𝑎𝑏 , 𝑄] = 2 𝛾 𝑎𝑏 𝑄,
[𝑅, 𝑄] = 𝑖𝛾5 𝑄, (217)
[P𝑎 , P𝑏 ] = 0,
[P𝑎 , M 𝑏𝑐 ] = −2𝑖𝜂 𝑎 [𝑏 P𝑐] ,
[M 𝑎𝑏 , M 𝑐𝑑 ] = 2𝑖𝜂 𝑐 [𝑎 M 𝑏 ]𝑑 − 2𝑖𝜂 𝑑 [𝑎 M 𝑏 ]𝑐 ,
48 Gianguido Dall’Agata and Marco Zagermann

where we used Hermitian generators P𝑎 and M 𝑎𝑏 for the Poincaré algebra. Just as
in (185), we sometimes also use their anti-Hermitian counterparts,

𝑃 𝑎 = 𝑖 P𝑎 , 𝑀𝑎𝑏 = 𝑖 M 𝑎𝑏 (218)

when this is more convenient.

References

1. G. Dall’Agata and M. Zagermann, “Supergravity: From First Principles to Modern Appli-

cations,” Lect. Notes Phys. 991 (2021), 1-263 2021, ISBN 978-3-662-63978-8, 978-3-662-
63980-1
2. D. Z. Freedman, P. van Nieuwenhuizen and S. Ferrara, Progress Toward a Theory of Super-
gravity, Phys. Rev. D13 (1976) 3214–3218.
3. S. Deser and B. Zumino, Consistent Supergravity, Phys. Lett. B62 (1976) 335.
4. P. Van Nieuwenhuizen, Supergravity, Phys. Rept. 68 (1981) 189–398.
5. H. P. Nilles, Supersymmetry, Supergravity and Particle Physics, Phys. Rept. 110 (1984) 1.
6. B. de Wit and D. Z. Freedman, Supergravity: The basics and beyond, Bonn [Link]
1984:0135. MIT-CTP-1238.
7. L. Castellani, R. D’Auria and P. Fre, Supergravity and superstrings: A Geometric perspective.
Vol. 1: Mathematical foundations, . Singapore, Singapore: World Scientific (1991) 1-603.
8. L. Castellani, R. D’Auria and P. Fre, Supergravity and superstrings: A Geometric perspective.
Vol. 2: Supergravity, . Singapore, Singapore: World Scientific (1991) 607-1371.
9. J. Wess and J. Bagger, “Supersymmetry and supergravity,” Princeton University Press (1992).
10. A. Van Proeyen, Tools for supersymmetry, hep-th/9910030.
11. J.-P. Derendinger, Introduction to supergravity, . Lectures at the Summer School “Gif 2000"
in Paris.
12. B. de Wit, Supergravity, hep-th/0212245.
13. A. Van Proeyen, Structure of supergravity theories, hep-th/0301005. To be published in the
series Publications of the Royal Spanish Mathematical Society.
14. W. Rarita and J. Schwinger, “On a theory of particles with half integral spin,” Phys. Rev. 60
(1941) 61.
15. M. Fierz, "On the relativistic theory of force-free particles with any spin", Helvetica Physica
Acta 12, I, 3-37, 1939.
M. Fierz and W. Pauli, “On relativistic wave equations for particles of arbitrary spin in an
electromagnetic field,” Proc. Roy. Soc. Lond. A 173 (1939) 211.
16. M. A. Vasiliev, “Higher spin gauge theories in four-dimensions, three-dimensions, and two-
dimensions,” Int. J. Mod. Phys. D 5 (1996), 763-797 [arXiv:hep-th/9611024 [hep-th]]; “Higher
spin gauge theories: Star product and AdS space,” [arXiv:hep-th/9910096 [hep-th]].
17. A.H. Chamseddine, P.C. West, Nucl. Phys. B 129, 39 (1977).
18. S.W. MacDowell, F. Mansouri, Phys. Rev. Lett. 38, 739 (1977). [Erratum: [Link]. 38,
1376 (1977)]
19. T. W. B. Kibble, J. Math. Phys. 2 (1961), 212-221 doi:10.1063/1.1703702
20. P. van Nieuwenhuizen, Supergravity as a Yang-Mills theory, hep-th/0408137.
21. S. Ferrara, Algebraic Properties of Extended Supergravity in de Sitter Space, Phys. Lett. B69
(1977) 481.
22. K. Pilch, P. van Nieuwenhuizen and M. F. Sohnius, De Sitter Superalgebras And Supergravity,
Commun. Math. Phys. 98 (1985) 105.
23. I. Antoniadis, E. Dudas, S. Ferrara and A. Sagnotti, “The Volkov–Akulov–Starobinsky super-
gravity,” Phys. Lett. B 733 (2014), 32-35 [arXiv:1403.3269 [hep-th]].
Simple Supergravity 49

24. E. A. Bergshoeff, D. Z. Freedman, R. Kallosh and A. Van Proeyen, “Pure de Sitter Supergrav-
ity,” Phys. Rev. D 92 (2015) no.8, 085040 [erratum: Phys. Rev. D 93 (2016) no.6, 069901]
[arXiv:1507.08264 [hep-th]].
25. G. Dall’Agata, E. Dudas and F. Farakos, “On the origin of constrained superfields,” JHEP 05
(2016), 041 [arXiv:1603.03416 [hep-th]].
26. N. Cribiori, G. Dall’Agata and F. Farakos, “From Linear to Non-linear SUSY and Back Again,”
JHEP 08 (2017), 117 [arXiv:1704.07387 [hep-th]].
27. P. K. Townsend, Cosmological Constant in Supergravity, Phys. Rev. D15 (1977) 2802–2804.
28. S. Cecotti, L. Girardello and M. Porrati, Constraints on partial superhiggs, Nucl. Phys. B268
(1986) 295–316.
29. E. Cremmer, S. Ferrara, L. Girardello and A. Van Proeyen, “Coupling Supersymmetric Yang-
Mills Theories to Supergravity,” Phys. Lett. B 116 (1982), 231-237
30. E. Cremmer, S. Ferrara, L. Girardello and A. Van Proeyen, “Yang-Mills Theories with Local
Supersymmetry: Lagrangian, Transformation Laws and SuperHiggs Effect,” Nucl. Phys. B 212
(1983), 413
31. J. A. Bagger, “Coupling the Gauge Invariant Supersymmetric Nonlinear Sigma Model to
Supergravity,” Nucl. Phys. B 211 (1983), 302
32. H. Elvang, D. Z. Freedman and B. Kors, “Anomaly cancellation in supergravity with Fayet–
Iliopoulos couplings,” JHEP 11 (2006), 068 [arXiv:hep-th/0606012 [hep-th]].
33. J. De Rydt, J. Rosseel, T. T. Schmidt, A. Van Proeyen and M. Zagermann, Symplectic structure
of N=1 supergravity with anomalies and Chern-Simons terms, Class. Quant. Grav. 24 (2007)
5201–5220 [0705.4216].
34. P. Binetruy, G. Dvali, R. Kallosh and A. Van Proeyen, Fayet–Iliopoulos terms in supergravity
and cosmology, Class. Quant. Grav. 21 (2004) 3137–3170 [hep-th/0402046].