Part 3 SuperSymmetry

Prepared for submission to JHEP
Supersymmetry
B. C. Allanacha
a
Department of Applied Mathematics and Theoretical Physics, Centre for Mathematical Sciences,
University of Cambridge, Wilberforce Road, Cambridge CB3 0WA, United Kingdom
Copyright 2013 University of Cambridge. Not to be quoted or reproduced without permission.
E-mail: B.C.Allanach@damtp.cam.ac.uk
Abstract: These are lecture notes for the Cambridge mathematics tripos Part III Supersymmetry course, based on Ref. [1]. You should have attended the required courses: Quantum Field Theory, and Symmetries and Particle Physics. You will nd the latter parts of
Advanced Quantum Field theory (on renormalisation) useful. The Standard Model course
will aid you with the last topic (the minimal supersymmetric standard model), and help
with understanding spontaneous symmetry breaking. The three accompanying examples
sheets may be found on the DAMTP pages, and there will be classes organised for each
sheet. You can watch videos of my lectures on the web by following the link from
http://users.hepforge.org/~allanach/teaching.html
where these notes may also be found. I have a tendency to make trivial transcription errors
on the board - please stop me if I make one.
In general, the books contain several typographical errors. The last two books on the
list have a dierent metric convention to the one used herein (switching metric conventions
is surprisingly irksome!)
Books
Bailin and Love, Supersymmetric gauge eld theory and string theory, Institute of
Physics publishing has nice explanations.
Lykken Introduction to supersymmetry, arXiv:hep-th/9612114 - particularly good
on extended supersymmetry.
Aithchison, Supersymmetry in particle physics, Cambridge University Press is super clear and basic.
Martin A supersymmetry primer, arXiv:hep-ph/9709356 a detailed and phenomenological reference.
Wess and Bagger, Supersymmetry and Supergravity, Princeton University Publishing is terse but has no errors that I know of.
I welcome questions during lectures.
Contents
1 Physical Motivation
1.1 Basic theory: QFT
1.2 Basic principle: symmetry
1.3 Classes of symmetries
1.4 Importance of symmetries
1.4.1 The Standard Model
1.5 Problems of the Standard Model
1.5.1 Modications of the Standard Model
1
1
2
2
2
4
5
6
2 Supersymmetry algebra and representations

2.1 Poincar symmetry and spinors
e
2.1.1 Properties of the Lorentz group
2.1.2 Representations and invariant tensors of SL(2, C)
2.1.3 Generators of SL(2, C)
2.1.4 Products of Weyl spinors
2.1.5 Dirac spinors
2.2 SUSY algebra
2.2.1 History of supersymmetry
2.2.2 Graded algebra
2.3 Representations of the Poincar group
e
2.4 N = 1 supersymmetry representations
2.4.1 Bosons and fermions in a supermultiplet
2.4.2 Massless supermultiplet
2.4.3 Massive supermultiplet
2.4.4 Parity
2.5 Extended supersymmetry
2.5.1 Algebra of extended supersymmetry
2.5.2 Massless representations of N > 1 supersymmetry
2.5.3 Massive representations of N > 1 supersymmetry and BPS states
7
7
7
8
10
10
12
13
13
13
16
17
17
18
19
21
21
22
22
25
3 Superspace and Superelds

3.1 Basics about superspace
3.1.1 Groups and cosets
3.1.2 Properties of Grassmann variables
3.1.3 Denition and transformation of the general scalar supereld
3.1.4 Remarks on superelds
3.2 Chiral superelds
27
27
27
29
30
32
33
4 Four dimensional supersymmetric Lagrangians

4.1 N = 1 global supersymmetry
4.1.1 Chiral supereld Lagrangian
4.1.2 Vector superelds
4.1.3 Denition and transformation of the vector supereld
4.1.4 Wess Zumino gauge
4.1.5 Abelian eld strength supereld
4.1.6 Non - abelian eld strength
4.1.7 Abelian vector supereld Lagrangian
4.1.8 Action as a superspace integral
4.2 N = 2, 4 global supersymmetry
4.2.1 N = 2
4.2.2 N = 4
4.3 Non-renormalisation theorems
4.3.1 History
4.4 A few facts about local supersymmetry
34
34
34
37
37
38
38
39
40
42
43
43
44
44
45
45
5 Supersymmetry breaking
5.1 Preliminaries
5.1.1 F term breaking
5.1.2 ORaifertaigh model
5.1.3 D term breaking
5.1.4 Breaking local supersymmetry
46
46
47
48
49
50
6 Introducing the minimal supersymmetric standard model (MSSM)

6.1 Particles
6.2 Interactions
6.3 Supersymmetry breaking in the MSSM
6.4 The hierarchy problem
6.5 Pros and Cons of the MSSM
51
51
52
55
58
60
Physical Motivation
Let us review some relevant facts about the universe we live in.
1.1
Basic theory: QFT
Microscopically we have quantum mechanics and special relativity as two fundamental theories.
A consistent framework incorporating these two theories is quantum eld theory (QFT). In
this theory the fundamental entities are quantum elds. Their excitations correspond to
the physically observable elementary particles which are the basic constituents of matter
as well as the mediators of all the known interactions. Therefore, elds have a particle-like
character. Particles can be classied in two general classes: bosons (spin s = n Z) and
1
fermions (s = n + 2 Z + 1 ). Bosons and fermions have very dierent physical behaviour.
2
The main dierence is that fermions can be shown to satisfy the Pauli exclusion principle
, which states that two identical fermions cannot occupy the same quantum state, and
therefore explaining the vast diversity of atoms.
All elementary matter particles: the leptons (including electrons and neutrinos) and quarks
(that make protons, neutrons and all other hadrons) are fermions. Bosons on the other
hand include the photon (particle of light and mediator of electromagnetic interaction),
and the mediators of all the other interactions. They are not constrained by the Pauli
principle. As we will see, supersymmetry is a symmetry that unies bosons and fermions
despite all their dierences.
1.2
Basic principle: symmetry
If QFT is the basic framework to study elementary processes, one tool to learn about these
processes is the concept of symmetry.
A symmetry is a transformation that can be made to a physical system leaving the physical
observables unchanged. Throughout the history of science symmetry has played a very
important role to better understand nature.
1.3
Classes of symmetries
For elementary particles, we can dene two general classes of symmetries:

Space-time symmetries: These symmetries correspond to transformations on a eld
theory acting explicitly on the space-time coordinates,
x x (x ) ,
, = 0, 1, 2, 3 .
Examples are rotations, translations and, more generally, Lorentz- and Poincar
e
transformations dening special relativity as well as general coordinate transformations that dene general relativity.
Internal symmetries: These are symmetries that correspond to transformations of
the dierent elds in a eld theory,
a (x) M a b b (x) .
Roman indices a, b label the corresponding elds. If M a b is constant then the symmetry is a global symmetry; in case of space-time dependent M a b (x) the symmetry
is called a local symmetry.
1.4
Importance of symmetries
Symmetry is important for various reasons:
Labelling and classifying particles: Symmetries label and classify particles according
to the dierent conserved quantum numbers identied by the space-time and internal
symmetries (mass, spin, charge, colour, etc.). In this regard symmetries actually
dene an elementary particle according to the behaviour of the corresponding eld
with respect to the dierent symmetries.
Symmetries determine the interactions among particles, by means of the gauge principle, for instance. It is important that most QFTs of vector bosons are sick: they
are non-renormalisable. The counter example to this is gauge theory, where vector
bosons are necessarily in the adjoint representation of the gauge group.
As an
illustration, consider the Lagrangian
L = V (, )
which is invariant under rotation in the complex plane
exp(i) ,
as long as is a constant (global symmetry). If = (x), the kinetic term is no
longer invariant:
exp(i) + i( ) .
However, the covariant derivative D , dened as
D := + iA ,
transforms like itself, if the gauge - potential A transforms to A :
D exp(i) + i( ) + i(A )
= exp(i) D ,
so rewrite the Lagrangian to ensure gauge - invariance:

L = D D V (, ) .
The scalar eld couples to the gauge - eld A via A A , similarly, the Dirac
Lagrangian
L = D
has an interaction term A . This interaction provides the three point vertex that
describes interactions of electrons and photons and illustrate how photons mediate
the electromagnetic interactions.
Symmetries can hide or be spontaneously broken: Consider the potential V (, ) in
the scalar eld Lagrangian above.
If V (, ) = V (||2 ), then it is symmetric for exp(i). If the potential is of
the type
V = a ||2 + b ||4 ,
a, b 0 ,
Figure 1. The Mexican hat potential for V = a b ||2
with a, b 0.
then the minimum is at = 0 (here 0||0 denotes the vacuum expectation

value (VEV) of the eld ). The vacuum state is then also symmetric under the
symmetry since the origin is invariant. However if the potential is of the form
V
a b ||2
a, b 0 ,
the symmetry of V is lost in the ground state = 0. The existence of hidden

symmetries is important for at least two reasons:
(i) This is a natural way to introduce an energy scale in the system, determined
by the non vanishing VEV. In particular, we will see that for the standard
model Mew 103 GeV, denes the basic scale of mass for the particles of the
standard model, the electroweak gauge bosons and the matter elds, through
their Yukawa couplings, obtain their mass from this eect.
(ii) The existence of hidden symmetries implies that the fundamental symmetries
of nature may be huge despite the fact that we observe a limited amount of
symmetry. This is because the only manifest symmetries we can observe are
the symmetries of the vacuum we live in and not those of the full underlying
theory. This opens-up an essentially unlimited resource to consider physical
theories with an indenite number of symmetries even though they are not
explicitly realised in nature. The standard model is the typical example and
supersymmetry and theories of extra dimensions are further examples.
1.4.1
The Standard Model
The Standard Model is well-dened and currently well conrmed by experiments.

space-time symmetries: Poincar in 4 dimensions
e
gauged GSM =SU(3)c SU(2)U(1)Y symmetry, where SU(3)c denes the strong interactions. SU(2)L U(1)Y is spontaneously broken by the Higgs mechanism to
U(1)em . The gauge elds are spin-1 bosons, for example the photon A , or gluons Ga=1,...,8 . Matter elds (quarks and leptons) have spin 1/2 and come in three
families (successively heavier copies). The Higgs boson (a particle has just been discovered at the LHC whose properties are consistent with the Higgs boson) is the spin
zero particle that spontaneously breaks the SU(2)L U(1)Y . The W and Z particles
get a mass via the Higgs mechanism and therefore the weak interactions are short
range. This is also the source of masses for all quarks and leptons. The sub-index
L in SU (2)L refers to the fact that the Standard Model does not preserve parity
and dierentiates between left-handed and right-handed particles. In the Standard
Model only left-handed particles transform non-trivially under SU (2)L . The gauge
particles have all spin s = 1 and mediate each of the three forces: photons () for
U (1) electromagnetism, gluons for SU (3)C of strong interactions, and the massive
W and Z for the weak interactions.
1.5
Problems of the Standard Model
The Standard Model is one of the cornerstones of all science and one of the great triumphs
of the past century. It has been carefully experimentally veried in many ways, especially
during the past 20 years. However, there are still some unresolved issues or mysteries:
The hierarchy problem. The Higgs vacuum expectation value is v 246 GeV, whereas
the gravitational scale is MP lanck G 1019 GeV. The hierarchy problem is: why
is v/MP lanck 1017 so much smaller than 1? In a fundamental theory, one might
expect them to be the same order. In QFT, one sees that quantum corrections (loops)
to v are expected to be of order of the heaviest scale in the theory divided by 16 2 .
The question of why the hierarchy is stable with respect to the quantum corrections
is called the technical hierarchy problem, and is arguably the main motivation for
weak-scale supersymmetry.
The cosmological constant () problem: probably the biggest problem in fundamental
physics. is the energy density of free space time. Why is (/MP lanck )4 10120
1?
The Standard Model has around 20 parameters, which must be measured then set
by hand.
What particle constitutes the dark matter observed in the universe? It is not contained in the Standard Model.
We wish to nd extensions that could solve some or all of the problems mentioned above
in order to generalise the Standard Model. The the Part III Standard Model course for
more details. Experiments are a traditional way of making progress in science. We need
experiments to explore energies above the currently attainable scales and discover new
particles and underlying principles that generalise the Standard Model. This approach is
of course being followed at the LHC. The experiment will explore physics at the 103 GeV
scale and new physics beyond the Standard Model. Notice that exploring energies closer
to the Planck scale MP lanck 1019 GeV is out of the reach for many years to come.
1.5.1
Modications of the Standard Model
In order to go beyond the Standard Model we can follow several avenues, for example:
Add new particles and/or interactions (e.g. a dark matter particle).
More symmetries. For example,
(i) internal symmetries, for example grand unied theories (GUTs) in which the symmetries of the Standard Model are themselves the result of the breaking of a yet larger
symmetry group.
GGUT
M 1016 GeV
GSM
M 102 GeV
SU (3)c U (1)Y ,
This proposal is very elegant because it unies, in one single symmetry, the three
gauge interactions of the Standard Model. It leaves unanswered most of the open
questions above, except for the fact that it reduces the number of independent parameters due to the fact that there is only one gauge coupling at large energies. This is
expected to run at low energies and give rise to the three dierent couplings of the
Standard Model (one corresponding to each group factor). Unfortunately, with our
present precision understanding of the gauge couplings and spectrum of the Standard
Model, the running of the three gauge couplings does not unify at a single coupling
at higher energies but they cross each other at dierent energies.
(ii) Supersymmetry. Supersymmetry is an external, or space-time symmetry. Supersymmetry solves the technical hierarchy problem due to cancellations between the
contributions of bosons and fermions to the electroweak scale, dened by the Higgs
mass. Combined with the GUT idea, it also solves the unication of the three gauge
couplings at one single point at larger energies. Supersymmetry also provides the
most studied example for dark matter candidates. Moreover, it provides well dened QFTs in which issues of strong coupling can be better studied than in the
non-supersymmetric models.
(iii) Extra spatial dimensions. More general space-time symmetries open up many more
interesting avenues. These can be of two types. First we can add more dimensions to
space-time, therefore the Poincar symmetries of the Standard Model and more genere
ally the general coordinate transformations of general relativity, become substantially
enhanced. This is the well known Kaluza Klein theory in which our observation of a 4
dimensional universe is only due to the fact that we have limitations about seeing
other dimensions of space-time that may be hidden to our experiments. In recent
years this has been extended to the brane world scenario in which our 4 dimensional
universe is only a brane or surface inside a larger dimensional universe. These ideas
may lead to a dierent perspective of the hierarchy problem and also may help unify
internal and space-time symmetries.
Beyond QFT: A QFT with Supersymmetry and extra dimensions does not address
the problem of quantising gravity. For this purpose, the current best hope is string
theory which goes beyond our basic framework of QFT. It so happens that for its
consistency, string theory requires supersymmetry and extra dimensions also. This
gives a further motivation to study supersymmetry.
Supersymmetry algebra and representations
2.1
Poincar symmetry and spinors

e
The Poincar group corresponds to the basic symmetries of special relativity, it acts on
e
space-time coordinates x as follows:
x x =
x +
a
translation
Lorentz
Lorentz transformations leave the metric tensor = diag(1, 1, 1, 1) invariant:

T =
They can be separated between those that are connected to the identity and this that are
not (i.e. parity reversal P = diag(1, 1, 1, 1) and time reversal T = diag(1, 1, 1, 1)).
We will mostly discuss those continuously connected to identity, i.e. the proper ore
thochronous group1 SO(1, 3) . Generators for the Poincar group are the hermitian M
(translations) with algebra
(rotations and Lorentz boosts) and P
P , P
M , P
M , M
= 0
= i P P
= i M + M M M
A 4 dimensional matrix representation for the M is

(M )
2.1.1
= i .
Properties of the Lorentz group
We now show that locally (i.e. in terms of the algebra), we have a correspondence
SO(1, 3) SU (2) SU (2).
=
The generators of SO(1, 3) (Ji of rotations and Ki of Lorentz boosts) can be expressed as
Ji =
1
ijk Mjk ,
2
Ki = M0i ,
and the Lorentz algebra written in terms of Js and Ks is

[Ki , Kj ] = iijk Jk ,
1
[Ji , Kj ] = iijk Kk ,
[Ji , Jj ] = iijk Jk .
These consist of the subgroup of transformations which have det = +1 and 0 1.

0
We now construct the linear2 combinations (which are neither hermitian nor anti hermitian)
Ai =
1
Ji + iKi ,
2
Bi =
1
Ji iKi
2
(2.1)
which satisfy SU (2) SU (2) commutation relations

A i , Aj
= iijk Ak ,
Bi , Bj
= iijk Bk ,
Ai , Bj
= 0
Under parity P , (x0 x0 and x x) we have

Ji Ji ,
Ki Ki
Ai
Bi .
We can interpret J = A + B as the physical spin.

On the other hand, there is a homeomorphism (not an isomorphism)
SO(1, 3) SL(2, C) .
=
To see this, take a 4 vector X and a corresponding 2 2 - matrix x,
= x e = (x0 , x1 , x2 , x3 ) ,
x = x =
x0 + x3 x1 ix2
x1 + ix2 x0 x3
where is the 4 vector of Pauli matrices

=
10
01
01
10
0 i
i 0
1 0
0 1
Transformations X X under SO(1, 3) leaves the square

|X|2 = x2 x2 x2 x2
0
1
2
3
invariant, whereas the action of SL(2, C) mapping x N xN with N SL(2, C) pre
serves the determinant

det x = x2 x2 x2 x2 .
0
1
2
3
The map between SL(2, C) and SO(1, 3) is 2-1, since N = both correspond to = ,
but SL(2, C) has the advantage of being simply connected, so SL(2, C) is the universal
covering group.
2.1.2
Representations and invariant tensors of SL(2, C)
The basic representations of SL(2, C) are:

The fundamental representation
= N ,
, = 1, 2
The elements of this representation are called left-handed Weyl spinors.

2
NB 123 = +1 = 123 .
(2.2)
The conjugate representation
, = 1, 2

= N ,
Here are called right-handed Weyl spinors.
The contravariant representations are
= (N 1 ) ,
= (N 1 ) .
The fundamental and conjugate representations are the basic representations of SL(2, C)
and the Lorentz group, giving then the importance to spinors as the basic objects of special
relativity, a fact that could be missed by not realising the connection of the Lorentz group
and SL(2, C). We will see next that the contravariant representations are however not
independent.
To see this we will consider now the dierent ways to raise and lower indices.
The metric tensor = ( )1 is invariant under SO(1, 3) and is used to raise/lower
indices.
The analogy within SL(2, C) is
= , 12 = +1, 21 = 1.

since
= N N
= det N
= .
That is why is used to raise and lower indices

= ,
= ,
so contravariant representations are not independent from covariant ones.

To handle mixed SO(1, 3)- and SL(2, C) indices, recall that the transformed components x should look the same, whether we transform the vector X via SO(1, 3) or
the matrix x = x via SL(2, C)

(x ) N (x ) N
= x ( ) ,
so the correct transformation rule is

( ) = N ( ) () N .
Similar relations hold for
( ) := ( )

= (, ) .
2.1.3
Generators of SL(2, C)
Let us dene tensors , as antisymmetrised products of matrices:
( )
( )
:=
:=
4
i

which satisfy the Lorentz algebra
= i + ,
and analagously for . They thus form representations of the Lorentz algebra (the spinor
representation).
Under a nite Lorentz transformation with parameters , spinors transform as follows:
i
exp
exp
(left-handed)
(right-handed)
Now consider the spins with respect to the SU (2)s spanned by the Ai and Bi :
:
(A, B) =
(A, B) =
1
, 0
2
1
0,
2
Ji =
Ji =
1
i ,
2
1
i ,
2
i
Ki = i
2
i
Ki = + i
2
Some useful identities concerning the and can be found on the examples sheets.
For now, let us just mention the identities3

2i
1
= ,
2i
known as self duality and anti self duality. They are important because naively being
antisymmetric seems to have 43 components, but the self duality conditions reduces this
2
by half. A reference book illustrating many of the calculations for two - component spinors
is [2].
2.1.4
Products of Weyl spinors
Dene the product of two Weyl spinors as

:= =
:= = ,
0123 = 1 = 0123
10
where in particular
= = = 2 1 1 2 .
Choosing the to be anticommuting Grassmann numbers, 1 2 = 2 1 , so =

1
22 1 . Thus = 2 ().
We note that eq. 2.1 implies that A B under Hermitian conjugation. Therefore, the
Hermitian conjugate of a left (right)-handed spinor is a right (left)-handed spinor. Thus
we dene
( ) := ,
:= ( 0 )
it follows that
() = ,
( ) =
which justies the contraction of implicit dotted indices in contrast to the implicit
contraction of undotted ones.
In general we can generate all higher dimensional representations of the Lorentz group by
1
1
products of the fundamental representation ( 2 , 0) and its conjugate (0, 2 ). The computa1
r s
tion of tensor products ( 2 , 2 ) = ( 2 , 0)r (0, 1 )s can be reduced to successive application
2
j
of the elementary SU (2) rule ( 2 ) ( 1 ) = ( j1 ) ( j+1 ) (for j = 0).
2
2
2
Let us give two examples for tensoring Lorentz representations:
1
1
( 1 , 0) (0, 2 ) = ( 2 , 1 )
2
2
Bi-spinors with dierent chiralities can be expanded in terms of the . Actually,
the matrices form a complete orthonormal set of 2 2 matrices with respect to

the trace Tr{ } = 2 :
1
( )

2
Hence, two spinor degrees of freedom with opposite chirality give rise to a Lorentz
vector .
1
( 1 , 0) ( 2 , 0) = (0, 0) (1, 0)
2
Alike bi-spinors require a dierent set of matrices to expand, and ( T ) :=

( ) . The former represents the unique antisymmetric 2 2 matrix, the latter
provides the symmetric ones.

1
1
() +
2
2
( )
The product of spinors with alike chiralities decomposes into two Lorentz irreducible
representations, a scalar and a self-dual antisymmetric rank two tensor .
The counting of independent components of from its self-duality property precisely provides the right number of three components for the (1, 0) representation.
Similarly, there is an anti-self dual tensor in (0, 1).

These expansions are also referred to as Fierz identities.
11
2.1.5
Dirac spinors
To connect the ideas of Weyl spinors with the more standard Dirac theory, dene
0
0
:=
then these satisfy the Cliord algebra

= 2 .
The matrix 5 , dened as

0
0
5 := i 0 1 2 3 =
can have eigenvalues 1 (chirality). The generators of the Lorentz group are
0
0
We dene Dirac spinors to be the direct sum of two Weyl spinors of opposite chirality,
D
:=
such that the action of 5 is given as

5 D
0
0
We can dene the following projection operators PL , PR ,

PL :=
1
2
5 ,
1
2
PR :=
+ 5 ,
eliminating one part of denite chirality, i.e.

PL D
PR D
Finally, dene the Dirac conjugate D and charge conjugate spinor D C by
:= ( , ) = 0
D
D C
:= C D
where C denotes the charge conjugation matrix

C
:=
0

0
12
Majorana spinors M have property = ,

M
= M C ,
so a general Dirac spinor (and its charge conjugate) can be decomposed as

D
2.2
D C
= M 1 iM 2 .
SUSY algebra
2.2.1
= M 1 + iM 2 ,
History of supersymmetry
In the 1960s, from the study of strong interactions, many hadrons have been discovered and were successfully organised in multiplets of SU (3)f , the f referring to
avour. This procedure was known as the eightfold way of Gell-Mann and Neeman. Questions arouse about bigger multiplets including particles of dierent spins.
In a famous No-go theorem (Coleman, Mandula 1967) said that the most general
symmetry of the S - matrix is Poincar internal, that cannot mix dierent spins
e
(for example), if you still require there to be interactions
Golfand and Licktman (1971) extended the Poincar algebra to include spinor
e
generators Q , where = 1, 2.
Ramond,Neveu-Schwarz, Gervais, Sakita (1971): devised supersymmetry in 2
dimensions (from string theory).
Wess and Zumino (1974) wrote down supersymmetric eld theories in 4 dimensions.
They opened the way for many other contributions to the eld. This is often seen as
the actual starting point on systematic study of supersymmetry.
Haag, Lopuszanski, Sohnius (1975): generalised the Coleman Mandula theorem
to show that the only non-trivial quantum eld theories have a symmetry group of
super Poincare group in a direct product with internal symmetries.
e
2.2.2
Graded algebra
We wish to extend the Poincar algebra non-trivially. The Coleman Mandula theorem
e
stated that in 3+1 dimensions, one cannot do this in a non-trivial way and still have nonzero scattering amplitudes. In other words, there is no non-trivial mix of Poincar and
e
internal symmetries with non-zero scattering except for the direct product
Poincar internal.
e
However (as usual with no-go theorems) there was a loop-hole because of an implicit axiom:
the proof only considered bosonic generators.
We wish to turn bosons into fermions, thus we need to introduce a fermionic generator Q.
Heuristically:
Q|boson |fermion ,
Q|fermion |boson .
13
For this, we require a graded algebra - a generalisation of Lie algebra. If Oa is an operator

of an algebra (such as a group generator), a graded algebra is
e
Oa Ob (1)a b Ob Oa = iCab Oe ,
(2.3)
where a = 0 if Oa is a bosonic generator, and a = 1 if Oa is a fermionic generator.

For supersymmetry, the bosonic generators are the Poincar generators P , M and the
e
fermionic generators QA , QA , where A = 1, ..., N . In case N = 1 we speak of a simple
SUSY, in case N > 1 of an extended SUSY. In this section, we will only discuss N = 1.
We know the commutation relations [P , P ], [P , M ] and [M , M ] already from
the Poincar algebra, so we need to nd
e
(a) Q , M , (b) Q , P ,
(c)
, (d) Q , Q
Q , Q
also (for internal symmetry generators Ti )

(e) Q , Ti .
We shall be using the fact that the right hand sides must be linear and that they must
transform in the same way as the commutators under a Lorentz transformation, for in
stance. The relations for Q Q may then be obtained from these by taking hermitian
conjugates.
(a) Q , M : we can work this one out by knowing how Q transforms as a
spinor and as an operator.
Since Q is a spinor, it transforms under the exponential of the SL(2, C) generators
:
i
i
Q = exp
Q
Q .
2
2
Under an active transformation, as an operator. |

i
|U Q U | , where U = exp 2 M . Hence
Q = U Q U
i
M
2
U |
|Q |
i
M
2
Compare these two expressions for Q up to rst order in ,
i
( )
2
i
Q M M Q
2
= Q
Q , M
= ( ) Q
Similarly,
Q , M
= ( ) Q
14
+ O( 2 )

(b) Q , P : c( ) Q is the only way of writing a sensible term with free indices
, which is linear in Q. To x the constant c, consider [Q , P ] = c ( ) Q

(take adjoints using (Q ) = Q and ( Q) = (Q ) ). The Jacobi identity for
P , P and Q
0 =
P , P , Q
P , Q , P
Q , P , P
= c ( ) P , Q
+ c ( ) P , Q
= |c|2 ( ) ( ) Q |c|2 ( ) ( ) Q

= |c|2 ( ) Q
=0
can only hold for general Q , if c = 0, so

Q , P
(c)
Q , P
= 0
Q , Q
Due to index structure, that commutator should look like

= k ( ) M .
Q , Q
Since the left hand side commutes with P and the right hand side doesnt, the only
consistent choice is k = 0, i.e.
Q , Q
(d)
Q , Q
= 0,
= 0
Q , Q
This time, index structure implies an ansatz
Q , Q
= t ( ) P .
There is no way of xing t, so, by convention, set t = 2, dening the normalisation

of the operators:
= 2 ( ) P
Q , Q
Notice that two symmetry transformations Q Q have the eect of a translation. Let |B
be a bosonic state and |F a fermionic one, then
Q |F
= |B ,
Q |B
= |F
15
QQ : |B
|B (translated) .
(e)
Q , Ti
Usually, this commutator vanishes due to the Coleman-Mandula theorem. Exceptions

are U (1) automorphisms of the supersymmetry algebra known as R symmetry. The
algebra is invariant under the simultaneous change
Q exp(i) Q .
Q exp(i) Q ,
Let R be a global U (1) generator, then, since Q eiR Q eiR ,

Q , R
2.3
Q , R
= Q ,
= Q .
Representations of the Poincar group

e
Since we are changing the Poincar group, we must check to see if anything happens to
e
the Casimirs of the changed group, since these are used to label irreducible representations
(remember that one needs a complete commuting set of observables to label them). Recall
the rotation group {Ji : i = 1, 2, 3} satisfying
J i , Jj
= iijk Jk .
The Casimir operator

3
J2 =
Ji2
i=1
commutes with all the Ji and labels irreducible representations by eigenvalues j(j + 1) of
J 2 . Within these irreducible representations, the J3 eigenvalues j3 = j, j + 1, ..., j 1, j
label each element. States are labelled like |j, j3 .
Also recall the two Casimirs in the Poincar group, one of which involves the Pauli Ljubanski
e
vector W describing generalised spin
W =
1
P M
2
(where 0123 = 0123 = +1).

The Poincar Casimirs are then given by
e
C1 = P P ,
C2 = W W ,
since the Ci commute with all generators.

Poincar multiplets are labelled |m, , where m2 is the eigenvalue of C1 and is the eigene
value of C2 . States within those irreducible representations carry the eigenvalue p of the
generator P as a label. Notice that at this level the Pauli Ljubanski vector only provides
a short way to express the second Casimir. Even though W has standard commutation
relations with the generators of the Poincar group M (since it transforms as a vector
e
under Lorentz transformations) and commutes with P (it is invariant under translations),
16
the commutator [W , W ] = i W P implies that the W s by themselves are not

generators of a closed algebra.
To nd more labels we take P as given and look for all elements of the Lorentz group that
commute with P . This denes little groups:
Massive particles, p = (m,
0, 0, 0
), have rotations as their little group,
invariant under rot.
since they leave P invariant. From the denition of W , it follows that
W0 = 0 ,
Wi = m Ji .
Thus, C1 = P 2 with eigenvalue m2 , C2 = P 2 J 2 with eigenvalue m2 j(j + 1), hence

a particle with non-zero mass is an irreducible representation of the Poincar group
e
, j .
with labels |m, j; p 3
Massless particles have p = (|p|, p) and W eigenvalues p (see Part III Particles
and Symmetries course). Thus, = j p/|p| is the helicity.
States are thus labelled |0, 0; p , =: |p , . Under CPT4 , those states transform
to |p , . must be integer or half integer5 = 0, 1 , 1, ..., e.g. = 0 (Higgs),
2
1
= 2 (quarks, leptons), = 1 (, W , Z 0 , g) and = 2 (graviton). Note that
massive representations are CPT self-conjugate.
2.4
N = 1 supersymmetry representations
For N = 1 supersymmetry, C1 = P P is still a good Casimir, C2 = W W , however, is

not. One can have particles of dierent spin within one multiplet. To get a new Casimir
C2 (corresponding to superspin), we dene

B := W
Q ( ) Q ,

4
:= B P B P
C2 := C C .
2.4.1
Bosons and fermions in a supermultiplet
In any supersymmetric multiplet, the number nB of bosons equals the number nF of

fermions,
nB = nF .
To prove this, consider the fermion number operator (1)F = ()F , dened via
()F |B
()F |F
= |B ,
= |F .
This new operator ()F anticommutes with Q since

()F Q |F
= ()F |B
= |B
= Q |F
= Q ()F |F
()F , Q
= 0.
See the Standard Model Part III course for a rough proof of the CPT theorem, which states that any
local Lorentz invariant quantum eld theory is CPT invariant.
5
See the Part II Principles of Quantum Mechanics course.
17
Next, consider the trace (in the operator sense, i.e. over elements of the multiplet)
Tr ()F Q , Q
()F Q Q + ()F Q Q
= Tr
anticommute
cyclic perm.
= Tr Q () Q + Q ()F Q
F
= 0.
On the other hand, it can be evaluated using {Q , Q } = 2( ) P ,
Tr ()F Q , Q
= Tr ()F 2 ( ) P
= 2 ( ) p Tr ()F
where P is replaced by its eigenvalues p for the specic state. The conclusion is
0 = Tr ()F
B| ()F |B +
=
bosons
=
bosons
Tr ()F
2.4.2
B|B
fermions
F |F
fermions
F | ()F |F
= nB nF .
is known as the Witten index.
Massless supermultiplet
States of massless particles have P - eigenvalues p = (E, 0, 0, E). The Casimirs
C1 = P P and C2 = C C are zero. Consider the algebra (implicitly acting on our

, on the right hand side)
massless state |p
Q , Q
= 2 ( ) P = 2 E 0 + 3
= 4E
10
00
which implies that Q2 is zero in the representation:
p , | Q2 , Q2 |p ,
= 0 Q2 |p ,
= Q2 |p ,
= 0.
We may also nd one element |p , such that Q1 |p , = 0.

From our previous commutation relation,
1
1
[W , Q ] = P [M , Q ] = P ( ) Q

2
2
(2.4)
and the denition of W , in this representation

i
[W0 , Q ]|p , = 03jk p3 [ j , k ]Q

8
1

|p , = p3 ( 3 Q) |p , .
2
(2.5)
So, remembering that p3 = p0 and, for massless representations, W0 |p , = p0 |p , ,

1
W0 Q2 |p , = [W0 , Q2 ] + Q2 p0 |p , = ( )p0 Q2 |p , .
2
18
Thus, Q2 = Q1 decreases the helicity by 1/2 a unit6 . The normalised state is then
|p ,
Q
1
= 1 |p ,
2
4E
and there are no other states, since Eq. 2.6 Q1 |p ,
Q1 |p ,
1
1
1
= Q1 Q1 |p , =
2
4E
4E
(2.6)
1
2
= 0 and
Q1 , Q1 Q1 Q1 |p , =
4E|p , ,
Thus, we have two states in the supermultiplet: a boson and a fermion, plus CPT conjugates:
|p , ,
|p , 1 .
2
1
1
There are, for example, chiral multiplets with = 0, 2 , vector- or gauge multiplets ( = 2 , 1
gauge and gaugino)
= 0 scalar
squark
slepton
Higgs
= 1 fermion
2
quark
lepton
Higgsino
1
2
fermion
photino
gluino
W ino, Zino
= 1 boson
photon
,
gluon
W, Z
as well as the graviton with its partner:

=
3
2
fermion = 2 boson
gravitino graviton
Question: Why do we put matter elds in the = {0, 1 } supermulti2

plets rather than in the = { 1 , 1} ones?
2
2.4.3
Massive supermultiplet
In case of m = 0, in the centre of mass frame there are P - eigenvalues p = (m, 0, 0, 0)

and Casimirs
C1 = P P = m2 ,
C2 = C C
= 2 m4 Y i Yi ,
where Yi denotes superspin

Yi = Ji
1
Q i Q ,
4m
Yi , Yj
= iijk Yk .
The eigenvalues of Y 2 = Y i Yi are y(y +1), so we label irreducible representations by |m, y .
Again, the anticommutation - relation for Q and Q is the key to get the states:
Q , Q
6
= 2 ( ) P = 2 m ( 0 )
Note that we have used natural units, therefore
= 1.
19
= 2m
10
01
Let | be the ground state, annihilated by Q1,2 . Consequently,

Yi |
= Ji |
1
Q i Q|
4m
= Ji | ,
i.e. for | , the spin j and superspin y are the same. So for given m, y:
= |m, j = y; p , j3
We may obtain the rest of the supersymmetry multiplet by deriving the commutation
relations
1
1

[Q , Ji ] = (i ) Q ,
[Ji , Q ] = (i ) Q
(2.7)
2
2
from the supersymmetry algebra. Thus,
Q1
1
a |j3 := |j3 = |j3 ,
1
2
2m
1
Q2
a |j3 := |j3 = |j3 + .
2
2
2m
(2.8)
We may use Eq. 2.7 to derive

3
[J 2 , Q ] = Q (i ) Q Ji ,
(2.9)
[J3 , a a ] = [J 2 , a a ] = 0
1 2
1 2
(a) y = 0
Let us now consider a specic case, y = 0. We dene J := J1 iJ2 , which lowers/raises
spin by 1 unit in the third direction (see Part II Principles of Quantum Mechanics notes)
but leaves the total spin unchanged. Using Eq. 2.9, and | := |m, 0, 0 ,
3
a1
J 2 a | = Q1 | a J | a J3 | =: j(j + 1) | .
1
2
1
4
zero
zero
Hence a | has j = 1/2 and you can check that j3 = 1/2. Similarly, a | = |m, 1/2, 1/2 .
1
2
The remaining state
| := a a | = a a |
2 1
1 2
represents a dierent spin j object.
Question: How do we know that | = | ?
Thus, for the case y = 0, we have states
|
= |m, j = 0; p , j3 = 0
a a |
1 2
= |m, j = 0; p , j3 = 0
a |
1,2
1
1
= |m, j = 2 ; p , j3 = 2
20
=: |
(b) y = 0
The case y = 0 proceeds slightly dierently. The doublet Q is a doublet (i.e. spin 1/2)
of the right-handed SU (2) in SL(2, C), as Eq. 2.2 shows. The doublet (a , a ) acting on
1
2
| behaves like the combination of two spins: 1 and j, from Eq. 2.8. This yields a linear
2
1
1
combination of two possible total spins j + 2 and j 2 with Clebsch Gordan coecients
1
1
1
ki (recall j 2 = (j 2 ) (j + 2 )):
a |
2
a |
1
1
= k1 |m, j = y + 2 ; p , j3 +
= k3 |m, j = y + 1 ; p , j3
2
We also have a1 |j3 = |j3

2 |m, j = y; p , j3 ,
1
2
1
2
1
2
and a2 |j3 = |j3 +
+ k2 |m, j = y 1 ; p , j3 +
2
+ k4 |m, j = y 1 ; p , j3
2
1
2
. In total, we have
1
1 |m, j = y + 2 ; p , j3 ,
(4y+2) states
1
2
1
2
1 |m, j = y 1 ; p , j3 ,
2
(2y+2) states
(2y) states
in a |m, y multiplet, which is of course an equal number of bosonic and fermionic states.
Notice that in labelling the states we have the value of m and y xed throughout the
multiplet and the values of j change state by state (as is proper, since in a supersymmetric
multiplet there are states of dierent spin).
2.4.4
Parity
Since {Q , Q } = 2( ) P ,
under parity P (with phase
we need the following transformation rules for Q and Q

factor P such that |P | = 1):
1
Parity interchanges (A, B) (B, A), i.e. ( 2 , 0) (0,
1
2 ).
P Q P 1 = P ( 0 ) Q

P Q P 1 = P ( 0 ) Q

This ensures P P P 1 = (P 0 , P )

Question: Calculate P {Q , Q }P 1 , checking that it is equivalent to

2( ) P P P 1 .
and has the eect that P 2 Q P 2 = Q . Moreover, consider the two j = 0 massive states
| and | : Since Q | = 0, whereas Q | = 0, and since parity swaps Q Q ,

it also swaps | | . To get ground states with a dened parity, we need linear
combinations
| := | | ,
P | = | .
These states are called scalar (|+ ) and pseudo-scalar (| ) states.
2.5
Extended supersymmetry
Having discussed the algebra and representations of simple (N = 1) supersymmetry, we

will turn now to the more general case of extended supersymmetry N > 1.
21
2.5.1
Algebra of extended supersymmetry
Now, the spinor generators get an additional label A, B = 1, 2, ..., N . The algebra is the
same as for N = 1 except for
QA , QB
= 2 ( ) P A B
= Z AB ,
QA , QB

QA , QB
= (Z )AB

with antisymmetric central charges Z AB = Z BA commuting with all the generators

Z AB , P
Z AB , M
Z AB , QA
Z AB , Z CD
Z AB , Ta
= 0.
They form an abelian invariant sub-algebra of internal symmetries. Recall that [Ta , Tb ] =
iCabc Tc . Let G be an internal symmetry group, then dene the R symmetry H G to
be the set of G elements that do not commute with the supersymmetry generators, e.g.
Ta G satisfying
QA , Ta = Sa A B QB = 0
is an element of H. If the eigenvalues of Z AB are all zero, then the R symmetry is

H = U (N ), but with some eigenvalues of Z AB = 0, H will be a subgroup of U (N ). The
existence of central charges is the main new ingredient of extended supersymmetries. The
derivation of the previous algebra is a straightforward generalisation of the one for N = 1
supersymmetry.
2.5.2
Massless representations of N > 1 supersymmetry
As we did for N = 1, we will proceed now to discuss massless and massive representations.
We will start with the massless case which is simpler and has very important implications.
Let p = (E, 0, 0, E), then (similar to N = 1).
QA , QB |p ,
= 4E
10
00
A
B |p ,
QA |p ,
2
= 0
We can immediately see from this that the central charges Z AB vanish since QA |p , = 0
2
implies Z AB |p , = 0 from the anticommutator QA , QB |p , = 0 = 12 Z AB |p , .

1
2
In order to obtain the full representation, we now dene N creation- and N annihilation operators
A
:=
QA
1 ,
2 E
:=
QA
1
2 E
22
aA , a
B
= A B ,
to get the following states (starting from ground state | , which is annihilated by all the
aA ):
states
helicity
number of states
|
0
1= N
0
1
A |
0 + 2
N= N
a
1
1
aA aB |
N (N 1) = N
0 + 1
2
2!
1
0 + 3 3! N (N 1)(N 2) = N
aA aB aC |
3
2
.
.
.
.
.
.
.
.
.
N a(N 1) ...a1 | + N
1= N
a
0
N
2
Note that the total number of states is given by
N
k=0
N
k
=
k=0
N
k
1k 1N k
= 2N .
Consider the following examples

N = 2 vector multiplet, as shown in Fig. 2a: so-called because it contains a vector
particle, which must be in the adjoint (i.e. vector-like, or real) representation if the
quantum eld theory is to be renormalisable. We can see that this N = 2 multiplet
can be decomposed in terms of N = 1 multiplets: one N = 1 vector and one N = 1
chiral multiplet.
N = 2 CPT self-conjugate hyper - multiplet, see Fig. 2b. Again this can be decomposed in terms of two N = 1 multiplets: one chiral, one anti-chiral.
N = 4 vector - multiplet (0 = 1)
1
4
6
4
1
= 1
1
= 2
= 0
= +1
2
= +1
3
This is the single N = 4 multiplet with states with || < 2 . It consists of one
N = 2 vector supermultiplet plus a CPT conjugate and two N = 2 hypermultiplets.
Equivalently, it consists of one N = 1 vector and three N = 1 chiral supermultiplets
plus their CPT conjugates.
N = 8 maximum - multiplet (0 = 2)
1
8
28
56
70
= 2
3
= 2
= 1
= 1
2
= 0
23
=0
a
1
N = 1 chiral supermultiplet
a
2
1
2
=
a
2
a
1
1
2
N = 1 vector supermultiplet
=1
(a) Vector supermultiplet
a
1
N = 1 anti-chiral supermultiplet
= 1
2
a
2
=0
=0
a
2
a
1
=
1
2
N = 1 chiral supermultiplet
(b) hyper supermultiplet
Figure 2. N = 2 vector and hyper multiplets.
From these results we can extract very important general conclusions:

In every multiplet: max min =
N
2
Renormalisable theories have || 1 implying N 4. Therefore N = 4 supersymmetry is the largest supersymmetry for renormalisable eld theories. Gravity is not
renormalisable!
The maximum number of supersymmetries is N = 8. There is a strong belief that
no massless particles of helicity || > 2 exist (so only have N 8). One argument
1
against || > 2 is the fact that massless particles of || > 2 and low momentum
couple to some conserved currents ( j = 0 in = 1 - electromagnetism, T in
= 2 - gravity). But there are no conserved currents for || > 2 (something that
can also be seen from the Coleman Mandula theorem). Also, N > 8 would imply
that there is more than one graviton. See chapter 13 in [4] on soft photons for a
24
detailed discussion of this and the extension of his argument to supersymmetry in an

article by Grisaru and Pendleton (1977). Notice this is not a full no-go theorem,
in particular the limit of low momentum has to assumed.
N > 1 supersymmetries are non-chiral. We know that the Standard Model particles
live on complex fundamental representations. They are chiral since right handed
quarks and leptons do not feel the weak interactions whereas left-handed ones do feel
it (they are doublets under SU (2)L ). All N > 1 multiplets, except for the N = 2
hypermultiplet, have = 1 particles transforming in the adjoint representation
1
which is non-chiral. Then the = 2 particles within the multiplet would transform
in the same representation and therefore be non-chiral. The only exceptions are the
N = 2 hypermultiplets - for these, the previous argument doesnt work because they
1
do not include = 1 states, but since = 2 - and = 1 states are in the same
2
multiplet, there cant be chirality either in this multiplet. Therefore only N = 1, 0
1
can be chiral, for instance N = 1 with 2 predicting at least one extra particle
0
for each Standard Model particle. These particles have not been observed, however.
Therefore the only hope for a realistic supersymmetric theory is: broken N = 1
supersymmetry at low energies E 102 GeV.
2.5.3
Massive representations of N > 1 supersymmetry and BPS states
Now consider p = (m, 0, 0, 0), so
QA , QB
= 2m
10
01
A B .
Contrary to the massless case, here the central charges can be non-vanishing. Therefore
we have to distinguish two cases:
Z AB = 0
There are 2N creation- and annihilation operators
aA :=
QA
,
2m
aA :=
QA
2m
leading to 22N states, each of them with dimension (2y + 1). In the N = 2 case, we
nd:
|
1 spin 0
A
4 spin 1
a |
2
A B
a a | 3 spin 0 , 3 spin 1 ,
aA aB aC |
C
aA aB a aD |
4 spin
1
2
1 spin 0
i.e. as predicted 16 = 24 states in total. Notice that these multiplets are much
larger than the massless ones with only 2N states, due to the fact that in that case,
half of the supersymmetry generators vanish (QA = 0).
2
25
Z AB = 0
Dene the scalar quantity H to be (again, implicitly sandwiching in a bra/ket)
H := ( 0 ) QA A , QA A
0.
As a sum of products AA , H is positive semi-denite, and the A are dened as

A := U AB QB ( 0 )
for some unitary matrix U (satisfying U U = ). We derive

H = 8 m N 2 Tr Z U + U Z
0.
Due to the polar decomposition theorem, each matrix Z can be written as a product
Z = HV of a positive semi-denite hermitian matrix H = H and a unitary phase
matrix V = (V )1 . Choosing U = V ,
0.
H = 8 m N 4 Tr H
= 8 m N 4 Tr Z Z
This is the BPS - bound for the mass m:
m
1
Tr Z Z
2N
1
States of minimal m = 2N Tr
Z Z are called BPS states (due to Bogomolnyi,
Prasad and Sommerfeld). They are characterised by a vanishing combination
QA A , so the multiplet is shorter (similar to the massless case in which Qa = 0)

2
having only 2N instead of 22N states.

For N = 2, we dene the components of the antisymmetric Z AB to be
Z AB
0 q1
q1 0
q1
.
2
More generally, if N > 2 (but N even) we may perform a similarity transform7 such
that
0 q1
0 0 0
q 0
0 0 0
0 q2 0
0 0
0 0 q2 0 0
AB
,
..
(2.10)
Z
=
0 0 .
0 0
. .
. .
..
. .
. .
.
. .
. .
0 qN
2
q N 0
2
If N > 2 but N is odd, we obtain Eq. 2.10 with the block matrices extending to q(N 1)/2 and an extra
column and row of zeroes.
26
xs the BPS conditions holds block by block: m 1 maxi (qi ), since we could dene
2
one H for each block. If k of the qi are equal to 2m, there are 2N 2k creation
operators and 22(N k) states.
k = 0
N
0 < k <
2
N
k =
2
22N states, long multiplet
22(N k) states, short multiplets
2N states, ultra - short multiplet
Let us conclude this section about non-vanishing central charges with some remarks:
(i) BPS states and bounds came from soliton (monopole-) solutions of Yang Mills
systems, which are localised nite energy solutions of the classical equations of
motion. The bound refers to an energy bound.
(ii) The BPS states are stable since they are the lightest centrally charged particles.
(iii) Extremal black holes (which are the end points of the Hawking evaporation and
therefore stable) happen to be BPS states for extended supergravity theories.
Indeed, the equivalence of mass and charge reminds us of charged black holes.
(iv) BPS states are important in understanding strong-weak coupling dualities in
eld- and string theory.
(v) In string theory extended objects known as D branes are BPS.
Superspace and Superelds
So far, we have just considered 1 particle states in supermultiplets. Our goal is to arrive at
a supersymmetric eld theory describing interactions. Recall that particles are described
by elds (x ) with the properties:
they are functions of the coordinates x in Minkowski space-time
transforms under the Poincar group
e
In the supersymmetric case, we want to deal with objects (X) which
are function of coordinates X of superspace
transform under the super Poincar group.
e
But what is that superspace?
3.1
3.1.1
Basics about superspace

Groups and cosets
We know that every continuous group G denes a manifold MG via its parameters {a }
: G MG ,
g = exp(ia T a )
where dim G = dim MG . Consider for example:
27
G = U (1) with elements g = exp(iQ), then [0, 2], so the corresponding

manifold is the 1 - sphere (a circle) MU (1) = S 1 .
G = SU (2) with elements g =
||2
+ ||2
, where complex parameters and satisfy
= 1. Write = x1 + ix2 and = x3 + ix4 for xk R, then the constraint

for p, q implies 4 x2 = 1, so MSU (2) = S 3
k=1 k
G = SL(2, C) with elements g = ea V , V SU (2) and a is traceless and hermitian,

i.e.
x1
x2 + ix3
a=
x2 ix3 x1
for xi R, so MSL(2,C) = R3 S 3 .
To be more general, lets dene a coset G/H where g G is identied with gh h H G,
e.g.
G = U1 (1) U2 (1) g = exp i(1 Q1 + 2 Q2 ) , H = U1 (1) h = exp(iQ1 ). In
G/H = U1 (1) U2 (1) /U1 (1), the identication is
g h = exp i (1 + ) Q1 + 2 Q2
= exp i (1 Q1 + 2 Q2 )
= g,
so only 2 contains an eective information, G/H = U2 (1).

G/H = SU (2)/U (1) SO(3)/SO(2): Each g SU (2) can be written as g =
=

i i ) makes eectively real.
, identifying this by a U (1) element diag(e , e
2
2
Hence, the parameter space is the 2 sphere (1 + 2 + 2 = 1), i.e. MSU (2)/U (1) = S 2 .
More generally, MSO(n+1)/SO(n) = S n .

Minkowski = Poincar / Lorentz = { , a }/{ } simplies to the translations
e
{a = x } which can be identied with Minkowski space.
We dene N = 1 superspace to be the coset
Super Poincar / Lorentz =
e
, a , ,
Recall that the general element g of super Poincar group is given by

e

g = exp i ( M + a P + Q + Q ) ,
where Grassmann parameters , reduce anticommutation relations for Q , Q to com

mutators because eg {Q , } = 0:
Q , Q
= 2 ( ) P
28
Q , Q
= 2 ( ) P .
Figure 3. Illustration of the coset identity G/H = U1 (1) U2 (1) /U1 (1) = U2 (1): The blue horizontal
line shows the orbit of some G = U1 (1) U2 (1) element g under the H = U1 (1) group which is divided out.
All its points are identied in the coset. Any red (dark) vertical line contains all the distinct coset elements
and is identied with its neighbours in 1 direction.
3.1.2
Properties of Grassmann variables
Superspace was rst introduced in 1974 by Salam and Strathdee [6, 7]. Recommendable
books about this subject are [8] and [9].
Let us rst consider one single variable . When trying to expand a generic (analytic)
function in as a power series, the fact that squares to zero, 2 = 0, cancels all the terms
except for two,
f () =
fk k
= f0 + f1 + f2 2 + ...
k=0
= f0 + f1 .
So the most general function f () is linear. Of course, its derivative is given by

For integrals, dene
df
d
:= 0 =
d = 0 ,
d
df
d
= f1 .
as if there were no boundary terms. Integrals over are left to talk about: To get a
non-trivial result, dene
d := 1
29
() = .
The integral over a function f () is equal to its derivative,

d f () =
df
.
d
d (f0 + f1 ) = f1 =
Next, let , be spinors of Grassmann numbers. Their squares are dened by

:=
:= ,
1

= ,

2
Derivatives work in analogy to Minkowski coordinates:
:=
.
2
:=

where { , } = { , } = 0. As for multi-dimensional integrals,
d1
1
2
d2 2 1 =
d1
d2 = 1 ,
which justies the denition

d2 :=
1
2
d1
Note that { d1 ,
d2
d2 = 1 ,
d2 } = { d1 ,
d2 } = 0. Written in terms of :
1
d2 = d d ,
4
or
1
d2 = d d ,
4
3.1.3
d2 () () = 1 .
d2
d2 =
d d .

4

d2 = d d .
Denition and transformation of the general scalar supereld
To dene a supereld, recall properties of scalar elds (x ):

function of space-time coordinates x
transformation under Poincar
e
Treating as an operator, a translation with parameter a will change it to
exp(ia P ) exp(ia P ) .
(3.1)
But (x ) is also a Hilbert vector in some function space F, so

(x ) exp(ia P ) (x ) =: (x a )
P = i .
(3.2)
P is a representation of the abstract operator P acting on F. Comparing the two

transformation rules Eqs. 3.1,3.2 to rst order in a , we get the following relationship:
1 ia P 1 + ia P
1 ia P = i , a P
30
= ia P = a .
We shall perform a similar (but super-) transformation on a supereld.
For a general scalar supereld S(x , , ), one can do an expansion in powers of ,
with a nite number of nonzero terms:
S(x , , )
(x) + (x) + (x) + M (x) + N (x) + ( ) V (x)
+ () (x) + () (x) + () ( ) D(x)

(3.3)
We have the transformation of S(x , , ) under the super Poincar group, rstly as a
e
eld operator
S(x , , ) exp i (Q + Q) S exp i (Q + Q) ,
(3.4)
secondly as a Hilbert vector

S(x , , ) exp i (Q + Q) S(x , , ) = S x +x , + , + . (3.5)
Here, denotes a parameter, Q a representation of the spinorial generators Q acting on
functions of , , and c is a constant to be xed later, which is involved in the translation
x = ic ( ) + ic ( ) .
The translation of arguments x , , implies,

c ( )
= +i + c ( )
= i ,
Q = i
Q
P
where c can be determined from the commutation relation which, of course, holds in any
representation:
Q , Q
= 2 ( ) P
Re{c} = 1
It is convenient to set c = 1. Again, a comparison of the two expressions (to rst order in
) for the transformed supereld S is the key to get its commutation relations with Q :
i S , Q + Q
= i Q + Q S
= S
Considering an innitesimal; transformation S S + S = (1 + iQ + iQ)S, where
:=
(x) + (x) + (x) + M (x) + N (x) + ( ) V (x)

+ () (x) + () (x) + () () D(x).

(3.6)
Substituting for Q , Q and S, we get explicit terms for the changes in the dierent parts
of S:
= + ,
31
= 2M + ( )(i + V )
i
M =
2
i
N = +
2
i
D = ( )
= 2N ( )(i V )
V = + + ( )
2
i
= 2D + ( ) V + i( ) M
2
i
= 2D ( ) V + i( ) N
2
as on the second examples sheet. Note that D is a total derivative. Also, we have bosons
and fermions transforming into each other).
3.1.4
Remarks on superelds
S is a supereld it satises S = i(Q + Q)S. Thus:
If S1 and S2 are superelds then so is the product S1 S2 :

(S1 S2 )
= S1 S2 + (S1 )S2
= S1 i (Q + Q) S2
i (Q + Q) (S1 S2 )
i (Q + Q) S1 S2
(3.7)
In the last step, we used the Leibnitz property of the Q and Q as dierential operators.
Linear combinations of superelds are superelds again (straightforward proof).
S is a supereld but S is not:
( S) = (S) = i [(Q + Q)S] = i(Q + Q) ( S)
since [ , Q + Q] = 0. We need to dene a covariant derivative,
D := + i( ) ,
D := i ( )
which satises
D , Q
D , Q
D , Q
D , Q
= 0
and therefore
D , Q + Q
= 0
D S
is supereld.
Also note that super-covariant derivatives satisfy anticommutation relations
D , D
= 2i ( ) ,
D , D
D , D
= 0.
S = f (x) is a supereld only if f = const, otherwise, there would be some f .

For constant spinor c, S = c is not a supereld due to = c.
32
S is not an irreducible representation of supersymmetry, so we can eliminate some of its

components keeping it still as a supereld. In general we can impose consistent constraints
on S, leading to smaller superelds that are irreducible representations of the supersymmetry algebra. There are dierent types depending upon the constraint:
chiral supereld such that D = 0
anti-chiral supereld such that D = 0
vector (or real) supereld V = V

linear supereld L such that DDL = 0 and L = L .
3.2
Chiral superelds
We want to nd the components of a superelds satisfying D = 0. We dene for

convenience
y := x + i .
If = (y, , ), then, since D is a dierential operator,
D = (D )
y,
+ (D y )
+ (D )
.
y,
We have (D ) = 0 and (D y ) = ( i )(x +i ) = i( ) i( ) = 0,
hence the chiral supereld condition becomes = 0. Thus there is no - dependence

and depends only on y and . In components, one nds
(y , ) = (y ) + 2 (y ) + F (y ) ,

where the left handed supercovariant derivative acts as D = +2i( ) y on (y , ).
The physical components of a chiral supereld are as follows: represents a scalar part
1
(squarks, sleptons, Higgs), some s = 2 particles (quarks, leptons, Higgsino) and F is an
auxiliary eld in a way to be dened later. O shell, there are 4 bosonic (complex , F )
and 4 fermionic (complex ) components. Performing a Taylor expansion of around x :
(x , , ) = (x) + 2 (x) + F (x) + i (x)

1
i
() () (x)
() (x)
4
2
Under a supersymmetry transformation

= i Q + Q ,
we nd for the change in components
33
= i 2 + 2 F
F = i 2 .
So F is another total derivative term, just like D in a general supereld. Note that:
The product of chiral superelds is a chiral supereld, since D (S1 S2 ) = (D S1 )S2 +

S2 = 0 if D Si = 0. In general, any holomorphic function f () of a chiral
S1 D
supereld is a chiral supereld.
If is chiral, then = is anti-chiral.

and + are real superelds but neither chiral nor anti-chiral.
Four dimensional supersymmetric Lagrangians

N = 1 global supersymmetry
4.1
We want to determine couplings among superelds which include the particles of the Standard Model. For this we need a prescription to build Lagrangians which are invariant
(up to a total derivative) under a supersymmetry transformation. We will start with the
simplest case of only chiral superelds.
4.1.1
Chiral supereld Lagrangian
In order to nd an object L() such that L is a total derivative under a supersymmetry

transformation, we exploit that:
For a general scalar supereld S = ... + ()( )D(x), the D term transforms as:
D =
For a chiral supereld = ... + ()F (x), the F term transforms as:
F = i 2 .
Since F and D are total derivatives, they have no eect on local physics in the action,
and integrate to zero. For a chiral supereld = . . . + ()F , thus the F term |F is
dened to be whatever multiplies (). Thus, for example, under a SUSY transformation,
d4 x|F = d4 xF d4 x(F + F ) = d4 xF is invariant. Therefore, the most general
Lagrangian for a chiral supereld s can be written as:
L =
K(, )
Khler - potential
a
W ()
super - potential
34
+ h.c.
Where |D refers to the D term of the corresponding supereld (whatever multiplies ()().
. W () is
The function K is known as the Khler potential, a real function of and
a
known as the superpotential, a holomorphic function of the chiral supereld (and therefore
is a chiral supereld itself).
In order to obtain a renormalisable theory, we need to construct a Lagrangian in terms
of operators of dimensionality such that the Lagrangian has dimensionality 4. We know
[] = 1 (where the square brackets stand for dimensionality of the eld) and want [L] = 4
. Terms of dimension 4, such as , m2 and g||4 , are renormalisable, but
couplings with negative mass dimensions are not. The mass dimension of the supereld
is the same as that of its scalar component and the dimension of of is as the same any
standard fermion, that is
[] = [] = 1 ,
From the expansion = +
3
2
[] =
2 + F + ... it follows that
[] =
1
,
2
[F ] = 2 .
This already hints that F is not a standard scalar eld. In order to have [L] = 4 we need:
[KD ] 4 in K
= ... + () ( ) KD
[WF ] 4 in W
= [K] 2 ,
= ... + () WF
[W ] 3 .
A possible renormalisable term for K is , but not + or + since these

contain no Dterms.
Therefore we are lead to the following general expressions for K and W :
K
= ,
= + +
g 3
m 2
+
,
2
3
whose Lagrangian is known as Wess Zumino model:

LW Z
We get the expression for
= +
W ()
+ h.c. .
(4.1)
by substituting
i
1
2 + F + i () () () . (4.2)
4
2
We also perform a Taylor expansion around = (where

W () = W () + ( )
W
+
... + F + ...
35
W
= ):
1
( )2
2
... + () () + ...
2W
2
(4.3)
Substituting Eqs. 4.3,4.2 into Eq. 4.1, we obtain
LW Z = i + F F +
W
1
F + h.c.
2W
+ h.c. .
2
The part of the Lagrangian depending on the auxiliary eld F takes the simple form:
L(F ) = F F +
W
W
F +
F
Notice that this is quadratic and without any derivatives. This means that the eld F does
not propagate. Also, we can easily eliminate F using the eld equations
S(F )
F
S(F )
F
= 0
= 0
W
F +
F +
=
=
= 0
= 0
and substitute the result back into the Lagrangian,

L(F )
=: V(F ) () ,
This denes the scalar potential. From its expression we can easily see that it is a positive
denite scalar potential V(F ) ().
We nish the section about chiral supereld Lagrangian with two remarks,
The N = 1 Lagrangian is a particular case of standard N = 0 Lagrangians: the
scalar potential is positive semi-denite (V 0). Also the mass for scalar eld
(as it can be read from the quadratic term in the scalar potential) equals the one for
1 2
the spinor (as can be read from the term 2 W ). Moreover, the coecient g of
2
Yukawa coupling g() also determines the scalar self coupling, g 2 ||4 . This is the
source of some miraculous cancellations in SUSY perturbation theory: divergences
are removed from some loop corrections, a la Fig. 4.
Figure 4. One loop diagrams which yield corrections to the scalar mass squared. SUSY relates the 4
coupling to the Yukawa couplings ( ) and therefore ensures cancellation of the leading divergence.
36
In general, we may expand K(i , ) and W (i ) around i = i in components,

j
from whence we get the kinetic terms, e.g.
K( , i )
j
= ... +
2K
i
j
i = . . . + Ki i .
j
j
j
Ki is a metric in a space which is a complex Khler - manifold with coordinates i .

a
j
Vector superelds
4.1.3
4.1.2
Denition and transformation of the vector supereld
The most general vector supereld V (x, , ) = V (x, , ) has the form
i
i
V (x, , ) = C(x) + i(x) i(x) +

M (x) + iN (x)
M (x) iN (x)
2
2
1

+ V (x) + () (x) (x)
2
1
1
1
() () D(x) C(x) ,
+ () (x) (x) +
2
2
2
where we have shifted some elds (notably D and ) for convenience. There are 8 bosonic
components C, M , N , D, V and 4 + 4 fermionic ones ( , ).
If is a chiral supereld, then i( ) is a vector supereld. It has components:
= i
2
=
1
(M + iN ) = F
2
V = +
= D = 0
Question: Can you derive these relations by substituting in for ?

We can dene a generalised gauge transformations of vector elds via
V
V + i ,
which induces a standard gauge transformation for the vector component of V

V V 2Re()
=: V + .
Then we can choose , , F within to gauge away some of the components of V , as

long as we have constructed a Lagrangian that is invariant under the generalised gauge
transformation.
37
4.1.4
Wess Zumino gauge
We can choose the components of above: , , F in such a way to set C = = M =

N = 0. This denes the Wess Zumino (WZ) gauge, and we shall use this from now on. A
vector supereld in Wess Zumino gauge reduces to the form
VWZ (x, , ) = ( ) V (x) + () (x)
+ () (x)
() () D(x) .
2
The physical components of a vector supereld are: V corresponding to gauge particles
(, W , Z, gluon), the and to gauginos and D is an auxiliary eld, to be dened later.

Powers of VWZ are given by
2
VWZ =
() () V V ,
2
2+n
VWZ = 0 n N .
Note that the Wess Zumino gauge is not supersymmetric, since VWZ VWZ under su
persymmetry. However, under a combination of supersymmetry and generalised gauge
transformation VWZ VWZ we can end up with a vector eld in Wess Zumino gauge.
4.1.5
Abelian eld strength supereld
Recall that a non-supersymmetric complex scalar eld coupled to a gauge eld V via
covariant derivative D = iqV transforms like
(x) exp iq(x) (x) ,
V (x) V (x) + (x)
under local U (1) with charge q and local parameter (x).

Under supersymmetry, these concepts Generalized to chiral superelds and vector superelds V . To construct a gauge invariant quantity out of and V , we impose the following
transformation properties:
exp(2iq)
V V + i
exp(2qV ) K
is gauge invariant.
Here, is the chiral supereld dening the generalised gauge transformations. Note that
exp(2iq) is also chiral if is.
Before supersymmetry, we dened
F
= V V
as an abelian eld - strength. The supersymmetric analogy is

1
W := (DD) D V
4
which is chiral.
Question: How does one know that W is chiral?
38
To obtain W in components, it is most convenient to rewrite V in the shifted y = x +

i
i variable (where V (x) = V (y) 2 2 2 V (y)), then the supercovariant

derivatives simplify to
D = + 2i( ) and D = :
W (y, ) = (y) + D(y) + ( ) F (y) i() ( ) (y)
Hence, we see generalised gauge invariance of W : , D and F are all gauge invariant!
4.1.6
Non - abelian eld strength
In this section supersymmetric U (1) gauge theories are generalised to non-abelian gauge
groups. The gauge degrees of freedom then take values in the associated Lie algebra
spanned by hermitian generators T a :
= a T a ,
= Va T a ,
Ta , Tb
= if abc Tc
Just like in the abelian case, we want to keep e2qV invariant under the gauge transformation eiq , but the non-commutative nature of and V enforces a nonlinear
transformation law V V :
exp(2qV ) = exp(iq ) exp(2qV ) exp(iq)
iq
i
( )
V , +
V = V
2
2
+ ...
The commutator terms are due to the Baker Campbell Hausdor formula for matrix exponentials
1
exp(X) exp(Y ) = exp X + Y +
X , Y + ... .
2
The eld strength supereld W also needs some modication in non-abelian theories. Recall that the eld strength tensor F of non-supersymmetric Yang Mills theories transforms
to U F U 1 under unitary transformations. Similarly, we dene
W :=
1
(DD) exp(2qV ) D exp(2qV )
8q
and obtain a gauge covariant quantity.

In Wess Zumino gauge, the supersymmetric eld strength can be evaluated as
1
a
W (y, ) = (DD) D V a (y, , ) + i V b (y, , ) V c (y, , ) f a bc

4
a

= a (y) + Da (y) + ( ) F (y) i() ( ) D a (y)
where
a
a
b
F := Va V + q f a bc V Vc
b
D a := a + q V c fbc a
39
4.1.7
Abelian vector supereld Lagrangian
Before attacking vector supereld Lagrangians, let us rst discuss how we ensured gauge invariance of under local transformations exp iq(x) in the non-supersymmetric
case.
Introduce covariant derivative D depending on gauge potential A
D := iq A ,
A A +
and rewrite kinetic term as

L = D (D ) + ...
Add a kinetic term for A to L
L = ... +
1
F F ,
4g 2
= A A .
With SUSY, the Khler potential K = is not invariant under

a
exp(2iq) ,
exp 2iq( )
for chiral . Our procedure to construct a suitable Lagrangian is analogous to the nonsupersymmetric case (although the expressions look slightly dierent):
Introduce V such that
K
= exp(2qV ) ,
V + i ,
i.e. K is invariant under our generalised gauge transformation.

Add kinetic term for V with coupling
Lkin = f () (W W )
+ h.c.
which is renormalisable if f () is a constant f = . Sometimes in this case we write

( ) = 1/g 2 . For general f (), however, it is non-renormalisable. We will call f the
gauge kinetic function.
A new ingredient of supersymmetric theories is that an extra term can be added to
L. It is also SUSY/gauge invariant (for U (1) gauge theories) and known as the Fayet
Iliopoulos term:
1
D
LF I = V
=
2
D
The parameter is a constant. Notice that the FI term is gauge invariant for a U (1)
theory because the corresponding gauge eld is not charged under U (1) (the photon
is chargeless), whereas for a non-abelian gauge theory the gauge elds (and their
corresponding D terms) would transform under the gauge group and therefore have
to be forbidden. This is the reason the FI term only exists for abelian gauge theories.
40
The renormalisable Lagrangian of super QED involves f = = 1 :

4
L =
exp(2qV )
W ()
+ h.c.
1
W W
4
+ h.c.
+ V
If there were only one supereld charged under U (1) then W = 0. For several superelds
the superpotential W is constructed out of holomorphic combinations of the superelds
which are gauge invariant. In components (using Wess Zumino gauge):
exp(2qV )
= F F + i + q V + i i
+ 2 q + + q (D + q V V ) ||2
Note that
V n3 = 0 due to Wess Zumino gauge
we can augment to D = + iqV by soaking up the terms qV
only chargeless products of i may contribute in W (i ), since for example 1 2 3
exp(2i(q1 + q2 + q3 ))1 2 3 under a U (1) gauge transformation.
In gauge theories, we have W () = 0 if there is only one with a non-zero charge.
Let us examine the W W - term:
W W
= D2
i
1
F F 2i
F F .
2
4
1
In the QED choice f = 4 , the kinetic terms for the vector superelds are given by
Lkin =
1
W W
4
+ h.c. =
1
1 2
D
F F i .
2
4
The last term in W W F involving F = F drops out whenever f () is chosen to

1
be real. Otherwise, it couples as 2 Im{f ()}F F where F F itself is a total derivative

without any local physics.
1
With the FI contribution V D = 2 D, the collection of the D dependent terms in L
L(D) = q D ||2 +
1 2
1
D +
D
2
2
yields eld equations

S(D)
D
= 0
D =
q ||2 .
2
Substituting those back into L(D) ,

L(D) =
1
2
+ q ||2
2
41
=: V(D) () ,
we get a scalar potential V(D) (). Together with V(F ) () from the previous section, the
total potential is given by
V () = V(F ) () + V(D) () =
1
2
+ q ||2
2
0.
Note that one always expands elds around their VEVs. The VEVs are nearly always zero,
but if the scalar potential predicts a non-zero VEV v for the real part of a complex scalar
eld , say, one writes: = (v + h0 + iA0 )/ 2, where h0 and A0 are real scalar elds.
a
In the non-abelian extension, 0 and V(D) () := 1 Da Da , where Da = Tij j ,
i
2
where a is an adjoint group label, and i, j are elements of the representation of . Also,
:= a T a , V := Va T a , and there are other less trivial complications in W and in the
generalised gauge transformations as well. See Bailin and Love for all of the details.
4.1.8
Action as a superspace integral
Without SUSY, the relationship between the action S and L is

S
d4 x L .
To write down a similar expression for SUSY - actions, recall
d4 () () = 1 .
d2 () = 1 ,
This provides elegant ways of expressing K
L = K
=
d4 K +
+ h.c.
and so on:
f W W
d2 W + h.c.
+ h.c. + V |D
d2 f W W + h.c.
We end up with the most general action involving several chiral superelds i
S K , exp(2qV ), i , W i , f i ,
i
d4 x
=
+
d4 K + V
d4 x
d2 W + f W W + h.c. .
Recall that the FI term V can only appear in abelian U (1) gauge theories and that the
non-abelian generalisation of the W W term requires an extra trace to keep it gauge
invariant:
Tr W W
Tr eiq W W eiq
= Tr W W eiq eiq
= 1
Thus, we have seen that in general the functions K, W, f and the FI constant determine
the full structure of N = 1 supersymmetric theories (up to two derivatives of the elds as
usual). If we know their expressions we know all the interactions among the elds.
42
N = 2, 4 global supersymmetry
4.2
For N = 1 SUSY, we had an action S depending on K, W , f and . What will the N 2

actions depend on?
We know that in global supersymmetry, the N = 1 actions are particular cases of nonsupersymmetric actions (in which some of the couplings are related, the potential is positive,
etc.). In the same way, actions for extended supersymmetries are particular cases of N = 1
supersymmetric actions and will therefore be determined by K, W , f and . The extra
supersymmetry will put constraints to these functions and therefore the corresponding
actions will be more rigid. The larger the number of supersymmetries the more constraints
on actions arise.
4.2.1
N =2
Consider the N = 2 vector multiplet

A
where the A and are described by a vector supereld V and the , by a chiral
supereld .
N = 2 SUSY enforces W = 0 in the action. K and f can be written in terms of a single
holomorphic function F() called the prepotential:
f () =
2F
,
2
K(, ) =
1
2i
exp(2V )
F
h.c.
The full perturbative action does not contain any corrections for more than 1 loop,
F
2
2
: (tree level)
ln
2
2
: (1 loop)
where denotes some cuto. These statements apply to the Wilsonian eective action.
Note that:
Perturbative processes usually involve series
exp gc2
n an g
with small coupling g 1.
is a non-perturbative example (no expansion in powers of g possible).
There are obviously more things in QFT than Feynman diagrams can tell, e.g. instantons
and monopoles.
We decompose the N = 2 prepotential F as
F() = F1loop + Fnon-pert
where Fnon-pert for instance could be the instanton expansion k ak exp gc2 k . In 1994,
Seiberg and Witten achieved such an expansion in N = 2 SUSY [11].
43
4.2.2
N =4
As an N = 4 example, consider the vector multiplet,
A
2
1 + 3
2 .
1
3
N =2 vector
N =2 hyper
In N = 4, there are no free functions at all, but we have a free parameter:

f
F F
4
g2
F F
N = 4 is a nite theory, moreover its function vanishes. Couplings remain constant at any
scale, therefore we have conformal invariance. There are nice transformation properties
under modular S duality,
a + b
,

c + d
where a, b, c, d form a SL(2, Z) matrix. Finally, as an aside, major developments in string
and eld theories have led to the realisation that certain theories of gravity in Anti de Sitter
space are dual to eld theories (without gravity) in one less dimension, that happen
to be invariant under conformal transformations. This is the AdS/CFT correspondence
allowing one to describe gravity (and string) theories in domains where they are not well
understood (the same benet applies to eld theories as well). The prime example of this
correspondence is AdS in 5 dimensions dual to a conformal eld theory in 4 dimensions
that happens to possess N = 4 supersymmetry.
4.3
Non-renormalisation theorems
There are some important properties of K, W , f and in N = 1 SUSY. It was shown

by using supergraph perturbation theory (a generalisation of the usual Feynman rules to
superspace), that any radiative corrections in a SUSY theory can be written as d4 g,
where the function g contains no functions of or . This result (and some other similar
ones) imply that:
The interactions in K are corrected order by order in perturbation theory
W () and are not renormalised in perturbation theory
f () only receives one loop - corrections
The non-renormalisation of the superpotential is one of the most important results in
supersymmetric eld theory. The simple behaviour of f and the non-renormalisation of
also have interesting consequences.
44
4.3.1
History
In 1977 Grisaru, Siegel, Rocek showed using supergraphs that, except for 1 loop
corrections in f , quantum corrections only come in the form
d4 x
d4
...
In 1993, Seiberg (based on string theory arguments by Witten 1985) used symmetry
and holomorphicity arguments to establish these results in a simple an elegant way [10].
For more details, see Ref. [5] (section 27.6).
4.4
A few facts about local supersymmetry
We have seen that a supereld transforms under supersymmetry as

= i (Q + Q) .
The question arises if we can make a function of space-time coordinates (x), i.e. extend
SUSY to a local symmetry. The answer is yes, and the corresponding theory is supergravity.
How did we deal with local (x) in internal symmetries? We introduced a gauge eld A
coupling to a current J via an interaction term A J . The current J is conserved and
the corresponding charge q is constant
q =
d3 x J 0 = const .
When we make the Poincar parameters space-time dependent, we obtain a theory of grave
ity. The metric g as a gauge eld couples to the current T via g T . Conservation
T = 0 implies constant total momentum
P =
d3 x T 0 = const .
Now consider local SUSY. The generalised gauge eld is the spin 3/2 gravitino with
associated supercurrent J and SUSY charge

0
d3 x J .
Q =
The scalar potential of global SUSY VF is modied in supergravity to (where i =

VF
= exp
Di W
K
2
Mpl
(K 1 )ij Di W D W 3
j
:= i W +
i ):
|W |2
2
Mpl
1
2 (i K) W .
Mpl
Note that in the Mpl limit, gravity is decoupled and the global supersymmetric scalar
potential VF = (K 1 )ij i W W restored. Notice that for nite values of the Planck mass,
j
the potential VF above is no longer positive. The extra (negative) factor proportional to
3|W |2 comes from the auxiliary elds of the gravity multiplet.
45
Supersymmetry breaking
5.1
Preliminaries
We know that elds i of gauge theories transform as

i
exp(ia T a )
j
i
i = ia (T a )i j j
j ,
under nite and innitesimal group elements. By Goldstones theorem, gauge symmetry is
broken8 if the vacuum state (vac )i transforms in a non-trivial way, i.e.
(a T a )i j (vac )j
= 0.
vac is the value that the eld takes when it minimises the potential V (). Suppose
we have a U (1) symmetry, and let = exp(i) in complex polar coordinates, then
innitesimally
= i = = 0 ,
= .
corresponds to the massless Goldstone boson (this is eaten by the gauge boson via the
Higgs mechanism if the U (1) is a gauge symmetry).
Similarly, we speak of broken SUSY if the vacuum state |vac satises
Q |vac
= 0.
Let us consider the anticommutation relation {Q , Q } = 2( ) P contracted with
( ) ,
( ) Q , Q
= 2 ( ) ( ) P = 4 P = 4 P ,
in particular the ( = 0) component using 0 = :
( 0 ) Q , Q
Q (Q ) + (Q ) Q
= 4P0 = 4E
=1
This has two very important implications:

E 0 for any state, since Q (Q ) + (Q ) Q is positive semi-denite
In broken SUSY, Q |vac = 0, so vac|[Q (Q ) + (Q ) Q ]|vac > 0, hence the
energy density is strictly positive, E > 0
Since W is not renormalised to all orders in perturbation theory, we have an important
result: If global SUSY is unbroken at tree level, then it also unbroken to all orders in
perturbation theory. This means that in order to break supersymmetry spontaneously, one
has to do it non-perturbatively.
46
(a) intact
$
(b) $$
gauge
$
SUSY
(c) $$
$
$
SUSY gauge
(d) $$ $$
Figure 5.
Various symmetry breaking scenarios: SUSY is broken, whenever the minimum potential
energy V (min ) is nonzero. Gauge symmetry is broken whenever the potentials minimum is attained at a
nonzero eld conguration min = 0 of a gauge non-singlet.
5.1.1
F term breaking
Consider the transformation - laws under SUSY for components of a chiral supereld ,
2
=
2 F + i 2
F = i 2 .
If one of , , F = 0, then SUSY is broken. But to preserve Lorentz invariance,
we need
= = 0
as they both transform non-trivially under the Lorentz group. So our SUSY breaking
condition simplies to
SUSY
F = 0.
$$$
8
See spontaneous symmetry breaking notes in the Standard Model course.
47
Only the fermionic part of will change,
= 0,
2 F
= 0,
so call a Goldstone fermion or the goldstino (although it is not the SUSY partner of
some Goldstone boson). Remember that the F term of the global SUSY scalar potential
is given by
W W
,
V(F ) = Ki1
j
i
j
and F term SUSY breaking is equivalent to a positive vacuum expectation value

F term $$$
SUSY
5.1.2
V(F )
> 0.
ORaifertaigh model
The ORaifertaigh model involves a triplet of chiral superelds 1 , 2 , 3 for which the
Khler potential and superpotential are given by
a
= i ,
i
= g 1 (2 m2 ) + M 2 3 ,
3
m.
From the F eld equations,
F1 =
F2 =
F3 =
W
1
W
2
W
3
= g (2 m2 )
3
= M 3
= 2 g 1 3 + M 2 .
We cannot have Fi = 0 for all i = 1, 2, 3 simultaneously, so this form of W indeed breaks

SUSY. In order to see some eects of the SUSY breaking, we determine the spectrum. For
this, we need to minimise the scalar potential:
V
If m2 <
M2
,
2g 2
W
i
W
j
= g 2 2 m2
3
+ M 2 |3 |2 + 2 g 1 3 + M 2
then the minimum of the potential is at

=
= 0,
1 arbitrary =
= g 2 m4 > 0 .
As usual, we expand the elds around the vacuum expectation values 1,2,3 . For simplicity,
we take the example of 1 = 0 and compute the spectrum of fermions and scalars.
Consider the fermion mass term

0 0 0
1
2W
1
1

i
j =
1 2 3 0 0 M 2
2
i j
2
0M 0
3
48
Figure 6. Example of a at direction: If the potential takes its minimum for a continuous range of eld
congurations (here: for any 2 R), then it is said to have a at direction. As a result, the scalar eld
1 will be massless.
in the Lagrangian, which yields the i masses

m1
= 0,
m2
= m3
= M .
1 turns out to be the goldstino (due to 1 F1 = 0). To determine the scalar masses,
we examine the quadratic terms in V :
2
Vquad = m2 g 2 (2 + 3 ) + M 2 |3 |2 + M 2 |2 |2
3
m1 = 0 ,
m2 = M
3 is a complex eld, which we must split into its real and imaginary parts 3 =
since they have dierent masses:
m2 = M 2 2 g 2 m2 ,
a
1
(a+ib),
2
m2 = M 2 + 2 g 2 m2 .
b
Summarising, we have the following spectrum:

We generally get heavier and lighter superpartners since the supertrace of M i.e. STr M 2
(which treats bosonic and fermionic parts dierently) vanishes:
STr M 2
(1)2j+1 (2j + 1) m2 = 0 ,
j
:=
j
where j represents the spin of the particles. This is generic for tree level directly broken
SUSY.
5.1.3
D term breaking
Consider a vector supereld V = (, V , D),

D
= 0
SUSY
$$$ .
is a goldstino (which, again, is not the fermionic partner of any Goldstone boson). See
examples sheet 3, where you are asked to work out some details.
49
Figure 7.
Mass splitting of the real- and imaginary part of the third scalar 3 in the ORaifertaigh
model.
5.1.4
Breaking local supersymmetry
The supergravity multiplet contains new auxiliary - elds Fg with Fg = 0 for broken
SUSY.
The F - term is proportional to
F
DW
W
1 K
+
W .
2
Mpl
The scalar potential V(F ) has a negative gravitational term,

V(F ) = exp
K
2
Mpl
(K 1 )ij Di W D W 3
j
|W |2
2
Mpl
That is why both V = 0 and V = 0 are possible after SUSY breaking in supergravity, whereas broken SUSY in the global case required V > 0. This is very
important for the cosmological constant problem (which is the lack of understanding
of why the vacuum energy density today is almost zero, O(103 eV)4 . The vacuum
energy density essentially corresponds to the vacuum expectation value of the scalar
potential at its minimum. In global supersymmetry, we need to make super-particles
heavy, of order 100 GeV or heavier. Thus, global SUSY would naturally give a
contribution to the cosmological constant that is far too large, O(100 GeV)4 , since
the SUSY breaking scale squared appears in the potential with no negative terms. In
supergravity however, it is possible to break supersymmetry at an empirically viable
large energy scale and still to keep the vacuum energy zero. This does not solve the
cosmological constant problem, though.
50
The super Higgs eect: Spontaneously broken gauge theories realise the Higgs mechanism in which the corresponding Goldstone boson is eaten by the corresponding
gauge eld to get a mass. A similar phenomenon happens in supersymmetry. The
goldstino eld joins the originally massless gravitino eld (which is the gauge eld of
N = 1 supergravity) and gives it a mass, in this sense the gravitino receives its mass
by eating the goldstino. The graviton remains massless, however.
Introducing the minimal supersymmetric standard model (MSSM)
The MSSM is based on SU (3)C SU (2)L U (1)Y N = 1 SUSY. We must t all of the
experimentally discovered eld states into N = 1 supermultiplets.
6.1
Particles
First of all, we have vector superelds containing the Standard Model gauge bosons. We
write their representations under (SU (3)C , SU (2)L U (1)Y ) as (pre-Higgs mechanism):
gluons/gluinos
G = (8, 1, 0)
W bosons/winos
B bosons/gauginos
= (1, 3, 0)
B = (1, 1, 0),
which contains the gauge boson of U (1)Y .

Secondly, there are chiral superelds containing Standard Model matter and Higgs elds.
Since chiral superelds only contain left-handed fermions, we place charge conjugated, i.e.
anti right handed fermionic elds (which are actually left-handed), denoted by c
(s)quarks: lepton number L = 0, whereas baryon number B = 1/3 for a (s)quark,
B = 1/3 for an anti-quark.
Qi =
3, 2, 1 ,
6
1, 2
3,
3
uc =
i
left-handed
dc =
i
1,
3,
1
3
anti (right-handed)
(s)leptons L = 1 for a lepton, L = 1 for an anti-lepton. B = 0.

Li =
1
1, 2, 2 ,
ec = (1, 1, +1)
i
anti (right-handed)
left-handed
higgs bosons/higgsinos: B = L = 0.
H2 =
1, 2,
1
2
H1 =
1, 2, 1
2
the second of which is a new Higgs doublet not present in the Standard Model. Thus,
the MSSM is a two Higgs doublet model. The extra Higgs doublet is needed in order
to avoid a gauge anomaly, and to give masses to down-type quarks and leptons.
51
Note that after the breaking of electroweak symmetry (see the Standard Model course),
SU (2)L
the electric charge generator is Q = T3
+ Y /2. Baryon and lepton number correspond
to multiplicative discrete perturbative symmetries in the SM, and are thus conserved,
perturbatively.
Chiral fermions may generate an anomaly in the theory, as shown by Fig. 8. This is where
a symmetry that is present in the tree-level Lagrangian is broken by quantum corrections.
Here, the symmetry is U (1)Y : all chiral fermions in the theory travel in the loop, and yield
a logarithmic divergence proportional to
A :=
LH fi
Yi3
Yi3
RH fi
multiplied by some kinematic factor which is the same for each fermion. If A is non-zero, one
must renormalise the diagram away by adding a B B B counter term in the Lagrangian.
But this breaks U (1)Y , meaning that U (1)Y would not be a consistent symmetry at the
quantum level. Fortunately, A = 0 for each fermion family in the Standard Model.
Figure 8. Anomalous graph proportional to Tr{Y 3 } which must vanish for U (1)Y to be a valid
symmetry at the quantum level. Hyper-charged chiral fermions f travel in the loop contributing to
a three-hypercharge gauge boson B vertex.
Question: Can you show that A = 0 in a Standard Model family?
In SUSY, we add the Higgsino doublet H1 , which yields a non-zero contribution to A. This
must be cancelled by another Higgsino doublet with opposite Y : H2 .
6.2
Interactions
K = exp(2V )i is renormalisable, where
i
Y
1
V := g3 T a Ga + g2 i W i + gY B,
2
2
T a being the Gell-Mann matrices and i being the Pauli matrices.
fa = a where Re{a } =
4
2
ga
determines the gauge coupling constants.
Gauge couplings are renormalised, which ends up giving them renormalisation scale
dependence, which matches onto dependence upon the energy scale at which one is
52
g3
g3
g3 +
g3
g3
g3
Figure 9. Contribution to the one loop QCD beta function 3 from gluon G loops and gluino
G loops. There are other contributing diagrams, some involving loops of quarks and squarks, for
instance.
probing them:
dga ()
3
2
2
= a ga (), ga () = ga (0 ) 2a ln
d
0
(6.1)
where a is a constant determined by which particles travel in the loop in the theory.
For ordinary QCD it is 3 = 7/(16 2 ) whereas for the MSSM, it is 3 = 3/(16 2 )
because of additional contributions from squarks and gluinos to the loops, as in Fig. 9.
Eq. 6.1 is used to extrapolate gauge couplings measured at some energy scale 0
(often taken to be MZ , from LEP constraints) to some other scale . With the SUSY
contributions in the MSSM, the gauge couplings all meet at a renormalisation scale
E 2 1016 GeV, whereas with just the Standard Model contributions, they do
not meet each other at all: see Fig. 10. The meeting of the gauge couplings is a
necessary condition for a Grand Unied Theory, which only has one gauge coupling
(above MGU T 2 1016 GeV).
2
Figure 10. Renormalisation of the structure constants a := ga /4 associated with the
SU (3)C , SU (2)L and U (1)Y groups.
53
For the FI term: we must have = 0, otherwise the scalar potential breaks charge
and colour (because one generates a non-zero vacuum expectation value for a squark,
for instance).
We write down a superpotential containing all terms which are renormalisable and
consistent with our symmetries. If one does this, one obtains two classes of terms,
W = WRp + WRP V . The terms in WRp all conserve baryon number B and lepton
number L, whereas those in WRP V break either B or L:
WRp
WRP V
= (YU )ij Qi H2 uc + (YD )ij Qi H1 dc + YE Li H1 ec + H1 H2

j
j
j
= ijk Li Lj ec + Li Qj dc + uc dc dc + i Li H2 ,
k
ijk
k
ijk i j k
where we have suppressed gauge indices.

Question: Which terms break L and which break B? Why is
there no term H1 H1 ec in WRP V ?
k
k
The rst three terms in WRp correspond to standard Yukawa couplings and give
masses to up quarks, down quarks and leptons, as we shall see. Writing x = 1, 2, 3 as
a fundamental SU (3) index, a, b = 1, 2 as fundamental SU (2) indices, the rst term
in WRp becomes
+
b
0
(YU )ij Qxa H2 uc ab = (YU )ij [ux H2 uc dx H2 uc ].
i
jx
L
jx
L
jx
0
Once the neutral Higgs component develops a vacuum expectation value, H2 := (v2 +
x c
h0 )/ 2, the rst term becomes (YU )ij v2 / 2uLi ujx +. . ., yielding a Dirac mass matrix
2
mu := (YU )ij v2 / 2 for the up quarks. The down quark and lepton masses proceed in
an analogous manner. The fourth term is a mass term for the two Higgs(ino) elds.
If all of the terms in WRP V are present, the interaction shown in Fig. 11 would allow
proton decay p e+ + 0 within seconds, whereas experiments say that it should be
> 1034 years. In order to forbid proton decay an extra symmetry should be imposed.
One symmetry that works is a discrete multiplicative symmetry R parity dened as
R := (1)3(BL)+2S
+1 : Standard Model particles

.
1 : superpartners
It forbids all of the terms in WRP V , but there exist other examples which only ban
some subset.
R parity would have important physical implications:
The lightest superpartner (LSP) is stable.
Cosmological constraints then say that a stable LSP must be electrically and colourneutral (higgsino, photino, zino). It is then a good candidate for cold weakly interacting dark matter.
54
Figure 11. Proton decay due to baryon- and lepton number violating interactions. Both B and
L violating terms must be present for the proton to decay. The matrix element is proportional to
.
1j1
11j
In colliders, the initial state is Rp = +1, implying that superparticles are produced
in pairs. When a superparticle decays, it must do to another (lighter) superparticle
plus some standard model particles.
One ends up with LSPs at the end of the decays. These do not interact with the
detector, and hence appear as unbalanced or missing momentum.
Note that the terms in WRP V can lead to Majorana fermion structure9 . For instance,
W = uc dc dc : we take the F terms as usual in order to nd the Lagrangian in terms
112 1 1 2
of components:
1
L=
u d1 Cd2 ( ) u1 d1 T C d2R
R
112
R
2 112 1 R
plus supersymmetric copies, where C is the charge conjugation matrix and
transpose.
6.3
denotes
Supersymmetry breaking in the MSSM
We cannot break supersymmetry directly in the MSSM, since it preserves STr M 2 = 0.

Applying this to the photon, say: 3m2 +2m2 = 0, which would predict a massless photino
that hasnt been observed. Applying it to up quarks: 2m2 m2 L m2 R = 0, thus one up

u
u
squark must be lighter than the up quark, again this hasnt been observed. We introduce
a hidden sector, which breaks SUSY and has its own elds (which do not directly interact
with MSSM elds) and interactions, and an additional messenger sector
observable
sector, MSSM
9
messenger sector
hidden
sector
This is a familiar structure for people extending the Standard Model to include neutrino masses.
55
This gets around the supertrace rule. There is typically an overall gauge group
SU (3) SU (2) U (1)
$
$
G$$ =: GSM G$$ ,
SUSY
SUSY
$
where the MSSM elds are singlets of G$$ and the hidden sector elds are singlets of
SUSY
GSM .
We have already seen several examples of SUSY breaking theories. One popular SUSYbreaking sector in the MSSM context is that of gaugino condensation: here, some asymptotically free gauge coupling g becomes large at some energy scale . g will renormalise
like Eq. 6.1 with some beta function coecient. Solving the equation, with g 2 () 0,
we obtain = M exp[g 2 (M )/]. M could be some large scale such as the string scale,
5 1017 GeV. It is easy to arrange for M . When the gauge coupling becomes large,
and the theory becomes non-perturbative, one can obtain g g O(3 ), breaking SUSY
10 .
dynamically
The SUSY breaking elds have couplings with the messenger sector, which in turn
have couplings with the MSSM elds, and carry the SUSY breaking over to them. There
are several possibilities for the messenger sector elds, which may determine the explicit
$
form of SUSY breaking terms in the MSSM, including (note here that M$$ is the SUSY
SUSY
breaking in the hidden sector, whereas m is the SUSY breaking that ends up in the
MSSM elds):
gravity mediated $$$

SUSY
If the mediating eld couples with gravitational strength to the standard model, the
couplings are suppressed by the inverse Planck mass Mpl , the natural scale of gravity.
The SUSY breaking mass splitting between MSSM particles and superparticles, m,
becomes
2
M$$
$
SUSY
.
m =
Mpl
We want m 1 TeV and know Mpl 1018 GeV, so
$
M$$ =
SUSY
m Mpl 1011 GeV .
The gravitino gets a mass m 3 of m order TeV from the super Higgs mechanism.
2
gauge mediated $$$

SUSY
$
Messenger elds are charged under both GSM and G$$ . Gauge loops transmit
SUSY
$
SUSY breaking to the MSSM elds. Thus, m gives M$$ /(16 2 ) O(m),
SUSY
i.e. TeV. In that case, the gravitino mass m 3

2
M2 $
SUSY
$ $
Mpl
eV and is the LSP.
anomaly mediated $$$

SUSY
In this case, the auxiliary elds of supergravity get a vacuum expectation value. The
eects are always present, but suppressed by loop factors. They may be dominant if
the tree-level contribution is suppressed for some reason.
10
Here, g is the gaugino of the hidden sector gauge group, and is the hidden gauge group beta function
coecient.
56
log scale
~ MSUSY
Figure 12. Gaugino condensation and supergravity mediated SUSY breaking
Each of these scenarios has phenomenological advantages and disadvantages and solving
their problems is an active eld of research. In all scenarios, the Lagrangian for the observable sector has contributions
$
L = LSUSY + L$$ .
SUSY
In the second term, we write down all renormalisable symmetry invariant terms which do
not reintroduce the hierarchy problem. They are of the form (where i and j label dierent
elds):
1
2
$
L$$ = m2 j + m ij (i j + h.c.) +
SUSY
ij i
2 M
scalar masses
gaugino masses
+ Aijk i j k
trilinear couplings
+ h.c. .
M , m 2 , m2 , Aijk are called soft SUSY breaking terms: they do not reintroduce quadratic
ij
ij
divergences into the theory. Particular forms of SUSY breaking mediation can give relations
between the dierent soft SUSY breaking terms. They determine the amount by which
supersymmetry is expected to be broken in the observable sector, and the masses of the
superparticles for which the LHC is searching.
Explicitly, we parametrise all of the terms that softly break SUSY in the Rp preserving
MSSM, suppressing gauge indices:
$
$
SUSY
L$
= (AU )ij QLi H2 u + (AD )ij QLi H1 d + (AE )ij LLi H1 e +
Rj
Rj
Rp
Rj
2
2
QLi (mQ )ij QLj + Li (mL )ij Lj + uRi (mU )ij uRj + dRi (m2 )ij d + eRi (m2 )ij e +
Rj
Rj
D
E
1
1
1
2

(m2 H1 H2 + h.c.) + m2 |H1 | + m2 |H2 |2 + M3 g g + M2 W W + M1 B B.

3
1
2
2
2
2
Sometimes, m2 is written as B. Often, specic high scale models provide relations between
3
these many parameters. For instance, the Constrained MSSM (which may come from some
57
SPS1a
600
mQl
(2+mHd2)1/2
GeV
400
M3
M2
200
M1
mEr
0
(2+mHu2)1/2
-200
2
SOFTSUSY3.0.5
8 10 12
log10(/GeV)
14
16
Figure 13. An example of renormalisation in the MSSM. A particular high energy theory is
assumed, which has GUT symmetry and implies that the gauginos are all mass degenerate at the
GUT scale. The scalars (e.g the right-handed electron Er and the left-handed squarks Ql) are also
mass-degenerate at the GUT scale. Below the GUT scale though, the masses split and renormalise
separately. When we are scattering at energies O(1) GeV, it is a good approximation to use the
masses evaluated at that renormalisation scale E. We see that one of the Higgs mass squared
2
parameters, 2 + MHu , becomes negative at the electroweak scale, triggering electroweak symmetry
breaking.
string theory or other eld theory) gives the constraints

M1 = M2 = M3 =: M1/2
m2 = m2 = m2 = m2 = m2 := m2 I3
0
Q
L
U
D
E
m2 = m2 = m2
1
2
0
AU = A0 YU , AD = A0 YD , AE = A0 YE
where I3 is the 3 by 3 identity matrix. Thus in the CMSSM, we reduce the large number
of free SUSY breaking parameters down to11 3: M1/2 , m0 and A0 . These relations hold at
the GUT scale, and receive large radiative corrections, as Fig. 13 shows.
6.4
The hierarchy problem
The Planck mass Mpl 1019 GeV is an energy scale associated with gravity and the
electroweak scale Mew 102 GeV is an energy scale associated with symmetry breaking
scale of the Standard Model. The hierarchy problem involves these two scales being so
dierent in magnitude. Actually the problem can be formulated in two parts:
11
One should really include tan = v2 /v1 as well, the ratio of the two Higgs vacuum expectation values.
58
(i) Why is Mew Mpl at tree level? Answering this question is the hierarchy problem.
There are many solutions.
(ii) Once we have solved (i), why is this hierarchy stable under quantum corrections?
This is the technical hierarchy problem and does not have many solutions, aside
from SUSY.
Let us now think some more about the technical hierarchy problem. In the Standard Model
we know that:
Vector bosons are massless due to gauge invariance, that means, a direct mass term
for the gauge particles M 2 A A is not allowed by gauge invariance (A A +
for a U (1) eld, for example).
Chiral fermion masses m are also forbidden for all quarks and leptons by gauge
invariance.
Question: Which symmetry bans say meR eR ?
Recall that these particles receive a mass only through the Yukawa couplings to the
Higgs (e.g. H L R giving a Dirac mass to after H gets a non-zero value12 ).

The Higgs is the only scalar particle in the Standard Model. There is no symmetry
banning its mass term m2 H H in the Standard Model Lagrangian. If the heaviest
H
state in the theory has a mass of , loops give corrections of order 2 /(16 2 ) to
the scalar mass. The corrections come from both bosons and fermions running in
loops. On the other hand, the Z and W bosons are connected to the Higgs mass
parameter by the minimisation of the Higgs potential, and come out to be of the
same order of magnitude. We need the Higgs mass to be mH 125 GeV. This
is unnatural since the loop corrections are much larger: the largest are expected
to be13 O(1017 ) GeV. Therefore even if we start with a Higgs mass of order the
electroweak scale, loop corrections would bring it up to the highest scale in the theory,
/(16 2 ). This would ruin the hierarchy between large and small scales. It is possible
to adjust or ne tune the loop corrections such as to keep the Higgs light, but
this would require cancellations between the apparently unrelated tree-level and loop
contributions to some 15 signicant gures. This ne tuning is considered unnatural
and an explanation of why the Higgs mass (and the whole electroweak scale) can be
naturally maintained to be hierarchically smaller than the Planck scale or any other
large cuto scale is required.
In SUSY, bosons have the same masses as the fermions. Since quarks and leptons are
massless because of gauge invariance, SUSY implies that the squarks and sleptons are
protected too.
12
Notice that with Rparity, the MSSM does not give neutrinos mass. Thus one must augment the
model in some way.
13
This does rely on quantum gravity yielding an eective quantum eld theory that acts in the usual
way.
59
Secondly, SUSY implies that in the explicit computation of loop diagrams (see Fig. 4), the
leading divergences of the bosonic loops cancel against the fermionic loops. This is due to
the fact that the couplings dening SUSY relates the vertices in each diagram to involve
the same coupling. Even when SUSY is softly broken, these leading divergences cancel,
1
leaving us with only a term of O( 162 MSU SY ln ), where MSU SY is the SUSY breaking
mass of some particle in the loop.
Therefore if supersymmetry were exact, fermions and bosons would be degenerate, but if
MSU SY is close to the electroweak scale then it will protect the Higgs from becoming too
heavy. Thus, we expect the superparticle masses to be close to the electroweak scale, and
therefore accessible at the LHC.
6.5
Pros and Cons of the MSSM
We start with a list of unattractive features of the MSSM:

There are 100 extra free parameters in the SUSY breaking sector, making for a
complicated parameter space.
Nearly all of this parameter space is ruled out from avour physics constraints: SUSY
particles could heavily mix in general, then this mixing could appear in loops and
make the quarks mix in a avour changing neutral current, upon which there are
very strong experimental bounds. It could be that this clue is merely telling us that
there is more structure to the MSSM parameter space, though (like in the CMSSM).
The problem. in WRp must be < O(1) TeV, since it contributes at tree-level to
mH . Why should this be, when in principle we could put it to be O(MP l ), because
it does not break any SM symmetries?
These SUSY problems can be solved with further model building.
We close with an ordered list of weak-scale SUSYs successes:
SUSY solves the technical hierarchy problem.
Gauge unication works.
The MSSM contains a viable dark matter candidate, if Rp is conserved.
Electroweak symmetry breaks radiatively.
Acknowledgements
These lecture notes are heavily based on Ref. [1].
Appendix: the Part III Exam

There is a 2 hour examination for this course. You will be asked to complete 2 out of 3
possible questions. I have a habit of putting useful equations, and conventions which I
wish you to follow on the rst side of the exam paper. As ever, you should work through
some past papers to get an idea for the kind of questions you can expect.
60
References
[1] F. Quevedo, S. Krippendorf and O. Schlotterer, Cambridge Lectures on
Supersymmetry and Extra Dimensions, arXiv:1011.1491 [hep-th].
[2] H.J.W. Mller-Kirsten, A. Wiedemann, Supersymmetry, an introduction with
u
conceptual and calculational details, World Scientic
[3] I.L. Buchbinder and S.M. Kuzenko, Ideas and methods of supersymmetry and
supergravity, or, A walk through superspace , CRC Press
[4] S. Weinberg, The quantum theory of elds, Volume I Foundations, Cambridge

University Press
[5] S. Weinberg, The quantum theory of elds, Volume III: Supersymmetry, Cambridge
University Press
[6] A. Salam and J. A. Strathdee, Supergauge Transformations, Nucl. Phys. B 76 (1974)
477.
[7] A. Salam and J. A. Strathdee, On Superelds And Fermi-Bose Symmetry, Phys. Rev.
D 11 (1975) 1521.
[8] F.A. Berezin, A.A. Kirillov, D. Leites Introduction to superanalysis, Reidel (1987).
[9] B. de Witt, Supermanifolds, CUP (1992).
[10] N. Seiberg, Naturalness Versus Supersymmetric Non-renormalization Theorems, Phys.
Lett. B 318 (1993) 469 [arXiv:hep-ph/9309335].
[11] N. Seiberg and E. Witten, Monopole Condensation, And Connement In N = 2
Supersymmetric Yang-Mills, Nucl. Phys. B 426 (1994) 19 [Erratum-ibid. B 430
(1994) 485] [arXiv:hep-th/9407087].
61

Part 3 SuperSymmetry

Hochgeladen von

Dokumentinformationen

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

Part 3 SuperSymmetry

Hochgeladen von

Copyright:

Verfügbare Formate

Prepared for submission to JHEP

Copyright 2013 University of Cambridge. Not to be quoted or reproduced without permission.

Copyright 2013 University of Cambridge. Not to be quoted or reproduced without permission.

2 Supersymmetry algebra and representations

3 Superspace and Superelds

Copyright 2013 University of Cambridge. Not to be quoted or reproduced without permission.

4 Four dimensional supersymmetric Lagrangians

6 Introducing the minimal supersymmetric standard model (MSSM)

Basic theory: QFT

Copyright 2013 University of Cambridge. Not to be quoted or reproduced without permission.

Basic principle: symmetry

For elementary particles, we can dene two general classes of symmetries:

Symmetry is important for various reasons:

Copyright 2013 University of Cambridge. Not to be quoted or reproduced without permission.

so rewrite the Lagrangian to ensure gauge - invariance:

Copyright 2013 University of Cambridge. Not to be quoted or reproduced without permission.

Figure 1. The Mexican hat potential for V = a b ||2

then the minimum is at = 0 (here 0||0 denotes the vacuum expectation

the symmetry of V is lost in the ground state = 0. The existence of hidden

The Standard Model

The Standard Model is well-dened and currently well conrmed by experiments.

Copyright 2013 University of Cambridge. Not to be quoted or reproduced without permission.

Problems of the Standard Model

Modications of the Standard Model

Copyright 2013 University of Cambridge. Not to be quoted or reproduced without permission.

Supersymmetry algebra and representations

Poincar symmetry and spinors

Copyright 2013 University of Cambridge. Not to be quoted or reproduced without permission.

Lorentz transformations leave the metric tensor = diag(1, 1, 1, 1) invariant:

A 4 dimensional matrix representation for the M is

Properties of the Lorentz group

and the Lorentz algebra written in terms of Js and Ks is

These consist of the subgroup of transformations which have det = +1 and 0 1.

which satisfy SU (2) SU (2) commutation relations

Copyright 2013 University of Cambridge. Not to be quoted or reproduced without permission.

Under parity P , (x0 x0 and x x) we have

We can interpret J = A + B as the physical spin.

where is the 4 vector of Pauli matrices

Transformations X X under SO(1, 3) leaves the square

serves the determinant

Representations and invariant tensors of SL(2, C)

The basic representations of SL(2, C) are:

The elements of this representation are called left-handed Weyl spinors.

The conjugate representation

Here are called right-handed Weyl spinors.

The contravariant representations are

Copyright 2013 University of Cambridge. Not to be quoted or reproduced without permission.

That is why is used to raise and lower indices

so contravariant representations are not independent from covariant ones.

so the correct transformation rule is

Similar relations hold for

Let us dene tensors , as antisymmetrised products of matrices:

which satisfy the Lorentz algebra

Copyright 2013 University of Cambridge. Not to be quoted or reproduced without permission.

Products of Weyl spinors

Dene the product of two Weyl spinors as

Copyright 2013 University of Cambridge. Not to be quoted or reproduced without permission.

Choosing the to be anticommuting Grassmann numbers, 1 2 = 2 1 , so =

Bi-spinors with dierent chiralities can be expanded in terms of the . Actually,

the matrices form a complete orthonormal set of 2 2 matrices with respect to

Alike bi-spinors require a dierent set of matrices to expand, and ( T ) :=

then these satisfy the Cliord algebra

Copyright 2013 University of Cambridge. Not to be quoted or reproduced without permission.

The matrix 5 , dened as