# Field theory and the Standard Model

W. Buchmüller and C. Lüdeling Deutsches Elektronen-Synchrotron DESY, 22607 Hamburg, Germany Abstract We give a short introduction to the Standard Model and the underlying concepts of quantum ﬁeld theory. 1

Introduction

In these lectures we shall give a short introduction to the Standard Model of particle physics with emphasis on the electroweak theory and the Higgs sector, and we shall also attempt to explain the underlying concepts of quantum ﬁeld theory. The Standard Model of particle physics has the following key features: – As a theory of elementary particles, it incorporates relativity and quantum mechanics, and therefore it is based on quantum ﬁeld theory. – Its predictive power rests on the regularization of divergent quantum corrections and the renormalization procedure which introduces scale-dependent ‘running couplings’. – Electromagnetic, weak, strong and also gravitational interactions are all related to local symmetries and described by Abelian and non-Abelian gauge theories. – The masses of all particles are generated by two mechanisms: conﬁnement and spontaneous symmetry breaking. In the following chapters we shall explain these points one by one. Finally, instead of a summary, we brieﬂy recall the history of ‘The making of the Standard Model’ [1]. From the theoretical perspective, the Standard Model has a simple and elegant structure: it is a chiral gauge theory. Spelling out the details reveals a rich phenomenology which can account for strong and electroweak interactions, conﬁnement and spontaneous symmetry breaking, hadronic and leptonic ﬂavour physics etc. [2, 3]. The study of all these aspects has kept theorists and experimenters busy for three decades. Let us brieﬂy consider these two sides of the Standard Model before we discuss the details. 1.1 Theoretical perspective

The Standard Model is a theory of ﬁelds with spins 0, 1 and 1. The fermions (matter ﬁelds) can be 2 arranged in a big vector containing left-handed spinors only: ΨT = qL1 , uC , eC , dC , lL1 , (nC ), qL2 , . . ., . . . , (nC ) , L R1 R1 R1 R1 R3

1st family 2nd 3rd

(1)

where the ﬁelds are the quarks and leptons, all in threefold family replication. The quarks come in triplets of colour, i.e., they carry an index α, α = 1, 2, 3, which we suppressed in the above expression. The left-handed quarks and leptons come in doublets of weak isospin,

α qLi =

uαi L dαi L

and

lLi =

νLi eL i

,

where i is the family index i = 1, 2, 3. We have included a right-handed neutrino n R because there is evidence for neutrino masses from neutrino oscillation experiments.

1

W. B UCHMÜLLER AND C. L ÜDELING

The subscripts L and R denote left- and right-handed ﬁelds, respectively, which are eigenstates of the chiral projection operators P L or PR . The superscript C indicates the charge conjugate ﬁeld (the antiparticle). Note that the charge conjugate of a right-handed ﬁeld is left-handed: 1 − γ5 ψL = ψ L , 2 1 + γ5 ψR = ψ R , PR ψR ≡ 2 PL ψL ≡

C C P L ψR = ψ R , C C P R ψL = ψ L , C P L ψR = P L ψL = 0 , C P R ψL = P R ψR = 0 .

(2) (3)

So all ﬁelds in the big column vector of fermions have been chosen left-handed. Altogether there are 48 chiral fermions. The fact that left- and right-handed fermions carry different weak isospin makes the Standard Model a chiral gauge theory. The threefold replication of quark-lepton families is one of the puzzles whose explanation requires physics beyond the Standard Model [4]. The spin-1 particles are the gauge bosons associated with the fundamental interactions in the Standard Model, GA , A = 1, . . . , 8 : µ

I Wµ

the gluons of the strong interactions , the W and B bosons of the electroweak interactions.

(4) (5)

, I = 1, 2, 3 , Bµ :

These forces are gauge interactions, associated with the symmetry group GSM = SU(3)C × SU(2)W × U(1)Y , where the subscripts C, W , and Y denote colour, weak isospin and hypercharge, respectively. The gauge group acts on the fermions via the covariant derivative D µ , which is an ordinary partial derivative, plus a big matrix Aµ built out of the gauge bosons and the generators of the gauge group: Dµ ΨL = (∂µ + gAµ ) ΨL . From the covariant derivative we can also construct the ﬁeld strength tensor, i Fµν = − [Dµ , Dν ] , g which is a matrix-valued object as well. The last ingredient of the Standard Model is the Higgs ﬁeld Φ, the only spin-0 ﬁeld in the theory. It is a complex scalar ﬁeld and a doublet of weak isospin. It couples left- and right-handed fermions together. Written in terms of these ﬁelds, the Lagrangian of the theory is rather simple: 1 L = − tr [Fµν F µν ] + ΨL iγ µ Dµ ΨL + tr (Dµ Φ)† D µ Φ 2 2 1 T 1 Ψ ChΦΨL + h.c. . + µ 2 Φ† Φ − λ Φ† Φ + 2 2 L (8)

(6)

(7)

(9)

The matrix C in the last term is the charge conjugation matrix acting on the spinors, h is a matrix of Yukawa couplings. All coupling constants are dimensionless, in particular, there is no mass term for any quark, lepton or vector boson. All masses are generated via the Higgs mechanism which gives a vacuum expectation value to the Higgs ﬁeld, Φ ≡ v = 174 GeV . (10)

The Higgs boson associated with the Higgs mechanism has not yet been found, but its discovery is generally expected at the LHC. 2

2

F IELD THEORY AND THE S TANDARD M ODEL

1.2

Phenomenological aspects

The Standard Model Lagrangian (9) has a rich structure which has led to different areas of research in particle physics: – The gauge group is composed of three subgroups with different properties: – The SU(3) part leads to quantum chromodynamics, the theory of strong interactions [5]. Here the most important phenomena are asymptotic freedom and conﬁnement: The quarks and gluons appear as free particles only at very short distances, probed in deep-inelastic scattering, but are conﬁned into mesons and baryons at large distances. – The SU(2) × U(1) subgroup describes the electroweak sector of the Standard Model. It gets broken down to the U(1)em subgroup of quantum electrodynamics by the Higgs mechanism, leading to massive W and Z bosons which are responsible for charged and neutral-current weak interactions, respectively. – The Yukawa interaction term can be split into different pieces for quarks and leptons: 1 T ¯ Ψ ChΦΨL = hu ij uRi qLj Φ + hd ij dRi qLj Φ + he ij eRi lLj Φ + hn ij nRi lLj Φ , ¯ ¯ ¯ 2 L (11)

where i, j = 1, 2, 3 label the families and Φa = ab Φ∗ . When the Higgs ﬁeld develops a vacuum b expectation value Φ = v, the Yukawa interactions generate mass terms. The ﬁrst two terms, mass terms for up-type- and down-type-quarks, respectively, cannot be diagonalized simultaneously, and this misalignment leads to the CKM matrix and ﬂavour physics [6]. Similarly, the last two terms give rise to lepton masses and neutrino mixings [7]. 2

Quantization of ﬁelds

In this section we cover some basics of quantum ﬁeld theory (QFT). For a more in-depth treatment, there are many excellent books on QFT and its application in particle physics, such as Refs. [2, 3]. 2.1 2.1.1 Why ﬁelds? Quantization in quantum mechanics

q(t) ˙ Quantum mechanics is obtained from classical mechanics by a method called quantization. Consider, for example, a particle moving in one dimension along a trajectory q(t), with velocity q(t) (see Fig. 1). Its mo˙ tion can be calculated in the Lagrangian or the Hamiltonian approach. The Lagrange function L(q, q) is a function of the position and the ve˙ locity of the particle, usually just the kinetic minus the potential energy. The equation of motion is obtained by requiring that the action, the time q(t) integral of the Lagrange function, be extremal, or, in other words, that its Fig. 1: Particle moving in variation under arbitrary perturbations around the trajectory vanishes: one dimension δS = δ dtL (q(t), q(t)) = 0 . ˙ (12)

The Hamiltonian of the system, which corresponds to the total energy, depends on the coordinate q and its conjugate momentum p rather than q: ˙ H(p, q) = pq − L (q, q) , ˙ ˙ p= ∂L . ∂q ˙ (13)

3

3

W. B UCHMÜLLER AND C. L ÜDELING

To quantize the system, one replaces the coordinate and the momentum by operators q and p acting on some Hilbert space of states that we shall specify later. In the Heisenberg picture, the states are time-independent and the operators change with time as q(t) = eiHt q(0)e−iHt . (14)

Since p and q are now operators, they need not commute, and one postulates the commutation relation [p(0), q(0)] = −i , where h = 2π is Planck’s constant. In the following we shall use units where commutator (15) leads to the uncertainty relation ∆q · ∆p ≥ 1 . 2 (15) = c = 1. The

(16)

Note that on Schrödinger wave functions the operator q is just the coordinate itself and p is −i∂/∂q. In this way the commutation relation (15) is satisﬁed. As an example example of a quantum mechanical system, consider the harmonic oscillator with the Hamiltonian H= 1 2 p + ω2 q2 , 2 (17)

which corresponds to a particle (with mass 1) moving in a quadratic potential with a strength characterized by ω 2 . Classically, H is simply the sum of kinetic and potential energy. In the quantum system, we can deﬁne new operators as linear combinations of p and q: 1 q=√ a + a† , 2ω ω 1 a= q+i p, 2 2ω p = −i a† = ω a − a† 2 ω q−i 2 1 p. 2ω , (18a) (18b)

i.e. ,

a and a† satisfy the commutation relations a, a† = 1 . In terms of a and a† the Hamiltonian is given by H= ω aa† + a† a 2 . (20) (19)

Since Eqs. (18) are linear transformations, the new operators a and a † enjoy the same time evolution as q and p: a(t) = eiHt a(0)e−iHt = a(0)e−iωt , where the last equality follows from the commutator of a with the Hamiltonian, [H, a] = −ωa , H, a† = ωa† . (22) (21)

We can now construct the Hilbert space of states that the operators act on. We ﬁrst notice that the commutators (22) imply that a and a† decrease and increase the energy of a state, respectively. To see this, suppose we have a state|E with ﬁxed energy, H|E = E|E . Then Ha|E = (aH + [H, a])|E = aE|E − ωa|E = (E − ω) a|E . 4

4

(23)

F IELD THEORY AND THE S TANDARD M ODEL

i.e., the energy of the state a|E is (E −ω). In the same way one ﬁnds Ha † |E = (E + ω)|E . From the form of H we can also see that its eigenvalues must be positive. This suggests constructing the space of states starting from a lowest-energy state|0 , the vacuum or no-particle state. This state needs to satisfy a|0 = 0 , (24)

**so its energy is ω/2. States with more ‘particles’, i.e., higher excitations, are obtained by successive application of a† : |n = a† 2.1.2
**

n

|0 ,

with

H|n =

n+

1 2

ω|n .

(25)

**Special relativity requires antiparticles
**

A2 +e− →B2 ∆Q=−1 (t2 ,x2 )

So far, we have considered non-relativistic quantum mechanics. A theory of elementary particles, however, has to incorporate special relativity. It is very remarkable that quantum mechanics together with special relativity implies the existence of antiparticles. To see this (following an argument in Ref. [8]), consider two systems (e.g., atoms) A 1 and A2 at positions x1 and x2 . Assume that at time t1 atom A1 emits an electron and turns into B1 . So the charge of B1 is one unit higher than that of A1 . At a later time t2 the electron is absorbed by atom A2 which turns into B2 with charge lower by one unit. This is illustrated in Fig. 2.

(t1 ,x1 ) A1 →B1 +e− ∆Q=1

Fig. 2: Electron moving from A1 to A2

According to special relativity, we can also watch the system from a frame moving with relative velocity v. One might now worry whether the process is still causal, i.e., whether the emission still precedes the absorption. In the boosted frame (with primed coordinates), one has t2 − t1 = γ (t2 − t1 ) + γv (x2 − x1 ) , 1 . γ=√ 1−v2 (26)

Here t2 − t1 must be positive for the process to remain causal. Since |v| < 1, t 2 − t1 can only be negative for spacelike distances, i.e., (t2 − t1 )2 − (x1 − x2 )2 < 0. This, however, would mean that the electron travelled faster than the speed of light, which is not possible according to special relativity. Hence, within classical physics, causality is not violated. This is where quantum mechanics comes in. The uncertainty relation leads to a ‘fuzzy’ light cone, which gives a non-negligible propagation probability for the electron even for slightly spacelike distances, as long as (t2 − t1 )2 − (x1 − x2 )2 Does this mean causality is violated?

2 A1 +e+ →B1 ∆Q=1 (t1 ,x1 ) (t2 ,x2 ) A2 →B2 +e+ ∆Q=−1

−

m2

.

(27)

Fortunately, there is a way out: The antiparticle. In Fig. 3: Positron moving from A2 to A1 the moving frame, one can consider the whole process as the emission of a positron at t = t2 , followed by its absorption at a later time t = t1 (see Fig. 3). So we see that quantum mechanics together with special relativity requires the existence of antiparticles for consistency. In addition, particle and antiparticle need to have the same mass.

5

5

b† (k1 )b† (k2 )|0 . a† (k ) = b(k). and construct the space of states just as we did for the harmonic oscillator in the previous section. the uncertainty relation (16) also implies that particles cannot be localized below their Compton wavelength ∆x ≥ mc . The expression on the right-hand side is the Lorentz-invariant way to say that only operators with the same momentum do not commute [the (2π) 3 is just convention]. and second. a† ) and antiparticles (b. we also have to postulate the commutation relations of these operators. For each oscillator. These operators will then be combined into a ﬁeld operator. We want to construct the quantum ﬁelds for particles of mass m. (28)
For shorter distances the momentum uncertainty ∆p > mc allows for contributions from multiparticle states. so we can combine each momentum k with the associated energy ωk = k 0 = k 2 + m2 to form the momentum 4-vector k. Each creation operator a † (k) (b† (k)) creates a (anti)particle with momentum k.
6
6
. corresponding to different momenta. L ÜDELING
In a relativistic theory. The four components of k are not independent. B UCHMÜLLER AND C. so the space of states is vacuum: |0 . Furthermore. that is k 0 = ωk > 0. For each k we deﬁne creation and annihilation operators. there are only two non-vanishing commutators (‘canonical commutation relations’): a(k). a† (k1 )b† (k2 )|0 .1 States. and we also require positive energy. A ﬁeld here is a collection of inﬁnitely many harmonic oscillators. which is annihilated by both particle and antiparticle annihilation operators. (29)
which are the counterparts of relation (19). 2. both for particles (a. creation and annihilation
The starting point is a continuous set of harmonic oscillators which are labelled by the spatial momentum k. particle and antiparticle operators should commute with each other. the quantum analogue of the classical ﬁeld. particle number is not well deﬁned. Like in the harmonic oscillator case. First.2 Multiparticle states and ﬁelds
In the previous section we saw that the combination of quantum mechanics and special relativity has important consequences.2.W. This 4-vector satisﬁes k 2 ≡ k µ kµ = m2 . we can construct operators and states just as before in the quantum mechanical case. b† ). . b† (k)|0 two-particle states: a† (k1 )a† (k2 )|0 . a(k)|0 = b(k)|0 = 0
one-particle states: a† (k)|0 . but satisfy k 2 ≡ kµ k µ = m2 . and we choose them in a similar way: operators with different momenta correspond to different harmonic oscillators and hence they commute. Since we now have a continuous label for the creation and annihilation operators. Hence. we need antiparticles. For the states we again postulate the vacuum state. b† (k ) = (2π)3 2ωk δ 3 k − k . . These results can be obtained by applying the method of canonical quantization to ﬁelds. we need a Lorentz-invariant way to sum over operators with different momentum. and one can no longer talk about a single particle. These properties can be conveniently described by means of ﬁelds. 2.

of spatial momentum and energy. Its construction is obvious.. the ﬁeld is real. which is the 0-component of P µ . the net charge of a state is proportional to the number of particles minus the number of antiparticles: Q= and one easily veriﬁes Q. a† (k) = k µ a† (k) . dt (35)
So far. (31)
This gives the correct commutation relations: P µ . b† (k) = −b† (k) . 2. this construction applied to the case of a complex ﬁeld..F IELD THEORY AND THE S TANDARD M ODEL
Taking these things into account.2. one is led to the integration measure dk ≡ = = = d4 k 2 2 Θ k0 4 2π δ k − m (2π) d4 k δ k 0 − ωk k 0 + ωk Θ k 0 (2π)3 d4 k 1 δ k 0 − ωk + δ k 0 + ωk (2π)3 2ωk d3 k 1 . P µ . This obviously commutes with the momentum operator. P µ . (2π)3 2ωk
(30) Θ k
0
The numerical factors are chosen such that they match those in Eq. i. (32a) (32b)
Another important operator is the charge. Q. a† (k) = a† (k) . H] = 0 . That means we just have to count the number of particles with each momentum and sum the contributions: Pµ = dk k µ a† (k)a(k) + b† (k)b(k) . but also with the charge: i d Q = [Q. a(k) = −k µ a(k) . one has a = b and Q = 0.2 Charge and momentum
Now we have the necessary tools to construct operators which express some properties of ﬁelds and states. (33)
We have now conﬁrmed our intuition that a † (k) b† (k) creates a particle with 4-momentum k and charge +1 (–1). The ﬁrst one is the operator of 4-momentum. (29) for the commutator of a(k) and a† (k). P µ . since we interpret a † (k) as a creation operator for a state with 4-momentum k. i. b(k) = −k µ b(k) . For the special case of neutral particles.e. Both momentum and charge are conserved: The time derivative of an operator is equal to the commutator of the operator with the Hamiltonian.e. b† (k) = k µ b† (k) . Since particles and antiparticles have opposite charge. (34) dk a† (k)a(k) − b† (k)b(k) .
7
7
.

Q=+1
Fig. x)|0 = −φ(t.
The commutator with the charge operator is [Q.
(42)
is the d’Alambert operator. The ﬁeld operator obeys the (free) ﬁeld equation.2. The vacuum expectation value of this expression is the Feynman propagator: i∆F (x2 − x1 ) = 0 T φ(x2 )φ† (x1 ) 0 =i (45)
d4 k eik(x2 −x1 ) . depending on the temporal order interpret the propagation of charge either as a particle going from x1 to x2 or an antiparticle going the other way. x)|0 . Q=−1 2
Q. x2 ). The ﬁeld operator creates 2 a state with charge ±1 ‘at position (t. φ† = φ† . we particle. x)|0 = φ† (t. This transformation can be derived from the transformation of the a’s: eiyP a† (k)e−iyP = a† (k) + iyµ P µ .
(41)
dk −k 2 + m2
e−ikx a(k) + eikx b† (k) = 0 . L ÜDELING
2. x1 ) 1 and xµ = (t2 . see Fig.4 = ∂ 2 /∂t2 − Propagator
t2 >t1 .3
Field operator
We are now ready to introduce ﬁeld operators. this is expressed as the time-ordered product [using the Θ-function. + m2 φ(x) = where 2. (2π)4 k 2 − m2 + iε
(46)
8
8
. a† (k) + O y 2 = (1 + iyk + · · · ) a† (k) =e
iyk †
(37)
(38) (39) (40)
a (k) . We consider the causal propagation of a charged particle between x µ = (t1 . 4.W. (43) (44)
(t1 . Formally. (36)
A spacetime translation is generated by the 4-momentum in the following way: eiyP φ(x)e−iyP = φ(x + y) . B UCHMÜLLER AND C.
Now we can tackle the problem of causal propagation that led us to introduce antiparticles. Θ(τ ) = 1 for τ > 0 and Θ(τ ) = 0 for τ < 0]: T φ(x2 )φ† (x1 ) = Θ(t2 − t1 )φ(x2 )φ† (x1 ) + Θ(t1 − t2 )φ† (x1 )φ(x2 ) . x)’. Q φ(t. which can be thought of as the Fourier transform of creation and annihilation operators: φ(x) = dk e−ikx a(k) + eikx b† (k) . φ(x)] = −φ(x) .x2 ) ∆Q=−1
t1 >t2 .x1 ) ∆Q=+1
(t2 . Q φ† (t.2. x)|0 . 4: Propagation of a particle or an antiDepending on the temporal order of x 1 and x2 .

Since we always assume that all ﬁelds vanish at spatial inﬁnity. (15). x). x ) = −iδ 3 x − x 9
9
. i. j µ = iφ† ∂ µ φ = i φ† ∂ µ φ − ∂ µ φ† φ . From Noether’s theorem. + m2 ∆F (x2 − x1 ) = d4 k −p2 + m2 −ip(x2 −x1 ) = −δ 4 (x2 − x1 ) . ω+i
(47)
This Feynman propagator is a Green function for the ﬁeld equation. x). the surface term vanishes.
↔
∂0 Q = 0 . and we again have to construct the conjugate momentum variables and impose commutation relations among them. We now start from classical ﬁeld theory.e. there is a conserved current j µ associated with this symmetry. π(x) = ∂L ˙ = φ† (x) .3 Canonical quantization
All the results from the previous section can be derived in a more rigorous manner by using the method of canonical quantization which provides the step from classical to quantum mechanics. π(t.
(50)
it stays invariant. . ˙ ∂ φ(x) π † (x) = ∂L ˙ = φ. under the transformation of the ﬁeld φ → φ = eiα φ . Let us consider the Lagrange density for a complex scalar ﬁeld φ. φ† (t. ˙ ∂ φ† (x) (53)
At this point. Like the Lagrangian in classical mechanics. where the ﬁeld at point x corresponds to the position q in classical mechanics. The commutation relations we have to impose are analogous to Eq. the free Lagrange density is just the kinetic minus the potential energy density.
(54)
..F IELD THEORY AND THE S TANDARD M ODEL
where we used the Θ-function representation Θ(τ ) = − 1 2πi
∞
dω
−∞
e−iωτ . at different spatial points but at equal times. i. φ(t. 4 p2 − m2 + iε e (2π) (48)
It is causal.
↔
(49)
α = const. So we are left with an integral of a total derivative which we can transform into a surface integral via Gauss’s theorem.e. we again replace the classical ﬁelds by operators which act on some Hilbert space of states and which obey certain commutation relations. 2. The Lagrangian has a U(1)-symmetry. it is given by the derivative of the Lagrangian with respect to the time derivative of the ﬁeld. The only non-vanishing commutators are the ones between ﬁeld and conjugate momentum. x ) = π † (t. Like in classical mechanics. L = ∂ µ φ† ∂ µ φ − m 2 φ† φ . Now we need to construct the ‘momentum’ π(x) conjugate to the ﬁeld φ. ∂µ j µ = 0 . (51)
The space integral of the time component of this current is conserved in time: Q= d3 x iφ† ∂ 0 φ . it propagates particles into the future and antiparticles into the past..
(52)
The time derivative vanishes because we can interchange derivation and integration and then replace ∂0 j 0 by ∂i j i since ∂µ j µ = ∂0 j 0 + ∂i j i = 0.

for other purposes. These four-byfour matrices are labelled by a vector index and act on spinor indices. fermions with spin 1 . Various rules and identities related to γ matrices are collected in Appendix A. 2 To describe fermionic particles. other representations are more convenient. Z0 and the gluons). (55)
From the Lagrangian and the momentum. In the previous section we considered a scalar ﬁeld which describes particles with spin 0. Furthermore. gauge ﬁelds which carry spin 1 (photons. γν } = 2gµν . although it requires the choice of a particular time direction. These are four-component objects (but not vectors!) ψ. the operator ∂ ≡ γ µ ∂µ .) To derive the ﬁeld equation. {γµ . one can choose among different possible representations. ∂ψ (60)
since L does not depend on derivatives of ψ 1 .
Of course one can shift the derivative from ψ to ψ via integration by parts. (The ﬁrst guess ψ † ψ is not Lorentz invariant. just as for a scalar. (56)
Note that canonical quantization yields Lorentz-invariant results. In the Standard Model. which still remains to be discovered. (36) via the (anti)particle creation and annihilation operators. 2. the quarks and leptons. (59)
/ The kinetic term contains only a ﬁrst-order derivative.
1
0
2
0
.
(58)
10
10
. −. The Lagrangian for a free fermion contains. we can also construct the Hamiltonian density. the kinetic term and the mass: / L = ψi∂ψ − mψψ . B UCHMÜLLER AND C. A common representation is the so-called chiral or Weyl representation. However.W.
γi =
0 σi −σ i 0
. one deﬁned as ψ ≡ ψ has to treat ψ and ψ as independent variables. which is constructed from the Pauli matrices:
γ0 =
2
This representation is particularly useful when one considers spinors of given chiralities.
(57)
with the metric gµν = diag(+. The Euler–Lagrange equation for ψ is the familiar Dirac equation: 0= ∂L / = i∂ − m ψ . the Higgs ﬁeld. W ± . there are the matter ﬁelds. rather. These relations are satisﬁed by the ﬁeld operator deﬁned in Eq. Those are described by vector ﬁelds which will be discussed in Section 3. −). There are other bosonic ﬁelds. ∂µ ∂L ∂L − = ∂(∂µ φ) ∂φ + m 2 φ† = 0 .4 Fermions
Fermions make calculations unpleasant. which are deﬁned via a set of γ matrices. −. The adjoint spinor ψ is † γ 0 . L ÜDELING
all other commutators vanish. Its ﬁeld equation can be derived from the Lagrangian. spinor ﬁelds. They fulﬁl the anticommutation relations (the Clifford or Dirac algebra). ˙ ˙ H = π φ + π † φ† − L = π † π + φ† φ + m 2 φ† φ . The numerical form of the γ matrices is not ﬁxed. This slightly modiﬁes the computation. we need to introduce new quantities. but the result is still the same. there is just one fundamental scalar ﬁeld.

x ) = −iδαβ δ 3 x − x . x). we again expand the ﬁeld operator in terms of plane waves. The relations are again postulated at equal times (‘canonical anticommutation relations’): πα (t. (64a) (64b)
πα (t. x). For the Weyl representation they are given in the Appendix. / = p+m
αβ
/ p + m v (i) (p) = 0 .F IELD THEORY AND THE S TANDARD M ODEL
The Lagrangian again has a U(1) symmetry. . Because of the four-component nature of the ﬁeld. (61)
which leads to a conserved current and. (66) (65)
This is an eigenvalue equation for the 4 × 4 matrix p µ γ µ . x). ψβ (t. . we obtain two more solutions.4. . ψ → ψ = eiα ψ . The form of these solutions depends on the representation of the γ matrices. to a conserved charge. 4).2) (p) that can be interpreted as antiparticles. 2).
(69)
These are the ingredients we need to deﬁne creation and annihilation operators in terms of the spinor ﬁeld ψ(x) and its conjugate ψ(x): ψ(x) = dp
i
bi (p)u(i) (p)e−ipx + d† (p)v (i) (p)eipx . πα = ∂L = i ψγ 0 ˙ ∂ ψα
α † = iψα . They are denoted u(1. ψβ (t. j µ = ψγ µ ψ . which has two solutions for p2 = m2 and p0 > 0.
In order to obtain creation and annihilation operators.
(67)
(68)
αβ
(i) u(i) (p)uβ (p) α i
. which is equivalent to considering p 0 < 0. Taking a positive sign in the exponential in Eq. (62)
Canonical quantization of fermions
Quantization proceeds along similar lines as in the scalar case. obey the identities: u(i) (p)u(j) (p) = −v (i) (p)v (j) (p) = 2mδij .
(63)
Instead of imposing commutation relations. πβ (t.1 ∂ µ jµ = 0 . x ) = ψα (t. One ﬁrst deﬁnes the momentum π α conjugate to the ﬁeld ψα (α = 1. v (1. This is a manifestation of the Pauli exclusion principle which can be derived from the spin-statistics theorem. where p is the momentum four-vector of the plane wave: / i∂ − m u(p)e−ipx = 0 . / p − m u(i) (p) = 0 . 2. now a spinor u(p) occurs. The eigenspinors determined from the equations (i = 1. for fermions one has to impose anticommutation relations. (65). correspondingly. ψ → ψ = e−iα ψ . which implies / p − m u(p) = 0 . x ) = 0 . Q= d3 x ψγ 0 ψ .2) (p) and represent positive energy particles. . the multiplication of ψ by a constant phase. however.
i
(i) (i) vα (p)v β (p)
/ = p−m
. i 11
11
(70a)
.

Their conjugates d ± (p) and b± (p) annihilate those particles. i i (76) (77)
An operator we did not encounter in the scalar case is the spin operator Σ . L ÜDELING
ψ(x) = Here.
(72)
The creation and annihilation operators inherit the anticommutator algebra from the ﬁeld operators. (73a) (73b)
bi (p ). i. Creation operators for particles with deﬁnite spin satisfy the commutation relations s · Σ. B UCHMÜLLER AND C. a spatial unit vector s. ± 2 ± (78)
This immediately leads to the construction of the Fock space of fermions: We again start from a vacuum state |0 . and construct particle states by successive application of creation operators: vacuum: |0 . charge +1 (−1) and spin orientation ± 2 ( 2 ) relative to the chosen axis s. (2π)3 2Ep
Ep =
p2 + m2 . however.
12
12
.2 fermions with four-momentum pµ . which is annihilated by the annihilation operators.e. b† (p) = ± b† (p) .. It has three components.
2
In summary. 2 ± 1 s · Σ. d† (p) = ± 1 † d (p) . Σ is constructed as a commutator of γ matrices and as such has six independent components. since P µ and Q are bosonic operators (not changing the number of fermions in a given state): P µ . dk b† (k)b(k) − d† (k)d(k) . i i Q. i i Q. all these commutation relations tell us how to interpret the operators d † (p) (b† (p)): ± ± 1 1 1 They create spin. d† (p) = −d† (p) . i
(70b)
dp = Inverting Eq. But three of these correspond to Lorentz boosts which mix time and spatial directions. measurable. as before. d† (p ) = (2π)3 2Ep δ 3 p − p j j . Pµ = Q= dk k µ b† (k)b(k) + d† (k)d(k) . (70a) one obtains
d3 p 1 . b† (p ) = di (p).
bi (p)|0 = di (p)|0 = 0
Actually.
dp
i
b† (p)u(i) (p)eipx + di (p)v (i) (p)e−ipx . b† (p) = b† (p) .
(71)
bi (p) = and similar equations for the other operators. which involve commutators. dj (p ) = all other anticommutators = 0 .
d3 x u(i) (p)eipx γ 0 ψ(x) . This is speciﬁed by a choice of quantization axis. i i P µ . bi (p).
The momentum and charge operators are again constructed from the creation and annihilation operators by ‘counting’ the number of particles in each state and summing over all states. The operator that measures the spin of a particle is given by the scalar product s · Σ. d† (p) = pµ d† (p) . (74) (75)
These operators have the correct algebraic relations.W. Only one combination of these components is. b† (p) = pµ b† (p) . Σ is the spin operator in the rest frame. corresponding to the three components of an angular momentum vector 2 .

while L I is the interaction term. it is obtained as the time-ordered product of two ﬁeld operators. but then the theory would not have a ground state since the energy would not be bounded from below): L = L 0 + LI 1 1 λ = ∂µ φ∂ µ φ − m2 φ2 − φ4 . 5: n particles with momenta p i interact. resulting in m particles
13
13
.5 Interactions (80)
So far. A theory of elementary particles obviously needs interactions. . By far the most important approximation method is perturbation theory where one treats the interaction as a small effect. 2 2 4! (81)
L0 is the free Lagrangian. the coupling constant. d† (p)|0 i i two-particle states: b† (p1 )d† (p2 )|0 . whose strength is given by the dimensionless coupling constant λ. and one expresses physical quantities as power series in this parameter. The ﬁnal expression we need for the further discussion is the propagator. By the same reasoning as in the scalar case. in particular scattering crosssections for processes like the one in Fig.5. we have considered free particles and their propagation. The Feynman propagator SF for fermions. which is now a matrix-valued object. i i i i (79)
anticommuting
where |X is an arbitrary state. . i j . to the free theory. like the precise prediction of the anomalous magnetic moment of the electron. despite the fact that important conceptual problems still remain to be resolved.F IELD THEORY AND THE S TANDARD M ODEL
one-particle states: b† (p)|0 . a perturbation. containing kinetic and mass term. For a state of two fermions with identical quantum numbers. we have to look for approximations. on account of the choice of anticommutation relations in Eq. 2. they are much more difﬁcult to handle. At this point we can verify that the Pauli principle is indeed satisﬁed. the fermions would have the wrong (i. (64). and little is known rigorously (except in two dimensions). involving only one real scalar ﬁeld with a quartic self-interaction (a cubic term would look even simpler. .. Bose) statistics.e. Hence. (2π)4 p2 − m2 + iε This completes our discussion on the quantization of free scalar and spinor ﬁelds. Unfortunately. The interaction strength is quantiﬁed by a numerical parameter. 2. Had we quantized the theory with commutation relations instead. . is given by iSF (x1 − x2 )αβ = 0 T ψα (x1 )ψ β (x2 ) 0 / p + m αβ d4 p =i e−ip(x1 −x2 ) .1 φ4 theory
Let us consider the simplest example of an interacting theory. This approach has been very successful and has led to many celebrated results. we would get b† (p) b† (p) |X = −b† (p) b† (p)|X = 0 . In perturbation theory we can calculate various physical quantities. .

propagators. . . . . . The Feynman rules read: i. . . . . the Feynman diagrams are all composed out of three building blocks: External lines corresponding to incoming or outgoing particles. pn . pn ) one then obtains the amplitude for the scattering process p1 . . . pm . particles are free at inﬁnite past and future.
p
1
External lines: For each external line. . 6). . L ÜDELING
p1 . . the total momentum is conserved. . The momentum direction is indicated by the arrow. . and 4-vertices. pm pn p1 pn . one ﬁrst draws all possible Feynman diagrams with a given number of vertices and then translates them into an analytic expression using the Feynman rules. . . . . (83)
Detailed techniques have been developed to obtain a perturbative expansion for the S-matrix from the deﬁnition (83). one needs to keep track of the momentum of each particle entering or leaving the interaction. p1 . disconnected pieces involving non-interacting particles have to be subtracted (see Fig. . .
p1 p1 p2 . .
(86)
where the matrix element M contains all the dynamics of the interaction. . However. Because of the translational invariance of the theory. external lines do not contribute to the matrix element in this theory). The basis is Wick’s theorem and the LSZ formalism.. . . pn . . . . From the Fourier transform ˜ τ (x1 . producing m outgoing ones with momenta p 1 . . . . xm . . One starts from a generalization of the propagator. pn . pn . . . . . In other words. pm S p1 . in at . which deﬁnes a unitary matrix. pm . . . (82)
The transition amplitude for the scattering process is determined by the scalar product of incoming and outgoing states. x1 . . The matrix element can be calculated perturbatively up to the desired order in the coupling λ via a set of Feynman rules. . . xn ) −→ τ (p1 . . pm . respectively
Fig. φ(xk )| 0 . pn and p1 . 5: Scattering of n incoming particles. we have free asymptotic states |p1 . 5 decomposes into a smaller blob and straight lines just passing from the left to the right side. pn = (2π)4 δ 4
out F. multiply by 1 (i. pm S p1 . . . . . . . . 14
14
. . . . τ (x1 . . . p1 . pm . (84)
First. . . pm
Fig. . out at t = +∞ . . . in = p1 . . Since the interaction is localized in a region of spacetime. B UCHMÜLLER AND C. . .W. the so-called S-matrix (S for scattering). . xk ) = 0 | T φ(x1 ). 6: A disconnected diagram: One particle does not participate in the interaction
with momenta pj . . For the φ4 theory. t = −∞ and p1 . . .T.e. the time-ordered product of k ﬁelds. To calculate the cross-section for a particular process. . and the blob in Fig. .
(85)
pi −
pi
in
iM . . out p1 . .

That means that the integral depends logarithmically on the upper integration limit. For each undetermined loop momentum p. 7. p2 = m2 need not hold! −iλ Vertices yield a factor of the coupling constant. The relevant diagrams are collected in Fig. The momenta of internal loops are not ﬁxed by the incoming momenta. 4 p2 − m 2 (2π) (p1 + p2 − p)2 − m2
(87)
The factors of 1 are symmetry factors which arise if a diagram is invariant under interchange of internal 2 lines. m i i d4 p −− − − − − − − −→ 4 p2 − m 2 2 (2π) (p + p1 − p3 ) − m2 15
15
Λ
d4 p −1 ∝ ln Λ . Note that particles of internal lines need not be on-shell. so there is only one vertex. 7: Feynman graphs for 2 → 2 scattering in φ 4 theory to second order. the integrand behaves like p−4 . while the second-order diagrams involve an integration: iM = −iλ + + + 1 (−iλ)2 2 1 (−iλ)2 2 1 (−iλ)2 2 i d4 p i 4 p2 − m 2 (2π) (p + p1 − p3 )2 − m2 d4 p i i 4 p2 − m 2 (2π) (p + p1 − p4 )2 − m2 d4 p i i + O λ3 .
d4 p (2π)4
As an example. i. The ﬁrst-order diagram simply contributes a factor of −iλ. This can be seen by counting the powers of the integration variable p. (2π)4 p4
(88)
. For p much larger than incoming momenta and the mass.
p1 p3
p2
p4
(a) Tree graph
p1 p p2
p3 p + p 1 − p3 p4
p1 p p2
p3
p1 p2
p p1 + p 2 − p
p3 p4
p4
(b) One-loop graphs
Fig. one integrates over all values of p.
iv. The expression for M has a serious problem: The integrals do not converge. and the interaction term does not contain derivatives. and it does not depend on the momenta.F IELD THEORY AND THE S TANDARD M ODEL
ii.e.
Λ
p pi . there is only one species of particles..
p
i Propagators between vertices are free propagators corresponding p2 − m2 + iε to the momentum of the particle. In this theory. The one-loop graphs are all 1 invariant under the interchange of the internal lines and hence get a symmetry factor of 2 .
iii. let us calculate the matrix element for the 2 → 2 scattering process to second order in λ.

1
Global symmetries versus gauge symmetries
Consider a complex scalar ﬁeld with the Lagrangian L = ∂ µ φ† ∂ µ − V φ† φ . (91)
16
16
.
p −→ p −→ p −→ p −→ p −→
u(p) u(p) v(p) v(p)
Incoming or outgoing particles get a factor of u(p) or u(p). Usually it is sufﬁcient to consider the variation of the ﬁelds and the Lagrangian under inﬁnitesimal transformations. The lines carry two arrows.
/ i p+m Free propagator for fermion with momentum p.2 Fermions
We can augment the theory by adding a fermionic ﬁeld ψ.
iv. This theory has a U(1) symmetry under which φ → φ = exp{iα} φ with constant parameter α.5. δφ = φ − φ = iαφ .e. The additional rules are: i. and renormalization. respectively. Again.W.
3
Gauge theories
1 In addition to spin-0 and spin. (90)
which is a generalization of the one considered in Eq. respectively. They are the quanta of vector ﬁelds which can describe strong and electroweak interactions.
iii. making the integrals ﬁnite by introducing some cutoff parameter. i. / Lψ = ψ i∂ − m ψ − gψφψ . where this additional parameter is removed in the end. L ÜDELING
Divergent loop diagrams are ubiquitous in quantum ﬁeld theory. The corresponding theories come with a local (“gauge”) symmetry and are called gauge theories.
3.
ii. 2. the standard model contains spin-1 particles.2 particles. which basically distinguishes particles and antiparticles. They can be cured by regularization. one for the momentum as for the scalars and one for the fermion number ﬂow. (49).
Incoming or outgoing antiparticles get a factor of v(p) or v(p).. yielding ﬁnite results for observables. This will be discussed in more detail in the section on quantum corrections. δφ† = −iαφ† . with a Lagrangian including an interaction with the scalar φ.
free Lagrangian interaction
(89)
There are additional Feynman rules for fermions. B UCHMÜLLER AND C. 2 − m2 + iε p −ig The fermion-fermion-scalar vertex yields a factor of the coupling constant. there is no momentum dependence.

The setup is invariant under spatial rotations around the star. does not mean that the solutions to the equations of motion derived from this Lagrangian are invariant under such a transformation. it is independent of the form of δφ. The above procedure can be generalized to more complicated Lagrangians and symmetries. Indeed. and only φ ≡ 0 is invariant. generically they are not. and up to the second line of (92). around the three axes of a Cartesian coordinate system). δL = 0. we compute the variation of the Lagrangian under such a transformation: δL = ∂L ∂L † ∂L ∂L δφ + δφ + δ (∂µ φ) + δ ∂µ φ† ∂φ ∂φ† ∂ (∂µ φ) ∂ (∂µ φ† )
=∂µ δφ
=
∂L ∂L − ∂µ ∂φ ∂ (∂µ φ)
=0 by equation of motion
δφ +
∂L ∂L − ∂µ † ∂φ ∂ (∂µ φ† )
=0
δφ† (92)
+ ∂µ
∂L ∂L δφ + δφ† ∂ (∂µ φ) ∂ (∂µ φ† )
= α∂µ i∂ µ φ† φ − iφ† ∂ µ φ = −α∂µ j µ . φ+ ∂V φ. What is the meaning of such a symmetry? Loosely speaking.g. Eq. a symmetry of the Lagrangian always implies a conserved current. i. φ† = ∂φ φ − φ† µ2 − 2λφ† φ = 0 . This situation is analogous to the Kepler problem: A planet moving around a stationary (very massive) star. This group is three-dimensional (meaning that any rotation can be built from three independent rotations. consider the Mexican hat potential. Thus there are three 17
17
. This means that any constant φ with this modulus is a solution to the equation of motion. Since the Lagrangian is invariant.
(95)
This potential has a ring of minima.e.. As a general result.. The derivation does not depend on the precise form of L . This. but it can be angular momentum or energy as well). it states that “physics does not change” under such a transformation. namely all ﬁelds for which |φ| 2 = µ2 /(2λ). ∂µ j µ = 0 . As an example. V (φ† φ) = −µ2 φ† φ + λ φ† φ
2
(93)
∂L ∂L δφ − ∂µ ∂ (∂µ φ) ∂ (∂µ φ)
δφ
(94)
. To derive the Noether current. it is obvious that any solution to the equations of motion will be mapped into another solution under such a transformation. From the ﬁrst to the second line we have used ∂L ∂µ δφ = ∂µ ∂ (∂µ φ) by the Leibniz rule. however. the symmetries form the group SO(3). (96)
These solutions are not invariant under U(1) phase rotations. (51). which in turn gives a conserved quantity (often referred to as charge. e. we obtain a conserved current for solutions of the equations of motion. On the other hand.F IELD THEORY AND THE S TANDARD M ODEL
where terms O α2 have been neglected.

difference between this and another type of symmetry. the variation is not multiplicative. · E = ρ. ∂t B= × A. most experiments are scattering experiments at colliders. i. ∂t (99)
where Λ is a scalar ﬁeld. Rotated solutions. it is a local transformation.2 Abelian gauge theories
The easiest way to come up with a gauge symmetry is to start from a global symmetry and promote it to a gauge one. The solutions of this problem—the planet’s orbits—are ellipses in a plane. but it does relate identical states which describe exactly the same physics. are again solutions. Potentials that are related to φ = 0 and A = 0 by a gauge transformation are called pure gauge. since different potentials can lead to the same E and B ﬁelds. but important. To see this. L = ∂µ φ† ∂ µ φ − V (φ† φ) . and the gauge transformation (99) quantiﬁes just this redundancy. demand invariance of the Lagrangian under local transformations (where the transformation parameter is a function of spacetime). There is a subtle. together with charge and current densities ρ and j. One formulation uses electric and magnetic ﬁelds E and B. give the same electric and magnetic ﬁelds. that is. If some potentials φ and A lead to certain E and B ﬁelds. the result will be the same as if one had done the experiment with the untransformed state and transformed the result. i.. ×B− ∂t · B = 0. but it is different from the global symmetries we considered before. φ =φ− ∂Λ . A gauge transformation is also a transformation which leaves the Lagrangian invariant. Second. 3. since solutions of the equations of motion for electric and magnetic ﬁelds are invariant. ∂t ∂E =j. It is important to note that this gauge transformation is inhomogeneous. However. we have expressed the physical ﬁelds E and B in terms of the potentials φ and A. but can generate non-vanishing potentials from zero.e.e. These potentials still contain too many degrees of freedom for the physical ﬁelds E and B. the correspondence between the physical ﬁelds and the potentials is not unique. so they are not at all invariant under spatial rotations. it relates physically indistinguishable ﬁeld conﬁgurations. First. This might be familiar from electrodynamics. For those. Physical states and observables have to be invariant under gauge transformations. These ﬁelds and sources are related by Maxwell’s equations: ×E+ ∂B = 0. gauge symmetry. not even under rotations in the plane of motion. the transformed potentials A =A+ Λ. (98)
So we have reduced the six components of E and B down to the four independent ones φ and A. 18
18
. the statement that “physics does not change” translates into “transformed initial states lead to transformed ﬁnal states”: If one applies the transformation to the initial state and performs the experiment. recall the Lagrangian with the global U(1) symmetry from the preceding section. It is a symmetry of the theory. which yield E and B via E =− φ− ∂A . the transformation parameter Λ varies in space and time..W. In particle physics. however. L ÜDELING
conserved charges which correspond to the three components of angular momentum. Phrased differently. This transformation (98) is called gauge transformation. (97a) (97b)
The ﬁrst two of these can be identically solved by introducing the potentials φ and A. So the description in terms of potentials is redundant. B UCHMÜLLER AND C.

(102)
This is called covariant derivative because the differentiated object D µ φ transforms in the same way as the original ﬁeld. ieAν ] + [ieAµ . (∂ν + ieAν )] e e i = − [∂µ . there is no kinetic term for it. the Lagrangian is no longer invariant. 0 −E1 −E2 −E3 E 0 −B3 B2 . The way to restore invariance of the Lagrangian is to add another ﬁeld. ∂ν ] − e2 [Aµ . we ﬁrst construct the ﬁeld strength tensor from the commutator of two covariant derivatives: i i Fµν = − [Dµ . ∂ν ] + [∂ν . The gauge ﬁeld A µ .
If we now allow spacetime-dependent parameters α(x). we can redecompose A µ into the scalar and vector potential φ and A and spell out the ﬁeld strength tensor in electric and magnetic ﬁelds.F IELD THEORY AND THE S TANDARD M ODEL
and the transformation φ → φ = eiα φ . (104)
1 = ∂µ eiα(x) φ + ie Aµ (x) − ∂µ α(x) eiα(x) φ e
(103)
So far this is a theory of a complex scalar with U(1) gauge invariance. e (101) (100)
The factor 1 is included for later convenience. but the kinetic term picks up derivatives of α(x). (106) F µν = 1 E2 B3 0 −B1 E3 −B2 B2 0 19
19
..
(105)
To check that this is a sensible object to construct. however. combined into a four-vector Aµ = (φ. In order to ﬁnd such a kinetic term.e. A): 1 Aµ (x) → Aµ (x) = Aµ (x) − ∂µ α(x) . is not a dynamical ﬁeld. i. δφ = φ − φ = iαφ . the gauge ﬁeld. so the variation of the Lagrangian is δL = i∂µ α ∂ µ φ† φ − φ† ∂ µ φ = −∂µ α j µ . Aν ] e = ∂ µ Aν − ∂ ν Aµ . the derivative of α times the Noether current of the global symmetry derived before. Dν ] = − [(∂µ + ieAµ ) . with a gauge transformation just like the electromagnetic potentials in the previous section. Dµ φ −→ (Dµ φ) = ∂µ + ieAµ φ = eiα(x) Dµ φ . This kinetic term should be gauge invariant and contain derivatives up to second order. So we can construct an invariant Lagrangian from the ﬁeld and its covariant derivative: L = (Dµ φ)† (D µ φ) − V φ† φ . We can now combine the inhomogeneous transformation e of Aµ with the inhomogeneous transformation of the derivative in a covariant derivative D µ : Dµ φ = (∂µ + ieAµ ) φ . The potential part still is.

we have to count: A n × n complex matrix U has n 2 complex entries. This is all we need to construct the Lagrangian for QED.. i. the gauge ﬁeld Aµ . This covariant derivative was the object that coupled the gauge ﬁeld to the other ﬁelds of the theory.e. L = 1 E2 − B2 2 . in terms of E and B ﬁelds. just in a way to make a covariant derivative. L ÜDELING
This shows that the ﬁeld strength is gauge invariant. as are E and B. quarks and leptons indeed do). 1 δFµν = ∂µ δAν − ∂ν δAµ = − (∂µ ∂ν − ∂ν ∂µ ) α(x) = 0 .
(111)
What have we done so far? We started from a Lagrangian. To make this into a dynamical theory. SU(n) is the group of n × n complex unitary matrices with determinant 1. we ﬁrst have to discuss non-Abelian groups. † but not all component equations are independent.3 Non-Abelian gauge theories
To construct non-Abelian theories in the same way as before. this can also be shown by straightforward calculation. For ﬁelds with charge q (in units of elementary charge).
(q) Dµ ψq = (∂µ + iqeAµ ) ψq . We shall focus on the groups SU(n). The desired kinetic term is now just the square of the ﬁeld strength tensor. (108b) (108a) (107)
The coupling to scalar ﬁelds via the covariant derivative can also be applied to fermions. if we call the ﬁelds electron and photon: 1 / LQED = − Fµν F µν + ψ iD − m ψ . B UCHMÜLLER AND C. We imposed invariance under local transformations. e so it is just the antisymmetry in µ and ν that ensures gauge invariance. one again replaces the ordinary derivative with the covariant one. Alternatively. one simply imposes the gauge transformation ψ → ψ = eiα ψ . we added a kinetic term for the gauge ﬁeld.W. however. For QCD and the electroweak theory. U † U is Hermitian. 4 or. 3. 4 (110)
Finally. since they are most relevant for the Standard Model. Actually. This ﬁeld transformed inhomogeneously under gauge transformations. 1 Lgaugekin = − Fµν F µν . groups whose elements do not commute. let us note that for a U(1) gauge theory. The Lagrangian for a fermion coupled to a U(1) gauge ﬁeld is quantum electrodynamics (QED). U † U = U † U . (90) with a global U(1) symmetry (91). different ﬁelds may have different charges under the gauge group (as. we could have started with the gauge ﬁeld and tried to couple it to other ﬁelds. using the ﬁeld strength tensor. we have to replace the gauge transformations and consequently the covariant derivative as follows: ψq → ψq = eiqα ψq . The unitarity constraint. Of course. and we would have been led to the transformation properties (91). we need a richer structure: non-Abelian gauge theories. is a matrix equation. so we had to introduce a new ﬁeld. for example. (109)
In the Lagrangian. To see how many degrees of freedom there are. so 20
20
. equivalent to 2n2 real ones. To couple a fermion ψ to the gauge ﬁeld. Eq. U † U = .

the relevant groups are SU(2) for the electroweak theory and SU(3) for 1 QCD. · · · . SU(2) has three parameters. (Dµ Φ) = [(∂µ + igAµ ) Φ] = ∂µ + igAµ (U Φ) = U ∂µ + U −1 (∂µ U ) + igU −1 Aµ U Φ = U Dµ Φ . The group elements are usually written in terms of these parameters and n 2 − 1 matrices T a . n (called a multiplet of SU(n)). −→ Φ = U Φ . 1 there are n + 2 · 2 n(n − 1) real constraints. Thus. (113)
with the antisymmetric structure constants f abc . 2 To construct a model with a global SU(n) symmetry. but an ncomponent vector Φi . we again need to introduce a covariant derivative consisting of a partial derivative plus a gauge ﬁeld. n . The generators are usually chosen to be the Pauli matrices.
21
21
. (114) Φ = . the generators live in the Lie algebra of the group. Hermitian or not. i. where µ T a are the generators of the group.
U = exp {iαa T a } =
+ iαa T a + O α2 . T a = 2 σ a . which of course also depend on the choice of generators.
(112)
The generators are usually chosen as Hermitian matrices 3 . . T a = 1 λa . as each generator represents an independent transformation in the group. All in all. we immediately encounter the same problem as before: The derivative term is not invariant. Hence. A µ = Aa T a . Finally. This means that we can generalize the Lagrangian (90) in a straightforward way to include a non-Abelian symmetry: L = (∂µ Φ)† (∂ µ Φ) − V Φ† Φ . the vector ﬁeld needs to be matrix-valued.
(115)
If we allow for local transformations U = U (x). det U † U = |det U |2 = 1.
3
∂µ Φ → ∂µ Φ = ∂µ (U Φ) = U ∂µ Φ + (∂µ U ) Φ . In the Standard Model. Φ† = Φ† . The transformation law of Aµ is chosen such that the covariant derivative is covariant. . . Φ† −→ Φ† = Φ† U † . σ b = iεabc σ c . by taking the determinant of the unitarity constraint. however. . the generators of the group. We clearly need one vector ﬁeld per generator. This time. we consider not a single ﬁeld. on which the matrices of SU(n) act by multiplication: Φ1 . This means that any U ∈ SU(n) can be speciﬁed by n 2 −1 real parameters αa . 1 Φn
Now we see why we want unitary matrices U : A product Φ † Φ is invariant under such a transformation. because the derivatives act on the matrix U as well. i = 1. as an exponential and one often considers only inﬁnitesimal parameters. whose commutation relations are σ a . To save the day. restricting to det U = 1 eliminates one more real degree of freedom..
(116)
(117)
!
Actually. The product of group elements translates into commutation relations for the generators. The common generators of SU(3) are the eight Gell-Mann matrices. and so one can choose any basis one likes.F IELD THEORY AND THE S TANDARD M ODEL
the diagonal entries are real and the lower triangle is the complex conjugate of the upper one. T a . T b = if abc T c .e. we have 2n2 − n − 2 · 1 n(n − 1) − 1 = n2 − 1 real degrees of freedom in the elements of 2 SU(n).

e. 8 counts the gluons and q is a three-component (i. Since tr AB = tr BA. (101). the matrix U = exp{iα} = 1 + iα. g or for each component 1 Aa = Aa − ∂µ αa − f abc αb Ac . 22
22
(126)
. unlike in QED. Furthermore. In the second step we have used a normalization convention. µ µ µ g Sometimes it is convenient to write down the covariant derivative in component form:
a (Dµ Φ)i = ∂µ δij + igTij Aa Φj . . . the quarks. B UCHMÜLLER AND C. g In the Abelian case this reduces to the known transformation law. the commutator of covariant derivatives: i a Fµν = − [Dµ . µ µ ν ν (122)
Now we see that the ﬁeld strength is more that just the derivative: There is a quadratic term in the potentials. it is covariant. when we calculate the transformation of the ﬁeld strength. This leads to a self-interaction of gauge ﬁelds. Eq. By choosing the gauge group SU(3) and coupling the gauge ﬁeld to fermions. we can write down the Lagrangian of quantum chromodynamics (QCD): 1 / LQCD = − Ga Ga µν + q iD − m q . (118) becomes 1 Aµ = Aµ − ∂µ α + i [α. (123)
i. µ
(118)
(119)
(120)
(121)
Next we need a kinetic term. Dν ] = ∂µ Aν − ∂ν Aµ + ig [Aµ . where the gluons interact with each other. There is an easy way to produce an invariant quantity out of this: the trace. 2 with compensating changes in the coefﬁcient of the kinetic term. three-colour) quark. the Lagrangian 1 a 1 L = − tr (Fµν F µν ) = − Fµν F a µν 2 4 (124)
is indeed invariant. Aµ ] . like in QCD. aν ] = Fµν T a . and Eq. This is the basic reason for conﬁnement. which again involves the ﬁeld strength. The factor 1 is arbitrary and could be chosen differently. . 2 (125)
and every generator is necessarily traceless. but transforms as Fµν → Fµν = U Fµν U −1 .. we ﬁnd that it is not invariant. For inﬁnitesimal parameters α = αa T a . g a Fµν = ∂µ Aa − ∂ν Aa − gf abc Ab Ac .W.e. . as tr U F 2 U −1 = tr U −1 U F 2 = tr F 2 . 1 tr T a T b = δ ab . L ÜDELING
This requirement ﬁxes the transformation of A µ to be i Aµ = U Aµ U −1 − U ∂µ U −1 . where the photon is not charged. 4 µν where a = 1.

A common (class of) gauge is the covariant gauge which depends on a parameter ξ. In QED. For external states.4
Quantization
So far we have discussed only classical gauge theories. On the more technical side.
k a b
−iδ ab k 2 + iε ieγ µ
The propagator for ghosts is the one of scalar particles. One way out would be to ﬁx the gauge. strange ﬁelds which are scalars but which anticommute and do not show up as physical states but only as internal lines in loop calculations. so it has zero eigenvalues. namely gauge modes. there is just one vertex between photon and electron. k µ µ = k = 0. it annihilates every pure gauge mode. simply demand a certain gauge condition like ·A = 0 (Coulomb gauge) or nµ Aµ = 0 with a ﬁxed 4-vector (axial gauge). the propagator must be deﬁned in a more clever way. It turns out that gauge invariance is not lost but rather traded for a different symmetry. BRST symmetry. The form of the gauge ﬁxing and ghost terms depends on the gauge condition we want to take.
ii.. They are labelled by two polarization vectors ± which are transverse. There are no external ghost states. orthogonal to the µ momentum four-vector and the spatial momentum. (127)
The Green function should be the inverse of the differential operator in brackets. Indeed. If we want to quantize the theory and ﬁnd the Feynman rules for diagrams involving gauge ﬁelds. however.e.e.. the naïve Green function for the free equation of motion does not exist. we add two terms to the Lagrangian. nor count separately ﬁelds which differ only by a gauge transformation. we have to restrict to physical states. which becomes Feynman gauge (Landau/Lorenz gauge) for ξ = 1 (ξ = 0). since those are meant to be physically identical. µ
a
−iδ ab k 2 + iε × gµν + (1 − ξ)
kµ kν k2
The propagator for gauge bosons contains the parameter ξ.
iii. The fact that it is not gauge invariant means that now the propagator is well deﬁned. the gauge-ﬁxing term and the ghost term. of which there are two for massless bosons. ( g µν − ∂ ν ∂ µ ) ∂µ Λ = 0 .2. as it should. i. We now list the Feynman rules for a non-Abelian gauge theory (QCD) coupled to fermions (quarks) and ghosts. i. that the loss of Lorentz invariance causes many problems in calculations. In this approach. The fermionic external states and propagators are listed in Section 2. It turns out. This is compensated by the propagation of ghosts.
k −→ µ k −→ µ p ν b
µ (k) ∗ (k) µ
(128)
For each external line one has a polarization vector.
µ
. Hence. which ensures that we get physically sensible results. we have a problem: we have to make sure we do not count ﬁeld conﬁgurations of Aµ which are pure gauge. but the operator is not invertible. The gauge-ﬁxing term is not gauge invariant. but the price to pay is that it propagates too many degrees of freedom. 23
23
iv. A better way makes use of Faddeev–Popov ghosts. the equation is ∂µ F µν = Aν − ∂ ν ∂µ Aµ = ( g µν − ∂ ν ∂ µ ) Aµ = 0 . i. but rather represents a certain gauge condition which can be chosen freely. In the Abelian case.F IELD THEORY AND THE S TANDARD M ODEL
3.5.

(133)
. To lowest order. one of the most important tests of quantum ﬁeld theory. one encounters divergent loop integrals.
µ
i g γ µ λa 2
In QCD. the basic quark–quark–gluon vertex involves the Gell-Mann matrices. e/(2m). its magnitude would be the Bohr magneton. and still is today. the magnetic moment is aligned in the direction of s. the magnetic moment is different. 24
24
γi =
0 σi −σ i 0
. In the following sections we shall study these divergences and show how to remove them by renormalization. / iD − m ψ = [γ µ (i∂µ − eAµ ) − m] ψ = 0 .W. A).
viii. we are ready to calculate quantum corrections [3. this just means solving the Dirac equation in an external electromagnetic ﬁeld A µ = (φ. which is expressed by the Landé factor g e . B UCHMÜLLER AND C. In most cases. however.
− 1 g 2 f abc f ade g µν g ρσ 4 +permutations
Four-gluon self-interaction. 2m (130)
We now want to calculate ge in QED. Finally.
4
Quantum corrections
Now that we have the Feynman rules. 5. As a ﬁrst example we consider the anomalous magnetic moment of the electron at one-loop order.
For a particle with spin s. The calculation is still quite simple because the one-loop expression is ﬁnite.
p −→ b
µ. and for a classical spinning particle of mass m and charge e. µe = ge e s. as an application. This was historically. Hmag = −µ · B . 4. we shall discuss the running of coupling constants and asymptotic freedom. In the quantum theory. (132) (131)
It is convenient to use the following representation of the Dirac matrices:
γ0 =
0 0 −
. For a bound non-relativistic electron a stationary solution takes the form ψ(x) = ϕ(x) −iEt e .
vi.1 Anomalous magnetic moment (129)
The magnetic moment of the electron determines its energy in a magnetic ﬁeld.
gf abc kµ + permutations
Three-gluon self-interaction. L ÜDELING
v. a c
−gf abc pµ
vii. χ(x) with E−m m 1. 9]. The ghosts couple to the gauge ﬁeld.

the Landé factor of the electron is g e = 2. such as an electron–positron bubble on the incoming photon.
(134a) (134b)
The coefﬁcient of χ in the second equation is approximately independent of φ. it is contained in the blob on the left side of Fig. All other one-loop diagrams concern only external legs. 1 −i 2m − eA
2
+ eφ −
e B · σ ϕ = (E − m) ϕ . Diagrammatically.F IELD THEORY AND THE S TANDARD M ODEL
q
µ =
q
µ +
q
µ +···
p
p
p
p
p
k
p
Fig. and will be removed by wave-function renormalization. − eA · σϕ = 0 .
(139)
. (135)
Inserting this into (134a). we get the Pauli equation. 2m (137)
Note that the photon becomes on-shell only for q → 0. (138) e sB . so no polarization vector is included. 8: Tree-level and one-loop diagram for the magnetic moment
One then obtains the two coupled equations [(E − eφ) − m] ϕ − −i − (E − eφ) − m χ + −i
≈−2m
− eA · σχ = 0 . where an internal photon connects the two electron lines. χ= 1 −i m − eA · σϕ . There are several one-loop corrections to this diagram. which denotes the complete electron– photon coupling. In QED. The magnetic moment is the spin-dependent coupling of the electron to a photon in the limit of vanishing photon momentum. u(p )γ u(p) = u(p )
µ
i µν (p + p )µ + σ p −p 2m 2m 25
25
ν
u(p) . 2m
(136)
This is a Schrödinger-like equation which implies (since s = 1 σ). 8. the magnetic moment is modiﬁed by quantum corrections. gives a contribution to the magnetic moment. The tree-level diagram is the fundamental electron–photon coupling. The matrix element of the electromagnetic current can be decomposed via the Gordon identity into convection and spin currents. but only the so-called vertex correction. so we can solve the equation to determine χ in terms of ϕ. The expression for the tree-level diagram is iu(p )eγ µ u(p) . 2 Hmag = −2 Hence.

W. the part ∝ σ µν qµ is ﬁnite and can be extracted via some tricks: – Consider ﬁrst the denominator of the integral (141). Using the Feynman rules. It is the product of three terms of the form (momentum)2 − m2 . L ÜDELING
Here the ﬁrst term can be viewed as the net ﬂow of charged particles. we ﬁrst note that the corresponding expression will contain the same external states. as can be seen by power counting. B UCHMÜLLER AND C. – For the numerator. 2 32π (x1 + x2 )2 m2
(145)
Now all that is left are the parameter integrals over x 1 and x2 . ieΓµ (p. We can now decompose Γµ into different parts according to index structure and extract the term ∝ σ µν . since it gives the spin-dependent coupling of the electron. (144)
ν
Here we have used again the Gordon formula to trade (p + p )ν for σνρ q ρ . where A1 x1 + A2 x2 + A3 (1 − x1 − x2 ) = k − x1 p − x2 p
k 2
1 . q = p − p. 26
26
.
(143)
Note that one must be careful when manipulating divergent integrals. 1 =2 A1 A2 A3
1 0
dx1
0
1−x1
dx2
– After introducing the Feynman parameters. q) is the correction to the vertex due to the exchange of the photon. (140)
where Γµ (p. linear and quadratic in k. q) = (−ie)3 / / d4 k −igρσ ρ i p − k + m γµ 4 k 2 + iε γ 2 2 + iε (2π) (p − k) − m / / i p−k+m × γσ . the second one is the spin current. so it can be written as iu(p )eΓµ (p. In order to isolate the magnetic moment from the loop diagram. which can be transformed into a sum at the expense of further integrations over the so-called Feynman parameters x 1 and x2 . but in this case. A standard calculation gives (see Appendix) / / / / γ ν p − k + m γ µ p − k + m γν = −2m2 γµ − 4imσ µν p − p / / − 2pγµ p + O(k) + O k 2 . there is no problem. On the other hand. since the leading term is ∝ k 2 in the numerator and ∝ k 6 in the denominator. the next trick is to shift the integration momentum k → k . Only this one is relevant for the magnetic moment. The quadratic piece leads to a divergent contribution which we shall discuss later. one should ﬁrst regularize them and then perform the shifts on the regularized integrals. 2 2 + iε (p − k) − m
(141)
This integral is logarithmically divergent. The integral over the k-independent part in the limit q µ → 0 yields d4 k 1 4 (2π) k 2 − (x + x )2 m2 + iε 1 2
3
=−
i 1 . In principle. q)u(p) . which is allowed only if the expression is sandwiched between on-shell spinors u(p ) and u(p). The linear term can be dropped under the integral. [A1 x1 + A2 x2 + A3 (1 − x1 − x2 )]3
(142)
− x1 p + x 2 p
2
+ iε . the important part is the Dirac algebra of γ matrices. – Now the numerator is split into pieces independent of k. we ﬁnd for Γµ in Feynman gauge (ξ = 1).

factor.5) · 10−12 .
Comparison with the Gordon decomposition (139) gives the one-loop correction to the Landé g =2 1+
α . The agreement of theory and experiment is impressive: aexp = (1159652185. so a good regularization scheme should preserve as many symmetries as possible. (148) 2 Today. the regulator is removed.F IELD THEORY AND THE S TANDARD M ODEL
Finally. we have to introduce a new parameter of mass dimension 1. One such example is the vertex function Γµ we already considered. Of course. however. The key idea is that the ‘bare’ parameters which appear in the Lagrangian are not physical. ae is known up to three loops analytically and to four loops numerically [10]. We shall restrict ourselves to dimensional regularization. usually expressed in terms of the ﬁne structure constant α = e2 / (4π). It is often expressed as the anomalous magnetic moment ae . An obvious choice would be a cutoff Λ which serves as an upper bound for the momentum integration. Since all divergences have been absorbed into the parameters of the theory. ieu(p )Γµ u(p) = +ieu(p ) α i µν σ qν + · · · 2π 2m u(p) . In this step. but at least the result was ﬁnite. which is the most common scheme used nowadays. so they can be divergent. one obtains the result. 4. one has to make sure the results do not depend on the regularization method. – Finally. the results remain ﬁnite for inﬁnite regulator. although a regularization parameter has generally no direct physical meaning. Symmetries. Most other expressions. We can isolate these by setting q = 0.8) · 10−12 . e (149)
This is one of the cornerstones of our conﬁdence in quantum ﬁeld theory.9 ± 8. while conceptually simple. which yields Γµ (p. as it breaks Lorentz and gauge invariance. have divergent momentum integrals. are very important for all calculations. (147) 2π This correction was ﬁrst calculated by Schwinger in 1948.2 Divergences
The anomalous magnetic moment we calculated in the last section was tedious work. however. Their divergences are chosen such as to cancel the divergences coming from the divergent integrals. It has contributions which are logarithmically divergent. 0) = −2ie2
1 0
dx1
0
1−x1
dx2
/ / γ ν kγµ kγν d4 k 2 (2π) k 2 − (x + x )2 m2 + iε 1 2
3
. – The second step is renormalization. is not a convenient method. One might even argue that there should be a cutoff at a scale where quantum gravity becomes important. e ath = (1159652175.
(150)
This expression is treated in two steps: – First we make the integral ﬁnite in a step called regularization. The divergences are absorbed into the parameters of the theory.9 ± 3. 27
27
. (146)
where the dots represent contributions which are not ∝ σ µν qν . The cutoff regularization. g−2 ae = .

pi = d gµν (2π) d4 k 2 k f k 2 . Derivatives have mass dimension 1. Let us consider some technical issues. scalars and fermions reads L = (∂µ Aν )2 + e∂µ Aµ Aν Aν + e2 (Aµ Aµ )2 / / + (∂µ φ)2 + ψ i∂ − m ψ + eψ Aψ + m2 φ2 + · · · . 2 (156) (157)
How do we evaluate a d-dimensional integral? One ﬁrst transforms to Euclidean space replacing k 0 by ik4 . translates into [L ] = d. This procedure is well deﬁned and just an intermediate step in the calculation.2. so [dxµ ] = −1 and [∂µ ] = 1. L ÜDELING
4.1
Dimensional regularization
The key idea is to deﬁne the theory not in four. (153)
The tensor structure of diagrams can be simpliﬁed as follows. we need to compensate the change in dimensionality by a factor of µ .. 2 d−1 3 [ψ] = → . the Lorentz indices ‘range from 0 to d’. and so do masses. the integrals do converge.
tr ( ) = 4 . so everything can be expressed in powers of GeV. That implies for the dimensions of the ﬁelds (and the limit as d → 4). and the coefﬁcient can be obtained by contracting with g µν to yield 1 d4 k 2 2 4 kµ kν f k . but in d = 4 − dimensions [9].e. the only dimension is mass. a Lagrangian of gauge ﬁelds. In d dimensions.
4
28
28
. γ µ γ ν γ ρ γµ = 4g νρ − γ ν γ ρ . The mass dimensions of ﬁelds and parameters also change. They can be derived from the condition that the action. Schematically (i. since dd x = −d.
(151)
{γ µ . this must be proportional to gµν k 2 . In Euclidean space. If is not an integer. where µ is an arbitrary parameter of mass dimension 1. without all numerical factors). B UCHMÜLLER AND C. which is the d-dimensional integral over the Lagrangian. one can easily convert
In our units where = c = 1. 2 2 [φ] = d−2 →1. 2 d [e] = 2 − → 0 . be dimensionless. but in the end we shall take the limit of → 0 and return to four dimensions. p2 . such as γ µ γ ν γµ = − (2 − ) γ ν . [S] = 0.
(152)
The γ-matrix contractions are also modiﬁed due to the change in the trace of g µν . The only symmetric tensor we have is the metric (as long as the remaining integrand depends only on the square of k and the squares of the external momenta p i ). i (2π)4 (154)
The measure of an integral changes from d 4 k to dd k. and there are d γ-matrices obeying the usual algebra. The basic quantities have [mass] = [energy] = [momentum] = 1 and [length] = [time] = −1. Since k is a dimensionful quantity 4 (of mass dimension 1). Non-integer dimensionality might seem weird. γ ν } = 2g µν .W. [Aµ ] = d−2 →1. so that the Lorentzian measure dd k becomes dd kE . (155)
The condition of dimensionless action. since it is of second order in k and symmetric in (µν). If a momentum integral over k contains a factor of kµ kν . in the sense that g µν gνµ = d .

consider the logarithmically divergent integral [cf. the situation is better than before: We have separated the divergent part from the ﬁnite one and can take care of the divergence before taking the limit → 0. (159)
As an example. The Γ function has poles at negative integers and at zero. but as we let d → 4. (150)] 1 d4 k . or equivalently.
(163)
. 2π 2π 29
29
What have we achieved? In four dimensions. the result is still divergent.
(158)
The remaining integral can then be evaluated. the original divergence appears again in the form of Γ(2 − d/2). → 0. again often using Γ functions. In the limit d → 4. 9: One-loop corrections to the propagators of electron and photon
to spherical coordinates and perform the integral over the angular variables. 4 2 + C)2 (2π) (k where C = (x1 + x2 )2 m2 . The result is ﬁnite for d = 4. this becomes µ 1 d4 kE 4 2 (2π) kE + C
2
(160)
=
(4π)
µ Γ 2−
d/2
d 2
1 C 2−d/2
Γ(2)
=
1 1 + ··· 8π 2
(161)
For the original expression (150) we thus obtain Γµ (p. 9(b) (quadratically divergent).58. In d Euclidean dimensions. Σ(p) = α 1 3α 1 / m− p − m + O(1) . 9(a) (linearly divergent) and the photon self-energy or vacuum polarization Π µν in Fig. However. so the integral exists for non-integer dimension. d 2 =Γ 2 = 2 − γE + O ( ) . dd kE (2π) f k2 = d dΩd (2π)
d 0 1 1 2d−1 π d/2 Γ(d/2) ∞ d−1 dkE kE f k 2 . 0) = α 1 µ γ + O(1) . This is done in the procedure of renormalization. which gives the ‘area’ of the d-dimensional ‘unit sphere’.F IELD THEORY AND THE S TANDARD M ODEL
k p p−k p = −iΣ (p) µ
p+q q p
(b) The vacuum polarization
q
ν = −iΠµν (q)
(a) The electron self-energy
Fig. one has Γ 2− with the Euler constant γE 0. The self-energy graph has two divergent terms. 2π (162)
There are more divergent one-loop graphs where we can achieve the same: the electron self-energy Σ in Fig.

0 1 L = − (∂µ A0 ν − ∂ν A0 µ ) ∂ µ A0 ν − ∂ ν A0 µ + ψ 0 (γ µ (i∂µ − e0 A0 µ ) − m0 ) ψ0 .W. m = m(µ) .2 Renormalization 2α 1 + O(1) . the tensor structure is ﬁxed by gauge invariance which requires q µ Πµν (q) = 0 . leaving physical observables ﬁnite. 30
30
. How do we get rid of them? The crucial insight is that the parameters of the Lagrangean. are not observable. 1 / ∆L = − (Z3 − 1) Fµν F µν + (Z2 − 1) ψi∂ψ 4 / − (Zm − 1) mψψ − (Z1 − 1) eψ Aψ . B UCHMÜLLER AND C. Therefore. Πµν (q) = gµν q 2 − qµ qν Π q 2 . (170)
In terms of the renormalized ﬁelds and parameters the Lagrangian (167) reads 1 L = − (∂µ Aν − ∂ν Aµ ) (∂ µ Aν − ∂ ν Aµ ) + ψ (γ µ (i∂µ − eAµ ) − m) ψ + ∆L . To make this more explicit. let us express. Rather. The remaining scalar quantity Π(q 2 ) has the divergent part Π q2 = 4. the QED Lagrangian in terms of bare ﬁelds Aµ and ψ0 and bare parameters m0 and e0 . divergencies of bare parameters can cancel against divergent loop corrections. 4 where ∆L contains the divergent counterterms. (172) (171)
The counter-terms have the same structure as the original Lagrangian and lead to new vertices in the Feynman rules: i. A0 µ = Z 3 Aµ . respectively. m0 = Z2 ψ0 = Z2 ψ . Zm m. Z1 e0 = √ µ2−d/2 e . 4 (167)
The ‘renormalized ﬁelds’ Aµ and ψ and the ‘renormalized parameters’ e and m are then obtained from the bare ones by multiplicative rescaling. because of Lorentz invariance. It has the same tensor structure as the vacuum polarization.2. e = e(µ) . but they are still there. the ‘bare’ parameters. Hence. However. Z2 Z 3 (168) (169)
Note that coupling and electron mass now depend on the mass parameter µ. 3π (166) (165) (164)
So far we have isolated the divergences. The vacuum polarization seems more complicated since it is a second rank tensor. the sum of bare parameters and loop-induced corrections are physical. as an example.
q µ ν
−i (Z3 − 1) × gµν q 2 − qµ qν
Photon wave function counter-term (counter-terms are generically denoted by ). L ÜDELING
which contribute to the mass renormalization and the wave function renormalization.

The proof is highly non-trivial and has been a major achievement in quantum ﬁeld theory! 4. we can safely take the limit → 0. the O (α) vacuum polarization now has two contributions. A Ward identity.2. which follows from gauge invariance. the renormalized coupling e(µ) depends on the renormalization scale µ [cf. for example.F IELD THEORY AND THE S TANDARD M ODEL
p p
ii. 2π Similarly. Electron mass counter-term. There are different ways to ﬁx them. All schemes give the same results for physical quantities. corresponding to different renormalization schemes. (169) in and e(µ). (174)
(175)
The other constants Z2 and Zm are ﬁxed analogously. one has
+
= −ieγ µ
α 1 + (Z1 − 1) + O (1) 2π
. 3π (176) = −i gµν q 2 − qµ qν 2α 1 + (Z3 − 1) + O (1) 3π . Z1 = 1 − + which yields Z3 = 1 − 2α 1 + O (1) . e0 = Z1 √ µ−2+d/2 e(µ) = e(µ)µ− Z2 Z 3
/2
Z3
1 −2
. Having absorbed the divergences into the renormalized parameters and ﬁelds. 9]. the O (α) correction to the electron–photon vertex. iii.
where we have used the Ward identity Z 1 = Z2 .
(173)
Demanding that the whole expression be ﬁnite determines the divergent part of Z 1 .
/ −i (Z2 − 1) p −i (Zm − 1) m −ie (Z1 − 1) γ µ
Electron wave function counter-term. It is very remarkable that the scale dependence is determined by the divergences.
iv.
Vertex counter-term. expand Eq. Calculating. (169)]. The theory now yields well-deﬁned relations between physical observables.
The renormalization constants Zi are determined by requiring that the counter-terms cancel the divergences. Divergences can be removed to all orders in the loop expansion for renormalizable theories [3. They can be determined as power series in α. The ﬁnite parts of the renormalization constants are still undetermined. yields the important relation Z 1 = Z2 .
. e0 = e(µ) 1 − = e(µ) 1 2 ln µ + · · · 1+ 1 α + ··· 3π .3 Running coupling in QED
Contrary to the bare coupling e0 . Quantum electrodynamics and the Standard Model belong to this class. e (µ)
4
e2 (µ)
e2 (µ) +1− ln µ + O 12π 2 24π 2 31
31
(177) . α 1 + O (1) . To see this. but differ at intermediate steps. The lowest order counter-terms are O (α) and have to be added to the one-loop diagrams.

L ÜDELING
where we have used α = e2 /(4π). the coupling increases with µ until it approaches the so-called Landau pole where the denominator vanishes and perturbation theory breaks down. (179) is the so-called β function.
4. Such terms make the expansion unreliable unless one chooses µ2 ∼ q 2 . It is.W. e2 q 2 represents the effective interaction strength at a momentum (or energy) scale q 2 or. Qualitatively. Quantitatively. β(e) = b0 3 e + O e5 . ∂µ ∂µ 24π 2 (178)
This equation is known as the renormalization group equation. and we shall not be able to discuss this in detail. since there are more diagrams to calculate. which again are absorbed by renormalization constants.
2 increases to α MZ =
the value conveniently used in electroweak precision tests. Using a given value of e at a scale µ 1 . B UCHMÜLLER AND C. (4π)2 with b0 = 2 . alternatively. such as a scattering amplitude at some momentum transfer q 2 . much more complicated. The positive β function in QED implies that the effective coupling strength decreases at large distances. Schematically. and they lead to similar divergences. Since the bare mass e0 does not depend on µ. in particular to QCD [5]. (182)
+ +
Z2 .4
Running coupling in QCD
Everything we did so far for QED can be extended to non-Abelian gauge theories. this can be understood as the effect of ‘vacuum polarization’: Electron–positron pairs etc. the coupling α at another scale µ is given by α(µ) = α (µ1 ) µ . however.
1 127 . from the vacuum screen any bare charge at distances larger than the corresponding Compton 1 wavelength. 3 (180)
The differential equation (179) can easily be integrated. one ﬁnds that the value α(0) = 137 . b0 1 − α (µ1 ) (2π) ln µ1 (181)
Since b0 > 0. these are + + Z1 . at a distance of r ∼ 1/q. The additional diagrams contain gluon self-interactions and ghosts. (184)
. What is the meaning of a scale-dependent coupling? This becomes clear when one calculates physical quantities. and the function on the right-hand side of Eq. + 32
32
(183) + Z3 . differentiation with respect to µ yields 0=µ and therefore µ ∂ e3 e= + O e5 ≡ β(e) . ∂µ 24π 2 (179) ∂ e3 ∂ e0 = µ e − + O e5 . measured in Thompson scattering.2. In the perturbative expansion one then ﬁnds terms ∝ e2 (µ) log q 2 /µ2 . Hence.

i. Correspondingly. G EW = SU(2)W × U(1)Y . the building blocks are massless left. 5. Neutral-current interactions.e. which couple quarks and leptons differently. ∂µ (4π)2 b0 = − 4 11 Nc − Nf 3 3 . Eq. QCD 5
Electroweak theory
So far we have studied QED.and right-handed fermions.
(187)
The analogue of the Landau pole now occurs at small µ or large distances. one obtains the β function for the gauge coupling g.F IELD THEORY AND THE S TANDARD M ODEL
The renormalized coupling can again be deﬁned as in QED. seemingly inconsistent with gauge invariance. we omit the right-handed one. only involve left-handed fermions and readily change ﬂavour. Z2 Z 3 (185)
The coefﬁcients of the 1/ -divergences depend on the gauge group and on the number of different fermions. the simplest gauge theory. Charged-current interactions. on the other hand. weak interactions come in two types.and right-handed fermions. and ﬂavourchanging neutral currents are strongly suppressed. They are shortranged. the inverse of Λ QCD gives roughly the size of hadrons. we have seven chiral spinors: Two each for up. 2 1 1 + γ 5 ψR . (169). The coupling at a scale µ can again be expressed in terms of the coupling at a reference scale µ 1 . and QCD. weak interactions also turn out to be described by a non-Abelian gauge theory. which is the quantum number associated with the SU(2) W factor. But there also are the weak interactions. g0 = Z1 √ µ−2+d/2 g . Despite these differences from QED and QCD. 2
In a chiral gauge theory. as in the strange quark decay s → ue − ν e . (186)
Note that for Nf < 11Nc /4 the coefﬁcient is negative! Hence. and the U(1) charge is the hypercharge Y .. The calculation of this coefﬁcient earned the Nobel Prize in 2004 for Gross. α(µ) = α (µ1 )
|b0 | µ 1 + α (µ1 ) (2π) ln µ1
. mediated by the W ± bosons. Furthermore. which is the basis of the parton model. charged and neutral current–current interactions. For one generation of Standard Model particles. which seem rather different. Yet the electroweak theory is different for two reasons: it is a chiral gauge theory. one can treat in deep-inelastic scattering quarks inside the proton as quasi-free particles. and just one for the neutrino which we shall treat as massless in this section. and the gauge symmetry is spontaneously broken. For a SU(Nc ) gauge group with Nf ﬂavours of fermions. which requires massive messenger particles. couple both left. µ b0 3 ∂ g= g + O g5 . The decrease of the coupling at short distances is the famous phenomenon of asymptotic freedom. The electroweak gauge group is a product of two groups.7 fm. ψL = ψR = (188)
with different gauge quantum numbers. rhad ∼ Λ−1 ∼ 0. For QCD withN c = 3 and Nf = 6. the coupling decreases at high-momentum transfers or short distances.and down-type quark and charged lepton.1 Quantum numbers 1 1 − γ 5 ψL . the prime example of a non-Abelian gauge theory. As a consequence. At the QCD scale gluons and quarks are strongly coupled and colour is conﬁned [5]. 33
33
. Politzer and Wilczek. Here the subscript W stands for ‘weak isospin’. the pole is at the ‘QCD scale’ ΛQCD 300 MeV.

we know that charged currents connect up. like the U(1) boson Bµ . σ1 = 0 1 1 0 . B UCHMÜLLER AND C. since weak interactions do not change colour.
5 Here we use ‘representation’ as meaning ‘irreducible representation’. respectively.and down-type quarks. we can write down the covariant derivatives. and that the W ± bosons couple only to left-handed fermions. L ÜDELING
The assignment of quantum numbers. while W µ does not. one has
I Lint = −gJW. 2. so it has only one-dimensional representations. The SU(2) W has three I generators. eR . (193) (192)
1 2 3 we see that Wµ and Wµ mix up.
I where Wµ = 1 σ I Wµ . which takes the form (current) · (vector ﬁeld).
(190)
With these representations. lL = νL eL . nor left. to allow a consistent coupling to gauge bosons. (189)
with the hypercharges (which we shall justify later) ﬁeld: qL uR hypercharge:
1 6 2 3
dR
1 −3
lL −1 2
eR −1
.with down-type quarks and charged leptons with neutrinos. and therefore three gauge ﬁelds W µ . and from eL and νL .W.
It is often convenient to split the Lagrangian into the free (kinetic) part and the interaction Lagrangian. So we obtain the SU(2)W multiplets qL = uL dL . The covariant derivatives acting on the left-handed ﬁelds are Dµ ψL = ∂µ + igWµ + ig Y Bµ ψL . µ = q L γµ qL − lL γµ lL + uR γµ uR − dR γµ dR − eR γµ eR . which corresponds to the grouping into representations of the gauge group. In the electroweak theory.and right-handed ﬁelds.
34
34
. µ = q L γµ 1 σ I qL + lL γµ 1 σ I lL . σ3 = 1 0 0 −1 . ∂ µ J µ = 0. 2
(191)
while the right-handed ﬁelds are singlets under SU(2) W . which would violate Lorentz symmetry. From the explicit form of the Pauli matrices. and hence do not couple to the W bosons. This suggests to form doublets from uL and dL . 6 2 3 3
(195) (196)
These currents have to be conserved. is obtained as follows: The non-Abelian group SU(2) W has a chargeless onedimensional singlet (1) representation and charged multidimensional representations. The U(1)Y gauge ﬁeld is Bµ . dR . and to keep the right-handed ﬁelds as singlets. µ W I µ − g JY. I = 1. Furthermore. Dµ ψR = ∂µ + ig Y Bµ ψR . This means we can assign different hypercharges to the various singlets and doublets of SU(2) W . which we choose to be the Pauli matrices. uR . µ B µ . starting with the two-dimensional doublet (2) representation 5 . The U(1)Y factor is Abelian. We are not allowed to mix quarks and leptons. and the coupling constants are g and g . σ2 = 0 −i i 0 . 3.
(194)
with the currents
I JW. 2 2 1 1 2 1 JY. Of course we can build reducible representations of any dimension.

The left. T B T C
L
− tr
T A. 10). In general. T B and T C in the left. whose anticommutator is 1 σ I . W Y
1 The SU(2)W generators are 1 σ I . A = tr T A. J B and J C . Z and γ bosons of the Standard Model. nor in the quark and lepton sector individually. for currents J A . two. where the quantized theory just lacks that particular symmetry. the so-called triangle diagrams (see Fig. It does not vanish in either the left. We shall see that the choice of hypercharges in (190) is severely constrained by the consistency of the theory. in principle there are four combinations of currents. which thus are automatically anomaly free. For gauge symmetries. which means the current is no longer conserved. the resulting quantum ﬁeld theory might not have that symmetry any more. For the electroweak theory. otherwise the theory is inconsistent. However.and right-handed sectors. JB JB
ψL
ψR
A ∝
JA JC
−
JA JC
Fig. σ 2 2
Y
L
1 1 = δ IJ tr [Y ]L = δ IJ 2 2
Nc
1 1 3 · − 6 2
=0. This vanishing of the anomaly is again related to the number of colours. Furthermore. however. one or no SU(2) W current. After quantizing the theory. Y } Y ]L − tr [{Y. let us brieﬂy discuss anomalies. the currents have to be conserved.
!
(197)
Here the trace is taken over all fermions. Suppose we have a classical ﬁeld theory with a certain symmetry and associated conserved current.1
Anomalies
Before considering the Higgs mechanism which will lead to the identiﬁcation of the physical W ± .or right-handed sector. 10: The gauge anomaly is given by triangle diagrams with chiral fermions in the loop Anomalies are caused by certain one-loop diagrams. Hence the SU(2)2 U(1)Y anomaly is W
A = tr
1 I 1 J σ . T B T C
R
= 0. containing three.and right-handed fermions contribute with different sign. so we only have to check the SU(2) 2 U(1)Y and U(1)3 anomalies. the anomaly A is the difference of the traces of the generators T A . Anomalies are not a problem for global symmetries.1. This is the case in QED and QCD. since the right-handed ones are SU(2) W singlets. Hence the vanishing of 35
35
L
− tr Y 3 1 3
3
R
1 6
3
+2 −
1 2
3
−3
2 3
3
−3 −
− (−1)3
(199)
. so if they have the same quantum numbers. the anomaly vanishes.
(198)
We see that it vanishes only if quarks come in three colours! The U(1)3 anomaly also vanishes: Y A = tr [{Y. Y } Y ]R = 2 tr Y 3 =2 3·2 =0. the trace of any odd number of σ I matrices vanishes.F IELD THEORY AND THE S TANDARD M ODEL
5. 2 σ J = 1 δ IJ . 2 2 2 only the left-handed ﬁelds contribute. This is called an anomaly.

The key ingredient for the Higgs mechanism is a complex scalar ﬁeld Φ. the Standard Model is anomaly free. such a term is not gauge invariant. the Higgs ﬁeld settles in this minimum. mψψ = m ψ L ψR + ψ R ψL . would not be a consistent theory. Anomaly cancellation is not restricted to the electroweak gauge currents. A simpler version of this effect is what happens in superconductors: the condensate of Cooper pairs induces an effective mass for the photon. A ‘Standard Model’ without quarks.
(203)
Here we have eliminated the upper component and the imaginary part of the lower one. Hence. since it is a complete singlet. at Φ † Φ = v 2 ≡ µ2 /λ. nor would a ‘Standard Model’ with four colours of quarks. leading to the Meissner–Ochsenfeld effect where external magnetic ﬁelds are expelled from the superconductor. levitating it. the minimization of the potential only ﬁxes the modulus Φ† Φ.and right-handed fermions. suggested by neutrino masses. The same is true for the C last possible anomaly. as it should be. In the vacuum. At ﬁrst sight. µ2 > 0 . one of the four degrees of freedom. 36
36
. 5. implying long-range forces. Furthermore. The I gauge bosons Wµ and Bµ are massless. the gravitational one. i. but applies to the strong force and gravity as well: Mixed SU(3) C -U(1)Y anomalies vanish by the same argument as above: Only the SU(3)2 U(1)Y triangle contributes. so that the dynamical ﬁeld H(x) vanishes in the vacuum. so that electromagnetic interactions become short-ranged. The other three. Note that a right-handed neutrino. does not pose any problem. without any charge. all particles of one generation with their strange hypercharges have to conspire to cancel the different anomalies. which is a hint to grand uniﬁed theories where anomaly cancellation is often automatic. and we can choose the following form of Φ. for instance. the fermions are massless as well. again because of gauge invariance: a mass term mixes left. The crucial feature of the Higgs 2 ﬁeld is its potential.. which is often referred to as unitary gauge: Φ= 0 v+
1 √ 2
H(x)
. but it is tr [Y ] L − tr [Y ]R = 0. because a mass term m 2 Wµ W µ would violate gauge invariance. which has four real degrees of freedom. The way out is the celebrated Higgs mechanism: Spontaneous symmetry breaking generates masses for the gauge bosons and fermions without destroying gauge invariance.2 Higgs mechanism
The electroweak model discussed so far bears little resemblance to the physics of weak interactions. where two non-Abelian gauge currents are replaced by the energy-momentum tensor Tµν . (200)
and since these have different gauge quantum numbers. however. L ÜDELING
anomalies provides a deep connection between quarks and leptons in the Standard Model. For this. can be eliminated by a gauge transformation. B UCHMÜLLER AND C.
H = H∗ . 2 Dµ Φ = (201)
(202)
This potential has a minimum away from the origin.W. with i ∂µ + igWµ − g Bµ Φ . and thus it does not contribute to any gauge anomaly. 2 2 1 V Φ† Φ = −µ2 Φ† Φ + λ Φ† Φ .e. which is of the Mexican hat form: L = (Dµ Φ)† (D µ Φ) − V Φ† Φ . We have also shifted the lower component to the vacuum value. which is a doublet under SU(2)W with hypercharge − 1 .

and the three remaining ones are g 2 . which was given in terms of the old ﬁelds W µ and Bµ and their currents (195) and (196). The associated currents are separated into a charged ± current (for Wµ ) and neutral currents (for Aµ and Zµ ): g LCC = − √ 2 LNC =
+ (uLi γ µ dLi + ν Li γ µ eLi ) Wµ + h. i. 2 2 – two neutral vector bosons with masses M Z = 1 g 2 + g 2 v 2 = MW cos−2 θW and Mγ = 0. the photon.e.µ.τ
with the electric charge
Qi = Ti3 + Yi . the Higgs Lagrangian (201) becomes L = λ 4 v 2 1 λ λ + ∂µ H ∂ µ H − λv 2 H 2 + √ vH 3 + H 4 2 8 2 2 g 0 2 0 g2 1 1 1 2 3 + v+√ H Wµ . it describes one massless particle. = −eJem µ Am − sin 2θW
i=1. .d. however. These 4 potentials just differ by the shift ∼ v . The massless eigenstate of the mass matrix. is the linear com3 bination Aµ = − sin θW Wµ + cos θW Bµ . 3µ gg W Bµ g2
(204)
.
6
37
37
..
(205)
To summarize. 2 – and one neutral Higgs boson with mass m 2 = 2λv 2 .F IELD THEORY AND THE S TANDARD M ODEL
In unitary gauge.
(206)
JY µ B µ e JZ µ Z µ . the theory contains the following mass eigenstates:
2 – Two charged vector bosons W ± with mass MW = 1 g 2 v 2 .e. 3 + sin θ B . B µ 4 2
0
g2 gg
0
The ﬁrst line could be interpreted as vacuum energy density. H
The Higgs mechanism and the diagonalization of the vector boson mass matrix allow us to rewrite I the interaction Lagrangian (194). so that its expectation value vanishes in the vacuum. nothing prevents us from adding an arbitrary constant to the Lagrangian. the orthogonal combination is the Z boson. is the last one: It contains mass terms for the vector bosons! A closer look at the mass matrix reveals that it only is of rank three..c.t. and are indistinguishable within QFT. such an interpretation is on shaky grounds in quantum ﬁeld theory. W µ . H The most important line. Here we have introduced the Weinberg angle θ .b.
i=u.
(208)
Generally. the Higgs potential is often written as Φ† Φ − v 2 . obtaining any desired ‘vacuum energy’.2.3 3 −gJµ W 3 µ − g
(207)
with the electromagnetic and Z currents Jem µ = ψ i γµ Qi ψi . W ± and Z bosons.
cos θW =
g g2 +g2
. However. i. which is deﬁned Zµ = cos θW Wµ W µ W by sin θW = g g2 +g
2
W1µ 2 µ W .e. and one which is even heavier. The second line describes a real scalar ﬁeld H of mass m 2 = 2λv 2 with cubic and quartic self-interactions. i. we have identiﬁed the physical γ. and g 2 + g 2 . so it has one zero eigenvalue. two of equal non-zero mass. ` ´2 For example. In other words. in terms of the physical ﬁeld. g 2 . W µ .e.. s. so we shall ignore this term 6 . a cosmological constant.c.

The Z boson. allowed Yukawa couplings of the Higgs doublet to two fermions. The coupling constant e is related to the original couplings and the weak mixing angle: e = g sin θW = g cos θW . however. but also for the fermions.νµ . (211)
The photon couples only vector-like. So there are still twelve degrees of freedom. B UCHMÜLLER AND C. direct mass terms are not allowed in the Standard Model. There are. but only hidden. three Higgs degrees of freedom are gone (one remaining). but they have resurfaced as extra components of three massive vector ﬁelds 7 (nine). i. The Higgs mechanism can also be described in a manifestly gauge-invariant way. on the other hand.s.or down-type quark singlets. . zero for singlets) and the hypercharge Y . a fact known as lepton universality. Each term is parametrized by a 3 × 3 matrix in generation space. and to lepton doublet and charged lepton singlets.
Remember that a massless vector has only two (transverse) degrees of freedom.e. it does not distinguish between different chiralities.c. The ‘spontaneous breaking of gauge invariance’ reshufﬂes the degrees of freedom of the theory: Before symmetry breaking.
7
38
38
.t. After symmetry breaking (and going to unitary gauge).
(212)
These Yukawa couplings effectively turn into mass terms once the electroweak symmetry is spontaneously broken: A vacuum expectation value Φ X = v inserted in the Lagrangian (212) yields Lm = (mu )ij uL i uR j + (md )ij dL i dR j + (me )ij eL i eR j + h.. In particular.b e. the Z couples in the same way to all leptons. is somewhat misleading: Gauge symmetries are never broken.W. L ÜDELING
JZ µ =
i=u.3 Fermion masses and mixings
The Higgs mechanism generates masses not only for the gauge bosons.and right-handed ﬁelds. couples to the vector and axial-vector currents of different fermions ψ i (i. so twelve in total.
(209)
Here the fermions ψi are the sum of left. we have the complex Higgs doublet (four real degrees of freedom) and four massless vector ﬁelds with two degrees of freedom each. is given by the sum of the third component of weak isospin T 3 (± 1 for doublets. so it seems we have lost three gauge symmetries. is just a consequence of choosing the unitary gauge. The Higgs mechanism described above is also called spontaneous symmetry breaking. and uL .. They are given by the respective couplings vi and ai . the electric charge Q. dL and eL denote the respective components of the quark and lepton doublets qL and lL .c. mode.e. This reproduces the 2 known electric charges of quarks and leptons. however.. and all currents remain conserved. couplings to quark doublets and either up. which justiﬁes the hypercharge assignments in (190). (213)
Here the mass matrices are mu = hu v etc. which are universal for all families. while a massive one has a third. As already emphasized. The Lagrangian (204) has only a manifest U(1) symmetry associated with the massless vector ﬁeld. LY = (hu )ij qL i uR j Φ + (hd )ij q L i dR j Φ + (he )ij lL i eR j Φ + h. ψi = ψLi + ψRi . their left-and right-handed components) with different strengths. longitudinal.ντ
ψ i γµ vi − a i γ 5 ψi . 5. however.τ. (210)
The coupling to the photon.µ. and one vector ﬁeld stays massless (another two). . where Φ is given by Φa =
∗ ab Φb .d. This.νe .c. This term. They come in three classes.

V
(d) † †
(214a) (214b) (214c)
md V (d) = diag(md . 2 where i.and down-type matrices V (u) and V (d) are not identical. β.
etc. mc .α .
(u)
dL i = Viα dL.) to mass eigenstates (with indices α. there is no ﬂavour mixing in the neutral current. There is one possible exception: a right-handed neutrino. 39
39
mν = −mD M −1 mT . and mD ∼ mt ∼ 100 GeV for the largest Dirac mass. D
(220)
. .
V (e) me V (e) = diag(me .c. .c. . For M ∼ 10 15 GeV. mb ) . . which is gauge invariant only for singlet ﬁelds. which one may add to the Standard Model to have also neutrino masses.mass = hν ij lL i νR j Φ + Mij ν R i νR j + h. mτ ) . . but rather g + LCC = − √ Vαβ uL α γ µ dL β Wµ + h. and a diagonalization of the (νL . one ﬁnds m ν ∼ 10−2 eV.
(d)
uR i = Viα uR α . with unitary matrices V . 1 Lν. j again are family indices.F IELD THEORY AND THE S TANDARD M ODEL
The mass matrices thus obtained are in general not diagonal in the basis where the charged current is diagonal. mt ) .
†
V (u) V (u) =
†
. .
This amounts to a change of basis from the weak eigenstates (indices i.
(d)
(215)
The up. j. the charge conjugate spinor C has charges opposite to the original one. which has an important consequence: The charged-current couplings are now no longer diagonal. (218)
where C = iγ 2 γ 0 is the charge conjugation matrix. ms .
(u)
dR i = Viα dR α . (219)
The Higgs vacuum expectation value v turns the coupling matrix h ν into the Dirac mass matrix mD = hν v. It also has opposite chirality. νR ) system leads to three light modes ν i with the mass matrix Large Majorana masses naturally appear in grand uniﬁed theories.
†
(217)
which carries the information about ﬂavour mixing in charged-current interactions. successfully relates neutrino physics to grand uniﬁed theories. V (u) mu V (u) = diag(mu . It is a singlet of the Standard Model gauge group and can therefore have a Majorana mass term which involves the charge conjugate fermion ψ C = Cψ T . This ‘see-saw mechanism’. Because of the unitarity of the transformations. which explains the smallness of neutrino masses masses as a consequence of large Majorana mass terms. . Thus we can produce C a mass term ψ ψ (remember that a mass term always requires both chiralities). As the name suggests. (216) 2 with the CKM matrix Vαβ = V (u) αi V (d) iβ . So a right-handed neutrino νR can have the usual Higgs coupling and a Majorana mass term. The eigenvalues of the Majorana mass matrix M can be much larger than the Dirac masses. We saw that the Higgs mechanism generates fermion masses since direct mass terms are not allowed due to gauge invariance. which is consistent with results from neutrino oscillation experiments. P L ψR = ψR .): uL i = Viα uL α . . They can be diagonalized by bi-unitary transformations. mµ .

i. and the Higgs mass m H which is not yet known. fb Af = fb
f f σf + σ b f f σf − σ b
. Γ(Z → hadrons) (221)
– Forward–backward asymmetries: In e + + P em → Z/γ → f f reactions. the Fermi constant G F and the Z boson mass MZ . γ
f e− θ
f
e+ e+
(a)
f
f
(b)
Fig. c . W and Z bosons were produced in huge numbers. there is a correlation between the directions of the outgoing fermion and the incoming electron. This asymmetry has been measured for several types of ﬁnal-state fermions.. which are more easily measured: The ﬁne-structure constant α. This is quantiﬁed by the asymmetries Af . left–right and forward–backward asymmetries. – Ratios of partial decay widths. Fig. The grey blobs indicate higher order corrections which must be included to match the experimental precision. the two gauge couplings and the two parameters of the Higgs potential: g.W. τ. while σb is the cross-section for backward scattering. There are many observables related to their production and decay (Fig. In W decays.e. π/2] in f Fig. 12: The forward–backward asymmetry Afb : In the process e+ e− → Z/γ → f f. the fermion and antifermion can have different ﬂavour. b.
(222)
f where σf is the cross-section for an outgoing fermion in the forward direction. L ÜDELING
5. the ratio of the partial Z width into bottom quarks to that into all hadrons. Rb = 1 Γ Z → bb . which are known to great accuracy. θ ∈ [0. the direction of the outgoing fermion is correlated with the incoming electron.
f W f
(a) (b)
f Z f
At LEP. Afb = LR
f σLf f f f f σLf − σLb − σRf + σRb f σLb f σRf f σRb
+
+
+
3 ≡ Af . 4
(223)
40
40
.
for f = µ. 12. µ2 and λ. 11: Decays of the W and Z bosons into two fermions. They can be traded for four other parameters. 11). These include – The W mass MW and the decay widths ΓW and ΓZ .4
Predictions
The electroweak theory contains four parameters. for example. g .
Also important are double. mostly at LEP with centre-of√ mass energy s = MZ .
e− Z. B UCHMÜLLER AND C.

Z
W ±.
(225)
At tree level. which lead to different cross-sections for the processes Z → f L f R and Z → fR f L . ∆αhad (mZ ). which differs from the low energy value α(0) in particular by hadronic corrections. on which the vector and axial-vector couplings of the Z boson depend.
b W+ W+ Z
t Z
t
(a) Heavy quark corrections
t
H
H
W ±. The leading correction is ∆ρ(t) = 3GF m2 m2 t √t ∝ 2 . are an important prediction of the electroweak theory. and are dominated by the top quark. which depend on the masses of the particles in the loop. as in Fig 13(a). which is only broken by the U(1) Y gauge interaction and by Yukawa couplings. Z
(b) Higgs corrections
Fig.F IELD THEORY AND THE S TANDARD M ODEL
The reason for these asymmetries is the presence of the axial couplings a i in the Z boson current (209).
lept sin2 θeff =
1 4
1−
vl al
. α(mZ ). Thus the corrections depend on the fermion masses. Z
W ±. one can deduce the ai and vi couplings for fermions from the forward–backward asymmetries. 13. ρ = 1. Z
W ±. and therefore to ρ. so that they test the non-Abelian nature of the electroweak theory. MW 8π 2 2 (226)
This led to the correct prediction of the top mass from electroweak precision data before the top quark was discovered at the Tevatron.
(224)
– Electroweak measurements by now are very precise. Diagrams with gauge boson self-interactions have been omitted. The tree-level value ρ = 1 is protected by an approximate SU(2) symmetry. 41
41
. 13: Radiative corrections to the masses of the W and Z bosons. and ﬁnally the weak mixing angle. called custodial symmetry. deﬁned by ρ=
2 MW 2 MZ cos2 θW
. The theoretical predictions critically depend on the the electromagnetic coupling at the electroweak scale. An important observable is the ρ parameter. and require the inclusion of W boson loops in theoretical calculations. Loop corrections to the masses of the gauge bosons. Thus. due to quark or Higgs boson loops as in Fig.

For the process e +ein → and on the show hand.W. MW (227)
From this relation. as for the W-pair cross section. Combined W W γ cross section as a function of the 160 180 200 ter-of-mass energy. which affects ρ quadratically. the full calculation including the results from LEP. Only L3 and DELPHI results are used
0
180
190
200
√s (GeV)
√s (GeV)
(a) W boson pair production cross-section at LEP2. Fig. 4. . Combined production. I have also to thank my colleagues from the four LEP ex42 periments and the members of the LEP Electroweak Working 42 Group for providing the data which were presented in this report. Clearly. which Pre. Although not yet ﬁnal. Fig. and from precision measurements one can obtain a ﬁt for the Higgs mass. NC02 diagrams for the production of Z–pair events level of accuracy has shows the measured cross-section The tree-level diagrams are given in Fig. Natale: W and Z pair production at LEP2
σWW (pb)
0. In the electroweak theory. L3 and while omission of the γWW and ZWW vertices leads to large aborations have beendiscrepancies. values using the YFSZZ Montecarlo program tially. all diaThe results obtainedgrams agrees well with data. [11]) analysed and provides a powerful tool to test the Standard Model.5
YFSWW/RacoonWW
only ν exchange (Gentle)
0 . but this time the effect is only logarithmic: ∆ρ(H) = −C ln m2 H 2 . 6.
Acknowledgements. This is shown in the famous blue-band plot. this can be seen in the process e +e− → W +W −. – A characteristic prediction of any non-Abelian gauge theory is the self-interaction of the gauge bosons. there is no triple Z vertex. L ÜDELING
e− νe
W−
e− Z/γ
W−
Fig. don’t other deviations from the Standard Model. OPALthe herein presented are based on the full available statistics by the DELPHI. so at is observed. compared with theoretical predictions. In many cases. the Z-pair ﬁnal state is produced by means of o main born level diagrams referred to as NC02 (Neutral rrent) and shown in Fig.corrections.
Fig. Let me thank the organisers of EPS03 Conference for their professionalism and warm hospitality. S.4
20
1
an/Λ2=0.(b) For Z pair Z-pair cross sections compared to predicted dictions which ignore the WWW vertex deviate substan. the g. No statistically signiﬁcant deviation tree level one only has the t-channel diagram which is similar to the diagram or γZZ) vertex. the Higgs mass (weakly) inﬂuences many other quantities. 5. 15: Gauge boson pair production cross-sections at LEP2 energies (from Ref. 15(a) reached the sensitivity threshold for radiative Clearly. the accuracy of this prediction strongly depends on the experimental error on the top mass.
ZZ production
LEP. there is no triple gauge boson (ZZZ combined and they are shown − ZZ.1
0. 14: The process e+e− → W +W−. 21. 4.2
10
0 180 190 200 210 ZWW vertex (Gentle) no
e √s (GeV)
0. 14. The signal deﬁnition for the
. γWW and ZWW. 5. However.6
LEP
σZZ (pb)
σWWγ (pb)
30
13/07/2003
PRELIMINARY
LEP
02/08/2004
11/07/2003
EEWWG RacoonWW
n PRELIMINARY
a /Λ =0. The diagrams of panel (b) contain triple gauge boson vertices. one can obtain a prediction for the mass of the Higgs boson.
e+
(a)
W+
e+
(b)
W+
The correction due to the Higgs boson diagrams in Fig. 13(b) again depends on the Higgs mass.2
2
LEP PRELIMINARY
ZZTO and YFSZZ
0.agrees well with the experimental result. and Fig. B UCHMÜLLER AND C.

one can use it as an effective theory at energies small compared to the W mass. kaons and B mesons. 15(b). 17). The agreement between theory and data is evident from Fig. There is one deviation of almost 3σ. and they can be reliably calculated in perturbation theory. A four-fermion theory for the weak interactions was ﬁrst introduced by Fermi in 1934. 2
where GF is Fermi’s constant. requires hadronic processes which are known experimentally with less accuracy and are also theoretically subject to larger uncertainties. which allow for very precise measurements. = √ 2 4 2 MW 2 2 v2 (229)
which is inversely proportional to the Higgs vacuum expectation value v 2 .F IELD THEORY AND THE S TANDARD M ODEL
in Fig. on the other hand. The electroweak theory is extremely well tested experimentally. 5.4. it cannot be considered a fundamental theory.
q2 W
2 MW
5. but with an electron instead of the neutrino. an effective theory for momentum transfers small compared to the W mass. and also all fermion masses.
43
43
. g2 1 GF = √ . This is sufﬁcient for many applications in ﬂavour physics. 17: W boson exchange can be described in terms of the Fermi theory. to the level of 0. This symmetry is spontaneously broken down to U(1) em by the Higgs mechanism which generates the gauge boson and Higgs masses. 16: µ decay (228)
GF µ † = − √ JCC JCCµ .1%. all other quantities agree within less than 2σ. However. the momentum transfer is much smaller than the mass of the W boson. since direct mass terms are forbidden by gauge invariance. 16). Hence to good approximation one can ignore q 2 and replace the −2 propagator by MW . Fig. 18.1 Fermi theory
2 MW
The exchange of a W boson with momentum q in a Feynman diagram contributes a factor of − to the amplitude. This impressive agreement is only possible because of two properties of the electroweak interactions: they can be tested in lepton–lepton collisions. where the energy scale is set by the masses of leptons. This amounts to introducing an effective four-fermion vertex (see Fig. The results of a global electroweak ﬁt are shown in Fig. For low-energy processes like muon decay (see Fig. QCD. The W propagator is replaced by a four-fermion vertex ∝ GF . 14(a). q2
eff LCC −2
νµ µ−
e−
νe
Fig.5
Summary
The electroweak theory is a chiral gauge theory with gauge group SU(2) W × U(1)Y . which probes loop effects of the non-Abelian gauge theory. Since it is not renormalizable.

670 ± 0.115 ± 0. [11].2324 ± 0.0032 0.0023 41.W.02758 ± 0.020 0.0992 ± 0.478 20.0012 ΓW [GeV] mt [GeV] mW [GeV]
0
1
2
3
Fig.1480 0. L ÜDELING
Measurement ∆αhad(mZ) ΓZ [GeV] σhad [nb] Rl Afb Rb Rc Afb Afb Ab Ac Al(SLD)
2 lept 0.
44
44
.3 91.058 172.0035 0.c 0.1875 ± 0.1037 0.376 2.935 0.2314 80.025 0.1465 ± 0.4959 41.0742 0.02767 91. The right column shows the deviation of the ﬁt from measured values in units of the standard deviation.923 ± 0.668 0.l 0 (5)
Fit
0.037 20. 18: Results of a global ﬁt to electroweak precision data.00035 0.01643 0.9
|O 0
meas
−O |/σ 1 2
fit
meas
3
mZ [GeV]
0.01714 ± 0.540 ± 0. From Ref.00066 0.5 ± 2.0707 ± 0.0030 0.21581
Al(Pτ)
sin θeff (Qfb) 0.1874 2.21629 ± 0.0021 80.4952 ± 0.1480 0.030 2.1513 ± 0.00095 0.1721 ± 0.092 172.743 0.027 0.b 0.1722 0. B UCHMÜLLER AND C.404 ± 0.0021 2.767 ± 0.0016 0.

For a heavy Higgs. bb and tt pairs. for instance. Γ(H → ZLZL) = Γ(H → W LW L) = 2 32π 2
The branching fractions of the Higgs into different decay products strongly depend on the Higgs mass. These theories generically have problems with electroweak precision tests and the generation of fermion masses. as the ﬁfth component of a ﬁve-dimensional vector ﬁeld. and so the discovery of the Higgs boson and the detailed study of its interactions is a topic of prime importance for the LHC and the ILC. – Technicolour theories model the Higgs as a composite particle of size 1/Λ TC . and the Higgs ﬁeld can be identiﬁed. with m H > 2MW .F IELD THEORY AND THE S TANDARD M ODEL
6
The Higgs proﬁle
The only missing building block of the Standard Model is the Higgs boson. as shown in Fig. It 45
45
. is a cornerstone of the Standard Model. there are important loop-induced couplings to massless gluons and photons [see Fig. Since the Higgs is a scalar particle. All couplings are proportional to the mass of the particle. In addition. How can one establish that it indeed is the Higgs? The Higgs boson can be distinguished from other scalar particles as they occur. – In theories with large extra dimensions new degrees of freedom occur. Spontaneously broken electroweak symmetry. 6. where ΛTC ∼ 1 TeV is the conﬁnement scale of a new non-Abelian gauge interaction. – A related idea regards the Higgs as a pseudo-Goldstone boson of some approximate global symmetry spontaneously broken at an energy scale above the electroweak scale. All such ideas can be tested at the LHC and the ILC. the decay into two photons might be the best possible detection channel given the large QCD background for the decay into two gluons at the LHC. which are tt or. in supersymmetric theories. 19(c)]. The investigation of the Higgs sector can be expected to to give important insight also on physics beyond the Standard Model. Such considerations have motivated various extensions of the Standard Model: – Supersymmetry retains an elementary scalar Higgs (and actually adds four more). for a light Higgs. since it is generated by the Higgs mechanism. The tree-level diagrams are given in Figs. MW are given by (230a) (230b)
GF m3 1 H √ . Hence the Higgs decays dominantly into the heaviest particles kinematically allowed. since the unitarity of WW scattering implies that the Standard Model Higgs and/or other effects related to electroweak symmetry breaking become manifest at energies below ∼ 1 TeV. by its special couplings to Standard Model particles. The tree-level decay widths in the approximation m H Γ H → ff = GF mH m2 f √ Nc . It also has a strong coupling ∝ m H to the longitudinal component of W and Z bosons. m H > 114 GeV. however. while radiative corrections with opposite signs from bosons and fermions cancel. 20. and it almost equals the Higgs mass at mH ∼ 1 TeV where the Higgs dynamics becomes non-perturbative. the decay into a pair of W bosons dominates. and an enormous ‘ﬁne-tuning’ of the tree-level mass term is needed to keep the Higgs light (this is usually referred to as the ‘naturalness problem’ of the Higgs sector). The Higgs mass is then related to the explicit breaking of this symmetry. 19(a) and 19(b). At the threshold the width increases by two orders of magnitude. 4π 2 mf .1 Higgs couplings and decay Suppose a resonance is found at the LHC with a mass above 114 GeV and zero charge. For a light Higgs with a mass just above the present experimental limit. its mass is subject to quadratically divergent quantum corrections. for instance.

is clearly an experimental challenge to establish the mass dependence of the Higgs couplings. [12]. From Ref. It increases by two orders of magnitude at the WW threshold.W. L ÜDELING
f H f
(a)
∝ mf
W L/ZL H
g/γ H g/γ
(c)
∝ mH
W L/ZL
(b)
H H H
(d)
∝ mH
Fig.
46
46
. The cubic Higgs self-coupling can be probed at the ILC and possibly at the LHC. 20: Left: Higgs branching ratios as function of the Higgs mass. Tree-level couplings are proportional to masses. Right: Higgs decay width as function of the Higgs mass. B UCHMÜLLER AND C. 19: Higgs boson decays. but there also are loopinduced decays into massless particles. so the true discovery of the Higgs is likely to take several years of LHC data!
Fig.

on the other hand. i. The mass bounds in the Standard Model arise from the scale dependence of couplings. that the loop corrections used to determine the Higgs mass strongly depend on the top mass as well. the Higgs self-coupling is given by the gauge couplings.F IELD THEORY AND THE S TANDARD M ODEL
6. . the gauge couplings empirically unify at the GUT scale. the Higgs mass is constrained to lie in a narrow region. well within the current uncertainties. if there are no new particles between ∼ 10 2 GeV and ΛGUT . 47
47
. m 2 = 2λv 2 is a free parameter H which cannot be predicted. ∼ 102 GeV. Fig. ht = h − 8gs + .e. (22a)]. one needs to satisfy the following two conditions in the range v < µ < Λ: – The triviality bound: λ(µ) < ∞. . There are. For a given top mass. and the electroweak vacuum would no longer be the ground state of the theory.e. 21. assumes that the Standard Model is valid up to Λ GUT . Before we present these arguments. second. we ﬁrst recall the experimental bounds: – The Higgs has not been seen at LEP. For increasing Λ. In the Minimal Supersymmetric Standard Model (MSSM). The result of a global ﬁt is shown in the blue-band plot. especially in the supersymmetric Standard Model. the Higgs potential would not be bounded from below any more.2
Higgs mass bounds
We now turn to the issue of the Higgs mass. two indications for such a ‘desert’ between the electroweak scale and the GUT scale: First. as explained in Section 4. this translates into an upper and lower bound on the Higgs mass. as can be seen by comparing the plots in Fig. (22b)]. can shift the Higgs mass best ﬁt by several tens of GeV. . the theory would be ‘trivial’. i. – The vacuum stability bound: λ(µ) > 0. and for the known top quark mass and Λ ∼ Λ GUT ∼ 1016 GeV. in particular for the ρ parameter. λ(µ) = βλ λ. Theoretical bounds on the Higgs mass arise. an impressive result! One should keep in mind. 2 ∂µ 2 t (4π) µ (231a) (231b)
These equations imply that ht decreases with increasing µ whereas the behaviour of λ(µ) depends on the initial condition λ(v). For the Standard Model to be a consistent theory from the electroweak scale v up to some highenergy cutoff Λ. 21. This gives a lower bound on the mass. – The Higgs contributes to radiative corrections. which implies the upper bound mH 135 GeV. the evidence for small neutrino masses is also consistent with an extrapolation to ΛGUT without new physics at intermediate scales. . If λ would hit the Landau pole at some scale µ L < Λ. however. given the fact that our present experimental knowledge ends at the electroweak scale. the scale of grand uniﬁcation. This might seem a bold extrapolation. ht = t ∂µ (4π)2 ht ∂ 9 2 2 µ ht (µ) = βλ λ.. from two consistency requirements: (Non-)Triviality and vacuum stability. The current 95% conﬁdence level upper bound is m H < 185 GeV. via the see-saw mechanism. m H > 114 GeV. however. . The renormalization group equations for the couplings λ(µ) and h t (µ) are 1 ∂ 12λ2 − 12h4 + .. These two requirements deﬁne allowed regions in the m H –mt plane as a function of the cutoff Λ [see Fig. . theoretical consistency arguments which yield stringent upper and lower bounds on the Higgs mass. If λ would become negative. which one obtains from the triviality and vacuum stability bounds. Other Yukawa couplings are much smaller and can be ignored. Within the Standard Model. a ﬁnite value λ(µL ) would require λ(v) = 0. on the Higgs mass. however. There are. 130 GeV < m H < 180 GeV [see Fig. the allowed region shrinks. even in the Standard Model. Hence precision measurements yield indirect constraints on the Higgs mass. Most relevant are the quartic Higgs self-coupling λ and the top quark Yukawa coupling h t which gives the top mass via mt = ht v. A shift of a few GeV in the top mass. The impressively narrow band of allowed Higgs masses.

00012 incl. Furthermore.02738±0. L ÜDELING
6
theory uncertainty
6
theory uncertainty
∆α(5) = had
0.02761±0.02761±0. A ‘good idea’ was the quark model. 21: The blue-band plot showing the constraints on the Higgs mass from precision measurements. It is very instructive to look at this process as the interplay of some ‘good ideas’ and some ‘misunderstandings’ which often prevented progress for many years. 2003 and 2005 (left to right). From Ref. But were quarks real or just some mathematical entities? Many physicists did not believe in quarks 48
48
.00036 0. proposed in 1964 independently by Gell-Mann and Zweig.02749±0. instead of a summary. B UCHMÜLLER AND C.02749±0.00012 incl.W. low Q2 data
∆χ2
3 2 1 0 Excluded 30 100 300
mH [GeV]
Fig.02761±0.02758±0.00020
6
theory uncertainty
∆α(5) = had
0.00036 0. [11].00035 0.00012
6 5 4
2
Theory uncertainty
∆α(5) = had
0. The hypothesis that hadrons are made out of three quarks and antiquarks allowed one to understand their quantum numbers and mass spectrum in terms of an approximate SU(3) ﬂavour symmetry. most notable due to shifts in the top mass and its uncertainty. The small plots show the same plot from winter conferences of different years: 1997. the deep-inelastic scattering experiments at SLAC in 1968 could be interpreted as elastic scattering of electrons off point-like partons inside the proton. 1. 2001.02747±0. low Q2 data
4
4
4
2
Without NuTeV
∆χ
2
2
∆χ
∆χ
2 Excluded 0 10
Preliminary
2
2
∆χ
0 Excluded 20
Preliminary
3 2 1
mH [GeV]
10
2
10
3
0
Excluded
Preliminary
mH [GeV]
10
2
100
400
0
Excluded 30
100
500
mH [GeV]
mH [GeV]
6 5 4
Theory uncertainty
∆α(5) = had
0. Weinberg [1]. The big plot dates from winter 2006. 7
History and outlook
Finally. and it was natural to identify these partons with quarks. the ‘eightfold way’.00036 0. The best ﬁt and the width of the parabola vary. we shall brieﬂy recall the history of ‘The making of the Standard Model’ following a review by S.

This led to the gauge group 49
49
. and one hoped to obtain in this way a theory of strong interactions with the ρ-mesons as gauge bosons. 22: Bounds on the Higgs and top mass from triviality and vacuum stability. Glashow. Then the Goldstone boson becomes the helicity-zero part of the gauge boson. after the V − A structure of the weak interactions had been identiﬁed. breaking explicitly the local gauge symmetry and thereby destroying the rationale for introducing non-Abelian local symmetries in the ﬁrst place. and in 1967 Weinberg still considered the chiral SU(2) L × SU(2)R symmetry of strong interactions to be a gauge theory with the ρ and a1 mesons as gauge bosons. neither in strong nor weak interactions. however. Only several years later. 3. Another ‘good idea’ was the invention of non-Abelian gauge theories by Yang and Mills in 1954. which thereby acquires a mass. [14]). [13]). eL ) and eR (he did not believe in quarks!). these new developments were applied to broken symmetries in strong interactions. Salam and Ward and others develop theories of the weak interactions with intermediate vector bosons. it was realized that non-Abelian gauge theories with mass terms would be nonrenormalizable.
since no particles with third integer charges were found despite many experimental searches. On the other hand. Panel (a) shows the combined bounds for different values of Λ (from Ref. 2. Panel (b) gives the bounds on the Higgs mass for the known top mass (from Ref. In the same year. But again. he then applied the idea of spontaneous symmetry breaking to the weak interactions of the leptons of the ﬁrst family. A further ‘good idea’ was spontaneous symmetry breaking: There can be symmetries of the Lagrangian that are not symmetries of the vacuum. Such mass terms had to be inserted by hand. plagued by the same divergences as the four-fermion theory of weak interactions. (νL .F IELD THEORY AND THE S TANDARD M ODEL
600 500
M H [GeV/c 2 ]
400 300 EW Precision
Triviality
200 100 0
EW vacuum is absolute minimum
3
5
7
9
11
Λ
13
15
17
19
log 10 Λ [GeV]
(a) (b)
Fig. did Bludman. Furthermore. there is no experimental evidence for any massless scalar with strong or weak interactions. According to the Goldstone theorem there must be a massless spinless particle for every spontaneously broken global symmetry. The local symmetry was the isospin group SU(2). In 1964 Higgs and Englert and Brout found a way to circumvent Goldstone’s theorem: The theorem does not apply if the symmetry is a gauge symmetry as in electrodynamics or the non-Abelian Yang–Mills theory. But all physical applications of non-Abelian gauge theories seemed to require massive vector bosons because no massless ones had been found.

γ0 = γ0 is Hermitian while the γi = −γi are anti-Hermitian.1
Vectors. These include grand uniﬁcation. and all γ µ are traceless. B UCHMÜLLER AND C. there are again some ‘misunderstandings’ among theorists. run from 0 to 3. (A. {γµ . purely spatial vectors are indicated by a vector arrow.
Appendix
A A.
Today. and derivatives with respect to x µ are denoted by ∂µ = ∂ = ∂xµ ∂ . there are also a number of ‘good ideas’ on the market. . and there are several common representations. gµν = diag(+. ν. So.W. Finally. τ.2 γ matrices
In four dimensions. 3 . γν } = 2gµν . dynamical symmetry breaking. The matrix form of the γ matrices is not ﬁxed by the algebra. spinors and γ algebra
Metric conventions
Our spacetime metric is mostly minus. .1)
so timelike vectors v µ have positive norm vµ v µ > 0.
(A.
µ = 0. massive W and Z bosons. and of QCD at LEP.
50
50
. . ∂t . supersymmetry and string theory. but we can soon hope for clariﬁcations from the results of the LHC. x) (with upper index). . Very likely. and the generation of hadron masses. . . The coordinate four-vector is x µ = (t. it was realized that the infrared properties of non-Abelian gauge theories lead to the conﬁnement of quarks and massless gluons. A. L ÜDELING
SU(2) × U(1). HERA and Tevatron. which lead beyond the Standard Model. by 1973 ‘The making of the Standard Model’ was completed! Since 1973 many important experiments have conﬁrmed that the Standard Model is indeed the correct theory of elementary particles: – – – – – 1973: discovery of neutral currents 1979: discovery of the gluon 1983: discovery of the W and Z bosons 1975–2000: discovery of the third family. (A. −) . −. −.2)
Greek indices µ. b. . t and ντ During the past decade impressive quantitative tests have been performed of the electroweak theory at LEP.3)
† † In addition. SLC and Tevatron. the γ matrices are deﬁned by their anticommutators.
Acknowledgements
We would like to thank the participants of the school for stimulating questions and the organizers for arranging an enjoyable and fruitful meeting in Kitzbühel. a massless photon and the Higgs boson! The next steps on the way to the Standard Model are well known: The proof by ’t Hooft and Veltman that non-Abelian gauge theories are renormalizable and the discovery of asymptotic freedom by Gross and Wilczek and Politzer. ρ.

⇒
γ5 =
In this basis. 1) T . γ5. one often needs to contract several γ matrices such as γ µ γµ = 4 γ γ γ γµ = 4g
µ ν ρ
(A. A.7c) etc.
γi =
. The product of all γ matrices is γ 5 = iγ 0 γ 1 γ 2 γ 3 which is Hermitian.
.
γ0 =
2
2
0
. the positive-energy spinor becomes E + pz 0 (A. Choosing the momentum along the z-axis and e. squares to one and anticommutes with all γ matrices. ξ = (1. the spinors u(p) and v(p) are given by us (p) =
2
Here ξ and η are two-component unit spinors. The chiral projectors PL/R are deﬁned as PL/R = 1 1 ± γ5 . and similar for η and the 2 negative energy spinors.F IELD THEORY AND THE S TANDARD M ODEL
like the Dirac and Weyl representations.
(A.7d)
νρ
γ µ γ ν γ ρ γ σ γµ = −2γ σ γ ρ γ ν / For a vector v µ we sometimes use the slash v = γ µ vµ . However. respectively. (A.
2 PL/R = PL/R .
51
51
.10) u+ = .
v s (p) =
E E
+ p · σ ξs s 2−p·σ ξ
E 2 + p · σ ηs − E 2 − p · σ ηs
0
0 σi i −σ 0
− 2 0
2
.6)
To evaluate Feynman diagrams like for the anomalous magnetic moment.7b) (A. the following identities hold regardless of the representation. (A. For ξ = (0.7a)
ν
γ γ γµ = −2γ
µ ν
(A.4)
(A. Eqs.5)
(A. Weyl and Majorana spinors
The solutions of the Dirac equation in momentum space are ﬁxed by the equations / p − m u(i) (p) = 0 / p + m v (i) (p) = 0 . 0)T . (133) and (58). 2 P L PR = P R PL = 0 . the spin is reversed.8)
Here it is convenient to choose the Weyl representation (58) of the Dirac matrices. E − pz 0 which has spin + 1 along the z-axis.g.
(A. γµ = 0 .3 Dirac.9)
.

web. New York. 46 (1998) 203 [arXiv:hep-ph/9705337]. Phys. Weyl or chiral spinors are subject to the constraint PL ψL = ψ L or P R ψR = ψ R (A. (A. Fortschr. [10] T. see these proceedings. V. [13] C. Weyl and Majorana spinors. Reading.W. B UCHMÜLLER AND C. Phys. MA. 157 (2006) 101. Suppl. but no usual mass term. J. which have only two degrees of freedom. E. Veltman. [14] C. like right-handed neutrinos. 18 (2003) 77 [arXiv:hep-ph/0307138]. there is the possibility of a Majorana mass term via the charge conjugate spinor ψ C : ψ C = Cψ T with the charge conjugation matrix C = iγ 0 γ 2 . see these proceedings. An Introduction to Quantum Field Theory (Addison-Wesley. Fleischer. There are two restricted classes of spinors. CERN report 73–9 (1973). Weinberg. Schroeder. Weinberg. chiral spinors correspond to sums u ± γ 5 v. Acta Phys.14)
ψ C is of opposite chirality to ψ. C 34 (2004) 5 [arXiv:hep-ph/0401010]. since
† (ψL ) = PL ψL = (PL ψL )† γ 0 = ψL PL γ 0 = ψ L PR
(A. this term violates all symmetries under which ψ is charged. Berlin. so it is acceptable only for complete singlets. Chiral spinors can have a kinetic term.
=0
(A. Lindner. Peskin and D.
References
[1] S. The making of the Standard Model. Phys.cern. [11] LEP Electroweak Working Group. [12] M. Spira. Nucl.or right-handed fermions. 1995). [9] G. Diagrammar. Ecker. Froggatt. spin up and spin down). Polon. Quigg. lecture given at this School (unpublished). D. B 30 (1999) 2145 [arXiv:hep-ph/9905369]. ’t Hooft and M. so it can be used to build a bilinear ψ C ψ for a mass term.ch. Proc. lecture given at this School. [8] S. [5] G. [6] R. 1990). Gravitation and Cosmology (Wiley. [7] M. pp. [2] O. L ÜDELING
The spinors considered so far are called Dirac spinors: They are restricted only by the Dirac equation and have four degrees of freedom (particle and antiparticle.12)
and hence ψL ψL = ψ L PR PL ψL = 0 . Kinoshita. Surveys High Energ. Elementary Particle Physics: Concepts and Phenomena (Springer. Eur.13)
However. 1972). [4] J. http://lepewwg.11)
and correspond to purely left. Ellis. 61–63. Nachtmann.
52
52
. In the language of u’s and v’s. [3] M. However. lecture given at this School. lecture given at this School (unpublished). Phys.

42 Gauge conditions. 6 Dirac algebra. 32 Magnetic moment. 4 Heisenberg picture. 10 Gauge boson self-interaction. 20 Landé factor. 23 Field strength. 25 Group generators. 3 Counterterms.F IELD THEORY AND THE S TANDARD M ODEL
Index
2 → 2 scattering in φ4 theory. 14 Electroweak theory W I and B bosons. 27 one-loop correction. 43 Feynman parameters. 22 QCD. 21 Pauli matrices for SU(2). 19 Forward–backward asymmetries. 4 for creation and annihilation operators. 10 Antiparticle. 22 Gauge transformation. 24 anomalous. 37 quantum numbers. 14 Bare ﬁelds. 17 Momentum operator. 12 for spinor ﬁelds. 7 Naturalness problem. 18 Gordon identity. 24 Weyl representation. 3 Lagrangian non-Abelian gauge ﬁeld. 45 53
53
. 33 isospin and hypercharge currents. 9 in quantum mechanics. 40 γ matrices. 6 Lagrange density. 39 Mass dimension. 26 Feynman propagator. 37 gauge group. 15 Adjoint spinor. 4 Higgs mechanism. 36 potential. 19 Creation and annihilation operators. 21 Gell-Mann matrices for SU(3). 28 Maxwell’s equations. 30 β function. 39 Compton wavelength. 9 for raising and lowering operators. 36 Hilbert space for the spinor ﬁeld. 12 of the harmonic oscillator. 22 QED. 10 Disconnected diagrams. 10 Dirac representation. 21 Hamiltonian. 5 Asymptotic freedom. 27 Majorana mass. 34 charged current. 10 Dirac equation. 34 Faddeev–Popov ghosts. 14 counterterms. 18 transformation of. 32 for QCD coupling. 30 Covariant derivative. 13 Feynman rules for φ4 theory. 4 of the scalar ﬁeld. 18 Mexican hat potential. 23 Fermi constant GF . 30 for fermions. 30 Bare parameters. 24 Landau pole. 9 Lagrange function. 11 Canonical commutation relations for p and q. 4 Charge operator. 11 for scalar ﬁelds. 8 for fermions. 34 neutral current. 6 for ﬁeld operators. 33 Canonical anticommutation relations for creation and annihilation operators. 7 CKM matrix. 23 Gauge potential electromagnetic. 3 Harmonic oscillator. 33 Asymptotic states. 16 for non-Abelian gauge theories. 6 Conjugate momentum for fermions.

40 Regularization. 13 Polarization vector. 20 Θ function.W. 11 Uncertainty relation. 12 Spinors. 5 Vertex correction. 10 Spontaneous symmetry breaking. 36 Vacuum. 9 Triangle diagrams. 17 Noether’s theorem. 38 Structure constants. 31 S matrix. 25 Pauli principle. L ÜDELING
Noether current. 31 Weak mixing angle. 14 See-saw mechanism. 35 u and v spinors. 21 SU(n). 31 Renormalized ﬁelds. 29 Spin operator. 27 constants. 9 Pauli equation. 31 group equation. 41 Running coupling. 4 Unitary gauge. 32 schemes. 29 photon (vacuum polarisation). 13 φ4 theory. B UCHMÜLLER AND C. 39 Self-energy electron. 25 Ward identity. 37
54
54
. 28 Renormalization. 23 R ratios. 27 dimensional. 30 ρ parameter.