# QUANTUM FIELD THEORY

P.J. Mulders Department of Theoretical Physics, Department of Physics and Astronomy, Faculty of Sciences, VU University, 1081 HV Amsterdam, the Netherlands E-mail: mulders@few.vu.nl

November 2011 (version 6.04)

Contents

1 Introduction 1.1 Quantum ﬁeld theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.2 Units . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.3 Conventions for vectors and tensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Relativistic wave equations 2.1 The Klein-Gordon equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2 Mode expansion of solutions of the KG equation . . . . . . . . . . . . . . . . . . . . . 2.3 Symmetries of the Klein-Gordon equation . . . . . . . . . . . . . . . . . . . . . . . . . 3 Groups and their representations 3.1 The rotation group and SU (2) . . . . 3.2 Representations of symmetry groups . 3.3 The Lorentz group . . . . . . . . . . . 3.4 The generators of the Poincar´ group . e 3.5 Representations of the Poincar´ group e 4 The 4.1 4.2 4.3 4.4 4.5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1 2 4 8 8 9 10 13 13 15 18 20 21 27 27 29 31 34 36 39 39 39 41 44 44 46 48 49 50 54 54 56 57 60

Dirac equation The Lorentz group and SL(2, C) . . . . . . . . . . . . . Spin 1/2 representations of the Lorentz group . . . . . . General representations of γ matrices and Dirac spinors Plane wave solutions . . . . . . . . . . . . . . . . . . . . γ gymnastics and applications . . . . . . . . . . . . . . .

5 Vector ﬁelds and Maxwell equations 5.1 Fields for spin 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2 The electromagnetic ﬁeld . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.3 The electromagnetic ﬁeld and topology∗ . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Classical lagrangian ﬁeld theory 6.1 Euler-Lagrange equations . . . . . . . . . . . 6.2 Lagrangians for spin 0, 1/2 and 1 ﬁelds . . . 6.3 Symmetries and conserved (Noether) currents 6.4 Space-time symmetries . . . . . . . . . . . . . 6.5 (Abelian) gauge theories . . . . . . . . . . . . 7 Quantization of ﬁelds 7.1 Canonical quantization . . . . . . . . 7.2 Creation and annihilation operators 7.3 The real scalar ﬁeld . . . . . . . . . 7.4 The complex scalar ﬁeld . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1

2 7.5 7.6 The Dirac ﬁeld . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The electromagnetic ﬁeld . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60 62 66 66 67 68 69 70 73 73 75 78 79 81 82

8 Discrete symmetries 8.1 Parity . . . . . . . . . 8.2 Charge conjugation . . 8.3 Time reversal . . . . . 8.4 Bi-linear combinations 8.5 Form factors . . . . .

9 Path integrals and quantum mechanics 9.1 Time evolution as path integral . . . . . . . . . . . . . 9.2 Functional integrals . . . . . . . . . . . . . . . . . . . 9.3 Time ordered products of operators and path integrals 9.4 An application: time-dependent perturbation theory . 9.5 The generating functional for time ordered products . 9.6 Euclidean formulation . . . . . . . . . . . . . . . . . . 10 Feynman diagrams for scattering amplitudes 10.1 Generating functionals for free scalar ﬁelds . . . . 10.2 Generating functionals for interacting scalar ﬁelds 10.3 Interactions and the S-matrix . . . . . . . . . . . . 10.4 Feynman rules . . . . . . . . . . . . . . . . . . . . 10.5 Some examples . . . . . . . . . . . . . . . . . . . . 11 Scattering theory 11.1 kinematics in scattering processes 11.2 Crossing symmetry . . . . . . . . 11.3 Cross sections and lifetimes . . . 11.4 Unitarity condition . . . . . . . . 11.5 Unstable particles . . . . . . . . 12 The 12.1 12.2 12.3 12.4 12.5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

85 . 85 . 89 . 92 . 95 . 100 . . . . . . . . . . 103 103 106 107 109 111 114 114 118 122 123 128

standard model Non-abelian gauge theories . . . . . . . . . . . . . . . Spontaneous symmetry breaking . . . . . . . . . . . . The Higgs mechanism . . . . . . . . . . . . . . . . . . The standard model SU (2)W ⊗ U (1)Y . . . . . . . . . Family mixing in the Higgs sector and neutrino masses

1 4. I refer to the book of Srednicki [1] and Ryder [2]. 1994. Schroeder. L. 5.4
2.1 6.2 3. Cambridge University Press. Quantum Field Theory. 3. Vol.3 2. Other text books of Quantum Field Theory that are useful are given in refs [3-6]. Vol.3 2. Cambridge University Press. Peskin and D. M.1 1. Ryder. McGraw-Hill. Quantum Field Theory. Cambridge University Press.2 2. Veltman.2 6.1 3.E. An introduction to Quantum Field Theory. M. but are rather compact.4 3.
Corresponding chapters in books of Ryder.H. The schedule in the Fall of 2011 is Chapters 1 through 10 in period 2 (7 weeks in November and December 2011) and Chapters 11 and 12 in period 3 (January 2012).2 3.4 2. 4.2 36 22
.7 2.5 4.3 3. 1980. 2. 1996. Weinberg. Itzykson and J.2 4.2 3.3 2. I: Foundations.2 2. 6.2 6.3 2.3.1. Cambridge University Press.2 3.1 5.3 3.3 3. 2007. M.3 6.4 3. The quantum theory of ﬁelds. The notes contain all essential information.2 1. These notes Section 1. Addison-Wesly. 1985.8 3.1 2. 3.V.
1.1 2.2.2
2.5 2.1 3. Quantum Field Theory.2 3.3
3. 1995.4 3. 2.5 Ryder Peskin & Schroeder Srednicki 1 1 1 1 3
2.3 3.3 References As most directly related books to these notes. Zuber. Diagrammatica.-B. Peskin & Schroeder and Srednicki.3 4. 3. S. II: Modern Applications.4 6.5 2.5 5.4 4. Cambridge University Press. C.3 2. Srednicki.1
2 2
3. 1995.

5 6.6 4.3 8.5
6 7 6
6. 6.3 7.5 11.1 12. 8.1 20.4 3.5 7. 9. 9.4.3 10.3.1. 20.7 4.2. 6.3.5 9.3 8.2 4.2.1.5 4.3 6.5
.4 4.1 8.2.7 5.1 20.1.6 6.2 11.2 12.4 12.6 8.4 These notes Section 7.5 9.5 7.3 9. 6.2. 6.6 3. 2.4.2 8.3 12.2 5.1 5. 4.1 7.4 9.2 8. 6.8 6.6 3.5.2 7. 2.3 6 8 9 Srednicki 3 3 3 37
5.4 10.4 12.5 Ryder 4 4 4 4 4 4 Peskin & Schroeder 2.3 20.2 10.3 3.1 11.4 2.7.1 10.10
8.3 11.3 2.8 5 4.1 20.2 9. 4.6 10.3.4 8.4.5 4.3 7.1 9. 4.6 3.1 9.6 9.4 7.5 2. 4.

• Slight rewriting of discussion of Majorana mass term in section 6.2
.3 • First two sentences of Chapter 5 have been modiﬁed.5 Corrections 2011/2012) • Extension of Exercise 3.

Corresponding position and momentum operators do not commute. In this coordinate representation the position operators r op simply acts as rop ψ(r) = r ψ(r).1 Quantum ﬁeld theory
In quantum ﬁeld theory the theories of quantum mechanics and special relativity are united. In special relativity a special role is played by the velocity of light c.054 571 68 (18) × 10−34 J s = 6.Chapter 1
Introduction
1. the action of the momentum operator in the coordinate representation is not as simple as the position operator. In quantum mechanics a special role is played by Planck’s constant h. the position can in principle be determined at any time with any accuracy. c = 299 792 458 m s−1 . ˜ ψ(p) = i d3 r exp − p · r ψ(r). ˜ The corresponding states are |p and the wave functions ψ(p) = p||ψ are the Fourier transforms of the coordinate space wave functions. ≡ h/2π = 1. (1. being eigenvalues of the position operators. They satisfy the well-known (canonical) operator commutation relations [ri .3)
The uncertainty principle tells us that in this representation the momenta cannot be fully determined. S ≫ . (1.1)
In the limit that the action S is much larger than .4)
where δij is the Kronecker δ function. It is given by pop ψ(r) = −i ∇ψ(r). Indeed.2) In the limit that v ≪ c one reaches the non-relativistic domain. 1 (1. One can talk about states |r and the wave function ψ(r) = r||ψ . (1.5)
One can also choose a representation in which the momenta of the particles are the dynamical variables. In quantum mechanics.6)
. In the framework of classical mechanics as well as quantum mechanics the position of a particle is a well-deﬁned concept and the position coordinates can be used as dynamical variables in the description of the particles and their interactions. quantum eﬀects do not play a role anymore and one is in the classical domain. usually given divided by 2π.582 119 15 (56) × 10−22 MeV s.
(1. pj ] = i δij . (1.

Introduction

2

and ψ(r) =

d3 p exp (2π )3

i

˜ p · r ψ(p).

(1.7)

The existence of a limiting velocity, however, leads to new fundamental limitations on the possible measurements of physical quantities. Let us consider the measurement of the position of a particle. This position cannot be measured with inﬁnite precision. Any device that wants to locate the position of say a particle within an interval ∆x will contain momentum components p ∝ /∆x. Therefore if we want ∆x ≤ /mc (where m is the rest mass of the particle), momenta of the order p ∝ mc and energies of the order E ∝ mc2 are involved. It is then possible to create a particle - antiparticle pair and it is no longer clear of which particle we are measuring the position. As a result, we ﬁnd that the original particle cannot be located better than within a distance /mc, its Compton wavelength, ∆x ≥ mc . (1.8)

For a moving particle mc2 → E (or by considering the Lorentz contraction of length) one has ∆x ≥ c/E. If the particle momentum becomes relativistic, one has E ≈ pc and ∆x ≥ /p, which says that a particle cannot be located better than its de Broglie wavelength. Thus the coordinates of a particle cannot act as dynamical variables (since these must have a precise meaning). Some consequences are that only in cases where we restrict ourselves to distances ≫ /mc, the concept of a wave function becomes a meaningful (albeit approximate) concept. For a massless particle one gets ∆x ≫ /p = λ/2π, i.e. the coordinates of a photon only become meaningful in cases where the typical dimensions are much larger than the wavelength. For the momentum or energy of a particle we know that in a ﬁnite time ∆t, the energy uncertainty is given by ∆E ≥ /∆t. This implies that the momenta of particles can only be measured exactly when one has an inﬁnite time available. For a particle in interaction, the momentum changes with time and a measurement over a long time interval is meaningless. The only case in which the momentum of a particle can be measured exactly is when the particle is free and stable against decay. In this case the momentum is conserved and one can let ∆t become inﬁnitely large. The result thus is that the only observable quantities that can serve as dynamical coordinates are the momenta (and further the internal degrees of freedom like polarizations, . . . ) of free particles. These are the particles in the initial and ﬁnal state of a scattering process. The theory will not give an observable meaning to the time dependence of interaction processes. The description of such a process as occurring in the course of time is just as unreal as classical paths are in non-relativistic quantum mechanics. The main problem in Quantum Field Theory is to determine the probability amplitudes between well-deﬁned initial and ﬁnal states of a system of free particles. The set of such amplitudes p′ , p′ ; out|p1 , p2 ; in ≡ p′ , p′ ; in|S|p1 , p2 ; in determines the scattering matrix or S-matrix. 1 2 2 1 Another point that needs to be emphasized is the meaning of particle in the above context. Actually, the better name might be ’degree of freedom’. If the energy is low enough to avoid excitation of internal degrees of freedom, an atom is a perfect example of a particle. In fact, it is the behavior under Poincar´ transformations or in the limit v ≪ c Gallilei transformations that determine the description e of a particle state, in particular the free particle state.

1.2

Units

It is important to choose an appropriate set of units when one considers a speciﬁc problem, because physical sizes and magnitudes only acquire a meaning when they are considered in relation to each other. This is true speciﬁcally for the domain of atomic, nuclear and high energy physics, where the typical numbers are diﬃcult to conceive on a macroscopic scale. They are governed by a few fundamental units and constants, which have been discussed in the previous section, namely and

Introduction

3

c. By making use of these fundamental constants, we can work with less units. For instance, the quantity c is used to deﬁne the meter. We could as well have set c = 1. This would mean that one of the two units, meter or second, is eliminated, e.g. given a length l the quantity l/c has the dimension of time and one ﬁnds 1 m = 0.33 × 10−8 s or eliminating the second one would use that, given a time t, the quantity ct has dimension of length and hence 1 s = 3 × 108 m. Table 1.1: Physical quantities and their canonical dimensions d, determining units (energy)d . quantity time t length l energy E momentum p angular momentum ℓ mass m area A force F charge (squared) e2 Newton’s constant GN velocity v quantity with dimension energyd t/ l/( c) E pc ℓ/ mc2 A/( c)2 F c α = e2 /4πǫ0 c GN /( c5 ) v/c dimension (energy)−1 (energy)−1 (energy)1 (energy)1 (energy)0 (energy)1 (energy)−2 (energy)2 (energy)0 (energy)−2 (energy)0 canonical dimension d -1 -1 1 1 0 1 -2 2 0 -2 0

In ﬁeld theory, it turns out to be convenient to work with units such that and c are set to one. All length, time and energy or mass units then can be expressed in one unit and powers thereof, for which one can use energy (see table 1.1). The elementary unit that is most relevant depends on the domain of applications, e.g. the eV for atomic physics, the MeV or GeV for nuclear physics and the GeV or TeV for high energy physics. To convert to other units of length or time we use appropriate combinations of and c, e.g. for lengths c = 0.197 326 968 (17) GeV fm or for order of magnitude estimates c ≈ 0.2 GeV fm = 200 eV nm, implying (when 1 fm = 10−15 m ≈ 5 GeV−1 . For areas, e.g. cross sections, one needs

2 2

(1.9) = c = 1) that (1.10)

c

= 0.389 379 323 (67) GeV2 mbarn

(1 barn = 10−28 m2 = 102 fm2 ). For times one needs = 6.582 119 15 (56) × 10−22 MeV s, (1.11)

implying (when = c = 1) that 1 s ≈ 1.5 × 1024 GeV−1 . Depending on the speciﬁc situation, of course masses come in that one needs to know or look up, e.g. those of the electron or proton, me mp = 9.109 382 6 (16) × 10−31 kg = 0.510 998 918 (44) MeV/c , = 1.672 621 71 (29) × 10

−27 2

(1.12) (1.13)

kg = 0.938 272 029 (80) GeV/c .

2

Furthermore one encounters the strength of the various interactions. In some cases like the electromagnetic and strong interactions, these can be written as dimensionless quantities, e.g. for electromagnetism the ﬁne structure constant α= e2 = 1/137.035 999 11 (46). 4π ǫ0 c (1.14)

Introduction

4

For weak interactions and gravity one has quantities with a dimension, e.g. for gravity Newton’s constant, GN = 6.708 7 (10) × 10−39 GeV−2 . (1.15) c5 By putting this quantity equal to 1, one can also eliminate the last dimension. All masses, lengths and energies are compared with the Planck mass or length (see exercises). Having many particles, the concept of temperature becomes relevant. A relation with energy is established via the average energy of a particle being of the order of kT , with the Boltzmann constant given by k = 1.380 650 5(24) × 10−23 J/K = 8.617 343(15) × 10−5 eV/K. (1.16)

Quantities that do not contain or c are classical quantities, e.g. the mass of the electron me . Quantities that contain only are expected to play a role in non-relativistic quantum mechanics, e.g. the Bohr radius, a∞ = 4πǫ0 2 /me e2 or the Bohr magneton µe = e /2me . Quantities that only contain c occur in classical relativity, e.g. the electron rest energy me c2 and the classical electron radius re = e2 /4πǫ0 me c2 . Quantities that contain both and c play a role in relativistic quantum mechanics, e.g. the electron Compton wavelength −e = /me c. It remains useful, however, to use λ and c to simplify the calculation of quantities.

1.3

**Conventions for vectors and tensors
**

3

**We start with vectors in Euclidean 3-space E(3). A vector x can be expanded with respect to a basis ei (i = 1, 2, 3 or i = x, y, z), ˆ x=
**

i=1

xi ei = xi ei , ˆ ˆ

(1.17)

to get the three components of a vector, xi . When a repeated index appears, such as on the right hand side of this equation, summation over this index is assumed (Einstein summation convention). Choosing an orthonormal basis, the metric in E(3) is given by ei · ej = δij , where the Kronecker delta ˆ ˆ is given by 1 if i = j δij = . (1.18) 0 if i = j, The inner product of two vectors is given by x · y = xi yi ei · ej = xi yj δij = xi yi . ˆ ˆ (1.19)

The inner product of a vector with itself gives its length squared. A vector can be rotated, x′ = Rx or x′ = Rij xj leading to a new vector with diﬀerent components. Actually, rotations are those real, i linear transformations that do not change the length of a vector. Tensors of rank n are objects with n components that transform according to Ti′1 ...in = Ri1 j1 . . . Rin jn Tj1 ...jn . A vector is a tensor of rank 1. The inner product of two vectors is a rank 0 tensor or scalar. The Kronecker delta is a constant rank-2 tensor. It is an invariant tensor that does not change under rotations. The only other invariant constant tensor in E(3) is the Levi-Civita tensor if ijk is an even permutation of 123 1 −1 if ijk is an odd permutation of 123 ǫijk = (1.20) 0 otherwise.

in this case an inertial system exists in which the two points are at the same time and ds2 just represents minus the spatial distance squred ds2 = −dx2 . The lower indices are constructed in the following way.22) (1. (1. .2.21) (1. such as eigentime τ for the position vector τ 2 ≡ x2 = t2 − x2 . One can distinguish: • ds2 > 0 (timelike intervals).Introduction
5
that can be used in the cross product of two vectors z = x × y. The real. ˆ ˆ (1. x3 ) = (t. The transformations that leave invariant the distance ds2 = dt2 − (dx2 + dy 2 + dz 2 ) between two points are called inhomogeneous Lorentz transformations or Poincar´ transformations. in this case an inertial system exists in which the two points are at the same space point and in that frame ds2 just represents the time diﬀerence ds2 = dt2 . For four-vectors in Minkowski space we will use the notation with upper indices and write x = (t. No diﬀerence is made between upper or lower. The scalar product of two diﬀerent vectors x and y is denoted x · y = xµ y ν gµν = xµ yµ = xµ y µ = x0 y 0 − x · y. Vectors are denoted x = xµ nµ .. x1 . linear transformations that do not change the length of a four-vector are called the Lorentz transformations. the invariant length or distance (squared) is not positive deﬁnite.24) The quantity gµν ≡ nµ · nν is the metric tensor.µn = Λµ1ν1 . Λµn n T ν1 .1. These transformations do change the components of a vector.25)
Within Minkowski space the real. one must distinguish tensors with upper or lower indices and one can have mixed tensors.26) (1.23)
δjm δkn − δjn δkm . denoted as V ′µ = Λµν V ν . Useful relations are ǫijk ǫℓmn ǫijk ǫimn ǫijk ǫijl = = = δiℓ δjℓ δkℓ 2 δkl . Relations relating tensor expressions. In Minkowski space. given by g00 = −g11 = −g22 = −g33 = 1 (the ˆ ˆ other components are zero). x3 ). Because of the diﬀerent signs occurring in gµν .νn can be conν structed. The (invariant) lengths often have special names.. xµ = gµν xν . x1 . and are given by (x0 . where the coordinate t = x0 is referred to as the time component. however. . xi are the three space components.3). linear transformations that leave the length of a vector invariant are called (homogeneous) Lorentz transformations.
We note that for Euclidean spaces (with a positive deﬁnite metric) vectors and tensors there is only one type of indices. are
. (1. −x). So we could have used all upper indices in the above equations. Unlike in Euclidean space. The invariant distance between two points x and y in Minkowski space is determined from the length dsµ = (x − y)µ . The length (squared) of a vector is obtained from the scalar ˆ product. we will use upper indices for the three-vectors. x2 . x) = (x0 .. In special relativity we start with a four-dimensional real vector space E(1.3) with basis nµ (µ = ˆ 0. When 3-dimensional space is considered as part of Minkowski space. The Poincar´ transformations include e e Lorentz transformations and translations. the points lie on the lightcone and they can be connected by a light signal. x2 = x · x = xµ xν nµ · nν = xµ xν gµν . One has x2 = xµ xµ = t2 − x2 . Many other four vectors and tensors transforming like T ′ µ1 . independent of a coordinate system. in which case zi = ǫijk xj yk . • ds2 < 0 (spacelike intervals). x2 .. δim δjm δkm δin δjn δkn . • ds2 = 0 (lightlike or null intervals). it is convenient to distinguish lower indices from upper indices.

The product of two epsilon tensors is given by g µµ ′ g νµ ′ g ρµ ′ g σµ g νν ′ g ρν ′ g σν
′ ′ ′
ǫµνρσ ǫµ ν
′
′ ′
ρ σ′
= −
g µν ′ g νν ′ g ρν ′ g σν g νρ ′ g ρρ ′ g σρ
′ ′
′
g µρ ′ g νρ ′ g ρρ ′ g σρ g νσ ′ g ρσ ′ g σσ
′ ′
′
g µσ ′ g νσ ′ g ρσ ′ g σσ . Examples are the scalar products above but also relations like pµ = m dxµ /dτ for the momentum four vector.
.35)
ρ ǫµνρσ ǫµν σ σ ǫµνρσ ǫµνρ
′
′
= −2 g ρρ g σσ − g ρσ g σρ = −6g σσ . . but more important has the same eﬀect on lefthandside and righthandside. (1.36)
Exercises
Exercise 1.31)
ǫµνρσ ǫµν
′ ′
ρσ
′
= −
(1. ∂1 .
=
∂ . explicitly written as (p0 .
′
′
.34) (1. (1. . ∂t ∂x ∂y ∂z
ν ∂µ xν = gµ . for instance. Each of these permutations leads to a minus sign. is easily proven for ǫ ǫ from which the general case can be obtained by making permutations of indices on the lefthandside and permutations of rows or columns on the righthandside.27)
It is easy to convince oneself of the nature of the indices in the above equation.and righthandside a four vector because τ is a scalar quantity! The equation with t = x0 instead of τ simply would not make sense! The momentum four vector.∇ . It is an invariant tensor. gµ = gµρ g ρν and 0 1 2 3 is in essence a ’Kronecker delta’. p). constructed via the metric tensor itself.33) (1. The length squared of ∂ is the d’Alembertian operator. (1. The derivative ∂µ is deﬁned ∂µ = ∂/∂xµ and we have a four vector ∂ with components (∂0 . a∞ = 4πǫ0 2 /me e2 .32) .28)
ν ν Note that gµ with one upper and lower index. p) = (E.1
(a) In the the Hydrogen atom (quantum system) the scale is set by the Bohr radius. because one has (1. ∂t
(1. g0 = g1 = g2 = g3 = 1.
(1. Note that in this equation one has on left. For the contraction of a vector with the antisymmetric tensor one often uses the shorthand notation ǫABCD = ǫµνρσ Aµ Bν Cρ Dσ . (1. starting from ǫ0123 = 1. Relate this quantity to the electron Compton wavelength −e via the dimensionless λ ﬁne structure constant α. is timelike with invariant length (squared) p2 = p · p = pµ pµ = E 2 − p2 = m2 . deﬁned by ∂2 − ∇2 .Introduction
6
called covariant.30)
(Note that there are diﬀerent conventions around and sometimes the opposite sign is used). not aﬀected by Lorentz transformations. ∂2 .
0123 0123
′
′
ǫ
µνρσ
ǫµνρσ
The ﬁrst identity.29) 2 = ∂ µ ∂µ = ∂t2 The value of the antisymmetric tensor ǫµνρσ is determined in the same way as for ǫijk . = −24. ∂3 ) = ∂ ∂ ∂ ∂ . where m is called the mass of the system.

ˆ ˆ
. g+− and g−+ . ǫ0 .3. 3} the index α can only be equal to one of the indices in ǫµνρσ . n+ . 0. a1 . 0. c. Also for the coordinates [a− . α and me c2 = 0.2
Prove the identity A × (B × C) = (A · C) B . . 0. a+ . 0. 0. This demonstrates how a careful use of units can save a lot of work. re = e2 /4πǫ0 me c2 to the Compton wavelength. n2 .4
Lightcone coordinates for a four vector a (which we will denote with square brackets as [a− . which is the Compton wavelength for the Planck mass. [Note: what is the MKS unit for V/T?]
Exercise 1. a3 ) are the expansion coeﬃcients using the basis vectors n0 . ρ. 2. σ} is a permutation of {0.(A · B) C using the properties of the tensor ǫijk given in section 1. ]. a+ . 0.Introduction
7
(b) Relate the classical radius of the electron (a relativistic concept). a1 . 1. 1. a2 . 0) ˆ n2 = (0. One does not need to know . n3 . (b) The coordinates (a0 .511 MeV. 1) ˆ . Use that to give the ˆ ˆµ ˆ components of the basis vectors n+ and n− . n2 . . me . Note that ˆ ˆ ˆ ˆ a · nα = aα and derive from that the components nα of the four vector nα . by a simple few-line reasoning [For instance: If {µ. n1 . e.3
Prove the following relation ǫµνρσ g αβ = ǫανρσ g µβ + ǫµαρσ g νβ + ǫµνασ g ρβ + ǫµνρα g σβ . How are a+ and a− related to a+ and a− . 0) ˆ n3 = (0. (e) Note that /m = − c and use this to calculate in a simple way (avoiding putting in the value λ of e) the Bohr magneton µe = e /2me and the nuclear magneton µp = e /2mp in electronvolts per Tesla (eV/T).71 × 10−39 GeV−2 to construct a mass Mpl (Planck mass).13 to give its actual value in kg. a1 . 1. n1 . aT ]) are deﬁned through √ a± ≡ (a0 ± a3 )/ 2.
Exercise 1. . a+ . ˆ ˆ ˆ ˆ These are n1 = (0. 1. a2 ] or [a− . but only appropriate combinations. 0) ˆ n0 = (1. ν. Compare it with the proton mass and use Eq. (c) Calculate the Compton wavelength of the electron and the quantities under (a) and (b) using the value of c. Also construct and calculate the Planck length Lpl . a2 ] we can ﬁnd basis vectors n− . (d) Use the value of the gravitational constant GN / c5 = 6. g−− . (a) Express the scalar product a · b in lightcone coordinates and deduce from this the values of g++ .
Exercise 1.

we just want to play a bit with covariant equations and study their behavior under Lorentz transformations. t) = 0. ∂t 2M (2. namely plane waves characterized by a wave number k. But the replacement 2. p2 = pµ pµ = E 2 − p2 = M 2 .6)
The energy spectrum is not bounded from below: considering the above stationary plane wave solutions. written as pµ −→ i∂µ is covariant (the same in every frame of reference).3 are not covariant. t). ∂t2 (2. (2. (2. 2 + M 2 φ(r.3)
Equations 2. i ∂ ∇2 ψ(r. t) = ∂2 − ∇2 + M 2 φ(r. Substitution of operators gives the Klein-Gordon (KG) equation for a real or complex function φ. E = p2 .7) k 0 = ± k2 + M 2 = ±Ek . φk (r. one obtains (2.
Although it is straightforward to ﬁnd the solutions of this equation. (2. t) = exp(−i k 0 t + i k · r). 2M (2.2. 8
.1 and 2. The Schr¨dinger equation in quantum mechanics is the operator equation o corresponding to the non-relativistic expression for the energy. the interpretation of this equation as a single-particle equation in which φ is a complex wave function poses problems because the energy spectrum is not bounded from below and the probability is not positive deﬁnite. Thus a covariant equation can be obtained by starting with the (covariant) equation for the invariant length of the four vector (E. ∂t p −→ pop = −i∇. t) = − ψ(r.1)
under the substitution (in coordinate representation) E −→ Eop = i ∂ .1 The Klein-Gordon equation
In this chapter.2)
Acting on the wave function one ﬁnds for a free particle. p).Chapter 2
Relativistic wave equations
2.4)
where M is the particle mass.5)
with (k 0 )2 = k2 + M 2 .

11)
Therefore.5). −i φ∗ ∇ φ . however. k) ≡ a(k) and φ(−Ek . This leads to the existence of negative densities. The appropriate current corresponding to the KG equation (see Excercise 2. φ(x) = d4 k ˜ 2π δ(k 2 − M 2 ) e−i k·x φ(k) (2π)4 (2. there are solutions with negative energy. φ(x) = d3 k (2π)3 2Ek ˜ ˜ e−i k·x φ(Ek .e. For this we note that an arbitrary solution for the ﬁeld φ can always be written as a superposition of plane waves. ∂µ j µ = 0. is a second order equation and φ and ∂φ/∂t can be ﬁxed arbitrarily at a given time.
∂ρ = −∇ · j. rather than as a wave function. for which we need the interpretation of φ itself as an operator.13)
˜ with (in principle complex) coeﬃcients φ(k). (2.10) ∂t which follows directly from the Schr¨dinger equation. This operator has all possible solutions in it multiplied with creation (and annihilation) operators.3). At that point the dependence on position r and time t is just a dependence on numbers/parameters on which the operator depends. The KG equation. Probability is not positive: in quantum mechanics one has the probability and probability current ρ j = = ψ∗ ψ ↔ i i − (ψ ∗ ∇ψ − (∇ψ ∗ )ψ) ≡ − ψ ∗ ∇ ψ. As we will see later both problems are related and have to do with the existence of particles and antiparticles. j) of the four-current j. Then. (2.
↔
↔
(2. 2M 2M (2. we want the most general solution. (2π)3 2Ek (2. having the KG equation as a space-time symmetric (classical) equation.2) is j µ = i φ∗ ∂ µ φ or (ρ.Relativistic wave equations
9
i. but rather the zero component of a four vector.14)
˜ ˜ Introducing φ(Ek .
2. 2. relativistically the density is not a scalar quantity. just as the dependence on time was in ordinary quantum mechanics. k) + ei k·x φ(−Ek . it is simply a matter of being careful to ﬁnd a consistent (covariant) theory.15)
.2
Mode expansion of solutions of the KG equation
Before quantizing ﬁelds. It is possible to rewrite it as an integration over positive energies only but this gives two terms (use the result of exercise 2. −k) . −k) ≡ b∗ (k) one has φ(x) = d3 k e−i k·x a(k) + ei k·x b∗ (k) = φ+ (x) + φ− (x). And. The integration over k-modes clearly is covariant and restricted to the ‘mass’-shell (as required by Eq.9)
They satisfy the continuity equation. there are no longer fundamental objections to mix up space and time. This continuity equation can be written down o covariantly using the components (ρ. (2. which is what Lorentz transformations do.12)
It is easy to see that this current is conserved if φ (and φ∗ ) satisfy the KG equation.8) (2. j) =
↔
i φ∗ ∂0 φ.

3
Symmetries of the Klein-Gordon equation
We arrived at the Klein-Gordon equation by constructing a covariant operator (∂µ ∂ µ + M 2 ) acting on a complex function φ. Since a · b = a · ˜ it is easy to see that ˜ b. i. (2. which implies (xµ ) = (t. (2π)3 2Ek (2. Another symmetry is found by complex conjugating the KG equation. x ∂µ ∂ µ + M 2 φ(˜) = 0.e. (2.18) (2.15 one has elimated k 0 and in both equations k · x = Ek t − k · x.20)
one has for the mode amplitudes aP (k) = a(−k) and bP (k) = b(−k).
2.14 and 2.6 We will explicitly discuss the example of a discrete symmetry. The coeﬃcients a(k) and b∗ (k) are the amplitudes of the two independent solutions (two. one thus must have that the function φ → φ′ such that φ′ (x′ ) = φ(x) or φ′ (x) = φ(Λ−1 x). x) → (t.Relativistic wave equations
10
In Eqs 2. This shows how parity transforms k-modes into −k modes. Performing some Lorentz transformation x → x′ = Λx. x Transforming everywhere in the KG equation x → x one obtains ˜ ˜ ˜ x ∂µ ∂ µ + M 2 φ(˜) = 0. after restricting the energies to be positive).21) d3 k x x e−i k·˜ a(k) + ei k·˜ b∗ (k) (2π)3 2Ek d3 k ˜ ˜ e−i k·x a(k) + ei k·x b∗ (k) (2π)3 2Ek d3 k e−i k·x a(−k) + ei k·x b∗ (−k) . (2π)3 2Ek
(2. −x) ≡ (˜µ ).22)
showing that with each solution there is a corresponding charge conjugated solution φC (x) = φ∗ (x). φP (x) ≡ φ(˜) (P for parity). for which we consider space inversion. In terms of modes one has φC (x) = φ∗ (x) = ≡ d3 k e−i k·x b(k) + ei k·x a∗ (k) (2π)3 2Ek d3 k e−i k·x aC (k) + ei k·x bC∗ (k) . They are referred to as mode and anti-mode amplitudes (or because of their origin positive and negative energy modes). (2π)3 2Ek (2. It is easy to show that x φP (x) = φ(˜) x = = = or since one can deﬁne φP (x) ≡ d3 k e−i k·x aP (k) + ei k·x bP ∗ (k) . It is trivial to see that ∂µ ∂ µ + M 2 φ∗ (x) = 0. changing the sign of the spatial coordinates.23)
.16) The consequence of this is discussed in Exercise 2. The choice of a and b∗ allows an easier distinction between the cases that φ is real (a = b) or complex (a and b are independent amplitudes).17)
(2.19) implying that for each solution φ(x) there exists a corresponding solution with the same energy. (2.

(2π)3 2Ek
where Ek =
Exercise 2. Use these to show that ˜ ˜ a(k) = ei Ek t i ∂0 φ(k. For the real ﬁeld one has aC (k) = a(k).
↔ ↔
Exercise 2. d3 k F (Ek . t). ˜ b(k) = ei Ek t i ∂0 φ∗ (−k.e. Q = 0. Note that a(k) and b(k) are independent of t. QV ≡ satisﬁes ˙ QV = − ds · j. This shows how charge conjugation transforms ’particle’ modes into ’antiparticle’ modes and vice versa. t) = ei Ek t (i∂0 + Ek ) φ(k.
S V
d3 x j 0 (x).
1 Use
↔
↔
the following property of delta functions δ (f (x)) =
zeros xn
1 δ(x − xn ).1
Show that for a conserved current (∂µ j µ = 0) the charge in a ﬁnite volume.15. k). is conserved. t) for the mode expansion in Eq. for the mode amplitudes aC (k) = b(k) and bC (k) = a(k).Relativistic wave equations
11
i. |f ′ (xn )|
.
Exercises
Exercise 2.
Exercise 2. k) = (2π)4 k2 + M 2 . t). t) ≡ d3 x φ(x.
˙ and thus for any normalized solution the full ‘charge’.2
Show that if φ and φ∗ are solutions of the KG equation.3
Show that1 d4 k 2π δ(k 2 − M 2 ) θ(k 0 ) F (k 0 . letting V → ∞. 2. t) exp(−i k · x) and its time ˜ derivative i∂0 φ(k. that j µ = i φ∗ ∂ µ φ is a conserved current Note: A ∂µ B ≡ A∂µ B − (∂µ A)B).4
˜ Write down the 3-dimensional Fourier transform φ(k.

∂ ψ(r. i (a) Show that relativistic invariance. 2. αj } = 2 δ ij I. β} = 0.2) in terms of the a(k) and b(k) using the expansion in Eq.6
Write down the mode expansion for the Lorentz transformed scalar ﬁeld φ′ (x) and show that it implies that the Lorentz transformed modes satisfy a′ (Λk) = a(k). B] = AB − BA and the anticommutator as {A.15. and β 2 = I.
Exercise 2. 1 2 (b) Express now the full charge QV (t) (exercise 2.1) for a complex scalar ﬁeld current (exercise 2. t). i. {αi . also using the result in (a).7
To solve the problem with positive and negative energies and get a positive deﬁnite density. Is the time dependence of the result as expected? Note that the mode amplitudes will become creation and annihilation operators after quantization of the ﬁelds (Chapter 7).e. a{µ bν} ≡ aµ bν + aν bµ .
in terms of the a(k) and b(k). B} = AB + BA. t) = (−i α · ∇ + m β) ψ(r.
Exercise 2. (c) Similarly. making sure that each component of ψ satisﬁes the KleinGordon equation (in terms of diﬀerential operators ∂µ ∂ µ +m2 = 0) requires the anticommutation relation2 {αi .
. We will encounter these quantities later as energy and momentum. j 0 = ψ†ψ and j = ψ † αψ. an attempt to construct a ﬁrst order diﬀerential equation for the time evolution would be to write.
From this one concludes that α and β must be matrix-valued and ψ must be a multi-component wave function. show that for two ﬁelds φ1 (with coeﬃcients a1 and b∗ ) 1 and φ2 (with coeﬃcients a2 and b∗ ) one has 2 d3 x φ∗ (x) φ2 (x) 1 = d3 k 2 (2π)3 4Ek a∗ (k)a2 (k) + b1 (k)b∗ (k) 1 2 + a∗ (k)b∗ (−k) e2iEk t + b1 (k)a2 (−k) e−2iEk t . Note that a{µ bν} indicates symmetrization. We will see a∗ (k) → a† (k) and a(k) → aop (k) with aop and a† op op being annihilation and creation operators (as known from a harmonic oscillator) and the same for the coeﬃcients b(k). express the quantities E(t) = P i (t) = d3 x (∂0 φ)∗ (∂0 φ) + ∇φ∗ · ∇φ + M 2 φ∗ φ .Relativistic wave equations
12
Exercise 2. One then ﬁnds that Q is expressed as an inﬁnite sum over number operators a† a (particles) and b† b (anti-particles). ∂t where ψ is a wave function and the nature of α and β is left open for the moment. d3 x (∂ {0 φ)∗ (∂ i} φ).5
(a) As an introduction to parts (b) and (c).
(b) Show that one has current conservation for the current
2 We
denote the commutator as [A.

A general rotation — we will consider SO(3) as an example — is of the form ′ Vx V′ y ′ Vz cos ϕ sin ϕ 0 Vx = − sin ϕ cos ϕ 0 V y 0 0 1 Vz (3. Since the KG equation expresses just the relativistic relation between energy and momentum. i. we want to investigate if there exist objects other than just a scalar (real or complex) ﬁeld φ.Chapter 3
Groups and their representations
The simple systems that we want to describe in a relativistically invariant way are free particles with spin. R(π.3)
ϕ=0
13
. z ) Lz = i ∂ϕ 0 = i 0 −i 0 0 0 .g.1
The rotation group and SU (2)
The rotation groups SO(3) and SU (2) are examples of Lie groups.g. e. The parameter-space of SO(3) is a ˆ ˆ ˆ sphere with radius π. In this section we will investigate the requirements imposed by Poincar´ e invariance. Particles with spin then will be described by certain spinors.1)
for a rotation around the z-axis or shorthand V ′ = R(ϕ. −n). two-component ﬁelds in analogy to the two-component wave functions used to include spin in a quantum mechanical description of an electron in the atom. Before proceeding with the Lorentz group we will ﬁrst discuss the rotation group as an example of a Lie group with and a group we are familiar with in ordinary quantum mechanics. 0 0 (3. in which the parameter space forms locally a Euclidean space.
3. The KG equation will actually remain valid. provided we identify the antipodes.e. that is groups characterized by a ﬁnite number of real parameters. and we must study the representations of the Lorentz group.2)
(3. however. e. z )V . In quantum mechanics spin is described by a vector. For an inﬁnitesimal rotation around the z-axis one has ˆ R(δϕ. n) where n is a unit vector and ˆ ˆ ϕ is the rotation angle. Any rotation can be uniquely written as R(ϕ. In particular. the symmetry group describing rotations is embedded in the Lorentz group.e. Locally this parameter-space is 3-dimensional and correspondingly one has three generators. n) ≡ R(π. 0 ≤ ϕ ≤ π. i. in particular each component of these spinors will satisfy this equation. In a relativistic theory. it has 3 components that we know the behavior of under rotations. z ) = 1 + i δϕ Lz with as generator ˆ 1 ∂R(ϕ. electrons. we also want it to hold for particles with spin.

and thus (σ · n)2 = 1. Writing a0 = cos(ϕ/2) and a = n sin(ϕ/2) we have1 A = A(ϕ. [z. Lj ] = i ǫijk Lk . 0
(3. 2
(3.
.4)
R
. These generators form a threedimensional Lie algebra SO(3).8)
ˆ that (σ · a)(σ · b) = a · b + i σ · (a × b). x] = 0 (thus [x. y] ∈ A. which reﬂects itself in the noncommutation of the generators. π)
(n. π )
π -sphere (antipodes identified)
2 π -sphere (surface identified)
Figure 3. [y. ˆ R(ϕ. is a 3-dimensional Lie-group. ˆ therefore. • [x. x]] + [z. the rotations in SO(3) can be generated from inﬁnitesimal rotations that can be expressed in terms of a basis of three generators Lx . [x. They satisfy the commutation relations [Li . Lx = Ly = 0 0 0 i 0 −i 0 ϕ ˆ . One way of viewing the parameter space.7) = a0 0 −1 i 0 1 0 0 1 • [x.2 π)
(-n. x]).z N
N
or (Lk )ij = −i ǫijk . (3. These matrices can be deﬁned as acting on 2-component spinors (χ → Aχ) or equivalently as acting on 2 × 2 matrices (B → ABA† ). y] = −[y. In the same way we can consider rotations around the x. we turn to the group SU (2) of special (det A = 1) unitary (A† = A−1 ) 2 × 2 matrices.5)
Rotations in general do not commute. It is straightforward to check that any (ﬁnite) rotation can be obtained from a combination of inﬁnitesimal rotations. It is straightforward to check that the conditions require a0 + i a3 +i a1 + a2 = a0 1 + i a · σ A = +i a1 − a2 a0 − i a3 1 0 + i a1 0 1 + i a2 0 −i + i a3 1 0 (3.Groups and their representations
14
(n.6)
Summarizing. Locally this is a 3-dimensional Euclidean space and SU (2).
with real a’s and 3 (ai )2 = 1.1: the parameter spaces of SO(3) (left) and SU (2) (right). thus is as the surface i=0 of a sphere in 4 Euclidean dimensions. ] that satisﬁes • ∀ x. z]] + [y. Ly and Lz .
(3. y ∈ A ⇒ [x. y]] = 0 (Jacobi identity). z ) = lim
N →∞
that are generated by i 0 . namely that there exists a bilinear product [. for rotations around z for instance.and y-axes 0 0 0 0 0 0 0 −i . Next. With matrix commutation this algebra satisﬁes the requirements for a Lie algebra. n) = 1 cos ˆ
1 Note
ϕ ϕ ˆ + i (σ · n) sin 2 2
= exp i
ϕ ˆ σ·n .

There exists a corresponding mapping of the groups given by µ: SU (2) −→ SO(3) ˆ ˆ A(ϕ. sometimes in more than one way (but this will be discussed later). ˆ ˆ A(ϕ.14)
For a Lie group. thus.
3. (3. the SU (2) → SO(3) mapping is a 2 : 1 mapping where both A = ±1 are mapped into R = I.Groups and their representations
15
where we have used the (for operators new!) deﬁnition eA ≡ An 1 = 1 + A + A2 + .12)
One. SU (2) ≃ SO(3). 2 2 2 (3. Suppose we have a system described by a Hamiltonian H. (3. For instance in a function space translations. (3.e. H] = 0 for g ∈ G. n)
ˆ −→ R(2π − ϕ.13)
0≤ϕ≤π
The relation A(σ · a)A−1 = σ · RA a (valid for any vector a) can be used to establish the homomorphism (Check that it satisﬁes the requirements of a homomorphism. They satisfy σi σj σk = i ǫijk . In the full parameter space. i. The inﬁnitesimal generators of SU (2) are ˆ obtained by considering inﬁnitesimal transformations. i. Near the identity. it is suﬃcient that the generators commute with H. . The next issue is to ﬁnd the appropriate symmetry operators in the Hilbert space. n! 2! n=0
∞
(3. n) −→ R(ϕ. i.e. .15) [g.9)
The parameter-space of SU (2). one has a Lie algebra isomorphism that is linear and preserves the bilinear product. .11)
Thus σx /2. with ˆ J ·n ≡ ˆ 1 ∂A(ϕ.2
Representations of symmetry groups
The presence of symmetries simpliﬁes the description of a physical system and is at the heart of physics. . φ′ (r) = φ(r + a) = φ(r) + a · ∇ φ + 1 (a · ∇)2 φ + . n)
π ≤ ϕ ≤ 2π. however. . 2!
. The existence of symmetries means that there are operators g belonging to a symmetry group G that commute with the Hamiltonian. the above mapping corresponds to the trivial mapping of the Lie algebras. since any ﬁnite transformation can be constructed from the inﬁnitesimal ones. σy /2 and σz /2 form the basis of the Lie-algebra SU (2). thus.e.10)
σ ˆ · n. n) ≈ 1 + i ϕ J · n. now with radius 2π and with all points at the surface identiﬁed (see ﬁgure). for ﬁxed n. 2
(3. H] = 0 for g ∈ G. in particular that AB → RAB = RA RB ). immediately sees that the Lie algebras are identical. n) i ∂ϕ =
ϕ=0
(3. [g. . also can be considered as a ﬁlled 3-sphere. .

m with eigenvalues of Jz . for rotations the operator Jz and the (quadratic Casimir) operator J 2 . the Poisson bracket operation satisﬁes all properties of a Lie algebra. m = j(j + 1) − m(m ± 1)|j. In particular the quantum operators ℓop obey in quantum mechanics commutation relations [ℓi . it is suﬃcient to consider the representations Φ of the Lie-algebra G. 3. In many cases one can simply construct a suitable Hilbert space or part of it by considering the representations Φ of a group G. J+ = 0 1 . m with m-values running from the heighest to the lowest. The ﬁnite dimensional vector space just represents new degrees of freedom. with H = i ∂/∂t and the rotation operator. j − 1. J = 0 0 + 0 0 . Indeed. pj ] = i δij . J− = 0 1 0 0 0 .e. but it is at the heart of being able to construct a suitable Hilbert space. one looks among the generators for a maximal commuting set of operators. m ± 1 with 2j + 1 being integer and m = j. m = j. It may seem trivial. ˆ U (α. Similarly one has the time translation operator φ(t + τ ) = U (τ )φ(t). z ) = exp (+iα ℓz ) . U (τ ) = exp (−iτ H) .6. These are mappings of G into a ﬁnite dimensional vector space. just starting oﬀ with the (basic) canonical commutation relations [ri . Somehow the nontrivial structure of rotations should show up and it. 0
for j = 1:
1 0 Jz = 0 0 0 0
√ 0 0 2 √ 0 √ 0 0 . while J± |j. J− = 2 √ 0 0 0 0 2 0
. . which preserve the Lie-algebra structure. the commutation relations. ˆ (3. ℓj ] = i ǫijk ℓk that are identical to those in Eq. . does in the nontrivial behavior of Poisson brackets of quantities A(x. ϕ + α) = U (α. m . From the algebra one derives J 2 |j. m = j(j + 1)|j. . e. θ. for j = 1/2: 0 1/2 Jz = 0 −1/2 0 0 −1 . for the function space. Casimir operators commute with all the generators and the eigenvalue of J 2 can be used to label the representation (j). These are mappings from G into the same ﬁnite dimensional vector space (its dimension is the dimension of the representation).18) (3.Groups and their representations
16
are generated by the derivative acting on the functions. p) (we will come back to this also in Chapter 7). . . more precise φ(r + a) = U (a)φ(r) where U (a) = exp +ia pop . The other generators Jx and Jy (or J± ≡ Jx ± iJy ) then transform between the states in V (j) . . z )φ(r. . J+ = 0 J− = 0 . p) and B(x. −j.16)
with ℓz = −i∂/∂ϕ = −i(x ∂/∂y − y ∂/∂x) or ℓop = r op × pop . j − 1. 2 . The most well-known example is spin (or the total angular momentum of a system) in non-relativistic quantum mechanics as representations of SU (2). pop and ℓop work in Hilbert space and their commutators indeed follow the required commutation relations as symmetry operators. m = m|j.17) (3. preserving the group structure. . Jz |j. −j one has for j = 0: Jz = 0 . Within the (2j + 1)-dimensional representation space V (j) one can label the eigenstates |j. i. Note that the same requirements would apply to classical mechanics.g. with pop = −i∇. To get explicit representations. φ(r. θ. The latter is suﬃcient to get the commutation relations of the (quantum) rotation operators. ϕ). Explicit representations using the basis states |j. In order to ﬁnd local representations Φ of a Lie-group G. The above set of operators H. indeed. m .

one can look at the conjugate representation. the extension from a local one must be unique. ǫy . χ)|j. χ) ≡ e−iϕJz e−iθJy e−iχJz . 0 ≡ ǫ0 ≡ ǫz . Jy = 0 0 0 . one obtains the unitary representations Φ(g) = exp(i Φ(g)) of the full group G. Of all groups of which the Lie algebras are homeomorphic. 2
From the (hermitean) representation Φ(g) of G. For those global representations.3). For SU(2).
If rotations leave H invariant. θ.20)
By moving through the parameter space the D-functions can be extended to global functions for all allowed angles. e. For a decent (welldeﬁned) global representation. the simply connected group is called the (universal) covering group. if χ → σχ. J = 0 0 + 0 0 0 0 0 0 0 0 √ 2 0 √ 0 . however. i. The matrix elements of these unitary representations are known as D-functions. 2 |1. any point in the parameter space can be reached in a unique way and any local (inﬁnitesimal) representation can be extended to a global one.19)
(3. 1 |1. m = Dm′ m (ϕ. m′ |UE (ϕ. If the group is simply connected. however. Jz = i 0 i 0 −i 0 0 0
−i 0 0
0 0 0
. is not simply connected. or equivalently the Hilbert space can be written as a direct product space of spaces V (j) . Therefore the representation and conjugate representation are equivalent in this case (see Excercise 3. Consider the j = 1/2 representation of SU(2). For the transformations A in SU (2) and R in SO(3) these are commonly parametrized with Euler angles. the possibility thus exist that some extensions will not be well-deﬁned representations.Groups and their representations
17
and for j = 3/2: 0 0 3/2 0 0 1/2 0 0 Jz = 0 0 −1/2 0 0 0 0 −3/2 √ 2 √ 0 0 3 . Explicitly.
. For SO(3). If a transformation U acts on χ. Because there exists a matrix ǫ such that −σ ∗ = ǫ σǫ−1 one immediately sees that appropriate linear combinations of conjugate states. UE (ϕ. dm′ m (θ) ≈ δm′ m − iθ(J 2 )m′ m . Inﬁnitesimally (around the identity) the D-functions for SU (2) and SO(3) are the same. The group SO(3). This turns out to be the case for all half-integer representations of the Lie algebra. ǫz } are 0 0 0 0 0 i 0 Jx = 0 0 −i . the conjugate representation is not a new one. all states in the representation space have the same energy. J− = 2 0 0 0 0 √ 3 0 0 0 0 √ 2 0 0 0 0 .
For j = 1 another commonly used representation starts with three Cartesian basis states ǫ related to the previous basis via 1 |1. There exist two diﬀerent types of paths. the topological structure of the group is important. The jump U → U ∗ implies for the generators σ/2 → −σ ∗ /2. −1 ≡ ǫ−1 ≡ √ (ǫx − i ǫy ) . that is any closed curve in the parameter space can be contracted to a point. This works for SU (2). 1 ≡ ǫ1 ≡ − √ (ǫx + i ǫy ) . j. For an element in the group G the corresponding point in the parameter space can be reached in two ways. SU (2) is the covering group of SO(3). Given a representation. θ.g. however.
(j) (j)
′
The spin matrices for that Cartesian basis {ǫx .e. χ) = eim ϕ dm′ m (θ) e−imχ . then ǫχ → ǫσχ = ǫσǫ−1 ǫχ = −σ∗ ǫχ. to be precise the states ǫχ transform via −σ ∗ .
(j)
(3. contractable and paths that run from a point at the surface to its antipode. the conjugate transformation U ∗ acts on χ∗ . θ. however.

explicitly x′µ = (Λ)µν xν + aµ or x′µ = Λµν xν + aµ .Groups and their representations
18
3. Invariance of the length of a vector requires for the Lorentz transformations x′2 = gµν x′µ x′ν = gµν Λµρ xρ Λνσ xσ = x2 = gρσ xρ xσ or Λµρ gµν Λνσ = gρσ . e.24)
L↑ + L↓ + L↑ − L↓ −
proper orthochrone proper non-orthochrone improper orthochrone improper non-orthochrone
. Using (i) and (ii) the Lorentz transformations can be divided into 4 classes (with disconnected parameter spaces) det Λ +1 +1 -1 -1 Λ00 ≥1 ≤ -1 ≥1 ≤ -1 (3. σ/2 in a two-dimensional (spin 1/2) case. which as a matrix equation with (Λ)µν = Λµν and (G)µν = gµν gives (ΛT )ρµ (G)µν (Λ)νσ = (G)ρσ . proof: (ΛT GΛ)00 = (G)00 = 1 → Λµ0 gµν Λν0 = 1 → (Λ00 )2 − i (Λi 0 )2 = 1.3
The Lorentz group
In the previous section spin has been introduced as a representation of the rotation group SU (2) without worrying much about the rest of the symmetries of the world. Writing x as a column vector and the metric tensor in matrix form. det Λ = −1 is called improper).g.27) (3. proof: det(ΛT GΛ) = det(G) → (det Λ)2 = 1.26) (3. (det Λ = +1 is called proper. consisting of the Lorentz group and translations. it is easy to derive some properties of the matrices Λ: (i) det(Λ) = ±1. (3.21) g 20 g21 g22 g23 0 0 −1 0 g30 g31 g32 g33 0 0 0 −1 the scalar product can be written as x2 = xT G x (3. it is convenient to use a vector notation for the points in Minkowski space.23)
The proper tensor structure of the matrix element (Λ)µν is a tensor with one upper and one lower index. From this property.25) (3. e To derive some of the properties of the Lorentz group. Denoted in terms of column vectors and 4-dimensional matrices one writes for the poincar´ transe formations x′ = Λ x + a. In this section we consider the Poincar´ group. 0 0 g00 g01 g02 g03 1 0 g g11 g12 g13 0 −1 0 0 = . G = 10 (3.22) (Note that xT is a row vector). We considered the generators and looked for representations in ﬁnite dimensional spaces. Thus one has the (pseudo-orthogonality) relation ΛT GΛ = G ⇔ GΛT G = Λ−1 ⇔ ΛGΛT = G. (ii) |Λ00 | ≥ 1.

K 2 . Using these explicit transformations. the diﬀerence of ﬁrst performing a boost in the y-direction and thereafter in the x-direction and the boosts in reversed order) contains a rotation (in the example around the z-axis). βγ = sinh η. z ) ≈ I + i ϕ J 3 . we have found the generators of rotations. since the generators K do not form a closed algebra. Rotations around the z-axis are given by ΛR (ϕ. one distinguishes rotations 0 + and boosts. The ﬁrst two sets of commutation relations exhibit the rotational behavior of J and K as vectors in E(3) under rotations. it is easy to ﬁnd the following typical examples from each of the four classes. The commutator of two boosts in diﬀerent directions (e. since I ∈ L↑ and + + det Λ and Λ0 change continuously along a path from the identity.31) 0 − 0 +1 0 0 0 0 +1 0 0 +1 0 0 −1 0 0 ∈ L↑ (space inversion) Is = (3. Thus ˆ 0′ 0 0 0 V 0 V 1 0 0 0 0 1′ 1 V 0 cos ϕ sin ϕ 0 V 0 0 −i 0 2′ = 2 −→ J3 = V 0 − sin ϕ cos ϕ 0 V 0 i 0 0 . K j ] = i ǫijk K k . J 2 . z ) ≈ ˆ ˆ I − i η K 3 .g. i proof: use ΛT GΛ = G and ΛGΛT = G. This is the origin of the Thomas precession. Thus 0′ 0 V cosh η 0 0 sinh η V 0 0 0 i 1′ 1 V V 0 0 0 0 0 1 0 0 2′ = 2 −→ K3 = V V 0 0 0 0 . which satisfy the commutation relations (check!) [J i . J = (J 1 . From the commutation relations one sees that the boosts (pure Lorentz transformations) do not form a group. Returning to the global group. inﬁnitesimally given by ΛB (η.
3 i 2 i=1 (Λ0 ) 3
19
(iii)
Boosts along the z-direction are given by ΛB (η. inﬁnitesimally given by ˆ ΛR (ϕ. K j ] = −i ǫijk J k .29) 0 0 1 0 3′ 3 V sinh η 0 0 cosh η V i 0 0 0
Note that Lorentz transformations generated from the identity must belong to L↑ . [K i . J j ] = i ǫijk J k . J 3 ). K = (K 1 . and those of the boosts. K 3 ).32) 0 − 0 −1 0 0 0 0 −1
. (3. z ) = exp(−i ηK 3 ).30) 0 + 0 +1 0 0 0 0 +1 0 0 −1 0 0 +1 0 0 ∈ L↓ (time inversion) It = (3.28) 3′ 3 V 0 0 0 1 V 0 0 0 0
The parameter η runs from −∞ < η < ∞. In L↑ . [J i . Note that the velocity β = v = v/c and the Lorentz contraction factor γ = (1 − β 2 )−1/2 corresponding to the boost are related to η as γ = cosh η. 0 0 +1 0 0 +1 0 0 ∈ L↑ (identity) I = (3. (3.Groups and their representations = i=1 (Λ0 )2 . z ) = exp(i ϕ J 3 ).

We will just require Poincar´ invariance for the generators themselves. We now want to ﬁnd a covariant form of the six generators of the Lorentz transformations.36) (3. J 3 = M 12 ) and for the (inﬁnitesimal) boosts (along (3.33)
These four transformations form the Vierer group (group of Klein). explicitly reading
x′µ
= =
(Λ)µν xν + aµ = ((I)µν + (ω)µν ) xν + ǫµ Λµν xν + aµ = (g µν + ω µν ) xν + ǫµ .38) Explicitly. we have for the (inﬁnitesimal) rotations (around z-axis) ΛR = I + i ϕ3 J 3 = I − i ω12 M 12 . (3. The e
2 These
commutation relations are the same as those for the ’quantummechanics’ operators i(xµ ∂ ν − xν ∂ µ )
.
inf
(3.35)
thus (ω)ij = −(ω)ji and (ω)0i = (ω)i0 . In order to ﬁnd the commutation relations including the translation generators. we continue using covariance arguments.4
The generators of the Poincar´ group e
↑ The transformations belonging to P+ are denoted as (Λ. Thus the two vector operators J and K form under Lorentz transformations an antisymmetric tensor M µν .g. which are obtained by writing the inﬁnitesimal parameters (ω)µν in terms of six antisymmetric matrices (M αβ )µν .40)
ΛB = I + i η 3 K 3 = I − i ω30 M 30 . with ϕi = − 1 ǫijk ω jk and J i = 2 z-axis). M ρσ ] = −i (g µρ M νσ − g νρ M µσ ) − i (g µσ M ρν − g νσ M ρµ ) .Groups and their representations 0 0 −1 0 0 −1 0 0 = 0 0 −1 0 0 0 0 −1 ∈ L↓ +
20
Is It
(space-time inversion)
(3. The extension to the Poincar´ group is e ↑ straightforward. inﬁnitesimally approximated by (I + ω. 2 (M αβ )µν = i g αµ g βν − g α g βµ ν (3. three of which only involve spatial coordinates (rotations) and three others involving time components (boosts). a). There are six generators.
i i0 i i0 1 ijk M jk 2ǫ
(3. ǫ). P+ etc. This is a + normal subgroup and the factor group is the Vierer group. Of the four parts only L↑ forms a group. We therefore ﬁnd (again) that there are six generators for the Lorentz group. Multiplying the proper orthochrone transformations with one of them gives all Lorentz transformations.37)
The algebra of the generators of the Lorentz transformations can be obtained by an explicit calculation2 . Summarizing.34)
The condition ΛT GΛ = G yields g ρµ + ω ρµ gρσ (g σν + ω σν ) = gµν =⇒ ωνµ + ωµν = 0.
3. Also this group can be divided into four parts. (3.
3 30
with η = ω and K = M (e. the Lorentz transformations form a (Lie) group (Λ1 Λ2 and Λ−1 are again Lorentz transformations. [M µν . One can easily convince oneself that i (ω)µν = ω µν = − ωαβ (M αβ )µν .39)
(e.g. K = M ).

P j ] = i ǫijk P k . H] = i P i . Explicitly. [J i . the generators J± in the case of the rotation group. .47)
(3.48)
. (3. H] = [J i .43)
giving the following commutation relations by equating the coeﬃcients of ǫµ and ωµν . [J i . (3. (1 + iǫγ P γ − i i ωαβ M αβ )P µ (1 − iǫδ P δ + ωρσ M ρσ ) = P µ + ω µν P ν . P µ |p. P ) in terms of the Hamiltonian and the three-momentum operators. Inﬁnitesimally.
In order to label the states in an irreducible representation we construct a maximal set of commuting operators. = pµ |p.
3. because one then sees that by letting c → ∞ the commutation relations of the Galilei group.
We have here reinstated c. Taking any one of the states in an irreducible representation. a) = Λ−1 µ P ν . K j ] = −i ǫijk J k /c2 . (3. the tensor M µν in terms of boosts cK i = M i0 and rotations J i = 1 ǫijk M jk . other states in that representation are obtained by the action of operators outside the maximal commuting set. as part of the set. a) = Λµν P ν or U † (Λ.44) (3. one obtains 2 [P i . writing the generator P = (H/c. . The generators of the translations are the (momentum) operators P µ . 2 2 (3. . .45)
Note that the commutation relations among the generators M µν in Eq.5
Representations of the Poincar´ group e
where pµ is a set of four arbitrary real numbers.41) 2 At this point. . For instance. ǫ) = 1 − ωαβ M αβ + i ǫα P α .Groups and their representations
21
inﬁnitesimal form of any representation of the Poincar´ group (thus including the translations) is of e the form i U (I + ω. the generators M αβ no longer exclusively act in Minkowski space. [J i . P ν ] [M µν . Of the generators of the Poincar´ group we choose ﬁrst of all the generators P µ . P j ] = [P i .46) [K i . H] = 0. known from non-relativistic quantum mechanics are obtained. Λµν pν ⇒ ωµν pν ⇒ ωµν = = = pµ + ω µν pν = pµ 0 ǫµνρσ pρ sσ . that commute e among themselves. the generators J 2 and J 3 in the case of the rotation group. [P µ . The requirement that P µ transforms as a four-vector (for which we know the explicit behavior from the deﬁning four-dimensional representation). J j ] = i ǫijk J k . a)P µ U † (Λ. To ﬁnd other generators that commute with P µ we look for Lorentz transformations that leave the four vector pµ invariant. These form a group called the little group associated with that four vector. P ρ ] = 0. . = −i (g µρ P ν − g νρ P µ ) .42)
(to decide on where the inverse is. ν (3. 3. check that U1 U2 corresponds with Λ1 Λ2 ). The eigenvalues of these will be the four-momentum pµ of the state. [K i . They just state that M µν transforms as a tensor with two Lorentz indices. K j ] = i ǫijk K k . a)P µ U (Λ. [K i . . For instance. (3.38 could have been obtained in the same way. These deﬁne states with speciﬁed quantum numbers that are eigenvalues of these generators. leads to U (Λ. P j ] = i δ ij H/c2 .

Wν ] = −i (gµα Wν − gνα Wµ ) . while the others have no physical signiﬁcance. of which the components generate the little group of pµ or (for the third of the following relations) from explicit calculation [Mµν . . .Groups and their representations
22
where s is an arbitrary vector of which the length and the component along pσ are irrelevant. 0. Pν ] = [Wµ .
(3.55) commute with all generators and therefore are invariants under Poincar´ transformations.54)
This will enable us to pick a suitable commuting ’spin’ operator. i exp − ωµν M µν 2 i exp − ǫµνρσ M µν pρ sσ 2
(3. for which we distinguish the following cases p 2 = m2 > 0 p2 = 0 pµ ≡ 0 p 2 = m2 > 0 p2 = 0 p2 < 0 p0 > 0 p0 > 0 p0 < 0 p0 < 0
Only the ﬁrst two cases correspond to physical states. . .50)
where the Pauli-Lubanski operators W µ are given by 1 Wµ = − ǫµνρσ M νρ P σ . 2 (3. (3. The elements of the little group thus are U (Λ(p)) = = and U (Λ(p))|p.
Given the momentum four vector pµ we choose a tetrade consisting of three orthogonal spacelike unit vectors ni (p). M ǫijk .53) (3.56)
.49)
(3.51)
The following properties follow from the fact that W µ is a four-vector by construction. i ǫµνρσ W ρ P σ . . p0 > 0. These e operators are the Casimir operators of the algebra and can be used to deﬁne the representations of ↑ P+ . From the algebra of the generators one ﬁnds that P 2 = Pµ P µ and W 2 = Wµ W µ (3.52) (3. . 2 exp (−isµ W µ ) |p. the third case represents the vacuum.
0. which can thus be chosen spacelike. . = = i exp − ǫµνρσ M µν pρ sσ |p. . . satisfying gµν nµ (p)nν (p) = j i nµ (p) pµ i ǫµνρσ pµ nν nρ nσ i j k = = −δij . Wα ] = [Wµ .
Massive particles: p2 = M 2 > 0.

In this case a set of four independent vectors is chosen starting with a reference frame in which pµ = pµ 0 in the following way: p0 = (p0 .e. The commutation relations [W i (p). W j (p)] = i M ǫijk W k (p). (3. 1. ms .64)
such that in an arbitrary frame where the four vectors p. 0). n2 (p) and s(p) are obtained by a Lorentz boost from the reference frame one has the property ǫµνρσ sµ nν nρ pσ > 0. M2
(3. The explicit construction of the wave function can be done using the D-functions (analogous as the rotation functions). (0. |p3 |). We will not do this but construct them as solutions of a wave equation to be discussed explicitly in chapter 4
Massless particles: p2 = M 2 = 0.57 The best insight in the meaning of the operators W µ is to sit in the rest frame of the particle and put P µ = (M. Noting that S i (p)
i 2
=
1 W µ (p)W ν (p) M2
niµ (p)niν (p). Having made a covariant decomposition. 0). 1. 0).61)
one sees that S i (p)
i
=−
W2 .62)
1 Thus W 2 has the eigenvalues −M 2 s(s + 1) with s = 0. p. In that case the vectors ni (p) are just the space directions and W = (0.
More generally (fully covariantly) one can proceed by deﬁning S i (p) ≡ W i (p)/M and obtain [S i (p). .
(3.
i
(3.. (1. 0. 0. 0. . the S i (p) form the generators of an SU (2) subgroup that belongs to the maximal set of commuting operators.60)
and using the completeness relation nµ (p)nν (p) = − g µν − i i
2
i
pµ pν M2
.63)
where E = p2 + M 2 and ms is the z-component of the spin −s ≤ ms ≤ +s (in steps of one).e. s. show that W /M can be interpreted as the intrinsic spin. (0.65)
. n1 (p).Groups and their representations
23
We can write W µ (p) =
3
W i (p) nµ (p) i
i=1
(3. −1)
n1 (p0 ) ≡
n2 (p0 ) ≡ s(p0 ) ≡
(3. 2 . W i (p) = −W · ni (p)). . it is suﬃcient to choose a particular frame to investigate the coeﬃcients in Eq.58)
i. S j (p)] = −i nµ (p)nν (p) j i ǫµνρσ W ρ P σ = i ǫijk S k (p). 0. 1 2 (3. 3. M2 (3.59)
(3. M J) with components W i = (M/2)ǫijk M jk . 1. 0.57)
(i. p0 > 0. Together with the four momentum states thus can be labeled as |M. 0.

−i W 1 (p). thus. W 1 (p)] [λ(p). 1 2
24
(3. Wν ]. is characterized by its momentum and the helicity. The algebra of W 1 . the action of RA b · RA a = = ˆ ˆ ˆ b · a cos ϕ + n · (b × a) sin ϕ + 2(b · n)(a · n) sin2 (ϕ/2) ˆ ˆ b · a + b · (ˆ × n) sin ϕ − 2 (b · a − (b · n)(a · n)) sin2 (ϕ/2) a
.71)
which is called the helicity. P µ . |p| (3.
Exercises
Exercise 3. But if M = 0 one has a diﬀerent algebra (Eq. which can take any integer or half-integer value.67) (3. W µ (p) = λ(p) P µ . The eigenvalues of W 1 and W 2 (in that case corresponding to the translations) thus can take any continuous values. W 2 (p)] [λ(p). Note that angular momentum does not contribute to helicity as L · p = 0. (3. W 2 (p)] = = = i M 2 λ(p). σ · RA a ≡ A(σ · a)A−1 ˆ (valid for any vector a). W 2 (3.70) as a commuting set with zero eigenvalues for P 2 and W 2 .Groups and their representations The above does actually include the massive case. and the eigenvalues of W 1 and W 2 are set to zero (corresponding to the limit M → 0 for the W µ Wµ eigenvalues of M 2 s(s + 1) → 0).68) (3. (3. where A = exp(i ϕ · σ/2) = exp(iϕ σ · n/2). however. W 2 and λ is derived from the general commutation relations for [Wµ .13. implying continuous values for the spin (actually for W i = M S i ).1
(a) Derive from the relation that deﬁnes the mapping of SU (2) matrices A into SO(3) matrices RA in Eq. This state is in physical applications nondegenerate and is denoted by |0 . The vector W µ (p) now can be expanded W µ (p) = W 1 (p) nµ (p) + W 2 (p) nµ (p) + λ(p) pµ + W s (p) sµ (p). We don’t know of any physical relevance. A massless particle.69)
We reproduce the algebra for the massive case with W 3 (p) = M λ(p). λ .73) (3. Thus 2 λ(p) = J ·p . 3. This algebra is (for instance) isomorphic to the one generated by rotations and translations in a 2-dimensional Euclidean plane (see exercises). Thus we have P 2 . a)|0 = |0 .72)
One has P µ |0 = W µ |0 = 0 and the state is invariant under Lorentz transformations. U (Λ. The vacuum: pµ = 0. 3. i W 2 (p). and gives [W 1 (p).66)
where W · p = 0 implies that W s (p) = −(M 2 /p · s) λ(p).67 has a vanishing right hand side). The meaning of λ(p) is most easily seen by comparing for a momentum eigenstate W 0 (p) = λ(p) |p| with W 0 (p) = − 1 ǫ0ijk Mij pk = J · p. |p.

d) doublet representation for SU (2)isospin or the Higgs (φ+ .
Exercise 3.Groups and their representations
25
for any vectors a and b. Examples are the quark (u. Is this easy to understand. For a spin 1/2 antiparticle the representation A∗ is needed. the pions and the isospin 0 meson state. The appropriate isospin doublet to be used for antiquarks then is (−d. This is true for SU (2) but it is. . the operators Jk must be hermitean. (c) In situations in which going to the conjugate representation (which is used to describe antiparticles) cannot be followed by a redeﬁnition of the basis. φ0 ) doublet representation for SU (2)weak . show that the transformation for the operators is given by O′ = U O U −1 = O + i [J.2
Show that for a unitary operator U = exp[iαk Jk ] with real coeﬃcients αk . Note: Recall (or derive) that the Pauli matrices σ satisfy σi σj = δij + i ǫijk σk and thus Tr(σi σj ) = 2 δij . what are the matrices J c in the Lie-algebra representation describing the spin rotations for two-component ’anti-spinors’. and calculate U (a) p U −1 (a)
. triplet and anti-triplet representation. show that one matrix ǫ exist such that J c = ǫ J ǫ† . (b) This establishes that for spin 1/2 one doesn’t need to talk about ’anti-spin’ for the antiparticle. Show actually that ǫ = 2 eiπJ (up to a phase).4
(a) If the states in a Hilbert space (unitarily) transform as φ′ = U φ = eiJ φ and O is an operator acting on φ in that Hilbert space. i. O] + . .e. the trace of the operator Jk is zero (Convince yourself that det(eA ) = eTr A by diagonalizing A.
Exercise 3. for instance. . not true for SU (3) (ﬂavor or color properties of particles know two distinct deﬁning 3-dimensional representation. u).
Excercise 3. respectively). Show that the Lie-algebra representations J and J c are equivalent.3
(a) Consider the representation A = exp(i α · J ) with the Lie-algebra representations J = σ/2 describing spin rotations for a spin 1/2 particle. one has to explicitly include this rotation. (b) Use the result under (a) to derive the matrix elements of a rotation around the z-axis and show that it indeed represents the SO(3) rotation RA = exp(i ϕ Lz ) in which the Lz is one of the (deﬁning) generators of SO(3). In both of these cases the symmetry is broken by the charges ¯¯ of the particles. Tr(σi σj σk σl ) = 2 (δij δkl − δik δjl + δil δjk ) . Now write down the three isospin 1 meson quark-antiquark (meson) states. Note the diﬀerence with the (usual) spin 0 and spin 1 states composed from two spin 1/2 states. Show that if det(U ) = 1. Writing A∗ = exp(i α · J c ). (b) Show that this transformation for the position operator in quantum mechanics indeed gives U (a) r U −1 (a) = exp (i a · p) r exp (−i a · p) = r + a. the (non-strange) η-meson state. Both particle and antiparticle have ’spin 1/2’. Tr(σi σj σk ) = 2i ǫijk .

sj ] = [pi .Groups and their representations
26
Exercise 3. r2 ) inevitably leads to interaction terms in the boost operators. that’s why many-particle non-relativistic quantum mechanics is ’easy’. [ri . sj ] = 0. = r × p + s. the others vanish.67 to 3. in particular those with the momentum operator pj and the Hamiltonian H = mc2 + p2 /2m. pj ] = [ri .6
Consider rotations and translations in a plane (two-dimensional Euclidean space). if you don’t want to do this in general. ∂t
Show that boost invariance for a free-moving particle (the commutation relations under (a) being valid). H P J K = p 2 c 2 + m2 c4 . and U (u) p U −1 (u) = p − m u.
Exercise 3. boost invariance leads to a conserved quantity. momentum and spin operators e satisfy the canonical commutation relations. Hint: for the Hamiltonian. there is a one to one mapping) the algebra in Eqs 3. i.
Exercise 3. i. Use a 3-dimensional matrix representation or consider the generators of rotations and translations in the space of functions on the (x-y) plane.e. although that in this case is a relation that is found already from the Ehrenfest relation for O = ri . Comment: extending this to more particles is a highly non-trivial procedure. sj ] = iǫijk sk . For a single free particle or for the free-moving center of mass system that is the quantity K = mr − tp. the operator U (u) that gives U (u) r U −1 (u) = r − u t. f (p)] = i ∇p f (p). = 2c2 H + mc2
satisfy the commutation relations of the Poincar´ group if the position.e.
= p. From either of these.5
In classical mechanics (Gallilean invariance) boosts describe invariance of physical laws when viewed from a moving frame.
. derive the commutation relations and show that the algebra is isomorphic to (i. p×s 1 (rH + Hr) − t p + . although the presence of an interaction term V (r1 . [ri . H] + i dt ∂O . they describe r −→ r′ = r − u t while t′ = t. rj ] = [pi . but it can be done. indeed assures d K /dt = 0. (c) Give the unitary operator for boosts. These do not matter in the non-relativistic limit (c → ∞). Do this by showing that the set of operators. (b) In quantum mechanics one has the Ehrenfest relation i dO = [O. you might just check relations involving J or K by taking some (relevant) explicit components. (a) In classical mechanics.. show ﬁrst the operator identity [r. consistent with the (canonical) commutation relations of a quantum theory.e. pj ] = iδ ij and [si . This is possible for a single particle. Determine the commutation relations for the corresponding operators Ki in quantum mechanics.69 of the Pauli-Lubanski operators for massless particles.7 (optional)
One might wonder if it is actually possible to write down a set of operators that generate the Poincar´ e transformations.

where we restrict (for ﬁxed n) the parameters 0 ≤ ϕ ≤ 2π and 0 ≤ η < ∞. It is simply connected and forms the covering group of L↑ . j ′ ). They precisely correspond to the two types of representations that we have seen before. C): Type I (denoted M ): Type II (denoted M ): 27 J= σ σ . 2 2 σ σ J = .7)
From the considerations above. Aj ] = [B i .Chapter 4
The Dirac equation
4.5) = 1 (J + i K). i ǫijk B k . 2 1 (J − i K). (4. 2 (4. C). The group SL(2. K = +i . M = exp i ϕ·σ 2 1 exp ± η · σ 2 = U (ϕ) H(η) . j) K = −i J (B = 0). similarly as the homeomorphism between SO(3) and SU (2). becoming the deﬁning representations of SL(2. 2 2 (4. This tells us how to ﬁnd the representations of the group.2)
This shows that the Lie algebra of the Lorentz group is identical to that of SU (2) ⊗ SU (2). 0. C) can be written as a product of a unitary + matrix U and a hermitean matrix H.1) (4. They will be labeled by two angular momenta corresponding to the A and B groups. it also follows directly that the Lorentz group is homeomorphic with the group SL(2.6) (4. The matrices in SL(2. (4.4) (4. respectively.3) (4.10)
. K = −i . B j ] = [Ai . (j. B j ] = i ǫijk Ak . U (ϕ) H(η) (4. Special cases are the following representations: Type I : (j.8)
ˆ ˆ ˆ with η = η n and ϕ = ϕn. C)
Instead of the generators J and K of the homogeneous Lorentz transformations we can use the (hermitean) combinations A = B which satisfy the commutation relations [Ai . 0) Type II : (0.1 The Lorentz group and SL(2. K = i J (A = 0). C) is the group of complex 2 × 2 matrices with determinant one.9) (4. With this choice of parameter-spaces the plus and minus signs are actually relevant.

One deﬁnes spinors ξ and η transforming similarly under unitary rotations (U † = U −1 . the ﬁrst equation shows that the conjugate representation with spinors χ∗ transforming according to −σ∗ /2 is equivalent with the ordinary two-dimensional representation (±ǫχ∗ transforms according to σ/2). Eqs 4. C). With E = M γ = M cosh(η) and p = M βγ n = M sinh(η) n.17)
that can be used to relate σ µ∗ = ǫ† σ µ ǫ = ǫ−1 σ µ ǫ and hence ˜ ¯ U ∗ = ǫ† U ǫ = ǫ† U ǫ. ξ −→ Hξ. H(η) = exp(−η · σ/2). ˜
Non-equivalence of M and M
One might wonder if type I and type II representations are not equivalent. one can use a boost to the frame with momentum ˆ ˆ p. ǫ−1 = ǫ† = ǫT = −ǫ). ˜ The sets of four operators deﬁned by deﬁnitions σ µ ≡ (1.19)
. U σ µ aµ U ˜
−1
H σ µ aµ H −1 = σ µ ΛB aµ H σ µ aµ H ˜
−1
(4. Note that M ∗ = ǫ† M ǫ has its equivalent for the Lorentz transformations Λ(M ) and Λ(M ∗ ). i. H ≡ (H † )−1 = H −1 ). (4. ˜ (4. η → U η. while ±ǫη ∗ transforms as a type-I + spinor.18)
As already discussed for SU (2).
(4. σ µ ≡ (1. −σ). (ǫ∗ = ǫ. One has +
−1 Λ(M ∗ ) = I2 Λ(M ) I2 .14)
satisfying Tr(σ µ σ ν ) = −2 gµν and Tr(σ µ σ ν ) = 2gµν = 2 δ µν (the matrices.12) ξ → U ξ. are not covariant!) ˜ can be used to establish the homeomorphism between L↑ and SL(2.18 show that ±ǫξ ∗ transforms as a type-II (η) spinor.11)
Considering ξ and η as spin states in the rest-frame. C) or L↑ . (4. thus. however. (4. The following relations for + rotations and boosts are useful.15) (4.1). C) and.The Dirac equation
28
Let us investigate the deﬁning (two-dimensional) representations of SL(2. In fact. (4. Also useful is the relation H 2 (p) = σ µ pµ /M = (E + σ · p)/M .e. namely H(η) = exp(η · σ/2). if there does not exist a unitary matrix S. ¯ ¯ H ∗ = ǫ† Hǫ = ǫ−1 Hǫ. there exists the matrix ǫ = iσ 2 . They are connected via Λ(M ∗ ) = Λ−1 (ǫ)Λ(M )Λ(ǫ) within L↑ because iσ 2 = U (πˆ ) showing that y + ↑ ∗ ǫ ∈ SL(2. This shows that M ∗ ≃ M . cannot be connected by a Lorentz transformation belonging to L↑ . This boost then is given by H(p) = exp η·σ 2 = cosh η η ˆ + σ · n sinh 2 2 = M +E+σ·p (4. U ≡ (U † )−1 = U ) U (ϕ) = exp(i ϕ · σ/2) but diﬀerently under hermitean boosts (H † = H. thus. The Lorentz transformations Λ(M ) and Λ(M ). Λ(ǫ) ∈ L+ . C).13)
2M (E + M )
(exercise 4. σ). η → Hη. U σ µ aµ U −1 = σ µ ΛR aµ . such that M = SM S −1 . This is not true for the two-component spinors in SL(2.16)
= σ µ Λ R aµ . ˜
= σ µ Λ B aµ .

Examples (with between brackets the Euclidean assigments) are
r t p H J λ(p) K
−→ −λ(p) −→ −K
−→ H −→ J
−→ t (scalar). i. 0) of SL(2. (pseudoscalar). −a). i. and the angular momentum in the rest frame is what we are familiar with as the spin of a particle.
The behavior is the same for classical quantities. as we have seen in section 4. that have the correct commutation relations. From the deﬁnition of the representations (0.
(4.The Dirac equation
29
where
The matrix I2 does not belong to L↑ .
−→ −r
(vector). but by a transformation belonging ↑ ↑ to the class P− .22) u= η η M (Λ) 0 where M (Λ) = ǫM ∗ (Λ)ǫ−1 with ǫ given in Eq. Thus M ∗ and M are not equivalent within SL(2. For a particle at rest only angular momentum is important and we can choose ξ(0. we have seen that in the rest frame W i (p)/M → J i . the well-known two-component spinor for a spin 1/2 particle. 4. (axial vector).e. We also have seen that the representations (0. The i angular momentum operators J are represented by σ i /2. we obtain for the two components of u which we will refer to as chiral right and chiral left components. m) = χm uL H(p) 0
. ǫµνρσ will change sign. It has the same eﬀect as lowering the indices. must combine the repe resentations. 0) (via operators J and K) one sees that under parity 2 2 1 1 (0. the aij elements of a tensor will not change sign. Although these operators cannot be used to label the representations. 0 χm uR H(p) . 2 ) and ( 2 . (vector). −→ M (Λ) (4. + −
1 0 0 1 I2 = 0 0 0 0
0 0 −1 0
0 0 0 1
. they cannot be connected by a unitary 2 2 transformation.
(scalar). 0). where a = (a0 . but rather the operators W i (p) should be used. Taking M (Λ) = H(p). m) = η(0. Within the Lorentz group. −→ −p (vector). 1 ) and ( 1 .g.17. C) are suitable for representing spin 1/2 particles. consider the four component spinor that transforms under a Lorentz transformation as 0 ξ ξ .21)
In nature parity turns (often) out to be a good quantum number for elementary particle states. m) = χm . = (4.23) u(p.13. 1 ) and ( 1 . For the spin 1/2 representations of the Poincar´ group including parity we. etc.20)
4. generators. This provides the easiest way of seeing ˜ what is happening. C). therefore. A parity transformation changes aµ into aµ .1. e.2
Spin 1/2 representations of the Lorentz group
1 1 Both representations (0. 4. but to L↑ . 2 2 (4. 0) are inequivalent. they can be connected. ) −→ ( . The representing member of the class P− is the parity or space inversion operator Is .e. the boost in Eq. a) and ˜ a = (a0 .

32)
suppressing on the RHS the identity matrix in Dirac space. γ } ∂µ ∂ν + M 2 ψ(x) = 0.33)
deﬁne for a four vector a the contraction a = aµ γµ /
. = (4. ψ † −iγ 0 ∂0 +iγ i ∂i −M
1 We
←
←
= 0.27)
where γ µ are 4 × 4 matrices called the Dirac matrices1 . We will discuss another explicit realization of this algebra in the next section. = H −1 (p).29)
which is an explicit realization of the (momentum space) Dirac equation. 0 H 2 (p) uR uR . To see this consider the equation for the hermitean † † conjugate spinor ψ † (noting that γ0 = γ0 and γi = −γi ). γ ν } = 2 g µν . one must require that for ψ(x) the Klein-Gordon relation 2 + M 2 ψ(x) = 0 is valid for each component separately). Applying (iγ µ ∂µ + M ) from the left gives (4. which in general is a linear equation in pµ .26) −2 uL uL H (p) 0 or explicitly in the socalled Weyl representation −M E+σ·p E−σ·p −M uR uL = 0.30) γ µ γ ν ∂µ ∂ν + M 2 ψ(x) = 0. The ﬁrst indeed is solved. (4. (pµ γ µ − M ) u(p) ≡ (/ − M ) u(p) = 0. From this one obtains the Cliﬀord algebra for the Dirac matrices.The Dirac equation
30
with H(p) = ¯ H(p) = 2M (E + M ) E+M −σ·p 2M (E + M ) E+M +σ·p . (4. (4. The explicit realization appearing in Eq. 4. The general requirements for the γ matrices are thus easily obtained.31)
To achieve also that the energy-momentum relation p2 = M 2 is satisﬁed for u(p).28)
which is a covariant (linear) ﬁrst order diﬀerential equation. We ﬁrst want to investigate if the Dirac equation solves the problem with the negative densities and negative energies. It is of a form that we also played with in Exercise 2. Since ∂µ ∂ν is symmetric. This leads to the Dirac equation for ψ(x) = u(p)e−i p·x in coordinate space. {γ µ . (4. As in section 2 we can use Pµ = i∂µ as a representation for the momenta (translation operators) in function space. p (4. 2 (4.27 is known as the Weyl representation. (iγ µ ∂µ − M ) ψ(x) = 0.25)
It is straightforward to eliminate χm and obtain the following constraint on the components of u. .24) (4. this can be rewritten 1 µ ν {γ .
(4.7.

The Dirac equation
31
E empty levels M -M filled levels
E particle M -M hole
Figure 4.
. E = +M (twice) and E = −M (twice).27).39)
use ρi ⊗ σj with both ρ and σ being the standard 2 × 2 Pauli matrices. the one of the negative energy states. opposite charge).34)
for the adjoint spinor ψ ≡ ψ † γ0 . From the equations for ψ and ψ one immediately sees that j µ = ψγ µ ψ is a conserved probability current (exercise!).3
General representations of γ matrices and Dirac spinors
{γ µ . .1: The Dirac sea of negative energy states and antiparticles. where γ 0 p0 ψ = M ψ
0
−→
Eψ = M γ 0 ψ. Relying on the Pauli exclusion principle for spin 1/2 particles.
4.37)
Using the explicit form of γ in the Weyl representation (see Eq. The second problem encountered before. 4.35)
is indeed positive and j 0 can serve as a probability density. one sees that there are two positive and two negative eigenvalues. but with properties such that it can be annihilated by the particle (e. This can most easily be seen in the particle rest frame. Multiplying with γ 0 on the right and pulling it through one ﬁnds a covariant equation ψ(x) iγ µ ∂µ +M
←
= 0. γ ν } = 2 g µν . remains. From the vacuum a particle can be removed. The density j 0 = ψγ 0 ψ = ψ † ψ (4. We will see how this is properly implemented (also for bosons!) when quantizing ﬁelds. Dirac supposed that the negative energy states were already completely ﬁlled and the exclusion principle prevents any more particles to enter the sea of negative energy states.38)
The general algebra for the Dirac matrices is
Two often used explicit representations are the following2 : The standard representation: σk 1 0 0 0 3 k 2 k .g.
(4. The Dirac sea forms the vacuum.
(4. This problem was solved by Dirac through the introduction of a negative energy sea. (4. This hole forms an antiparticle with the same mass.36) (4. γ =ρ ⊗1= γ = iρ ⊗ σ = 0 −1 −σ k 0
2 We
(4.

γ µ } = 0 and explicitly one has 0 1 1 .
0 γ k = −iρ2 ⊗ σ k = k σ
−σ k .40)
(4.9 and 4.49)
i 1 with −i S ρσ = 2 σ ρσ = − 4 [γ ρ . Another matrix which is often used is γ5 deﬁned as γ5 = i γ 0 γ 1 γ 2 γ 3 = −i γ0 γ1 γ2 γ3 = It satisﬁes {γ5 . γ 2 ].The Dirac equation
32
The Weyl (or chiral) representation: 0 1 .3)
. K 3 = M 30 = 0 iσ 3 2 4 or in general one has the transformation i ψ → Lψ = exp − ωρσ S ρσ ψ. (γµ )S.
(4. the rotation and boost generators in Weyl representation in Eqs 4.46) (4. while L−1 γ µ L = Λµ γ ν .R. ν The latter assures Lorentz invariance of the Dirac equation (see Exercise 4.44)
and an adjoint spinor deﬁned by ψ = ψ † γ0 .R. =ρ ⊗1= 0 −1
3
(4. γ 0 ]. (γ5 )S. We note that ψ → Lψ and ψ → ψ L−1 . For example.R. = ρ ⊗ 1 = 1 0 i ǫµνρσ µ ν ρ σ γ γ γ γ .47)
(4. −→ Sγµ S −1 . that project out chiral right/left states. 0 σ3 2 4 1 −iσ 3 0 = − i [γ 3 . = S(γµ )W. γ 0 = ρ1 ⊗ 1 = 1 0 γµ ψ
Diﬀerent representations can be related to each other by unitary transformations. −→ Sψ. 2 (4. 1 −1 2 For all representations one has
† γµ = γ0 γµ γ0 . which in the case of the Weyl representation are just the upper and lower components.42)
We note that the explicit matrix taking us from the Weyl representation to the standard representation.41) (4. it is easy to see that PR/L = 1 (1 ± γ5 ) 2
1 0 .45)
(γ5 )W.10 are 1 σ3 0 i J 3 = M 12 = = [γ 1 .43) S=√ . γ σ ]. 4! (4.R. 0
(4. S −1 . is 1 1 1 (4.48)
are projection operators.
Lorentz invariance
The Lorentz transformations can also be written in terms of Dirac matrices.
For instance in the Weyl representation (but valid generally).

6). In S. reverse the charge of the particle. but it does. in Weyl representation we ﬁnd in Dirac space T C ξ −ǫ η ∗ . Note that we have (as expected) explicitly in Weyl x representation in Dirac space P ξ −→ ψ p = γ 0 ψ = η .51)
The existence of positive and negative energy solutions implies another symmetry in the Dirac equation.54)
e.
T
(4. that brings ψ → ψ c .R.50)
Charge conjugation
but with ψ p (x) ≡ ηp γ0 ψ(˜) (Exercise 4. In any representation a matrix C exist. ∂ (4.57) (4. ψ= (4. which is again a solution of the Dirac equation. One has ψ c = −ψ T C −1 .59) −→ ψ c = C ψ = η ǫ ξ∗
. T
(4. C is real and one has C −1 = C † = C T = −C c and [C.53)
(4. Starting with (i/ − M ) ψ(x) = 0 we note ∂ that by hermitean conjugating and transposing the Dirac equation one obtains iγ µT ∂µ + M ψ (x) = 0. As with parity we look for a transformation. such that
T Cγµ C −1 = −γµ .
−iσ 2 0 −ǫ = . is a left-handed spinor. called charge conjugation.g. ψR . ∂ with the solution ψ c (x) = ηc Cψ (x). x) → x = (t.R.) and all representations connected via a real (up to an overall phase) matrix S. (or W.58)
where ηc is an arbitrary (unobservable) phase. This symmetry does not change the spin 1/2 nature. −x) and via a few manipulations obtain again the ˜ Dirac equation (i/ − M ) ψ p (x) = 0. (C)W.g. ∂ (4. x = (t. (C)S. The latter implies that the conjugate of a right-handed spinor.55) (4. ψ= (4. = 0 ǫ iσ 2
(4. (i/ − M ) ψ(x) = 0. It is easy to convince oneself that if ψ(x) is a solution of the Dirac equation.56)
(i/ − M ) ψ c (x) = 0.52) η ξ
one can apply space inversion. parity.R. Some properties of C are C −1 = C † and C T = −C. = = 0 iγ γ = −iρ ⊗ σ = −iσ 2 −iσ 2 iγ 2 γ 0 = −iρ3 ⊗ σ 2 = 0
2 0 1 2
Thus we ﬁnd back the Dirac equation. 0 −ǫ 0 0 −ǫ 0 . e. Explicitly. for instance. usually to be taken unity. γ5 ] = 0.R.The Dirac equation
33
Parity
There are a number of symmetries in the Dirac equation.

Choosing χ = −ǫχ∗ (the equivalent spin ¯ 1/2 conjugate representation).. ⇔ (/ − M ) u(p) = 0. ¯
(4. s) = Ep + M σ ·p Ep +M χs
. 2M 2M −/ + M p v(p. s) = v(p. s) = u(p. where u satisﬁes −σ · p Ep − M u(p) = 0.70) = Ps = ˆ 0 1∓σ·s 2 2
. (2π)3 2Ek
(4. the spinors satisfy C uT (p.66)
An arbitrary spin 1/2 ﬁeld can be expanded in the independent solutions. Note that the spinors in the negative energy modes (antiparticles) could be two diﬀerent spinors. s) = u(p.60) ψ(x) = 0. s)u(p. s)¯(p. In an arbitrary frame one has s · p = 0 and s(p) can e. The explicit Dirac equation in the standard representation reads ∂ iσ · ∇ i ∂t − M (4. s) v = . ¯ ¯ † ′ u (p. s) and C v T (p. s ) = v (p. 2M 2M
(4. s ) = 0.
v (p. In that frame is a spacelike unit vector sµ = (0. ψ(x) = u±s (p) e−i p·x . the spin polarization vector in the rest frame is the starting ˆ point.1). s′ ) = 2M δss′ .65) (4. be obtained by a Lorentz transformation.64) (4. s). ∂ −iσ · ∇ −i ∂t − M
(4. The ¯ ¯ solutions are normalized to u(p. s). s) u = . s)u(p.62)
where χs are two independent (s = ±) two-component spinors. where v satisﬁes σ·p −(Ep + M ) v(p) = 0. the best representation to describe particles at rest is the standard representation. s)v(p.63)
u(p. s) ei k·x d∗ (k. s′ ) = −2M δss′ . with E = Ep = + p2 + M 2 . s)u(p. in which γ 0 is diagonal (see discussion of negative energy states in section 4.
(4. s) e−i k·x b(k. p (4. s)v(p. v(p. ψ(x) = v ±s (p) ei p·x . After separating positive and negative energy solutions as done in the case of the scalar ﬁeld one has ψ(x) =
s
d3 k u(k. ¯
′ ′
Explicit solutions in the standard representation are χs . s) . p −σ · p (Ep − M ) Ep + M σ ·p ¯ Ep +M χs χs ¯
For a free massive particle.68) (4. s ) = v † (p. s)¯(p. ⇔ (/ + M ) v(p) = 0. s)v(p. s′ ) = 2Ep δss′ . It is easy to check that 1 1±σ·s 1 ± γ5 / s ˆ 0 (4.The Dirac equation
34
4.4
Plane wave solutions
Looking for positive energy solutions ∝ exp(−iEt) one ﬁnds two solutions.67)
It is straightforward to ﬁnd projection operators for the positive and negative energy states P+ P− =
s
p /+M u(p.g. s) + v(k.69)
=
−
s
In order to project out spin states.61) σ·p −(Ep + M ) There are also two negative energy solutions.

λ = + 1 ) = √ + E +M − E −M . λ = −1/2) = √ − E+M √ 0 0 E+M
(4. p3 E +M E+M 1 2 3 p + ip −p 1 2 p − ip 3 1 −p v(p. p ψR/L ≡ PR/L ψ are independent solutions. chirality. We note that for high energy ¯ ¯ the positive helicity (λ = +1/2) is in essence righthanded. λ = −1/2) = u(p. One ﬁnds for the helicity states in Weyl representation: √ √ 0√ E +M + E −M √ 1 √ 1 0√ 1 .74)
.The Dirac equation
35
(the last equality in the restframe and in standard representation) projects out spin ±1/2 states (check ˆ ˆ this in the restframe for s = z ). Therefore. √ 0 u(p. For instance with the z-axis as spin quantization axis.72)
ˆ Helicity states (p along z ) in Standard representation are: √ √ 0 E+M E +M . +1/2) = √ 0 E +M E+M . etc. while the negative helicity (λ = −1/2) is in essence lefthanded. −1/2) = √ E + M −(E + M ) 0 . λ = + 2 ) = √ E +M − E −M 2 0√ 2 2 √ 0 E +M + E −M √ √ 0√ √ − E +M − E −M 1 1 E+M − E−M 0 √ . λ) = v(p. v(p. λ) = u(p. one has in standard representation: 0 E+M E+M 1 1 0 .73)
Also for helicity states C uT (p. This means that in the solution space for massless fermions the chirality states. 1 u(p. 2 2 0 √ 2 √ 2 − E+M − E−M 0
. In principle massless fermions can be described by twocomponent spinors.48 replace the spin projection operators which are not deﬁned (by lack of a rest frame). +1/2) = √ u(p. . λ = − 1 ) = √ E + M − E − M . u(p.71) p − i p2 . 4.
(4. λ = +1/2) = v(p. Note that for solutions of the massless Dirac equation /ψ = 0. v(p. λ) and C uT (p.
By writing the helicity states in Weyl representation it is easy to project out righthanded (upper components) and lefthanded (lower components). γ5 /ψ = 0 but also p p p /γ5 ψ = −γ5 /ψ = 0. The ratio of the components is (directly or inversely) proportional to √ √ E+M − E−M 2M M ≪E M √ √ √ .
Explicit examples of spinors are useful to illustrate spin eigenstates. = √ =⇒ 2E E+M + E−M ( E + M + E − M )2 which vanishes in the ultra-relativistic limit E ≫ M or in the massless case. The chirality projection operators in Eq. −1/2) = √ (4. λ = +1/2) = E−M √ 0 − E −M 0 √ √ 0 − E−M − E −M 0 . λ = − 1 ) = √ √ v(p. helicity states. −p3 −p1 − i p2 1 v(p. (4. λ). u(p.

Properties of traces of γ-matrices (1) Tr(γ µ1 · · · γ µn ) = 0 if n is odd. s)(γ ν )kl u u
s. s′ )γ µ u(k. however. we will give one example. s). (7) Tr(γ5 γ µ γ ν ) = 0 or Tr(γ5 //) = 0. etc.) In order to show the use of the above relations. (7) γµ γ5 = -γ5 γµ = i ǫµνρσ γ ν γ ρ γ σ /3!. (5) γ µ γ λ γ ν γ ρ γ σ γµ = 2 γ σ γ λ γ ν γ ρ + γ ρ γ ν γ λ γ σ . k k 2
. u ¯ This leads with explicit Dirac indices to Lµν = = = = 1 2 1 2 ui (k ′ . u u (4. (Use (γ5 )2 = 1 and pull one γ5 through.76)
ul (k ′ . γν ] (9) σµν γ5 = γ5 σµν = ǫµνρσ γ ρ γ σ /2. (5) Tr(γ µ1 · · · γ µn ) = g µ1 µ2 Tr(γ µ3 · · · γ µn ) . s′ )¯i (k ′ . In principle one can take a representation and just calculate the quantity u(k. Some elementary properties are: Properties of products of γ-matrices: (1) γ µ γµ = 4 (2) γ µ γ ν γµ = −2 γ ν or γ µ /γµ = −2 a. s′ )(γ µ )ij uj (k.5
γ gymnastics and applications
An overview of properties can be found in many text books.75) Lµν (k. s)¯k (k. k ′ ) = 2 ′
s. (8) σµν ≡ (i/2)[γµ .s′
1 ′ (/ + m)li (γ µ )ij (/ + m)jk (γ ν )kl k k 2 1 Tr [(/ ′ + m)γ µ (/ + m)γ ν ] . It is. namely the calculation of the quantity 1 ∗ (¯(k. For this we ﬁrst have to prove (do this) that (¯(k. s)(γ ν )kl ul (k ′ .s
which appears in quantum electrodynamics calculations (see Chapter 10) for the emission of a photon from an electron changing its momentum from k to k ′ (k 2 = k ′2 = m2 ). s)γ ν u(k ′ . s′ )) (¯(k. (Use matrices C to transpose the matrices in the expression under the trace and use Tr AT = Tr A. s)γ µ u(k ′ . s)¯k (k. (6) γ5 = i γ 0 γ 1 γ 2 γ 3 = −i ǫµνρσ γ µ γ ν γ ρ γ σ /4!. s′ )) .g µρ g νσ + g µσ g νρ ). (6) Tr(γ5 ) = 0.) (2) Tr(1) = 4. more convenient ¯ to use the projection operators introduced earlier. ab (4) γ µ γ ν γ ρ γ σ γµ = −2 γ σ γ ρ γ ν . s)γ µ u(k ′ . s′ ). s)γ µ u(k ′ . s′ )(γ µ )ij uj (k. s′ ) ¯ u
s. (9) Tr(γ µ1 · · · γ µ2n ) = Tr(γ µ2n · · · γ µ1 ).The Dirac equation
36
4. s′ )) = u(k ′ . ab (4) Tr(γ µ γ ν γ ρ γ σ ) = 4 S µνρσ .s′ ∗
(4. a / (3) γ µ γ ρ γ σ γµ = 4 g ρσ or γ µ //γµ = 4 a · b. (3) Tr(γ µ γ ν ) = 4 g µν or Tr(//) = 4 a · b. ab (8) Tr(γ5 γ µ γ ν γ ρ γ σ ) = −4i ǫµνρσ . (10) γ µ γ ν = g µν − i σ µν (11) γ µ γ ν γ ρ = S µνρσ γσ + i ǫµνρσ γσ γ5 with S µνρσ = (g µν g ρσ .g µ1 µ3 Tr(γ µ2 γ µ4 · · · γ µn ) + · · · .

s) and v(p.g. a 4 a · b. s).
2M (E + M )
ˆ ˆ where E = M γ = M cosh(η) and p = M βγ n = M sinh(η) n. 1 i −i S ρσ = − [γ ρ .g. by proving that H 2 (p) = exp (η · σ) = (E + σ · p)/M = σ µ pµ /M. −x) is also a solution of the Dirac equation. Show that the equality holds.1
Prove Eq. of which only the traces of four and two gamma-matrices are nonzero. H(p) = exp η·σ 2 = M +E+σ·p .
Excercise 4.13. s)¯(p. x ˜
. γ σ ] = σ ρσ . ˜
Exercise 4.
(4. e.4
Show that in P+ ≡
s=±1/2
u(p. They lead to Lµν 1 Tr [(/ ′ + m)γ µ (/ + m)γ ν ] k k 2 = 2 k µ k ′ν + k ν k ′µ − g µν (k · k ′ − m2 ) =
= 2 k µ k ′ν + 2 k ν k ′µ + (k − k ′ )2 g µν .77)
Exercises
Excercise 4. x → x. 4.6
Apply space-inversion. Do this e. i S ρσ ] = g ρµ γ σ − g σµ γ ρ . ν µ µ Look at the inﬁnitesimal forms Λµ = gν + ων and L = 1 − (i/2)ωρσ S ρσ to obtain ν [γ µ . γν } = 2gµν that γµ / γ µ a = Tr(//) = ab µn µ1 Tr(γ · · · γ ) = −2 /.The Dirac equation
37
The trace is linear and can be split up in parts containing traces with up to four gamma-matrices. 4 2
Exercise 4. s) u p /+M = 2M 2M
both sides are projection operators.5
Show starting from the relation {γµ . to the Dirac equation and use this to show that the spinor ψ p (x) = ˜ γ0 ψ(˜). 0 if n is odd. where x = (t.2
Show that j µ = ψγ µ ψ is a conserved current if ψ en ψ are solutions of the Dirac equation
Exercise 4.3
Imposing Lorentz invariance on the Dirac equation can be achieved by requiring L−1 γ µ L = Λµ γ ν .
Exercise 4. by letting them act on the four basis spinors u(p.

you need to determine the corresponding conﬁnement condition for ψ. We will not pursue this chiral freedom at this point and take α = 0. The pion has spin 0.66 MeV. In this solution the wave number k = |k| √ = E 2 − M 2 and χm a two-component spinor while j0 and j1 are spherical Bessel functions. namely i / ψ = eiαγ5 ψ.
. Do you understand why? The calculation of the life-time is completed in chapter 11. |M (π − → µ− νµ )|2 ¯ Actually if the lepton would be massless.The Dirac equation
38
Exercise 4. Calculate for now just |M |2 . For such a cavity nµ = (0. (d) Plot for these cases the vector densities ψ † ψ and the scalar densities ψψ as a function of r/R.
(b) Next.8
A negatively charged pion (π − ) decays in a lepton ℓ (ℓ can be an electron or a muon) and a corresponding anti-neutrino (¯ℓ ). t) = N k ˆ i E+M σ · r j1 (kr)χm −i Et e .57 MeV.e. the decay would even be forbidden. Note: use ﬁrst the appropriate equation (which?) that is obeyed by each individual component of the Dirac equation to show that the upper two components are allowed solutions. Our starting point is the transition amplitude given by M = ℓ(kℓ )¯ℓ (kν )|Hint |π(p) = GF fπ u(kℓ )/(1 − γ5 ) v(kν ). Note: the outcome shows that the decay into muons is much more likely that the decay into electrons (mπ = 139. which is the most important factor needed to ﬁnd the decay width (inverse lifetime). there is no current ﬂowing through the surface of the cavity. Show that this n condition is suﬃcient to guarantee that nµ j µ = 0. i. n It turns out that there is actually a whole family of conditions that provide conﬁnement. A condition for conﬁning the fermion in the cavity is i / ψ = ψ (where n is the surface normal). we want to conﬁne this solution to a spherical cavity with radius R. n (c) Apply the conﬁnement condition to ﬁnd the value of k for the lowest eigenmode in the spherical ˆ cavity. me = 0. Note: To prove this. r).7
(a) Show that (in standard representation) j0 (kr)χm ψ(x.
Excercise 4.8 × 10−6 .511 MeV). ν p Note: how such parametrizations work will be explained in chapter 8. Calculate the value of k for M R = 0. Show that the ratio is |M (π − → e− νe )|2 ¯ ≈ 54. M R = 1. mµ = 105. the lepton and neutrino have spin 1/2 and are ν described by momentum-space spinors.
is a stationary solution of the (free) Dirac equation.5 and M → ∞. The conditions for ψ and ψ imply that nµ j µ = ψ/ ψ = 0.

λ) + ǫ(λ)∗ (k) eik·x c∗ (k.5) −E B2 .6)
where Aµ is the four-vector potential
(5. one has ǫ(λ) (k) ǫ(λ)∗ (k) = −gµν + µ ν kµ kν .3)
Since the vectors ǫµ (k) together with the momentum k µ form a complete set of four-vectors.Chapter 5
Vector ﬁelds and Maxwell equations
5. 39 (5. In an arbitrary frame ǫµ (k) with λ = 0. ǫλ ).1)
The Lorentz condition implies that k µ ǫµ (k) = 0.2
The electromagnetic ﬁeld
F µν = = ∂ µ Aν − ∂ ν Aµ 1 −E 2 0 −E 1 E 0 −B 3 2 E 3 B 0 3 E −B 2 B 1 Aµ = (φ. These are solutions of the Klein-Gordon equation with in addition a condition. −B 1 0
3
With the electromagnetic ﬁeld tensor
(5.±
5.1 Fields for spin 1
In section 3. (2 + M 2 )Vµ = 0. 0). with ∂µ V µ = 0 (Lorentz condition).± (λ)
d3 k ǫ(λ) (k) e−ik·x c(k. where k = k0 = (M. In the restframe. The covariant generalization is to describe them with Vµ (x) = ǫµ (k) e±ik·x . µ (2π)3 2E µ
(5. namely ǫµ (k0 ) = (0. there are then indeed three independent possibilities.4)
λ=0. just as the two-component Dirac spinors did for spin 1/2. (5.7)
. ±(1) are obtained by boosting the restframe vectors. λ) . The vectorﬁeld is therefore (λ) suited to describe spin 1.2 we have discussed the vectors ǫλ (with λ = 0 and ±1) as basis states of the spin 1 representation. These can describe spin 1 states in the rest-frame. A). M2 (5.2) (5. A real vector ﬁeld Vµ (x) then can be expanded in modes as Vµ (x) =
λ=0.

Furthermore.17) (radiation. Although the Lorenz condition is a constraint. It states the absence of magnetic charges and Faraday’s law.e. In that case one also has ∇·A = 0 or ki ǫi (k) = 0.15 the d’Alembertian is replaced by 2 + M 2 . One possible choice is the gauge A0 = 0 (5. it will vanish at all times. The equations of motion for the ﬁelds Aµ become 2Aµ − ∂µ (∂ν Aν ) = jµ . The gauge freedom can ﬁrst of all be used to impose the Lorentz condition for the electromagnetic ﬁeld.9)
1 ˜ where F µν ≡ − 2 ǫµνρσ Fρσ is the tensor where E → B and B → −E. (5.15) (5. Under a gauge transformation Aµ −→ Aµ + ∂µ χ. The Lorenz condition can actually be incorporated by changing the equations of motion to 2Aµ + (λ − 1)∂µ (∂ν Aν ) = jµ . in accordance with what we found in section 3.Vector ﬁelds and Maxwell equations and the (conserved) current j µ = (ρ. ˜ x Note that via parity transformation one obtains another solution. 0). Note that the incorporation of the Lorentz condition can also be done for the case of the massive vector ﬁeld. Fµν −→ Fµν + (∂µ ∂ν χ − ∂ν ∂µ χ) = Fµν . it does not eliminate the freedom of gauge transformations but they are now restricted to Aµ → Aµ + ∂µ χ with 2χ = 0. ∂µ Aµ = 0. if in Eq. 2Aµ = 0.12) (5. As remarked above. shows that the electromagnetic ﬁelds correspond to the case of a massless spin 1 ﬁeld.11)
This equation is not aﬀected by a gauge transformation.16) (5. transverse or Coulomb gauge). if ∂µ Aµ = 0 for large times |t|. only two independent vectors remain ǫ(±) µ (k) = (0. (5. (5. one ﬁnds for a conserved current and λ = 0 2∂µ Aµ = 0. while the fact µ C ∗ that the Minkowski space is real implies that Aµ (x) = Aµ (x) = Aµ (x). there is still a gauge freedom left (Aµ → Aµ + ∂µ χ with 2χ = 0). 5. For electrodynamics one has furthermore the freedom of gauge transformations. ∓1. moree over. Therefore if k µ = (|k|. The equation in vacuum. (5. for M = 0. j) the Maxwell equations read ∂µ F µν = j ν . |k|). i.14) of which the solutions give the Li´nard-Wiechert potentials.13)
Hence.
. −i. AP (x) = Aµ (˜).18) corresponding to the helicity states λ = ±1 of the massless photon ﬁeld. in which case one has 2Aµ = jµ . Taking the divergence of this equiation.8)
(5.10)
the electric and magnetic ﬁelds are unchanged. the antisymmetry of F µν implies that ˜ ∂µ F µν = 0 or ∂λ Fµν + ∂µ Fνλ + ∂ν Fλµ = 0. 0. 0.
40
(5.5. (5.

. The energy density 2 (E 2 + B 2 ) = 0 (hence a vacuum).0 . sin ϕ. illustrated in ﬁg.Vector ﬁelds and Maxwell equations
41
L ∆x d a
(infinitely) long solenoid
Figure 5. This phase diﬀerence is actually observed. In the presence of an electromagnetic ﬁeld Aµ an additional phase diﬀerence is observed.1. ϕ= − ˆ 2r 2r2 2r2
where r = (cos ϕ. The situation. This occurs in the region outside the solenoid. It is however. 0) and ϕ = (− sin ϕ. λ L − λ λ
or ∆x = (L− λ/d)δ. Nevertheless the signiﬁcance of Aµ is shown in the Aharonov-Bohm experiment. The phase diﬀerence between the electrons travelling two diﬀerent paths is given by δ = 2π a a ∆x d = −≈ . ˆ B = 0. Inside solenoid: Outside solenoid: B = B z. F = e E + ev × B (5. may look a bit akward. . cos ϕ. This causes an interference pattern as a function of ∆x.
= exp −i
The phase diﬀerence between the two paths of the electron then are given by ∆δ = − = e
1
A · dr +
e
2
A · dr = e
e
A · dr e Φ.20)
e
(∇ × A) · ds =
B · ds =
where Φ is the ﬂux through the solenoid. nevertheless. nothing more than a manisfes1 tation of a nontrivial vacuum (a la the Dirac sea for fermions). while the electromagnetic ﬁeld Aµ is not. but A = 0 (hence there is structure in the vacuum).0 .
. A= By Bx Br ϕ= − ˆ .1: The Aharonov-Bohm experiment
5. (5. In this experiment an observable phase diﬀerence is measured for electrons moving through ﬁeld-free space (E = B = 0 but Aµ = 0). but where E = B = 0. To be precise. 5. where Aµ = 0. 2 2 2 BR2 BR2 y BR2 x A= .3
The electromagnetic ﬁeld and topology∗
The electric and magnetic ﬁelds E and B as appearing for instance in the Lorentz force on a moving charge. The explanation of the signiﬁcance of the vector ˆ ˆ potential A for the phase of the electron wave function is found in minimal substitution ψ ∝ exp i p·r −→ exp i (p − eA) · r e A · r exp i p·r .19) are gauge invariant. 0).

Consider a superconductor. however. the relevant charge turns out to be 2e because the electrons appear in Cooper pairs
. giving no observable phase2 . Another situation in which the topology of space can be used is in the case of superconductors. Therefore space outside the solenoid doesn’t care that χ is multi-valued.Vector ﬁelds and Maxwell equations
U(1) S (space with solenoid)
42
class 0 ei φ class 1
Π1(S) = Z
class 2
Figure 5.
ϕ=2π
The function χ in Eq. The only way for ﬂux tubes to be formed and move around without global consequences is when each ﬂux tube contains a ﬂux Φ such that ∆δ = (e/ )Φ = n · 2π. the number of times they go around the defect) cannot be continuously deformed into one another.
1 The occurrence of nontrivial possibilities. A = ∇χ with χ= BR2 ϕ 2 (5. forming a simple (connected) space. Below the critical temperature the magnetic ﬁeld is squeezed out of the superconducting material. organizing itself in tiny ﬂux tubes (Abrikosov strings). This situation that A can be written as ∇χ is always true when ∇ × A = 0. nonobservable phases ϕ = 2π n. has been employed by Dirac in constructing magnetic monopoles in electrodynamics 2 Actually.e. 5. is multivalued as ϕ = 0 and ϕ = 2π are the same point in space. The AharonovBohm experiment shows a realization of this possibility. illustrating also that the group structure of the mapping of U (1) into this space is that of the group of integers Indeed when one considers the A-ﬁeld in the example of the Aharonov-Bohm eﬀect outside the solenoid. it can be obtained from the situation A = 0 by a gauge transformation. implying that it must be single-valued1 . If this would be be happening in empty space one would be in trouble.2: The topology of the space in the Aharonov-Bohm experiment. In such a space loops with diﬀerent winding number (i. The observable phase going around in general is ∆δ = e A · dr = e ∇χ · dr = e χ|ϕ=0 .e. i. By going around an arbitrary loop a diﬀerent number of times the electron wave function would acquire diﬀerent phases or it would acquire an arbitrary phase in a point by contracting a loop continuously into that point. The diﬀerence here is that we are working in a space with a ’defect’ (the inﬁnitely long solenoid).21)
(ϕ being the azimuthal angle). Now back to the Aharonov-Bohm experiment. Therefore the gauge transformation must be uniquely deﬁned in the space one is working in. minimizing the space occupied by B ﬁelds.21.

9 explicitly give the equations ∇ · E = ρ. ∂t
∂B = 0. ∂t ∇ · B = 0.
. 5.8 and 5.Vector ﬁelds and Maxwell equations
43
Exercises
Exercise 5.1
Show that Eqs. ∇×B =j+ ∇×E+ ∂E .

Chapter 6
Classical lagrangian ﬁeld theory
6. ∂x ˙ (6. ˙ p= ∂L . (6.
t1
(6. Dirac ﬁelds (Dirac equation) and massless vector ﬁelds (Maxwell equations) and corresponding to these ﬁelds conserved currents describing the probability and probability current.3) (6.6)
The ﬁrst term leads to the Euler-Lagrange equations.5) with ∆x(t) = δx(t) + x(t) δτ . x) = K − V = ˙ 1 mx2 − V (x). As an example. ˙ 2 (6. These equations can be obtained from a lagrangian using the action principle.4)
δS
=
t1 t2
dt dt
t1
∂L ∂L δx + δ x + L δτ ˙ ∂x ∂x ˙ d ∂L − ∂x dt ∂L ∂x ˙ δx +
t2 t1
=
∂L δx + L δτ ∂x ˙
t2
. ∂x (6.7)
The quantity ∂L/∂ x plays a special role and is known as the canonical momentum.1 Euler-Lagrange equations
In the previous chapter we have seen the equations for scalar ﬁelds (Klein-Gordon equation).8)
44
. The requirement δS = 0 with ﬁxed boundaries x(t1 ) = x1 and x(t2 ) = ˙ x2 leads to
t2 ′
(6. x (t) = x(t) + δx(t). thus t′ = t + δτ.1)
and as an example the lagrangian L(x. d dt ∂L ∂x ˙ = ∂L . and the total change x′ (t′ ) = x(t) + ∆x(t). recall classical mechanics with the action
t2
S=
t1
dt L(x.2)
The principle of minimal action looks for a stationary action under variations in the coordinates and time. x) ˙
(6.

˙ x The second term in Eq. L (φ(x). think for instance of a temperature or density distribution or of an electromagnetic ﬁeld).
t1
(6. indicated as x′µ φ (x) or combined φ′ (x′ ) = φ(x) + ∆φ(x). but which in 3 dimensions includes conserved quantities related to rotations and boosts. t2 ).
R
(6. In general any continuous transformation. 6.17)
Furthermore the variations δφ and δ∂µ φ contribute to δS.t1 )
(6.2.6 can be rewritten as δS = . + ∂L ∆x − H δτ ∂x ˙
t2
. Variations in the action can come from the coordinates or the ﬁelds. . their derivatives and possibly on the position. i. ∂µ φ′ . Consider a lagrangian density L which depends on these functions. −
d3 x . x). . The resulting variation of the action is δS =
R µ ′
= xµ + δxµ . .16)
The change in variables x → x′ in the integration volume involves a surface variation of the form dσµ L δxµ . ∂µ φ(x). .
∂R
Note for the speciﬁc choice of the surface for constant times t1 and t2 . also indicated by ∂R (a more general volume in four-dimensional space-time with some boundary ∂R can also be considered). which in classical mechanics vanishes at the boundary) does not play a role.
(R3 . x′ ) −
d4 x L (φ. δ(∂µ φ)
(6.15)
d4 x′ L (φ′ .10)
One sees that invariance under time translations requires that H(t1 ) = H(t2 ). ˙ (6. gives rise to a conserved quantity (at least if H is invariant!) Q=p dx dλ −H dt dλ . We will come back to it in a bit more formal way in section 9. dσµ . In classical ﬁeld theory one proceeds in complete analogy but using functions depending on space and time (classical ﬁelds. ∂µ φ(x)) =
R
d4 x L (φ(x). . ∂µ φ(x)).18)
1 Taking a functional derivative. The hamiltonian H is deﬁned by H(p.e.14) (6. ∆x = (dx/dλ)λ=0 δλ and δt = (dt/dλ)λ=0 δλ.
. ∂µ φ.13) (6.Classical lagrangian ﬁeld theory
45
For the lagrangian speciﬁed above this leads directly to Newton’s law p = m¨ = −∂V /∂x. t1 ) and R3 . x) and an action
t2
S[φ] =
t1
dt L =
dt d3 x L (φ(x). H is a conserved quantity. .11)
λ=0
of which p and H are the simplest examples related to space and time translations.9)
which is done because the ﬁrst term (multiplying ∆x.
(6. . . = φ(x) + δφ(x)
(6. with ∆φ(x) = δφ(x) + (∂µ φ)δx . . giving1 δS = = δL δL δ(∂µ φ) + δφ + δ(∂µ φ) δφ R δL δL δφ + d4 x − ∂µ δφ δ(∂µ φ) R d4 x dσµ L δxµ
∂R
dσµ
∂R
δL δφ + L δxµ .12)
Here R indicates a space-time volume bounded by (R3 . x) ≡ p x − L .t2 )
d3 x .
λ=0
(6. indicated with δF [φ]/δφ should pose no problems. =
∂R (R3 .

26)
The Dirac ﬁeld
The appropriate lagrangian from which to derive the equations of motion is L = i → i ← i ↔ ψ / ψ − M ψψ = ψ / ψ − ψ / ψ − M ψψ ∂ ∂ ∂ 2 2 2 = ψ (i/ − M ) ψ.27) (6.29)
. δL δL ∂µ = . The integrand of the ﬁrst term must vanish.21)
which diﬀer only by surface terms. Using the variations in ψ (in the symmetric form).
The scalar ﬁeld
It is straightforward to derive the equations of motion for a real scalar ﬁeld φ from the lagrangian densities. 2
ψ = 0. δL δ(∂µ ψ) δL δψ one obtains immediately i / −M ∂
→
i = − γ µψ 2 i → = ∂ / ψ − M ψ.19) δ(∂µ φ) δφ
6.24) (6. (2 + M )φ (x) = 0. L = 1 1 ∂µ φ∂ µ φ − M 2 φ2 2 2 1 µ = − φ ∂µ ∂ + M 2 φ.22)
which can be considered as the sum of the lagrangian densities for two real scalar ﬁelds φ1 and φ2 √ with φ = (φ1 + iφ2 )/ 2. the second term is irrelevant.2
Lagrangians for spin 0.23) (6. 1/2 and 1 ﬁelds
By an appropriate choice of lagrangian density the equations of motion discussed in previous chapters for the scalar ﬁeld (spin 0). leading to the Euler-Lagrange equations.28)
where the second line is not symmetric but in the action only diﬀers from the symmetric version by a surface term (partial integration). leading to (2 + M 2 )φ(x) = 0.
2 ∗
(6.20) (6. (6. (6. One easily obtains (2 + M 2 )φ(x) = 0.
(6. the Dirac ﬁeld (spin 1/2) and the vector ﬁeld (spin 1) can be found. ∂ (6.Classical lagrangian ﬁeld theory
46
With for the situation of classical ﬁelds all variations of the ﬁelds and coordinates at the surface vanishing.25) (6. For the complex scalar ﬁeld one conventionally uses L = = ∂µ φ∗ ∂ µ φ − M 2 φ∗ φ −φ∗ ∂µ ∂ µ + M 2 φ. 2 (6.

but note the factor 1/2 as compared to the Dirac lagrangian.36) (6. Peshkin and Schroeder or Exercise 12.g. Using the twospinors ξ and η. which comes because we in essence use ’real’ spinors.
(6. since a lagrangian with both Dirac and Majorana mass terms can be rewritten as the sum of two Majorana lagrangians after redeﬁning the ﬁelds (See e. namely2 LM (Majorana) = + (6.Classical lagrangian ﬁeld theory
47
and similarly from the variation with respect to ψ ψ i / +M ∂
←
= 0.and lefthanded ones.
(6. 2 (6. L = =
↔ 1 1 c c M ψL ψL + M ∗ ψL ψL ψL i / ψL − ∂ 2 2 ↔ 1 1 M ΥR ΥL + M ∗ ΥL ΥR . the mass term in the Dirac lagrangian 6.38) (6. ∂ ∂ 2 2
(6. that the lagrangian separates into two independent parts for M = 0.33)
We now introduce a ’real’ (four-component) spinor satisfying Υc = Υ with ψL . 1 M η † ǫη ∗ − M ∗ η T ǫη .g. The reasons for Grassmann variables will become clear in the next chapter.31 is given by LM (Dirac) = −M ξ † η + η † ξ .32)
There exists another possibility to write down a mass term with only one kind of ﬁelds.30)
It is often useful to link to the two-component spinors ξ and η which we started with in chapter 4. ∗ 0 ⇒ Υ ≡ −ǫ η ΥL = ψL = η η
∗ for which η0 = ǫ η0 .
Vector ﬁeld
From the lagrangian density L 1 1 λ 2 = − Fµν F µν + M 2 Vµ V µ − (∂µ V µ ) 4 2 2 1 µ = V (∂ 2 + M 2 )gµν − (1 − λ)∂µ ∂ν V ν . it is in the absence c of a Dirac mass term possible to introduce a lagrangian in which only left ﬁelds ψL and ψL appear or one can work with ’real’ spinors or Majorana spinors. ψL ≡ (ψL )c = C ψL = 0
(6.
and thus (β ∗ α)∗ = α∗ β = −βα∗ .35)
Since the kinetic term in L separates naturally in left and right parts (or ξ and η). In that case one ﬁnds trivially L =
↔ ↔ 1 1 ψR i / ψR + ψL i / ψL − M (ψR ψL + ψL ψR ).7). We note that
of which the left part coincides . or equivalently separate the ﬁeld into right.31)
showing e.34) (6.39)
2 This
term is written down with η being an anticommuting Grassmann number for which (αβ)∗ = β ∗ α∗ = −α∗ β ∗ .37)
This gives an expression with a mass term that is actually of the same form as the Dirac mass term. The Majorana case is in fact more general. Υi / Υ − ∂ 4 2
T −ǫ η ∗ c = ΥR .
. 2
αβ = −βα. (6.

. Q(t1 ) = Q(t2 ). we consider the second term in Eq. ∂µ J µ (x) = 0. discussed in the next chapter. implying ∂µ F µν + M 2 V ν = 0 and ∂µ V µ = 0.40)
6. Returning to δS the surface term is rewritten to δS =
R
d4 x .3
Symmetries and conserved (Noether) currents
Next. δ(∂µ φ)
(6. one can consider the variations at the initial or ﬁnal surface.. (6.46)
For quantum ﬁelds (φ operator!) things become more subtle as it it is not possible to specify ˙ for instance φ and φ on the surface σ1 .42)
Θµν = The variation δS thus can be expressed as δS with J µ (x) = − =
δL ∂ ν φ − L g µν .47)
.+
R
4
δL δL ∆φ − ∂ ν φ − L g µν δxν δ(∂µ φ) δ(∂µ φ) δL ∆φ − Θµν (x)δxν . 6. +
∂R
dσµ dσµ
∂R
≡ where
d x . ψ(x) → ei eΛ ψ(x) or ∆ψ(x) = i eΛ ψ(x).Classical lagrangian ﬁeld theory
48
where Fµν = ∂µ Vν − ∂ν Vµ . These do not aﬀect the dynamics and will give rise to conserved currents. (6. we get (omitting the parameter δλ) dxν dφ δL + Θµν (x) .45) J µ (x) = − δ(∂µ φ) dλ λ=0 dλ λ=0 Thus considering ∂R = σ2 .
(6. δ(∂µ φ)
For continuous transformations. .41) (6. requires the presence of a conserved quantity. (6. In this case. which in a consistent picture precisely generate the symmetries.. As an example for the classical ﬁeld.43)
F (σ1 ) − F (σ2 ) =
dσµ J µ (x) =
∂R R
d4 x ∂µ J µ (x). consider U (1) transformations of the Dirac ﬁeld proportional to the charge e.18. ∆φ = (dφ/dλ)λ=0 δλ and δxµ = (dxµ /dλ)λ=0 δλ. Q= d3 x J 0 (x. t). these conserved quantities become operators. which is the space-integral over the zero-component of a conserved current. In particular when the lagrangian is invariant under symmetries.σ1 with σ = (R3 . one immediately obtains δL = −∂ µ V ν + ∂ ν V µ − λ g µν (∂ρ V ρ ) = −F µν − λ g µν (∂ρ V ρ ) δ(∂µ Vν ) leading to the equations of motion ∂µ F µν + M 2 V ν + λ ∂ ν (∂ · V ) = 2 + M 2 V ν − (1 − λ)∂ ν (∂ · V ) = 0. δ(∂µ φ)
(6.44)
δL ∆φ + Θµν (x)δxν . the presence of a symmetry that leaves the lagrangian invariant. (6. t).

(6.49)
These currents are conserved as discussed already in chapter 2.55)
Note that the coordinate transformations can be written in a form similar to that for the ﬁelds (matrices in internal space). For quantized ﬁelds this will become the expressions of the hamiltonian and the momentum operators in terms of the ﬁelds. H = d3 x Θ00 (x).
6. (6.
(6. The integral over the zero-component. From Noether’s theorem one sees that the current Θµν ǫν is conserved. = −(∂µ φ)δxµ = −ǫµ ∂µ φ(x). including e the space-time translations and the Lorentz transformations. (ω µν antisymmetric) 1 ωρσ (−i S ρσ )ij φj (x).51) (6. Summarizing.56) δxµ = ωρσ (aρσ )µν xν 2 with (aρσ )µν = aρσ µν = g ρµ g σν − g σµ g ρν .Classical lagrangian ﬁeld theory From the lagrangian for the Dirac ﬁeld.48)
For the complex scalar ﬁeld the U (1) transformations leave the lagrangian invariant and lead to the conserved current (see Exercise 6.53)
These are the energy and momentum.
Lorentz transformations
In this case the transformation of the coordinates and ﬁelds are written as δxµ ∆φi (x) = = ω µν xν . = 0.52)
The behavior of the ﬁeld under translations (δφ) is governed by the translational behavior of the argument in such a way that ∆φ = 0. 1 (6.54) (6.50) (6.g.3) j µ = i e φ∗ ∂µ φ. Therefore there are four conserved currents Θµν (ν labeling the currents!) and four conserved quantities Pν = dσµ Θµν = d3 x Θ0ν . 2 (6.
↔
(6. ˙ Q = d3 x j 0 (x.4
Space-time symmetries
One kind of symmetries that leave the lagrangian invariant are the Poincar´ transformations. e. (6.57) For the ﬁelds the behavior under Lorentz transformations has been the subject of previous chapters. t) is the conserved charge (Q = 0).
Translations
µ Under translations (generated by Pop ) we have
δxµ ∆φ(x) δφ(x)
= ǫµ . we obtain (since δxµ = 0). omitting the parameter Λ jµ = − δL δL (i e ψi ) (−i e ψi ) − δ(∂µ ψi ) δ(∂µ ψ i )
49
= e ψγ µ ψ. In the next section we will see the quantity show up as the charge operator.
.

In some applications (speciﬁcally coupling to gravity) it is advantageous to have an equivalent current that is symmetric. one has a tensor that satisﬁes (exercise 6. (6.
(6. the angle of the transformation depends on the space-time point x.e.67)
i. γσ ] This result for the Dirac ﬁeld has been discussed in chapter 3 (see also Exercise 4.64) (6. The current deduced from Noether’s theorem is Jµ = = = = − δL ∆φi + Θµν δxν δ(∂µ φi ) δL 1 ωρσ − (−i S ρσ )ij φj + Θµρ xσ − Θµσ xρ 2 δ(∂µ φi ) 1 ωρσ {H µρσ + Θµρ xσ − Θµσ xρ } 2 1 ωρσ M µρσ .60)
Therefore there are six conserved currents M µρσ labeled by ρ and σ (antisymmetric) and corresponding to it there exist conserved quantities M ρσ = d3 x M 0ρσ . (6. The lagrangians which we have considered sofar are invariant under global gauge transformations and corresponding to this there exist a conserved Noether current.61)
A ﬁnal note concernes the symmetry properties of Θµν . • Vector ﬁeld Aµ : −i Sρσ = aρσ .66)
In section 6.62) with Gρµν = (H ρµν + H µνρ + H νµρ )/2. 2
(6. e. φ(x) φ (x) ∂µ φ(x)
∗
→ e
→ e
→ ei eΛ(x) φ(x).
−i eΛ(x) ∗ i eΛ(x)
(6.65)
M µρσ = T µρ xσ − T µσ xρ . φ(x). Any lagrangian containing derivatives.68) (6.g.63)
µν
= ∂µ Θ
.3 we have seen global transformations or gauge transformations of the ﬁrst kind. Deﬁning T µν = Θµν − ∂ρ Gρµν . (6.58) (6.
6.6) T µν = T νµ . In general this tensor is not symmetric. is not invariant under local gauge transformations.70)
.59) (6.
in which the U (1) transformation involves an angle e Λ.
∂µ φ(x) + i e ∂µ Λ(x) e
(6.5
(Abelian) gauge theories
φ(x) → ei eΛ φ(x). independent of x. where H ρµν is the quantity appearing in M ρµν . however.3).Classical lagrangian ﬁeld theory
50
• Scalar ﬁeld φ: −i Sρσ = 0.69)
i eΛ(x)
φ (x). ∂µ T
µν
(6. (6. Gauge transformations of the second kind or local gauge transformations are transformations of the type φ(x) → ei eΛ(x) φ(x). • Dirac ﬁeld ψ: −i Sρσ = −(1/4)[γρ .

73) (6.75)
The ﬁeld φ is used here in a general sense standing for any possible ﬁeld. (6. A solution is the one known as minimal substitution in which the derivative ∂µ is replaced by a covariant derivative Dµ which satisﬁes Dµ φ(x) → ei eΛ(x) Dµ φ(x).Classical lagrangian ﬁeld theory
51
where the last term spoils gauge invariance. The equation of motion for the fermion follow from δL δ(∂µ ψ) δL δψ = = i − γ µ ψ. Dµ φ) − Fµν F µν . D ∂ A (6. ∂ 2 4 (6. 4 (6. 2 Minimal substitution ∂µ ψ → (∂µ + i eAµ )ψ leads to the gauge invariant lagrangian L = 1 i ↔ ψ / ψ − M ψψ − e ψγ µ ψ Aµ − Fµν F µν . For this purpose it is necessary to introduce a vector ﬁeld Aµ . As an example consider the Dirac lagrangian. (6. −e j µ Aµ = −e ρφ + e j · A. Dµ φ(x) ≡ (∂µ + i eAµ (x))φ(x). i L = ψγ µ ∂µ ψ − (∂µ ψ)γ µ ψ − M ψψ. 1 L (φ.76)
We note ﬁrst of all that the coupling of the Dirac ﬁeld (electron) to the vector ﬁeld (photon) can be written in the familiar interaction form Lint = −e ψγ µ ψ Aµ = −e j µ Aµ .71)
Thus the covariant derivative has the correct transformation behavior provided Aµ → Aµ − ∂µ Λ.72) (6. (6.78)
.74)
the behavior which we have encountered before as a gauge freedom for massless vector ﬁelds with the (free) lagrangian density L = −(1/4)Fµν F µν .77)
involving the interaction of the charge (ρ = j 0 ) and three-current density (j) with the electric potential (φ = A0 ) and the vector potential (A). ∂µ φ) =⇒ L (φ. Replacing derivatives by covariant derivatives and adding the (free) part for the massless vector ﬁelds to the original lagrangian therefore produces a gauge invariant lagrangian. 2 i ∂ / ψ − M ψ − e/ ψ A 2
giving the Dirac equation in an electromagnetic ﬁeld. (i/ − M ) ψ = (i/ − e/ − M ) ψ = 0. The required transformation for Dµ then demands Dµ φ(x) = = (∂µ + i eAµ (x))φ(x) ei eΛ ∂µ + i e(A′ + ∂µ Λ) φ µ → ei eΛ ∂µ φ + i e (∂µ Λ) ei eΛ φ + i eA′ ei eΛ φ µ ≡ ei eΛ (∂µ + i eAµ ) φ.

↔
Exercise 6. 2 2 (b) Show that the Klein-Gordon equation for the complex scalar ﬁeld (considering φ and φ∗ as independent ﬁelds can be derived from the lagrangian density L = ∂µ φ∗ ∂ µ φ − M 2 φ∗ φ.3
Show that the current jµ = i φ∗ ∂µ φ for a complex scalar ﬁeld is connected to a U (1) transformation on the ﬁelds. ∂µ F µν = j ν .4
Given the Dirac equation for a (negatively charged) spin-1/2 particle in an external electromagnetic T ﬁeld. where j µ = e ψγ µ ψ. ∂µ F µν = j ν . [iγ µ (∂µ − i e Aµ ) − M ]ψ = 0. (6.
Exercise 6. −eψγ ν ψ. 4 (b) What is the form of the interaction term involving a current jµ and the ﬁeld Aµ that will give the inhomogeneous Maxwell equations. φ → eiΛ φ. and deduce what is the charge of the antiparticle as compared to a particle.e. give the equation which is satisﬁed by ψ c = Cψ .
Exercise 6.
giving the Maxwell equation coupling to the electromagnetic current.
. ∂µ j µ = 0 (Note that the addition of a total derivative to the lagrangian density does not modify the equations of motion).79)
Exercises
Exercise 6.1
(a) Show that the Klein-Gordon equation for the real scalar ﬁeld can be derived from the lagrangian density 1 1 L = ∂µ φ∂ µ φ − M 2 φ2 .Classical lagrangian ﬁeld theory
52
For the photon the equations of motion follow from δL δ(∂µ Aν ) δL δAν = = −F µν . i. (c) Show that the interaction term is invariant under gauge transformations only if the current jµ is conserved.2
(a) Show that the homogeneous Maxwell equations can be derived from the lagrangian density 1 L = − Fµν F µν .

2M 2M with Enr ≡ E − M ≪ M . Hints: it may be useful to realize that Gµρσ is antisymmetric in ﬁrst two indices.Classical lagrangian ﬁeld theory
53
Exercise 6.4) for a stationary solution in the non-relativistic (|Enr | ≪ M ) and weak-ﬁeld (|A0 | ≪ M ) limits yields in the standard representation for the ’upper (two) components’ ψu . This relation is known as the Pauli equation.6 (optional)
Prove the properties of the tensor T µν in section 6. Comment: this relation is known as the Pauli equation. A) (see also exercise 6. 2me
Exercise 6. t).4.
. ﬁnally note that it is suﬃcient to show the last equation for dσµ M µρσ . e 1 (−i∇ + eA)2 + σ · B − eA0 ψu (r. Hint ≡ −gs µe · B = −gs Qe s · B. use ∂µ T µν = 0 and ∂µ M µρσ = 0. t) = Enr ψu (r. It shows that for a spin 1/2 particle the g-factor in the coupling of spin to a magnetic ﬁeld takes the value gs = 2.5
Show that the Dirac equation for an electron with charge −e in an external electromagnetic ﬁeld Aµ = (A0 . in order to show that Θρσ − Θσρ = ∂µ H µρσ .

6. t). q) = p q − L(q. ˙ L(q. Quantization of the theory is achieved by imposing canonical commutation relations between q and p. q) = ˙ 1 mq 2 − V (q). ˙ 2 (7. there are conserved ˙ ˙ quantities Q (Eq.11) found through Noether’s theorem. An immediate generalization for ﬁelds can be obtained by considering them as coordinates. starting from the lagrangian L(q.Chapter 7
Quantization of ﬁelds
7. labeled by the position. is the Poisson bracket [A. in our example p = mq. q) also considered in the previous chapter. q) ≈ (dO/dλ)λ=0 δλ under symmetry transformations one has dO = [O. p]P . The bilinear product for the conserved quantities. Actually the conserved quantities satisfy a similar algebra as the Lie algebras for continuous symmetries that they are obtained from.2) H(p. (7. p]P = 1.4) with a possible realization as operators in the Hilbert space of (coordinate space) wave functions through qop ψ(q) = qψ(q) and pop ψ(q) = −idψ/dq.1 Canonical quantization
We will ﬁrst recall the example of classical mechanics for one coordinate q(t). Q]P . 1 qx (t) = 3 d3 x φ(x.5) ∆ x ∆3 x 54
. namely [q. H]P dt and dO = [O. B]P = dB dA dA dB − . p)) = ˙ ˙ 2m where the (canonical) momentum p = ∂L/∂ q. q(q. p] = i. [q.3)
For variations δO(p.1)
The Hamiltonian (also corresponding to a conserved quantity because of time translation invariance) is given by p2 + V (q). dq dp dq dp (7. dq
The basic Poisson bracket is actually the one between q and its canonical momentum p. Further. however. (7. This is nothing else as representing the classical Poisson algebra as operators in the Hilbert space. (7. dλ Examples are dO = [O.

˙ δ φ(x) (7. py (t)] = i δxy or for the ﬁelds φ(x) and Π(x) the socalled equal time commutation relations [φ(x. t)] = i δ 3 (x − y). ∇φ) (7. a) = Rj (Λ−1 ) φj (Λ−1 x − a).10) x ∆3 x [Πx qx − L x ] ˙ ∆3 x Hx (7. for symmetry transformations. The lagrangian is given by L = = x In order to construct the hamiltonian it is necessary to ﬁnd the canonical momenta. px (t) = ∂L ∂Lx = ∆3 x = ∆3 x Πx (t). Φ(f ) = d4 x φ(x) f (x). t)] = 0.2.12) (7. Π(y. the ﬁelds φ(x) can be considered as regular operators in the Hilbert space after smearing with a test function f .9)
Note that this indeed corresponds to the zero-zero component (Θ00 ) of the conserved energy-momentum stress tensor Θµν .
(7. qx . ∂ qx ˙ ∂ qx ˙ (7.7) d3 x L (φ(x). t)] = [Π(x. the discretization procedure above is an explicit example. δ∂0 φ) 1 1 1 = Θ00 (x) = (∂0 φ)2 + (∇φ)2 + M 2 φ2 . This implies speciﬁc transformation properties in the Hilbert space for the quantum ﬁelds φ. t). ∂µ φ(x)) = ˙ d3 x L (φ. 2 2 2
(7. φ(y.14)
In fact. Sometimes it is useful to keep in mind that.8) (7. qx+∆x ) . Π(y.15) with furthermore the relations [φ(x.
i U † (Λ. ∆φ. t). As an example. The hamiltonian then is H= x px qx − L = ˙ = x where H (x) = δL ˙ φ(x) − L (x) = Θ00 (x). for the scalar ﬁeld theory.16)
. φ.6)
∆3 x L x (qx .Quantization of ﬁelds
55
etc. t). The essential conditions to consistently quantize a theory then are the following.13)
Π(x) H (x)
For the quantization procedure we can formulate a number of basic axioms of quantum ﬁeld theory. discussed in the section 6. (7. • Poincar´ invariance e The full variation of the ﬁelds. • Canonical commutation relations The quantization of the theory is achieved by imposing the canonical quantization condition [qx (t). albeit with ’sharp’ functions. we have L (x) = 1 1 ∂µ φ∂ µ φ − M 2 φ2 2 2 1 1 1 2 (∂0 φ) − (∇φ)2 − M 2 φ2 = 2 2 2 δL = = ∂0 φ.11) (7. a) φi (x)U (Λ. (7. formally. ˙
where Πx (t) is obtained from the continuous ﬁeld Π(x) ≡ δL /δ(∂0 φ).

20)
where the coordinate Q and the momentum P satisfy the canonical commutation relations [Q. (7.25)
. a] = −a. φ(y)] = 0 if (x − y)2 < 0. a† ] = a† . a] = [a† . Q] = [P. j (7.19)
7. Microscopic causality implies local commutativity. [φ(x). P ] = i. In simpliﬁed form the hamiltonian is given by H= 1 2 1 2 2 P + ω Q .4. 1 Q = √ (a + a† ) 2ω and P = −i ω (a − a† ) 2 (7. (7. i(xµ ∂ν − xν ∂µ )φ (x) +
i
(7. P ] 2 (7. a) =
i Rj (Λ−1 ) =
1 + iǫµ P µ − 1−
i ωµν M µν .24)
It is straightforward to ﬁnd the commutation relations between N and a and a† .21)
Writing Q and P in terms of creation (a† ) and annihilation (a) operators.2
Creation and annihilation operators
Before discussing (real and complex) scalar ﬁelds and Dirac ﬁelds we recall the analogy with the wellknown harmonic oscillator as an example of quantization using creation and annihilation operators. ω a† a + 2 2 ω i − [Q.22)
it is straightforward to check that the commutation relations between Q and P are equivalent with the commutation relations [a. (7. This must be valid for the – now operators – generators P µ and M µν that in the previous chapter have been found in terms of the ﬁelds through Noether’s theorem. [Q. P ] = 0 (7. • Causality Operators Φ(f ) and Φ(g) for which the test functions are space-like separated can be measured simultaneously (macroscopic causality). The measurements cannot inﬂuence each other or [Φ(f ). Φ(g)] = 0. 2
i ωµν (S µν )i . j 2
one has [φi (x).23) The hamiltonian in this case can be expressed in the number operator N = a† a. and [N. Mµν ] =
i
i∂µ φi (x). [a.Quantization of ﬁelds
56
or if U (Λ.17) (Sµν )i φj (x). Pµ ] = [φ (x). H = = ω ω P P Q−i √ Q+i√ 2 2 2ω 2ω 1 1 =ω N+ . [N.18)
with S µν given in 6. sometimes referred to as second quantization. a† ] = 1. 2 2 (7. a† ] = 0.

is equivalent with [a(k). N |n = n|n one ﬁnds N a† |n N a|n = = (n + 1) a† |n . Π(x )]x0 =x′0 = 0. Note that we will often write a(k) or a† (k). a(k ′ )] = [a† (k). the real scalar ﬁeld φ(x) becomes d3 k (7.
7. a† (k ′ )] = (2π)3 2Ek δ 3 (k − k′ ) [a(k).e. and
(7. H = = = d3 x 1 1 1 (∂0 φ)2 + (∇φ)2 + M 2 φ2 2 2 2 d3 k Ek † a (k)a(k) + a(k)a† (k) (2π)3 2Ek 2 d3 k Ek N (k) + Evac . The canonical momentum becomes −i d3 k ˙ Π(x) = φ(x) = a(k) e−i k·x − a† (k) ei k·x . Π(x′ )]x0 =x′0 = i δ 3 (x − x′ )
′ ′
and (7.like’ commutation relations between a(k) and a† (k ′ ). d3 x φ(x) i ∂0 e−i k ·x
↔
′
↔
(7. and we see that a state |0 must exist for which N |0 = a|0 = 0. a† (k ′ )] = 0.26) a(k) e−i k·x + a† (k) ei k·x . (n − 1) a|n
† i.g.28) (7. † √ a and a act as raising and lowering operators. which we have split into positive and negative energy pieces with (complex) coeﬃcients a(k) and a∗ (k) multiplying them. [φ(x). The quantization of the ﬁeld is achieved by quantizing the coeﬃcients in the Fourier expansion. In this way one has found for the harmonic oscillator the spectrum of eigenstates |n (with n a non-negative integer) with En = (n + 1/2)ω.e.3
The real scalar ﬁeld
We have expanded the (classical) ﬁeld in plane wave solutions. (7.32) (7. φ(x) = (2π)3 2Ek where the Fourier coeﬃcients a(k) and a† (k) are now operators.27) 2 (2π)3 It is easy to check that these equations can be inverted (see Exercise 2. From the normalizations one obtains a |n = √ n + 1 |n + 1 and a|n = n |n − 1 . i.4 for the classical ﬁeld) a(k) a† (k ′ ) = = d3 x ei k·x i ∂0 φ(x).33)
. which represents the ’number of particles’ with momentum k. (2π)3 2Ek
(7.31)
The hamiltonian can be rewritten in terms of a number operator N (k) = N (k) = a† (k)a(k).Quantization of ﬁelds
57
Deﬁning states |n as eigenstates of N with eigenvalue n. φ(x )]x0 =x′0 = [Π(x).30)
[φ(x). but one needs to realize that in that case k 0 = Ek = k2 + M 2 .29)
It is straightforward to prove that the equal time commutation relations between φ(x) and Π(x′ ) are equivalent with ’harmonic oscillator . e.

The 1-particle wave function e−ip·x is obtained via 0|φ(x)|p p|φ(x)|0 = = 0|φ+ (x)|p = d3 k 0|a(k)a† (p)|0 e−i k·x = e−i p·x (2π)3 2Ek (7.38)
d4 k 2π δ(k 2 − M 2 ) θ(k 0 ) |k k| = (2π)4
(7. assuring that the vacuum (by deﬁnition) has eigenvalue 0!. d3 k a(k) e−i k·x .34)
where V = (2π)3 δ 3 (0) is the space-volume. This term will be adressed below. and multiparticle states |(k1 )n1 (k2 )n2 . writing all annihilation operators to the right of the creation operators. . (2π)3 (7. . (2π)3 2Ek (7. φ(x) φ+ (x) φ− (x) = = = φ+ (x) + φ− (x).43)
The normal ordered product can be expressed as : φ(x)φ(y) := φ+ (x)φ+ (y) + φ− (x)φ+ (y) + φ− (y)φ+ (x) + φ− (x)φ− (y).39)
The problem with the vacuum or zero-point energy. (2π)3 2Ek
(7. (2π)3 2Ek (7. The rest of the states are then obtained from the groundstate via the creation operator. k 0 = Ek = k2 + M 2 ).
. a(k)|0 = 0.45) (7.42) (7. in ﬁeld theory also referred to as vacuum energy Evac = 1 V 2 d3 k Ek . deﬁning particle states |k = |k (with positive energy. .36)
and satisfying the completeness condition 1 =
k|k ′ = (2π)3 2Ek δ 3 (k − k′ ).46) (7.e. is solved by subtracting it as an (inﬁnite) constant.40)
This procedure is known as normal ordering. i. which amounts to redeﬁning H as H= d3 x : H (x) := d3 k Ek N (k).37) (7.44)
p|φ− (x)|0 = ei p·x . |k = a† (k)|0 . For the purpose of normal ordering it is convenient to decompose the ﬁeld in positive and negative frequency parts. |0 . = normalized as (a† (k1 ))n1 (a† (k2 ))n2 √ √ . d3 k |k k|.Quantization of ﬁelds
58
where the necessity to commute a(k)a† (k) (as in the case of the quantum mechanics case) leads to a zero-point energy. which now contains an inﬁnite number of oscillators. n1 ! n2 ! (7. For the momentum operator one has Pi = d3 x Θ0i (x) = 1 2 d3 x ∂ {0 φ ∂ i} φ = d3 k k i N (k). (2π)3 2Ek d3 k a† (k) ei k·x . . Just as in the case of the harmonic oscillator it is essential (axiom) that there exists a ground state |0 that is annihilated by a(k). (2π)3 2Ek (7.35)
where the vacuum contribution disappears because of rotational symmetry.41) (7.

i∆+ (x) i∆− (x) = = = d4 k θ(k 0 ) 2π δ(k 2 − M 2 ) e−i k·x .54) = i∆(x − y) = i (∆+ (x − y) + ∆− (x − y)) .56) = −δ 3 (x). The last item to be checked for the scalar ﬁeld are the causality condition. (7.55) d4 k ǫ(k 0 ) 2π δ(k 2 − M 2 ) e−i k·x . (2π)3 2Ek ≡ i∆− (x − y). In order to calculate [φ(x).52)
The result for the invariant commutator function is [φ(x). x) = 0 and hence ∆(x) = 0 for x2 < 0. i.18). (7. (2π)4 d4 k − θ(k 0 ) 2π δ(k 2 − M 2 ) ei k·x (2π)4 d4 k − θ(−k 0 ) 2π δ(k 2 − M 2 ) e−i k·x = −i∆+ (−x) (2π)4 (7. φ− (y)]
≡ i∆+ (x − y). φ(y)] which has the properties (i) i∆(x) = i∆+ (x) + i∆− (x) can be expressed as i∆(x) = where ǫ(k 0 ) = θ(k 0 ) − θ(−k 0 ).53)
t=0
. (iv) The equal time commutation relations follow from ∂ ∆(x) ∂t (v) For M = 0. d3 k ei k·(x−y) = −i∆+ (y − x). a† (k ′ )] 3 2E ′ (2π) k
[φ− (x).17) and (7. (ii) ∆(x) is a solution of the homogeneous Klein-Gordon equation. 2π (7.47)
′ d3 k ′ e−i k·x+i k ·y [a(k). φ(y)] consider
µν
[φ+ (x). (iii) ∆(0. φ+ (y)]
d3 k e−i k·(x−y) .Quantization of ﬁelds
59
In order to ensure the consistency of the theory it is necessary to check that the operators P ν and M obtained from the conserved currents Θρσ and M µρσ are indeed the generators of the Poincar´ e group. = − (2π)3 2Ek
(7.49) (7.51) (7. that they satisfy the required commutation relations in Eqs (7. ∆(x) = − ǫ(x0 ) δ(x2 ). (2π)4 (7.50)
or as integrals over d4 k.48) (7. d3 k = (2π)3 2Ek =
(7.e.

4
The complex scalar ﬁeld
In spite of the similarity with the case of the real ﬁeld. which is equivalent to the relations (only nonzero ones) [a(k).63)
i.59) (7.57) (7.5
The Dirac ﬁeld
L =
↔ i ψγ µ ∂µ ψ − M ψψ. extending it with the charge operator and the introduction of particle and antiparticle operators.61)
(7. a† (k ′ )] = [b(k). The ﬁeld satisﬁes the Klein-Gordon equation and the density current (U (1) transformations) and the energy-momentum tensor are jµ Θµν The quantized ﬁelds are written as φ(x) = φ† (x) = d3 k a(k) e−i k·x + b† (k) ei k·x . (2π)3 2Ek
(7. (2π)3 2Ek (7.65)
7. (2π)3 2Ek (7. Also the charge operator requires normal ordering (in order to give the vacuum eigenvalue zero). ∂0 φ† (y)]x0 =y0 = i δ 3 (x − y). b† (k ′ )] = (2π)3 2Ek δ 3 (k − k′ ).Quantization of ﬁelds
60
7.60) = i φ∗ ∂µ φ. φ† (y)] = i∆(x − y). = ∂{µ φ∗ ∂ν} φ − L gµν . 2
From the lagrangian density
(7. The hamiltonian is as before given by the normal ordered expression H = = = d3 x : Θ00 (x) : d3 k Ek : a† (k)a(k) + b(k)b† (k) : (2π)3 2Ek d3 k Ek a† (k)a(k) + b† (k)b(k) .66)
.58)
and satisfy the equal time commutation relation (only nonzero ones) [φ(x). we will consider it as a repetition of the quantization procedure.
↔
(7.64)
The commutator of φ and φ† is as for the real ﬁeld given by [φ(x). Q = = = i d3 x : φ† ∂0 φ − ∂0 φ† φ(x) : d3 k : a† (k)a(k) − b(k)b† (k) : (2π)3 2Ek d3 k a† (k)a(k) − b† (k)b(k) . (2π)3 2Ek d3 k b(k) e−i k·x + a† (k) ei k·x . particles (created by a† ) and antiparticles (created by b† ) with the same momentum contribute equally to the energy.62) (7. (7.e.

s) e−i k·(x−y) j
s † + vi (k.77)
Note that achieving normal ordering. s)vj (k. s).79)
Also for the ﬁeld and the canonical conjugate momentum anticommutation relations are considered. The solution is the introduction of anticommutation relations. The quantized ﬁelds are written ψ(x) =
s
d3 k b(k. s)u(k.67) i ψ / ψ − M ψψ gµν . (2π)3 2Ek d3 k b† (k.68)
The canonical momentum and the hamiltonian are given by Π(x) H (x) = = δL = i ψ † (x). (7. s)u† (k. {b(k. b† (k ′ . ∂ 2
↔
(7.69)
(7. then leads to additional minus signs and H Q = = d3 k Ek b† (k)b(k) + d† (k)d(k) (2π)3 2Ek d3 k b† (k)b(k) − d† (k)d(k) . jµ Θµν = = ψγµ ψ.70)
where the last line is obtained by using the Dirac equation. s)ei k·x + d(k. s) ei k·(x−y) x0 =y 0
. s)v(k. (7. s)e−i k·x . ψj (y)}x0 =y0
=
d3 k (2π)3 2Ek
ui (k.75) (7. (2π)3 2Ek (7. s)ei k·x .74) (7.
.Quantization of ﬁelds
61
the conserved density and energy-momentum currents are easily obtained. s)e−i k·x + d† (k. s′ )} = {d(k. u v (2π)3 2Ek
(7. s)¯(k. ˙ δ ψ(x) ↔ i Θ00 (x) = − ψγ i ∂i ψ + M ψψ 2 = iψγ 0 ∂0 ψ = iψ † ∂0 ψ.71) (7. s)¯(k. s).78) (7. s′ )} = (2π)3 2Ek δ 3 (k − k′ ) δss′ . ↔ i ψγµ ∂ν ψ − 2 (7. interchanging creation and annihilation operators.72)
ψ(x) =
s
In terms of the operators for the b and d quanta the hamiltonian and charge operators are (omitting mostly the spin summation in the rest of this section) H = = Q = = d3 x : ψ † (x) i∂0 ψ(x) : d3 k Ek : b† (k)b(k) − d(k)d† (k) : (2π)3 2Ek d3 x : ψ † ψ : d3 k : b† (k)b(k) + d(k)d† (k) : .76)
which seems to cause problems as the antiparticles (d-quanta) contribute negatively to the energy and the charges of particles (b-quanta) and antiparticles (d-quanta) are the same. (2π)3 2Ek (7.73) (7. i.e.
† {ψi (x). d† (k ′ .

but we must restrict ourselves to the weaker condition B|∂µ Aµ |A = 0.
−λ(∂ρ Aρ ). 4 Π0 Πi δL = 0.6
The electromagnetic ﬁeld
1 L = − Fµν F µν .86) (7. we would have obtained [ψi (x). but has the problem of being noncovariant.81)
where i∆ = i∆+ + i∆− is the same invariant commutator function as encountered before. ψ j (y)] = (i/x + M )ij i∆1 (x − y).83)
From the lagrangian density
the canonical momenta are = = (7.88)
. ψ j (y)} at arbitrary times one has {ψi (x). ∂ (7. as the vanishing of Π0 induces a constraint. It is possible to continue in a covariant way with the lagrangian 1 λ L = − Fµν F µν − (∂µ Aµ )2 .Quantization of ﬁelds
62
Using the positive and negative energy projection operators discussed in section 4.87)
If one wants to impose canonical commutation relations ∂µ Aµ = 0 cannot hold as an operator identity.
(7. For the scalar combination {ψi (x). it implies the Lorentz constraint and leads to the canonical momenta Π0 Π
i
= =
E i. ψj (y)}x0 =y0
=
d3 k (/ + M )il (γ0 )lj e−i k·(x−y) k (2π)3 2Ek + (/ − M )il (γ0 )lj ei k·(x−y) k
x0 =y 0
=
d k 2Ek (γ0 γ0 )ij ei k·(x−y ) (2π)3 2Ek (7. ψ j (y)} = d3 k (/ + M )ij e−i k·(x−y) + (/ − M )ij ei k·(x−y) k k (2π)3 2Ek d3 k e−i k·(x−y) + −ei k·(x−y) = (i/x + M )ij ∂ (2π)3 2Ek = (i/x + M )ij i∆(x − y).85)
which reﬂects the gauge freedom. ˙ δ A0 δL = F i0 = E i . one has
† {ψi (x).
7.80)
3
= δ 3 (x − y) δij . (7. ˙ δ Ai (7.84) (7. 4 2 This gives the equations of motion discussed before. which however has wrong causality properties! Therefore the relation between spin and statistics is required to get micro-causality (which is also known as the spin statistics theorem). ∂
(7.82) where i∆1 = i∆+ −i∆− . When we would have started with commutation relations for the ﬁeld ψ and the canonical momentum.

(7. Πν (y)]x0 =y0 = i gµν δ 3 (x − y). containing a time-like photon. 0. Πν (y)] = 0.89)
with four independent vectors ǫµ . i.90)
where Πν = F ν0 − λ g ν0 (∂ρ Aρ ) and we have furthermore [Aµ (x).e. (7. λ)c(k. a longitudinal photon and two transverse photons. and are equivalent with [c(k. 0. In fact the commutation relations imply ˙ [Aµ (x). what are [ri . where Aµ + + is the part of the vector ﬁeld containing the annihilation operators. The quantized ﬁeld is expanded as Aµ (x) = d3 k (2π)3 2Ek
(λ) 3
ǫ(λ) (k) c(k.93)
which does exhibit problems with the time-like photon. pj ]P . λ). Aν (y)] = [Πµ (x). i. λ′ )] = −g λλ 2Ek (2π)3 δ 3 (k − k′ ).Quantization of ﬁelds
63
for physical states |A and |B . c† (k ′ . Aν (y)]x0 =y0 = i gµν δ 3 (x − y). j = 1. which reproduce the Hamilton equations. and [ℓi .95)
Exercises
Exercise 7.94)
Choosing k µ = (|k 3 |. (b) Proof for a quantity O(p. [ℓi . λ)e−i k·x + c† (k. (c) Calculate the Poisson brackets for the coordinates x and the standard conserved quantities p and ℓ.e. q) that dO/dλ = [O. µ
λ=0
(7. where Q is the conserved quantity in Eq. Q]P . 0)c(k. (7. (d) Use Poisson brackets to calculate q and p. λ)|A = 0. k 3 ) this reads |k 3 |c(0) (k) − k 3 c(3) (k) |A = 0 c(0) (k) ∓ c(3) (k) |A = 0. The canonical equal time commutation relations are [Aµ (x).91)
(7. λ) − c† (k. λ)ei k·x . 7. The hamiltonian in terms of the creation and annihilation operators is (after normal ordering) given by H= d3 k Ek (2π)3 2Ek
3
λ=1
c† (k.
′
(7. These problems are solved by the Lorentz constraint between physical states given above. µ
λ=0
(7.11. pj ]P . 2. 6. ˙ ˙
. 0) . It gives
3
k µ ǫ(λ) (k)c(k. for which it is suﬃcient that ∂µ Aµ |A = 0.1 (optional)
If you haven’t seen Poisson brackets in classical mechanics you should (a) Proof that Poisson brackets (Eq. rj ]P . one time-like photon by itself is not allowed! This solves the problems with the normalization and the negative energies.92)
Note that for the transverse states there are no problems with the normalization and the statistics of the states as −g ij = δ ij for i.3) satisfy the conditions of a Lie algebra.

2
Prove that the equal time commutation relations between quantum ﬁelds φ(x) and Π(x′ ) are equivalent with the commutation relations between a(k) and a† (k ′ )
Exercise 7.3
(a) Show that the correct implimentation of symmetries in the Hilbert space.Quantization of ﬁelds
64
Exercise 7. (2 + M 2 ) ∆inhom (x) = −δ 4 (x). (b) Show that ∆inhom (x) = 1 (2π)4 d4 k e−ikx 1 k2 − M 2
is a solution of the inhomogeneous Klein-Gordon equation. Is the subtraction of ’zero point’ contributions essential in this check?
µ
Pµ
φ(x) ei a
µ
Pµ
= φ(x − a)
Exercise 7. ∆hom (Λx) = ∆hom (x) if f is an invariant function.4 (Green’s functions)
(a) Show that ∆hom (x) = 1 (2π)4 d4 k e−ik·x δ(k 2 − M 2 )f (k)
is a solution of the homogeneous Klein-Gordon equation. Pµ ] for the (real) scalar ﬁeld using the expressions for the ﬁelds and momentum operator in terms of creation and annihilation operators. (c) What are the poles in the integral under (b) in the (complex) k0 plane? (d) Depending on the paths in the k0 plane going from k0 = −∞ to k0 = +∞ one can distinguish four diﬀerent Green’s functions (solutions to the above inhomogeneous Klein-Gordon equation):
k0 -E E
∆ R (retarded) ∆ C (causal) ∆ ∆A (advanced)
Show that by appropriate shifts of the poles into the complex plane. f (Λk) = f (k) for Lorentz transformations Λ. e−i a requires [φ(x). (Similarly one can show that for the advanced Green’s function ∆A (x) = 0 if x0 > 0). (b) Check the above commutation relation [φ(x). It is invariant. we can write for ∆R an integral where the k 0 integration remains along the real axis. (2 + M 2 ) ∆hom (x) = 0.
. Pµ ] = i∂µ φ(x). (e) Show that ∆R (x) = 0 if x0 < 0. Give also the expression for ∆C . ∆R = 1 (2π)4 d4 k e−ikx k2 1 − M 2 + iǫk 0
(with the implicit prescription to take ǫ → 0 after integration).

(i) (Optional) Give similar expressions for the other Green’s functions under (d) in terms of the homogeneous solutions under (f).Quantization of ﬁelds
65
(f) By performing the k0 integration over any closed path C (encircling one or both of the ples) show that 1 1 d4 k e−ikx 2 ∆(x) = − 4 (2π) C k − M2 are homogeneous solutions for the contours:
k0 −E ∆_ ∆+ E
∆
One way of showing this is to write these contours as integrals of the form under (a). f (k) ∝ θ(k 0 ). or ǫ(k 0 ) = θ(k 0 ) − θ(−k 0 ). In that case one actually needs discontinuous functions f (k) such as the step function. (g) Show that the homogeneous solution ∆ in (f) satisﬁes ∆(0. (h) Show that the causal Green’s function ∆C (x) = θ(x0 ) ∆+ (x) − θ(−x0 ) ∆− (x).
. x) = 0 and argue that one can use Lorentz invariance to show that ∆(x) = 0 for x2 < 0.

Chapter 8
Discrete symmetries
In this chapter we discuss the discrete symmetries.
8. r) −→ xµ ≡ xµ = (t. x
(8. Table 8.1
Parity
xµ = (t. The consequences of P. −r). ˜ (8. (iγ µ ∂µ − M ) ψ(x) = 0.2)
where ηP is the intrinsic parity of the ﬁeld and A is a 4 × 4 matrix acting in the spinor space. T. Both ψ p and ψ satisfy the Dirac equation. We can determine A. T and C for classical quantities is shown in the table 1.1)
The parity operator transforms
We will consider the transformation properties for a fermion ﬁeld ψ(x). and C
quantity t r xµ E p pµ L s ˆ λ = s·p
P t -r xµ ≡ xµ ˜ E -p pµ ˜ L s −λ
T -t r −˜µ x E -p pµ ˜ -L -s λ
C t r xµ E p pµ L s λ
66
.1: The behavior of classical quantities under P. starting with the Dirac equation for ψ(x). time reversal (T) and charge conjugation (C). parity (P). writing
−1 ψ(x) −→ Pop ψ(x)Pop = ηP Aψ(˜) ≡ ψ p (x).

−λ)v(k. m). m). the above operation reverses the sign of λ). x Therefore γ0 ψ(˜) is again a solution of the Dirac equation and we have x ψ p (x) = γ0 ψ(˜). x µ (iγ ∂µ − M ) γ0 ψ(˜) = 0. ˜ Pop d(k.13) ψ(x) −→ ψ c (x) = ηC C ψ (x).
8. λ)u(k. if helicity λ is used instead of the z-component of the spin m. λ)v(k. −λ)u(k. op P
(8. m) γ0 v(k. λ) e−ik·x − d† (k. γ x
(8. (2π)3 2Ek (8. m) = = ˜ u(k.e.
(i˜ µ ∂µ − M ) ψ(˜) = 0. λ)γ0 v(k. λ) P −1 = −η ∗ d(k.8)
3
=
λ
=
λ
=
λ
From this one sees immediately that
−1 ˜ Pop b(k.9) (8.6)
(check this for the standard representation. −λ) eik·x 3 2E (2π) ˜ k d3 k ˜ ˜ ηP b(k. one can also consider the scalar ﬁeld and vector ﬁelds. The result is ψ p (x)
−1 = Pop ψ(x) Pop = ηP γ0 ψ(˜) x
(8. choosing ηP is real (ηP = ±1) particle and antiparticle have opposite parity. x It is straightforward to apply this to the explicit ﬁeld operator ψ(x) using γ0 u(k. λ) eik·x . λ)v(k. λ) Pop = ηP b(k.11) (8.10)
i. λ)γ0 u(k.Discrete symmetries
67
After parity transforming x to x the Dirac equation becomes after some manipulations ˜ ˜ x iγ µ ∂µ − M ψ(˜) = 0. −λ).
iγ µ† ∂µ − M ψ(˜) = 0. −λ) e−ik·x − d† (k. ˜ −v(k. −λ) eik·x (2π)3 2Ek ˜ d3 k ˜ ˜ ˜ ˜ ηP b(k.12)
and for the vector ﬁeld
−1 Aµ (x) −→ Pop Aµ (x) Pop = −Aµ (˜).4)
(8.7)
=
λ
d k x x ηP b(k. λ) eik·˜ (2π)3 2Ek d3 k ˜ ˜ ˜ ˜ ηP b(k. For the scalar ﬁeld we have seen
−1 φ(x) −→ Pop φ(x) Pop = ηP φ(˜). x
(8.3)
(8. x
The latter behavior of the vector ﬁeld will be discussed further below.
. λ)u(k. −λ) e−ik·x − d† (k.5) (8. −λ). In the same way as the Fermion ﬁeld. λ) e−ik·˜ + d† (k.2
Charge conjugation
We have already seen the particle-antiparticle symmetry with under what we will call charge conjugation the behavior T (8.

we start with the complex conjugated x Dirac equation for ψ. x
(8. (2π)3 2Ek (8.
. λ). m). i. writing
−1 ψ(x) −→ Top ψ(x)Top = ηT Aψ(−˜) ≡ ψ t (−˜). λ)C uT (k.
(i˜ µ∗ ∂µ − M ) ψ(−˜) = 0.
(8.22)
= Aψ|Aφ
∗
is antilinear if A(λ|φ + µ|ψ ) = λ∗ A|φ + µ∗ A|ψ . it is antilinear1 . λ) e−ik·x + b† (k. λ)v(k.16)
d k ηC d(k. m) ¯ C v T (k. x
iγ µT ∂µ − M ψ(−˜) = 0.3
Time reversal
xµ = (t.17)
3
=
λ
This shows that
−1 Cop b(k. to ﬁnd ψ t (−˜) that is a solution of the Dirac equation. antilinear operator is anti-unitary if A† = A−1 . The same relations hold for helicity states. x
1A 2 An
−iC −1 γ µ C∂µ − M ψ(−˜) = 0. λ) e−ik·x + b† (k. x (8. r) −→ −˜µ ≡ −xµ = (−t.14) (8. λ) eik·x .21)
where A is a 4 × 4 matrix acting in the spinor space.Discrete symmetries
68
the latter being also a solution of the Dirac equation. m). λ). The action on the spinors (using C = iγ 2 γ 0 = −iρ1 σ 2 in standard representation) gives C uT (k. As time reversal will transform ’bra’ into ’ket’. λ) eik·x ¯ ¯ (2π)3 2Ek d3 k ηC d(k. as discussed in section 4). λ)C v T (k. r). ˜ x −(iγ µ )∗ ∂µ − M ψ(−˜) = 0.19)
Cop d(k. Top |φ = φt | = (|φt )∗ . u(k. x x
(8. λ) Cop = ηC d(k.20)
The time reversal operator transforms
We will again consider the transformation properties for a fermion ﬁeld ψ(x). x µ (iγ ∂µ − M ) γ5 Cψ(−˜) = 0. m) ¯ = = v(k.e.
8. λ)u(k. ((iγ µ )∗ ∂µ − M ) ψ(x) = 0. One has Aφ|Aψ = φ|ψ
∗
= ψ|A† Aφ = ψ|φ . γ x
i(γ5 C)−1 γ µ γ5 C∂µ − M ψ(−˜) = 0.15)
(where one must be aware of the choice of spinors made in the expansion. λ)
−1 Cop
=
∗ ηC
b(k. Therefore ψ c (x) = =
λ −1 ¯ Cop ψ(x) Cop = ηC C ψ T (x)
(8. The (time-reversed) Dirac equation becomes. For a quantized ﬁeld one has −1 ∗ −1 Top fk (x)bk Top = fk (x)Top bk Top . (8.18) (8. Norm conservation requires Top to be anti-unitary2 .

λ)v ∗ (k.31) (8. λ) ψ t (x)
−1 Top ψ(x) Top = i ηT γ5 Cψ(−˜) x
= =
˜ u∗ (k. ∗ ˜ v (k. λ a |a. λ) e−ik·x (2π)3 2Ek (8. λ |¯. p. λ)v ∗ (k. λ) e−ik·x 3 2E (2π) k d3 k ˜ ˜ ηT b(k. x (8. Since there are 16 independent 4 × 4 matrices. λ)u∗ (k. −λ |¯. there are 16 independent of these bilinear combinations.2: The transformation properties of physical states for particles (a) and antiparticles (¯). λ
Therefore γ5 Cψ(−˜) is again a solution of the (ordinary) Dirac equation and we can choose (phase is x convention) ψ t (x) = i γ5 C ψ(−˜). op T
(8. λ). λ) eik·˜ + d† (k.29)
In table 2 the behavior of particle states under the various transformations has been summarized. λ). λ a P |a. a state |a. λ) i γ5 C v(k. λ)iγ5 Cv(k. ˜ Top d(k. λ)u∗ (k. p. λ| a. λ)v ∗ (k.23) In the standard representation iγ5 C = σ2 and it is straightforward to apply this to the explicit ﬁeld operator ψ(x) using i γ5 C u(k.27)
3
=
λ
=
λ
=
λ
From this one obtains
−1 ˜ Top b(k.
(8. −p.Discrete symmetries
69
Table 8.34)
. λ) eik·x + d† (k. −p. λ)u∗ (k. Therefore one has A|k = A|b† (k)|0 = A|T † T b† (k) T † T |0 ∗ = At |b† (k)|0 ∗ ˜ t = k|At .25)
(check this for the standard representation). ˜ = 0|b(k)|A
8. Note that applying an anti-unitary transformation such as Top one must take for the matrix element ˜ the complex conjugate. −λ a T a. λ) eik·x + d† (k.32) (8. λ| ¯ C |¯.4
Bi-linear combinations
In quantities such as currents and lagrangians often bilinear combinations of spinor ﬁelds are encountered. λ) e−ik·˜ (2π)3 2Ek d3 k ˜ ˜ ˜ ˜ ηT b(k. λ) eik·x + d† (k. λ) T −1 = η ∗ d(k. λ). −p. −p. p. λ).28) (8.30) (8. λ)iγ5 Cu(k.33) (8. p. The result is = =
λ
(8.26)
d k x x ηT b(k. λ) e−ik·x 3 2E (2π) k ˜ d3 k ˜ ˜ ˜ ˜ ηT b(k. λ) Top = ηT b(k.24) (8. They are the following S(x) V µ (x) T µν (x) Aµ (x) P (x) ¯ = ψ(x)ψ(x) (scalar) ¯ = ψ(x)γ µ ψ(x) (vector) µν ¯ = ψ(x)σ ψ(x) (tensor) µ ¯ = ψ(x)γ5 γ ψ(x) (axial vector) ¯ = iψ(x)γ5 ψ(x) (pseudoscalar) (8.

38)
(8.. V µ (x) = ei Pop ·x V µ (0)e−i Pop ·x . γ µ . s .37)
(8. 2M (8.. p′ . for the vector current V µ (x). Note that the lagrangian density L (x) → L (−x) under Θ. As the coupling of the photon ﬁeld to fermions is given by an interaction term in the ¯ lagrangian of the form : ψ(x)γ µ ψ(x) : Aµ (x) and behaves as a scalar one sees immediately that the ¯ photon ﬁeld Aµ (x) behaves in the same way as the vector combination ψ(x)γ µ ψ(x). (q 2 ). C. C. (8.39)
where Γµ (p′ . In many applications the expectation values of currents are needed. e.5
Form factors
Currents play an important role in ﬁeld theory. F. called form factors.3: The behavior of the independent bilinear combinations of fermi ﬁelds under P. and T P S(˜) x Vµ (˜) x Tµν (˜) x −Aµ (˜) x −P (˜) x C S(x) −V µ (x) −T µν (x) Aµ (x) P (x) T S(−˜) x Vµ (−˜) x −Tµν (−˜) x Aµ (−˜) x −P (−˜) x Θ = PCT S(−x) −V µ (−x) T µν (−x) −Aµ (−x) P (−x)
S(x) V µ (x) T µν (x) Aµ (x) P (x)
The matrix σ µν ≡ (i/2)[γ µ .
8. in this case only q 2 = (p′ − p)2 . and T. (8. F.Discrete symmetries
70
Table 8. p)u(p) e−i (p−p )·x . For instance for nucleons one has Γµ (p′ . γ5 γ µ or γ5 ).36)
(8. ¯ In general the expectation value between momentum states can be more complicated. As an example consider the vector current for a point fermion. γ ν ]. momenta (pµ or p′µ ) or constant tensors (ǫµνρσ or gµν ). Applying the results from the previous sections it is straightforward to determine the behavior of the combinations under P. p) = γµ F1 (q 2 ) + iσµν q ν F2 (q 2 ). p) can be built from any combination of Dirac matrices (1. σ µν .35)
The x-dependence can be accounted for straightforwardly using translation invariance. of which the expectation value between momentum eigenstates can be simply found. There are several other terms that also have the correct tensorial behavior such as (pµ + p′µ ) γ5 σ µν qν q µ F3 (q 2 ).g. ¯
′ ′
(8. s′ |V µ (x)|p. This implies p′ |V µ (x)|p = e−i (p−p )·x p′ |V µ (0)|p .41) 2M 2M
. (q 2 ). p′ |V µ (x)|p = u(p′ )Γµ (p′ . ¯ V µ (x) =: ψ(x)γ µ ψ(x) :. as well as under the combined operation Θ = PCT (see Table 3). The 16 combinations of Dirac matrices appearing above are linearly independent.40)
where the coeﬃcients are functions of invariants constructed out of the momenta. p′ |V µ (0)|p = u(p′ )γ µ u(p).

p) that γ0 Γµ† (p. p)γ0 . p)γ0 u(p). C and T invariance. e.42)
where q = p′ − p. p′ )γ0 = Γµ (p′ . or relations that follow from the equations of motion. ¯ p ˜
Therefore parity invariance implies for the structure of Γµ (p′ . • Parity p′ |V µ (x)|p = ei q·x u(p′ )Γµ (p′ .g. γ5 σµν = (1/2)ǫµνρσ γ ρ γ σ . the Gordon decomposition u(p′)γ µ u(p) = ¯ 1 u(p′ ) [(p′ + p)µ + iσ µν qν ] u(p). p) that Γµ (p′ . p)u(p) ¯
† † = p′ |Pop Pop V µ (x)Pop Pop |p ˜x = p′ |Vµ (˜)|˜ = ei q ·˜ u(˜′ )Γµ (˜′ . p)(iγ5 C)u(p). p′ )u(p′ )
= u† (p′ )Γµ† (p. ¯ µ p ˜
Therefore time reversal invariance implies for the structure of Γµ (p′ . p) that Γµ (p′ . p) = (i γ5 C)Γ∗ (˜′ .44)
= p′ |Vµ (−˜)|˜ ˜ x p
∗
˜x = ei q ·˜ [¯(˜′ )]∗ Γ∗ (˜′ . p).1
The matrix element of the electromagnetic current between nucleon states is written as < p′ |Jµ (x)|p >= e−i(p−p )·x u(p′ )Γµ u(p) ¯ Here Γµ iσµν q ν F2 (q 2 ) 2M pµ + pµ ′ = γµ H1 (q 2 ) − H2 (q 2 ) 2M = γµ F1 (q 2 ) +
′
. e. µ p ˜ (8.Discrete symmetries
71
but that are eliminated because of relations between Dirac matrices. • Hermiticity p′ |V µ (0)|p = p′ |V µ† (0)|p = u(p′ )Γµ (p′ . p)(i γ5 C). ¯ 2M (8. p)u(p) ¯
† † = p′ |Top Top V µ (x)Top Top |p ∗
(8. p′ )u(p′ )) u =
∗ †
u† (p)γ0 Γµ (p.45)
Exercises
Exercise 8. p)u(p) ¯ = p|V µ (0)|p′ ∗ = (¯(p)Γµ (p. p) = γ0 Γµ (˜′ . or relations based on hermiticity of the operator or P .43)
= ei q·x u(p′ )γ0 Γµ (˜′ . Therefore hermiticity implies for the structure of Γµ (p′ . p ˜ • Time reversal p′ |V µ (x)|p = ei q·x u(p′ )Γµ (p′ . p)u∗ (˜) up p µ p ˜
= ei q·x u(p′ ) (iγ5 C)Γ∗ (˜′ .g. p′ )γ0 u(p). p)u(˜) ˜ x p ¯ p p ˜ p
(8.

p′ = (E. |q|). (b) Show that such a term is also not allowed by time-reversal symmetry.
Exercise 8. 0. (c) Show that hermiticity of the current requires that F1 en F2 are real. 0. |q|/2) with E 2 = M 2 + |q|2 /4 and express them in terms of the Sachs form factors depending on Q2 = −q 2 .3
(a) Show that because of parity conservation no term of the form γ5 can appear in the current matrix element. 0. −|q|/2). σµν q ν F3 (q 2 ) 2M
.2
(a) Calculate the current matrix element (using the explicit fermion spinors) in the Breit-frame in which q = (0. p = (E.Discrete symmetries
72
(a) Use the Dirac equation to prove the Gordon-decomposition u(p′)γ µ u(p) = ¯ 1 u(p′ )[(p′ + p)µ + iσ µν (p′ − p)ν ]u(p) ¯ 2M
(b) Give the relation between Hi en Fi . 4M 2 = F1 + F2 . GE GM = F1 − Q2 F2 .
Exercise 8. 0. 0. 0. L
int
= ejµ Aµ
the charge is given by e GE (0) and the magnetic moment by e GM (0)/2M . it would correspond to an electric dipole moment d = −e F3 (0)/2M (Interaction term d σ · E). (c) Show that if a term of this form would exist.
(b) Show that in the interaction with the electromagnetic ﬁeld.

≡ 0. t0 ) ∂t (9. [AH . (9. in which the operators are time-independent. (9. t
′
|q .
(9.5) (9.3) (9. t′ ≡ q ′ |q ′ . t0 ). |ψS (t) = U (t.1 Time evolution as path integral
U (t.6)
Of these the Heisenberg picture is most appropriate for quantum ﬁeld theory since the ﬁeld operators do depend on the position and one would like to have position and time on the same footing. i ∂ |ψH ∂t ∂ i AH ∂t ≡ = 0. AS (t) = AS and the states o are time dependent. t0 )|ψS (t0 ) .
(9. i ∂ |ψS ∂t ∂ i AS ∂t = H |ψS .Chapter 9
Path integrals and quantum mechanics
9. AH (t) = U −1 (t. t
′
QH (t′ )|q ′ .1)
The time evolution from t0 → t of a quantum mechanical system is generated by the Hamiltonian. t ≡ q|q. t′ .4) ∂ U (t. in which the states are time-independent. Consider the two (time-independent) Heisenberg states: |q.2)
(ii) Heisenberg picture.7)
73
.
QH (t)|q. and the operators are time-dependent. H]. t . |ψH (t) = |ψH . t0 ) = H U (t. t0 ) AH (t0 ) U (t. t0 ) = e−i(t−t0 )H or i Two situations can be distinguished: (i) Schr¨dinger picture.

with the correction1 being of order (∆τ )2 . Q) = K(P ) + V (Q) = (P 2 /2M ) + V (Q) one can split e−i H∆τ ≈ e−i K∆τ e−i V ∆τ .8)
Choose t as the starting point with |q = |q. H(P.qj )∆τ e e 2π dpj i pj (qj+1 −qj )−i ∆τ H(pj . B] + . . . QS |q ′ ≡ q ′ |q ′ . t′ |q. qj )]) . tj+1 |qj . .10)
dqn−1 q ′ |e−i H∆τ |qn−1 qn−1 | . Q) expressed in terms of the operators P and Q. tj ≈ = qj+1 |e−i K∆τ e−i V ∆τ |qj dpj qj+1 |pj pj |e−i K∆τ e−i V ∆τ |qj .
′
(9. t = q ′ |e−i H(t −t) |q .13)
At least for a simple hamiltonian such as consisting of a kinetic energy term and a local potential. t = = q ′ |e−i n H∆τ |q = q ′ | e−i H∆τ dq1 . ˙ 2π
1 1 1 [A.9)
Dividing the interval from t ≡ t0 to t′ ≡ tn into n pieces of length ∆τ and using completeness (at each time ti ) one writes q ′ . These can be written in coordinate or momentum representation as Q = P = dq |q q q| dq |q −i ∂ ∂q q| = dp |p p p|..15)
By letting the kinetic and potential parts act to left and right respectively one can express the expectation values in terms of integrals containing H(pj . B] + [A. 2π
(9. Then qj+1 .11) (9. .16)
this.14)
. tj+1 |qj . t′ |q.
n
|q (9.
The purpose of this is to calculate the evolution for an inﬁnitesimal time interval. [A. Combining the two terms gives qj+1 . t and QH (t) = QS and study the evolution of the system by calculating the quantum mechanical overlap amplitude q ′ . B]] + [[A. |q1 q1 |e−i H∆τ |q .
(9. The hamiltonian is an operator H = H(P. B].12)
where the transformation between coordinate and momentum space involves q|p = ei p. (9.qj ) e 2π dpj exp (i ∆τ [pj qj − H(pj . which is a hamiltonian function in which the operators are replaced by real-numbered variables.q . eA eB = eC with C = A+B+ (9. qj ). 2 12 12
(9.. 2π (9. tj = = =
1 For
dpj i pj (qj+1 −qj ) −i H(pj . use the Campbell-Baker-Hausdorﬀ formula.Path integrals and quantum mechanics
74
and Schr¨dinger states o |q |q ′ QS |q ≡ q|q .

. y) in the above examples can be (hermitean) operators acting on the functions. i. q1 . dx dy α(x) K(x.2
Functional integrals
In this section I want to give a fairly heuristic discussion of functional integrals.
F3 [α. t =
τ
dq(τ ) dp(τ ) exp i 2π p exp i 2π
t′ t
t′ t
∆τ [pq − H(p. β] = α K β ≡ 1 F4 [α] = exp − α2 2
F5 [α.e. while Dα F [α] =
x
dαx N
F (αx )
(9. What one is after is the meaning of Dα F [α]. .
′ ′ q1 . . etc.
9.17)
=
Dq D
dτ [pq − H(p. . . y) β(y).e. q)] .19)
where F [α] is a functional that represents a mapping from a function space F of (real) functions α (R → R) into the real numbers R. . . y) α(y) . t′ |q1 . Two working approaches for functional integration are the following: (i) the heuristic division of the space on which α acts into cells. α∗ ] = exp (−α∗ K α) ≡ exp −
The kernels K(x. including diﬀerential operators. Before proceeding we also give the straightforward extension to more than one degree of freedom. .Path integrals and quantum mechanics
75
or for the full interval q ′ . F [α] ∈ R. pN . dx dy α∗ (x) K(x. . t N
(9.18)
t
′
=
n=1
Dqn D
pn 2π
exp i
t
dτ
n
pn qn − H(p1 . Examples are F1 [α] = α ≡ F2 [α] = α2 ≡ dx α(x). ˙
where Dq and Dp indicate functional integrals. . .20)
becomes a multidimensional integral. ≡ exp − 1 2 dx α2 (x) . (9. . t′ |q. q)] ˙ (9. qN ) ˙
. The importance of this expression is that it expresses a quantum mechanical amplitude as a path integral with in the integrand a classical hamiltonian function. dx α2 (x). α(x) → αx and F [α] = F (αx ) changes into a multivariable function. . qN . qN . i.

27) (9. (9.
.21)
again a multidimensional integral. In both cases N is an appropriately choosen normalization constant in such a way that the integral is ﬁnite.24) (9.Path integrals and quantum mechanics
76
(ii) Write α in terms of a sum of orthonormal basis functions.28)
Dα Dα∗ exp (−α∗ α − α∗ ω − ω ∗ α) = exp (ω ∗ ω) . (9. 1 = √ det K 1 Dα Dα∗ exp (−α∗ K α) = det K 1 Dα exp − α K α 2 (α real-valued) (α complex-valued) (9.22)
=
x
The last equality is obtained by deﬁning the right measure (normalization N ) in the integration. α(x) = n αn fn and K(x. A useful property of functional integration is the translation invariance.23)
Having deﬁned the Gaussian integral.n fm (x) Kmn fn (y) and consider F [α] = F (αn ) as a multivariable function. Physical answers will usually come out as the ratio of two functional integrals and are thus independent of the chosen measure. y) = ∗ m. the following integrals can be derived for a real symmetric or complex hermitean kernel K. with Dα F [α] =
n
dαn N
F (αn )
(9. Dα F [α] = Dα F [α + β].26)
As an important application of translation invariance we mention the identities 1 Dα exp − α2 − ωα 2 = exp 1 2 ω . Note that procedure (i) is an example of the more general procedure under (ii).25)
(see Exercises). N
(9. Using method 1 one writes 1 Dα exp − α2 2 =
x
dαx N
exp −
1 2
∆x α2 x
x
=
x
dαx 1 exp − ∆x α2 x N 2 1 N 2π ∆x ≡ 1. Consider the gaussian functional as an example. 2 (9.n
dx fm (x)fn (x)
=
n
=
n
dαn 1 exp − α2 N 2 n √ 2π ≡ 1. Using the second method and expanding in a basis set of functions one obtains for the (real) gaussian functional 1 Dα exp − α2 2 =
n
dαn N
exp −
1 αm αn 2 m.

32) (9. It is now easy to check that Dθ∗ Dθ exp (−θ∗ K θ) = det K. Examples of functional diﬀerentiation are
(9.37)
We note that from the ﬁrst to the second line one needs to realize that a pair of diﬀerent Grassmann variables behaves as ordinary complex variables.34) (9.
(9. products of the same pair appear. θ2 = 0.
(9. δα(x) δα(x) δ αKβ = dy K(x.38)
. i. δα(x) or in discretized form δ 1 ∂ ∂ with . y) β(y) δα(x) δ 1 1 exp − α2 = −α(x) exp − α2 . however. dx dy θ∗ (x) K(x. δα(x) 2 2 δ exp (−α β) = −β(x) exp (−α β) . the Gaussian-type functionals. In principle the deﬁnitions of functionals is the same. α(y) = δ(x − y).e. θ∗ ] = exp (−θ∗ K θ) ≡ exp − Integration for Grassmann variables is deﬁned as dθ 1 = 0 and dθ θ = 1. θ∗ ] = exp (−θ∗ θ) ≡ exp − F [θ. −→ fn (x) δα(x) ∂αn ∂αm n Note that δ(x − y) =
n ∗ fn (x) fn (y). which vanish. e.g. In the expansion of the exponential (from second to third line).35)
For applications to fermion ﬁelds. αy = δxy .29)
(9. θη = −ηθ. −→ δα(x) ∆x ∂αx ∂αx δ ∂ ∂ ∗ with .
n θn fn (x)
dx θ∗ (x) θ(x) . δα(x)
(9. we need to consider Grassmann-valued functions involving anticommuting Grassmann variables.Path integrals and quantum mechanics
77
Functional diﬀerentiation is deﬁned as δ . y) θ(y) .31)
δ δ α = dy α(y) = dy δ(x − y) = 1.33) (9.30) (9. αn = δmn .36) where the coeﬃ-
This gives for the Gaussian functional integral after expanding θ(x) = cients θn are Grassmann variables the result Dθ∗ Dθ exp (−θ∗ θ) =
n ∗ dθn dθn ∗ dθn n
exp −
∗ θn θn n
= =
n
∗ dθn exp (−θn θn )
∗ ∗ dθn dθn (1 − θn θn ) = 1. (9. F [θ.

Path integrals and quantum mechanics

78

**Functional diﬀerentiation of Grassmann-valued functions is given by δ , θ(y) δθ(x) ¯ leading for two Grassmann variables θ and η to e.g.
**

∗ ∗ δ e−θ η = θ∗ (x) = θ∗ (x) e−θ η , δη(x) ∗ ∗ δ e−η θ = −θ(x) = −θ(x) e−η θ . δη ∗ (x)

= δ(x − y),

(9.39)

(9.40) (9.41)

The translation invariance property and the ’Fourier transform’ identity of Gaussian integrals, remains in essence the same, e.g. Dθ∗ Dθ exp (−θ∗ θ − θ∗ η − η ∗ θ) = exp (η ∗ η) . (9.42)

9.3

Time ordered products of operators and path integrals

As an example of working with functional integrals consider the expression for K(q ′ , t′ ; q, t) for a lagrangian L(q, q) = 1 q 2 − V (q) and the corresponding hamiltonian H(p, q) = 1 p2 + V (q). The ˙ 2 ˙ 2 expression ′ t 1 p exp i dτ pq − p2 − V (q) , ˙ (9.43) q ′ , t′ |q, t = Dq D 2π 2 t

1 ˙ ˙ ˙ can be rewritten after rewriting the integrand as − 2 (p − q)2 + 1 q 2 − V (q) = − 1 (p − q)2 + L(q, q). 2 ˙ 2 The result is ′ t

q ′ , t′ |q, t =

Dq exp i

t

dτ L(q, q) , ˙

(9.44)

**which was considered the starting point for path integral quantization by Feynman.
**

Note, however, that not always the Dp integration can be removed that easily. A counter example is the lagrangian L(q, q) = 1 q 2 f (q) for which H(p, q) = p2 /[2f (q)]. As discussed for instance in ˙ ˙ 2 Ryder the Dp integration can still be removed but one ends up with an eﬀective lagrangian in the path integral q ′ , t′ |q, t =

t′

Dq exp i

t i 2

dτ Leﬀ (q, q) , ˙

(9.45)

Making use of path integrals it is straightforward to calculate the expectation value q ′ , t′ |Q(s)|q, t of an operator Q(s) if t ≤ s ≤ t′ . By sandwiching the time s in one of the inﬁnitesimal intervals, tj ≤ s ≤ tj+1 , we have q ′ , t′ |Q(s)|q, t =

i

which is of the form Leﬀ (q, q) = L(q, q) − ˙ ˙

δ(0) ln f (q).

dqi q ′ , t′ |qn , tn . . . (9.46)

**× qj+1 , tj+1 |Q(s)|qj , tj . . . q1 , t1 |q, t . Using Q(s)|qj , tj = q(s)|qj , tj , one gets q ′ , t′ |Q(s)|q, t = = Dq D p q(s) exp i 2π
**

t′ t′ t

dτ [pq − H(p, q)] ˙

(9.47) (9.48)

Dq q(s) exp i

t

dτ L(q, q) . ˙

Path integrals and quantum mechanics

79

Deﬁning the time ordered product of operators T Q(t1 ) . . . Q(tn ) ≡ Q(ti1 ) . . . Q(tin ), where ti1 ≥ · · · ≥ tin is a permutation of {t1 , . . . , tn }, one has q ′ , t′ |T Q(t1 ) . . . Q(tn )|q, t = Dq q(ti1 ) . . . q(tin ) exp i

t t′

(9.49)

dτ L(q, q) . ˙

(9.50)

Thus not only the quantum mechanical overlap of states can be calculated via a classical path integral, but also expectation of operators, at least if they appear time-ordered.

9.4

An application: time-dependent perturbation theory

From quantum mechanics one should be familiar with the procedure of time-dependent perturbation theory, the lowest order leading to Fermi’s golden rule. One works in the socalled interaction picture, in which a separation is made of the hamiltonian H = H0 + HI . The fast time evolution is described in H0 while HI is considered as a perturbation. The interaction picture is deﬁned as UI (t′ , t) ≡ e−i H0 (t −t) , ψI (t) ≡ ei H0 t ψS (t) = ei H0 t e−i Ht ψS (0) = e−iHI t ψS (0), AI (t) ≡ ei H0 t AS (t)e−i H0 t = ei H0 t AS e−i H0 t ,

′

(9.51) (9.52) (9.53)

i.e. if HI = 0 it is the Heisenberg picture and the evolution q ′ , t′ |q, t is described through the operators by H0 . The evolution of the (interaction) states is described only by HI , ∂ ψI ∂t ∂ i AI ∂t i and ψI (t′ ) = U (t′ , t) ψI (t), ∂ i U (t′ , t) = HI U (t′ , t) ∂t with U (t, t) = 1 or U (t′ , t) = 1 − i which can be solved by iteration, i.e. writing U (t′ , t) = one has U (0) (t′ , t) = U (1) (t′ , t) = = 1

t′ ∞ n=0 t′

= HI (t) ψI (t), = [AI , H0 ],

(9.54) (9.55)

(9.56) (9.57)

dτ HI (τ ) U (τ, t),

t

(9.58)

U (n) (t′ , t),

(9.59)

1−i (−i)

t t′ t

dτ HI (τ ) U (0) (τ, t) − U (0) (t′ , t) dτ HI (τ )

**Path integrals and quantum mechanics
**

t′ t t′ t

80

U

(2)

(t , t) = 1 − i = (−i) . . .

′

dτ HI (τ ) U (0) (τ, t) + U (1) (τ, t) − U (0) (t′ , t) + U (1) (t′ , t) dτ HI (τ ) U (1) (τ, t)

t′

U

(n)

(t , t) = (−i)

t

′

dτ HI (τ ) U (n−1) (τ, t)

t′ τ1 τn−1

= (−i)n

t

dτ1

t t′ t′

dτ2 . . .

t t′

dτn HI (τ1 ) . . . HI (τn ) dτn T HI (τ1 ) . . . HI (τn ).

t

=

(−i)n n!

dτ1

t t

dτ2 . . .

(9.60)

**The last equality is illustrated for the second term U (2) in the following. The integration
**

t′ τ1

dτ1

t t

dτ2 HI (τ1 ) HI (τ2 )

can also be performed by ﬁrst integrating over τ2 but changing the integration limits (check!). Thus we can write the integration as the sum of the two expressions (multiplying with 1/2), = 1 2

t′ τ1

dτ1

t t

dτ2 HI (τ1 ) HI (τ2 )

t′ t′

+

1 2

dτ2

t τ2

dτ1 HI (τ1 ) HI (τ2 ),

**Now the integration can be extended by adding theta functions, = 1 2
**

t′ t′

dτ1

t t

dτ2 HI (τ1 ) HI (τ2 )θ(τ1 − τ2 )

t′

+

1 2

t′

dτ2

t t

dτ1 HI (τ1 ) HI (τ2 )θ(τ1 − τ2 ),

**which can be rewritten (by interchanging in the second term the names of the integration variables) = 1 2
**

t′ t′

dτ1

t t

dτ2 [HI (τ1 ) HI (τ2 )θ(τ1 − τ2 ) + HI (τ2 ) HI (τ1 )θ(τ2 − τ1 )] ,

the desired result.

**The time evolution operator, therefore, can be written as U (t′ , t) = (−i)n n! n=0
**

∞ t′ t′ t′

dτ1

t t′ t

dτ2 . . .

t

dτn T HI (τ1 ) . . . HI (τn ) (9.61)

≡ T exp −i

dτ HI (τ ) .

t

one has q ′ .62)
p Dq D exp i 2π
t
dτ [pq − H(p. say. t′ |q. i. t|Q.Path integrals and quantum mechanics
source experiment detector
81
T L
t L+Jq
t’ L
T’
Figure 9. t |T exp −i p exp −i 2π
t′
′
′
t′ t
dτ HI (Q(τ )) |q.1: Physical picture of vacuum to vacuum amplitude in presence of a source term This is not a surprising result. for example x|n x|n = φn (x) = ei kn x = e
−ω 2 x2 /2
(9.65)
J
dq ′ dq Q′ . t
t′
V =0
Dq D
dτ HI (q) exp i
t t′ t
dτ [pq − H0 (p. Before and after these processes there is only the vacuum or ground-state |0 . ˙
The combination of exponentials in the last step is allowed because we are simply dealing with classical quantities (not operators!). for the physical states o q. t|Q. t = ei Ht |q . L(q.66)
. t Q′ . T . it is possible to switch on an interaction.
for g. a set |n of physical eigenstates.64) (9. T with q. t|n n|Q. We can again rewrite the time-ordered products as functional integrals as derived in the previous section. i. harmonic oscillator. n
(9. T =
n
q.s. Consider. q) + J(τ )q] . ˙
J
(9. q) → L(q.
φn (q. t|n = q|n e−iEn t with q|n the time-independent wave function. t′ q ′ . The Heisenberg state |q. q)] ˙ (9. t |q. t q. T < t < t′ < T ′ .
9. t = =
′ ′ V
= q . t is related to the Schr¨dinger state |q by |q.63)
for plane waves. furthermore. T ′ |Q. T =
n J t′
= =
Dq exp i
t
dτ [L(q. making the electron) and the absorption of an electron (think of a detector).e. t) φ∗ (Q) e+i En T . q) + J(t) · q.
Considering the source to be present between times t and t′ . t′ |q. ˙ ˙ physically pictured as. which in turn are embedded between an early time T and a future time T ′ . q)] . the creation of an electron (think of a radio-tube. T ′ |q ′ .5
The generating functional for time ordered products
By introducing a source-term.e. q .

t)
(9. thus
T →i∞
lim
q. in which case eiEn T → e−En ·∞ and the term eiE0 T is dominant. t 0 0out |0in J .
(9.
dq dτ
+ J(τ )q(τ )
. = (i)n 0|T Q(t1 ) .
. The generating functional precisely represents the vacuum to vacuum amplitude from initial (’in’) to ﬁnal(’out’) situation in the presence of a source. t|Q. T = φ0 (q. t′ |q. such that t′ → −i∞ t → i∞ ↔ ↔ ˜ t → −∞. T → i ∞ and T ′ → −i ∞.2: Analytic continuation of boundaries T and T ′ We can project out the groundstate by an analytic continuation in the time.71)
9. t′ ) q ′ . eiE0 T φ∗ (Q) 0
We deﬁne the generating functional Z[J] as Z[J] = Q′ . δJ(tn ) hence the name generating functional.68) (9. . ˜ t′ → ∞. T ′ |Q.69)
The factor that has been divided out in the ﬁrst line of this equation is a numerical factor depending on the boundaries of the space-time volume (T and T ′ ).
J=0
(9. 9.67)
= =
φ0 (q.6
Euclidean formulation
The above expression (Eq. The importance of Z[J] is that the time ordered product of operators can be obtained from it. Q(tn )|0 . . Better is the use ˜ of imaginary time t = −it.Path integrals and quantum mechanics
t
82
T
t
t’
T’
Figure 9.70)
one immediately ﬁnds δ n Z[J] δJ(t1 ) . . t).
J
(9. .70) for the generating functional is in fact ill-deﬁned.
T′
Z[J] =
lim T ′ → −i∞ T → i∞
Dq exp i
T
dτ L q. Since we have (neglecting multiplicative factors). T J lim −i E0 (T ′ −T ) φ (Q′ ) φ∗ (Q) 0 0 T ′ → −i∞ e T → i∞ dq ′ dq φ∗ (q ′ .

(9.
Exercises
Exercise 9. complex. Dα Dα∗ exp (−α∗ K α) = for complex-valued functions and Dθ∗ Dθ exp (−θ∗ K θ) = det K. δJ(tn ) J = 0 ˜ t = it
(9.77)
where the expressions have been divided by Z[J] and ZE [J] in order to get rid of dependence on multiplicative factors. In all of these cases the determinant of the (hermitean) operator K is deﬁned as det K ≡ p Kp . As discussed in the previous section the interesting quantities are obtained from functional diﬀerentiation with respect to the sources.74)
− V (q)
2
(9. dq dt = 1 2 1 2 dq dt
2
Dq exp Dq exp −
∞
d˜ L q.and Grassmann-valued functions.72) (9.75)
= − LE q.2
Check the examples of functional derivation for real-. ˜ ˜ ZE [J] δJ(t1 ) . y) = ∗ m. .73)
d˜ LE q. dq ˜ dt = 1 2
dq ˜ dt dq ˜ dt
2
− V (q).Path integrals and quantum mechanics
83
In terms of the imaginary time we can write the Euclidean generating functional.n fm (x) Kmn fn (y). The diﬀerentiations in Z[J] and ZE [J] are related.g.76)
which is a positive deﬁnite quantity. ensuring convergence for the functional integral ZE [J]. δ n Z[J] 1 Z[J] δJ(t1 ) .
. + V (q). . i
dq . 1 = √ . i τ
−∞ ∞ −∞
dq d˜ τ
+ J(˜)q(˜) τ τ dq d˜ τ − J(˜)q(˜) τ τ .1
Convince yourself by choosing an appropriate (orthonormal) set of functions that 1 Dα exp − α K α 2 for real-valued functions. ˜ dt
(9. . for Grassmann valued functions. τ
dq ˜ dt
≡ −L q. when L q. δJ(tn ) = (i)n
J=0
1 δ n ZE [J] .
(9. ZE [J] = = where LE q. . det K 1 . where Kp are the eigenvalues of the matrix Kmn in K(x. det K
Exercise 9. e.

Path integrals and quantum mechanics
84
Exercise 9.79) (9.80)
Dα∗ Dα exp (−α∗ K α − α∗ ω − ω ∗ α) =
1 exp ω ∗ K −1 ω . 2 (9. z) = δ(x − z). where dy K(x.78) (9.3
Prove the following relations 1 Dα exp − α K α − αω 2 1 = √ exp det K 1 ω K −1 ω . det K
Dθ∗ Dθ exp (−θ∗ K θ − θ∗ η − η ∗ θ) = det K exp η ∗ K −1 η .
. y) K −1 (y.

(10. .2)
=
∞ −∞
d4 xE [LE (φ. .1 Generating functionals for free scalar ﬁelds
The generating functional for quantum ﬁelds is a generalization of the results in the previous section to a system with more degrees of freedom. δJ(xn ) 85 ≡ G(n) (x1 . . φ(x. Π) (10. t′ |φ. i. ∂µ φ) + Jφ]
(10. .e. . implying at the level of four vectors in coordinate and momentum space (when k · x ≡ kE · xE ) for instance x4 = i x0 = i t x · x = (x0 )2 − d4 xE = i d4 x L =
1 2 3 i 2 i=1 (x )
k 4 = −i k 0 =−
4 µ 2 µ=1 (x )
k · k = (k 0 )2 − d4 kE = −i d4 k
3 i 2 i=1 (k )
=−
4 µ 2 µ=1 (k )
∂µ φ∂ µ φ −
1 2
M 2 φ2
LE =
1 2
µ (∂E φ)2 +
1 2
M 2 φ2
Furthermore we ﬁnd the n-point Green functions 0|T φ(x1 ) . .4)
J=0
.Chapter 10
Feynman diagrams for scattering amplitudes
10. ∂µ φ) − Jφ]
(10. xn ).1)
Dφ exp i
σ
d4 x L (x)
and
T′
Z[J] =
lim T → i∞ T ′ → −i∞ Dφ exp −
Dφ exp i
T
d4 x [L (φ. . x′ . x. t) is considered as a set of quantum operators φx (t) in the Heisenberg picture and
t′
φ′ . t
= =
Dφ DΠ exp i
t σ′
˙ dτ d3 x Π(x)φ(x) − H (φ. φ(xn )|0 = (−i)n δ n Z[J] δJ(x1 ) .3)
The Euclidean formulation is as before. .

13)
.3) one ﬁnds 1 Z0 [J] = exp − i 2 d4 x d4 y J(x) ∆F (x − y) J(y) ≡ exp (i W0 [J]) . . with Z0 [0] = 1.10)
J=0
In order to determine ∆F . one immediately sees that i∆F (x1 − x2 ) = δ 2 Z0 [J] (−i)2 Z0 [0] δJ(x1 ) δJ(x2 ) = G0 (x1 − x2 ) ≡
(2)
x1
x2
(10. . . d4 xn G(n) (x1 . (10. xn ) J(x1 ) . .
These Green functions appear in the expansion Z[J] =
n
in n!
d4 x1 .6)
the generating functional is (using ∂µ = − ∂µ ) given by Z0 [J] = Dφ exp −i
where Kxy = (∂µ ∂ µ + M 2 ) δ 4 (x − y). .Feynman diagrams for scattering amplitudes
86
for which we introduce the picture
x1
G(n) (x1 . J(xn ). ∆F (the socalled Feynman propagator) is the Green’s function of the Klein-Gordon equation. . .5)
Z0 [J] for the free scalar ﬁeld
For the (free) scalar ﬁeld lagrangian L =
←
1 1 ∂µ φ∂ µ φ − M 2 φ2 .e. As discussed in the previous section (exercise 9. .7)
where we made the choice Z0 [0] = 1 and ∂µ ∂ µ + M 2 ∆F (x − y) = −δ 4 (x − y). (2π)4 (k 0 + iǫ)2 − E 2 (10. = 0 2 2 2 k2 − M 2 (k ) − E 2 (k ) − k − M (10.12) d4 k −i k·x ˜ e ∆(k) (2π)4 (10. Furthermore.9)
i. (10. .8)
1 d4 x φ (∂µ ∂ µ + M 2 ) φ − Jφ 2
K
(10.4) one distinguishes the retarded Green’s function ∆R (x) = d4 k e−i k·x . which can be written as ∆(x) = ˜ with (k 2 − M 2 )∆(k) = 1 or ˜ ∆(k) = 1 1 1 = 0 2 . xn ) = x2
xn
. . 2 2
→
(10. .
(10.11)
Depending on the path choosen in the complex k 0 plane (see exercise 7. consider the general solution of ∂µ ∂ µ + M 2 ∆(x) = −δ 4 (x). .

The ﬁrst possibility is to consider the well-deﬁned Euclidean formulation. (10. (10.22) (10.1 for countours)
µ (∂E )2 − M 2 (i ∆F (x − y)) = −δ 4 (x − y)
0 kE =∞
(10. t) = 0 for t > 0. φ− (0)] = (2π)3 2E d3 k e−i k·x = [φ− (x). 10.15) e−i k·x d4 k . t) = 0 for t < 0. the advanced Green’s function ∆A (x) = satisfying ∆A (x. ∆(x) = − d4 k e−i k·x .16)
C
where C is a closed contour in the complex k 0 -plane.19) φ ∂E ∂E − M 2 φ + Jφ 2 1 d4 xE d4 yE J(x) (−i ∆F (x − y)) J(y) . In order to see which is the appropriate Green’s function to be used in the generating functional Z0 [J] we will take two routes.4). (2π)4 k 2 − M 2 (10. in the (well-deﬁned) Euclidean formulation starting with LE = 2 (∂E φ)2 + 2 M 2 φ2 µ 2 2 Jφ with the equation of motion (∂E ) − M φ(x) = −J(x). (2π)4 k 2 − M 2
(10.23)
k0 =i∞ k0 =−i∞
d4 k e−i k·x . the second is to explicitly consider 0|T φ(x1 ) φ(x2 )|0 .14)
They are solutions of the inhomogeneous equation. (2π)4 k 2 − M 2
where the latter contour can be deformed to the contour for ∆C . φ+ (0)] = (2π)3 2E
are also shown in exercise (7.Feynman diagrams for scattering amplitudes
87
satisfying ∆R (x.
. 4 k 2 − M 2 + iǫ (2π) (10. φ(0)] = i∆+ (x) + i∆− (x) d3 k e−i k·x = [φ+ (x). It is straightforward (closing contours in the appropriate half of the complex plane) to prove for instance that i ∆C (x) = θ(x0 ) i ∆+ (x) − θ(−x0 ) i ∆− (x).e. the causal Green’s function ∆C (x) = and ﬁnally the Green’s function d4 k e−i k·x = 4 (k 0 )2 − E 2 + iǫ (2π) ∆(x) = PV d4 k e−i k·x .20) = exp − 2
where
or (see ﬁg.
µ 1 1 • Firstly. ∆F = ∆C .18)
Thus we see various solutions. (2π)4 (k 0 − iǫ)2 − E 2 (10.21)
∆F (x)
= =
−i
0 kE =−∞
d4 kE e−i kE ·xE 2 (2π)4 kE + M 2
(10. The solution of the homogeneous equation ∂µ ∂ µ + M 2 ∆(x) = 0 can also be written as an integral in k-space.17)
d4 k e−i k·x . The contours for i ∆(x) i ∆+ (x) i ∆− (x) = [φ(x). the generating functional can be written 1 µ µ Z0 [J] = Dφ exp d4 xE (10. i.

x2 .Feynman diagrams for scattering amplitudes
0 1 1 0 0 0 1 1 1 0 1 0 0 0 1 0 1 0 11 00 k 1 0 1 0 k 11 00 E 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 -E 1 0 1 0 1 0 1111111111111111111111111111111111111 0000000000000000000000000000000000000 1111111 1 0000000 0 11111111111111111 00000000000000000 11111111111111 00000000000000 11 00 11 00 1 0 1 0 1 0 11111111 1 00000000 0 1 0 1 0 E 1 0 1 0 1 0 1 0 11 00 1 0 1 0 1 0 1 0 1 0 1 0 0 1 1 0 1 0 1 0 1 0 1 0
88
Figure 10. δJ(x4 ) 2 J=0 = − [∆F (x1 − x2 ) ∆F (x3 − x4 ) + ∆F (x1 − x3 ) ∆F (x2 − x4 ) + ∆F (x1 − x4 ) ∆F (x2 − x3 )] .24)
Knowing the explicit form of Z0 [J] it is straightforward to calculate the 4-points Green’s function. or diagrammatically
(4)
(10.25)
x1 x2
x3 = x4 + +
.3) i ∆F (x − y)
= 0|T φ(x) φ(y)|0 = θ(x0 − y 0 ) 0|φ(x) φ(y)|0 + θ(y 0 − x0 ) 0|φ(y) φ(x)|0 = θ(x0 − y 0 ) 0|φ+ (x) φ− (y)|0 + θ(y 0 − x0 ) 0|φ+ (y) φ− (x)|0
= i ∆C (x − y). φ− (y)]|0 − θ(y 0 − x0 ) 0|[φ− (x). the same result as above. We ﬁnd (see also section 7.
= θ(x0 − y 0 ) 0|[φ+ (x). . x4 ) = 0|T φ(x1 ) φ(x2 ) φ(x3 ) φ(x4 )|0 1 δ4 exp − i J ∆F J = (−i)4 δJ(x1 ) . φ+ (y)]|0 = θ(x0 − y 0 ) i ∆+ (x − y) − θ(y 0 − x0 ) i ∆− (x − y)
(10. we have G0 (x1 . Neglecting multiplicative factors or equivalently assuming that Z0 [0] = 1. because this is the only quantity entering Z0 [J].1: Contour in the k 0 plane for the Feynman propagator in Euclidean and Minkowski space • As the second possibility. . (4) (2) It should be clear that G0 can be expressed in terms of G0 = i ∆F . we use the fact that i∆F (x − y) = G(2) (x − y) = 0|T φ(x)φ(y)|0 to calculate ∆F via the time-ordered product. x3 .

(10. x3 .
(10. x2 .g ) (x1 .27)
When interactions are present.e.2
Generating functionals for interacting scalar ﬁelds
d4 x [L0 (φ) + LI (φ) + Jφ] 1 δ i δJ(z) d4 x [L0 (φ) + Jφ] (10.g ) (x1 . with propagators (i ∆F ) being connected to vertices (i LI ) according to the above expression for the functional Z[J].Feynman diagrams for scattering amplitudes
89
10. (10. x2 .34)
Introducing a vertex point to which four propagators are connected.35)
z
.
(2)
(0)
(10.28)
1 exp − i 2 = exp 1 2 d4 x d4 y exp i
δ δ i ∆F (x − y) δφ(x) δφ(y) d4 z [LI (φ) + J(z)φ(z)]
φ=0
. ≡ −i g d4 z (10. To zeroth order in the coupling one has
0 1 Z (g ) [J] = Z0 [J] = exp − i 2
J ∆F J .30)
and in zeroth order G(0. the generating functional can be written.31)
G(2.g
0
)
= G0 = 1. L (φ) = L0 (φ) + LI (φ). Z[J] = = Dφ exp i exp i d4 z LI (10. x2 ) = G0 (x1 . x3 . x4 ) = To ﬁrst order in g one has Z (g ) [J] = =
1 0
0
x1
+ +
x2
(10. x4 ) = G0 (x1 . i.32) (10.26)
Dφ exp i = exp i d4 z LI
1 δ i δJ(z) d4 x d4 y J(x) ∆F (x − y) J(y) (10. x2 ) = i ∆F (x1 − x2 ) = G(4.33)
(4)
i −i
d4 z LI g 4! d4 z
1 δ i δJ(z)
1 exp − i 2
J ∆F J
2
−3 ∆2 (0) + 6i ∆F (0) F +
d4 x ∆F (z − x) J(x)
4
d4 x ∆F (z − x) J(x)
1 exp − i 2
J ∆F J .29)
This expression will be the one from which Feynman rules will be derived. Consider as an example the interaction LI (φ) = − g 4 φ 4!
in the scalar ﬁeld theory discussed sofar.

. . xn ) J(x1 ) . . which also was the only Green’s function for which the diagram was connected.42)
x1
G(n) (x1 . . G(n) (x1 . . .39)
We have calculated the function up to ﬁrst order in the coupling constant for φ4 -theroy.
J=0
(10.36)
J
J
from which one obtains G(0. Z[J] =
n
in n!
d4 x1 . i.
(10.Feynman diagrams for scattering amplitudes
90
one has
J
Z
(g1 )
[J] =
1 8
1 − 4
J
J
1 + 24
J
Z0 [J]. . . . . Diagrammatically this exponent only contained the two-point function. . x2 ) =
1
z
x1 x2
+
1 2
x1
z x2
. .
(10. . . . . (10. did not contain parts that could be written as products of simpler Green’s functions.37)
G(2.e.g
1
)
=
1 8
z
1 8
.38)
The result for the 4-points Green’s function is left as an exercise. In the free case we have seen that the exponent of Z[J] contained all essential information. . .40)
in−1 n!
d4 x1 . J(xn ).
(10.g ) (x1 . d4 xn G(n) (x1 . . δJ(xn ) . d4 xn G(n) (x1 . xn ) J(x1 ) .
Connected Green’s functions
We have now seen the Green’s functions as the quantities appearing in the expansion of the generating functional Z[J]. .
(10. . .41)
in terms of socalled connected Green’s functions. J(xn ) c
(10. . xn ) = x2 c
xn
Speciﬁc examples are G(0) = c G(1) (x1 ) = c G(2) (x1 . . . . To see this write Z[J] = exp (i W [J]) with (by deﬁnition) the expansion W [J] =
n
or i W [J] = ln Z[J]. x2 ) = c (vacuum bubble) (tadpole) (connected propagator)
. xn ) = (−i)n−1 c which are denoted δ n W [J] δJ(x1 ) . This remains also true for the interacting case.

If i W [J] = then exp(i W [J]) = 1 + i W [J] + = 1 + 1 2! +i d4 x1 i2 + d4 x1 d4 x2 2! i2 4 4 d x1 d x2 + 2! + + (i W [J])2 (i W [J])3 + + . but is easily illustrated by writing down the ﬁrst few terms. 1 2!
Z[0] = 1 +
+
+ ..Feynman diagrams for scattering amplitudes
91
To see that Z[J] = exp(i W [J]) contains all non-connected diagrams can be performed inductively.
+
+i
d4 x1
1 2!
+
1 3!
+i
d4 x1
+
1 2!
i2 2!
d4 x1 d4 x2
+ . 2! 3! +i d4 x1 + i2 2! d4 x1 d4 x2 ... (ii) Note that Z[0] = exp(iW [0]) appears as a multiplicative factor. in the expansion of which all vacuum blobs are contained.. (1) (1) (iii) Gsc = Gc . In the expansion of Z[J]/Z[0] one (by deﬁnition) has the socalled source (n) connected Green’s functions Gsc ... δJ(x) J=0
..
Notes: (n) (i) In G(n) the connected Green functions Gc appear with particular combinatorial factors (accounted for in the deﬁnition of Feynman rules to be discussed later).
These can be divided out. (iv) If δZ[J] = i 0|φ(x)|0 = 0.

Also for free ﬁeld (Heisenberg) operators a distinction is made between φin and φout . (2) The one-particle state is invariant (conservation of energy and momentum. . out| = eiϕ0 0.
x4
10. out = |p′ . x2 . in|S = β. (10. out = δαβ ↔ SS † = 1. in| (choose ϕ0 = 0). Proof: α. Next. For the free scalar theory.g. x4 ) = c
1
x3 z
(10. . out|α. in = S|α. φout and the interpolating ﬁeld φ(x). out| and S † |α. out (suppressing except momenta all other 1 m quantum numbers).46)
The properties of the S-matrix are (1) The vacuum is invariant or |S00 | = 1. out = p|p′ . out|p′ .3
Interactions and the S-matrix
The S-matrix
The S-matrix transforms initial state free particle states (in-states) |α.Feynman diagrams for scattering amplitudes
92
i. p. thus β. the vacuum expectation value of the ﬁeld φ(x) is zero. in|p′ . in|S = 0. in = |α. out|α.44)
J=0
For the interacting theory. Sβα ≡ β.43)
implying as expected as the only nonzero connected Green’s function G(2) (x1 − x2 ) = (−i) c δ 2 W [J] δJ(x1 ) δJ(x2 ) = G0 (x1 − x2 ) = i ∆F (x1 − x2 ). e. pn . in ⇔ β. this will be translated to the action on ﬁelds. in = p. in|S = α. in|SS † |α. In line with the consideration of the generating functional representing the vacuum to vacuum amplitude we consider ﬁelds φin . in = β. considered explicitly. ∂µ ∂ µ + M 2 φin (x) = 0. Proof: 0. then (n) (n) Gsc = Gc for n ≤ 3. in into ﬁnal state particle states (out-states) |β. .
(2)
(10. p′ .
x1
G(4. in = |p1 . . out|p′ .45)
x2
as the only surviving diagram (see exercises).g ) (x1 .
(10. x3 . (10. in = p. out . . (3) S is unitary (it conserves the scalar product from initial to ﬁnal state). out . implying the absence of tadpoles.47)
.e. translation invariance). t = −∞ φin (x) t = +∞ φout (x)
φ(x)
where φin and φout transform under the Poincar´ group as scalar ﬁelds and satisfy the homogeneous e Klein-Gordon equation with the physical mass M . one ﬁnds for the connected 4-point Green’s function at order g. in = p. out| ⇔ |α. we have W0 [J] = − 1 2 d4 x d4 y J(x) ∆F (x − y) J(y). in|S|p′ . .

e.50)
= φ(x) =
√ Zφin (x) − √ Zφout (x) −
d4 y d4 z ∆R (x − y) K(y. implies the strong (operator) convergence φ(x) → Zφin (x). (2π)3 2E (10. β. in|Sφout Finally we check that as expected S does not spoil Poincar´ invariance.52)
where Z is a constant. Proof: β.and out-ﬁelds is: φin (x) = S φout (x)S −1 . φ(y)] = Z [φin (x).49)
The ﬁeld φ can be expressed in φin/out using retarded or advanced Green’s functions. √ Although the above. this can actually not be used as it would imply [φ(x). 9.g. in|φin S → φin S = S φout (x).
2 ∂µ ∂ µ + M0 φ(x) = J(x). The convergence therefore must be weakened to t→−∞ √ (10. z) φ(z). S is invariant under e Poincar´ transformations: U (Λ.e.
The relation between S and Z[J]
To establish this relation. out|φout = β.53) α|φf (t)β −→ Z α|φf (t)β in for normalizable states |α and |β and φf (t) ≡ d3 (x) f ∗ (x) i ∂ 0 φ(x) with f a normalizable solution of the Klein-Gordon equation (wave packet). z) φ(z) d4 y d4 z ∆A (x − y) K(y. the source term Jφ in the lagrangian density is treated in the interaction picture. φin (y)] = iZ ∆(x − y).
(10.54)
. as it stands. a) = S.48)
The fact that φin and φout satisfy the homogeneous Klein-Gordon equation with the physical mass M implies that they create particles and antiparticles as discussed. e Proof: φin (Λx + a) = U φin (x)U −1 = U Sφout (x)S −1 U −1 = U S U −1 U φout (x) U −1 U S −1 U −1 = U SU −1 φout (Λx + a) U S −1 U −1 ⇔ U SU −1 = S. i. The time evolution operator (see Eq. φin (x) = d3 k ∗ a(k) fk (x) + a† (k) fk (x) .51) (10. which in a later stage (renormalization) will become more important.
(10.
2 ˜ ∂µ ∂ µ + M 2 φ(x) = J(x) + (M 2 − M0 )φ(x) = J(x) √ ˜ φ(x) = Zφin (x) − d4 y ∆R (x − y) J(y)
(10. The relation between S-matrix and in. a)SU −1 (Λ. a causality condition that can be proven to imply the absence of interactions.62) is then
T′ UT [J] T′
= T exp i
T
d4 x J(x)φ(x) .
(10. Considering plane waves one sees from √ √ 0|φ(x)|p = lim Z 0|φout (x)|p = Z e−i p·x t→∞ √ √ = lim Z 0|φin (x)|p = Z e−i p·x
t→−∞ ↔
that identical normalization of (single-particle) plane waves in initial and ﬁnal state implies the same wave function normalization Z for in and out ﬁelds.Feynman diagrams for scattering amplitudes
93
while φ satisﬁes the inhomogeneous Klein-Gordon equation with the bare mass M0 (this is the mass appearing in the lagrangian L0 ).

S U [J]] = √ Z
δ S U [J]. (10.61) e
φ+ f
where the normal ordered expression : e φ f : is used. we can.
t i Ut∞ [J] φ(x)U−∞ [J]
(10.59) (10. B] is a c-number). Noting that 0| : eA : |0 = 1 it follows that F [J] = 0|S U [J]|0 = 0|U [J]|0 = 0out |0in ∝ Z[J].57)
i [φin (x).
φ− f
(10. e = i. δU [J] δJ(x) = = √ i Z U [J] φin (x) − √ i Z φout (x) U [J] − d4 y d4 z ∆R (x − y) K(y. e e ] = [A. Thus S U [J] =: exp 1 √ Z d4 x d4 y φin (x) K(x.
(10.
: =i
d4 y ∆(x − y) f (y) : e
d4 z φ(z) f (z)
:. B]eB . eB ] = [A. : e
d4 z φ(z) f (z) d4 x φ− (x) f (x) B C B C
(10. φ(y)] f (y) e
e
. B + C]e e . [A.55)
δU [J] δJ(x)
=
t→∞ t→−∞
−→
−→
√ i φ(x) U [J] → i Z φout (x) U [J] √ i U [J] φ(x) → i Z U [J] φin (x). y) δ δJ(y) : F [J].63)
. one has [A. δJ(z)
Taking the diﬀerence between the two expressions √ Z (φout (x) U [J] − U [J] φin (x)) =i or d4 y d4 z (∆R (x − y) − ∆A (x − y)) K(y. δJ(z)
(10.e.62)
where F [J] is some (arbitrary) functional.60)
e
d4 y φ+ (y) f (y) d4 z φ− (z) f (z) d4 z φ+ (z) f (z)
d4 y [φ(x). express it in terms of advanced and retarded Green’s functions.58)
In order to ﬁnd a solution to this equation note that. provided [A. z) δU [J] δJ(z) δU [J] d4 y d4 z ∆A (x − y) K(y. δJ(x) ∞ or for U [J] ≡ U−∞ [J]
′
(10.56)
Since δU [J]/δJ(x) satisﬁes the same equations as φ(x) and we know the limits. which is equal to the expression e in which the creation operators are placed left of annihilation operators. φ(x). φ(x). δJ(z)
(10. with the use of the Baker-Campbell-Hausdorﬀ formula e−B A e+B = A + [A. just as we did for φ(x). B] (for the case that [A. z)
δU [J] . Thus applied to the ﬁeld φ(x) = φ+ (x) + φ− (x). C] are c-numbers. B] and [A. z) .Feynman diagrams for scattering amplitudes
94
and satisﬁes the property
T ′ δUT [J] t = i UtT [J] φ(x)UT [J]. z) d4 y d4 z ∆(x − y)K(y.

which is determined by the inverse of the operator found in the quadratic term in the lagrangian i. = .) iKx .
(10. u(p) and v (p) for fermions and ǫµ (p) for vector ﬁelds). . . pn ) = G (p1 . They start with the propagator (i∆F (k)) in momentum space.
n
(2π)4 δ 4 (p1 + .+ 1 √ Z
∗ d4 x′ .4
Feynman rules
The real scalar ﬁeld
The procedure to obtain the matrix element is commonly summarized by a set of rules known as Feynman rules. .e. . fp′ (x′ ) . . ¯ ¯
and the amputated or one-particle irreducible (1PI) Green’s functions n −i (n) Γ(n) (p1 . . u(p). . . y) δ δJ(y) : Z[J] Z[0] . . see exercise 10. Sf i = δf i − i (2π)4 δ 4 (Pi − Pf ) Mf i . It is straightforward to check that the S-matrix element now precisely is given by the amputated Green’s function multiplied with the momentum space wave functions of the particles in initial and ﬁnal state (by which we refer to the quantities multiplying the plane wave e±i p·x in the ﬁeld expansion. .. . . which is obtained considering only connected diagrams.68)
10. . y). . xn ). d4 x .67)
where ∆(p) is the Fourier transform of the (full) propagator ∆F (x). . . . iKx′ = (2x′ + M 2 ) precisely annihilating an external propagator i∆F in the Green’s function. .δ.64)
Therefore. . . . (2π)3 2E
n
(10. . ∆(pj ) j=1
(10. . pn ) =
i=1
d4 xi ei pi ·xi G(n) (x1 . pn ). c
(n)
(10. |S|p. .. 1 for scalar case.
k
=
k2
i − M 2 + iǫ
. . . Usually we are interested in the part of the S-matrix describing the scattering. . . i. (λ) v(p).66)
where Gc is the connected Green’s function. .2). iKx′ . . Note that the Green’s function connecting the external point xi with the bubble is annihilated by K(x. fp (x) .e. ←− − →
× G(n) (x′ .
J=0
(10. .. . using that φin (x) = d3 k ∗ a(k) fk (x) + a† (k) fk (x) . . S =: exp 1 √ Z d4 x d4 y φin (x) K(x. . .e. . Next we introduce the Fourier transform (after extracting a momentum conserving delta function coming from translation invariance. . . . an S-matrix element between momentum eigenstates in initial and ﬁnal states is found by considering those source-connected Green’s functions (action of δ/δJ on Z[J]/Z[0]) where the external sources J(xi ) are replaced by the particle wave functions (which are the result of acting with φin (xi ) on momentum eigenstates). x. + pn ) G(n) (p1 .. i. Explicitly.65)
we get the Lehman-Symanzik-Zimmerman (LSZ) reduction formula. . . p′ .Feynman diagrams for scattering amplitudes
95
while for J = 0 one has U [0] = 1. . . .

66 and 10. If problems arise go back to the deﬁning expression for the generating function in 10. (Rule 3) Carry out the integration over all internal momenta (keeping track of momentum conservation at all vertices!) (Rule 4) Add a symmetry factor 1/S corresponding to permutation of internal lines and vertices (keeping external lines ﬁxed). for example up to order g 2 the scattering of two neutral spin 0 particles (real scalar ﬁeld) is described by
+
+
+
(Rule 2) The contribution of each diagram is obtained by multiplying the contributions from propagators. At these vertices each line can be assigned a momentum. which is summarized in the following rules: (Rule 1) Start with external legs (incoming particles/outgoing particles) and draw all possible topologically diﬀerent connected diagrams. which cancels the factor 1/2 in the quadratic piece. ’t Hooft and M.69) LI (φ) = − φ3 − φ4 . or calculating full Green’s functions external lines are treated as propagators.27. = −i g
. Note that in calculating amputated Green’s functions external lines are neglected. to be precise iLI vertices in momentum space are introduced. For the symmetry factor consider the examples (given in G. Veltman. Diagrammar) in the case of the interaction terms g f (10.Feynman diagrams for scattering amplitudes
96
(in fact the inverse of the operator found in the quadratic term is for real scalar ﬁelds also multiplied by a factor 2.27. = −i g (multiplied with 4! corresponding to the allowed number of permutations of identical particles). vertices and external particle wave functions in that diagram. Corresponding to external particles wave functions are introduced
k k
1 1
In order to calculate the connected amplitude −i Mf i appearing in the S-matrix element these ingredients are combined using Eqs 10. but overall momentum conservation at a vertex is understood. the factor 2 corresponds to the two-points Green’s function having two identical ends). For the interaction terms in the lagrangian. 3! 4! The vertices are: = −i f Consider ﬁrst the lowest order self-energy diagram.

The total result is 6×3×2 1 1 = = . J ∗ ] = Dφ Dφ∗ exp i d4 z LI d4 x φ∗ −∂µ ∂ µ − M 2 φ + LI (φ) + J ∗ φ + Jφ∗ δ 1 1 δ . S 3! 3! 4! 2! 2
Complex scalar ﬁelds
The case of complex scalar ﬁelds can be considered as two independent ﬁelds.
. External line 1 can be attached in six. ∗ (z) i δJ(z) i δJ d4 x d4 y J ∗ (x) i∆F (x − y) J(y) . Dividing by vertex factors and permutations of identical vertices. (10. Divide by the permutational factors of the vertices.Feynman diagrams for scattering amplitudes
97
Draw two points corresponding to the two vertices and draw in each of these points the lines coming out of the vertices:
Now count in how many ways the lines can be connected with the same topological result. S 3! 3! 2! 2 As a second example consider the diagram
1 2
There are three vertices. after that line 2 in three ways. which have been included in the deﬁnition of vertices (here 3! for each vertex). denoted
k
=
i k 2 − M 2 + iǫ
Note that in this case the propagator does not have identical ends. Then there are two ways to connect the remaining lines such that the desired diagram results. i.70)
= exp i
exp −
In Feynman diagrams the propagator is still given by i∆F (k). there is no combinatorial factor like in the scalar case. but it connects a source with its complex conjugate and therefore is oriented. or equivalently as independent ﬁelds φ and φ∗ . the result is 1 6×4×6×3×2 1 = = .e.
After connecting line 1 (6 ways) and line 2 (4 ways) we have
1
2
leaving 6 × 3 × 2 ways to connect the rest as to get the desired topology. Finally divide by the number of permutation of the points that have identical vertices (here 2!). The generating functional in the interacting case can be written as Z[J.

.
which arises from the quadratic term in exp i d4 z LI . (Rule 6) Each closed fermion loop gets a sign −1.
.
= + .− i δη(z) i δη(z) d4 x d4 y η(x) iSF (x − y) η(y) .73)
=
i p / − M + iǫ
=
ji
i(/ + M )ji p p2 − M 2 + iǫ
The time ordered functions are obtained by functional derivatives from Z(η.71)
= exp i
where iSF is the Feynman propagator for fermions.g. η). in e− e− → e− e− scattering (Møller scattering) the lowest order contribution is
(see next section for e-e-γ vertex). The latter rule is illustrated in the example of an interaction term LI = : gψ(x)ψ(x)φ(x) : in an interacting theory with fermionic and scalar ﬁelds. The two-points Green’s function 0|T φ(x) φ(y)|0 contains a fermionic loop contribution. 4 2 − M 2 + iǫ (2π) p and the (oriented) propagator in Feynman diagrams involving fermions is
i p j
(10.e. minus the inverse of the operator appearing in the quadratic piece) and is given by iSF (x − y) = 0|T ψ(x)ψ(y)|0 = = (i/x + M ) i∆F (x − y) ∂ d4 p −i p·x p /+M = e . η]. − g2 2 dz dz ′ δ2 δ2 δη(z) δη(z) δη(z ′ ) δη(z ′ )
and the quadratic term in Z0 [η. (10. which is the solution of (i/ − M )SF (x) = −δ 4 (x) ∂ (i. namely (Rule 5) Feynman diagrams which only diﬀer by exchanging identical fermions in initial or ﬁnal state have a relative minus sign.. η] = Dψ Dψ exp i d4 z LI exp − d4 x ψ (i/ − M ) ψ + LI (ψ) + ηψ + ψη ∂ 1 δ 1 δ . − 1 2 dx dy dx′ dy ′ η(x)S(x − y)η(y) η(x′ )S(x′ − y ′ )η(y ′ ).Feynman diagrams for scattering amplitudes
98
Dirac ﬁelds
For fermions the generating functional is given by Z[η.72) (10.. e. but the anticommutating properties of Grassmann variables imply some additional minus sign in Feynman diagrams.

+ iǫ)2
where the kµ kν terms in the case where the photon couples to a conserved current (such as ψγ µ ψ) will not contribute. As an example some scattering processes in quantum electrodynamics will be discussed in the next section after the introduction of the Feynman rules for vector ﬁelds. 1 k 2 − M 2 + iǫ λ (k 2 − M 2 + iǫ)(k 2 − λ M 2 + iǫ)
leads to
.
i
p
p i
ui (p) ui (p) ¯ vi (p) ¯
incoming fermion outgoing fermion incoming antifermion outgoing antifermion
i
p p i
vi (p)
In writing down the expressions for Feynman diagrams one has to be aware that the wave functions and the propagators have more than one component. Particular choices of λ are λ = 1 (Feynman propagator or Feynman gauge) and λ = ∞ (Landau gauge).74) L = − Fµν F µν − (∂ · A)2 + ψ(i/ − M )ψ − e ψγ µ ψ Aµ .Feynman diagrams for scattering amplitudes
99
The result is which contains an extra minus sign as compared to a bosonic loop.
λ (∂µ Aµ )2 2
1 For a massive vector boson inversion of the quadratic term including the Lorentz constraint ∝ the propagator kµ kν 1 −gµν + 1− i Dµν (k) = i . λ 1 ∂ (10. The wave functions for fermions are given by ∝ g 2 S(z − z ′ ) S(z ′ − z) = −g 2 iS(z − z ′ ) iS(z ′ − z). 4 2 In addition to the fermion propagator and fermion wave functions discussed in the previous section we have the photon propagator1
µ k ν
=
−i
k2
gµν 1 − 1− + iǫ λ
(k 2
kµ kν .
Vector ﬁelds and Quantum Electrodynamics
As the most important example of vector ﬁelds consider the lagrangian density for quantum electrodynamics (QED). The wave functions are given by
µ k
ǫµ (k)
k µ
incoming photon outgoing photon
ǫ∗ (k) µ
The vertex for the coupling of photon to the electron is given by
µ
=
i j
−i e(γµ )ji . writing down the propagators till the beginning of the ¯ line (incoming fermion u or outgoing antifermion v) has been reached. It is necessary to start at the end of a fermion line (with the above arrow convention an outgoing fermion wave function u or an incoming antifermion ¯ wave function v ) and keep on following that line.

s4 ) and averaged over spins in the initial state (1/2 × 1/2 × s1 .s3 . ¯ q2 (10. s1 ) u e2 u(p′ .5
Some examples
eµ scattering
The ﬁrst example is the electromagnetic scattering of an electron and a muon. s′ ) ¯ u
s.s2 .s′
′ ′ = 2 kµ kν + 2 kν kµ + q 2 gµν
1 Tr [(/ ′ + m)γµ (/ + m)γν ] k k 2 ′ ′ = 2 kµ kν + kν kµ − gµν (k · k ′ − m2 )
(10. s2 ). q 4 µν
[using that (¯(k ′ )γµ u(k))∗ = u(k)γµ u(k ′ )].s2 ) which can be written as (see also chapter 4) |M |2 = = 1 4s |¯(k ′ . s′ )γµ u(k.Feynman diagrams for scattering amplitudes
100
10. s3 )(−ie)γ µ u(k. To lowest order (∝ α = e2 /4π) only one diagram contributes. t2
(10.77)
Combining Lµν and Lµν (M) one obtains L(m) Lµν (M) = 2 s2 + u2 + 4t(M 2 + m2 ) − 2(M 2 + m2 )2 µν and M 4π
2
(m)
(10.75)
Note that the q µ q ν term in the photon propagator are irrelevant because the photon couples to a conserved current. where u ¯ L(m) µν = = 1 2 u(k ′ . s1 ) ¯ −igµν u(p′ .79)
.s4
e4 (m) µν (M) L L . s4 )(−ie)γ ν u(p. s)γν u(k ′ .78)
=
2 α2 2 s + u2 + 4t(M 2 + m2 ) − 2(M 2 + m2 )2 . s)¯(k. If we are interested in the scattering process of an unpolarized initial state and we are not interested in the spins in the ﬁnal state we need |M |2 summed over spins in the ﬁnal state ( s3 .76)
1 . The diagram and momenta and the commonly used invariants (Mandelstam variables) for a 2 → 2 scattering process are
1 k
3 k’
s t u
= (k + p)2 = m2 + M 2 + 2 k · p = (k ′ + p′ )2 = m2 + M 2 + 2 k ′ · p′ = (k − k ′ )2 = q 2 = 2 m2 − 2 k · k ′ = (p − p′ )2 = 2 M 2 − 2 p · p′
p 2
p’ 4
s
= (k ′ − p)2 = m2 + M 2 − 2 k ′ · p + t+u= m2 = 2 m2 + 2 M 2 i
= (k − p′ )2 = m2 + M 2 − 2 k · p′
The scattering amplitude is given by −iM = u(k ′ . s3 )γ µ u(k. s4 )γµ u(p. s2 )|2 ¯ q2 (10.

(c) Do the same for the connected Green function Gc .
Exercises
Exercise 10.s4
=
1 2s ×
v (k ′ . for instance Møller scattering (e− e− → e− e− ) and Bhabha scattering (e− e+ → e− e+ ) are related using crossing symmetry. s2 ) ¯ ¯ 1 2s u(p. s3 )γ µ v(p′ .s4
=
1 1 e Tr(/ − m)γµ (/ ′ + m)γν k k Tr(/ + M )γ µ (/′ − M )γ ν p p s2 2 2 1 e4 1 ′ ′ = 4 2 kµ kν + kν kµ − gµν s pµ p′ν + pν p′µ − g µν s s 2 2 M 4π
2
4
and
=
2 α2 2 t + u2 + 4s(M 2 + m2 ) − 2(M 2 + m2 )2 . s1 ) u(k. s2
(10. . (b) Use the deﬁnition of the source-connected Green function Gsc and show that indeed sourceconnected diagrams survive. s4 )γ ν u(p. To lowest order (∝ α = e2 /4π) only one diagram contributes.Feynman diagrams for scattering amplitudes
101
The second example is the annihilation of an electron pair and creation of a muon pair. . the masses. s2 )γµ u(k. s2 )γµ u(k.s2 . Basically the same diagram is calculated and the result is the same after the interchange of s ↔ t. s3 )γ µ v(p′ . . s3 ) ¯ ¯
1 . Similarly. s1 )|2 ¯ s
1 .80)
Note the similarity in the amplitudes for eµ scattering and e− e+ → µ− µ+ . This is known as crossing symmetry.s3 .2. momenta and invariants are
e− e+ → µ− µ+ scattering
k
p
s t
= =
(k + k ′ )2 = (p + p′ )2 (k − p)2 = (k ′ − p′ )2 (k − p′ )2 = (k ′ − p)2
k’
p’
u =
The scattering amplitude squared (spins summed and averaged) is given by |M |2 = 1 4s e4 s2 |¯(p. The diagram.1
(a) Give in diagrammatic notation the full Green functions G(4) (x1 .
(4) (4)
. s1 )γν v(k ′ . x4 ) for the interacting case in φ4 -theory to ﬁrst order in the coupling constant g as obtained from the full expression for Z[J] in section 10. s4 ) v (p′ . s4 ) u e2 v (k ′ .s2
3 .

or if we in general would consider all momenta as incoming. (b) Show (by using x1 as shift-varible) that this implies that
n i=1
d4 xi ei pi ·xi G(n) (x1 . pn ). .Feynman diagrams for scattering amplitudes
102
Exercise 10. . Note that the function G(n) (p1 . . t and u. their sum is zero. xn ). .
. . .3 reproduce for the diagrams
precisely the combinatorial factors that appear in Eqs 10. xn ) ∝ (2π)4 δ 4 (p1 + .
which means overall momentum conservation in Green functions in momentum space.38. . . . e(p1 ) + e(p2 ) → e(p′ ) + e(p′ ) in lowest order in α. . .
hence we can write
n i=1
d4 xi ei pi ·xi G(n) (x1 . . Show that i the amplitude is symmetric under the interchange of t ↔ u. but the incoming and outgoing momenta are identical. for instance. . that the two-point Green’s function in momentum space has an incoming and an outgoing leg (as for the x-space Green’s function). . pn ) only has n − 1 independent variables in it. .3
Show that the combinatorial factors found using the rules given in section 10. + pn ) G(n) (p1 . . . . |M |2 = T11 + T22 − T12 − T21 . . . xn ) ≡ (2π)4 δ 4 (p1 + . .
Exercise 10. .37 and 10. + pn ). in the amplitude (Tij = A∗ Aj ). . . . . It is of the form −i M = A1 − A2 . Express the contributions in invariants s. . . 1 2 (b) Calculate the quadratic pieces and interference terms. xn + a) = G(n) (x1 . Pictorially this implies.2
(a) Show that the translation properties of the ﬁelds and the vacuum imply G(n) (x1 + a.
Exercise 10.4
(a) Write down the Feynman diagrams contributing to electron-electron scattering.

. the 1particle phase space is d3 p d4 p θ(p0 ) (2π)δ(p2 − m2 ).6) (11.5) √ is referred to as the invariant mass squared. q). for instance for determining the threshold energy for the production of a ﬁnal state 1 + 2 + . It is determined by the energy-momentum four vector p = (E. . pa ) and pb = (Eb . . To be speciﬁc let us consider two frequently used frames. . i. s is for obvious reasons known as the center of mass (CM) energy. The ﬁrst is the CM system. For two particles.
i n
d3 pi . .
103
.
(11. .7)
pb =
cm (Eb . . pn ). . pb ) satisfying p2 = m2 and p2 = m2 .4)
For two particle states |pa .1) = (2π)3 2E (2π)4 (proven in Chapter 2).1 kinematics in scattering processes
Phase space
The 1-particle state is denoted |p . p1 . s = (p1 + . . In the CM frame the threshold value for s obviously is
n 2
sthreshold =
i=1
mi
. The phase space is determined by the weight factors assigned to each state in the summation or integration over states. (11. + n. a a b b the quantity s = P 2 = (pa + pb )2 . pn ) = (2π)4 δ 4 (P − pi ) dR(p1 . .3)
which is useful because the total 4-momentum of the ﬁnal state usually is ﬁxed by overall momentum conservation. It is a useful quantity. . A physical state has positive energy.
(11. (11. In that case
cm pa = (Ea . . pn ) = and the reduced phase space element by dR(s.Chapter 11
Scattering theory
11. p) which satisﬁes p2 = E 2 −p2 = m2 . Here s is the invariant mass of the n-particle system. . pb we start with the four vectors pa = (Ea . + pn )2 . . (2π)3 2Ei i=1
(11. −q).e.2)
(11. Its square root. . and the total momentum four-vector P = pa + pb . This is generalized to the multi-particle phase space dR(p1 . .

Also in this case one can express the energy and momentum in the invariants. m2 . m2 .16)
.
(11. 2mb λ(s. m2 ) = 4(pa · pb )2 − 4p2 p2 . 0). ptrf ). 1 2 d(E1 + E2 ) √ |q| dΩ(ˆ) q δ( s − E1 − E2 ) 2 (2π) 4(E1 + E2 ) √ q |q| dΩ(ˆ) λ12 dΩ(ˆ) q √ = .13) (11. Prove that |q| =
cm Ea cm Eb
(s − m2 − m2 )2 − 4 m2 m2 a a b b = 4s s + m2 − m2 a b √ = . m2 ) a b . m2 ) is a function symmetric in its three arguments.
trf Ea (threshold) =
1 2 mb
(
i
mi )2 − m2 − m2 a b
.8) (11. a
(11. p2 ) =
CM
=
=
1 (2π)2 1 (2π)2 1 (2π)2
d3 p1 d3 p2 4 δ (P − p1 − p2 ) 2E1 2E2 √ d3 q δ( s − E1 − E2 ) 4 E1 E2 q 2 d|q| √ dΩ(ˆ) q δ( s − E1 − E2 ) 4 E1 E2
which using |q| d|q| = (E1 E2 /(E1 + E2 )) d(E1 + E2 ) gives dR(s. 4s
(11. for instance.14)
One can.9) (11. Prove that
trf Ea =
|ptrf | = a
s − m2 − m2 a b . In that case
trf pa = (Ea . 2 s s − m2 + m2 a b √ . 4π s 4π 8π s 4π
(11. m2 ). p1 . which in the speciﬁc case a b also can be expressed as λ(s. m2 . p2 ) = = where λ12 denotes λ(s.11) (11. = 2 s
λ(s.12)
pb = (mb . m2 .15)
Explicit calculation of the reduced two-body phase space element gives dR(s.10)
The function λ(s. a a b b The second frame considered explicitly is the socalled target rest frame in which one of the particles (called the target) is at rest. m2 ) a b . 2mb
(11. p1 . mb and s). m2 .Scattering theory
104
It is straightforward to prove that the unknowns in the particular system can be expressed in the invariants (ma . use the ﬁrst relation and the abovementioned threshold value for s to calculate the threshold for a speciﬁc n-particle ﬁnal state in the target rest frame.

23)
which are not independent as s + t + u = m2 + m2 + m2 + m2 .21) (11. The initial state. 2s
(11. 11.24)
Using the relation between t and cos θcm it is straightforward to express dΩcm in dt. An often used set of invariants are the Mandelstam variables. Of course there are not only 2-particle channels. Consider the process a + b → c + d. A speciﬁc reaction channel starts contributing at the threshold value (Eq. pd ) = 4π s 4π 8π s 4π dt dt √ √ . In inclusive measurements no particles are detected in the ﬁnal state. H1 + H2 → h + X. py ). dt = 2 qq ′ d cos θcm and obtain for the two-body phase space element √ q ′ dΩcm λcd dΩcm √ = (11. pz = mT sinh y.25) dR(s. for instance the momenta of incoming (colliding) hadrons and it is useful to use for the produced particle rapidity as a variable. tmax min λcd /4s. there is usually a preferred direction.17) (11.26) = 16π q s 8π λab
Kinematics of inclusive hard scattering processes
In high energy (hard) scattering processes. = (11.19) (11. .Scattering theory
105
Kinematics of 2 → 2 scattering processes
π− + p
The simplest scattering process is 2 particles in and 2 particles out. usually is a 2-particle state. however. E
. pc .18) (11. where the ﬁrst is referred to as elastic channel and the set of all other channels as the inelastic channels. usually many particles are produced.20) → π0 + n → π+ + π− + n
→ . s = (pa + pb )2 = (pc + pd )2 t = (pa − pc ) = (pb − pd ) u = (pa − pd )2 = (pb − pc )2
2 2
(11. Examples appear in → π− + p (11. Writing E = mT cosh y. At high energies.. in which one particle is detected. a c with q = λab /4s and q ′ = being 0 or 180 degrees. Consider the 1-particle inclusive case. in exclusive measurements all particles are detected. (11. which for the above 2 → 2 process in the case of azimuthal ˆ ˆ symmetry is deﬁned as pa · pc = cos θ one can use in the CM the invariant t ≡ (pa − pc )2 = m2 + m2 − 2 Ea Ec + 2 qq ′ cos θcm . The variable s is always larger than a c b d the minimal value (ma + mb )2 .4).. while the ﬁnal state often arises from a series of 2-particle processes combined with the decay of an intermediate particle (resonance). Instead of the scattering angle. The minimum and maximum values for t correspond to θcm
CM
= m2 + m2 − 2 Ea Ec ± 2 qq ′ a c = m2 a + m2 c
(s + m2 − m2 )(s + m2 − m2 ) a c b d − ± 2s
√ λab λcd .27)
with m2 = m2 − p2 = m2 + p2 + p2 .22) (11. the rapidity y is deﬁned T T x y y= 1 ln 2 E + pz E − pz = ln E + pz mT = tanh−1 pz .
The various possibilities are referred to as diﬀerent reaction channels. pT = (px .

t. u) with s = (p1 + p2 )2 = (p3 + p4 )2 . u = (p1 − p4 )2 = (p2 − p3 )2 . tt = s. ¯ tu = (pa − pc )2 = t. η = − ln(tan(θ/2)) = tanh−1 (cos θ). tu = s. b Analiticity of the ﬁeld theoretically calculated result implies
Ma¯→¯ (st = t. u).2
Crossing symmetry
In the previous chapter.30) (11. Given a two-to-two scattering process ab → cd one can relate the processes
b p p a
2 1
d p
4
_ c −p p
1
3
d p
4
_ d −p p a
1
4
_ b −p
2
p 3 c ab cd
a _ ac
−p 2 _ _ b bd
_ ad
p 3 _ c cb
They are referred to as s-channel. t-channel and u-channel processes respectively. b ut = (pa − pd )2 = u. ut = u) = Mab→cd (s. which means that rapidity distributions dN/dy maintain their shape.
11. The one-particle phase space in terms of these variables becomes d3 p dy d|pT |2 dφ dy d2 pT |p| dη d|pT |2 dφ = = = . (11. ¯ b
(11. With the momenta deﬁned as in the ﬁgures above one has for all these processes the same amplitude M (s. ¯ tt = (pa − p¯)2 = s. For the t-channel process (a¯ → ¯ one has c bd) st = (pa + pc )2 = t. c bd Mad→c¯(su = u. For large energies and not too small scattering angles (θ ≫ 1/γ) y is approximately equal to the pseudo-rapidity η.
uu = (pa − p¯)2 = s.31)
.29) 3 2E 3 2 (2π) 2 (2π) 16π 2π E 16π 2 2π with at high energies the factor |p|/E ≃ 1.28)
which (only involving angles) is easier to determine. we have seen that the amplitudes for the processes e− µ− → e− µ− and the process e− e+ → µ+ µ− are simply related by an interchange of the variables s and t. This is known as crossing symmetry. t. ¯ while for the u-channel process (ad → c¯ one has b) su = (pa + pd )2 = u. These variables are precisely the Mandelstam variables for the s-channel process (ab → cd). (11. t = (p1 − p3 )2 = (p2 − p4 )2 . u). uu = s) = Mab→cd (s.Scattering theory
106
It is convenient because under a boost (along z) with velocity β it changes as y → y − tanh−1 β. t.

3
Cross sections and lifetimes
Scattering process
For a scattering process a + b → c + . Nc /T indicates the number of particles c detected in the scattering process. . . . i. T where V and T indicate the volume and the time in which the experiment is performed. .e.
_ ac
cos
s= 0
_ bd
cos θ t =1
θt =− 1
t u
cos θ u = 1
sθ u=−
s
t=0 cos θ s = 1
u= 0
_ ad
_ cb
ab
1
cd
1
cos θs
=−
11.) is deﬁned as the proportionality factor in Nc = σ(a + b → c + . shown below for the case of equal masses. 1 N . u) that represents physical amplitudes in the physical regions for three scattering processes. The proportionality factor has the dimension of area and is called the cross section.32) σ= trf T · V ρa ρb va
co
. . To see this one can make a two-dimensional plot for the variables s. while the ﬂux of the beam particles a is trf ﬂux(a) = ρa · va . (consider for convenience the rest frame for the target. which for a density ρb is given by Nb = ρb · V .) · Nb · ﬂux(a). (11.Scattering theory
107
One can also phrase it in the following way: one has a single analytic function Mab→cd (s. t. 2 2 2 2 u(t − s) + (Ma − Md )(Mc − Mb ) . t and u. This is a consequence of the constraint 2 2 2 2 s + t + u = Ma + Mb + Mc + Md . Nb indicates the number of (target) particles b. To ﬁnd the physical regions one looks for the boundaries of cos θs = cos θt = cos θu = 1 λs λs ab cd 1 λt λt ac bd 1 λu λu ad cb
2 2 2 2 s(t − u) + (Ma − Mb )(Mc − Md ) . say b) the cross section σ(a + b → c + .
This deﬁnes the boundaries of the physical regions. 2 2 2 2 t(s − u) + (Ma − Mc )(Mb − Md ) . .

From the solution N (t) = N (0) e−Γ t one knows that the decay time τ = 1/Γ.33)
Decay of particles
For the decay of particle a one has macroscopically dN = −Γ N. .
. (2π)4 δ 4 (Pi − Pf )
2
= (2π)4 δ 4 (Pi − Pf ) = (2π)4 δ 4 (Pi − Pf )
d4 x ei(Pi −Pf )·x
V. .39) One of the δ functions can be rewritten as T · V (remember the normalization of plane waves). (11.36) (11.T
V.
N 1 .T
d4 x = V · T (2π)4 δ 4 (Pi − Pf ). σ= √ 2 λab T · V
(11.37) 2 ma T · V
Fermi’s Golden Rule
In both the scattering cross section and the decay constant the quantity N/T V appears. as expected. N and T · V are covariant.34)
i. the amount of decaying particles is proportional to the number of particles with proportionality factor the em decay width Γ. Using ρtrf = a (0) (0) (0) lab lab lab ρa · γa = ρa · Ea /ma (where ρa is the rest frame density) and va = plab /Ea we have a
trf ρa ρb va = (0)
ρa ρb 2 λab 4 ma mb
(0) (0)
or with ρa = 2 ma . the number of scattered or decayed particles is given by 2 N = (2π)4 δ 4 (Pi − Pf ) iMf i dR(p1 . it is. pn ). dt (11. .38)
(in which we can calculate −iMf i using Feynman diagrams). Microscopically one has Ndecay = Na · Γ T or Γ= N 1 . T · V ρa (11.Scattering theory
108
Although this at ﬁrst sight does not look covariant.35)
This quantity is not covariant. .e. For this we employ in essence Fermi’s Golden rule stating that when the S-matrix element is written as Sf i = δf i − (2π)4 δ 4 (Pi − Pf ) iMf i (11. The decay time for moving particles τ is related to the decay time in the rest frame of that particle (the proper decay time τ0 ) by τ = γ τ0 . For the (proper) decay width one thus has 1 N Γ0 = . (11.

pn ). 2 λab and for instance for two particles dσ = q ′ M (s.40) T ·V Combining this with the expressions for the width or the cross section one obtains for the decay width Γ = 1 2m = dR(m2 .43)
dΩcm =
π M (s. . .
11. . q n ) q n
qn √ Mni (q i . θ). implies for the scattering matrix M . δf n + i(2π)4 δ 4 (Pf − Pn ) (M † )f n or −i Mf i − (M † )f i = −
n
δni − i(2π)4 δ 4 (Pi − Pn ) Mni = δf i . The sign diﬀerence comes from the (conventional) sign in relation between S and quantummechanical and relativistic scattering amplitude.45)
Since the amplitudes also depend on all momenta the full result for two-particle intermediate states is (in CM. p1 .4
Unitarity condition
(S † )f n Sni = δf i
The unitarity of the S-matrix. ˆ ˆ Inserted in the unitarity condition for M . √ (2ℓ + 1)Mℓ (s) Pℓ (cos θ). θ)|2 .41) (11. .47) M (s. . q n ). see 11.
(11. p1 . i. The amplitude −M /8π s is the one to be compared with the quantum mechanical scattering amplitude f (E.e. t) λab 4π
dt. note the sign and cos θ = qi · qf ). q f ). (11. p1 . 16π 2 s
(11. . θcm ) √ q 8π s
2 2
(11. θ) in quantum mechanics. for which the √ amplitudes squared have been calculated in the previous chapter. (11. respectively.25) −i Mf i − (M † )f i = −
∗ dΩ(ˆn )Mnf (q f . pn ) |M |2 q 32π 2 m2 dΩ |M |2 . The result is N = |Mf i |2 dR(s.46)
Partial wave expansion
Often it is useful to make a partial wave expansion for the amplitude M (s. θ) = −8π s
ℓ
(in analogy with the expansion for f (E. . . 8π s 2π 8π s
. (M † )f n (2π)4 δ 4 (Pi − Pn ) Mni . for which one has dσ/dΩ = |f (E. .44)
This can be used to get the full expression for dσ/dt for eµ and e+ e− scattering. θ) or M (q i .Scattering theory
109
(Using normalized wave packets these somewhat ill-deﬁned manipulations can be made more rigorous).42)
2−body decay
The diﬀerential cross section (ﬁnal state not integrated over) is given by 1 dσ = √ |Mf i |2 dR(s. i M† M √ − √ 8π s 8π s =
fi n
dΩn
∗ Mnf qn Mni √ √ . (11. pn ). . (11. . .

σel = 4π
ℓ
(2ℓ + 1)|Mℓ (s)|2 (2ℓ + 1)
ℓ
=
4π q2
ηℓ e 2i δℓ − 1 2i
2
. Using σ= dΩ q ′ M (s. parametrized as Sℓ (s) = ηℓ (s) exp(2i δℓ (s)). 2i q 2i q (11. σinel = π q2
2 (2ℓ + 1) (1 − ηℓ ).
† −i (Mℓ )f i − (Mℓ )f i = 2
† (Mℓ )f n qn (Mℓ )ni .
(11.
(11. q ˆ
i.54)
ℓ
.53)
The diﬀerence is the inelastic cross section. Show that σT = = 4π q 2π q2 (2ℓ + 1) Im Mℓ (s)
ℓ
ℓ
(2ℓ + 1) (1 − ηℓ cos 2δℓ ). θcm ) √ q 8π s
2
.49)
or Im Mℓ = q |Mℓ |2 . n
(11.e.52)
From the imaginary part of M (s.50)
where Sℓ (s) satisﬁes |Sℓ (s)| = 1 and δℓ (s) is the phase shift. which allows writing Mℓ (s) = Sℓ (s) − 1 e2i δℓ (s) − 1 = .
(11.48)
If only one channel is present this simpliﬁes to
∗ ∗ −i (Mℓ − Mℓ ) = 2 qMℓ Mℓ .
in combination with the partial wave expansion for the amplitudes M and the orthogonality of the Legendre polynomials immediately gives for the elastic channel.Scattering theory
110
we obtain LHS = −i while for the RHS use is made of
ℓ
† (2ℓ + 1) Pℓ (ˆi · qf ) (Mℓ )f i − (Mℓ )f i . the total cross section can be determined. η = 1) the result σel = 4π q2 (2ℓ + 1) sin2 δℓ . q ˆ
Pℓ (ˆ · q ′ ) = q ˆ
(ℓ)
m
4π Y (ℓ) (ˆ) Ym (ˆ′ ) q (ℓ)∗ q 2ℓ + 1 m
and the orthogonality of the Ym functions to prove that RHS = 2
n ℓ † (2ℓ + 1)Pℓ (ˆi · qf ) (Mℓ )f n qn (Mℓ )ni .
(11. 0).
ℓ
(11. In general a given channel has |Sℓ (s)| ≤ 1.51)
and for the case that this is the only channel (purely elastic scattering.

56)
(again disregarding spin).57)
This shows that Γ is the width of the resonance.55)
(note that we have disregarded spin). in which case.
(11. however. . (q Mℓ )ij (s) = −M Γi Γj . We can also invert the situation and use unitarity to ﬁnd the imaginary part of an amplitude at some order from a lower order calculation. it is straightforward to write down the partial wave amplitude.e.58)
. It is important to note that the physical width of a particle is the imaginary part of the two-point vertex at s = M 2 . . This is (somewhat sloppy!) seen by considering the (amputated) 1PI two-point vertex Γ(2) = −i ∆
as the amplitude −iM for scattering a particle into itself (forward!) through the decay channels as intermediate states. The prescription for the pole structure. To check unitarity we need the amplitude at order e4 . η = 0.5
Unstable particles
i k 2 − M 2 + iǫ
For a stable particle the propagator is i∆(k) = (11. 0) ∝ e−i(E−iΓ/2)t . one has poles at k 0 = ±(Ek − iǫ) where Ek = + k2 + M 2 guarantees the correct behavior. We note that unitarity is generally broken in a ﬁnite order calculation. The quantity Γ is precisely the width for unstable particles. the time-evolution operator.
i. σel = σinel . which is given by a sum of the partial widths into the diﬀerent channels.
11. 0)|2 = e−Γt . s − M 2 + iM Γ (11. For the amplitude in a scattering process going through a resonance. . ∝ U (t. The unitarity condition then states 2 Im ∆−1 (k) = R
n † dR(p1 . such that |U (t. For an unstable particle one expects that U (t. pn )MRn (2π)4 δ 4 (k − Pn ) MnR
=
2M
n
Γn = 2M Γ. but still real. This is what we will do to discuss decay widths. For instance the tree-level calculation of Feynman diagrams at order e2 was real.e. This is achieved with a propagator i∆R (k) = i k2 − M2 + iM Γ (11.Scattering theory
111
Note that the total cross section is maximal in the case of full absorption. 0). The product of amplitudes is of order e4 . i. speciﬁcally one has for t > 0 that the Fourier transform is dk e
0 −ik0 t
∆(k) ∝
e−ik t dk (k 0 − Ek + iǫ)(k 0 + Ek − iǫ)
0
0
(t>0)
∝
e−iEk t . .

• The J/ψ resonance in e+ e− scattering.Scattering theory
112
(Prove this using the unitarity condition for partial waves). Knowing that each neutrino contributes about 160 MeV (see next chapter). The cross section at resonance is about 30 nb. dt s(s − 4m2 ) with f (t. a fast change in the phase shift for a narrow resonance. u) + g(t. The cross section σT (π − p) also shows a resonance with the same width with a value of about 70 mb. Note that because of the presence of a background the phase shift at resonance may actually be shifted. This implies that the resonance has spin J = 3/2 (decaying in a P-wave (ℓ = 1) pion-nucleon state) and isospin I = 3/2 (the latter under the assumption that isospin is conserved for the strong interactions). All these decays can be seen and leave an ’invisible’ width of 498 MeV.e. i. Three famous resonances are: • The ∆-resonance seen in pion-nucleon scattering.
Exercises
Exercise 11. which is attributed to neutrinos. the full width is Γ = 88 keV. u) = g(t. Essentially this resonance can decay into quark-antiquark pairs or into pairs of charged leptons. M2 − s (11. The phase shift in the resonating channel near the resonance is given by tan δℓ (s) = MΓ . its width Γ = 120 MeV. Limiting ourselves to a resonance in one channel.88 MeV. t) + g(u. This characteristic shape of a resonance is called the Breit-Wigner shape. This is a narrow resonance discovered in 1974. 2 q (s − M 2 )2 + M 2 Γ2 (11. one can reconstruct the resonance shape for diﬀerent numbers of neutrino species. u) = 1 1 2 (s + u2 ) + 4m2 (t − m2 ) t2 2 2 1 1 ( s − m2 )( s − 3m2 ) tu 2 2
.59)
reaching the unitarity limit for s = M 2 . the partial width into e+ e− is Γee = 5. From this one sees that a resonance has the same shape in all channels but diﬀerent strength. Three neutrinos explain the resonance shape. Its mass is M = 1232 MeV. u) + f (u. where furthermore σinel = 0. Its mass is M = 3096. • The Z 0 resonance in e+ e− scattering with M = 91.60)
showing that the phase shift at resonance rises through δ = π/2 with a ’velocity’ ∂δ/∂s = 1/M Γ. At resonance the cross section σT (π + p) is about 210 mb.4) can be written as 4πα2 dσ = {f (t. The half-width of the resonance is M Γ.26 keV.1
Show that the cross section for electron-electron scattering (exercise 10.2 GeV.49 GeV. it is furthermore easy to prove that the cross section is given by σel = M 2 Γ2 4π (2ℓ + 1) . Γ = 2. t)} .

999 877 Γ and Γ(π − → e− νe ) = 0.6 × 10−8 s and the relative decay widths Γ(π − → µ− νµ ) = 0. Compare the results with the experimental π − lifetime τ = 2.000 123 Γ. Calculate now fπ .
Exercise 11. ρcm ρcm a b 1 cm . 4. writing N (s) M1 (s) = 2 + iM Γ .3
Complete the calculation of the decay width and lifetime of the pion by including the appropriate phase space factors and using the calculation of the squared matrix element in Exc. Show that unitarity can be used one step further to ﬁx also the real part of the residue of the pole in the amplitude. As discussed for unstable particles. we have here already used unitarity to ﬁnd a ﬁnite imaginary part that is proportional to the total decay width. Compare the result with the factor trf 1/ρa ρb va that we used in the derivation of the basic expression for cross sections.Scattering theory
113
Exercise 11.2
Calculate the ﬂux factor. s − M∆ ∆ ∆
2 calculate Re N (M∆ )
.2). a quantity which we will ¯ ¯ encounter in the next chapter (section 12. Thus.4
The ∆ resonance in N π scattering appears in just one partial wave (a P-wave) as a pole at the position 2 s = s0 = M∆ − iM∆ Γ∆ .8. cm |va − vb |
in the center of mass system for two colliding particles a and b.
Exercise 11.

Chapter 12

**The standard model
**

12.1 Non-abelian gauge theories

In chapter 10 we have considered quantum electrodynamics as an example of a gauge theory. The photon ﬁeld Aµ was introduced as to render the lagrangian invariant under local gauge transformations. The extension to non-abelian gauge theories is straightforward. The symmetry group is a Lie-group G generated by generators Ta , which satisfy commutation relations [Ta , Tb ] = i cabc Tc , (12.1)

with cabc known as the structure constants of the group. For a compact Lie-group they are antisymmetric in the three indices. In an abelian group the structure constants would be zero (for instance the trivial example of U (1)). Consider a ﬁeld transforming under the group, φ(x) −→ U (θ) φ(x) = ei θ

a

La

φ(x) = (1 + i θa La ) φ(x)

inf

(12.2)

where La is a representation matrix for the representation to which φ belongs, i.e. for a threecomponent ﬁeld (written as φ) under an SO(3) or SU (2) symmetry transformation, φ −→ U (θ) φ = ei θ·L φ = φ − θ × φ.

inf

(12.3)

The complication arises (as in the abelian case) when one considers for a lagrangian density a L (φ, ∂µ φ) the behavior of ∂µ φ under a local gauge transformation, U (θ) = ei θ (x)La , φ(x) ∂µ φ(x) −→ −→ U (θ)φ(x), U (θ)∂µ φ(x) + (∂µ U (θ)) φ(x). (12.4) (12.5)

Introducing as many gauge ﬁelds as there are generators in the group, which are conveniently combined a in the matrix valued ﬁeld W µ = Wµ La , one deﬁnes Dµ φ(x) ≡ ∂µ − ig W µ φ(x), and one obtains after transformation D µ φ(x) −→ U (θ)∂µ φ(x) + (∂µ U (θ)) φ(x) − ig W ′ U (θ)φ(x). µ Requiring that Dµ φ transforms as Dµ φ → U (θ) Dµ φ (or Dµ → U (θ)Dµ U −1 (θ)) gives Dµ φ(x) −→ U (θ)∂µ φ(x) − ig U (θ) W µ φ(x), 114 (12.6)

The standard model

115

**which implies W ′ = U (θ) W µ U −1 (θ) − µ or inﬁnitesimal
**

′a a c Wµ = Wµ − cabc θb Wµ +

i (∂µ U (θ)) U −1 (θ), g

(12.7)

1 1 a ∂µ θa = Wµ + Dµ θa . g g It is necessary to introduce the free lagrangian density for the gauge ﬁelds just like the term −(1/4)Fµν F µν in QED. For abelian ﬁelds Fµν = ∂µ Aν − ∂ν Aµ = (i/g)[Dµ , Dν ] is gauge invariant. In the nonabelian case ∂µ W a − ∂ν W a does not provide a gauge invariant candidate for Gµν = Ga La , ν µ µν as can be checked easily. Generalizing the expression in terms of the covariant derivatives, provides a gauge invariant deﬁnition for Gµν . We have Gµν = with for the explicit ﬁelds transforming like

a a b c Ga = ∂µ Wν − ∂ν Wµ + g cabc Wµ Wν , µν

i [D , D ] = ∂µ W ν − ∂ν W µ − ig [W µ , W ν ], g µ ν

(12.8)

(12.9) (12.10)

The gauge-invariant lagrangian density is now constructed as

Gµν → U (θ) Gµν U −1 (θ).

1 1 L (φ, ∂µ φ) −→ L (φ, Dµ φ) − Tr Gµν Gµν = L (φ, Dµ φ) − Ga Gµν a 2 4 µν

(12.11)

1 with the standard normalization Tr(La Lb ) = 2 δab . Note that the gauge ﬁelds must be massless, as a 2 a µa mass term ∝ MW Wµ W would break gauge invariance.

**QCD, an example of a nonabelian gauge theory
**

As an example of a nonabelian gauge theory consider quantum chromodynamics (QCD), the theory describing the interactions of the colored quarks. The existence of an extra degree of freedom for each species of quarks is evident for several reasons, e.g. the necessity to have an antisymmetric wave function for the ∆++ particle consisting of three up quarks (each with charge +(2/3)e). With the quarks belonging to the fundamental (three-dimensional) representation of SU (3)C , i.e. having three components in color space ψr ψ = ψg , ψb the wave function of the baryons (such as nucleons and deltas) form a singlet under SU (3)C , 1 |color = √ (|rgb − |grb + |gbr − |bgr + |brg − |rbg ) . 6 (12.12)

The nonabelian gauge theory that is obtained by making the ’free’ quark lagrangian, for one speciﬁc species (ﬂavor) of quarks just the Dirac lagrangian for an elementary fermion, L = i ψ/ ψ − m ψψ, ∂ invariant under local SU (3)C transformations has proven to be a good candidate for the microscopic theory of the strong interactions. The representation matrices for the quarks and antiquarks in the fundamental representation are given by Fa Fa = = λa for quarks, 2 λ∗ − a for antiquarks, 2

**The standard model
**

xψ x+dx

116

(x)

ψ(x)

x+dx ψ(x+dx)

**x-space internal space
**

Figure 12.1: The vectors belonging to internal space located at each point in (one-dimensional) space which satisfy commutation relations [Fa , Fb ] = i fabc Fc in which fabc are the (completely antisymmetric) structure constants of SU (3) and where the matrices λa are the eight Gell-Mann matrices1 . The (locally) gauge invariant lagrangian density is 1 a L = − Fµν F µν a + i ψ/ ψ − m ψψ, D 4 with Dµ ψ = ∂µ ψ − ig Aa Fa ψ, µ (12.13)

a Fµν = ∂µ Aa − ∂ν Aa + g cabc Ab Ac . ν µ µ ν

Note that the term i ψ/ ψ = i ψ/ ψ + g ψ/ a Fa ψ = i ψ/ ψ + j µ a Aa with j µ a = g ψγ µ Fa ψ describes D ∂ A ∂ µ the interactions of the gauge bosons Aa (gluons) with the color current of the quarks (this is again µ precisely the Noether current corresponding to color symmetry transformations). Note furthermore that the lagrangian terms for the gluons contain interaction terms corresponding to vertices with three gluons and four gluons due to the nonabelian character of the theory. For writing down the complete set of Feynman rules it is necessary to account for the gauge symmetry in the quantization procedure. This will lead (depending on the choice of gauge conditions) to the presence of ghost ﬁelds. (For more details see e.g. Ryder, chapter 7.)

**A geometric picture of gauge theories
**

A geometric picture of gauge theories is useful for comparison with general relativity and topological considerations (such as we have seen in the Aharonov-Bohm experiment). Consider the space x x G (called a ﬁbre bundle). At each space-time point x there is considered to be a copy of an internal space G (say spin or isospin). In each of these spaces a reference frame is deﬁned. x ψ(x) denotes a ﬁeld vector ψ(x) which belongs to a representation of G, i.e. forms a vector in the internal space (see ﬁg. 12.1). The superscript x denotes that it is expressed with respect to the frame at point x, i.e. the basis of x G. Let ﬁelds Aa (x) determine the ’parallel displacements’ in the internal space, i.e. connect µ

1 The

Gell-Mann matrices are the eight traceless hermitean matrices generating SU (3) transformations, −i 1 1 i 1 −1 λ3 = λ2 = λ1 = λ4 = 1 1 λ5 = i −i 1 λ6 = 1 1

λ7 =

i

−i

1 1 λ8 = √ 3

−2

However.18)
or considering a path xµ (s) from a ﬁxed origin (0) to point x. (12. should be independent of such transformations.
(12.
x+dx
ψ(x + dx) = 1 + dxµ Dµ
x+dx
ψ(x). The relation in Eq.
-dx -dy
µ µ
dyµ
µ
dx
For a vector. if two diﬀerent paths to the same point give diﬀerent phases the eﬀects can be observed.e. x x+dx ψ(x + dx) = (1 + dxµ ∂µ ) ψ(x).19)
This gives rise to a (path dependent) phase in each point.15)
ψ(x).e. Dν ] ψ(x) 1 − ig dσ µν Gµν ψ(x). i.16.17) The three equations given so far immediately give Dµ = ∂µ − ig Aµ (x). ∂µ − ig Aµ (x) ψ(x) = 0 dxµ ∂µ − ig Aµ (x(s)) ψ(x) = 0.The standard model the basis for x G and
x+dx
117 G. =
.
x+dx
ψ(x)
= =
1 + ig dxµ Aa (x)Ta µ 1 + ig dx Aµ (x)
µ x
x
ψ(x)
(12. but expresses them with respect to diﬀerent bases. the real change thus is given by ψ(x) = (1 − dy ρ Dρ )(1 − dxσ Dσ )(1 + dy ν Dν )(1 + dxµ Dµ )ψ(x) = 1 + dxµ dy ν [Dµ . requiring that the ’connection’ Aµ (x) is such that Dµ → S(x)D µ S −1 (x). What is this physical eﬀect by which the ’connection’ Aµ can be observed? For this consider the phase around a closed loop. i. If there is a ’true’ diﬀerence in the vector ψ(x) and ψ(x + dx) it is denoted with the covariant derivative connecting the vectors expressed with respect to the same basis.
which connects two identical vectors.16)
which in the presence of the ’connection’ Aµ diﬀers from the total change between x ψ(x) and x+dx ψ(x+ dx). ds
which is solved by dψ(s) ds = dxµ . In principle such a phase in a given point is not observable.
We note that local gauge invariance requires that we can modify all local systems with a (local) unitary transformation S(x). 12.
(12. ds s dxµ ds′ P exp ig A (s′ ) ψ(0). A ’constant’ vector that only rotates because of the arbitrary deﬁnitions of local frames satisﬁes Dµ ψ(x) = 0. (12.14) (12. ds′ µ 0 ig Aµ (s) ψ(s)
dxµ Aµ (x)
ψ(s) =
which is the path-ordered integral denoted ψ(x) = P eig
P
ψ(0).

Similarly as the deﬁnition of the covariant derivative the eﬀect is thus frame-independent and we have the transformation law Gµν → S(x)Gµν S −1 (x). Nature will choose one. 2 2 4
−V (φ)
The potential V (φ) is shown in ﬁg. This characterizes spontaneous symmetry breaking.2
Spontaneous symmetry breaking
In this section we consider the situation that the groundstate of a physical system is degenerate. Consider as an example a ferromagnet with an interaction hamiltonian of the form H=− Jij Si · Sj . we can disregard those ’perturbations’ and look at the ideal situation. e.2: The symmetry-breaking ’potential’ in the lagrangian for the case that m2 < 0. Nevertheless. Classically the (time-independent) ground state is found for
V(φ )
F
|φ|
Figure 12.g. the groundstate itself appears degenerate. In this case there are many possible groundstates (determined by a ﬁxed direction in space).The standard model
118
where Gµν = (i/g)[Dµ .
12. external magnetic ﬁelds. φ= 2 φ3 (12. In geometric language the eﬀect on parallel transport of a vector depends on the ’curvature’ Gµν .2. in reality a not perfectly symmetric situation. and equivalently Aµ can be considered as a pure gauge eﬀect. If the temperature is below a certain critical temperature (T < Tc ) the kinetic energy is no longer dominant and the above hamiltonian prefers a lowest energy conﬁguration in which all spins are parallel.
i>j
which is rotationally invariant. As there can be one and only one groundstate. i. which means that by an appropriate transformation S(x) it can be gauged away (see the example of Aharonov-Bohm eﬀect).e. a theory for a scalar degree of freedom (a scalar ﬁeld) having three (real) components.20)
with a lagrangian density of the form L =
1 1 1 ∂µ φ · ∂ µ φ − m2 φ · φ − λ(φ · φ)2 . If it is zero one has dxµ Aµ (x) = 0. this means that there is more than one possibility for the groundstate. If the temperature is high enough the spins are oriented randomly and the (macroscopic) ground state is spherically symmetric. Only if this quantity is nonzero a physically observable eﬀect of Aµ exists.
. φ1 φ . 12. usually being (slightly) prejudiced by impurities. Dν ].

while the quadratic terms must correspond with non-negative masses. . It is suﬃcient to do this to second order in the ﬁelds as the higher (cubic. corresponding to the number of broken generators (in this case rotations around 1 and 2 axis) and 1 massive scalar particle with mass m2 = −2m2 . In order to see the consequences for the particle spectrum of the theory we construct the lagrangian in terms of the ﬁelds ϕ1 . ϕ2 and η.23) F +η
the latter only forming a minimum for m2 < 0. A quantum ﬁeld theory has only one nondegenerate groundstate |0 . The η massless particles are called Goldstone bosons. say 0 0 . has less symmetry. In this situation one speaks of spontaneous symmetry breaking. so one has 0|φquantum |0 = 0 and hence 0|φ|0 = ϕc . They are known as the Weyl mode or the Goldstone mode: Weyl mode. φ = ϕc + φquantum where for the (operator-valued) coeﬃcients in the quantum ﬁeld one wants 0|c† = c|0 = 0. while the lagrangian including the nonzero groundstate expectation value chosen by nature. (12. In this mode the lagrangian and the vacuum are both invariant under a set of symmetry transformations generated by Qa . In the case of degeneracy.22) F
Realization of symmetries (model independent)
In this section we want to discuss a bit more formal the two possible ways that a symmetry can be implemented.
The situation now is the following. . The result is 1 1 1 1 (∂µ ϕ1 )2 + (∂µ ϕ2 )2 + (∂µ η)2 − m2 (ϕ2 + ϕ2 ) L = 1 2 2 2 2 2 1 2 1 − m (F + η)2 − λ (ϕ2 + ϕ2 + F 2 + η 2 + 2 F η)2 (12.24) 1 2 2 4 1 1 1 = (∂µ ϕ1 )2 + (∂µ ϕ2 )2 + (∂µ η)2 + m2 η 2 + .) terms constitute interaction terms. . Any constant ﬁeld with ’length’ F is a possible groundstate. It is only invariant under rotations around the 3-axis.e. for the vacuum Qa |0 = 0. Writing the ﬁeld φ as a sum of a classical and a quantum ﬁeld. It is appropriate to redeﬁne the ﬁeld as ϕ1 φ = ϕ2 . The original lagrangian contained an SO(3) invariance under (length conserving) rotations among the three ﬁelds. The ﬁeld along the third axis plays a special role because of the choice of the vacuum expectation value. λ
such that 0|ϕ1 |0 = 0|ϕ2 |0 = 0|η|0 = 0. In this case the spectrum is described
. ϕc = 0|φ |0 = (12. (12. etc. therefore a choice must be made. The presence of a nonzero value for the classical groundstate value of the ﬁeld will have an eﬀect when the ﬁeld is quantized.25) 2 2 2 Therefore there are 2 massless scalar particles.The standard model
119
a constant ﬁeld (∇φ = 0) and the condition ∂V ∂φ
ϕc
=0
−→
ϕc · ϕc = 0 or ϕc · ϕc = −
m2 ≡ F 2. The classical groundstate appears degenerate.21) Stability of the action requires the classical groundstate ϕc to have a well-deﬁned value (which can be nonzero). (12. i.

e. i. T = τ /2. if the generators Qa generate a symmetry.32) (12. .
a 0|∂ µ Jµ (x)|π b (k) = fπ k 2 ei k·x δab = fπ m2 a ei k·x δab . H] = 0. Using translation invariance and as kµ is the only four vector on which this matrix element could depend one may write
a 0|Jµ (x)|π b (k) = fπ kµ ei k·x δab
(fπ = 0)
(12. Lx and Ly or instead of the latter two L+ and L− ) are used to label the multiplet members or transform them into one another. i.
(12. This lagrangian density then is invariant under unitary (vector) transformations in the ﬂavor space.e. T ] ψ. The conserved currents corresponding to this symmetry transformation are found directly using Noether’s theorem (see chapter 6). . The generators Qa (in that case the rotation operators Lz . down. . denoted |π a (k) . Goldstone mode. and |a and |a′ belong to the same multiplet (there is a Qa such that |a′ = Qa |a ) then H|a = Ea |a implies that H|a′ = Ea |a′ . M = . Note that for the ﬁelds π a (x) one would have a the relation 0|π a (x)|π a (k) = ei k·x .28)
where ψ is extended to a vector in ﬂavor space and M is a diagonal matrix in ﬂavor space. Taking the derivative. ψ −→ ei α·T ψ. suggesting the stronger relation ∂ µ Jµ (x) = fπ m2 a π a (x). ψu mu md ψ = ψd .) one can write L =
f
ψ f (i/ − Mf )ψf .30)
which in the case that we include only two ﬂavors form an SU (2)V symmetry (isospin symmetry) generated by the Pauli matrices.26)
for all the states labeled by a corresponding to ’broken’ generators. irrespective of the fact if they annihilate the vacuum. π
(12. a massless Goldstone boson for each ’broken’ generator. This means that they are operators that create states from the vacuum. etc. a and a′ are degenerate states. A bit more formal. i. Using the Dirac equation. . As the a generators for a symmetry are precisely the zero-components of a conserved current Jµ (x) integrated a a over space. [Qa . In this mode the lagrangian is invariant but Qa |0 = 0 for a number of generators. the lagrangian for the quarks consists of the free Dirac lagrangian for each of the types of quarks.27)
If the transformations in the lagrangian give rise to a symmetry the Noether currents are conserved.The standard model
120
by degenerate representations of the symmetry group. called ﬂavors.31)
. (12.e. = ψ(i/ − M )ψ.. π
Example: chiral symmetry
An example of spontaneous symmetry breaking is chiral symmetry breaking in QCD. there must be a nonzero expectation value 0|Jµ (x)|π (k) . Neglecting at this point the local color symmetry. Known examples are rotational symmetry and the fact that the spectrum shows multiplets labeled by angular momentum ℓ (with members labeled by m).29)
(Note that each of the entries in the vector for ψ in the above equation is still a 4-component Dirac spinor). strange. it is easy to see that one gets ∂µ V µ = i ψ [M. a ∂ µ Jµ = 0. (12. V µ = ψγ µ T ψ. ∂ ∂
(12. Including a sum over the diﬀerent ﬂavors (up. and one must have mπa = 0.

parity minus). including in Dirac space a γ5 matrix. Note that these transformations also work on the spinor indices. where the quark masses are not completely zero. which is approximately true for the up and down quarks. in terms of which the Dirac lagrangian 2 density looks like L = i ψL / ψL + i ψR / ψR − ψR M ψL − ψL M ψR . 12. fπ = 0 remains. T } γ5 ψ. with almost degenerate masses (Mp = 938. The chiral ﬁelds ψR and ψL are transformed into each other under parity.3 MeV/c2 and Mn = 939. it is easy to see that one gets ∂µ Aµ = i ψ {M. for which the ﬁeld (according to the discussion above) has the same behavior under parity.e. SU (2)V (isospin) symmetry is good if the up and down quark masses are identical. T ] = 0. proton and neutron. is what happens in the real world. Therefore realization in the Weyl mode would require that all particles come double with positive and negative parity.g. gives also rise to a nonzero mass for the Goldstone bosons according to Eq. but the fact that the symmetry is not perfect and the right hand side of Eq. ψR/L = 1 (1 ± γ5 )ψ. etc. while the groundstate is only invariant under isospin rotations (R = L). nucleons and atomic nuclei has not only an isospin or vector symmetry SU (2)V but also an axial vector symmetry SU (2)A . etc. From group theory (Schur’s theorem) one knows that the latter can only be true. T γ5 = τ γ5 /2.27. ∂ ∂ (12. chiral symmetry is not perfect. That the massless theory has this symmetry can also be seen by writing it down for the socalled lefthanded and righthanded fermions. We know that mesons and baryons (such as the nucleons) have a well-deﬁned parity that is conserved. behaves as a pseudoscalar particle (spin zero. i. if in ﬂavor space M is proportional to the unit matrix.e. generated by TR/L = 1 (1 ± γ5 )T . Still the basic fact that the generators acting on the vacuum give a nonzero result (i. both are very small. nature has choosen the path SU (2)L ⊗ SU (2)R =⇒ SU (2)V .33)
which in the case of two ﬂavors form SU (2)A transformations generated by the Pauli matrices. the lagrangian density is invariant under left (L) and right (R) rotations independently. The currents corresponding to this symmetry transformation are again found using Noether’s theorem.
. stated equivalently. much smaller than any of the other mesons or baryons. e. This symmetry. Aµ = ψγ µ T γ5 ψ.35)
In this case ∂µ Aµ = 0 will be true if the quarks have zero mass. The Goldstone bosons of QCD are the pions for which fπ = 93 MeV and which have a mass of mπ ≈ 138 MeV/c2 . (12. This symmetry is realized in the Weyl mode with the spectrum of QCD showing an almost perfect isospin symmetry. From the number of broken generators it is clear that one expects three massless Goldstone bosons.
i. M = m · 1. parity would not play a role in the world. (12. In the real world.36) If the mass is zero the lagrangian is split into two disjunct parts for L and R showing that there is a direct product SU (2)L ⊗ SU (2)R symmetry. 12. or. ψ −→ ei α·T γ5 ψ. as the quantity ∂µ Aµ (x). which is equivalent 2 to the V-A symmetry. i. There exists another set of symmetry transformations. This combined symmetry is what one calls chiral symmetry.35 is nonzero.34) Using the Dirac equation. but also a triplet (isospin 1) of pions.6 MeV/c2 ). (12. The conclusion is that the original symmetry of the lagrangian is spontaneously broken and as the vector part of the symmetry is the well-known isospin symmetry.e. This situation. however.The standard model
121
Furthermore ∂µ V µ = 0 ⇐⇒ [M. I. (leaving out the ﬂavor structure) the same as ψγ5 ψ. How can we see this. is by nature not realized in the Weyl mode.e. a doublet (isospin 1/2) of nucleons. Therefore the world of up and down quarks describing pions. socalled axial vector transformations.e.

41) φ3 F +η Explicitly one then has 2 2 gF Wµ + g Wµ η 1 1 −gF Wµ − g Wµ η Dµ φ = ∂µ φ + g Wµ × φ = ∂µ η .38)
∂µ Wν − ∂ν Wµ + g Wµ × Wν . Dµ φ = Gµν = ∂µ φ + g Wµ × φ. Note that the number of massless gauge bosons (in this case one) coincides with the number of generators corresponding to the remaining symmetry (in this case rotations around the 3-axis). having two independent spin components) and three scalar ﬁelds (one degree of freedom each). one sees that the ﬁelds Wµ and Gµν also can be represented as three-component ﬁelds. Therefore we can always rotate the ﬁeld φ into the 3-direction in order to simplify the calculation. as in this case there are no massless Goldstone bosons.39) (12.3
The Higgs mechanism
The Higgs mechanism occurs when spontaneous symmetry breaking happens in a gauge theory where gauge bosons have been introduced in order to assure the local symmetry. (12. 0 φ= (12.42) 2 from which one reads oﬀ that the particle content of the theory consists of one massless gauge boson 3 1 2 (Wµ ).
(12.40)
a Dµ φ = ∂µ φ − ig Wµ La φ.
The symmetry is broken in the same way as before and the same choice for the vacuum. made into a gauge theory.
is made. Initially there are 3 massless gauge ﬁelds (each. in this case three-dimensional) representation reads (La )ij = −i ǫaij . 4 2 where Since the explicit (adjoint. . (12. . while the number of massive gauge bosons coincides with the number of ’broken’ generators.e. 2 massive vector ﬁelds or spin 1 bosons (2 × 3) and one scalar ﬁeld (1). .37) (12. F
which gives for the lagrangian density up to quadratic terms L
1 1 λ 1 = − Gµν · Gµν + Dµ φ · Dµ φ − m2 φ · φ − (φ · φ)2 4 2 2 4 1 2 2 1 1 2 = − (∂µ Wν − ∂ν Wµ ) · (∂ µ W ν − ∂ ν W µ ) − g F Wµ W µ 1 + Wµ W µ 2 4 2 1 + (∂µ η)2 + m2 η 2 + . two massive bosons (Wµ and Wµ with MW = gF ) and a massive scalar particle (η with m2 = η 2 −2 m . The latter is a spin 0 particle (real scalar ﬁeld) called a Higgs particle. After symmetry breaking the same number (as expected) comes out. One may wonder about the degrees of freedom. We have the possibility to perform local gauge transformations. i. thus 9 independent degrees of freedom.
. but one has 1 massless gauge ﬁeld (2). Considering the same example with rotational symmetry (SO(3)) as for spontaneous symmetry breaking of a scalar ﬁeld (Higgs ﬁeld) with three components. again 9 degrees of freedom. like a photon. The diﬀerence comes when we reparametrize the ﬁeld φ. 0 ϕc = 0|φ |0 = 0 .The standard model
122
12. 1 1 L = − Gµν · Gµν + Dµ φ · Dµ φ − V (φ). 0 0 0 = .

implies certain nontrivial relations between masses and mixing angles that can be tested experimentally and sofar are in excellent agreement with experiment. the quarks (up. etc.46)
−→
3 One notes that the charges of the leptons can be obtained as Q = TW − 1/2 for lefthanded particles 3 and Q = TW − 1 for righthanded particles. For the neutrino only a lefthanded particle (and righthanded antiparticle) exist. while the lefthanded particles form a doublet. 1 As they are massless. In the standard model one starts with a very simple basic lagrangian for (massless) fermions which exhibits more symmetry than observed in nature.
(12. R −→ e−i β R.45) (12. R.The standard model
123
12.44)
which has an SU (2)W symmetry under transformations eiα·TW . under which the lefthanded and righthanded particles with YW (L) = −1 and YW (R) = −2 transform with eiβYW /2 . The procedure. The lagrangian for the leptons consists of three families each containing an elementary fermion (electron e− . By introducing gauge ﬁelds and breaking the symmetry a more complex lagrangian is obtained.51)
. muon µ− or tau τ − ). the leptons (elektrons.48) (12. ∂ ∂ ∂
(12. τ ). ψR/L = 2 (1 ± γ5 )ψ decouple.49)
Next the SU (2)W ⊗ U (1)Y symmetry is made into a local symmetry introducing gauge ﬁelds Wµ and Bµ in the covariant derivative Dµ = ∂µ − i g Wµ · TW − i g ′ Bµ YW /2.e. its corresponding neutrino (νe . i. 0 1 +1/2 L νe 3 − = with TW = and TW = L= −1/2 eL L 2 R = R − = eR
3 with TW = 0 and TW = 0. explicitly L R
SU(2)W SU(2)W
−→
ei α·τ /2 L. νµ and ντ ) and their antiparticles. explicitly Dµ L Dµ R i i g Wµ · τ L + g ′ Bµ L. however. 2 2 = ∂µ R + i g ′ Bµ R. down.4
The standard model SU (2)W ⊗ U (1)Y
The symmetry ideas discussed before play an essential role in the standard model that describes the elementary particles.and righthanded particles. Thus L
(f )
= i eR / eR + i eL / eL + i νeL / νeL + (µ. that gives a good description of the physical world. electromagnetic and weak forces.). neutrinos. muons.
(12. explicitly L −→ e−i β/2 L. = ∂µ L − (12.47)
and YW is considered as an operator that generates a U (1)Y symmetry. etc. left. 2
(12.50) (12.) and the gauge bosons responsible for the strong.
Thus the lagrangian density is L
(f )
∂ ∂ = i L/ L + i R/ R. This is written as 3 Q = TW +
YW .
U(1)Y U(1)Y
(12.43)
and
One introduces a (weak) SU (2)W symmetry under which eR forms a singlet.

.
(12. TW = ±1 or 0 and YW = 0 (thus Q = TW ) and 3 Bµ is a singlet under SU (2)W (TW = TW = 0) and also has YW = 0. h)
(12. (12. .
−V (φ)
L and
(h2)
= −Ge (LφR + Rφ L).
(12. For the classical ﬁeld the choice θ4 = v is made. + 8
(12. leading to the parametrization 1 0 φ(x) = √ (12. which assures with the choice of YW for the Higgs ﬁeld that Q generates the remaining U (1) symmetry. .53) φ= √ (θ4 − iθ3 ) φ 2 with TW = 1/2 and YW = 1 is introduced. 4 4
In order to break the symmetry to the symmetry of the physical world. .The standard model
124
3 3 where Wµ is a triplet of gauge bosons with TW = 1. + 8 g2 v2 1 + − (Wµ )2 + (Wµ )2 (∂µ h)2 + m2 h2 + 2 8 (g 2 + g ′2 ) v 2 (Zµ )2 + . a complex Higgs ﬁeld + 1 φ √2 (θ2 + iθ1 ) 0 = 1 (12. with the following lagrangian density consisting of a symmetry breaking piece and a coupling to the fermions.52)
L L
(f 1) (f 2)
= =
i Rγ µ (∂µ + ig ′ Bµ )R + i Lγ µ (∂µ +
i ′ i g Bµ − g Wµ · τ )L 2 2 1 1 − (∂µ Wν − ∂ν Wµ + g Wµ × Wν )2 − (∂µ Bν − ∂ν Bµ )2 .
†
i i g Wµ · τ − g ′ Bµ )φ.56) v + h(x) 2 Dµ φ = (∂µ − and Dµ φ = =
1 √ 2
− ig 2
∂µ h +
1 2 Wµ −iWµ √ (v + h) 2 3 gWµ −g′ Bµ i √ (v + 2 2
Up to cubic terms.57)
=
1 g2 v2 1 2 (Wµ )2 + (Wµ )2 (∂µ h)2 + m2 h2 + 2 8 v2 2 3 gWµ − g ′ Bµ + . Putting this in leads to
L
(f )
=L
(f 1)
+L
(f 2)
. . L where L
(h1) (h)
=L
(h1)
+L
(h2)
. the U (1)Q symmetry (generated by the charge operator). Using local gauge invariance θi for i = 1.55) 2 2 The Higgs potential V (φ) is choosen such that it gives rise to spontaneous symmetry breaking with ϕ† ϕ = −m2 /2λ ≡ v 2 /2. 2 and 3 may be eliminated (the necessary SU (2)W rotation is precisely e−iθ(x)·τ ).59)
. this leads to the lagrangian L
(h1)
.54)
= (Dµ φ)† (Dµ φ) −m2 φ† φ − λ (φ† φ)2 .58)
(12.

and correspond to three massive particle ﬁelds (W ± and Z 0 ) and one massless ﬁeld (photon γ) with
2 MW 2 MZ 2 Mγ
= = =
g2 v2 . 2
(12.69)
. 2
Zµ (12. we can read oﬀ e = g sin θW = g ′ cos θW .The standard model
125
where the quadratically appearing gauge ﬁelds that are furthermore eigenstates of the charge operator are
± Wµ
= = =
Zµ Aµ
1 1 2 √ Wµ ± i Wµ .66)
From the coupling to the photon. σ (νµ γ ρ µL ) 2 2 (eL γ νe ) 2 k + MW (12.62)
g 2 + g ′2
3 ≡ sin θW Wµ + cos θW Bµ .64) (12. g ′ /g = tan θW . for instance in the amplitude for the decay of the muon
νµ µ− W − e−
νµ
=
µ−
e− νe
νe
is given by −i M = − ≈ i g2 −i gρσ + .68)
g2 ρ 5 2 (νµ γρ (1 − γ5 )µ) (eγ (1 − γ )νe ) 8 MW
(µ)† (jL )ρ (e) (jL )ρ
GF (µ)† (e) ≡ i √ (jL )ρ (jL )ρ .65)
The weak mixing angle is related to the ratio of coupling constants. .61) (12.63) (12.60) (12. 2 3 g Wµ − g ′ Bµ 3 ≡ cos θW Wµ − sin θW Bµ . 2 + g ′2 g
3 g ′ Wµ + g Bµ
(12. (12.67)
The coupling of electrons or muons to their respective neutrinos. 4 2 MW g2 v2 = . The coupling of the fermions to the physical gauge bosons are contained in L (f 1) giving L
(f 1)
=
i eγ µ ∂µ e + i νe γ µ ∂µ νe − g sin θW eγ µ e Aµ g 1 1 + sin2 θW eR γ µ eR − cos 2θW eL γ µ eL + νe γ µ νe cos θW 2 2 g µ + µ − + √ νe γ eL Wµ + eL γ νe Wµ . 2θ 2θ 4 cos W cos W 0. .
(12.

40 GeV. Comparing this with the total width into (invisible!) channels. from the MW mass is about 250 GeV. (12. Finally we want to say something about the weak properties of the quarks. the mass of the electron comes from the spontaneous symmetry breaking but is not predicted (it is in the coupling Ge ). as appear for instance in the decay of the neutron or the decay of the Λ (quark content uds). The coupling to the Higgs particle is weak as the value for v calculated e.19 GeV.e. one has the relation g2 e2 1 GF √ = .72) (12.05 and CA = −1/2 we calculate Γ(e+ e− ) ≈ 78.73) (12.The standard model
126
the good old four-point interaction introduced by Fermi to explain the weak interactions.5 MeV (exp. Γinvisible = 480 MeV one sees that three families of (light) neutrinos are allowed. muon or tau. 80. With the choosen vacuum expectation value for the Higgs ﬁeld.
−5
(12.80)
First.166 4 × 10 0. Actually including corrections corresponding to higher order diagrams the agreement for the decay width into electrons can be calculated much more accurately and the number of allowed (light) neutrinos turns to be even closer to three.75)
The coupling of the Z 0 to fermions is given by (g/ cos θW )γ µ multiplied with
3 TW
1 1 1 (1 − γ5 ) − sin2 θW Q ≡ CV − CA γ5 . 2 2 2
(12. g ′ and v determine a number of experimentally measurable quantities. g2 MZ 2 2 (CV + CA ). giving rise to a running coupling constant after renormalization of the ﬁeld theory. one obtains L
(h2)
Ge v Ge = − √ (eL eR + eR eL ) − √ (eL eR + eR eL ) h 2 2 me = −me ee − eeh. such as2 α = e2 /4π GF sin2 θW MW MZ ≈ 1/127. me /v is extremely small.77) (12. 91.70) = 2 = 2 sin2 θ 8 MW 2 v2 8 MW 2 W In this way the parameters g. v
(12. 1. i.
. 3 = TW . i.231 2.78)
From this coupling it is straightforward to calculate the partial width for Z 0 into a fermion-antifermion pair.
(12.76)
with CV CA
3 = TW − 2 sin2 θW Q.
(12.g. The masses of the fermions and the coupling to the Higgs particle are contained in L (h2) . Γe ≈ Γµ ≈ Γτ ≈ 83 MeV). For each neutrino species (with CV = 1/2 and CA = 1/2 one expects Γ(νν) ≈ 155 MeV.74) (12.e.71) GeV
−2
= = = =
. leptons with CV = −1/2 + 2 sin2 θW ≈ −0.
2 The value of α deviates from the known value α ≈ 1/137 because of higher order contributions.79) Γ(Z 0 → f f ) = 48π cos2 θW For the electron. (12.

81)
n −→ pe− νe
⇐⇒
d −→ ue− νe . Decay of B-mesons containing b-quarks allow estimate of Vub . their antiparticles and the electroweak gauge bosons as appearing in each family. Q = TW + YW /2 and constant along speciﬁc diagonals as indicated in the ﬁgure. Comparing neutron decay and Λ decay one can get an estimate of the mixing parameter Vus in the socalled Cabibbo-Kobayashi-Maskawa mixing matrix. 12. In principle one complex phase is allowed in the most general form of the CKM matrix.3. The electric charge Q is then 3 ﬁxed. t c u ′ . for which SU (5) or SO(10) are actually nice candidates. i. suggesting an underlying larger unifying symmetry group.e. The
. ′ ′ b s d L L L ′ d Vud ′ s = V cd ′ b Vtd L Vus Vcs Vts Vub Vcb Vtb d s b L (12. As shown in Fig.
Λ −→ pe− νe ¯
⇐⇒
s −→ ue− νe . this requires particular YW -TW assignments to get the charges right. We will not discuss this any further in this chapter. A complication arises for quarks (and as we will discuss in the next section in more detail also for leptons) as it are not the ’mass’ eigenstates that appear in the weak isospin doublets but linear combinations of them.3: Appropriate YW and TW assignments of quarks. but with diﬀerent strength. The pattern is actually intriguing. which can account for the (observed) CP violation of the weak interactions. CP violation requires three generations. ¯
3 This mixing allows all quarks with TW = −1/2 to decay into an up quark. This is only true if the mixing matrix is at least three-dimensional. leptons.
eR
Q = −1
u d W eνe s W -
u
eνe
where
The quarks also turn out to ﬁt into doublets of SU (2)W for the lefthanded species and into singlets 3 for the righthanded quarks.The standard model
127
YW eL
uR
νR
+1
eR
dL dL
_
uL
W
0
W
−1/2
B0
+1/2
W
+
I3 W
uR dR eL
−1
dR
Q = +1
νL
Q=0
uL
3 Figure 12. etc.

The Higgs ﬁeld is still limited to one complex doublet. e.g.227.e.q)
= −QL φΛd DR − DR Λ† φ† QL − QL φc Λu UR − UR Λ† φc† QL u d
(12.q)
′ weak † mass mass DL = DL = Vu Vd DL = VCKM DL . the three righthanded quarks with charge +2/3 in UR and the three righthanded quarks with charges −1/3 in DR . respectively). d d
(12.
(12. for which we need the appropriate weak isospin doublet 0∗ 1 φ c = √ v +h .82 and ρ ≈ 0.85) √ √ with Mu = Gu v/ 2 (diagonal matrix containing mu . One then reads oﬀ that starting with the family basis as deﬁned via the left doublets that the mass eigenstates (and states coupling to the Higgs ﬁeld) involve the righthanded † mass † mass mass † mass states UR = Wu UR and DR = Wd DR and the lefthanded states UL = Vu UL and DL = † Vd DL . UR = uR cR tR .82)
space. The imaginary part i η gives rise to CP violation in decays of K and B-mesons (containing s and b quarks.
magnitudes of the entries in the CKM matrix are nicely represented parametrization 1 − 1 λ2 λ λ3 A(ρ − i η) 2 1 2 −λ 1− 2λ λ2 A V = 3 2 λ A(1 − ρ − i η) −λ A 1
12. allowing us to write
† Λu = Vu Gu Wu † and Λd = Vd Gd Wd . each of these containing the three families. A ≈ 0. The Λu and Λd are complex matrices in the 3×3 family
and Λd Λ† = Vd G2 Vd† . Working with the mass eigenstates one simply sees that the weak current coupling to the W ± mass † mass becomes U L γ µ DL = U L γ µ Vu Vd DL . the weak mass eigenstates are
L
(h2.The standard model
128 using the socalled Wolfenstein + O(λ4 )
with λ ≈ 0. ms and mb ).86)
the unitary CKM-matrix introduced above in an ad hoc way. i.83)
where Vu and Vd are unitary matrices. Note that we need the conjugate Higgs ﬁeld to get a U (1)Y singlet in the case of the charge +2/3 quarks.34.22 and η ≈ 0.
† Λu Λ† = Vu G2 Vu . u u
where we include now the three lefthanded quark doublets in QL .84)
with Gu and Gd being real and positive and Wu and Wd being diﬀerent unitary matrices.
. Thus one has
† † † =⇒ −DL Vd Md Wd DR − DR Wd Md Vd† DL − UL Vu Mu Wu UR − UR Wu Mu Vu UL (12.5
Family mixing in the Higgs sector and neutrino masses
The quark sector
Allowing for the most general (Dirac) mass generating term in the lagrangian one starts with L
(h2.
(12. φ = −φ− 0 2 For the (squared) complex matrices we can ﬁnd positive eigenvalues. mc and mt ) and Md = Gd v/ 2 (diagonal matrix containing md .

ℓ) =⇒ −Me EL Ve We ER − ER We Ve† EL . For the (massless) neutrino ﬁelds we just can redeﬁne ﬁelds into NL = Ve† NL . As before.
UPMNS
a parametrization that in principle also could have been used for quarks. since the weak current is the only place where they show up.ℓ)
= −LφΛe ER − ER Λ† φ† L.89)
with three righthanded neutrinos added to the previous case.5×10−3 eV2 .
(12. † † Again we continue as before now with matrices Λe = Ve Ge We and Λn = Vn Gn Wn . EL = Ve EL . Working with the mass eigenstates and the weak current becomes EL γ NL = EL e n L weak for the charged leptons we see that the weak eigenstates for the neutrinos are NL = Ve† NL with the relation to the mass eigenstates for the lefthanded neutrinos given by † ′ weak mass mass NL = NL = Ve† Vn NL = UPMNS NL . (12.92)
.
(12. there is no family mixing for massless neutrinos.91)
† with UPMNS = Vn Ve known as the Pontecorvo-Maki-Nakagawa-Sakata mixing matrix.ℓ)
= −LφΛe ER − ER Λ† φ† L − Lφc Λn NR − NR Λ† φc† L e n
(12.88) √ mass with Me = Ge v/ 2 the diagonal mass matrix with masses me . mµ and mτ . it is particularly useful because θ12 is essentially determined by solar neutrino oscillations requiring ∆m2 ≈ 8 × 10−5 12 eV2 (convention m2 > m1 ).The standard model
129
The lepton sector (massless neutrinos)
For a lepton sector with a lagrangian density of the form L in which
(h2. The mixing is intriguingly close to the Harrison-Perkins-Scott tri-bimaximal 23 mixing matrix 0 0 2/3 1/3 0 2/3 1/3 0 1 1/2 − 1/2 − 1/3 UHPS = 0 1/3 − 1/2 . one can write Λe = Ve Ge We and we ﬁnd † L (h2.90)
mass † mass mass † mass † We note that there are mass ﬁelds ER = We ER .87)
is a weak doublet containing the three families of neutrinos (NL ) and charged leptons (EL ) and ER is † a three-family weak singlet. The W -current then becomes mass mass E L γ µ NL = E L γ µ NL . e N L= L EL
(12.ℓ) † † † =⇒ −EL Ve Me We ER − ER We Me Ve† EL − NL Vn Mn Wn NR − NR Wn Mn Vn NL .
(12. For neutrino’s this matrix is parametrized in terms of three angles θij with cij = cos θij and sij = sin θij and one angle δ. we ﬁnd massless neutrinos.e. decoupling from all known interactions. i.
The lepton sector (massive Dirac neutrinos)
In principle a massive Dirac neutrino could be accounted for by a lagrangian of the type L
(h2. In this case. NL = Vn NL and NR = Wn NR µ mass γ µ V † V N mass . The mass ﬁelds ER † mass † mass = We ER . = − 1/6 2/3 0 1/2 1/2 0 0 1 0 1/3 1/2 − 1/6 (12. and obtain L
(h2. EL = Ve† EL . while θ23 then is determined by atmospheric neutrino oscillations requiring |∆m2 | ≈ 2.93)
1 = 0 0
0 c23 −s23
0 s23 c23
c13 0 −s13 eiδ
0 s13 eiδ 1 0 0 c13
c12 −s 12 0
s12 c12 0
0 0 1
.

2 R ψ = nR + nL . If the Majorana mass MR ≫ MD one actually obtains in a natural way one Majorana neutrino with a very small mass. 2 although this option is not attractive as it violates the electroweak symmetry. 1 ∗ c c (12.88 a Majorana mass term for the (lefthanded) neutrino mass eigenstates. is to add in Eq. α2 and δ).
The see-saw mechanism
Consider (for one family N = n) the most general Lorentz invariant mass term for two independent Majorana spinors. 12. The mass matrix can be written as ML |MD | eiφ M = (12. Actually. (12. with for the righthanded sector a mass term MR . the massless and massive Majorana neutrinos. For the leptons. 1 L Υ ′ = nR + nc .The standard model
130
The lepton sector (massive Majorana ﬁelds)
An even simpler option than sterile righthanded Dirac neutrinos.97)
As the most general mass term in the lagrangian density we have L
mass
=
−
=
which for MD = 0 is a pure Majorana lagrangian and for ML = MR = 0 and real MD represents the Dirac case. are equivalent to two decoupled Majorana neutrinos (see below).99)
. Thus (disregarding family structure) one has two Majorana neutrinos. − MD MR nR 2
(12.100) |MD | eiφ MR
1 1 ∗ ∗ M L nc nL + M L nL nc − M R nR nc + M R nc nR L R L R 2 2 1 1 ∗ ∗ − M D nc nc + M D nL nR − M D nR nL + M D nc nc L R R L 2 2 1 nc nR M L M D nL L c + h. Absorption of phases in the states is not possible for Majorana neutrinos. Υ′ and Υ′ (satisfying Υc = Υ and as discussed in chapter 6. one being massive. hence the mixing matrix becomes iα /2 0 0 e 1 (12.ν
=−
containing three (CP-violating) phases (α1 . 0 0 1 L
mass. We use here the primes starting with the weak eigenstates. a unitary matrix relating them to the weak eigenstates.98) (12.ν = − MR NR NR + MR NR NR .94) M L NL NL + M L NL NL . one must for the neutrinos as well as charged leptons.96) VPMNS = UPMNS K with K = 0 eiα2 /2 0 .95) 2 In order to have more than a completely decoupled sector. it is easy to see R that this incorporates the Dirac case by considering the lefthanded part of Υ′ and the righthanded 1 part of Υ′ as a Dirac spinor ψ.and lefthanded species through Dirac mass terms coming from the coupling to the Higgs sector as in the previous section. however. (12. as above. couple the right. For the charged leptons there cannot exist a Majorana mass term as this would break the U(1) electromagnetic symmetry. the left. 1 c ∗ c L mass. Υc ≡ (ΥL )c = ΥR 1 2 L and Υc = ΥL ).and righthanded species then just form a Dirac fermion.c. coupled by a Dirac mass term. This is called the see-saw mechanism (outlined below). For the neutrino sector. For these light Majorana neutrinos one has. The way to circumvent this is to introduce as in the previous section righthanded neutrinos. Thus 2 Υ ′ = nc + nL .

3
(a) The coupling of the Z 0 particle to fermions is described by the vertex −i g 2 cos θW
f f CV γ µ − CA γ µ γ5 .104)
↔ 1 1 Υi i / Υi − Mi Υi Υi ∂ 4 2
(12.e. H] = 0. [Qa .
Exercise 12. a and a′ are degenerate states.The standard model
131
taking ML and MR real and non-negative.
2 U (M M † ) U † = M0 . 2 with real masses Mi .
(12. Prove that if the generators Qa generate a symmetry. i. Υ2R nR
(12. L Υ2L nR
Υ1 and Υ2 .105)
for i = 1.
(12. For the situation ML = 0 and MR ≫ MD (taking MD real) one 2 ﬁnds M1 ≈ MD /MR and M2 ≈ MR .
Exercise 12. T } γ5 ψ. V µ = ψγ µ T ψ and Aµ = ψγ µ γ5 T ψ ∂µ V µ ∂µ A
µ
= =
i ψ [M.101)
Thus one obtains from 2 ML + |MD |2 † MM = |MD | ML e+iφ + MR e−iφ the eigenvalues
2 M1/2
|MD | ML e−iφ + MR e+iφ 2 MR + |MD |2
. i ψ {M.
(12.1
Consider the case of the Weyl mode for symmetries.
. The diagonalization is analogous to what was done for the Λ-matrices and one ﬁnds U M U T = M0 with a (unitary) matrix U . This choice is possible without loss of generality because the phases can be absorbed into Υ′ and Υ′ (real must be replaced by hermitean if one includes 1 2 families).102)
=
1 2 2 ML + MR + 2|MD |2 2 ±
2 2 2 2 (ML − MR )2 + 4|MD |2 (ML + MR + 2ML MR cos(2φ)) . related via c Υ1R = U nL . This is a mixing problem with a symmetric (complex) mass matrix leading to two (real) mass eigenstates. T ] ψ.e.
Exercises
Exercise 12.2
Derive for the vector and axial vector currents. i. which implies U ∗ M † U † = U ∗ M ∗ U † = M0 and a ’normal’ diagonalization of the (hermitean) matrix M M † . and |a and |a′ belong to the same multiplet (there is a Qa such that |a′ = Qa |a ) then H|a = Ea |a implies that H|a′ = Ea |a′ .103)
for each of which one ﬁnds the lagrangians L =
and we are left with two decoupled Majorana ﬁelds Υ1L = U ∗ nc .

to be precise one has for bosons (b) or fermions (f). mb ≈ 5 GeV and MW ≈ 80 GeV.4
Calculate the lifetime τ = 1/Γ for the top quark (t) assuming that the dominant decay mode is t → b + W +.231. ǫ(λ) (p)ǫ(λ)∗ (p) = −gµν + µ ν M2
λ=1
¯ to calculate the width Γ(Z → f f ). 3 TW . In the standard model this coupling is described by the vertex −i g µ √ (γ − γ µ γ5 ) . etc.5
Show that the couplings to the Higgs (for W + W − h.
Exercise 12. The mass of the Z 0 is MZ = 91 GeV. Neglect the masses of fermions and use the fact that the sum over polarizations is 3 pµ pν .The standard model
132
with CV CA = =
3 TW − 2Q sin2 θW . Mf M2 and ghf f = . the weak mixing angle is given by ¯ sin2 θW = 0. ghbb = b v v where v is the vacuum expectation value of the Higgs ﬁeld.
. 2 2 The masses are mt ≈ 175 GeV.
(b) Calculate the width to electron-positron pair.
Exercise 12. Γ(Z 0 → e+ e− ). at least up to a high (which?) order in λ.6
Check that the use of the Wolfenstein parametrization in the CKM matrix indeed gives a unitary matrix. and the width to a pair of neutrino’s. Γ(Z 0 → νe νe ).7
In this exercise two limits are investigated for the two-Majorana case.
Exercise 12. e+ e− h. Note that you can ﬁnd the answer without explicit construction of all interaction terms in the lagrangian.
0
¯ Γ(Z 0 → f f ) =
MZ g2 48π cos2 θW
f f CV 2 + CA 2 .) are proportional to the mass squared for bosons or mass for fermions of the particles. hhh.
Write down the matrix element squared (averaged over initial spins and summed over ﬁnal spins) for the decay of the Z 0 . ZZh.
Exercise 12.

the mass eigenvalues and show that the mixing matrix is 1 1 1 √ U= i −i 2 which enables one to rewrite the Dirac ﬁeld in terms of Majorana spinors.
. Give the explicit expressions that relate ψ and ψ c with Υ1 and Υ2 . (b) A more interesting situation is 0 = ML < |MD | ≪ MR . Given that neutrino masses are of the order of 0.05 eV. Calculate the eigenvalues ML = 0 and MR = MX .The standard model
133
(a) Calculate for the special choice ML = MR = 0 and MD real. what is the mass MX if we take for MD the electroweak symmetry breaking scale v (about 250 GeV). which leads to the socalled see-saw mechanism.