# 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

4 3. 1994. 2007. Veltman.1 4.1 1.2 36 22
.2 3.2 2. I: Foundations.5 2. 3.3 References As most directly related books to these notes.2
2.5 4.V. S.2 3. M. 1980. The notes contain all essential information. An introduction to Quantum Field Theory.4 4. C. II: Modern Applications.1
2 2
3.3 3. Vol.3 4. M. Diagrammatica. Peskin and D. 2. 1995. Cambridge University Press.4
2.3 2.3 3.5 Ryder Peskin & Schroeder Srednicki 1 1 1 1 3
2.3
3. 6.3 3. Other text books of Quantum Field Theory that are useful are given in refs [3-6].2 2.2 6. McGraw-Hill.E.1 2. Cambridge University Press.2 3. Quantum Field Theory. I refer to the book of Srednicki [1] and Ryder [2].4 3. 3.1 6.5 2. Zuber.3 6.8 3. Vol. These notes Section 1.2 4. 5. Schroeder.1.3. The quantum theory of ﬁelds. 4. Quantum Field Theory. Cambridge University Press.7 2.3 2.3 2. Addison-Wesly.4 2.1 2.1 5. 1995. Peskin & Schroeder and Srednicki.3 3.2 3. Itzykson and J.
Corresponding chapters in books of Ryder. 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).4 3.4 6.3 2. 3.H. M. Srednicki. Cambridge University Press. 1985.1 3. but are rather compact.5 5. L. Cambridge University Press. Quantum Field Theory. 1996. Ryder.2 1. 2.2 6. Weinberg.-B.
1.1 3.2.2 3.3 2.

6 3.8 6.1 20.5
6 7 6
6.1.7 4.1 8.5 11.2 5.5 6. 20.3 6.7 5.4 12.6 4.10
8. 9.3 10.3 12.2 12.6 3. 6.2 7. 4.3 20.2 10. 6.5 9.5 4.1 20.2.4.3 7.1 20. 6.5 Ryder 4 4 4 4 4 4 Peskin & Schroeder 2.4 9.6 9.5 7. 2.3.2 8.6 10.4 12.5.3 8.4 7.5 4.5
.2.3 3.3 9.3 8.1 7.6 6.1 9.5 2.4 4.1.2. 6.3 7. 2.6 3.4 2.4 10.3 6 8 9 Srednicki 3 3 3 37
5.1 10.2 11.7.4.4 These notes Section 7.5 9.5 7. 8.1 5. 6.4 8.1 12.1 9.2 4.1. 4.1 11.4 3.3.6 8.2 8.3. 9.2.8 5 4. 4.4.3 11.3 2. 4.2 9.

5 Corrections 2011/2012) • Extension of Exercise 3.3 • First two sentences of Chapter 5 have been modiﬁed.2
. • Slight rewriting of discussion of Majorana mass term in section 6.

1)
In the limit that the action S is much larger than . In quantum mechanics a special role is played by Planck’s constant h.4)
where δij is the Kronecker δ function. the position can in principle be determined at any time with any accuracy. In this coordinate representation the position operators r op simply acts as rop ψ(r) = r ψ(r).3)
The uncertainty principle tells us that in this representation the momenta cannot be fully determined. It is given by pop ψ(r) = −i ∇ψ(r).
(1.054 571 68 (18) × 10−34 J s = 6. c = 299 792 458 m s−1 . 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.Chapter 1
Introduction
1. ˜ The corresponding states are |p and the wave functions ψ(p) = p||ψ are the Fourier transforms of the coordinate space wave functions. (1. ˜ ψ(p) = i d3 r exp − p · r ψ(r). the action of the momentum operator in the coordinate representation is not as simple as the position operator.2) In the limit that v ≪ c one reaches the non-relativistic domain. In special relativity a special role is played by the velocity of light c.6)
. ≡ h/2π = 1. 1 (1.5)
One can also choose a representation in which the momenta of the particles are the dynamical variables. Corresponding position and momentum operators do not commute. In quantum mechanics.1 Quantum ﬁeld theory
In quantum ﬁeld theory the theories of quantum mechanics and special relativity are united. S ≫ . They satisfy the well-known (canonical) operator commutation relations [ri . Indeed. usually given divided by 2π. being eigenvalues of the position operators. One can talk about states |r and the wave function ψ(r) = r||ψ . quantum eﬀects do not play a role anymore and one is in the classical domain. pj ] = i δij . (1. (1.582 119 15 (56) × 10−22 MeV s. (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.

. So we could have used all upper indices in the above equations. Vectors are denoted x = xµ nµ . the invariant length or distance (squared) is not positive deﬁnite. xi are the three space components. ˆ ˆ (1. given by g00 = −g11 = −g22 = −g33 = 1 (the ˆ ˆ other components are zero). in which case zi = ǫijk xj yk . however. • ds2 < 0 (spacelike intervals). • ds2 = 0 (lightlike or null intervals). In Minkowski space. (1.24) The quantity gµν ≡ nµ · nν is the metric tensor. where the coordinate t = x0 is referred to as the time component. xµ = gµν xν .21) (1. x2 . we will use upper indices for the three-vectors. x2 = x · x = xµ xν nµ · nν = xµ xν gµν . x1 . −x). δim δjm δkm δin δjn δkn . One has x2 = xµ xµ = t2 − x2 . The lower indices are constructed in the following way. 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 .
We note that for Euclidean spaces (with a positive deﬁnite metric) vectors and tensors there is only one type of indices. x1 . x3 ). . In special relativity we start with a four-dimensional real vector space E(1. are
. it is convenient to distinguish lower indices from upper indices. x) = (x0 .µn = Λµ1ν1 . One can distinguish: • ds2 > 0 (timelike intervals).νn can be conν structed. Many other four vectors and tensors transforming like T ′ µ1 .23)
δjm δkn − δjn δkm . When 3-dimensional space is considered as part of Minkowski space.1.3). x2 . Because of the diﬀerent signs occurring in gµν .. Λµn n T ν1 . The Poincar´ transformations include e e Lorentz transformations and translations. (1.2.. one must distinguish tensors with upper or lower indices and one can have mixed tensors.22) (1. independent of a coordinate system. The (invariant) lengths often have special names. the points lie on the lightcone and they can be connected by a light signal. 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.. such as eigentime τ for the position vector τ 2 ≡ x2 = t2 − x2 . These transformations do change the components of a vector.Introduction
5
that can be used in the cross product of two vectors z = x × y. The invariant distance between two points x and y in Minkowski space is determined from the length dsµ = (x − y)µ .25)
Within Minkowski space the real. Unlike in Euclidean space. Useful relations are ǫijk ǫℓmn ǫijk ǫimn ǫijk ǫijl = = = δiℓ δjℓ δkℓ 2 δkl . . and are given by (x0 . x3 ) = (t. denoted as V ′µ = Λµν V ν . No diﬀerence is made between upper or lower.3) with basis nµ (µ = ˆ 0. For four-vectors in Minkowski space we will use the notation with upper indices and write x = (t. The real. linear transformations that leave the length of a vector invariant are called (homogeneous) Lorentz transformations.26) (1. 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. linear transformations that do not change the length of a four-vector are called the Lorentz transformations. The length (squared) of a vector is obtained from the scalar ˆ product. 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 . Relations relating tensor expressions.

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. Relate this quantity to the electron Compton wavelength −e via the dimensionless λ ﬁne structure constant α. ∂t
(1. (1.31)
ǫµνρσ ǫµν
′ ′
ρσ
′
= −
(1. It is an invariant tensor.33) (1. (1.36)
Exercises
Exercise 1. p) = (E. (1. The derivative ∂µ is deﬁned ∂µ = ∂/∂xµ and we have a four vector ∂ with components (∂0 .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. not aﬀected by Lorentz transformations.34) (1. gµ = gµρ g ρν and 0 1 2 3 is in essence a ’Kronecker delta’.28)
ν ν Note that gµ with one upper and lower index.Introduction
6
called covariant. p). ∂3 ) = ∂ ∂ ∂ ∂ . starting from ǫ0123 = 1.29) 2 = ∂ µ ∂µ = ∂t2 The value of the antisymmetric tensor ǫµνρσ is determined in the same way as for ǫijk .
. .∇ . g0 = g1 = g2 = g3 = 1. For the contraction of a vector with the antisymmetric tensor one often uses the shorthand notation ǫABCD = ǫµνρσ Aµ Bν Cρ Dσ .27)
It is easy to convince oneself of the nature of the indices in the above equation. but more important has the same eﬀect on lefthandside and righthandside.35)
ρ ǫµνρσ ǫµν σ σ ǫµνρσ ǫµνρ
′
′
= −2 g ρρ g σσ − g ρσ g σρ = −6g σσ . deﬁned by ∂2 − ∇2 .32) . ∂2 . where m is called the mass of the system. . a∞ = 4πǫ0 2 /me e2 . because one has (1. for instance.1
(a) In the the Hydrogen atom (quantum system) the scale is set by the Bohr radius. explicitly written as (p0 .
′
′
.
0123 0123
′
′
ǫ
µνρσ
ǫµνρσ
The ﬁrst identity. ∂1 . Examples are the scalar products above but also relations like pµ = m dxµ /dτ for the momentum four vector.
=
∂ . constructed via the metric tensor itself. The length squared of ∂ is the d’Alembertian operator. (1. ∂t ∂x ∂y ∂z
ν ∂µ xν = gµ .
(1. Note that in this equation one has on left. is timelike with invariant length (squared) p2 = p · p = pµ pµ = E 2 − p2 = m2 . 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 . Each of these permutations leads to a minus sign. = −24.30)
(Note that there are diﬀerent conventions around and sometimes the opposite sign is used).

2
Prove the identity A × (B × C) = (A · C) B . n1 . ρ. Also for the coordinates [a− .3
Prove the following relation ǫµνρσ g αβ = ǫανρσ g µβ + ǫµαρσ g νβ + ǫµνασ g ρβ + ǫµνρα g σβ .4
Lightcone coordinates for a four vector a (which we will denote with square brackets as [a− .Introduction
7
(b) Relate the classical radius of the electron (a relativistic concept). a2 . Compare it with the proton mass and use Eq. (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). e.
Exercise 1. a1 . a2 ] we can ﬁnd basis vectors n− . ν. which is the Compton wavelength for the Planck mass. . a1 .13 to give its actual value in kg.(A · B) C using the properties of the tensor ǫijk given in section 1. 0. 0. a1 . 0. ].3. but only appropriate combinations. a2 ] or [a− . c. re = e2 /4πǫ0 me c2 to the Compton wavelength. 0) ˆ n3 = (0. 0. 1) ˆ .71 × 10−39 GeV−2 to construct a mass Mpl (Planck mass). . 0. ˆ ˆ
. (a) Express the scalar product a · b in lightcone coordinates and deduce from this the values of g++ . (d) Use the value of the gravitational constant GN / c5 = 6. ˆ ˆ ˆ ˆ These are n1 = (0. 0) ˆ n2 = (0. α and me c2 = 0. 0) ˆ n0 = (1. 2. Note that ˆ ˆ ˆ ˆ a · nα = aα and derive from that the components nα of the four vector nα . ǫ0 . 1. n2 . This demonstrates how a careful use of units can save a lot of work. One does not need to know . Also construct and calculate the Planck length Lpl . a+ . n+ . a3 ) are the expansion coeﬃcients using the basis vectors n0 . me . n2 . g+− and g−+ . (c) Calculate the Compton wavelength of the electron and the quantities under (a) and (b) using the value of c. 1. g−− . . a+ .
Exercise 1. (b) The coordinates (a0 . 3} the index α can only be equal to one of the indices in ǫµνρσ . n3 . How are a+ and a− related to a+ and a− . [Note: what is the MKS unit for V/T?]
Exercise 1. 1. by a simple few-line reasoning [For instance: If {µ. n1 . σ} is a permutation of {0. aT ]) are deﬁned through √ a± ≡ (a0 ± a3 )/ 2.511 MeV. 0. Use that to give the ˆ ˆµ ˆ components of the basis vectors n+ and n− . 1. a+ .

∂t p −→ pop = −i∇.1 The Klein-Gordon equation
In this chapter. But the replacement 2. we just want to play a bit with covariant equations and study their behavior under Lorentz transformations. Thus a covariant equation can be obtained by starting with the (covariant) equation for the invariant length of the four vector (E. Substitution of operators gives the Klein-Gordon (KG) equation for a real or complex function φ.3 are not covariant.5)
with (k 0 )2 = k2 + M 2 . ∂t 2M (2. φk (r. p2 = pµ pµ = E 2 − p2 = M 2 . t) = 0. ∂t2 (2. (2. i ∂ ∇2 ψ(r. t) = ∂2 − ∇2 + M 2 φ(r.
Although it is straightforward to ﬁnd the solutions of this equation. 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.2)
Acting on the wave function one ﬁnds for a free particle. 2M (2. t) = exp(−i k 0 t + i k · r). The Schr¨dinger equation in quantum mechanics is the operator equation o corresponding to the non-relativistic expression for the energy.4)
where M is the particle mass. namely plane waves characterized by a wave number k. one obtains (2.1 and 2.2. p). E = p2 .7) k 0 = ± k2 + M 2 = ±Ek . (2.Chapter 2
Relativistic wave equations
2. t) = − ψ(r. (2. written as pµ −→ i∂µ is covariant (the same in every frame of reference). 2 + M 2 φ(r. t). 8
.1)
under the substitution (in coordinate representation) E −→ Eop = i ∂ .3)
Equations 2.6)
The energy spectrum is not bounded from below: considering the above stationary plane wave solutions.

−k) ≡ b∗ (k) one has φ(x) = d3 k e−i k·x a(k) + ei k·x b∗ (k) = φ+ (x) + φ− (x). 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. Then.2
Mode expansion of solutions of the KG equation
Before quantizing ﬁelds. This continuity equation can be written down o covariantly using the components (ρ. It is possible to rewrite it as an integration over positive energies only but this gives two terms (use the result of exercise 2. it is simply a matter of being careful to ﬁnd a consistent (covariant) theory. As we will see later both problems are related and have to do with the existence of particles and antiparticles. k) + ei k·x φ(−Ek . −i φ∗ ∇ φ . ∂µ j µ = 0. −k) .3).10) ∂t which follows directly from the Schr¨dinger equation. 2M 2M (2. rather than as a wave function. At that point the dependence on position r and time t is just a dependence on numbers/parameters on which the operator depends.
↔
↔
(2. j) of the four-current j. (2.9)
They satisfy the continuity equation. which is what Lorentz transformations do. k) ≡ a(k) and φ(−Ek .
∂ρ = −∇ · j. And. (2.11)
Therefore. there are no longer fundamental objections to mix up space and time. φ(x) = d4 k ˜ 2π δ(k 2 − M 2 ) e−i k·x φ(k) (2π)4 (2. The integration over k-modes clearly is covariant and restricted to the ‘mass’-shell (as required by Eq.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. φ(x) = d3 k (2π)3 2Ek ˜ ˜ e−i k·x φ(Ek . we want the most general solution.Relativistic wave equations
9
i. This leads to the existence of negative densities. For this we note that an arbitrary solution for the ﬁeld φ can always be written as a superposition of plane waves.5).
2. This operator has all possible solutions in it multiplied with creation (and annihilation) operators.2) is j µ = i φ∗ ∂ µ φ or (ρ. The KG equation. there are solutions with negative energy.13)
˜ with (in principle complex) coeﬃcients φ(k). is a second order equation and φ and ∂φ/∂t can be ﬁxed arbitrarily at a given time. having the KG equation as a space-time symmetric (classical) equation. however. for which we need the interpretation of φ itself as an operator.14)
˜ ˜ Introducing φ(Ek . but rather the zero component of a four vector. (2π)3 2Ek (2.15)
. just as the dependence on time was in ordinary quantum mechanics. The appropriate current corresponding to the KG equation (see Excercise 2. relativistically the density is not a scalar quantity.e. 2. (2.

This shows how parity transforms k-modes into −k modes. Since a · b = a · ˜ it is easy to see that ˜ b. (2π)3 2Ek
(2.Relativistic wave equations
10
In Eqs 2. one thus must have that the function φ → φ′ such that φ′ (x′ ) = φ(x) or φ′ (x) = φ(Λ−1 x). The coeﬃcients a(k) and b∗ (k) are the amplitudes of the two independent solutions (two.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) .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. which implies (xµ ) = (t.6 We will explicitly discuss the example of a discrete symmetry.18) (2. 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) . for which we consider space inversion.e. i. (2. x Transforming everywhere in the KG equation x → x one obtains ˜ ˜ ˜ x ∂µ ∂ µ + M 2 φ(˜) = 0.14 and 2. changing the sign of the spatial coordinates.23)
.
2. (2π)3 2Ek (2. (2.20)
one has for the mode amplitudes aP (k) = a(−k) and bP (k) = b(−k). 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) . x ∂µ ∂ µ + M 2 φ(˜) = 0. 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).15 one has elimated k 0 and in both equations k · x = Ek t − k · x. after restricting the energies to be positive). They are referred to as mode and anti-mode amplitudes (or because of their origin positive and negative energy modes). x) → (t. It is trivial to see that ∂µ ∂ µ + M 2 φ∗ (x) = 0. Another symmetry is found by complex conjugating the KG equation. (2π)3 2Ek (2. Performing some Lorentz transformation x → x′ = Λx.16) The consequence of this is discussed in Exercise 2.19) implying that for each solution φ(x) there exists a corresponding solution with the same energy. (2.22)
showing that with each solution there is a corresponding charge conjugated solution φC (x) = φ∗ (x). φP (x) ≡ φ(˜) (P for parity).17)
(2. −x) ≡ (˜µ ).

15. t) exp(−i k · x) and its time ˜ derivative i∂0 φ(k. Q = 0. (2π)3 2Ek
where Ek =
Exercise 2.2
Show that if φ and φ∗ are solutions of the KG equation. letting V → ∞. t) ≡ d3 x φ(x. k) = (2π)4 k2 + M 2 .
↔ ↔
Exercise 2. |f ′ (xn )|
. for the mode amplitudes aC (k) = b(k) and bC (k) = a(k).
˙ and thus for any normalized solution the full ‘charge’.Relativistic wave equations
11
i. is conserved. t) = ei Ek t (i∂0 + Ek ) φ(k. t) for the mode expansion in Eq. Note that a(k) and b(k) are independent of t.e. t). For the real ﬁeld one has aC (k) = a(k).
Exercises
Exercise 2. ˜ b(k) = ei Ek t i ∂0 φ∗ (−k.
S V
d3 x j 0 (x). QV ≡ satisﬁes ˙ QV = − ds · j. k). t).1
Show that for a conserved current (∂µ j µ = 0) the charge in a ﬁnite volume. 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.
Exercise 2. d3 k F (Ek . 2.3
Show that1 d4 k 2π δ(k 2 − M 2 ) θ(k 0 ) F (k 0 . This shows how charge conjugation transforms ’particle’ modes into ’antiparticle’ modes and vice versa.
1 Use
↔
↔
the following property of delta functions δ (f (x)) =
zeros xn
1 δ(x − xn ). Use these to show that ˜ ˜ a(k) = ei Ek t i ∂0 φ(k.

i.Relativistic wave equations
12
Exercise 2. B] = AB − BA and the anticommutator as {A. express the quantities E(t) = P i (t) = d3 x (∂0 φ)∗ (∂0 φ) + ∇φ∗ · ∇φ + M 2 φ∗ φ . 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 . and β 2 = I. 2.e. ∂t where ψ is a wave function and the nature of α and β is left open for the moment.
.7
To solve the problem with positive and negative energies and get a positive deﬁnite density.
Exercise 2. j 0 = ψ†ψ and j = ψ † αψ.
From this one concludes that α and β must be matrix-valued and ψ must be a multi-component wave function. One then ﬁnds that Q is expressed as an inﬁnite sum over number operators a† a (particles) and b† b (anti-particles). 1 2 (b) Express now the full charge QV (t) (exercise 2. t). We will encounter these quantities later as energy and momentum.
in terms of the a(k) and b(k).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. also using the result in (a). 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). (c) Similarly. an attempt to construct a ﬁrst order diﬀerential equation for the time evolution would be to write. β} = 0. αj } = 2 δ ij I.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). d3 x (∂ {0 φ)∗ (∂ i} φ). t) = (−i α · ∇ + m β) ψ(r.
Exercise 2. i (a) Show that relativistic invariance. B} = AB + BA. ∂ ψ(r.1) for a complex scalar ﬁeld current (exercise 2. Note that a{µ bν} indicates symmetrization. a{µ bν} ≡ aµ bν + aν bµ . 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 . 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).15.2) in terms of the a(k) and b(k) using the expansion in Eq. {αi .

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

(3. These generators form a threedimensional Lie algebra SO(3). is a 3-dimensional Lie-group. y ∈ A ⇒ [x.
(3. These matrices can be deﬁned as acting on 2-component spinors (χ → Aχ) or equivalently as acting on 2 × 2 matrices (B → ABA† ). Ly and Lz .2 π)
(-n.and y-axes 0 0 0 0 0 0 0 −i . for rotations around z for instance. One way of viewing the parameter space. ˆ R(ϕ. 0
(3. y]] = 0 (Jacobi identity). Lx = Ly = 0 0 0 i 0 −i 0 ϕ ˆ . which reﬂects itself in the noncommutation of the generators. Locally this is a 3-dimensional Euclidean space and SU (2). 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 (3.6)
Summarizing. It is straightforward to check that any (ﬁnite) rotation can be obtained from a combination of inﬁnitesimal rotations. x] = 0 (thus [x. With matrix commutation this algebra satisﬁes the requirements for a Lie algebra.z N
N
or (Lk )ij = −i ǫijk . Next. [x.5)
Rotations in general do not commute.Groups and their representations
14
(n.
. π)
(n. ] that satisﬁes • ∀ x. the rotations in SO(3) can be generated from inﬁnitesimal rotations that can be expressed in terms of a basis of three generators Lx .8)
ˆ that (σ · a)(σ · b) = a · b + i σ · (a × b). z ) = lim
N →∞
that are generated by i 0 . z]] + [y. n) = 1 cos ˆ
1 Note
ϕ ϕ ˆ + i (σ · n) sin 2 2
= exp i
ϕ ˆ σ·n . y] = −[y. [z. • [x. ˆ therefore. In the same way we can consider rotations around the x.1: the parameter spaces of SO(3) (left) and SU (2) (right).
with real a’s and 3 (ai )2 = 1.4)
R
. we turn to the group SU (2) of special (det A = 1) unitary (A† = A−1 ) 2 × 2 matrices. x]). namely that there exists a bilinear product [. 2
(3. Writing a0 = cos(ϕ/2) and a = n sin(ϕ/2) we have1 A = A(ϕ. They satisfy the commutation relations [Li . Lj ] = i ǫijk Lk . thus is as the surface i=0 of a sphere in 4 Euclidean dimensions. and thus (σ · n)2 = 1. x]] + [z. [y. y] ∈ A. π )
π -sphere (antipodes identified)
2 π -sphere (surface identified)
Figure 3.7) = a0 0 −1 i 0 1 0 0 1 • [x.

[g. 2!
.14)
For a Lie group. In the full parameter space. There exists a corresponding mapping of the groups given by µ: SU (2) −→ SO(3) ˆ ˆ A(ϕ. . the SU (2) → SO(3) mapping is a 2 : 1 mapping where both A = ±1 are mapped into R = I. Suppose we have a system described by a Hamiltonian H. Near the identity. however. n! 2! n=0
∞
(3.15) [g. i.9)
The parameter-space of SU (2).e. also can be considered as a ﬁlled 3-sphere. σy /2 and σz /2 form the basis of the Lie-algebra SU (2). They satisfy σi σj σk = i ǫijk . . n) −→ R(ϕ. For instance in a function space translations.e. i. H] = 0 for g ∈ G.Groups and their representations
15
where we have used the (for operators new!) deﬁnition eA ≡ An 1 = 1 + A + A2 + . The next issue is to ﬁnd the appropriate symmetry operators in the Hilbert space. (3. The inﬁnitesimal generators of SU (2) are ˆ obtained by considering inﬁnitesimal transformations. .e.10)
σ ˆ · n. . n)
π ≤ ϕ ≤ 2π.12)
One. SU (2) ≃ SO(3). in particular that AB → RAB = RA RB ). it is suﬃcient that the generators commute with H. immediately sees that the Lie algebras are identical. one has a Lie algebra isomorphism that is linear and preserves the bilinear product. n) i ∂ϕ =
ϕ=0
(3. ˆ ˆ A(ϕ. . thus. the above mapping corresponds to the trivial mapping of the Lie algebras. . n)
ˆ −→ R(2π − ϕ. n) ≈ 1 + i ϕ J · n. 2 2 2 (3.2
Representations of symmetry groups
The presence of symmetries simpliﬁes the description of a physical system and is at the heart of physics. with ˆ J ·n ≡ ˆ 1 ∂A(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. H] = 0 for g ∈ G. sometimes in more than one way (but this will be discussed later). for ﬁxed n.11)
Thus σx /2. The existence of symmetries means that there are operators g belonging to a symmetry group G that commute with the Hamiltonian. now with radius 2π and with all points at the surface identiﬁed (see ﬁgure). i. since any ﬁnite transformation can be constructed from the inﬁnitesimal ones. (3. thus. (3. 2
(3. .
3. φ′ (r) = φ(r + a) = φ(r) + a · ∇ φ + 1 (a · ∇)2 φ + .

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

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

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. Denoted in terms of column vectors and 4-dimensional matrices one writes for the poincar´ transe formations x′ = Λ x + a. e To derive some of the properties of the Lorentz group. G = 10 (3. proof: det(ΛT GΛ) = det(G) → (det Λ)2 = 1. e. it is easy to derive some properties of the matrices Λ: (i) det(Λ) = ±1. Thus one has the (pseudo-orthogonality) relation ΛT GΛ = G ⇔ GΛT G = Λ−1 ⇔ ΛGΛT = G.g. From this property. 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 (3.27) (3. it is convenient to use a vector notation for the points in Minkowski space.26) (3. explicitly x′µ = (Λ)µν xν + aµ or x′µ = Λµν xν + aµ . Writing x as a column vector and the metric tensor in matrix form.23)
The proper tensor structure of the matrix element (Λ)µν is a tensor with one upper and one lower index. We considered the generators and looked for representations in ﬁnite dimensional spaces. consisting of the Lorentz group and translations.25) (3. In this section we consider the Poincar´ group.Groups and their representations
18
3.24)
L↑ + L↓ + L↑ − L↓ −
proper orthochrone proper non-orthochrone improper orthochrone improper non-orthochrone
. det Λ = −1 is called improper). σ/2 in a two-dimensional (spin 1/2) case.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. (ii) |Λ00 | ≥ 1.22) (Note that xT is a row vector). (det Λ = +1 is called proper.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. (3. 0 0 g00 g01 g02 g03 1 0 g g11 g12 g13 0 −1 0 0 = . 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 .

J 2 . z ) = exp(i ϕ J 3 ). Using these explicit transformations.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↑ . The ﬁrst two sets of commutation relations exhibit the rotational behavior of J and K as vectors in E(3) under rotations.28) 3′ 3 V 0 0 0 1 V 0 0 0 0
The parameter η runs from −∞ < η < ∞. K 3 ).
3 i 2 i=1 (Λ0 ) 3
19
(iii)
Boosts along the z-direction are given by ΛB (η. [K i . 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 . 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. 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 . The commutator of two boosts in diﬀerent directions (e. J = (J 1 .32) 0 − 0 −1 0 0 0 0 −1
. J j ] = i ǫijk J k . i proof: use ΛT GΛ = G and ΛGΛT = G. Rotations around the z-axis are given by ΛR (ϕ. inﬁnitesimally given by ˆ ΛR (ϕ. [J i . Returning to the global group.Groups and their representations = i=1 (Λ0 )2 . K j ] = −i ǫijk J k . From the commutation relations one sees that the boosts (pure Lorentz transformations) do not form a group.30) 0 + 0 +1 0 0 0 0 +1 0 0 −1 0 0 +1 0 0 ∈ L↓ (time inversion) It = (3. since I ∈ L↑ and + + det Λ and Λ0 change continuously along a path from the identity. 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). K = (K 1 . (3. K j ] = i ǫijk K k . z ) ≈ ˆ ˆ I − i η K 3 . z ) ≈ I + i ϕ J 3 . 0 0 +1 0 0 +1 0 0 ∈ L↑ (identity) I = (3. This is the origin of the Thomas precession. since the generators K do not form a closed algebra. one distinguishes rotations 0 + and boosts. which satisfy the commutation relations (check!) [J i .g. βγ = sinh η. In L↑ . K 2 . and those of the boosts. we have found the generators of rotations. (3.31) 0 − 0 +1 0 0 0 0 +1 0 0 +1 0 0 −1 0 0 ∈ L↑ (space inversion) Is = (3. it is easy to ﬁnd the following typical examples from each of the four classes. inﬁnitesimally given by ΛB (η. J 3 ). z ) = exp(−i ηK 3 ).

This is a + normal subgroup and the factor group is the Vierer group. Summarizing. we have for the (inﬁnitesimal) rotations (around z-axis) ΛR = I + i ϕ3 J 3 = I − i ω12 M 12 .g.39)
(e.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. Thus the two vector operators J and K form under Lorentz transformations an antisymmetric tensor M µν . M ρσ ] = −i (g µρ M νσ − g νρ M µσ ) − i (g µσ M ρν − g νσ M ρµ ) . we continue using covariance arguments. We now want to ﬁnd a covariant form of the six generators of the Lorentz transformations.33)
These four transformations form the Vierer group (group of Klein). J 3 = M 12 ) and for the (inﬁnitesimal) boosts (along (3.4
The generators of the Poincar´ group e
↑ The transformations belonging to P+ are denoted as (Λ. We therefore ﬁnd (again) that there are six generators for the Lorentz group.35)
thus (ω)ij = −(ω)ji and (ω)0i = (ω)i0 . P+ etc. 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 .36) (3. three of which only involve spatial coordinates (rotations) and three others involving time components (boosts).
3 30
with η = ω and K = M (e. (3. inﬁnitesimally approximated by (I + ω. ǫ). (3. a). Also this group can be divided into four parts. In order to ﬁnd the commutation relations including the translation generators.
3. explicitly reading
x′µ
= =
(Λ)µν xν + aµ = ((I)µν + (ω)µν ) xν + ǫµ Λµν xν + aµ = (g µν + ω µν ) xν + ǫµ . 2 (M αβ )µν = i g αµ g βν − g α g βµ ν (3. One can easily convince oneself that i (ω)µν = ω µν = − ωαβ (M αβ )µν . There are six generators. The extension to the Poincar´ group is e ↑ straightforward.g. We will just require Poincar´ invariance for the generators themselves. [M µν . Multiplying the proper orthochrone transformations with one of them gives all Lorentz transformations.38) Explicitly.37)
The algebra of the generators of the Lorentz transformations can be obtained by an explicit calculation2 . the Lorentz transformations form a (Lie) group (Λ1 Λ2 and Λ−1 are again Lorentz transformations. Of the four parts only L↑ forms a group. The e
2 These
commutation relations are the same as those for the ’quantummechanics’ operators i(xµ ∂ ν − xν ∂ µ )
.
inf
(3. K = M ).34)
The condition ΛT GΛ = G yields g ρµ + ω ρµ gρσ (g σν + ω σν ) = gµν =⇒ ωνµ + ωµν = 0. with ϕi = − 1 ǫijk ω jk and J i = 2 z-axis).
i i0 i i0 1 ijk M jk 2ǫ
(3.

= −i (g µρ P ν − g νρ P µ ) . (3.5
Representations of the Poincar´ group e
where pµ is a set of four arbitrary real numbers. P ν ] [M µν .48)
. the tensor M µν in terms of boosts cK i = M i0 and rotations J i = 1 ǫijk M jk . leads to U (Λ. the generators J± in the case of the rotation group.42)
(to decide on where the inverse is. The generators of the translations are the (momentum) operators P µ . .43)
giving the following commutation relations by equating the coeﬃcients of ǫµ and ωµν . . For instance. other states in that representation are obtained by the action of operators outside the maximal commuting set.
In order to label the states in an irreducible representation we construct a maximal set of commuting operators.45)
Note that the commutation relations among the generators M µν in Eq. 3. a)P µ U † (Λ. Λµν pν ⇒ ωµν pν ⇒ ωµν = = = pµ + ω µν pν = pµ 0 ǫµνρσ pρ sσ . . . [K i . For instance. To ﬁnd other generators that commute with P µ we look for Lorentz transformations that leave the four vector pµ invariant. a) = Λ−1 µ P ν . (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 j ] = i ǫijk P k .41) 2 At this point.47)
(3.44) (3. Taking any one of the states in an irreducible representation. a)P µ U (Λ. . the generators J 2 and J 3 in the case of the rotation group. (3. one obtains 2 [P i . 2 2 (3. Explicitly. .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.
We have here reinstated c. ν (3. These form a group called the little group associated with that four vector. P ) in terms of the Hamiltonian and the three-momentum operators. that commute e among themselves. [J i . because one then sees that by letting c → ∞ the commutation relations of the Galilei group. writing the generator P = (H/c. [P µ . Of the generators of the Poincar´ group we choose ﬁrst of all the generators P µ . P j ] = [P i . a) = Λµν P ν or U † (Λ. P µ |p.
3. (3. J j ] = i ǫijk J k . as part of the set. . = pµ |p.38 could have been obtained in the same way. K j ] = i ǫijk K k . H] = [J i .46) [K i . (3. The eigenvalues of these will be the four-momentum pµ of the state. P j ] = i δ ij H/c2 . H] = i P i . ǫ) = 1 − ωαβ M αβ + i ǫα P α . Inﬁnitesimally. They just state that M µν transforms as a tensor with two Lorentz indices. check that U1 U2 corresponds with Λ1 Λ2 ). K j ] = −i ǫijk J k /c2 . [J i . These deﬁne states with speciﬁed quantum numbers that are eigenvalues of these generators. the generators M αβ no longer exclusively act in Minkowski space. [K i . known from non-relativistic quantum mechanics are obtained. P ρ ] = 0. The requirement that P µ transforms as a four-vector (for which we know the explicit behavior from the deﬁning four-dimensional representation). [J i . H] = 0.

. of which the components generate the little group of pµ or (for the third of the following relations) from explicit calculation [Mµν . The elements of the little group thus are U (Λ(p)) = = and U (Λ(p))|p. = = i exp − ǫµνρσ M µν pρ sσ |p. . the third case represents the vacuum. . while the others have no physical signiﬁcance.56)
. From the algebra of the generators one ﬁnds that P 2 = Pµ P µ and W 2 = Wµ W µ (3. which can thus be chosen spacelike. p0 > 0.
Massive particles: p2 = M 2 > 0.
Given the momentum four vector pµ we choose a tetrade consisting of three orthogonal spacelike unit vectors ni (p).51)
The following properties follow from the fact that W µ is a four-vector by construction. satisfying gµν nµ (p)nν (p) = j i nµ (p) pµ i ǫµνρσ pµ nν nρ nσ i j k = = −δij . i ǫµνρσ W ρ P σ . 2 exp (−isµ W µ ) |p.Groups and their representations
22
where s is an arbitrary vector of which the length and the component along pσ are irrelevant.50)
where the Pauli-Lubanski operators W µ are given by 1 Wµ = − ǫµνρσ M νρ P σ . (3.49)
(3.
(3. . . . 2 (3. Pν ] = [Wµ . . 0.
0. 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.55) commute with all generators and therefore are invariants under Poincar´ transformations.54)
This will enable us to pick a suitable commuting ’spin’ operator. These e operators are the Casimir operators of the algebra and can be used to deﬁne the representations of ↑ P+ . Wν ] = −i (gµα Wν − gνα Wµ ) . i exp − ωµν M µν 2 i exp − ǫµνρσ M µν pρ sσ 2
(3. Wα ] = [Wµ . . M ǫijk . .52) (3.53) (3.

s. show that W /M can be interpreted as the intrinsic spin.60)
and using the completeness relation nµ (p)nν (p) = − g µν − i i
2
i
pµ pν M2
. .61)
one sees that S i (p)
i
=−
W2 . Having made a covariant decomposition. 1. M2 (3. S j (p)] = −i nµ (p)nν (p) j i ǫµνρσ W ρ P σ = i ǫijk S k (p). |p3 |). (0. 0. 1. 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.
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 . 0). ms . 0.62)
1 Thus W 2 has the eigenvalues −M 2 s(s + 1) with s = 0.65)
. −1)
n1 (p0 ) ≡
n2 (p0 ) ≡ s(p0 ) ≡
(3. p. The explicit construction of the wave function can be done using the D-functions (analogous as the rotation functions). 1 2 (3. Together with the four momentum states thus can be labeled as |M.58)
i. 0.63)
where E = p2 + M 2 and ms is the z-component of the spin −s ≤ ms ≤ +s (in steps of one). Noting that S i (p)
i 2
=
1 W µ (p)W ν (p) M2
niµ (p)niν (p). M J) with components W i = (M/2)ǫijk M jk . W i (p) = −W · ni (p)).
(3. The commutation relations [W i (p). (0. (1.64)
such that in an arbitrary frame where the four vectors p. 0). it is suﬃcient to choose a particular frame to investigate the coeﬃcients in Eq. p0 > 0. n1 (p). In that case the vectors ni (p) are just the space directions and W = (0. 2 .e.
i
(3. the S i (p) form the generators of an SU (2) subgroup that belongs to the maximal set of commuting operators. 0. 0.
More generally (fully covariantly) one can proceed by deﬁning S i (p) ≡ W i (p)/M and obtain [S i (p).59)
(3. 3.e. . 0. (3.Groups and their representations
23
We can write W µ (p) =
3
W i (p) nµ (p) i
i=1
(3.57)
(i. W j (p)] = i M ǫijk W k (p). 0). . 1..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. 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.

We don’t know of any physical relevance. Wν ].66)
where W · p = 0 implies that W s (p) = −(M 2 /p · s) λ(p). λ . P µ . σ · RA a ≡ A(σ · a)A−1 ˆ (valid for any vector a). is characterized by its momentum and the helicity. W 1 (p)] [λ(p). Note that angular momentum does not contribute to helicity as L · p = 0.13. |p| (3. This algebra is (for instance) isomorphic to the one generated by rotations and translations in a 2-dimensional Euclidean plane (see exercises). The vacuum: pµ = 0. however. where A = exp(i ϕ · σ/2) = exp(iϕ σ · n/2). Thus 2 λ(p) = J ·p . (3.73) (3. W 2 (p)] [λ(p). The algebra of W 1 . 1 2
24
(3. 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. −i W 1 (p). (3. The eigenvalues of W 1 and W 2 (in that case corresponding to the translations) thus can take any continuous values. |p. W µ (p) = λ(p) P µ . W 2 and λ is derived from the general commutation relations for [Wµ .71)
which is called the helicity. 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). 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
. But if M = 0 one has a diﬀerent algebra (Eq.72)
One has P µ |0 = W µ |0 = 0 and the state is invariant under Lorentz transformations.67 has a vanishing right hand side). Thus we have P 2 .1
(a) Derive from the relation that deﬁnes the mapping of SU (2) matrices A into SO(3) matrices RA in Eq.Groups and their representations The above does actually include the massive case. thus. a)|0 = |0 . implying continuous values for the spin (actually for W i = M S i ). 3. 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). i W 2 (p). W 2 (3. This state is in physical applications nondegenerate and is denoted by |0 . 3. which can take any integer or half-integer value.
Exercises
Exercise 3. W 2 (p)] = = = i M 2 λ(p). and gives [W 1 (p).67) (3.70) as a commuting set with zero eigenvalues for P 2 and W 2 . U (Λ.69)
We reproduce the algebra for the massive case with W 3 (p) = M λ(p). A massless particle.68) (3.

e. Note the diﬀerence with the (usual) spin 0 and spin 1 states composed from two spin 1/2 states. Show actually that ǫ = 2 eiπJ (up to a phase). In both of these cases the symmetry is broken by the charges ¯¯ of the particles. O] + . φ0 ) doublet representation for SU (2)weak . . for instance.Groups and their representations
25
for any vectors a and b. The appropriate isospin doublet to be used for antiquarks then is (−d. i. Tr(σi σj σk ) = 2i ǫijk . d) doublet representation for SU (2)isospin or the Higgs (φ+ .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. (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. show that the transformation for the operators is given by O′ = U O U −1 = O + i [J.
Excercise 3. (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. Both particle and antiparticle have ’spin 1/2’. show that one matrix ǫ exist such that J c = ǫ J ǫ† . not true for SU (3) (ﬂavor or color properties of particles know two distinct deﬁning 3-dimensional representation. Tr(σi σj σk σl ) = 2 (δij δkl − δik δjl + δil δjk ) .
Exercise 3. Is this easy to understand.2
Show that for a unitary operator U = exp[iαk Jk ] with real coeﬃcients αk . the pions and the isospin 0 meson state. respectively).
Exercise 3. . u). one has to explicitly include this rotation. triplet and anti-triplet representation. . the operators Jk must be hermitean. Note: Recall (or derive) that the Pauli matrices σ satisfy σi σj = δij + i ǫijk σk and thus Tr(σi σj ) = 2 δij . Writing A∗ = exp(i α · J c ). 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)
. For a spin 1/2 antiparticle the representation A∗ is needed.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. (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). Show that the Lie-algebra representations J and J c are equivalent. the (non-strange) η-meson state. (b) This establishes that for spin 1/2 one doesn’t need to talk about ’anti-spin’ for the antiparticle. the trace of the operator Jk is zero (Convince yourself that det(eA ) = eTr A by diagonalizing A. Show that if det(U ) = 1. This is true for SU (2) but it is. Examples are the quark (u. Now write down the three isospin 1 meson quark-antiquark (meson) states.

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

2 2 σ σ J = . They precisely correspond to the two types of representations that we have seen before.4) (4. 2 2 (4. U (ϕ) H(η) (4. becoming the deﬁning representations of SL(2. K = −i . similarly as the homeomorphism between SO(3) and SU (2). B j ] = [Ai . It is simply connected and forms the covering group of L↑ . 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. (j.Chapter 4
The Dirac equation
4. M = exp i ϕ·σ 2 1 exp ± η · σ 2 = U (ϕ) H(η) . 2 1 (J − i K). it also follows directly that the Lorentz group is homeomorphic with the group SL(2. C): Type I (denoted M ): Type II (denoted M ): 27 J= σ σ . The matrices in SL(2.1 The Lorentz group and SL(2.3) (4.6) (4. The group SL(2. i ǫijk B k .1) (4.10)
. where we restrict (for ﬁxed n) the parameters 0 ≤ ϕ ≤ 2π and 0 ≤ η < ∞. Special cases are the following representations: Type I : (j. This tells us how to ﬁnd the representations of the group. K = +i . j ′ ). K = i J (A = 0).8)
ˆ ˆ ˆ with η = η n and ϕ = ϕn. C) can be written as a product of a unitary + matrix U and a hermitean matrix H.9) (4. 2 (4. Aj ] = [B i .2)
This shows that the Lie algebra of the Lorentz group is identical to that of SU (2) ⊗ SU (2). They will be labeled by two angular momenta corresponding to the A and B groups. (4. B j ] = i ǫijk Ak .5) = 1 (J + i K). j) K = −i J (B = 0). respectively.7)
From the considerations above. C). 0) Type II : (0. C) is the group of complex 2 × 2 matrices with determinant one. With this choice of parameter-spaces the plus and minus signs are actually relevant. (4.

H ≡ (H † )−1 = H −1 ). C) or L↑ . η → Hη.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. ¯ ¯ H ∗ = ǫ† Hǫ = ǫ−1 Hǫ.e. (4.18)
As already discussed for SU (2). while ±ǫη ∗ transforms as a type-I + spinor.19)
. (4.
(4. Λ(ǫ) ∈ L+ . One deﬁnes spinors ξ and η transforming similarly under unitary rotations (U † = U −1 . ǫ−1 = ǫ† = ǫT = −ǫ).12) ξ → U ξ.13)
2M (E + M )
(exercise 4. σ µ ≡ (1. Also useful is the relation H 2 (p) = σ µ pµ /M = (E + σ · p)/M . The Lorentz transformations Λ(M ) and Λ(M ).18 show that ±ǫξ ∗ transforms as a type-II (η) spinor. The following relations for + rotations and boosts are useful.16)
= σ µ Λ R aµ . ˜ The sets of four operators deﬁned by deﬁnitions σ µ ≡ (1. there exists the matrix ǫ = iσ 2 . H(η) = exp(−η · σ/2). namely H(η) = exp(η · σ/2). Eqs 4. −σ). i. U σ µ aµ U −1 = σ µ ΛR aµ . This is not true for the two-component spinors in SL(2. σ). thus. C) and. 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. such that M = SM S −1 . ξ −→ Hξ. In fact. With E = M γ = M cosh(η) and p = M βγ n = M sinh(η) n. C). Note that M ∗ = ǫ† M ǫ has its equivalent for the Lorentz transformations Λ(M ) and Λ(M ∗ ). U ≡ (U † )−1 = U ) U (ϕ) = exp(i ϕ · σ/2) but diﬀerently under hermitean boosts (H † = H. cannot be connected by a Lorentz transformation belonging to L↑ . however. ˜
= σ µ Λ B aµ .15) (4. one can use a boost to the frame with momentum ˆ ˆ p. 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). (4. This shows that M ∗ ≃ M . (4. are not covariant!) ˜ can be used to establish the homeomorphism between L↑ and SL(2. ˜ (4. 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. thus. U σ µ aµ U ˜
−1
H σ µ aµ H −1 = σ µ ΛB aµ H σ µ aµ H ˜
−1
(4.The Dirac equation
28
Let us investigate the deﬁning (two-dimensional) representations of SL(2. ˜
Non-equivalence of M and M
One might wonder if type I and type II representations are not equivalent. One has +
−1 Λ(M ∗ ) = I2 Λ(M ) I2 . (ǫ∗ = ǫ. if there does not exist a unitary matrix S.11)
Considering ξ and η as spin states in the rest-frame. η → U η.1).14)
satisfying Tr(σ µ σ ν ) = −2 gµν and Tr(σ µ σ ν ) = 2gµν = 2 δ µν (the matrices. C).

0 χm uR H(p) . 2 2 (4. e. Although these operators cannot be used to label the representations. ǫµνρσ will change sign.The Dirac equation
29
where
The matrix I2 does not belong to L↑ . For the spin 1/2 representations of the Poincar´ group including parity we.
(4. and the angular momentum in the rest frame is what we are familiar with as the spin of a particle.e.22) u= η η M (Λ) 0 where M (Λ) = ǫM ∗ (Λ)ǫ−1 with ǫ given in Eq. a) and ˜ a = (a0 . −a). we obtain for the two components of u which we will refer to as chiral right and chiral left components. that have the correct commutation relations. Taking M (Λ) = H(p). they can be connected. 2 ) and ( 2 . 1 ) and ( 1 . must combine the repe resentations. the aij elements of a tensor will not change sign. as we have seen in section 4. Thus M ∗ and M are not equivalent within SL(2. etc. Within the Lorentz group. The i angular momentum operators J are represented by σ i /2. i. From the deﬁnition of the representations (0. ) −→ ( . (axial vector).2
Spin 1/2 representations of the Lorentz group
1 1 Both representations (0.
−→ −r
(vector). generators. but to L↑ . but by a transformation belonging ↑ ↑ to the class P− . m) = χm uL H(p) 0
. where a = (a0 . i. It has the same eﬀect as lowering the indices. We also have seen that the representations (0.
The behavior is the same for classical quantities. Examples (with between brackets the Euclidean assigments) are
r t p H J λ(p) K
−→ −λ(p) −→ −K
−→ H −→ J
−→ t (scalar). = (4.13. but rather the operators W i (p) should be used. A parity transformation changes aµ into aµ . C). 0). the boost in Eq. 4. 0) (via operators J and K) one sees that under parity 2 2 1 1 (0. (pseudoscalar).17.21)
In nature parity turns (often) out to be a good quantum number for elementary particle states. they cannot be connected by a unitary 2 2 transformation.g. the well-known two-component spinor for a spin 1/2 particle. consider the four component spinor that transforms under a Lorentz transformation as 0 ξ ξ . C) are suitable for representing spin 1/2 particles.e.23) u(p. 0) of SL(2. therefore.20)
4. 4. we have seen that in the rest frame W i (p)/M → J i .
(scalar). −→ −p (vector). This provides the easiest way of seeing ˜ what is happening. For a particle at rest only angular momentum is important and we can choose ξ(0. m) = χm . (vector).1. + −
1 0 0 1 I2 = 0 0 0 0
0 0 −1 0
0 0 0 1
. −→ M (Λ) (4. m) = η(0. 1 ) and ( 1 . The representing member of the class P− is the parity or space inversion operator Is . 0) are inequivalent.

(iγ µ ∂µ − M ) ψ(x) = 0. To see this consider the equation for the hermitean † † conjugate spinor ψ † (noting that γ0 = γ0 and γi = −γi ). It is of a form that we also played with in Exercise 2.The Dirac equation
30
with H(p) = ¯ H(p) = 2M (E + M ) E+M −σ·p 2M (E + M ) E+M +σ·p . The general requirements for the γ matrices are thus easily obtained. = H −1 (p). .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. The ﬁrst indeed is solved.30) γ µ γ ν ∂µ ∂ν + M 2 ψ(x) = 0. (4. (4. (pµ γ µ − M ) u(p) ≡ (/ − M ) u(p) = 0. = (4. this can be rewritten 1 µ ν {γ . {γ µ .31)
To achieve also that the energy-momentum relation p2 = M 2 is satisﬁed for u(p). (4.24) (4. one must require that for ψ(x) the Klein-Gordon relation 2 + M 2 ψ(x) = 0 is valid for each component separately). The explicit realization appearing in Eq.
(4.29)
which is an explicit realization of the (momentum space) Dirac equation. We ﬁrst want to investigate if the Dirac equation solves the problem with the negative densities and negative energies. (4. We will discuss another explicit realization of this algebra in the next section. 0 H 2 (p) uR uR .32)
suppressing on the RHS the identity matrix in Dirac space.27)
where γ µ are 4 × 4 matrices called the Dirac matrices1 . Since ∂µ ∂ν is symmetric. As in section 2 we can use Pµ = i∂µ as a representation for the momenta (translation operators) in function space.27 is known as the Weyl representation. which in general is a linear equation in pµ . γ ν } = 2 g µν .7. Applying (iγ µ ∂µ + M ) from the left gives (4.33)
deﬁne for a four vector a the contraction a = aµ γµ /
. This leads to the Dirac equation for ψ(x) = u(p)e−i p·x in coordinate space. 4. 2 (4. γ } ∂µ ∂ν + M 2 ψ(x) = 0. From this one obtains the Cliﬀord algebra for the Dirac matrices. p (4.25)
It is straightforward to eliminate χm and obtain the following constraint on the components of u.28)
which is a covariant (linear) ﬁrst order diﬀerential equation. ψ † −iγ 0 ∂0 +iγ i ∂i −M
1 We
←
←
= 0.

We will see how this is properly implemented (also for bosons!) when quantizing ﬁelds.35)
is indeed positive and j 0 can serve as a probability density. (4. where γ 0 p0 ψ = M ψ
0
−→
Eψ = M γ 0 ψ. This problem was solved by Dirac through the introduction of a negative energy sea.
(4.37)
Using the explicit form of γ in the Weyl representation (see Eq. Relying on the Pauli exclusion principle for spin 1/2 particles.
(4. The Dirac sea forms the vacuum. Multiplying with γ 0 on the right and pulling it through one ﬁnds a covariant equation ψ(x) iγ µ ∂µ +M
←
= 0.34)
for the adjoint spinor ψ ≡ ψ † γ0 . one sees that there are two positive and two negative eigenvalues.g. opposite charge). This hole forms an antiparticle with the same mass.The Dirac equation
31
E empty levels M -M filled levels
E particle M -M hole
Figure 4.36) (4. This can most easily be seen in the particle rest frame.3
General representations of γ matrices and Dirac spinors
{γ µ .
. remains. 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. γ ν } = 2 g µν . γ =ρ ⊗1= γ = iρ ⊗ σ = 0 −1 −σ k 0
2 We
(4. the one of the negative energy states.27). The density j 0 = ψγ 0 ψ = ψ † ψ (4.1: The Dirac sea of negative energy states and antiparticles. .
4. but with properties such that it can be annihilated by the particle (e. From the equations for ψ and ψ one immediately sees that j µ = ψγ µ ψ is a conserved probability current (exercise!). The second problem encountered before. E = +M (twice) and E = −M (twice).39)
use ρi ⊗ σj with both ρ and σ being the standard 2 × 2 Pauli matrices. 4. From the vacuum a particle can be removed.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 .

49)
i 1 with −i S ρσ = 2 σ ρσ = − 4 [γ ρ . K 3 = M 30 = 0 iσ 3 2 4 or in general one has the transformation i ψ → Lψ = exp − ωρσ S ρσ ψ. We note that ψ → Lψ and ψ → ψ L−1 . = ρ ⊗ 1 = 1 0 i ǫµνρσ µ ν ρ σ γ γ γ γ . 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 .10 are 1 σ3 0 i J 3 = M 12 = = [γ 1 . γ µ } = 0 and explicitly one has 0 1 1 .9 and 4.45)
(γ5 )W.
0 γ k = −iρ2 ⊗ σ k = k σ
−σ k . 1 −1 2 For all representations one has
† γµ = γ0 γµ γ0 .
(4. which in the case of the Weyl representation are just the upper and lower components.R.47)
(4.48)
are projection operators. 4! (4. −→ Sγµ S −1 .44)
and an adjoint spinor deﬁned by ψ = ψ † γ0 .
Lorentz invariance
The Lorentz transformations can also be written in terms of Dirac matrices.
For instance in the Weyl representation (but valid generally). while L−1 γ µ L = Λµ γ ν . it is easy to see that PR/L = 1 (1 ± γ5 ) 2
1 0 . =ρ ⊗1= 0 −1
3
(4. 0
(4.40)
(4. γ 0 ]. (γµ )S.41) (4. 2 (4. 0 σ3 2 4 1 −iσ 3 0 = − i [γ 3 .R.46) (4. ν The latter assures Lorentz invariance of the Dirac equation (see Exercise 4.42)
We note that the explicit matrix taking us from the Weyl representation to the standard representation.R. the rotation and boost generators in Weyl representation in Eqs 4.43) S=√ . γ 2 ]. = S(γµ )W. is 1 1 1 (4.The Dirac equation
32
The Weyl (or chiral) representation: 0 1 . S −1 . (γ5 )S. For example. γ 0 = ρ1 ⊗ 1 = 1 0 γµ ψ
Diﬀerent representations can be related to each other by unitary transformations.3)
. that project out chiral right/left states. γ σ ]. −→ Sψ.R.

x = (t. One has ψ c = −ψ T C −1 .6). that brings ψ → ψ c . but it does. (or W. This symmetry does not change the spin 1/2 nature. parity. Some properties of C are C −1 = C † and C T = −C.g. −x) and via a few manipulations obtain again the ˜ Dirac equation (i/ − M ) ψ p (x) = 0. 0 −ǫ 0 0 −ǫ 0 . In any representation a matrix C exist.56)
(i/ − M ) ψ c (x) = 0.54)
e. e.R. called charge conjugation. ∂ with the solution ψ c (x) = ηc Cψ (x). C is real and one has C −1 = C † = C T = −C c and [C. = 0 ǫ iσ 2
(4.57) (4. ψR . x) → x = (t.g.52) η ξ
one can apply space inversion.50)
Charge conjugation
but with ψ p (x) ≡ ηp γ0 ψ(˜) (Exercise 4.R.The Dirac equation
33
Parity
There are a number of symmetries in the Dirac equation. = = 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. ∂ (4.55) (4. reverse the charge of the particle. ψ= (4. in Weyl representation we ﬁnd in Dirac space T C ξ −ǫ η ∗ . γ5 ] = 0. is a left-handed spinor. As with parity we look for a transformation. It is easy to convince oneself that if ψ(x) is a solution of the Dirac equation.
T
(4. The latter implies that the conjugate of a right-handed spinor.58)
where ηc is an arbitrary (unobservable) phase. (i/ − M ) ψ(x) = 0.R.53)
(4. which is again a solution of the Dirac equation.R. T
(4. (C)W. usually to be taken unity. for instance. 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) = 0.51)
The existence of positive and negative energy solutions implies another symmetry in the Dirac equation. such that
T Cγµ C −1 = −γµ .) and all representations connected via a real (up to an overall phase) matrix S. In S. ψ= (4. Explicitly. ∂ (4.59) −→ ψ c = C ψ = η ǫ ξ∗
. Note that we have (as expected) explicitly in Weyl x representation in Dirac space P ξ −→ ψ p = γ 0 ψ = η .
−iσ 2 0 −ǫ = . (C)S.

60) ψ(x) = 0. with E = Ep = + p2 + M 2 . ¯
(4.The Dirac equation
34
4.g. s) v = .68) (4.64) (4.65) (4. s) e−i k·x b(k.66)
An arbitrary spin 1/2 ﬁeld can be expanded in the independent solutions. ⇔ (/ − M ) u(p) = 0. s). s) and C v T (p. Note that the spinors in the negative energy modes (antiparticles) could be two diﬀerent spinors.69)
=
−
s
In order to project out spin states. ¯ ¯ † ′ u (p.. the spinors satisfy C uT (p. The explicit Dirac equation in the standard representation reads ∂ iσ · ∇ i ∂t − M (4. s′ ) = −2M δss′ . s)¯(p. s′ ) = 2M δss′ . s)u(p.1). s)u(p. s)u(p.62)
where χs are two independent (s = ±) two-component spinors. the best representation to describe particles at rest is the standard representation.4
Plane wave solutions
Looking for positive energy solutions ∝ exp(−iEt) one ﬁnds two solutions. s) = v(p. v(p. ¯
′ ′
Explicit solutions in the standard representation are χs .67)
It is straightforward to ﬁnd projection operators for the positive and negative energy states P+ P− =
s
p /+M u(p.70) = Ps = ˆ 0 1∓σ·s 2 2
. s ) = 0. The ¯ ¯ solutions are normalized to u(p. In an arbitrary frame one has s · p = 0 and s(p) can e. Choosing χ = −ǫχ∗ (the equivalent spin ¯ 1/2 conjugate representation). 2M 2M −/ + M p v(p. s) = Ep + M σ ·p Ep +M χs
. s) + v(k. s) u = . ψ(x) = u±s (p) e−i p·x . where v satisﬁes σ·p −(Ep + M ) v(p) = 0. 2M 2M
(4.63)
u(p. be obtained by a Lorentz transformation. 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. the spin polarization vector in the rest frame is the starting ˆ point. ψ(x) = v ±s (p) ei p·x . s)v(p. s) = u(p.61) σ·p −(Ep + M ) There are also two negative energy solutions. p −σ · p (Ep − M ) Ep + M σ ·p ¯ Ep +M χs χs ¯
For a free massive particle. s ) = v † (p. s)¯(p. It is easy to check that 1 1±σ·s 1 ± γ5 / s ˆ 0 (4. In that frame is a spacelike unit vector sµ = (0. s). s ) = v (p. s) = u(p. ⇔ (/ + M ) v(p) = 0. s) ei k·x d∗ (k. s)v(p. s)v(p.
v (p. s′ ) = 2Ep δss′ . s) . ∂ −iσ · ∇ −i ∂t − M
(4. p (4.
(4. where u satisﬁes −σ · p Ep − M u(p) = 0. (2π)3 2Ek
(4. in which γ 0 is diagonal (see discussion of negative energy states in section 4.

p ψR/L ≡ PR/L ψ are independent solutions.73)
Also for helicity states C uT (p. −1/2) = √ E + M −(E + M ) 0 . λ = −1/2) = u(p. λ = +1/2) = v(p. 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). −p3 −p1 − i p2 1 v(p.71) p − i p2 . 2 2 0 √ 2 √ 2 − E+M − E−M 0
.
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. λ = − 1 ) = √ √ v(p. +1/2) = √ 0 E +M E+M . v(p. λ). −1/2) = √ (4. λ) = v(p. For instance with the z-axis as spin quantization axis. One ﬁnds for the helicity states in Weyl representation: √ √ 0√ E +M + E −M √ 1 √ 1 0√ 1 . Therefore. This means that in the solution space for massless fermions the chirality states. etc.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 ). p3 E +M E+M 1 2 3 p + ip −p 1 2 p − ip 3 1 −p v(p. u(p. λ = + 1 ) = √ + E +M − E −M . λ = − 1 ) = √ E + M − E − M . λ) and C uT (p. 4. Note that for solutions of the massless Dirac equation /ψ = 0. . v(p. while the negative helicity (λ = −1/2) is in essence lefthanded. λ = +1/2) = E−M √ 0 − E −M 0 √ √ 0 − E−M − E −M 0 . The chirality projection operators in Eq. λ = −1/2) = √ − E+M √ 0 0 E+M
(4.74)
. u(p. helicity states. λ) = u(p.
(4. 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 √ . one has in standard representation: 0 E+M E+M 1 1 0 . We note that for high energy ¯ ¯ the positive helicity (λ = +1/2) is in essence righthanded. (4. +1/2) = √ u(p. The ratio of the components is (directly or inversely) proportional to √ √ E+M − E−M 2M M ≪E M √ √ √ . γ5 /ψ = 0 but also p p p /γ5 ψ = −γ5 /ψ = 0. √ 0 u(p.
By writing the helicity states in Weyl representation it is easy to project out righthanded (upper components) and lefthanded (lower components). 1 u(p.72)
ˆ Helicity states (p along z ) in Standard representation are: √ √ 0 E+M E +M . chirality.

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

Do this e. i S ρσ ] = g ρµ γ σ − g σµ γ ρ . 0 if n is odd. x ˜
. −x) is also a solution of the Dirac equation. a 4 a · b.6
Apply space-inversion. where x = (t. e.
(4. x → x. to the Dirac equation and use this to show that the spinor ψ p (x) = ˜ γ0 ψ(˜). 1 i −i S ρσ = − [γ ρ . 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 4.3
Imposing Lorentz invariance on the Dirac equation can be achieved by requiring L−1 γ µ L = Λµ γ ν . Show that the equality holds. ˜
Exercise 4.
Excercise 4.g. s) and v(p.13.4
Show that in P+ ≡
s=±1/2
u(p.The Dirac equation
37
The trace is linear and can be split up in parts containing traces with up to four gamma-matrices. γν } = 2gµν that γµ / γ µ a = Tr(//) = ab µn µ1 Tr(γ · · · γ ) = −2 /. s). s) u p /+M = 2M 2M
both sides are projection operators.
Exercise 4. ν µ µ Look at the inﬁnitesimal forms Λµ = gν + ων and L = 1 − (i/2)ωρσ S ρσ to obtain ν [γ µ . 4. s)¯(p.2
Show that j µ = ψγ µ ψ is a conserved current if ψ en ψ are solutions of the Dirac equation
Exercise 4.1
Prove Eq.5
Show starting from the relation {γµ . 4 2
Exercise 4. by proving that H 2 (p) = exp (η · σ) = (E + σ · p)/M = σ µ pµ /M. of which only the traces of four and two gamma-matrices are nonzero. H(p) = exp η·σ 2 = M +E+σ·p .g.
2M (E + M )
ˆ ˆ where E = M γ = M cosh(η) and p = M βγ n = M sinh(η) n. by letting them act on the four basis spinors u(p. γ σ ] = σ ρσ .

7
(a) Show that (in standard representation) j0 (kr)χm ψ(x.
. t) = N k ˆ i E+M σ · r j1 (kr)χm −i Et e . namely i / ψ = eiαγ5 ψ. 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. Show that the ratio is |M (π − → e− νe )|2 ¯ ≈ 54. ν p Note: how such parametrizations work will be explained in chapter 8.8
A negatively charged pion (π − ) decays in a lepton ℓ (ℓ can be an electron or a muon) and a corresponding anti-neutrino (¯ℓ ). 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. Note: To prove this. Calculate for now just |M |2 . 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 ). the lepton and neutrino have spin 1/2 and are ν described by momentum-space spinors.
Excercise 4. For such a cavity nµ = (0.66 MeV. you need to determine the corresponding conﬁnement condition for ψ. The conditions for ψ and ψ imply that nµ j µ = ψ/ ψ = 0. n (c) Apply the conﬁnement condition to ﬁnd the value of k for the lowest eigenmode in the spherical ˆ cavity.e.The Dirac equation
38
Exercise 4. i.
(b) Next. A condition for conﬁning the fermion in the cavity is i / ψ = ψ (where n is the surface normal). the decay would even be forbidden.57 MeV. Show that this n condition is suﬃcient to guarantee that nµ j µ = 0. The pion has spin 0. M R = 1. We will not pursue this chiral freedom at this point and take α = 0.511 MeV). Note: the outcome shows that the decay into muons is much more likely that the decay into electrons (mπ = 139. Do you understand why? The calculation of the life-time is completed in chapter 11. Calculate the value of k for M R = 0. we want to conﬁne this solution to a spherical cavity with radius R. mµ = 105.8 × 10−6 .5 and M → ∞. n It turns out that there is actually a whole family of conditions that provide conﬁnement. r). me = 0. |M (π − → µ− νµ )|2 ¯ Actually if the lepton would be massless. (d) Plot for these cases the vector densities ψ † ψ and the scalar densities ψψ as a function of r/R. which is the most important factor needed to ﬁnd the decay width (inverse lifetime).
is a stationary solution of the (free) Dirac equation. there is no current ﬂowing through the surface of the cavity.

The vectorﬁeld is therefore (λ) suited to describe spin 1. The covariant generalization is to describe them with Vµ (x) = ǫµ (k) e±ik·x .4)
λ=0.5) −E B2 . 39 (5. ǫλ ). where k = k0 = (M.± (λ)
d3 k ǫ(λ) (k) e−ik·x c(k. In the restframe. (5. (2 + M 2 )Vµ = 0. just as the two-component Dirac spinors did for spin 1/2.2 we have discussed the vectors ǫλ (with λ = 0 and ±1) as basis states of the spin 1 representation.1)
The Lorentz condition implies that k µ ǫµ (k) = 0.Chapter 5
Vector ﬁelds and Maxwell equations
5. A real vector ﬁeld Vµ (x) then can be expanded in modes as Vµ (x) =
λ=0. µ (2π)3 2E µ
(5. A). −B 1 0
3
With the electromagnetic ﬁeld tensor
(5. M2 (5.2) (5.2
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.6)
where Aµ is the four-vector potential
(5. one has ǫ(λ) (k) ǫ(λ)∗ (k) = −gµν + µ ν kµ kν .7)
. These can describe spin 1 states in the rest-frame. These are solutions of the Klein-Gordon equation with in addition a condition.1 Fields for spin 1
In section 3. ±(1) are obtained by boosting the restframe vectors. In an arbitrary frame ǫµ (k) with λ = 0.±
5. 0). namely ǫµ (k0 ) = (0. λ) . with ∂µ V µ = 0 (Lorentz condition). λ) + ǫ(λ)∗ (k) eik·x c∗ (k. there are then indeed three independent possibilities.3)
Since the vectors ǫµ (k) together with the momentum k µ form a complete set of four-vectors.

Taking the divergence of this equiation. (5. it does not eliminate the freedom of gauge transformations but they are now restricted to Aµ → Aµ + ∂µ χ with 2χ = 0.10)
the electric and magnetic ﬁelds are unchanged. The Lorenz condition can actually be incorporated by changing the equations of motion to 2Aµ + (λ − 1)∂µ (∂ν Aν ) = jµ . (5. For electrodynamics one has furthermore the freedom of gauge transformations. (5. ∓1. there is still a gauge freedom left (Aµ → Aµ + ∂µ χ with 2χ = 0). in which case one has 2Aµ = jµ .8)
(5.14) of which the solutions give the Li´nard-Wiechert potentials. j) the Maxwell equations read ∂µ F µν = j ν . if ∂µ Aµ = 0 for large times |t|. One possible choice is the gauge A0 = 0 (5. for M = 0. 0. Under a gauge transformation Aµ −→ Aµ + ∂µ χ. It states the absence of magnetic charges and Faraday’s law.16) (5. The equation in vacuum.18) corresponding to the helicity states λ = ±1 of the massless photon ﬁeld.5. Note that the incorporation of the Lorentz condition can also be done for the case of the massive vector ﬁeld.Vector ﬁelds and Maxwell equations and the (conserved) current j µ = (ρ. (5. 0. The gauge freedom can ﬁrst of all be used to impose the Lorentz condition for the electromagnetic ﬁeld. if in Eq. Therefore if k µ = (|k|. 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). Although the Lorenz condition is a constraint. AP (x) = Aµ (˜). Furthermore. 5.17) (radiation. (5.9)
1 ˜ where F µν ≡ − 2 ǫµνρσ Fρσ is the tensor where E → B and B → −E. only two independent vectors remain ǫ(±) µ (k) = (0.13)
Hence. −i. Fµν −→ Fµν + (∂µ ∂ν χ − ∂ν ∂µ χ) = Fµν . In that case one also has ∇·A = 0 or ki ǫi (k) = 0. 0).15 the d’Alembertian is replaced by 2 + M 2 .12) (5. ˜ x Note that via parity transformation one obtains another solution. shows that the electromagnetic ﬁelds correspond to the case of a massless spin 1 ﬁeld. As remarked above.
40
(5. ∂µ Aµ = 0. The equations of motion for the ﬁelds Aµ become 2Aµ − ∂µ (∂ν Aν ) = jµ . |k|). moree over. it will vanish at all times. transverse or Coulomb gauge).15) (5.e.
. i.11)
This equation is not aﬀected by a gauge transformation. the antisymmetry of F µν implies that ˜ ∂µ F µν = 0 or ∂λ Fµν + ∂µ Fνλ + ∂ν Fλµ = 0. 2Aµ = 0. in accordance with what we found in section 3.

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

organizing itself in tiny ﬂux tubes (Abrikosov strings). it can be obtained from the situation A = 0 by a gauge transformation. 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. has been employed by Dirac in constructing magnetic monopoles in electrodynamics 2 Actually. implying that it must be single-valued1 .e. Therefore the gauge transformation must be uniquely deﬁned in the space one is working in. The diﬀerence here is that we are working in a space with a ’defect’ (the inﬁnitely long solenoid). The AharonovBohm experiment shows a realization of this possibility. Therefore space outside the solenoid doesn’t care that χ is multi-valued. 5. i. 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. 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. giving no observable phase2 . the relevant charge turns out to be 2e because the electrons appear in Cooper pairs
. Another situation in which the topology of space can be used is in the case of superconductors. Consider a superconductor. 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. nonobservable phases ϕ = 2π n. In such a space loops with diﬀerent winding number (i.e.21. the number of times they go around the defect) cannot be continuously deformed into one another. forming a simple (connected) space. A = ∇χ with χ= BR2 ϕ 2 (5. The observable phase going around in general is ∆δ = e A · dr = e ∇χ · dr = e χ|ϕ=0 . This situation that A can be written as ∇χ is always true when ∇ × A = 0. minimizing the space occupied by B ﬁelds.
1 The occurrence of nontrivial possibilities. however. Below the critical temperature the magnetic ﬁeld is squeezed out of the superconducting material.
ϕ=2π
The function χ in Eq.21)
(ϕ being the azimuthal angle).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. is multivalued as ϕ = 0 and ϕ = 2π are the same point in space. Now back to the Aharonov-Bohm experiment.

∂t
∂B = 0. ∇×B =j+ ∇×E+ ∂E .8 and 5.Vector ﬁelds and Maxwell equations
43
Exercises
Exercise 5.
.1
Show that Eqs. 5.9 explicitly give the equations ∇ · E = ρ. ∂t ∇ · B = 0.

∂x (6. The requirement δS = 0 with ﬁxed boundaries x(t1 ) = x1 and x(t2 ) = ˙ x2 leads to
t2 ′
(6.1)
and as an example the lagrangian L(x. thus t′ = t + δτ. These equations can be obtained from a lagrangian using the action principle.
t1
(6.Chapter 6
Classical lagrangian ﬁeld theory
6.8)
44
. d dt ∂L ∂x ˙ = ∂L . recall classical mechanics with the action
t2
S=
t1
dt L(x. and the total change x′ (t′ ) = x(t) + ∆x(t).3) (6.6)
The ﬁrst term leads to the Euler-Lagrange equations.2)
The principle of minimal action looks for a stationary action under variations in the coordinates and time. ˙ 2 (6. ∂x ˙ (6. As an example. x) = K − V = ˙ 1 mx2 − V (x). ˙ p= ∂L . x (t) = x(t) + δx(t).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
.1 Euler-Lagrange equations
In the previous chapter we have seen the equations for scalar ﬁelds (Klein-Gordon equation).7)
The quantity ∂L/∂ x plays a special role and is known as the canonical momentum.5) with ∆x(t) = δx(t) + x(t) δτ . Dirac ﬁelds (Dirac equation) and massless vector ﬁelds (Maxwell equations) and corresponding to these ﬁelds conserved currents describing the probability and probability current. x) ˙
(6. (6.

t2 )
d3 x .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
(6.Classical lagrangian ﬁeld theory
45
For the lagrangian speciﬁed above this leads directly to Newton’s law p = m¨ = −∂V /∂x. The hamiltonian H is deﬁned by H(p.
λ=0
(6. . . but which in 3 dimensions includes conserved quantities related to rotations and boosts. indicated as x′µ φ (x) or combined φ′ (x′ ) = φ(x) + ∆φ(x). In classical ﬁeld theory one proceeds in complete analogy but using functions depending on space and time (classical ﬁelds. .
(R3 . ∂µ φ(x)) =
R
d4 x L (φ(x). + ∂L ∆x − H δτ ∂x ˙
t2
.15)
d4 x′ L (φ′ . gives rise to a conserved quantity (at least if H is invariant!) Q=p dx dλ −H dt dλ . δ(∂µ φ)
(6. think for instance of a temperature or density distribution or of an electromagnetic ﬁeld). also indicated by ∂R (a more general volume in four-dimensional space-time with some boundary ∂R can also be considered).6 can be rewritten as δS = . ∆x = (dx/dλ)λ=0 δλ and δt = (dt/dλ)λ=0 δλ. x′ ) −
d4 x L (φ. dσµ . In general any continuous transformation. . x) ≡ p x − L . ˙ (6. Consider a lagrangian density L which depends on these functions.17)
Furthermore the variations δφ and δ∂µ φ contribute to δS. −
d3 x . The resulting variation of the action is δS =
R µ ′
= xµ + δxµ . . . . x) and an action
t2
S[φ] =
t1
dt L =
dt d3 x L (φ(x). which in classical mechanics vanishes at the boundary) does not play a role. ∂µ φ. 6.2.11)
λ=0
of which p and H are the simplest examples related to space and time translations. ∂µ φ′ .
. .
t1
(6.18)
1 Taking a functional derivative. L (φ(x). t1 ) and R3 . ∂µ φ(x)). H is a conserved quantity. with ∆φ(x) = δφ(x) + (∂µ φ)δx .
(6. x). .t1 )
(6. 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 . Variations in the action can come from the coordinates or the ﬁelds. ˙ x The second term in Eq. indicated with δF [φ]/δφ should pose no problems. = φ(x) + δφ(x)
(6.14) (6.9)
which is done because the ﬁrst term (multiplying ∆x. i.10)
One sees that invariance under time translations requires that H(t1 ) = H(t2 ).e. =
∂R (R3 . We will come back to it in a bit more formal way in section 9. t2 ).
∂R
Note for the speciﬁc choice of the surface for constant times t1 and t2 .12)
Here R indicates a space-time volume bounded by (R3 . their derivatives and possibly on the position.13) (6.

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.Classical lagrangian ﬁeld theory
46
With for the situation of classical ﬁelds all variations of the ﬁelds and coordinates at the surface vanishing.
The scalar ﬁeld
It is straightforward to derive the equations of motion for a real scalar ﬁeld φ from the lagrangian densities. (2 + M )φ (x) = 0. 2
ψ = 0. the Dirac ﬁeld (spin 1/2) and the vector ﬁeld (spin 1) can be found. leading to (2 + M 2 )φ(x) = 0. 2 (6. (6.
(6. The integrand of the ﬁrst term must vanish.24) (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).19) δ(∂µ φ) δφ
6.21)
which diﬀer only by surface terms.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 the Euler-Lagrange equations. L = 1 1 ∂µ φ∂ µ φ − M 2 φ2 2 2 1 µ = − φ ∂µ ∂ + M 2 φ. (6. the second term is irrelevant. Using the variations in ψ (in the symmetric form).20) (6.25) (6.23) (6. For the complex scalar ﬁeld one conventionally uses L = = ∂µ φ∗ ∂ µ φ − M 2 φ∗ φ −φ∗ ∂µ ∂ µ + M 2 φ. ∂ (6.29)
.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. δL δL ∂µ = .2
Lagrangians for spin 0. One easily obtains (2 + M 2 )φ(x) = 0.
2 ∗
(6.27) (6. δL δ(∂µ ψ) δL δψ one obtains immediately i / −M ∂
→
i = − γ µψ 2 i → = ∂ / ψ − M ψ.

38) (6. In that case one ﬁnds trivially L =
↔ ↔ 1 1 ψR i / ψR + ψL i / ψL − M (ψR ψL + ψL ψR ).31)
showing e. The Majorana case is in fact more general.32)
There exists another possibility to write down a mass term with only one kind of ﬁelds.33)
We now introduce a ’real’ (four-component) spinor satisfying Υc = Υ with ψL . Peshkin and Schroeder or Exercise 12. ψL ≡ (ψL )c = C ψL = 0
(6.g. 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. 1 M η † ǫη ∗ − M ∗ η T ǫη .30)
It is often useful to link to the two-component spinors ξ and η which we started with in chapter 4. or equivalently separate the ﬁeld into right.
and thus (β ∗ α)∗ = α∗ β = −βα∗ .g. but note the factor 1/2 as compared to the Dirac lagrangian. 2
αβ = −βα.
(6.39)
2 This
term is written down with η being an anticommuting Grassmann number for which (αβ)∗ = β ∗ α∗ = −α∗ β ∗ . Using the twospinors ξ and η.37)
This gives an expression with a mass term that is actually of the same form as the Dirac mass term.36) (6.Classical lagrangian ﬁeld theory
47
and similarly from the variation with respect to ψ ψ i / +M ∂
←
= 0. The reasons for Grassmann variables will become clear in the next chapter. (6. Υi / Υ − ∂ 4 2
T −ǫ η ∗ c = ΥR . 2 (6. that the lagrangian separates into two independent parts for M = 0. which comes because we in essence use ’real’ spinors. ∂ ∂ 2 2
(6.34) (6.35)
Since the kinetic term in L separates naturally in left and right parts (or ξ and η).
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 .7).
(6.
.31 is given by LM (Dirac) = −M ξ † η + η † ξ . ∗ 0 ⇒ Υ ≡ −ǫ η ΥL = ψL = η η
∗ for which η0 = ǫ η0 . We note that
of which the left part coincides . 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. L = =
↔ 1 1 c c M ψL ψL + M ∗ ψL ψL ψL i / ψL − ∂ 2 2 ↔ 1 1 M ΥR ΥL + M ∗ ΥL ΥR .and lefthanded ones. the mass term in the Dirac lagrangian 6.

(6. t). consider U (1) transformations of the Dirac ﬁeld proportional to the charge e. These do not aﬀect the dynamics and will give rise to conserved currents. .47)
. the presence of a symmetry that leaves the lagrangian invariant. these conserved quantities become operators. Q= d3 x J 0 (x. (6.43)
F (σ1 ) − F (σ2 ) =
dσµ J µ (x) =
∂R R
d4 x ∂µ J µ (x). In particular when the lagrangian is invariant under symmetries. requires the presence of a conserved quantity.41) (6. which is the space-integral over the zero-component of a conserved current.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 . one can consider the variations at the initial or ﬁnal surface. In this case.σ1 with σ = (R3 . Returning to δS the surface term is rewritten to δS =
R
d4 x .42)
Θµν = The variation δS thus can be expressed as δS with J µ (x) = − =
δL ∂ ν φ − L g µν . δ(∂µ φ)
For continuous transformations. ∆φ = (dφ/dλ)λ=0 δλ and δxµ = (dxµ /dλ)λ=0 δλ. (6..44)
δL ∆φ + Θµν (x)δxν .40)
6. As an example for the classical ﬁeld. Q(t1 ) = Q(t2 ). we get (omitting the parameter δλ) dxν dφ δL + Θµν (x) . 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. . t). ∂µ J µ (x) = 0. δ(∂µ φ)
(6.18. which in a consistent picture precisely generate the symmetries. δ(∂µ φ)
(6. +
∂R
dσµ dσµ
∂R
≡ where
d x .45) J µ (x) = − δ(∂µ φ) dλ λ=0 dλ λ=0 Thus considering ∂R = σ2 . 6. (6.+
R
4
δL δL ∆φ − ∂ ν φ − L g µν δxν δ(∂µ φ) δ(∂µ φ) δL ∆φ − Θµν (x)δxν . implying ∂µ F µν + M 2 V ν = 0 and ∂µ V µ = 0. ψ(x) → ei eΛ ψ(x) or ∆ψ(x) = i eΛ ψ(x). we consider the second term in Eq. discussed in the next chapter.3
Symmetries and conserved (Noether) currents
Next.
(6.Classical lagrangian ﬁeld theory
48
where Fµν = ∂µ Vν − ∂ν Vµ ..

˙ Q = d3 x j 0 (x.
.51) (6.
Lorentz transformations
In this case the transformation of the coordinates and ﬁelds are written as δxµ ∆φi (x) = = ω µν xν .49)
These currents are conserved as discussed already in chapter 2.
↔
(6.
6. Summarizing.54) (6.
(6. 1 (6. In the next section we will see the quantity show up as the charge operator. For quantized ﬁelds this will become the expressions of the hamiltonian and the momentum operators in terms of the ﬁelds. = 0.57) For the ﬁelds the behavior under Lorentz transformations has been the subject of previous chapters. The integral over the zero-component.Classical lagrangian ﬁeld theory From the lagrangian for the Dirac ﬁeld.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. including e the space-time translations and the Lorentz transformations.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 . H = d3 x Θ00 (x).g. e.50) (6. (6. (ω µν antisymmetric) 1 ωρσ (−i S ρσ )ij φj (x).
Translations
µ Under translations (generated by Pop ) we have
δxµ ∆φ(x) δφ(x)
= ǫµ . 2 (6. = −(∂µ φ)δxµ = −ǫµ ∂µ φ(x).
(6.4
Space-time symmetries
One kind of symmetries that leave the lagrangian invariant are the Poincar´ transformations.55)
Note that the coordinate transformations can be written in a form similar to that for the ﬁelds (matrices in internal space). (6. t) is the conserved charge (Q = 0).56) δxµ = ωρσ (aρσ )µν xν 2 with (aρσ )µν = aρσ µν = g ρµ g σν − g σµ g ρν . we obtain (since δxµ = 0).48)
For the complex scalar ﬁeld the U (1) transformations leave the lagrangian invariant and lead to the conserved current (see Exercise 6. 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. From Noether’s theorem one sees that the current Θµν ǫν is conserved.53)
These are the energy and momentum.

φ(x) φ (x) ∂µ φ(x)
∗
→ e
→ e
→ ei eΛ(x) φ(x). (6. Gauge transformations of the second kind or local gauge transformations are transformations of the type φ(x) → ei eΛ(x) φ(x).e.3).68) (6.
∂µ φ(x) + i e ∂µ Λ(x) e
(6.g. 2
(6. In some applications (speciﬁcally coupling to gravity) it is advantageous to have an equivalent current that is symmetric.
(6. (6.69)
i eΛ(x)
φ (x). γσ ] This result for the Dirac ﬁeld has been discussed in chapter 3 (see also Exercise 4. (6.
−i eΛ(x) ∗ i eΛ(x)
(6. φ(x). Any lagrangian containing derivatives. ∂µ T
µν
(6.63)
µν
= ∂µ Θ
. • Vector ﬁeld Aµ : −i Sρσ = aρσ .66)
In section 6. e.58) (6.70)
. one has a tensor that satisﬁes (exercise 6.6) T µν = T νµ . • Dirac ﬁeld ψ: −i Sρσ = −(1/4)[γρ . In general this tensor is not symmetric.
6. the angle of the transformation depends on the space-time point x. where H ρµν is the quantity appearing in M ρµν .65)
M µρσ = T µρ xσ − T µσ xρ .5
(Abelian) gauge theories
φ(x) → ei eΛ φ(x). Deﬁning T µν = Θµν − ∂ρ Gρµν .
in which the U (1) transformation involves an angle e Λ.62) with Gρµν = (H ρµν + H µνρ + H νµρ )/2.64) (6.60)
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 . The lagrangians which we have considered sofar are invariant under global gauge transformations and corresponding to this there exist a conserved Noether current.59) (6. is not invariant under local gauge transformations.3 we have seen global transformations or gauge transformations of the ﬁrst kind.Classical lagrangian ﬁeld theory
50
• Scalar ﬁeld φ: −i Sρσ = 0.67)
i. however. (6.61)
A ﬁnal note concernes the symmetry properties of Θµν . 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 . independent of x.

Dµ φ) − Fµν F µν . (6.78)
. 2 i ∂ / ψ − M ψ − e/ ψ A 2
giving the Dirac equation in an electromagnetic ﬁeld.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). (6. For this purpose it is necessary to introduce a vector ﬁeld Aµ . Dµ φ(x) ≡ (∂µ + i eAµ (x))φ(x). 4 (6. The equation of motion for the fermion follow from δL δ(∂µ ψ) δL δψ = = i − γ µ ψ. −e j µ Aµ = −e ρφ + e j · A.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 .71)
Thus the covariant derivative has the correct transformation behavior provided Aµ → Aµ − ∂µ Λ. D ∂ A (6. (6. 1 L (φ. 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.73) (6.72) (6. (i/ − M ) ψ = (i/ − e/ − M ) ψ = 0. 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. i L = ψγ µ ∂µ ψ − (∂µ ψ)γ µ ψ − M ψψ.Classical lagrangian ﬁeld theory
51
where the last term spoils gauge invariance. ∂ 2 4 (6. As an example consider the Dirac lagrangian. 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 .75)
The ﬁeld φ is used here in a general sense standing for any possible ﬁeld.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 . ∂µ φ) =⇒ L (φ. 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).

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. where j µ = e ψγ µ ψ. i. and deduce what is the charge of the antiparticle as compared to a particle.
Exercise 6.
giving the Maxwell equation coupling to the electromagnetic current.
.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. [iγ µ (∂µ − i e Aµ ) − M ]ψ = 0. (c) Show that the interaction term is invariant under gauge transformations only if the current jµ is conserved.
Exercise 6. −eψγ ν ψ. φ → eiΛ φ. ∂µ j µ = 0 (Note that the addition of a total derivative to the lagrangian density does not modify the equations of motion).2
(a) Show that the homogeneous Maxwell equations can be derived from the lagrangian density 1 L = − Fµν F µν . give the equation which is satisﬁed by ψ c = Cψ .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 . (6.e.4
Given the Dirac equation for a (negatively charged) spin-1/2 particle in an external electromagnetic T ﬁeld.
↔
Exercise 6.79)
Exercises
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.Classical lagrangian ﬁeld theory
52
For the photon the equations of motion follow from δL δ(∂µ Aν ) δL δAν = = −F µν . ∂µ F µν = j ν . ∂µ F µν = j ν .

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. Hint ≡ −gs µe · B = −gs Qe s · B. Hints: it may be useful to realize that Gµρσ is antisymmetric in ﬁrst two indices. ﬁnally note that it is suﬃcient to show the last equation for dσµ M µρσ . t). This relation is known as the Pauli equation.4.4) 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 . Comment: this relation is known as the Pauli equation.
.5
Show that the Dirac equation for an electron with charge −e in an external electromagnetic ﬁeld Aµ = (A0 . 2M 2M with Enr ≡ E − M ≪ M . 2me
Exercise 6. use ∂µ T µν = 0 and ∂µ M µρσ = 0.6 (optional)
Prove the properties of the tensor T µν in section 6.Classical lagrangian ﬁeld theory
53
Exercise 6. in order to show that Θρσ − Θσρ = ∂µ H µρσ . t) = Enr ψu (r. A) (see also exercise 6. e 1 (−i∇ + eA)2 + σ · B − eA0 ψu (r.

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

2. ∆φ. t). discussed in the section 6. ∇φ) (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. the ﬁelds φ(x) can be considered as regular operators in the Hilbert space after smearing with a test function f . 2 2 2
(7. qx+∆x ) . δ∂0 φ) 1 1 1 = Θ00 (x) = (∂0 φ)2 + (∇φ)2 + M 2 φ2 . As an example. Π(y. px (t) = ∂L ∂Lx = ∆3 x = ∆3 x Πx (t).6)
∆3 x L x (qx .14)
In fact. This implies speciﬁc transformation properties in the Hilbert space for the quantum ﬁelds φ. φ.11) (7. py (t)] = i δxy or for the ﬁelds φ(x) and Π(x) the socalled equal time commutation relations [φ(x. ∂ qx ˙ ∂ qx ˙ (7. (7. 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. qx .12) (7. ∂µ φ(x)) = ˙ d3 x L (φ. a) = Rj (Λ−1 ) φj (Λ−1 x − a).
(7. t). a) φi (x)U (Λ. ˙
where Πx (t) is obtained from the continuous ﬁeld Π(x) ≡ δL /δ(∂0 φ). albeit with ’sharp’ functions. (7.9)
Note that this indeed corresponds to the zero-zero component (Θ00 ) of the conserved energy-momentum stress tensor Θµν . ˙ δ φ(x) (7.15) with furthermore the relations [φ(x. Φ(f ) = d4 x φ(x) f (x). t)] = [Π(x.16)
. t)] = i δ 3 (x − y). the discretization procedure above is an explicit example. Sometimes it is useful to keep in mind that.8) (7. 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. formally.13)
Π(x) H (x)
For the quantization procedure we can formulate a number of basic axioms of quantum ﬁeld theory. • Poincar´ invariance e The full variation of the ﬁelds.7) d3 x L (φ(x).Quantization of ﬁelds
55
etc. t)] = 0.
i U † (Λ. t). Π(y.10) x ∆3 x [Πx qx − L x ] ˙ ∆3 x Hx (7. • Canonical commutation relations The quantization of the theory is achieved by imposing the canonical quantization condition [qx (t). The essential conditions to consistently quantize a theory then are the following. φ(y.

and [N. Mµν ] =
i
i∂µ φi (x). Q] = [P.24)
It is straightforward to ﬁnd the commutation relations between N and a and a† . (7. a† ] = a† .25)
. (7. a] = −a. a) =
i Rj (Λ−1 ) =
1 + iǫµ P µ − 1−
i ωµν M µν . In simpliﬁed form the hamiltonian is given by H= 1 2 1 2 2 P + ω Q . j 2
one has [φi (x).22)
it is straightforward to check that the commutation relations between Q and P are equivalent with the commutation relations [a. 2
i ωµν (S µν )i .Quantization of ﬁelds
56
or if U (Λ. [φ(x). 2 2 (7. Φ(g)] = 0. i(xµ ∂ν − xν ∂µ )φ (x) +
i
(7. a† ] = 1. P ] = i. H = = ω ω P P Q−i √ Q+i√ 2 2 2ω 2ω 1 1 =ω N+ . Pµ ] = [φ (x). The measurements cannot inﬂuence each other or [Φ(f ). 1 Q = √ (a + a† ) 2ω and P = −i ω (a − a† ) 2 (7. j (7.20)
where the coordinate Q and the momentum P satisfy the canonical commutation relations [Q.4. P ] = 0 (7.17) (Sµν )i φj (x).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. 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. • Causality Operators Φ(f ) and Φ(g) for which the test functions are space-like separated can be measured simultaneously (macroscopic causality). [N. Microscopic causality implies local commutativity. φ(y)] = 0 if (x − y)2 < 0. P ] 2 (7.18)
with S µν given in 6.21)
Writing Q and P in terms of creation (a† ) and annihilation (a) operators. ω a† a + 2 2 ω i − [Q. a† ] = 0.19)
7. a] = [a† . [a. sometimes referred to as second quantization. (7. [Q.23) The hamiltonian in this case can be expressed in the number operator N = a† a.

and
(7.29)
It is straightforward to prove that the equal time commutation relations between φ(x) and Π(x′ ) are equivalent with ’harmonic oscillator .30)
[φ(x).
7. (n − 1) a|n
† i. (2π)3 2Ek
(7. which we have split into positive and negative energy pieces with (complex) coeﬃcients a(k) and a∗ (k) multiplying them.28) (7. [φ(x).4 for the classical ﬁeld) a(k) a† (k ′ ) = = d3 x ei k·x i ∂0 φ(x).32) (7.26) a(k) e−i k·x + a† (k) ei k·x .like’ commutation relations between a(k) and a† (k ′ ). 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. d3 x φ(x) i ∂0 e−i k ·x
↔
′
↔
(7.e. is equivalent with [a(k). a† (k ′ )] = 0.33)
. φ(x )]x0 =x′0 = [Π(x). Π(x )]x0 =x′0 = 0. Note that we will often write a(k) or a† (k).Quantization of ﬁelds
57
Deﬁning states |n as eigenstates of N with eigenvalue n. N |n = n|n one ﬁnds N a† |n N a|n = = (n + 1) a† |n .e. (7. the real scalar ﬁeld φ(x) becomes d3 k (7. Π(x′ )]x0 =x′0 = i δ 3 (x − x′ )
′ ′
and (7. and we see that a state |0 must exist for which N |0 = a|0 = 0. φ(x) = (2π)3 2Ek where the Fourier coeﬃcients a(k) and a† (k) are now operators. a(k ′ )] = [a† (k). which represents the ’number of particles’ with momentum k. The quantization of the ﬁeld is achieved by quantizing the coeﬃcients in the Fourier expansion.g. From the normalizations one obtains a |n = √ n + 1 |n + 1 and a|n = n |n − 1 .31)
The hamiltonian can be rewritten in terms of a number operator N (k) = N (k) = a† (k)a(k). e. i.3
The real scalar ﬁeld
We have expanded the (classical) ﬁeld in plane wave solutions. a† (k ′ )] = (2π)3 2Ek δ 3 (k − k′ ) [a(k). The canonical momentum becomes −i d3 k ˙ Π(x) = φ(x) = a(k) e−i k·x − a† (k) ei k·x . but one needs to realize that in that case k 0 = Ek = k2 + M 2 . 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 .27) 2 (2π)3 It is easy to check that these equations can be inverted (see Exercise 2. † √ a and a act as raising and lowering operators.

34)
where V = (2π)3 δ 3 (0) is the space-volume.
.38)
d4 k 2π δ(k 2 − M 2 ) θ(k 0 ) |k k| = (2π)4
(7. φ(x) φ+ (x) φ− (x) = = = φ+ (x) + φ− (x).41) (7.44)
p|φ− (x)|0 = ei p·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). .42) (7. writing all annihilation operators to the right of the creation operators. . (2π)3 2Ek (7.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. (2π)3 (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). This term will be adressed below. |k = a† (k)|0 . i.36)
and satisfying the completeness condition 1 =
k|k ′ = (2π)3 2Ek δ 3 (k − k′ ). (2π)3 2Ek (7. . 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). assuring that the vacuum (by deﬁnition) has eigenvalue 0!.46) (7. k 0 = Ek = k2 + M 2 ).39)
The problem with the vacuum or zero-point energy. deﬁning particle states |k = |k (with positive energy. is solved by subtracting it as an (inﬁnite) constant. (2π)3 2Ek d3 k a† (k) ei k·x .37) (7. n1 ! n2 ! (7. For the purpose of normal ordering it is convenient to decompose the ﬁeld in positive and negative frequency parts. a(k)|0 = 0.35)
where the vacuum contribution disappears because of rotational symmetry. . d3 k a(k) e−i k·x . (2π)3 2Ek
(7. The rest of the states are then obtained from the groundstate via the creation operator. and multiparticle states |(k1 )n1 (k2 )n2 .45) (7. (2π)3 2Ek (7. 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. which now contains an inﬁnite number of oscillators. |0 . = normalized as (a† (k1 ))n1 (a† (k2 ))n2 √ √ . in ﬁeld theory also referred to as vacuum energy Evac = 1 V 2 d3 k Ek . d3 k |k k|. which amounts to redeﬁning H as H= d3 x : H (x) := d3 k Ek N (k).e.40)
This procedure is known as normal ordering.

(ii) ∆(x) is a solution of the homogeneous Klein-Gordon equation.52)
The result for the invariant commutator function is [φ(x). 2π (7. d3 k ei k·(x−y) = −i∆+ (y − x).17) and (7. φ− (y)]
≡ i∆+ (x − y). (7. (7.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. φ+ (y)]
d3 k e−i k·(x−y) . The last item to be checked for the scalar ﬁeld are the causality condition. ∆(x) = − ǫ(x0 ) δ(x2 ).18).56) = −δ 3 (x). i. 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 ).e. (iv) The equal time commutation relations follow from ∂ ∆(x) ∂t (v) For M = 0. x) = 0 and hence ∆(x) = 0 for x2 < 0.53)
t=0
. (2π)4 (7. i∆+ (x) i∆− (x) = = = d4 k θ(k 0 ) 2π δ(k 2 − M 2 ) e−i k·x .48) (7.55) d4 k ǫ(k 0 ) 2π δ(k 2 − M 2 ) e−i k·x . φ(y)] consider
µν
[φ+ (x).51) (7. that they satisfy the required commutation relations in Eqs (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 (7.47)
′ d3 k ′ e−i k·x+i k ·y [a(k). = − (2π)3 2Ek
(7. (iii) ∆(0.50)
or as integrals over d4 k. a† (k ′ )] 3 2E ′ (2π) k
[φ− (x).49) (7. (2π)3 2Ek ≡ i∆− (x − y). In order to calculate [φ(x). d3 k = (2π)3 2Ek =
(7.54) = i∆(x − y) = i (∆+ (x − y) + ∆− (x − y)) .

b† (k ′ )] = (2π)3 2Ek δ 3 (k − k′ ).5
The Dirac ﬁeld
L =
↔ i ψγ µ ∂µ ψ − M ψψ. Also the charge operator requires normal ordering (in order to give the vacuum eigenvalue zero).60) = i φ∗ ∂µ φ.e. (2π)3 2Ek d3 k b(k) e−i k·x + a† (k) ei k·x .59) (7. which is equivalent to the relations (only nonzero ones) [a(k). (2π)3 2Ek (7. 2
From the lagrangian density
(7.64)
The commutator of φ and φ† is as for the real ﬁeld given by [φ(x).63)
i. 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) . 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) . we will consider it as a repetition of the quantization procedure.62) (7.Quantization of ﬁelds
60
7. (2π)3 2Ek (7.4
The complex scalar ﬁeld
In spite of the similarity with the case of the real ﬁeld.65)
7.
↔
(7.61)
(7.58)
and satisfy the equal time commutation relation (only nonzero ones) [φ(x). a† (k ′ )] = [b(k). (7. (2π)3 2Ek
(7. particles (created by a† ) and antiparticles (created by b† ) with the same momentum contribute equally to the energy.66)
. extending it with the charge operator and the introduction of particle and antiparticle operators.57) (7. φ† (y)] = i∆(x − y). ∂0 φ† (y)]x0 =y0 = i δ 3 (x − y). = ∂{µ φ∗ ∂ν} φ − L gµν . 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 .

s)v(k. s)vj (k. i.74) (7. (7.79)
Also for the ﬁeld and the canonical conjugate momentum anticommutation relations are considered. s) ei k·(x−y) x0 =y 0
.Quantization of ﬁelds
61
the conserved density and energy-momentum currents are easily obtained. b† (k ′ . s)e−i k·x + d† (k.67) i ψ / ψ − M ψψ gµν .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) : .71) (7. interchanging creation and annihilation operators. ψj (y)}x0 =y0
=
d3 k (2π)3 2Ek
ui (k. u v (2π)3 2Ek
(7. The solution is the introduction of anticommutation relations. s)ei k·x . (7. s). s) e−i k·(x−y) j
s † + vi (k. s).70)
where the last line is obtained by using the Dirac equation.
. 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) . s)u† (k. jµ Θµν = = ψγµ ψ. ∂ 2
↔
(7.73) (7. s′ )} = {d(k.e.77)
Note that achieving normal ordering. (2π)3 2Ek (7. (2π)3 2Ek (7.76)
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. s)u(k. s)ei k·x + d(k.69)
(7. ↔ i ψγµ ∂ν ψ − 2 (7.
† {ψi (x).78) (7.68)
The canonical momentum and the hamiltonian are given by Π(x) H (x) = = δL = i ψ † (x). (2π)3 2Ek d3 k b† (k. ˙ δ ψ(x) ↔ i Θ00 (x) = − ψγ i ∂i ψ + M ψψ 2 = iψγ 0 ∂0 ψ = iψ † ∂0 ψ.75) (7. s)e−i k·x . s′ )} = (2π)3 2Ek δ 3 (k − k′ ) δss′ . d† (k ′ . The quantized ﬁelds are written ψ(x) =
s
d3 k b(k. s)¯(k. {b(k. s)¯(k.

86) (7. 4 2 This gives the equations of motion discussed before. but has the problem of being noncovariant.83)
From the lagrangian density
the canonical momenta are = = (7. it implies the Lorentz constraint and leads to the canonical momenta Π0 Π
i
= =
E i.85)
which reﬂects the gauge freedom. 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 (7.87)
If one wants to impose canonical commutation relations ∂µ Aµ = 0 cannot hold as an operator identity.6
The electromagnetic ﬁeld
1 L = − Fµν F µν . ψ j (y)} at arbitrary times one has {ψi (x).88)
. ˙ δ Ai (7.Quantization of ﬁelds
62
Using the positive and negative energy projection operators discussed in section 4. but we must restrict ourselves to the weaker condition B|∂µ Aµ |A = 0.82) where i∆1 = i∆+ −i∆− . ˙ δ A0 δL = F i0 = E i . ∂
(7. as the vanishing of Π0 induces a constraint.
(7. (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). one has
† {ψi (x). It is possible to continue in a covariant way with the lagrangian 1 λ L = − Fµν F µν − (∂µ Aµ )2 . ∂ (7.81)
where i∆ = i∆+ + i∆− is the same invariant commutator function as encountered before.84) (7. For the scalar combination {ψi (x).80)
3
= δ 3 (x − y) δij . 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. ψ j (y)] = (i/x + M )ij i∆1 (x − y). 4 Π0 Πi δL = 0.
−λ(∂ρ Aρ ). When we would have started with commutation relations for the ﬁeld ψ and the canonical momentum. we would have obtained [ψi (x).

It gives
3
k µ ǫ(λ) (k)c(k. Πν (y)] = 0. where Q is the conserved quantity in Eq. rj ]P .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. λ′ )] = −g λλ 2Ek (2π)3 δ 3 (k − k′ ). and are equivalent with [c(k. ˙ ˙
. i. The quantized ﬁeld is expanded as Aµ (x) = d3 k (2π)3 2Ek
(λ) 3
ǫ(λ) (k) c(k.e. for which it is suﬃcient that ∂µ Aµ |A = 0. (7. Aν (y)] = [Πµ (x). j = 1. where Aµ + + is the part of the vector ﬁeld containing the annihilation operators. containing a time-like photon. The canonical equal time commutation relations are [Aµ (x). pj ]P . 0. (b) Proof for a quantity O(p. a longitudinal photon and two transverse photons. 7. These problems are solved by the Lorentz constraint between physical states given above. q) that dO/dλ = [O. λ)c(k.
′
(7.
(7. Πν (y)]x0 =y0 = i gµν δ 3 (x − y).95)
Exercises
Exercise 7. Q]P . k 3 ) this reads |k 3 |c(0) (k) − k 3 c(3) (k) |A = 0 c(0) (k) ∓ c(3) (k) |A = 0. λ)|A = 0. 2. 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. 6.91)
(7. λ)e−i k·x + c† (k.93)
which does exhibit problems with the time-like photon. 0)c(k. (7. λ) − c† (k. 0. what are [ri . i.Quantization of ﬁelds
63
for physical states |A and |B . In fact the commutation relations imply ˙ [Aµ (x). c† (k ′ . Aν (y)]x0 =y0 = i gµν δ 3 (x − y).90)
where Πν = F ν0 − λ g ν0 (∂ρ Aρ ) and we have furthermore [Aµ (x). and [ℓi .3) satisfy the conditions of a Lie algebra. λ).e.94)
Choosing k µ = (|k 3 |. 0) .89)
with four independent vectors ǫµ .1 (optional)
If you haven’t seen Poisson brackets in classical mechanics you should (a) Proof that Poisson brackets (Eq. [ℓi . λ)ei k·x . pj ]P . µ
λ=0
(7. one time-like photon by itself is not allowed! This solves the problems with the normalization and the negative energies. (d) Use Poisson brackets to calculate q and p. µ
λ=0
(7. (c) Calculate the Poisson brackets for the coordinates x and the standard conserved quantities p and ℓ.11. which reproduce the Hamilton equations.

f (Λk) = f (k) for Lorentz transformations Λ. (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. Pµ ] = i∂µ φ(x). Pµ ] for the (real) scalar ﬁeld using the expressions for the ﬁelds and momentum operator in terms of creation and annihilation operators. It is invariant. (2 + M 2 ) ∆inhom (x) = −δ 4 (x). Give also the expression for ∆C .
. e−i a requires [φ(x). ∆hom (Λx) = ∆hom (x) if f is an invariant function. ∆R = 1 (2π)4 d4 k e−ikx k2 1 − M 2 + iǫk 0
(with the implicit prescription to take ǫ → 0 after integration).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. (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.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. (2 + M 2 ) ∆hom (x) = 0. we can write for ∆R an integral where the k 0 integration remains along the real axis.3
(a) Show that the correct implimentation of symmetries in the Hilbert space.Quantization of ﬁelds
64
Exercise 7. (b) Check the above commutation relation [φ(x). (Similarly one can show that for the advanced Green’s function ∆A (x) = 0 if x0 > 0). (e) Show that ∆R (x) = 0 if x0 < 0. Is the subtraction of ’zero point’ contributions essential in this check?
µ
Pµ
φ(x) ei a
µ
Pµ
= φ(x − a)
Exercise 7.

. f (k) ∝ θ(k 0 ). In that case one actually needs discontinuous functions f (k) such as the step function. (h) Show that the causal Green’s function ∆C (x) = θ(x0 ) ∆+ (x) − θ(−x0 ) ∆− (x). (g) Show that the homogeneous solution ∆ in (f) satisﬁes ∆(0.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). x) = 0 and argue that one can use Lorentz invariance to show that ∆(x) = 0 for x2 < 0. (i) (Optional) Give similar expressions for the other Green’s functions under (d) in terms of the homogeneous solutions under (f). or ǫ(k 0 ) = θ(k 0 ) − θ(−k 0 ).

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 parity operator transforms
We will consider the transformation properties for a fermion ﬁeld ψ(x). parity (P). (iγ µ ∂µ − M ) ψ(x) = 0. starting with the Dirac equation for ψ(x). T. ˜ (8.1: The behavior of classical quantities under P. r) −→ xµ ≡ xµ = (t. x
(8.1
Parity
xµ = (t. We can determine A.2)
where ηP is the intrinsic parity of the ﬁeld and A is a 4 × 4 matrix acting in the spinor space. time reversal (T) and charge conjugation (C). Both ψ p and ψ satisfy the Dirac equation. Table 8.
8.Chapter 8
Discrete symmetries
In this chapter we discuss the discrete symmetries. The consequences of P. T and C for classical quantities is shown in the table 1. −r). writing
−1 ψ(x) −→ Pop ψ(x)Pop = ηP Aψ(˜) ≡ ψ p (x).

λ) P −1 = −η ∗ d(k. For the scalar ﬁeld we have seen
−1 φ(x) −→ Pop φ(x) Pop = ηP φ(˜). γ x
(8. x Therefore γ0 ψ(˜) is again a solution of the Dirac equation and we have x ψ p (x) = γ0 ψ(˜). x
(8.
(i˜ µ ∂µ − M ) ψ(˜) = 0. (2π)3 2Ek (8.4)
(8. −λ) eik·x 3 2E (2π) ˜ k d3 k ˜ ˜ ηP b(k. x
The latter behavior of the vector ﬁeld will be discussed further below.7)
=
λ
d k x x ηP b(k. λ) e−ik·˜ + d† (k. ˜ Pop d(k. λ)v(k.3)
(8. λ)γ0 v(k. −λ) eik·x (2π)3 2Ek ˜ d3 k ˜ ˜ ˜ ˜ ηP b(k. −λ). λ)u(k. 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. In the same way as the Fermion ﬁeld.10)
i. m) γ0 v(k. −λ). λ) eik·˜ (2π)3 2Ek d3 k ˜ ˜ ˜ ˜ ηP b(k.9) (8. λ) e−ik·x − d† (k. λ) eik·x . ˜ −v(k. one can also consider the scalar ﬁeld and vector ﬁelds. λ)γ0 u(k.6)
(check this for the standard representation. if helicity λ is used instead of the z-component of the spin m.
iγ µ† ∂µ − M ψ(˜) = 0. m).e.13) ψ(x) −→ ψ c (x) = ηC C ψ (x). x µ (iγ ∂µ − M ) γ0 ψ(˜) = 0.12)
and for the vector ﬁeld
−1 Aµ (x) −→ Pop Aµ (x) Pop = −Aµ (˜).8)
3
=
λ
=
λ
=
λ
From this one sees immediately that
−1 ˜ Pop b(k. λ)u(k. λ)v(k. m) = = ˜ u(k. −λ)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) e−ik·x − d† (k.11) (8. −λ) e−ik·x − d† (k. −λ)v(k. op P
(8. x It is straightforward to apply this to the explicit ﬁeld operator ψ(x) using γ0 u(k. λ) Pop = ηP b(k.
.5) (8.
8.2
Charge conjugation
We have already seen the particle-antiparticle symmetry with under what we will call charge conjugation the behavior T (8. the above operation reverses the sign of λ). m).

21)
where A is a 4 × 4 matrix acting in the spinor space. λ). γ x
i(γ5 C)−1 γ µ γ5 C∂µ − M ψ(−˜) = 0. λ).Discrete symmetries
68
the latter being also a solution of the Dirac equation. x
iγ µT ∂µ − M ψ(−˜) = 0. Norm conservation requires Top to be anti-unitary2 .15)
(where one must be aware of the choice of spinors made in the expansion. r) −→ −˜µ ≡ −xµ = (−t.16)
d k ηC d(k. The (time-reversed) Dirac equation becomes. we start with the complex conjugated x Dirac equation for ψ. λ) e−ik·x + b† (k. x x
(8.18) (8. ˜ x −(iγ µ )∗ ∂µ − M ψ(−˜) = 0.
. λ)C v T (k. λ) e−ik·x + b† (k. x
(8.17)
3
=
λ
This shows that
−1 Cop b(k. it is antilinear1 . λ)v(k. m). λ) Cop = ηC d(k. For a quantized ﬁeld one has −1 ∗ −1 Top fk (x)bk Top = fk (x)Top bk Top . λ) eik·x ¯ ¯ (2π)3 2Ek d3 k ηC d(k. λ)u(k. The same relations hold for helicity states. antilinear operator is anti-unitary if A† = A−1 . ((iγ µ )∗ ∂µ − M ) ψ(x) = 0. m).
(i˜ µ∗ ∂µ − M ) ψ(−˜) = 0.22)
= Aψ|Aφ
∗
is antilinear if A(λ|φ + µ|ψ ) = λ∗ A|φ + µ∗ A|ψ .3
Time reversal
xµ = (t. Therefore ψ c (x) = =
λ −1 ¯ Cop ψ(x) Cop = ηC C ψ T (x)
(8. One has Aφ|Aψ = φ|ψ
∗
= ψ|A† Aφ = ψ|φ . i. The action on the spinors (using C = iγ 2 γ 0 = −iρ1 σ 2 in standard representation) gives C uT (k. u(k.e. to ﬁnd ψ t (−˜) that is a solution of the Dirac equation. (8.20)
The time reversal operator transforms
We will again consider the transformation properties for a fermion ﬁeld ψ(x). as discussed in section 4).
8. Top |φ = φt | = (|φt )∗ . m) ¯ = = v(k. x (8. λ)
−1 Cop
=
∗ ηC
b(k. As time reversal will transform ’bra’ into ’ket’. m) ¯ C v T (k. (2π)3 2Ek (8. λ) eik·x . x µ (iγ ∂µ − M ) γ5 Cψ(−˜) = 0.
(8. λ)C uT (k.19)
Cop d(k. writing
−1 ψ(x) −→ Top ψ(x)Top = ηT Aψ(−˜) ≡ ψ t (−˜).14) (8. r). x
1A 2 An
−iC −1 γ µ C∂µ − M ψ(−˜) = 0.

λ)v ∗ (k.29)
In table 2 the behavior of particle states under the various transformations has been summarized. λ). op T
(8. Therefore one has A|k = A|b† (k)|0 = A|T † T b† (k) T † T |0 ∗ = At |b† (k)|0 ∗ ˜ t = k|At .30) (8. λ |¯. λ) eik·x + d† (k. −λ a T a. λ) T −1 = η ∗ d(k.28) (8.33) (8. λ) eik·x + d† (k. −p. λ)iγ5 Cu(k.4
Bi-linear combinations
In quantities such as currents and lagrangians often bilinear combinations of spinor ﬁelds are encountered. λ) e−ik·x (2π)3 2Ek (8. λ) Top = ηT b(k. Since there are 16 independent 4 × 4 matrices. λ) e−ik·x 3 2E (2π) k ˜ d3 k ˜ ˜ ˜ ˜ ηT b(k. λ a P |a. λ). λ)u∗ (k.27)
3
=
λ
=
λ
=
λ
From this one obtains
−1 ˜ Top b(k.31) (8. there are 16 independent of these bilinear combinations.
(8.32) (8.23) 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. λ a |a. p. Note that applying an anti-unitary transformation such as Top one must take for the matrix element ˜ the complex conjugate.26)
d k x x ηT b(k. λ)u∗ (k. λ) eik·˜ + d† (k. λ| ¯ C |¯. λ)v ∗ (k. 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(−˜). p. The result is = =
λ
(8.25)
(check this for the standard representation). λ)u∗ (k.24) (8.34)
. x (8.2: The transformation properties of physical states for particles (a) and antiparticles (¯). λ| a. λ). ˜ = 0|b(k)|A
8. −λ |¯. λ). λ)v ∗ (k. ∗ ˜ v (k. λ) e−ik·˜ (2π)3 2Ek d3 k ˜ ˜ ˜ ˜ ηT b(k.Discrete symmetries
69
Table 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. ˜ Top d(k. −p. −p. p. p. a state |a. λ) ψ t (x)
−1 Top ψ(x) Top = i ηT γ5 Cψ(−˜) x
= =
˜ u∗ (k. λ)iγ5 Cv(k. λ) e−ik·x 3 2E (2π) k d3 k ˜ ˜ ηT b(k. λ) i γ5 C v(k. −p. λ) eik·x + d† (k.

Discrete symmetries
70
Table 8. γ ν ]. for the vector current V µ (x). Note that the lagrangian density L (x) → L (−x) under Θ. s . This implies p′ |V µ (x)|p = e−i (p−p )·x p′ |V µ (0)|p . momenta (pµ or p′µ ) or constant tensors (ǫµνρσ or gµν ). in this case only q 2 = (p′ − p)2 .35)
The x-dependence can be accounted for straightforwardly using translation invariance.. C. called form factors. p′ .41) 2M 2M
. (q 2 ). 2M (8. (8. C. As an example consider the vector current for a point fermion.38)
(8.40)
where the coeﬃcients are functions of invariants constructed out of the momenta. The 16 combinations of Dirac matrices appearing above are linearly independent. V µ (x) = ei Pop ·x V µ (0)e−i Pop ·x . γ5 γ µ or γ5 ). σ µν . s′ |V µ (x)|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 . (q 2 ).36)
(8. p)u(p) e−i (p−p )·x . e.
8. In many applications the expectation values of currents are needed. and T. ¯ V µ (x) =: ψ(x)γ µ ψ(x) :. p′ |V µ (0)|p = u(p′ )γ µ u(p). ¯
′ ′
(8. For instance for nucleons one has Γµ (p′ .5
Form factors
Currents play an important role in ﬁeld theory. F..3: The behavior of the independent bilinear combinations of fermi ﬁelds under P. 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 ). p′ |V µ (x)|p = u(p′ )Γµ (p′ . Applying the results from the previous sections it is straightforward to determine the behavior of the combinations under P. as well as under the combined operation Θ = PCT (see Table 3). 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). p) can be built from any combination of Dirac matrices (1. γ µ . F.g. (8.37)
(8. of which the expectation value between momentum eigenstates can be simply found.39)
where Γµ (p′ . p) = γµ F1 (q 2 ) + iσµν q ν F2 (q 2 ). ¯ In general the expectation value between momentum states can be more complicated.

p)u(˜) ˜ x p ¯ p p ˜ p
(8.42)
where q = p′ − p. γ5 σµν = (1/2)ǫµνρσ γ ρ γ σ . C and T invariance. p′ )u(p′ )) u =
∗ †
u† (p)γ0 Γµ (p. p)u(p) ¯ = p|V µ (0)|p′ ∗ = (¯(p)Γµ (p. Therefore hermiticity implies for the structure of Γµ (p′ . p)u∗ (˜) up p µ p ˜
= ei q·x u(p′ ) (iγ5 C)Γ∗ (˜′ .Discrete symmetries
71
but that are eliminated because of relations between Dirac matrices. p)γ0 u(p). or relations based on hermiticity of the operator or P . or relations that follow from the equations of motion. p′ )γ0 u(p). the Gordon decomposition u(p′)γ µ u(p) = ¯ 1 u(p′ ) [(p′ + p)µ + iσ µν qν ] u(p). p) = γ0 Γµ (˜′ . ¯ µ p ˜
Therefore time reversal invariance implies for the structure of Γµ (p′ . p) that Γµ (p′ . p)γ0 .g. p)(i γ5 C). p)u(p) ¯
† † = p′ |Pop Pop V µ (x)Pop Pop |p ˜x = p′ |Vµ (˜)|˜ = ei q ·˜ u(˜′ )Γµ (˜′ . ¯ 2M (8. ¯ p ˜
Therefore parity invariance implies for the structure of Γµ (p′ .45)
Exercises
Exercise 8. e. • Hermiticity p′ |V µ (0)|p = p′ |V µ† (0)|p = u(p′ )Γµ (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 ) +
′
. p) = (i γ5 C)Γ∗ (˜′ . • Parity p′ |V µ (x)|p = ei q·x u(p′ )Γµ (p′ . p) that Γµ (p′ . p′ )u(p′ )
= u† (p′ )Γµ† (p.g. p ˜ • Time reversal p′ |V µ (x)|p = ei q·x u(p′ )Γµ (p′ . p)(iγ5 C)u(p). e. p′ )γ0 = Γµ (p′ .43)
= ei q·x u(p′ )γ0 Γµ (˜′ . p)u(p) ¯
† † = p′ |Top Top V µ (x)Top Top |p ∗
(8. p) that γ0 Γµ† (p.44)
= p′ |Vµ (−˜)|˜ ˜ x p
∗
˜x = ei q ·˜ [¯(˜′ )]∗ Γ∗ (˜′ . p). µ p ˜ (8.

−|q|/2). p′ = (E.3
(a) Show that because of parity conservation no term of the form γ5 can appear in the current matrix element. |q|). (b) Show that such a term is also not allowed by time-reversal symmetry. σµν q ν F3 (q 2 ) 2M
. GE GM = F1 − Q2 F2 .
Exercise 8. (c) Show that if a term of this form would exist. 0.
Exercise 8. 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). 0. 0. p = (E.
(b) Show that in the interaction with the electromagnetic ﬁeld. 0.2
(a) Calculate the current matrix element (using the explicit fermion spinors) in the Breit-frame in which q = (0. 4M 2 = F1 + F2 . (c) Show that hermiticity of the current requires that F1 en F2 are real. 0.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 . 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 .

5) (9. t′ .4) ∂ U (t. i ∂ |ψS ∂t ∂ i AS ∂t = H |ψS . t0 ) = H U (t.2)
(ii) Heisenberg picture. |ψH (t) = |ψH .1)
The time evolution from t0 → t of a quantum mechanical system is generated by the Hamiltonian. |ψS (t) = U (t. AH (t) = U −1 (t. i ∂ |ψH ∂t ∂ i AH ∂t ≡ = 0. t0 ) ∂t (9. Consider the two (time-independent) Heisenberg states: |q. (9. (9. t0 ) = e−i(t−t0 )H or i Two situations can be distinguished: (i) Schr¨dinger picture. t′ ≡ q ′ |q ′ .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.
(9. ≡ 0. t0 ).
QH (t)|q. [AH .1 Time evolution as path integral
U (t. AS (t) = AS and the states o are time dependent. t ≡ q|q. in which the states are time-independent. in which the operators are time-independent. t
′
QH (t′ )|q ′ . and the operators are time-dependent.Chapter 9
Path integrals and quantum mechanics
9. t0 ) AH (t0 ) U (t.
(9. t . H].7)
73
. t0 )|ψS (t0 ) .3) (9. t
′
|q .

tj+1 |qj . Q) expressed in terms of the operators P and Q.13)
At least for a simple hamiltonian such as consisting of a kinetic energy term and a local potential.qj ) e 2π dpj exp (i ∆τ [pj qj − H(pj .8)
Choose t as the starting point with |q = |q.
n
|q (9.14)
. Then qj+1 . B]. H(P. use the Campbell-Baker-Hausdorﬀ formula. 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 . qj )]) .qj )∆τ e e 2π dpj i pj (qj+1 −qj )−i ∆τ H(pj ..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. 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 .12)
where the transformation between coordinate and momentum space involves q|p = ei p. eA eB = eC with C = A+B+ (9. t and QH (t) = QS and study the evolution of the system by calculating the quantum mechanical overlap amplitude q ′ . tj = = =
1 For
dpj i pj (qj+1 −qj ) −i H(pj . B] + .q .Path integrals and quantum mechanics
74
and Schr¨dinger states o |q |q ′ QS |q ≡ q|q .11) (9. 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 ′ . 2π
(9. t′ |q. |q1 q1 |e−i H∆τ |q . t = q ′ |e−i H(t −t) |q . [A. (9. which is a hamiltonian function in which the operators are replaced by real-numbered variables. . 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|. .
(9. QS |q ′ ≡ q ′ |q ′ . t′ |q.10)
dqn−1 q ′ |e−i H∆τ |qn−1 qn−1 | . 2π (9. B]] + [[A. 2 12 12
(9.
′
(9. . .16)
this. Combining the two terms gives qj+1 .
The purpose of this is to calculate the evolution for an inﬁnitesimal time interval. The hamiltonian is an operator H = H(P. qj ).. tj+1 |qj .

˙
where Dq and Dp indicate functional integrals.
F3 [α. α(x) → αx and F [α] = F (αx ) changes into a multivariable function. Before proceeding we also give the straightforward extension to more than one degree of freedom.2
Functional integrals
In this section I want to give a fairly heuristic discussion of functional integrals. . . α∗ ] = exp (−α∗ K α) ≡ exp −
The kernels K(x. including diﬀerential operators. dx dy α(x) K(x.
′ ′ q1 . pN . while Dα F [α] =
x
dαx N
F (αx )
(9. etc. .e. qN ) ˙
. . q)] ˙ (9. What one is after is the meaning of Dα F [α]. . t N
(9. q)] . Examples are F1 [α] = α ≡ F2 [α] = α2 ≡ dx α(x). F [α] ∈ R. Two working approaches for functional integration are the following: (i) the heuristic division of the space on which α acts into cells. .17)
=
Dq D
dτ [pq − H(p. . dx dy α∗ (x) K(x. β] = α K β ≡ 1 F4 [α] = exp − α2 2
F5 [α.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.e. i. 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.Path integrals and quantum mechanics
75
or for the full interval q ′ . . q1 . t =
τ
dq(τ ) dp(τ ) exp i 2π p exp i 2π
t′ t
t′ t
∆τ [pq − H(p. (9.
9. qN . dx α2 (x).20)
becomes a multidimensional integral. t′ |q. ≡ exp − 1 2 dx α2 (x) . t′ |q1 . .
. qN . y) β(y). . y) in the above examples can be (hermitean) operators acting on the functions. . y) α(y) .18)
t
′
=
n=1
Dqn D
pn 2π
exp i
t
dτ
n
pn qn − H(p1 . . i.

2 (9.25)
(see Exercises). the following integrals can be derived for a real symmetric or complex hermitean kernel K.23)
Having deﬁned the Gaussian integral.Path integrals and quantum mechanics
76
(ii) Write α in terms of a sum of orthonormal basis functions. Note that procedure (i) is an example of the more general procedure under (ii).21)
again a multidimensional integral.24) (9.26)
As an important application of translation invariance we mention the identities 1 Dα exp − α2 − ωα 2 = exp 1 2 ω . In both cases N is an appropriately choosen normalization constant in such a way that the integral is ﬁnite. (9. 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.n fm (x) Kmn fn (y) and consider F [α] = F (αn ) as a multivariable function. y) = ∗ m. α(x) = n αn fn and K(x. A useful property of functional integration is the translation invariance. 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. 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. (9.22)
=
x
The last equality is obtained by deﬁning the right measure (normalization N ) in the integration.n
dx fm (x)fn (x)
=
n
=
n
dαn 1 exp − α2 N 2 n √ 2π ≡ 1. Dα F [α] = Dα F [α + β].27) (9. Physical answers will usually come out as the ratio of two functional integrals and are thus independent of the chosen measure.28)
Dα Dα∗ exp (−α∗ α − α∗ ω − ω ∗ α) = exp (ω ∗ ω) . N
(9.
. with Dα F [α] =
n
dαn N
F (αn )
(9.

products of the same pair appear. α(y) = δ(x − y). −→ fn (x) δα(x) ∂αn ∂αm n Note that δ(x − y) =
n ∗ fn (x) fn (y). θ2 = 0. (9.29)
(9.
n θn fn (x)
dx θ∗ (x) θ(x) . αy = δxy . dx dy θ∗ (x) K(x.30) (9. αn = δmn . which vanish.34) (9. F [θ.32) (9. the Gaussian-type functionals. y) β(y) δα(x) δ 1 1 exp − α2 = −α(x) exp − α2 .
(9.33) (9. It is now easy to check that Dθ∗ Dθ exp (−θ∗ K θ) = det K. θ∗ ] = exp (−θ∗ θ) ≡ exp − F [θ.Path integrals and quantum mechanics
77
Functional diﬀerentiation is deﬁned as δ . In principle the deﬁnitions of functionals is the same. however.e. y) θ(y) . e. we need to consider Grassmann-valued functions involving anticommuting Grassmann variables. In the expansion of the exponential (from second to third line). δα(x) or in discretized form δ 1 ∂ ∂ with .38)
.31)
δ δ α = dy α(y) = dy δ(x − y) = 1.35)
For applications to fermion ﬁelds. θη = −ηθ. Examples of functional diﬀerentiation are
(9. δα(x) 2 2 δ exp (−α β) = −β(x) exp (−α β) . −→ δα(x) ∆x ∂αx ∂αx δ ∂ ∂ ∗ with .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. δα(x)
(9. θ∗ ] = exp (−θ∗ K θ) ≡ exp − Integration for Grassmann variables is deﬁned as dθ 1 = 0 and dθ θ = 1.
(9.36) 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. i. δα(x) δα(x) δ αKβ = dy K(x.g.

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

Path integrals and quantum mechanics
source experiment detector
81
T L
t L+Jq
t’ L
T’
Figure 9. q)] ˙ (9.
φn (q. q) + J(t) · q.
9. T ′ |q ′ . T =
n
q. a set |n of physical eigenstates. Consider. L(q.62)
p Dq D exp i 2π
t
dτ [pq − H(p. t is related to the Schr¨dinger state |q by |q. T ′ |Q.e. furthermore. the creation of an electron (think of a radio-tube.5
The generating functional for time ordered products
By introducing a source-term. The Heisenberg state |q. q)] . We can again rewrite the time-ordered products as functional integrals as derived in the previous section.1: Physical picture of vacuum to vacuum amplitude in presence of a source term This is not a surprising result. q . making the electron) and the absorption of an electron (think of a detector). Before and after these processes there is only the vacuum or ground-state |0 .66)
. t q.
Considering the source to be present between times t and t′ . t Q′ . it is possible to switch on an interaction. for the physical states o q. t |T exp −i p exp −i 2π
t′
′
′
t′ t
dτ HI (Q(τ )) |q.s. t′ q ′ . i. ˙ ˙ physically pictured as. T . T < t < t′ < T ′ . t′ |q. t|n = q|n e−iEn t with q|n the time-independent wave function. t |q. t
t′
V =0
Dq D
dτ HI (q) exp i
t t′ t
dτ [pq − H0 (p. t = ei Ht |q . t′ |q.
for g. for example x|n x|n = φn (x) = ei kn x = e
−ω 2 x2 /2
(9. T with q. harmonic oscillator.65)
J
dq ′ dq Q′ . one has q ′ . t|Q.64) (9. n
(9. q) + J(τ )q] . t) φ∗ (Q) e+i En T . i. say. q) → L(q.e. ˙
The combination of exponentials in the last step is allowed because we are simply dealing with classical quantities (not operators!). t|n n|Q. t|Q. t = =
′ ′ V
= q .63)
for plane waves. ˙
J
(9. which in turn are embedded between an early time T and a future time T ′ . T =
n J t′
= =
Dq exp i
t
dτ [L(q.

. such that t′ → −i∞ t → i∞ ↔ ↔ ˜ t → −∞.70) for the generating functional is in fact ill-deﬁned.
J
(9.6
Euclidean formulation
The above expression (Eq. ˜ t′ → ∞.
(9. in which case eiEn T → e−En ·∞ and the term eiE0 T is dominant. thus
T →i∞
lim
q. The generating functional precisely represents the vacuum to vacuum amplitude from initial (’in’) to ﬁnal(’out’) situation in the presence of a source. δJ(tn ) hence the name generating functional.Path integrals and quantum mechanics
t
82
T
t
t’
T’
Figure 9.69)
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 ′ ). t 0 0out |0in J .70)
one immediately ﬁnds δ n Z[J] δJ(t1 ) . . T ′ |Q. t|Q.71)
9. Since we have (neglecting multiplicative factors). T → i ∞ and T ′ → −i ∞. .
J=0
(9. T = φ0 (q. eiE0 T φ∗ (Q) 0
We deﬁne the generating functional Z[J] as Z[J] = Q′ . . T J lim −i E0 (T ′ −T ) φ (Q′ ) φ∗ (Q) 0 0 T ′ → −i∞ e T → i∞ dq ′ dq φ∗ (q ′ . t′ |q.
dq dτ
+ J(τ )q(τ )
.
.68) (9. 9.2: Analytic continuation of boundaries T and T ′ We can project out the groundstate by an analytic continuation in the time.
T′
Z[J] =
lim T ′ → −i∞ T → i∞
Dq exp i
T
dτ L q.67)
= =
φ0 (q. t′ ) q ′ . t)
(9. Q(tn )|0 . = (i)n 0|T Q(t1 ) . t). Better is the use ˜ of imaginary time t = −it. The importance of Z[J] is that the time ordered product of operators can be obtained from it.

(9. + V (q). The diﬀerentiations in Z[J] and ZE [J] are related.72) (9. dq ˜ dt = 1 2
dq ˜ dt dq ˜ dt
2
− V (q). when L q. det K
Exercise 9. ˜ dt
(9.2
Check the examples of functional derivation for real-. δJ(tn ) J = 0 ˜ t = it
(9. δJ(tn ) = (i)n
J=0
1 δ n ZE [J] .1
Convince yourself by choosing an appropriate (orthonormal) set of functions that 1 Dα exp − α K α 2 for real-valued functions. . ˜ ˜ ZE [J] δJ(t1 ) . In all of these cases the determinant of the (hermitean) operator K is deﬁned as det K ≡ p Kp .g. y) = ∗ m. Dα Dα∗ exp (−α∗ K α) = for complex-valued functions and Dθ∗ Dθ exp (−θ∗ K θ) = det K. det K 1 . ZE [J] = = where LE q. . . where Kp are the eigenvalues of the matrix Kmn in K(x. As discussed in the previous section the interesting quantities are obtained from functional diﬀerentiation with respect to the sources. τ
dq ˜ dt
≡ −L q.
.and Grassmann-valued functions. i τ
−∞ ∞ −∞
dq d˜ τ
+ J(˜)q(˜) τ τ dq d˜ τ − J(˜)q(˜) τ τ .n fm (x) Kmn fn (y).Path integrals and quantum mechanics
83
In terms of the imaginary time we can write the Euclidean generating functional.76)
which is a positive deﬁnite quantity.73)
d˜ LE q. ensuring convergence for the functional integral ZE [J]. for Grassmann valued functions.
(9.
Exercises
Exercise 9.74)
− V (q)
2
(9. dq dt = 1 2 1 2 dq dt
2
Dq exp Dq exp −
∞
d˜ L q. i
dq .77)
where the expressions have been divided by Z[J] and ZE [J] in order to get rid of dependence on multiplicative factors. δ n Z[J] 1 Z[J] δJ(t1 ) .75)
= − LE q. e. 1 = √ . . complex.

80)
Dα∗ Dα exp (−α∗ K α − α∗ ω − ω ∗ α) =
1 exp ω ∗ K −1 ω . where dy K(x.Path integrals and quantum mechanics
84
Exercise 9.78) (9.
. det K
Dθ∗ Dθ exp (−θ∗ K θ − θ∗ η − η ∗ θ) = det K exp η ∗ K −1 η .79) (9. z) = δ(x − z). 2 (9.3
Prove the following relations 1 Dα exp − α K α − αω 2 1 = √ exp det K 1 ω K −1 ω . y) K −1 (y.

t) is considered as a set of quantum operators φx (t) in the Heisenberg picture and
t′
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.3)
The Euclidean formulation is as before. φ(x. t′ |φ. Π) (10. xn ). . . (10.Chapter 10
Feynman diagrams for scattering amplitudes
10.4)
J=0
. δJ(xn ) 85 ≡ G(n) (x1 .2)
=
∞ −∞
d4 xE [LE (φ. i. x′ . .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. t
= =
Dφ DΠ exp i
t σ′
˙ dτ d3 x Π(x)φ(x) − H (φ. ∂µ φ) − Jφ]
(10. 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 ) . . φ(xn )|0 = (−i)n δ n Z[J] δJ(x1 ) . . x. .e. ∂µ φ) + Jφ]
(10.

2 2
→
(10.10)
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.Feynman diagrams for scattering amplitudes
86
for which we introduce the picture
x1
G(n) (x1 . with Z0 [0] = 1. . . . As discussed in the previous section (exercise 9. = 0 2 2 2 k2 − M 2 (k ) − E 2 (k ) − k − M (10. . J(xn ). xn ) J(x1 ) . . .7)
where we made the choice Z0 [0] = 1 and ∂µ ∂ µ + M 2 ∆F (x − y) = −δ 4 (x − y). ∆F (the socalled Feynman propagator) is the Green’s function of the Klein-Gordon equation. which can be written as ∆(x) = ˜ with (k 2 − M 2 )∆(k) = 1 or ˜ ∆(k) = 1 1 1 = 0 2 .5)
Z0 [J] for the free scalar ﬁeld
For the (free) scalar ﬁeld lagrangian L =
←
1 1 ∂µ φ∂ µ φ − M 2 φ2 .13)
. xn ) = x2
xn
. (2π)4 (k 0 + iǫ)2 − E 2 (10. Furthermore. (10.6)
the generating functional is (using ∂µ = − ∂µ ) given by Z0 [J] = Dφ exp −i
where Kxy = (∂µ ∂ µ + M 2 ) δ 4 (x − y).
(10. . .3) one ﬁnds 1 Z0 [J] = exp − i 2 d4 x d4 y J(x) ∆F (x − y) J(y) ≡ exp (i W0 [J]) .4) one distinguishes the retarded Green’s function ∆R (x) = d4 k e−i k·x . d4 xn G(n) (x1 . . (10. .8)
1 d4 x φ (∂µ ∂ µ + M 2 ) φ − Jφ 2
K
(10. consider the general solution of ∂µ ∂ µ + M 2 ∆(x) = −δ 4 (x). . .12) d4 k −i k·x ˜ e ∆(k) (2π)4 (10.
These Green functions appear in the expansion Z[J] =
n
in n!
d4 x1 .9)
i.11)
Depending on the path choosen in the complex k 0 plane (see exercise 7.e.

(2π)4 k 2 − M 2
(10.19) φ ∂E ∂E − M 2 φ + Jφ 2 1 d4 xE d4 yE J(x) (−i ∆F (x − y)) J(y) . ∆(x) = − d4 k e−i k·x . φ+ (0)] = (2π)3 2E
are also shown in exercise (7.1 for countours)
µ (∂E )2 − M 2 (i ∆F (x − y)) = −δ 4 (x − y)
0 kE =∞
(10. 10.e. the generating functional can be written 1 µ µ Z0 [J] = Dφ exp d4 xE (10.16)
C
where C is a closed contour in the complex k 0 -plane. 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).20) = exp − 2
where
or (see ﬁg.
. i. the second is to explicitly consider 0|T φ(x1 ) φ(x2 )|0 . (2π)4 (k 0 − iǫ)2 − E 2 (10. the advanced Green’s function ∆A (x) = satisfying ∆A (x. The contours for i ∆(x) i ∆+ (x) i ∆− (x) = [φ(x). (10.22) (10.17)
d4 k e−i k·x . (2π)4 k 2 − M 2 (10. The ﬁrst possibility is to consider the well-deﬁned Euclidean formulation. ∆F = ∆C .Feynman diagrams for scattering amplitudes
87
satisfying ∆R (x.21)
∆F (x)
= =
−i
0 kE =−∞
d4 kE e−i kE ·xE 2 (2π)4 kE + M 2
(10. 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 k 2 − M 2 + iǫ (2π) (10.15) e−i k·x d4 k . The solution of the homogeneous equation ∂µ ∂ µ + M 2 ∆(x) = 0 can also be written as an integral in k-space. (2π)4 k 2 − M 2
where the latter contour can be deformed to the contour for ∆C .23)
k0 =i∞ k0 =−i∞
d4 k e−i k·x . 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).14)
They are solutions of the inhomogeneous equation.4).
µ 1 1 • Firstly. t) = 0 for t > 0. 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 . (10. t) = 0 for t < 0. φ(0)] = i∆+ (x) + i∆− (x) d3 k e−i k·x = [φ+ (x). φ− (0)] = (2π)3 2E d3 k e−i k·x = [φ− (x).18)
Thus we see various solutions.

Neglecting multiplicative factors or equivalently assuming that Z0 [0] = 1. .1: Contour in the k 0 plane for the Feynman propagator in Euclidean and Minkowski space • As the second possibility. δJ(x4 ) 2 J=0 = − [∆F (x1 − x2 ) ∆F (x3 − x4 ) + ∆F (x1 − x3 ) ∆F (x2 − x4 ) + ∆F (x1 − x4 ) ∆F (x2 − x3 )] . 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. φ+ (y)]|0 = θ(x0 − y 0 ) i ∆+ (x − y) − θ(y 0 − x0 ) i ∆− (x − y)
(10.24)
Knowing the explicit form of Z0 [J] it is straightforward to calculate the 4-points Green’s function.
= θ(x0 − y 0 ) 0|[φ+ (x). x2 . we have G0 (x1 . We ﬁnd (see also section 7. φ− (y)]|0 − θ(y 0 − x0 ) 0|[φ− (x). the same result as above.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. or diagrammatically
(4)
(10.25)
x1 x2
x3 = x4 + +
. . x4 ) = 0|T φ(x1 ) φ(x2 ) φ(x3 ) φ(x4 )|0 1 δ4 exp − i J ∆F J = (−i)4 δJ(x1 ) . x3 .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). (4) (2) It should be clear that G0 can be expressed in terms of G0 = i ∆F . because this is the only quantity entering Z0 [J].

Z[J] = = Dφ exp i exp i d4 z LI (10. with propagators (i ∆F ) being connected to vertices (i LI ) according to the above expression for the functional Z[J].27)
When interactions are present. L (φ) = L0 (φ) + LI (φ). the generating functional can be written. x3 .32) (10.e.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) (10.33)
(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 .30)
and in zeroth order G(0.29)
This expression will be the one from which Feynman rules will be derived.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
.
(2)
(0)
(10.31)
G(2. x4 ) = To ﬁrst order in g one has Z (g ) [J] = =
1 0
0
x1
+ +
x2
(10. x2 ) = G0 (x1 .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
0
)
= G0 = 1.g ) (x1 . i.g ) (x1 . x2 . To zeroth order in the coupling one has
0 1 Z (g ) [J] = Z0 [J] = exp − i 2
J ∆F J . x4 ) = G0 (x1 .35)
z
. Consider as an example the interaction LI (φ) = − g 4 φ 4!
in the scalar ﬁeld theory discussed sofar. x3 .34)
Introducing a vertex point to which four propagators are connected. x2 ) = i ∆F (x1 − x2 ) = G(4. (10.
(10. x2 .
(10.Feynman diagrams for scattering amplitudes
89
10. ≡ −i g d4 z (10.

x2 ) =
1
z
x1 x2
+
1 2
x1
z x2
. . . δJ(xn ) . .41)
in terms of socalled connected Green’s functions.g ) (x1 . d4 xn G(n) (x1 . d4 xn G(n) (x1 . . .
J=0
(10.37)
G(2. xn ) J(x1 ) . (10. In the free case we have seen that the exponent of Z[J] contained all essential information.e.g
1
)
=
1 8
z
1 8
. . . . . . J(xn ) c
(10. .
(10. J(xn ). . . xn ) = (−i)n−1 c which are denoted δ n W [J] δJ(x1 ) . . xn ) = x2 c
xn
Speciﬁc examples are G(0) = c G(1) (x1 ) = c G(2) (x1 .38)
The result for the 4-points Green’s function is left as an exercise. 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]. xn ) J(x1 ) . x2 ) = c (vacuum bubble) (tadpole) (connected propagator)
.
(10.39)
We have calculated the function up to ﬁrst order in the coupling constant for φ4 -theroy. . .36)
J
J
from which one obtains G(0. . i. .
(10. Diagrammatically this exponent only contained the two-point function. . .Feynman diagrams for scattering amplitudes
90
one has
J
Z
(g1 )
[J] =
1 8
1 − 4
J
J
1 + 24
J
Z0 [J]. This remains also true for the interacting case. .40)
in−1 n!
d4 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]. Z[J] =
n
in n!
d4 x1 .
(10. did not contain parts that could be written as products of simpler Green’s functions. .42)
x1
G(n) (x1 . which also was the only Green’s function for which the diagram was connected. . G(n) (x1 . . . .

. 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 + + .
+
+i
d4 x1
1 2!
+
1 3!
+i
d4 x1
+
1 2!
i2 2!
d4 x1 d4 x2
+ .Feynman diagrams for scattering amplitudes
91
To see that Z[J] = exp(i W [J]) contains all non-connected diagrams can be performed inductively. (ii) Note that Z[0] = exp(iW [0]) appears as a multiplicative factor. (iv) If δZ[J] = i 0|φ(x)|0 = 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). 1 2!
Z[0] = 1 +
+
+ . 2! 3! +i d4 x1 + i2 2! d4 x1 d4 x2 . 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
.... but is easily illustrated by writing down the ﬁrst few terms.
These can be divided out. in the expansion of which all vacuum blobs are contained. (1) (1) (iii) Gsc = Gc .

. in|S = α. x3 . one ﬁnds for the connected 4-point Green’s function at order g. Also for free ﬁeld (Heisenberg) operators a distinction is made between φin and φout . in|S|p′ . we have W0 [J] = − 1 2 d4 x d4 y J(x) ∆F (x − y) J(y). φout and the interpolating ﬁeld φ(x). Next. Proof: 0. x4 ) = c
1
x3 z
(10. ∂µ ∂ µ + M 2 φin (x) = 0. in = β. in = |α.3
Interactions and the S-matrix
The S-matrix
The S-matrix transforms initial state free particle states (in-states) |α. pn .g.
x1
G(4. out| and S † |α. out| ⇔ |α. e.
x4
10. p′ . in = p. translation invariance).46)
The properties of the S-matrix are (1) The vacuum is invariant or |S00 | = 1.45)
x2
as the only surviving diagram (see exercises). out . (2) The one-particle state is invariant (conservation of energy and momentum. in|p′ . out| = eiϕ0 0. in|SS † |α. Sβα ≡ β. .
(10.Feynman diagrams for scattering amplitudes
92
i. out = |p′ .
(2)
(10. thus β. . out . out|p′ . in|S = β. . considered explicitly.44)
J=0
For the interacting theory.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 ). (10.47)
. out (suppressing except momenta all other 1 m quantum numbers). For the free scalar theory. out|α. 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 . this will be translated to the action on ﬁelds. In line with the consideration of the generating functional representing the vacuum to vacuum amplitude we consider ﬁelds φin . in = p. in = p. then (n) (n) Gsc = Gc for n ≤ 3. p. out|p′ . in ⇔ β. . out = p|p′ . Proof: α. the vacuum expectation value of the ﬁeld φ(x) is zero. (10. in = S|α. in|S = 0.g ) (x1 . in into ﬁnal state particle states (out-states) |β. implying the absence of tadpoles. in| (choose ϕ0 = 0).e. . (3) S is unitary (it conserves the scalar product from initial to ﬁnal state). x2 . out|α. in = |p1 . out = δαβ ↔ SS † = 1.

(10.62) is then
T′ UT [J] T′
= T exp i
T
d4 x J(x)φ(x) .
(10. φ(y)] = Z [φin (x). e. (2π)3 2E (10. z) φ(z) d4 y d4 z ∆A (x − y) K(y.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).e. i. a) = S.54)
.and out-ﬁelds is: φin (x) = S φout (x)S −1 .
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. a causality condition that can be proven to imply the absence of interactions. in|Sφout Finally we check that as expected S does not spoil Poincar´ invariance. this can actually not be used as it would imply [φ(x).
The relation between S and Z[J]
To establish this relation.g.49)
The ﬁeld φ can be expressed in φin/out using retarded or advanced Green’s functions. as it stands. out|φout = β.
2 ∂µ ∂ µ + M0 φ(x) = J(x). √ Although the above. The convergence therefore must be weakened to t→−∞ √ (10. β. z) φ(z). The time evolution operator (see Eq. φin (y)] = iZ ∆(x − y). 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. which in a later stage (renormalization) will become more important. S is invariant under e Poincar´ transformations: U (Λ. Proof: β.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 ).50)
= φ(x) =
√ Zφin (x) − √ Zφout (x) −
d4 y d4 z ∆R (x − y) K(y. 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.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. implies the strong (operator) convergence φ(x) → Zφin (x). 9. The relation between S-matrix and in. φin (x) = d3 k ∗ a(k) fk (x) + a† (k) fk (x) .52)
where Z is a constant. in|φin S → φin S = S φout (x). the source term Jφ in the lagrangian density is treated in the interaction picture.51) (10.
(10. a)SU −1 (Λ.

y) δ δJ(y) : F [J].
φ− f
(10. [A. express it in terms of advanced and retarded Green’s functions. 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.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). with the use of the Baker-Campbell-Hausdorﬀ formula e−B A e+B = A + [A. B + C]e e .63)
. which is equal to the expression e in which the creation operators are placed left of annihilation operators. one has [A.62)
where F [J] is some (arbitrary) functional. e e ] = [A. eB ] = [A. we can.Feynman diagrams for scattering amplitudes
94
and satisﬁes the property
T ′ δUT [J] t = i UtT [J] φ(x)UT [J]. z) δU [J] δJ(z) δU [J] d4 y d4 z ∆A (x − y) K(y. z)
δU [J] . e = i. δJ(z)
(10. : e
d4 z φ(z) f (z) d4 x φ− (x) f (x) B C B C
(10. δJ(x) ∞ or for U [J] ≡ U−∞ [J]
′
(10.59) (10. C] are c-numbers.57)
i [φin (x). 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. S U [J]] = √ Z
δ S U [J]. Thus applied to the ﬁeld φ(x) = φ+ (x) + φ− (x).56)
Since δU [J]/δJ(x) satisﬁes the same equations as φ(x) and we know the limits. z) d4 y d4 z ∆(x − y)K(y. φ(y)] f (y) e
e
. φ(x). φ(x).61) e
φ+ f
where the normal ordered expression : e φ f : is used. provided [A. Noting that 0| : eA : |0 = 1 it follows that F [J] = 0|S U [J]|0 = 0|U [J]|0 = 0out |0in ∝ Z[J]. just as we did for φ(x).
t i Ut∞ [J] φ(x)U−∞ [J]
(10. B] and [A. B] (for the case that [A.58)
In order to ﬁnd a solution to this equation note that.
(10.60)
e
d4 y φ+ (y) f (y) d4 z φ− (z) f (z) d4 z φ+ (z) f (z)
d4 y [φ(x). B]eB . (10. z) .
: =i
d4 y ∆(x − y) f (y) : e
d4 z φ(z) f (z)
:. B] is a c-number). Thus S U [J] =: exp 1 √ Z d4 x d4 y φin (x) K(x.e. δJ(z)
(10.

i.Feynman diagrams for scattering amplitudes
95
while for J = 0 one has U [0] = 1. . . . |S|p. . .
J=0
(10. xn ). . . (λ) v(p).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. . . pn ). iKx′ . which is determined by the inverse of the operator found in the quadratic term in the lagrangian i. . using that φin (x) = d3 k ∗ a(k) fk (x) + a† (k) fk (x) . . (2π)3 2E
n
(10. . . 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. . y). Explicitly. .e. . d4 x . 1 for scalar case.
k
=
k2
i − M 2 + iǫ
. Next we introduce the Fourier transform (after extracting a momentum conserving delta function coming from translation invariance. i. c
(n)
(10.δ. .. iKx′ = (2x′ + M 2 ) precisely annihilating an external propagator i∆F in the Green’s function. u(p) and v (p) for fermions and ǫµ (p) for vector ﬁelds). . + pn ) G(n) (p1 . p′ .65)
we get the Lehman-Symanzik-Zimmerman (LSZ) reduction formula. 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). .67)
where ∆(p) is the Fourier transform of the (full) propagator ∆F (x). . fp′ (x′ ) . .
n
(2π)4 δ 4 (p1 + . fp (x) . . . . x. Usually we are interested in the part of the S-matrix describing the scattering. . ¯ ¯
and the amputated or one-particle irreducible (1PI) Green’s functions n −i (n) Γ(n) (p1 . ←− − →
× G(n) (x′ .64)
Therefore. . . . . Note that the Green’s function connecting the external point xi with the bubble is annihilated by K(x.2). . . . . pn ) = G (p1 . pn ) =
i=1
d4 xi ei pi ·xi G(n) (x1 .e. . . . They start with the propagator (i∆F (k)) in momentum space. S =: exp 1 √ Z d4 x d4 y φin (x) K(x..+ 1 √ Z
∗ d4 x′ . = . see exercise 10. ∆(pj ) j=1
(10. . y) δ δJ(y) : Z[J] Z[0] .. u(p). . .. .e.) iKx . Sf i = δf i − i (2π)4 δ 4 (Pi − Pf ) Mf i .68)
10. . .66)
where Gc is the connected Green’s function. . .
(10.

At these vertices each line can be assigned a momentum. For the symmetry factor consider the examples (given in G. the factor 2 corresponds to the two-points Green’s function having two identical ends). For the interaction terms in the lagrangian.69) LI (φ) = − φ3 − φ4 . to be precise iLI vertices in momentum space are introduced. 3! 4! The vertices are: = −i f Consider ﬁrst the lowest order self-energy diagram. = −i g
. or calculating full Green’s functions external lines are treated as propagators.27. (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). Veltman. 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.27. which cancels the factor 1/2 in the quadratic piece. Diagrammar) in the case of the interaction terms g f (10. 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 10.66 and 10. vertices and external particle wave functions in that diagram. Note that in calculating amputated Green’s functions external lines are neglected.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. If problems arise go back to the deﬁning expression for the generating function in 10. but overall momentum conservation at a vertex is understood. ’t Hooft and M. 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. = −i g (multiplied with 4! corresponding to the allowed number of permutations of identical particles).

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

g.
. η). in e− e− → e− e− scattering (Møller scattering) the lowest order contribution is
(see next section for e-e-γ vertex). 4 2 − M 2 + iǫ (2π) p and the (oriented) propagator in Feynman diagrams involving fermions is
i p j
(10. 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 . The two-points Green’s function 0|T φ(x) φ(y)|0 contains a fermionic loop contribution.
= + . 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.. e.Feynman diagrams for scattering amplitudes
98
Dirac ﬁelds
For fermions the generating functional is given by Z[η.− i δη(z) i δη(z) d4 x d4 y η(x) iSF (x − y) η(y) . which is the solution of (i/ − M )SF (x) = −δ 4 (x) ∂ (i. − g2 2 dz dz ′ δ2 δ2 δη(z) δη(z) δη(z ′ ) δη(z ′ )
and the quadratic term in Z0 [η.. namely (Rule 5) Feynman diagrams which only diﬀer by exchanging identical fermions in initial or ﬁnal state have a relative minus sign. but the anticommutating properties of Grassmann variables imply some additional minus sign in Feynman diagrams.e. − 1 2 dx dy dx′ dy ′ η(x)S(x − y)η(y) η(x′ )S(x′ − y ′ )η(y ′ ). (10. η]. (Rule 6) Each closed fermion loop gets a sign −1. η] = Dψ Dψ exp i d4 z LI exp − d4 x ψ (i/ − M ) ψ + LI (ψ) + ηψ + ψη ∂ 1 δ 1 δ ..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.
which arises from the quadratic term in exp i d4 z LI .72) (10.71)
= exp i
where iSF is the Feynman propagator for fermions.

The wave functions for fermions are given by ∝ g 2 S(z − z ′ ) S(z ′ − z) = −g 2 iS(z − z ′ ) iS(z ′ − z). 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 . Particular choices of λ are λ = 1 (Feynman propagator or Feynman gauge) and λ = ∞ (Landau gauge).Feynman diagrams for scattering amplitudes
99
The result is which contains an extra minus sign as compared to a bosonic loop. 1 k 2 − M 2 + iǫ λ (k 2 − M 2 + iǫ)(k 2 − λ M 2 + iǫ)
leads to
. + 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. 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 .
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 .
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. writing down the propagators till the beginning of the ¯ line (incoming fermion u or outgoing antifermion v) has been reached. 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. 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. λ 1 ∂ (10.
Vector ﬁelds and Quantum Electrodynamics
As the most important example of vector ﬁelds consider the lagrangian density for quantum electrodynamics (QED).74) L = − Fµν F µν − (∂ · A)2 + ψ(i/ − M )ψ − e ψγ µ ψ Aµ .

s4 )γµ u(p.s3 .s2 ) which can be written as (see also chapter 4) |M |2 = = 1 4s |¯(k ′ . t2
(10. s4 )(−ie)γ ν u(p. 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 ′ .78)
=
2 α2 2 s + u2 + 4t(M 2 + m2 ) − 2(M 2 + m2 )2 . s1 ) u e2 u(p′ .Feynman diagrams for scattering amplitudes
100
10.s2 . where u ¯ L(m) µν = = 1 2 u(k ′ . s3 )(−ie)γ µ u(k.5
Some examples
eµ scattering
The ﬁrst example is the electromagnetic scattering of an electron and a muon. q 4 µν
[using that (¯(k ′ )γµ u(k))∗ = u(k)γµ u(k ′ )].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. s)¯(k.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 ).75)
Note that the q µ q ν term in the photon propagator are irrelevant because the photon couples to a conserved current. s′ ) ¯ u
s. ¯ q2 (10.79)
.s4 ) and averaged over spins in the initial state (1/2 × 1/2 × s1 .s4
e4 (m) µν (M) L L . s3 )γ µ u(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 . s2 )|2 ¯ q2 (10.76)
1 . To lowest order (∝ α = e2 /4π) only one diagram contributes. s)γν u(k ′ . s1 ) ¯ −igµν u(p′ . s′ )γµ u(k.

s1 ) u(k. Basically the same diagram is calculated and the result is the same after the interchange of s ↔ t.2. s1 )|2 ¯ s
1 .
(4) (4)
. s2
(10. (b) Use the deﬁnition of the source-connected Green function Gsc and show that indeed sourceconnected diagrams survive. Similarly. s3 ) ¯ ¯
1 . s2 )γµ u(k.s3 . . 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.s4
=
1 2s ×
v (k ′ . s2 )γµ u(k. s1 )γν v(k ′ . The diagram. s2 ) ¯ ¯ 1 2s u(p. the masses.Feynman diagrams for scattering amplitudes
101
The second example is the annihilation of an electron pair and creation of a muon pair. 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.s2
3 . s3 )γ µ v(p′ . . (c) Do the same for the connected Green function Gc . . This is known as crossing symmetry.
Exercises
Exercise 10.s2 . To lowest order (∝ α = e2 /4π) only one diagram contributes. s4 )γ ν u(p.1
(a) Give in diagrammatic notation the full Green functions G(4) (x1 . for instance Møller scattering (e− e− → e− e− ) and Bhabha scattering (e− e+ → e− e+ ) are related using crossing symmetry. s3 )γ µ v(p′ .80)
Note the similarity in the amplitudes for eµ scattering and e− e+ → µ− µ+ . s4 ) v (p′ . s4 ) u e2 v (k ′ .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 .

. . in the amplitude (Tij = A∗ Aj ). . . xn ). . + pn ). Note that the function G(n) (p1 . . . . . pn ) only has n − 1 independent variables in it.
Exercise 10. xn + a) = G(n) (x1 . Express the contributions in invariants s.Feynman diagrams for scattering amplitudes
102
Exercise 10. . or if we in general would consider all momenta as incoming.
hence we can write
n i=1
d4 xi ei pi ·xi G(n) (x1 . .
which means overall momentum conservation in Green functions in momentum space. for instance. It is of the form −i M = A1 − A2 . . 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). . . .
Exercise 10. . .37 and 10. . + pn ) G(n) (p1 . . . but the incoming and outgoing momenta are identical. Pictorially this implies.
. t and u. .2
(a) Show that the translation properties of the ﬁelds and the vacuum imply G(n) (x1 + a. . . . . . pn ). their sum is zero.3
Show that the combinatorial factors found using the rules given in section 10. |M |2 = T11 + T22 − T12 − T21 . Show that i the amplitude is symmetric under the interchange of t ↔ u. .4
(a) Write down the Feynman diagrams contributing to electron-electron scattering.38. e(p1 ) + e(p2 ) → e(p′ ) + e(p′ ) in lowest order in α.3 reproduce for the diagrams
precisely the combinatorial factors that appear in Eqs 10. xn ) ∝ (2π)4 δ 4 (p1 + . 1 2 (b) Calculate the quadratic pieces and interference terms. (b) Show (by using x1 as shift-varible) that this implies that
n i=1
d4 xi ei pi ·xi G(n) (x1 . xn ) ≡ (2π)4 δ 4 (p1 + . .

. The phase space is determined by the weight factors assigned to each state in the summation or integration over states. pn ). . . s = (p1 + . .Chapter 11
Scattering theory
11.e. The ﬁrst is the CM system.3)
which is useful because the total 4-momentum of the ﬁnal state usually is ﬁxed by overall momentum conservation. p1 .5) √ is referred to as the invariant mass squared.1) = (2π)3 2E (2π)4 (proven in Chapter 2). + n. pb we start with the four vectors pa = (Ea . In that case
cm pa = (Ea .
(11. Here s is the invariant mass of the n-particle system. q). . a a b b the quantity s = P 2 = (pa + pb )2 . and the total momentum four-vector P = pa + pb . . . . It is a useful quantity. (11. pn ) = and the reduced phase space element by dR(s. Its square root. .
(11.
i n
d3 pi . In the CM frame the threshold value for s obviously is
n 2
sthreshold =
i=1
mi
. For two particles. −q). .2)
(11. the 1particle phase space is d3 p d4 p θ(p0 ) (2π)δ(p2 − m2 ).1 kinematics in scattering processes
Phase space
The 1-particle state is denoted |p . for instance for determining the threshold energy for the production of a ﬁnal state 1 + 2 + . (11. . . p) which satisﬁes p2 = E 2 −p2 = m2 .4)
For two particle states |pa . pa ) and pb = (Eb . A physical state has positive energy. To be speciﬁc let us consider two frequently used frames. (2π)3 2Ei i=1
(11. .7)
pb =
cm (Eb . . It is determined by the energy-momentum four vector p = (E. . pn ) = (2π)4 δ 4 (P − pi ) dR(p1 . This is generalized to the multi-particle phase space dR(p1 . i. + pn )2 .6) (11.
103
. pb ) satisfying p2 = m2 and p2 = m2 . . s is for obvious reasons known as the center of mass (CM) energy.

11) (11. 2mb λ(s. p2 ) = = where λ12 denotes λ(s. 0). m2 . In that case
trf pa = (Ea . Prove that |q| =
cm Ea cm Eb
(s − m2 − m2 )2 − 4 m2 m2 a a b b = 4s s + m2 − m2 a b √ = . 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.10)
The function λ(s. which in the speciﬁc case a b also can be expressed as λ(s.12)
pb = (mb . 4s
(11. = 2 s
λ(s. 2 s s − m2 + m2 a b √ . Also in this case one can express the energy and momentum in the invariants.
(11. 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 √ = .9) (11. ptrf ). 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. for instance. 2mb
(11. a
(11. Prove that
trf Ea =
|ptrf | = a
s − m2 − m2 a b . m2 . p1 .16)
. 4π s 4π 8π s 4π
(11.
trf Ea (threshold) =
1 2 mb
(
i
mi )2 − m2 − m2 a b
. p1 . m2 . m2 . 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.14)
One can. m2 ) = 4(pa · pb )2 − 4p2 p2 . m2 ) a b . m2 ). mb and s). m2 ) a b .13) (11.15)
Explicit calculation of the reduced two-body phase space element gives dR(s. m2 .Scattering theory
104
It is straightforward to prove that the unknowns in the particular system can be expressed in the invariants (ma . m2 ) is a function symmetric in its three arguments.8) (11.

27)
with m2 = m2 − p2 = m2 + p2 + p2 . 11. dt = 2 qq ′ d cos θcm and obtain for the two-body phase space element √ q ′ dΩcm λcd dΩcm √ = (11. Of course there are not only 2-particle channels. pd ) = 4π s 4π 8π s 4π dt dt √ √ . . for instance the momenta of incoming (colliding) hadrons and it is useful to use for the produced particle rapidity as a variable.26) = 16π q s 8π λab
Kinematics of inclusive hard scattering processes
In high energy (hard) scattering processes. At high energies. pT = (px . H1 + H2 → h + X.24)
Using the relation between t and cos θcm it is straightforward to express dΩcm in dt. usually is a 2-particle state. In inclusive measurements no particles are detected in the ﬁnal state. 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 . in which one particle is detected. usually many particles are produced. in exclusive measurements all particles are detected. Consider the process a + b → c + d.. (11.25) dR(s. Consider the 1-particle inclusive case. s = (pa + pb )2 = (pc + pd )2 t = (pa − pc ) = (pb − pd ) u = (pa − pd )2 = (pb − pc )2
2 2
(11. where the ﬁrst is referred to as elastic channel and the set of all other channels as the inelastic channels. The initial state. An often used set of invariants are the Mandelstam variables.17) (11. however. tmax min λcd /4s.23)
which are not independent as s + t + u = m2 + m2 + m2 + m2 . there is usually a preferred direction. The variable s is always larger than a c b d the minimal value (ma + mb )2 .20) → π0 + n → π+ + π− + n
→ . a c with q = λab /4s and q ′ = being 0 or 180 degrees.19) (11. pc . the rapidity y is deﬁned T T x y y= 1 ln 2 E + pz E − pz = ln E + pz mT = tanh−1 pz .18) (11.21) (11.4).
The various possibilities are referred to as diﬀerent reaction channels.22) (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 . while the ﬁnal state often arises from a series of 2-particle processes combined with the decay of an intermediate particle (resonance). Writing E = mT cosh y. 2s
(11. = (11. Examples appear in → π− + p (11. pz = mT sinh y.Scattering theory
105
Kinematics of 2 → 2 scattering processes
π− + p
The simplest scattering process is 2 particles in and 2 particles out. py ). E
. A speciﬁc reaction channel starts contributing at the threshold value (Eq.. Instead of the scattering angle.

c bd Mad→c¯(su = u. u). uu = s) = Mab→cd (s. tu = s.28)
which (only involving angles) is easier to determine. 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 = = = . u = (p1 − p4 )2 = (p2 − p3 )2 . t. For large energies and not too small scattering angles (θ ≫ 1/γ) y is approximately equal to the pseudo-rapidity η. t.2
Crossing symmetry
In the previous chapter.30) (11. (11. t. ¯ while for the u-channel process (ad → c¯ one has b) su = (pa + pd )2 = u. which means that rapidity distributions dN/dy maintain their shape. With the momenta deﬁned as in the ﬁgures above one has for all these processes the same amplitude M (s. u). 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.Scattering theory
106
It is convenient because under a boost (along z) with velocity β it changes as y → y − tanh−1 β. u) with s = (p1 + p2 )2 = (p3 + p4 )2 . (11. b ut = (pa − pd )2 = u. tt = s. These variables are precisely the Mandelstam variables for the s-channel process (ab → cd). ¯ b
(11. 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.
uu = (pa − p¯)2 = s. This is known as crossing symmetry. ut = u) = Mab→cd (s. η = − ln(tan(θ/2)) = tanh−1 (cos θ). b Analiticity of the ﬁeld theoretically calculated result implies
Ma¯→¯ (st = t. ¯ tt = (pa − p¯)2 = s.
11.29) 3 2E 3 2 (2π) 2 (2π) 16π 2π E 16π 2 2π with at high energies the factor |p|/E ≃ 1. t = (p1 − p3 )2 = (p2 − p4 )2 . ¯ tu = (pa − pc )2 = t. t-channel and u-channel processes respectively. For the t-channel process (a¯ → ¯ one has c bd) st = (pa + pc )2 = t.31)
.

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

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. N and T · V are covariant.e.35)
This quantity is not covariant. 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 . . For the (proper) decay width one thus has 1 N Γ0 = . dt (11. 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 (11.38)
(in which we can calculate −iMf i using Feynman diagrams). as expected. 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 .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.36) (11. (11. pn ).
N 1 .
. σ= √ 2 λab T · V
(11.Scattering theory
108
Although this at ﬁrst sight does not look covariant. T · V ρa (11. the number of scattered or decayed particles is given by 2 N = (2π)4 δ 4 (Pi − Pf ) iMf i dR(p1 . the amount of decaying particles is proportional to the number of particles with proportionality factor the em decay width Γ. it is. .T
V. . From the solution N (t) = N (0) e−Γ t one knows that the decay time τ = 1/Γ.39) One of the δ functions can be rewritten as T · V (remember the normalization of plane waves).T
d4 x = V · T (2π)4 δ 4 (Pi − Pf ).34)
i. (11. . Microscopically one has Ndecay = Na · Γ T or Γ= N 1 .

implies for the scattering matrix M .43)
dΩcm =
π M (s. . note the sign and cos θ = qi · qf ). . . (11.44)
This can be used to get the full expression for dσ/dt for eµ and e+ e− scattering. q f ). . q n ) q n
qn √ Mni (q i .4
Unitarity condition
(S † )f n Sni = δf i
The unitarity of the S-matrix. i M† M √ − √ 8π s 8π s =
fi n
dΩn
∗ Mnf qn Mni √ √ . The amplitude −M /8π s is the one to be compared with the quantum mechanical scattering amplitude f (E. .
(11. 16π 2 s
(11. for which the √ amplitudes squared have been calculated in the previous chapter. 2 λab and for instance for two particles dσ = q ′ M (s. (11. p1 .42)
2−body decay
The diﬀerential cross section (ﬁnal state not integrated over) is given by 1 dσ = √ |Mf i |2 dR(s. i.e.46)
Partial wave expansion
Often it is useful to make a partial wave expansion for the amplitude M (s. . The sign diﬀerence comes from the (conventional) sign in relation between S and quantummechanical and relativistic scattering amplitude. θ) = −8π s
ℓ
(in analogy with the expansion for f (E. The result is N = |Mf i |2 dR(s. . (M † )f n (2π)4 δ 4 (Pi − Pn ) Mni . θ) in quantum mechanics. ˆ ˆ Inserted in the unitarity condition for M . q n ). θ). p1 . √ (2ℓ + 1)Mℓ (s) Pℓ (cos θ). θ)|2 . respectively. pn ) |M |2 q 32π 2 m2 dΩ |M |2 . p1 . pn ).45)
Since the amplitudes also depend on all momenta the full result for two-particle intermediate states is (in CM. . (11. . θcm ) √ q 8π s
2 2
(11.47) M (s. for which one has dσ/dΩ = |f (E.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 .41) (11. t) λab 4π
dt.Scattering theory
109
(Using normalized wave packets these somewhat ill-deﬁned manipulations can be made more rigorous). 8π s 2π 8π s
. 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 .
11.25) −i Mf i − (M † )f i = −
∗ dΩ(ˆn )Mnf (q f . . . pn ). θ) or M (q i . see 11. . (11.

Using σ= dΩ q ′ M (s.
(11.52)
From the imaginary part of M (s. θcm ) √ q 8π s
2
.48)
If only one channel is present this simpliﬁes to
∗ ∗ −i (Mℓ − Mℓ ) = 2 qMℓ Mℓ .
(11.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.
(11. Show that σT = = 4π q 2π q2 (2ℓ + 1) Im Mℓ (s)
ℓ
ℓ
(2ℓ + 1) (1 − ηℓ cos 2δℓ ). 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 .50)
where Sℓ (s) satisﬁes |Sℓ (s)| = 1 and δℓ (s) is the phase shift.
in combination with the partial wave expansion for the amplitudes M and the orthogonality of the Legendre polynomials immediately gives for the elastic channel.54)
ℓ
. which allows writing Mℓ (s) = Sℓ (s) − 1 e2i δℓ (s) − 1 = . σel = 4π
ℓ
(2ℓ + 1)|Mℓ (s)|2 (2ℓ + 1)
ℓ
=
4π q2
ηℓ e 2i δℓ − 1 2i
2
. parametrized as Sℓ (s) = ηℓ (s) exp(2i δℓ (s)).53)
The diﬀerence is the inelastic cross section. 0). In general a given channel has |Sℓ (s)| ≤ 1. 2i q 2i q (11.
ℓ
(11. η = 1) the result σel = 4π q2 (2ℓ + 1) sin2 δℓ .51)
and for the case that this is the only channel (purely elastic scattering. q ˆ
i. σinel = π q2
2 (2ℓ + 1) (1 − ηℓ ).
† −i (Mℓ )f i − (Mℓ )f i = 2
† (Mℓ )f n qn (Mℓ )ni .
(11.e.49)
or Im Mℓ = q |Mℓ |2 . n
(11.

but still real. which is given by a sum of the partial widths into the diﬀerent channels. The unitarity condition then states 2 Im ∆−1 (k) = R
n † dR(p1 . 0) ∝ e−i(E−iΓ/2)t . For an unstable particle one expects that U (t. it is straightforward to write down the partial wave amplitude. 0).56)
(again disregarding spin).
(11.Scattering theory
111
Note that the total cross section is maximal in the case of full absorption. 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 . one has poles at k 0 = ±(Ek − iǫ) where Ek = + k2 + M 2 guarantees the correct behavior.e. η = 0. For instance the tree-level calculation of Feynman diagrams at order e2 was real.
11. however. For the amplitude in a scattering process going through a resonance. The product of amplitudes is of order e4 .58)
. . in which case. ∝ U (t. To check unitarity we need the amplitude at order e4 .57)
This shows that Γ is the width of the resonance.5
Unstable particles
i k 2 − M 2 + iǫ
For a stable particle the propagator is i∆(k) = (11. σel = σinel .55)
(note that we have disregarded spin). This is achieved with a propagator i∆R (k) = i k2 − M2 + iM Γ (11.e. We note that unitarity is generally broken in a ﬁnite order calculation. The quantity Γ is precisely the width for unstable particles. pn )MRn (2π)4 δ 4 (k − Pn ) MnR
=
2M
n
Γn = 2M Γ. . 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 time-evolution operator. 0)|2 = e−Γt . . 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 . s − M 2 + iM Γ (11. 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. This is what we will do to discuss decay widths. . (q Mℓ )ij (s) = −M Γi Γj . such that |U (t.
i. The prescription for the pole structure. i.

one can reconstruct the resonance shape for diﬀerent numbers of neutrino species. u) + g(t. u) = 1 1 2 (s + u2 ) + 4m2 (t − m2 ) t2 2 2 1 1 ( s − m2 )( s − 3m2 ) tu 2 2
. u) + f (u. • The J/ψ resonance in e+ e− scattering. its width Γ = 120 MeV. a fast change in the phase shift for a narrow resonance.26 keV. M2 − s (11. Three famous resonances are: • The ∆-resonance seen in pion-nucleon scattering.49 GeV. Three neutrinos explain the resonance shape. This is a narrow resonance discovered in 1974. Its mass is M = 3096.88 MeV. At resonance the cross section σT (π + p) is about 210 mb. The phase shift in the resonating channel near the resonance is given by tan δℓ (s) = MΓ . Knowing that each neutrino contributes about 160 MeV (see next chapter). Its mass is M = 1232 MeV. the partial width into e+ e− is Γee = 5. the full width is Γ = 88 keV. Γ = 2.60)
showing that the phase shift at resonance rises through δ = π/2 with a ’velocity’ ∂δ/∂s = 1/M Γ.59)
reaching the unitarity limit for s = M 2 .
Exercises
Exercise 11. t)} .Scattering theory
112
(Prove this using the unitarity condition for partial waves). The cross section at resonance is about 30 nb. Essentially this resonance can decay into quark-antiquark pairs or into pairs of charged leptons. which is attributed to neutrinos. t) + g(u. it is furthermore easy to prove that the cross section is given by σel = M 2 Γ2 4π (2ℓ + 1) .4) can be written as 4πα2 dσ = {f (t. The cross section σT (π − p) also shows a resonance with the same width with a value of about 70 mb.e. Limiting ourselves to a resonance in one channel. The half-width of the resonance is M Γ. u) = g(t. i. 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). where furthermore σinel = 0.2 GeV. Note that because of the presence of a background the phase shift at resonance may actually be shifted. From this one sees that a resonance has the same shape in all channels but diﬀerent strength. This characteristic shape of a resonance is called the Breit-Wigner shape. All these decays can be seen and leave an ’invisible’ width of 498 MeV.1
Show that the cross section for electron-electron scattering (exercise 10. • The Z 0 resonance in e+ e− scattering with M = 91. dt s(s − 4m2 ) with f (t. 2 q (s − M 2 )2 + M 2 Γ2 (11.

a quantity which we will ¯ ¯ encounter in the next chapter (section 12. writing N (s) M1 (s) = 2 + iM Γ .999 877 Γ and Γ(π − → e− νe ) = 0. s − M∆ ∆ ∆
2 calculate Re N (M∆ )
.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.8.2
Calculate the ﬂux factor. cm |va − vb |
in the center of mass system for two colliding particles a and b. ρcm ρcm a b 1 cm . Show that unitarity can be used one step further to ﬁx also the real part of the residue of the pole in the amplitude.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∆ .
Exercise 11. Calculate now fπ . Compare the results with the experimental π − lifetime τ = 2. Compare the result with the factor trf 1/ρa ρb va that we used in the derivation of the basic expression for cross sections. 4.
Exercise 11.2).Scattering theory
113
Exercise 11. 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.000 123 Γ.6 × 10−8 s and the relative decay widths Γ(π − → µ− νµ ) = 0. Thus.

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

The standard model the basis for x G and
x+dx
117 G. (12. i. =
.
(12.e. (12.
x+dx
ψ(x + dx) = 1 + dxµ Dµ
x+dx
ψ(x). ∂µ − ig Aµ (x) ψ(x) = 0 dxµ ∂µ − ig Aµ (x(s)) ψ(x) = 0.17) The three equations given so far immediately give Dµ = ∂µ − ig Aµ (x). The relation in Eq. i.
which connects two identical vectors. Dν ] ψ(x) 1 − ig dσ µν Gµν ψ(x).19)
This gives rise to a (path dependent) phase in each point.
We note that local gauge invariance requires that we can modify all local systems with a (local) unitary transformation S(x). requiring that the ’connection’ Aµ (x) is such that Dµ → S(x)D µ S −1 (x).
x+dx
ψ(x)
= =
1 + ig dxµ Aa (x)Ta µ 1 + ig dx Aµ (x)
µ x
x
ψ(x)
(12. In principle such a phase in a given point is not observable. x x+dx ψ(x + dx) = (1 + dxµ ∂µ ) ψ(x).16. What is this physical eﬀect by which the ’connection’ Aµ can be observed? For this consider the phase around a closed loop.
(12.e. 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). However. 12. A ’constant’ vector that only rotates because of the arbitrary deﬁnitions of local frames satisﬁes Dµ ψ(x) = 0.18)
or considering a path xµ (s) from a ﬁxed origin (0) to point x. but expresses them with respect to diﬀerent bases. ds s dxµ ds′ P exp ig A (s′ ) ψ(0).14) (12. if two diﬀerent paths to the same point give diﬀerent phases the eﬀects can be observed.16)
which in the presence of the ’connection’ Aµ diﬀers from the total change between x ψ(x) and x+dx ψ(x+ dx).
-dx -dy
µ µ
dyµ
µ
dx
For a vector. ds
which is solved by dψ(s) ds = dxµ . 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 . 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. should be independent of such transformations.15)
ψ(x).

and equivalently Aµ can be considered as a pure gauge eﬀect. Nature will choose one. Only if this quantity is nonzero a physically observable eﬀect of Aµ exists. 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).
i>j
which is rotationally invariant. φ= 2 φ3 (12. external magnetic ﬁelds.2: The symmetry-breaking ’potential’ in the lagrangian for the case that m2 < 0. the groundstate itself appears degenerate.The standard model
118
where Gµν = (i/g)[Dµ . As there can be one and only one groundstate. we can disregard those ’perturbations’ and look at the ideal situation.g. This characterizes spontaneous symmetry breaking. In this case there are many possible groundstates (determined by a ﬁxed direction in space). In geometric language the eﬀect on parallel transport of a vector depends on the ’curvature’ Gµν . If the temperature is high enough the spins are oriented randomly and the (macroscopic) ground state is spherically symmetric. Dν ]. Classically the (time-independent) ground state is found for
V(φ )
F
|φ|
Figure 12.2. 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. If it is zero one has dxµ Aµ (x) = 0. 2 2 4
−V (φ)
The potential V (φ) is shown in ﬁg.
12. φ1 φ .e. a theory for a scalar degree of freedom (a scalar ﬁeld) having three (real) components. e. this means that there is more than one possibility for the groundstate. which means that by an appropriate transformation S(x) it can be gauged away (see the example of Aharonov-Bohm eﬀect). in reality a not perfectly symmetric situation. Consider as an example a ferromagnet with an interaction hamiltonian of the form H=− Jij Si · Sj . 12. usually being (slightly) prejudiced by impurities.
.2
Spontaneous symmetry breaking
In this section we consider the situation that the groundstate of a physical system is degenerate.20)
with a lagrangian density of the form L =
1 1 1 ∂µ φ · ∂ µ φ − m2 φ · φ − λ(φ · φ)2 . Nevertheless. i.

23) F +η
the latter only forming a minimum for m2 < 0. i. In order to see the consequences for the particle spectrum of the theory we construct the lagrangian in terms of the ﬁelds ϕ1 . ϕc = 0|φ |0 = (12.21) Stability of the action requires the classical groundstate ϕc to have a well-deﬁned value (which can be nonzero). etc. . In this case the spectrum is described
. (12. while the quadratic terms must correspond with non-negative masses. therefore a choice must be made. say 0 0 . A quantum ﬁeld theory has only one nondegenerate groundstate |0 . It is only invariant under rotations around the 3-axis. . The classical groundstate appears degenerate. It is appropriate to redeﬁne the ﬁeld as ϕ1 φ = ϕ2 . has less symmetry. (12. The original lagrangian contained an SO(3) invariance under (length conserving) rotations among the three ﬁelds. φ = ϕc + φquantum where for the (operator-valued) coeﬃcients in the quantum ﬁeld one wants 0|c† = c|0 = 0. 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 . The presence of a nonzero value for the classical groundstate value of the ﬁeld will have an eﬀect when the ﬁeld is quantized. They are known as the Weyl mode or the Goldstone mode: Weyl mode. In the case of degeneracy. for the vacuum Qa |0 = 0.22) 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. In this mode the lagrangian and the vacuum are both invariant under a set of symmetry transformations generated by Qa . λ
such that 0|ϕ1 |0 = 0|ϕ2 |0 = 0|η|0 = 0. . It is suﬃcient to do this to second order in the ﬁelds as the higher (cubic. (12.25) 2 2 2 Therefore there are 2 massless scalar particles.e. Writing the ﬁeld φ as a sum of a classical and a quantum ﬁeld. In this situation one speaks of spontaneous symmetry breaking.
The situation now is the following. so one has 0|φquantum |0 = 0 and hence 0|φ|0 = ϕc .24) 1 2 2 4 1 1 1 = (∂µ ϕ1 )2 + (∂µ ϕ2 )2 + (∂µ η)2 + m2 η 2 + . Any constant ﬁeld with ’length’ F is a possible groundstate.) terms constitute interaction terms. The η massless particles are called Goldstone bosons. ϕ2 and η. The ﬁeld along the third axis plays a special role because of the choice of the vacuum expectation value.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. while the lagrangian including the nonzero groundstate expectation value chosen by nature. 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.

down. the lagrangian for the quarks consists of the free Dirac lagrangian for each of the types of quarks. M = . 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.32) (12. Goldstone mode. Note that for the ﬁelds π a (x) one would have a the relation 0|π a (x)|π a (k) = ei k·x . Taking the derivative. π
(12.31)
. a ∂ µ Jµ = 0. π
Example: chiral symmetry
An example of spontaneous symmetry breaking is chiral symmetry breaking in QCD. (12.e. a and a′ are degenerate states. it is easy to see that one gets ∂µ V µ = i ψ [M. This means that they are operators that create states from the vacuum.e.29)
(Note that each of the entries in the vector for ψ in the above equation is still a 4-component Dirac spinor).26)
for all the states labeled by a corresponding to ’broken’ generators. a massless Goldstone boson for each ’broken’ generator. irrespective of the fact if they annihilate the vacuum.
a 0|∂ µ Jµ (x)|π b (k) = fπ k 2 ei k·x δab = fπ m2 a ei k·x δab . 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. In this mode the lagrangian is invariant but Qa |0 = 0 for a number of generators. A bit more formal.. strange.The standard model
120
by degenerate representations of the symmetry group. i. 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′ .e.28)
where ψ is extended to a vector in ﬂavor space and M is a diagonal matrix in ﬂavor space. . Neglecting at this point the local color symmetry. i. and one must have mπa = 0. called ﬂavors. V µ = ψγ µ T ψ. Known examples are rotational symmetry and the fact that the spectrum shows multiplets labeled by angular momentum ℓ (with members labeled by m). etc. Using the Dirac equation. . . suggesting the stronger relation ∂ µ Jµ (x) = fπ m2 a π a (x). Including a sum over the diﬀerent ﬂavors (up. ψ −→ ei α·T ψ. (12. ψu mu md ψ = ψd . denoted |π a (k) . H] = 0. This lagrangian density then is invariant under unitary (vector) transformations in the ﬂavor space. The generators Qa (in that case the rotation operators Lz . = ψ(i/ − M )ψ. there must be a nonzero expectation value 0|Jµ (x)|π (k) . i. if the generators Qa generate a symmetry. [Qa .) one can write L =
f
ψ f (i/ − Mf )ψf . ∂ ∂
(12. As the a generators for a symmetry are precisely the zero-components of a conserved current Jµ (x) integrated a a over space.30)
which in the case that we include only two ﬂavors form an SU (2)V symmetry (isospin symmetry) generated by the Pauli matrices. T = τ /2. T ] ψ.
(12.27)
If the transformations in the lagrangian give rise to a symmetry the Noether currents are conserved. . The conserved currents corresponding to this symmetry transformation are found directly using Noether’s theorem (see chapter 6).

I. but the fact that the symmetry is not perfect and the right hand side of Eq. gives also rise to a nonzero mass for the Goldstone bosons according to Eq. From group theory (Schur’s theorem) one knows that the latter can only be true. including in Dirac space a γ5 matrix. This symmetry. chiral symmetry is not perfect.34) Using the Dirac equation. In the real world. i. Therefore realization in the Weyl mode would require that all particles come double with positive and negative parity. stated equivalently. if in ﬂavor space M is proportional to the unit matrix. ψR/L = 1 (1 ± γ5 )ψ. T ] = 0. (12. That the massless theory has this symmetry can also be seen by writing it down for the socalled lefthanded and righthanded fermions. This combined symmetry is what one calls chiral symmetry. How can we see this.e. is what happens in the real world. as the quantity ∂µ Aµ (x). Still the basic fact that the generators acting on the vacuum give a nonzero result (i. M = m · 1.27. both are very small. (leaving out the ﬂavor structure) the same as ψγ5 ψ. (12. There exists another set of symmetry transformations.e. for which the ﬁeld (according to the discussion above) has the same behavior under parity. We know that mesons and baryons (such as the nucleons) have a well-deﬁned parity that is conserved. Aµ = ψγ µ T γ5 ψ. etc. parity minus). Note that these transformations also work on the spinor indices. which is equivalent 2 to the V-A symmetry. This situation.3 MeV/c2 and Mn = 939. The currents corresponding to this symmetry transformation are again found using Noether’s theorem. (12.e. is by nature not realized in the Weyl mode. nature has choosen the path SU (2)L ⊗ SU (2)R =⇒ SU (2)V . however. while the groundstate is only invariant under isospin rotations (R = L). much smaller than any of the other mesons or baryons. generated by TR/L = 1 (1 ± γ5 )T . 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.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. in terms of which the Dirac lagrangian 2 density looks like L = i ψL / ψL + i ψR / ψR − ψR M ψL − ψL M ψR . or. but also a triplet (isospin 1) of pions.The standard model
121
Furthermore ∂µ V µ = 0 ⇐⇒ [M.33)
which in the case of two ﬂavors form SU (2)A transformations generated by the Pauli matrices. e. The Goldstone bosons of QCD are the pions for which fπ = 93 MeV and which have a mass of mπ ≈ 138 MeV/c2 . socalled axial vector transformations. which is approximately true for the up and down quarks. the lagrangian density is invariant under left (L) and right (R) rotations independently. etc. T } γ5 ψ.e. 12. This symmetry is realized in the Weyl mode with the spectrum of QCD showing an almost perfect isospin symmetry. T γ5 = τ γ5 /2. it is easy to see that one gets ∂µ Aµ = i ψ {M.6 MeV/c2 ). The chiral ﬁelds ψR and ψL are transformed into each other under parity. where the quark masses are not completely zero. From the number of broken generators it is clear that one expects three massless Goldstone bosons. nucleons and atomic nuclei has not only an isospin or vector symmetry SU (2)V but also an axial vector symmetry SU (2)A . proton and neutron.
i. behaves as a pseudoscalar particle (spin zero.g. 12.35 is nonzero. parity would not play a role in the world.35)
In this case ∂µ Aµ = 0 will be true if the quarks have zero mass. ψ −→ ei α·T γ5 ψ. a doublet (isospin 1/2) of nucleons.
. with almost degenerate masses (Mp = 938.e. Therefore the world of up and down quarks describing pions. fπ = 0 remains. i. ∂ ∂ (12. SU (2)V (isospin) symmetry is good if the up and down quark masses are identical.

38)
∂µ Wν − ∂ν Wµ + g Wµ × Wν . in this case three-dimensional) representation reads (La )ij = −i ǫaij .
. One may wonder about the degrees of freedom. two massive bosons (Wµ and Wµ with MW = gF ) and a massive scalar particle (η with m2 = η 2 −2 m . Initially there are 3 massless gauge ﬁelds (each. Considering the same example with rotational symmetry (SO(3)) as for spontaneous symmetry breaking of a scalar ﬁeld (Higgs ﬁeld) with three components.37) (12. We have the possibility to perform local gauge transformations.40)
a Dµ φ = ∂µ φ − ig Wµ La φ. having two independent spin components) and three scalar ﬁelds (one degree of freedom each). (12.The standard model
122
12. 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). one sees that the ﬁelds Wµ and Gµν also can be represented as three-component ﬁelds. 1 1 L = − Gµν · Gµν + Dµ φ · Dµ φ − V (φ).
is made. Therefore we can always rotate the ﬁeld φ into the 3-direction in order to simplify the calculation.
The symmetry is broken in the same way as before and the same choice for the vacuum. Dµ φ = Gµν = ∂µ φ + g Wµ × φ. 0 φ= (12. 2 massive vector ﬁelds or spin 1 bosons (2 × 3) and one scalar ﬁeld (1).42) 2 from which one reads oﬀ that the particle content of the theory consists of one massless gauge boson 3 1 2 (Wµ ).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. . 0 ϕc = 0|φ |0 = 0 . like a photon.e. again 9 degrees of freedom. After symmetry breaking the same number (as expected) comes out. i. . The diﬀerence comes when we reparametrize the ﬁeld φ. made into a gauge theory.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 . as in this case there are no massless Goldstone bosons. thus 9 independent degrees of freedom. 0 0 0 = . but one has 1 massless gauge ﬁeld (2). 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 + . (12. while the number of massive gauge bosons coincides with the number of ’broken’ generators. The latter is a spin 0 particle (real scalar ﬁeld) called a Higgs particle. .39) (12. 4 2 where Since the explicit (adjoint.
(12.

neutrinos.51)
. however. τ ).50) (12. i. explicitly L R
SU(2)W SU(2)W
−→
ei α·τ /2 L.44)
which has an SU (2)W symmetry under transformations eiα·TW .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. the quarks (up.43)
and
One introduces a (weak) SU (2)W symmetry under which eR forms a singlet. The lagrangian for the leptons consists of three families each containing an elementary fermion (electron e− . the leptons (elektrons. ∂ ∂ ∂
(12.and righthanded particles. muon µ− or tau τ − ). explicitly Dµ L Dµ R i i g Wµ · τ L + g ′ Bµ L.The standard model
123
12. By introducing gauge ﬁelds and breaking the symmetry a more complex lagrangian is obtained. while the lefthanded particles form a doublet. Thus L
(f )
= i eR / eR + i eL / eL + i νeL / νeL + (µ. implies certain nontrivial relations between masses and mixing angles that can be tested experimentally and sofar are in excellent agreement with experiment. down.48) (12. The procedure.
(12. etc. νµ and ντ ) and their antiparticles.). R −→ e−i β R.47)
and YW is considered as an operator that generates a U (1)Y symmetry. that gives a good description of the physical world. 1 As they are massless. under which the lefthanded and righthanded particles with YW (L) = −1 and YW (R) = −2 transform with eiβYW /2 .
U(1)Y U(1)Y
(12.
(12. R.e.) and the gauge bosons responsible for the strong. 2 2 = ∂µ R + i g ′ Bµ R. In the standard model one starts with a very simple basic lagrangian for (massless) fermions which exhibits more symmetry than observed in nature. etc.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. This is written as 3 Q = TW +
YW . For the neutrino only a lefthanded particle (and righthanded antiparticle) exist.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. electromagnetic and weak forces. = ∂µ L − (12. 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.
Thus the lagrangian density is L
(f )
∂ ∂ = i L/ L + i R/ R. its corresponding neutrino (νe . explicitly L −→ e−i β/2 L. muons. left.45) (12. 2
(12. ψR/L = 2 (1 ± γ5 )ψ decouple.

the U (1)Q symmetry (generated by the charge operator). Using local gauge invariance θi for i = 1. leading to the parametrization 1 0 φ(x) = √ (12. L where L
(h1) (h)
=L
(h1)
+L
(h2)
.53) φ= √ (θ4 − iθ3 ) φ 2 with TW = 1/2 and YW = 1 is introduced.
(12. .
(12. + 8
(12. Putting this in leads to
L
(f )
=L
(f 1)
+L
(f 2)
. .58)
(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 = 0.59)
.54)
= (Dµ φ)† (Dµ φ) −m2 φ† φ − λ (φ† φ)2 . (12.
−V (φ)
L and
(h2)
= −Ge (LφR + Rφ L).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 .The standard model
124
3 3 where Wµ is a triplet of gauge bosons with TW = 1. 2 and 3 may be eliminated (the necessary SU (2)W rotation is precisely e−iθ(x)·τ ).
†
i i g Wµ · τ − g ′ Bµ )φ. which assures with the choice of YW for the Higgs ﬁeld that Q generates the remaining U (1) symmetry. + 8 g2 v2 1 + − (Wµ )2 + (Wµ )2 (∂µ h)2 + m2 h2 + 2 8 (g 2 + g ′2 ) v 2 (Zµ )2 + . For the classical ﬁeld the choice θ4 = v is made. 4 4
In order to break the symmetry to the symmetry of the physical world. with the following lagrangian density consisting of a symmetry breaking piece and a coupling to the fermions.57)
=
1 g2 v2 1 2 (Wµ )2 + (Wµ )2 (∂µ h)2 + m2 h2 + 2 8 v2 2 3 gWµ − g ′ Bµ + . h)
(12.56) 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. . a complex Higgs ﬁeld + 1 φ √2 (θ2 + iθ1 ) 0 = 1 (12. . .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. this leads to the lagrangian L
(h1)
.

(12. 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 .
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 .63) (12.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 . .62)
g 2 + g ′2
3 ≡ sin θW Wµ + cos θW Bµ .65)
The weak mixing angle is related to the ratio of coupling constants. 2
(12. 2θ 2θ 4 cos W cos W 0. 2 + g ′2 g
3 g ′ Wµ + g Bµ
(12.67)
The coupling of electrons or muons to their respective neutrinos.66)
From the coupling to the photon.69)
.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 .64) (12.60) (12. 4 2 MW g2 v2 = . . (12. 2
Zµ (12. 2 3 g Wµ − g ′ Bµ 3 ≡ cos θW Wµ − sin θW Bµ . σ (νµ γ ρ µL ) 2 2 (eL γ νe ) 2 k + MW (12. we can read oﬀ e = g sin θW = g ′ cos 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 + .61) (12. g ′ /g = tan θW .

giving rise to a running coupling constant after renormalization of the ﬁeld theory. For each neutrino species (with CV = 1/2 and CA = 1/2 one expects Γ(νν) ≈ 155 MeV. 80. me /v is extremely small. The masses of the fermions and the coupling to the Higgs particle are contained in L (h2) .e.76)
with CV CA
3 = TW − 2 sin2 θW Q. g ′ and v determine a number of experimentally measurable quantities.73) (12.
(12. Γinvisible = 480 MeV one sees that three families of (light) neutrinos are allowed. (12. one obtains L
(h2)
Ge v Ge = − √ (eL eR + eR eL ) − √ (eL eR + eR eL ) h 2 2 me = −me ee − eeh. Γe ≈ Γµ ≈ Γτ ≈ 83 MeV). With the choosen vacuum expectation value for the Higgs ﬁeld. as appear for instance in the decay of the neutron or the decay of the Λ (quark content uds). g2 MZ 2 2 (CV + CA ). one has the relation g2 e2 1 GF √ = .5 MeV (exp. v
(12.78)
From this coupling it is straightforward to calculate the partial width for Z 0 into a fermion-antifermion pair.40 GeV. The coupling to the Higgs particle is weak as the value for v calculated e.77) (12.231 2. Comparing this with the total width into (invisible!) channels.g. i.
2 The value of α deviates from the known value α ≈ 1/137 because of higher order contributions.19 GeV.72) (12.75)
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 .79) Γ(Z 0 → f f ) = 48π cos2 θW For the electron. such as2 α = e2 /4π GF sin2 θW MW MZ ≈ 1/127.e. 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. 1.166 4 × 10 0.
−5
(12.The standard model
126
the good old four-point interaction introduced by Fermi to explain the weak interactions.71) GeV
−2
= = = =
. from the MW mass is about 250 GeV.80)
First.
(12. leptons with CV = −1/2 + 2 sin2 θW ≈ −0.74) (12.05 and CA = −1/2 we calculate Γ(e+ e− ) ≈ 78. Finally we want to say something about the weak properties of the quarks. 3 = TW . the mass of the electron comes from the spontaneous symmetry breaking but is not predicted (it is in the coupling Ge ). i. muon or tau. 2 2 2
(12. 91. (12.70) = 2 = 2 sin2 θ 8 MW 2 v2 8 MW 2 W In this way the parameters g.
.

81)
n −→ pe− νe
⇐⇒
d −→ ue− νe . ¯
3 This mixing allows all quarks with TW = −1/2 to decay into an up quark. We will not discuss this any further in this chapter. their antiparticles and the electroweak gauge bosons as appearing in each family.3: Appropriate YW and TW assignments of quarks.e. The
.3. Q = TW + YW /2 and constant along speciﬁc diagonals as indicated in the ﬁgure. for which SU (5) or SO(10) are actually nice candidates. leptons. Decay of B-mesons containing b-quarks allow estimate of Vub . 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. t c u ′ . This is only true if the mixing matrix is at least three-dimensional. The electric charge Q is then 3 ﬁxed.
Λ −→ pe− νe ¯
⇐⇒
s −→ ue− νe . The pattern is actually intriguing.
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. etc. ′ ′ 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. suggesting an underlying larger unifying symmetry group. As shown in Fig. 12. Comparing neutron decay and Λ decay one can get an estimate of the mixing parameter Vus in the socalled Cabibbo-Kobayashi-Maskawa mixing matrix. which can account for the (observed) CP violation of the weak interactions.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. this requires particular YW -TW assignments to get the charges right. CP violation requires three generations. but with diﬀerent strength. i. In principle one complex phase is allowed in the most general form of the CKM matrix.

each of these containing the three families. A ≈ 0.e.82)
space.227. e. Note that we need the conjugate Higgs ﬁeld to get a U (1)Y singlet in the case of the charge +2/3 quarks. i.g.82 and ρ ≈ 0.86)
the unitary CKM-matrix introduced above in an ad hoc way. d d
(12.
† Λu Λ† = Vu G2 Vu . u u
where we include now the three lefthanded quark doublets in QL . The Λu and Λd are complex matrices in the 3×3 family
and Λd Λ† = Vd G2 Vd† . 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 . φ = −φ− 0 2 For the (squared) complex matrices we can ﬁnd positive eigenvalues.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. the weak mass eigenstates are
L
(h2.
.22 and η ≈ 0.
(12. The imaginary part i η gives rise to CP violation in decays of K and B-mesons (containing s and b quarks. respectively).83)
where Vu and Vd are unitary matrices. 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 three righthanded quarks with charge +2/3 in UR and the three righthanded quarks with charges −1/3 in DR .q)
= −QL φΛd DR − DR Λ† φ† QL − QL φc Λu UR − UR Λ† φc† QL u d
(12.84)
with Gu and Gd being real and positive and Wu and Wd being diﬀerent unitary matrices. mc and mt ) and Md = Gd v/ 2 (diagonal matrix containing md .85) √ √ with Mu = Gu v/ 2 (diagonal matrix containing mu .The standard model
128 using the socalled Wolfenstein + O(λ4 )
with λ ≈ 0. The Higgs ﬁeld is still limited to one complex doublet. ms and mb ). Thus one has
† † † =⇒ −DL Vd Md Wd DR − DR Wd Md Vd† DL − UL Vu Mu Wu UR − UR Wu Mu Vu UL (12. for which we need the appropriate weak isospin doublet 0∗ 1 φ c = √ v +h .
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.
(12. allowing us to write
† Λu = Vu Gu Wu † and Λd = Vd Gd Wd .q)
′ weak † mass mass DL = DL = Vu Vd DL = VCKM DL .34.

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
. 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 .ℓ)
= −LφΛe ER − ER Λ† φ† L − Lφc Λn NR − NR Λ† φc† L e n
(12.
UPMNS
a parametrization that in principle also could have been used for quarks. = − 1/6 2/3 0 1/2 1/2 0 0 1 0 1/3 1/2 − 1/6 (12. it is particularly useful because θ12 is essentially determined by solar neutrino oscillations requiring ∆m2 ≈ 8 × 10−5 12 eV2 (convention m2 > m1 ).ℓ) † † † =⇒ −EL Ve Me We ER − ER We Me Ve† EL − NL Vn Mn Wn NR − NR Wn Mn Vn NL .87)
is a weak doublet containing the three families of neutrinos (NL ) and charged leptons (EL ) and ER is † a three-family weak singlet.92)
.90)
mass † mass mass † mass † We note that there are mass ﬁelds ER = We ER .5×10−3 eV2 .88) √ mass with Me = Ge v/ 2 the diagonal mass matrix with masses me .e.ℓ)
= −LφΛe ER − ER Λ† φ† L.ℓ) =⇒ −Me EL Ve We ER − ER We Ve† EL . since the weak current is the only place where they show up. The W -current then becomes mass mass E L γ µ NL = E L γ µ NL . decoupling from all known interactions. In this case. As before. mµ and mτ . † † Again we continue as before now with matrices Λe = Ve Ge We and Λn = Vn Gn Wn . while θ23 then is determined by atmospheric neutrino oscillations requiring |∆m2 | ≈ 2.
(12. one can write Λe = Ve Ge We and we ﬁnd † L (h2. The mass ﬁelds ER † mass † mass = We ER .89)
with three righthanded neutrinos added to the previous case. 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 . and obtain L
(h2.91)
† with UPMNS = Vn Ve known as the Pontecorvo-Maki-Nakagawa-Sakata mixing matrix. i.
(12. EL = Ve† EL . e N L= L EL
(12. NL = Vn NL and NR = Wn NR µ mass γ µ V † V N mass .The standard model
129
The lepton sector (massless neutrinos)
For a lepton sector with a lagrangian density of the form L in which
(h2. 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. For the (massless) neutrino ﬁelds we just can redeﬁne ﬁelds into NL = Ve† NL .
The lepton sector (massive Dirac neutrinos)
In principle a massive Dirac neutrino could be accounted for by a lagrangian of the type L
(h2.
(12. there is no family mixing for massless neutrinos. (12. we ﬁnd massless neutrinos.

(12. the massless and massive Majorana neutrinos. Absorption of phases in the states is not possible for Majorana neutrinos. the left.95) 2 In order to have more than a completely decoupled sector.94) M L NL NL + M L NL NL .ν = − MR NR NR + MR NR NR . Thus (disregarding family structure) one has two Majorana neutrinos. If the Majorana mass MR ≫ MD one actually obtains in a natural way one Majorana neutrino with a very small mass. This is called the see-saw mechanism (outlined below).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. as above. 2 R ψ = nR + nL . (12. For the neutrino sector. − MD MR nR 2
(12. Thus 2 Υ ′ = nc + nL . 1 c ∗ c L mass. 1 L Υ ′ = nR + nc . For the leptons.
The see-saw mechanism
Consider (for one family N = n) the most general Lorentz invariant mass term for two independent Majorana spinors. Υc ≡ (ΥL )c = ΥR 1 2 L and Υc = ΥL ).ν
=−
containing three (CP-violating) phases (α1 . coupled by a Dirac mass term.99)
. one must for the neutrinos as well as charged leptons. hence the mixing matrix becomes iα /2 0 0 e 1 (12.and righthanded species then just form a Dirac fermion. Actually.The standard model
130
The lepton sector (massive Majorana ﬁelds)
An even simpler option than sterile righthanded Dirac neutrinos. 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. with for the righthanded sector a mass term MR . are equivalent to two decoupled Majorana neutrinos (see below). Υ′ and Υ′ (satisfying Υc = Υ and as discussed in chapter 6. 2 although this option is not attractive as it violates the electroweak symmetry. We use here the primes starting with the weak eigenstates. a unitary matrix relating them to the weak eigenstates. The way to circumvent this is to introduce as in the previous section righthanded neutrinos.96) VPMNS = UPMNS K with K = 0 eiα2 /2 0 . couple the right. For the charged leptons there cannot exist a Majorana mass term as this would break the U(1) electromagnetic symmetry.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. one being massive. For these light Majorana neutrinos one has. The mass matrix can be written as ML |MD | eiφ M = (12. α2 and δ).88 a Majorana mass term for the (lefthanded) neutrino mass eigenstates. 1 ∗ c c (12.and lefthanded species through Dirac mass terms coming from the coupling to the Higgs sector as in the previous section. 0 0 1 L
mass. is to add in Eq.98) (12. however.c. 12.

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 .
. L Υ2L nR
Υ1 and Υ2 . For the situation ML = 0 and MR ≫ MD (taking MD real) one 2 ﬁnds M1 ≈ MD /MR and M2 ≈ MR .
(12. related via c Υ1R = U nL . 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′ .The standard model
131
taking ML and MR real and non-negative.1
Consider the case of the Weyl mode for symmetries. V µ = ψγ µ T ψ and Aµ = ψγ µ γ5 T ψ ∂µ V µ ∂µ A
µ
= =
i ψ [M. 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). T ] ψ. 2 with real masses Mi .105)
for i = 1.
Exercise 12. [Qa . i ψ {M.104)
↔ 1 1 Υi i / Υi − Mi Υi Υi ∂ 4 2
(12. T } γ5 ψ.
(12.
Exercise 12. H] = 0. a and a′ are degenerate states. 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 .
Exercises
Exercise 12.103)
for each of which one ﬁnds the lagrangians L =
and we are left with two decoupled Majorana ﬁelds Υ1L = U ∗ nc .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
.
(12. This is a mixing problem with a symmetric (complex) mass matrix leading to two (real) mass eigenstates.e. i.2
Derive for the vector and axial vector currents.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)) . i. Prove that if the generators Qa generate a symmetry. which implies U ∗ M † U † = U ∗ M ∗ U † = M0 and a ’normal’ diagonalization of the (hermitean) matrix M M † . Υ2R nR
(12.
2 U (M M † ) U † = M0 .e.

the weak mixing angle is given by ¯ sin2 θW = 0.7
In this exercise two limits are investigated for the two-Majorana case.
Write down the matrix element squared (averaged over initial spins and summed over ﬁnal spins) for the decay of the Z 0 . Note that you can ﬁnd the answer without explicit construction of all interaction terms in the lagrangian. The mass of the Z 0 is MZ = 91 GeV. to be precise one has for bosons (b) or fermions (f). Γ(Z 0 → νe νe ).
Exercise 12.
Exercise 12. and the width to a pair of neutrino’s.
(b) Calculate the width to electron-positron pair. 2 2 The masses are mt ≈ 175 GeV. etc. In the standard model this coupling is described by the vertex −i g µ √ (γ − γ µ γ5 ) . e+ e− h. at least up to a high (which?) order in λ.231.
Exercise 12. ghbb = b v v where v is the vacuum expectation value of the Higgs ﬁeld.
0
¯ Γ(Z 0 → f f ) =
MZ g2 48π cos2 θW
f f CV 2 + CA 2 . hhh. Neglect the masses of fermions and use the fact that the sum over polarizations is 3 pµ pν .4
Calculate the lifetime τ = 1/Γ for the top quark (t) assuming that the dominant decay mode is t → b + W +. ǫ(λ) (p)ǫ(λ)∗ (p) = −gµν + µ ν M2
λ=1
¯ to calculate the width Γ(Z → f f ). Mf M2 and ghf f = . Γ(Z 0 → e+ e− ).5
Show that the couplings to the Higgs (for W + W − h. ZZh.6
Check that the use of the Wolfenstein parametrization in the CKM matrix indeed gives a unitary matrix.
Exercise 12.
.The standard model
132
with CV CA = =
3 TW − 2Q sin2 θW . mb ≈ 5 GeV and MW ≈ 80 GeV.) are proportional to the mass squared for bosons or mass for fermions of the particles. 3 TW .

what is the mass MX if we take for MD the electroweak symmetry breaking scale v (about 250 GeV). Give the explicit expressions that relate ψ and ψ c with Υ1 and Υ2 .The standard model
133
(a) Calculate for the special choice ML = MR = 0 and MD real.05 eV.
. which leads to the socalled see-saw mechanism. Given that neutrino masses are of the order of 0. 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. Calculate the eigenvalues ML = 0 and MR = MX . (b) A more interesting situation is 0 = ML < |MD | ≪ MR .