This action might not be possible to undo. Are you sure you want to continue?

# 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

2 4.2 2.8 3. Quantum Field Theory. S.2. Zuber. 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).H.4 4.1 2. McGraw-Hill. Cambridge University Press.3 3. Diagrammatica. 2. L.4 3.2 3. 5.3 2. M. Schroeder. 3.3 3.3 2. 1995. An introduction to Quantum Field Theory.1 2.2 36 22 .4 3. Quantum Field Theory.1.1 1.1 6. I: Foundations.2 3. 1. Cambridge University Press.2 6. M.1 4. Other text books of Quantum Field Theory that are useful are given in refs [3-6]. 1985. Peskin & Schroeder and Srednicki.5 5. 1995. 3.4 6. Srednicki.5 4.2 3. Veltman. I refer to the book of Srednicki [1] and Ryder [2]. The quantum theory of ﬁelds.3 3.3 References As most directly related books to these notes. Weinberg.2 2.3 2.5 2. Ryder. 2007.5 2. 4.3 4. Cambridge University Press. Corresponding chapters in books of Ryder.4 2. Vol. C. Cambridge University Press.E.2 6. M. 6. 2.1 3.V.1 5.3 3.3 3. Addison-Wesly. Cambridge University Press. Vol.2 2.-B.1 3. but are rather compact.1 2 2 3.3 2.3. 1994. Quantum Field Theory. 3. Peskin and D.4 3.2 3. 1980. These notes Section 1.2 1. II: Modern Applications.3 2.3 6.7 2.5 Ryder Peskin & Schroeder Srednicki 1 1 1 1 3 2. The notes contain all essential information. Itzykson and J.2 3. 1996.4 2.

8 6. 9.4 10.7 4.4 7.4 2.1 7.4 12.6 3.5 9.5 9.6 6.2.1.6 10.5 .1 10.2.3.3.2 5.1 5.1 9.3.6 9.4 12.3 6.5.4. 2.4 3.5 7.5 4. 4. 6.1 11.1 20.4. 6.4 9. 6.3 3.3 12.2 8.4.3 11. 20.6 4.2 10. 4.5 11. 8.3 8. 4.5 4.1.3 2.10 8.3 20.2 4.4 4. 2. 4.7.4 8.6 3.3 7.5 6 7 6 6.5 2.5 7.2. 6.7 5.1 9.4 These notes Section 7.1 20.3 9.3 10.1 20.1 12. 9.1.2 7.3 6 8 9 Srednicki 3 3 3 37 5.2.5 Ryder 4 4 4 4 4 4 Peskin & Schroeder 2.2 9.8 5 4. 6.2 11.3 8.6 3.3 7.5 6.2 12.6 8.2 8.1 8.

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

3) The uncertainty principle tells us that in this representation the momenta cannot be fully determined. One can talk about states |r and the wave function ψ(r) = r||ψ .1) In the limit that the action S is much larger than . Corresponding position and momentum operators do not commute. c = 299 792 458 m s−1 . (1. 1 (1. (1. ˜ The corresponding states are |p and the wave functions ψ(p) = p||ψ are the Fourier transforms of the coordinate space wave functions. 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. In special relativity a special role is played by the velocity of light c. ˜ ψ(p) = i d3 r exp − p · r ψ(r). (1. ≡ h/2π = 1. the position can in principle be determined at any time with any accuracy.6) . the action of the momentum operator in the coordinate representation is not as simple as the position operator. (1. It is given by pop ψ(r) = −i ∇ψ(r).2) In the limit that v ≪ c one reaches the non-relativistic domain. In quantum mechanics. Indeed. In this coordinate representation the position operators r op simply acts as rop ψ(r) = r ψ(r). quantum eﬀects do not play a role anymore and one is in the classical domain.054 571 68 (18) × 10−34 J s = 6. In quantum mechanics a special role is played by Planck’s constant h. being eigenvalues of the position operators.Chapter 1 Introduction 1.5) One can also choose a representation in which the momenta of the particles are the dynamical variables. (1.582 119 15 (56) × 10−22 MeV s.1 Quantum ﬁeld theory In quantum ﬁeld theory the theories of quantum mechanics and special relativity are united. pj ] = i δij . usually given divided by 2π. S ≫ . They satisfy the well-known (canonical) operator commutation relations [ri .4) where δij is the Kronecker δ function.

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.

µn = Λµ1ν1 . • ds2 < 0 (spacelike intervals).23) δjm δkn − δjn δkm . 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.3) with basis nµ (µ = ˆ 0.2. the points lie on the lightcone and they can be connected by a light signal. Unlike in Euclidean space. (1. In special relativity we start with a four-dimensional real vector space E(1. are . x3 ). The length (squared) of a vector is obtained from the scalar ˆ product. xi are the three space components. independent of a coordinate system. 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 . Vectors are denoted x = xµ nµ . 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.1. x2 . x1 . We note that for Euclidean spaces (with a positive deﬁnite metric) vectors and tensors there is only one type of indices.Introduction 5 that can be used in the cross product of two vectors z = x × y. . x2 = x · x = xµ xν nµ · nν = xµ xν gµν .22) (1. The lower indices are constructed in the following way.21) (1. −x). x1 . where the coordinate t = x0 is referred to as the time component.26) (1. • ds2 = 0 (lightlike or null intervals). In Minkowski space. in which case zi = ǫijk xj yk .24) The quantity gµν ≡ nµ · nν is the metric tensor. For four-vectors in Minkowski space we will use the notation with upper indices and write x = (t.νn can be conν structed. linear transformations that do not change the length of a four-vector are called the Lorentz transformations. The (invariant) lengths often have special names. Useful relations are ǫijk ǫℓmn ǫijk ǫimn ǫijk ǫijl = = = δiℓ δjℓ δkℓ 2 δkl . the invariant length or distance (squared) is not positive deﬁnite. δim δjm δkm δin δjn δkn . xµ = gµν xν . No diﬀerence is made between upper or lower. we will use upper indices for the three-vectors. (1. So we could have used all upper indices in the above equations. it is convenient to distinguish lower indices from upper indices. denoted as V ′µ = Λµν V ν .25) Within Minkowski space the real. and are given by (x0 . The invariant distance between two points x and y in Minkowski space is determined from the length dsµ = (x − y)µ .. These transformations do change the components of a vector. given by g00 = −g11 = −g22 = −g33 = 1 (the ˆ ˆ other components are zero). The real. Relations relating tensor expressions. x) = (x0 . ˆ ˆ (1.. x3 ) = (t. When 3-dimensional space is considered as part of Minkowski space.. Many other four vectors and tensors transforming like T ′ µ1 . one must distinguish tensors with upper or lower indices and one can have mixed tensors. The Poincar´ transformations include e e Lorentz transformations and translations. . Because of the diﬀerent signs occurring in gµν . One can distinguish: • ds2 > 0 (timelike intervals). however. Λµn n T ν1 .3). such as eigentime τ for the position vector τ 2 ≡ x2 = t2 − x2 . x2 .. One has x2 = xµ xµ = t2 − x2 . 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 . linear transformations that leave the length of a vector invariant are called (homogeneous) Lorentz transformations.

= −24. p). because one has (1.30) (Note that there are diﬀerent conventions around and sometimes the opposite sign is used).35) ρ ǫµνρσ ǫµν σ σ ǫµνρσ ǫµνρ ′ ′ = −2 g ρρ g σσ − g ρσ g σρ = −6g σσ . It is an invariant tensor. g0 = g1 = g2 = g3 = 1. 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 .32) . a∞ = 4πǫ0 2 /me e2 .∇ . . The derivative ∂µ is deﬁned ∂µ = ∂/∂xµ and we have a four vector ∂ with components (∂0 .33) (1.34) (1. starting from ǫ0123 = 1. ′ ′ . Each of these permutations leads to a minus sign. (1. The length squared of ∂ is the d’Alembertian operator. deﬁned by ∂2 − ∇2 . ∂2 .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. (1. but more important has the same eﬀect on lefthandside and righthandside. For the contraction of a vector with the antisymmetric tensor one often uses the shorthand notation ǫABCD = ǫµνρσ Aµ Bν Cρ Dσ . (1. (1.36) Exercises Exercise 1. ∂t (1. 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. p) = (E. (1.31) ǫµνρσ ǫµν ′ ′ ρσ ′ = − (1. not aﬀected by Lorentz transformations. ∂t ∂x ∂y ∂z ν ∂µ xν = gµ . explicitly written as (p0 . Relate this quantity to the electron Compton wavelength −e via the dimensionless λ ﬁne structure constant α.27) It is easy to convince oneself of the nature of the indices in the above equation. = ∂ . constructed via the metric tensor itself. gµ = gµρ g ρν and 0 1 2 3 is in essence a ’Kronecker delta’. is timelike with invariant length (squared) p2 = p · p = pµ pµ = E 2 − p2 = m2 .29) 2 = ∂ µ ∂µ = ∂t2 The value of the antisymmetric tensor ǫµνρσ is determined in the same way as for ǫijk .28) ν ν Note that gµ with one upper and lower index. .Introduction 6 called covariant. 0123 0123 ′ ′ ǫ µνρσ ǫµνρσ The ﬁrst identity. Note that in this equation one has on left. where m is called the mass of the system. Examples are the scalar products above but also relations like pµ = m dxµ /dτ for the momentum four vector. ∂3 ) = ∂ ∂ ∂ ∂ .1 (a) In the the Hydrogen atom (quantum system) the scale is set by the Bohr radius. for instance. ∂1 . .

n1 . Also construct and calculate the Planck length Lpl . n1 . g−− .3 Prove the following relation ǫµνρσ g αβ = ǫανρσ g µβ + ǫµαρσ g νβ + ǫµνασ g ρβ + ǫµνρα g σβ . ˆ ˆ . a2 ] we can ﬁnd basis vectors n− . 0. a+ . a2 ] or [a− . [Note: what is the MKS unit for V/T?] Exercise 1. e. 1. re = e2 /4πǫ0 me c2 to the Compton wavelength. 1. . 1) ˆ . One does not need to know . Exercise 1. ρ. ǫ0 . which is the Compton wavelength for the Planck mass. Compare it with the proton mass and use Eq. 0. a1 . 0) ˆ n3 = (0. 0) ˆ n2 = (0. ν. σ} is a permutation of {0.4 Lightcone coordinates for a four vector a (which we will denote with square brackets as [a− . a2 .13 to give its actual value in kg. but only appropriate combinations. (b) The coordinates (a0 . aT ]) are deﬁned through √ a± ≡ (a0 ± a3 )/ 2. . 0.Introduction 7 (b) Relate the classical radius of the electron (a relativistic concept). (d) Use the value of the gravitational constant GN / c5 = 6.71 × 10−39 GeV−2 to construct a mass Mpl (Planck mass). (c) Calculate the Compton wavelength of the electron and the quantities under (a) and (b) using the value of c. a3 ) are the expansion coeﬃcients using the basis vectors n0 . This demonstrates how a careful use of units can save a lot of work. Also for the coordinates [a− . a1 . 3} the index α can only be equal to one of the indices in ǫµνρσ . 0. n2 . α and me c2 = 0. n3 . 1.2 Prove the identity A × (B × C) = (A · C) B . 0) ˆ n0 = (1. a+ . g+− and g−+ . by a simple few-line reasoning [For instance: If {µ. me . Use that to give the ˆ ˆµ ˆ components of the basis vectors n+ and n− . 2. n2 . n+ . c.511 MeV. (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). Exercise 1. 1. (a) Express the scalar product a · b in lightcone coordinates and deduce from this the values of g++ . ]. . 0.(A · B) C using the properties of the tensor ǫijk given in section 1. 0. a1 . ˆ ˆ ˆ ˆ These are n1 = (0. a+ . How are a+ and a− related to a+ and a− . Note that ˆ ˆ ˆ ˆ a · nα = aα and derive from that the components nα of the four vector nα .3.

E = p2 . 2 + M 2 φ(r. namely plane waves characterized by a wave number k. written as pµ −→ i∂µ is covariant (the same in every frame of reference).7) k 0 = ± k2 + M 2 = ±Ek . ∂t 2M (2. we just want to play a bit with covariant equations and study their behavior under Lorentz transformations.1) under the substitution (in coordinate representation) E −→ Eop = i ∂ . 8 . (2.5) with (k 0 )2 = k2 + M 2 . t) = − ψ(r. φk (r. Substitution of operators gives the Klein-Gordon (KG) equation for a real or complex function φ. (2. i ∂ ∇2 ψ(r.4) where M is the particle mass.3) Equations 2. 2M (2. The Schr¨dinger equation in quantum mechanics is the operator equation o corresponding to the non-relativistic expression for the energy.1 The Klein-Gordon equation In this chapter. 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. one obtains (2. (2.1 and 2. But the replacement 2. t) = exp(−i k 0 t + i k · r).2) Acting on the wave function one ﬁnds for a free particle. p). t) = ∂2 − ∇2 + M 2 φ(r. t) = 0. Thus a covariant equation can be obtained by starting with the (covariant) equation for the invariant length of the four vector (E. t). ∂t2 (2.6) The energy spectrum is not bounded from below: considering the above stationary plane wave solutions.3 are not covariant.Chapter 2 Relativistic wave equations 2. Although it is straightforward to ﬁnd the solutions of this equation. ∂t p −→ pop = −i∇. p2 = pµ pµ = E 2 − p2 = M 2 .2.

This leads to the existence of negative densities. relativistically the density is not a scalar quantity.3). And. but rather the zero component of a four vector. we want the most general solution. 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.Relativistic wave equations 9 i.13) ˜ with (in principle complex) coeﬃcients φ(k). for which we need the interpretation of φ itself as an operator. 2M 2M (2. This operator has all possible solutions in it multiplied with creation (and annihilation) operators. which is what Lorentz transformations do. −i φ∗ ∇ φ . The KG equation.2 Mode expansion of solutions of the KG equation Before quantizing ﬁelds. The integration over k-modes clearly is covariant and restricted to the ‘mass’-shell (as required by Eq. j) of the four-current j. it is simply a matter of being careful to ﬁnd a consistent (covariant) theory. is a second order equation and φ and ∂φ/∂t can be ﬁxed arbitrarily at a given time. For this we note that an arbitrary solution for the ﬁeld φ can always be written as a superposition of plane waves.14) ˜ ˜ Introducing φ(Ek .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. The appropriate current corresponding to the KG equation (see Excercise 2. (2. As we will see later both problems are related and have to do with the existence of particles and antiparticles. however. φ(x) = d4 k ˜ 2π δ(k 2 − M 2 ) e−i k·x φ(k) (2π)4 (2. (2π)3 2Ek (2. rather than as a wave function.15) . k) ≡ a(k) and φ(−Ek . there are no longer fundamental objections to mix up space and time. It is possible to rewrite it as an integration over positive energies only but this gives two terms (use the result of exercise 2.2) is j µ = i φ∗ ∂ µ φ or (ρ.e.11) Therefore.5). ∂ρ = −∇ · j. 2. −k) ≡ b∗ (k) one has φ(x) = d3 k e−i k·x a(k) + ei k·x b∗ (k) = φ+ (x) + φ− (x).10) ∂t which follows directly from the Schr¨dinger equation. φ(x) = d3 k (2π)3 2Ek ˜ ˜ e−i k·x φ(Ek . just as the dependence on time was in ordinary quantum mechanics. ∂µ j µ = 0. there are solutions with negative energy.9) They satisfy the continuity equation. −k) . having the KG equation as a space-time symmetric (classical) equation. At that point the dependence on position r and time t is just a dependence on numbers/parameters on which the operator depends. This continuity equation can be written down o covariantly using the components (ρ. k) + ei k·x φ(−Ek . ↔ ↔ (2. 2. (2. (2.

φP (x) ≡ φ(˜) (P for parity).18) (2. x) → (t.21) d3 k x x e−i k·˜ a(k) + ei k·˜ b∗ (k) (2π)3 2Ek d3 k ˜ ˜ e−i k·x a(k) + ei k·x b∗ (k) (2π)3 2Ek d3 k e−i k·x a(−k) + ei k·x b∗ (−k) . (2π)3 2Ek (2.22) showing that with each solution there is a corresponding charge conjugated solution φC (x) = φ∗ (x). i. for which we consider space inversion. −x) ≡ (˜µ ).16) The consequence of this is discussed in Exercise 2. Performing some Lorentz transformation x → x′ = Λx.15 one has elimated k 0 and in both equations k · x = Ek t − k · x. changing the sign of the spatial coordinates. Since a · b = a · ˜ it is easy to see that ˜ b.20) one has for the mode amplitudes aP (k) = a(−k) and bP (k) = b(−k). x ∂µ ∂ µ + M 2 φ(˜) = 0.e. It is trivial to see that ∂µ ∂ µ + M 2 φ∗ (x) = 0. 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) .6 We will explicitly discuss the example of a discrete symmetry. (2π)3 2Ek (2.3 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.14 and 2. after restricting the energies to be positive). This shows how parity transforms k-modes into −k modes. 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). 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) .17) (2. The coeﬃcients a(k) and b∗ (k) are the amplitudes of the two independent solutions (two. (2. (2.23) . x Transforming everywhere in the KG equation x → x one obtains ˜ ˜ ˜ x ∂µ ∂ µ + M 2 φ(˜) = 0.19) implying that for each solution φ(x) there exists a corresponding solution with the same energy. (2. one thus must have that the function φ → φ′ such that φ′ (x′ ) = φ(x) or φ′ (x) = φ(Λ−1 x).Relativistic wave equations 10 In Eqs 2. (2π)3 2Ek (2. which implies (xµ ) = (t. They are referred to as mode and anti-mode amplitudes (or because of their origin positive and negative energy modes). Another symmetry is found by complex conjugating the KG equation. 2.

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

j 0 = ψ†ψ and j = ψ † αψ. Exercise 2. From this one concludes that α and β must be matrix-valued and ψ must be a multi-component wave function. express the quantities E(t) = P i (t) = d3 x (∂0 φ)∗ (∂0 φ) + ∇φ∗ · ∇φ + M 2 φ∗ φ . i (a) Show that relativistic invariance. αj } = 2 δ ij I. 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).2) in terms of the a(k) and b(k) using the expansion in Eq. i. d3 x (∂ {0 φ)∗ (∂ i} φ). One then ﬁnds that Q is expressed as an inﬁnite sum over number operators a† a (particles) and b† b (anti-particles). a{µ bν} ≡ aµ bν + aν bµ . in terms of the a(k) and b(k). B} = AB + BA.1) for a complex scalar ﬁeld current (exercise 2. B] = AB − BA and the anticommutator as {A.7 To solve the problem with positive and negative energies and get a positive deﬁnite density. t).Relativistic wave equations 12 Exercise 2. an attempt to construct a ﬁrst order diﬀerential equation for the time evolution would be to write. Note that a{µ bν} indicates symmetrization.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). ∂ ψ(r. (b) Show that one has current conservation for the current 2 We denote the commutator as [A. also using the result in (a).5 (a) As an introduction to parts (b) and (c). 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 . 2. {αi . We will encounter these quantities later as energy and momentum. (c) Similarly. Exercise 2. ∂t where ψ is a wave function and the nature of α and β is left open for the moment.e. 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). . t) = (−i α · ∇ + m β) ψ(r.15. and β 2 = I. 1 2 (b) Express now the full charge QV (t) (exercise 2. β} = 0. 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 .

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

Next. Locally this is a 3-dimensional Euclidean space and SU (2). ˆ therefore. z]] + [y. 2 (3. It is straightforward to check that any (ﬁnite) rotation can be obtained from a combination of inﬁnitesimal rotations. They satisfy the commutation relations [Li . 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). is a 3-dimensional Lie-group. [x. with real a’s and 3 (ai )2 = 1.5) Rotations in general do not commute. x]] + [z. Lj ] = i ǫijk Lk . for rotations around z for instance. namely that there exists a bilinear product [. [y. y] = −[y. y]] = 0 (Jacobi identity). 0 (3. Writing a0 = cos(ϕ/2) and a = n sin(ϕ/2) we have1 A = A(ϕ.and y-axes 0 0 0 0 0 0 0 −i . With matrix commutation this algebra satisﬁes the requirements for a Lie algebra.7) = a0 0 −1 i 0 1 0 0 1 • [x. . x]). [z. thus is as the surface i=0 of a sphere in 4 Euclidean dimensions. π) (n.1: the parameter spaces of SO(3) (left) and SU (2) (right). and thus (σ · n)2 = 1. π ) π -sphere (antipodes identified) 2 π -sphere (surface identified) Figure 3. ] that satisﬁes • ∀ x.Groups and their representations 14 (n. (3.4) R . ˆ R(ϕ. y ∈ A ⇒ [x. x] = 0 (thus [x.z N N or (Lk )ij = −i ǫijk . These generators form a threedimensional Lie algebra SO(3).2 π) (-n. (3. z ) = lim N →∞ that are generated by i 0 .6) Summarizing. Lx = Ly = 0 0 0 i 0 −i 0 ϕ ˆ . which reﬂects itself in the noncommutation of the generators. 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. • [x. In the same way we can consider rotations around the x. y] ∈ A. Ly and Lz . One way of viewing the parameter space. n) = 1 cos ˆ 1 Note ϕ ϕ ˆ + i (σ · n) sin 2 2 = exp i ϕ ˆ σ·n . These matrices can be deﬁned as acting on 2-component spinors (χ → Aχ) or equivalently as acting on 2 × 2 matrices (B → ABA† ). we turn to the group SU (2) of special (det A = 1) unitary (A† = A−1 ) 2 × 2 matrices.

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

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

SU (2) is the covering group of SO(3). i. For a decent (welldeﬁned) global representation. For an element in the group G the corresponding point in the parameter space can be reached in two ways. if χ → σχ. the simply connected group is called the (universal) covering group. If a transformation U acts on χ. Consider the j = 1/2 representation of SU(2).20) By moving through the parameter space the D-functions can be extended to global functions for all allowed angles. m′ |UE (ϕ. Explicitly. dm′ m (θ) ≈ δm′ m − iθ(J 2 )m′ m . all states in the representation space have the same energy. θ. one can look at the conjugate representation. contractable and paths that run from a point at the surface to its antipode. Jz = i 0 i 0 −i 0 0 0 −i 0 0 0 0 0 . For those global representations. the conjugate representation is not a new one. −1 ≡ ǫ−1 ≡ √ (ǫx − i ǫy ) . (j) (j) ′ The spin matrices for that Cartesian basis {ǫx . UE (ϕ. There exist two diﬀerent types of paths. J = 0 0 + 0 0 0 0 0 0 0 0 √ 2 0 √ 0 . e. the topological structure of the group is important. The jump U → U ∗ implies for the generators σ/2 → −σ ∗ /2.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 . θ. (j) (3. the possibility thus exist that some extensions will not be well-deﬁned representations. The matrix elements of these unitary representations are known as D-functions. 2 From the (hermitean) representation Φ(g) of G. ǫy . Given a representation. the conjugate transformation U ∗ acts on χ∗ . χ)|j. If rotations leave H invariant. χ) ≡ e−iϕJz e−iθJy e−iχJz . J− = 2 0 0 0 0 √ 3 0 0 0 0 √ 2 0 0 0 0 . that is any closed curve in the parameter space can be contracted to a point. The group SO(3). For SO(3). 0 ≡ ǫ0 ≡ ǫz . m = Dm′ m (ϕ. j. is not simply connected. 1 |1. This turns out to be the case for all half-integer representations of the Lie algebra. however. then ǫχ → ǫσχ = ǫσǫ−1 ǫχ = −σ∗ ǫχ. θ. however.g. For j = 1 another commonly used representation starts with three Cartesian basis states ǫ related to the previous basis via 1 |1.3). Inﬁnitesimally (around the identity) the D-functions for SU (2) and SO(3) are the same. 1 ≡ ǫ1 ≡ − √ (ǫx + i ǫy ) . 2 |1. however. to be precise the states ǫχ transform via −σ ∗ . Because there exists a matrix ǫ such that −σ ∗ = ǫ σǫ−1 one immediately sees that appropriate linear combinations of conjugate states.e.19) (3. This works for SU (2). . 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 SU(2). χ) = eim ϕ dm′ m (θ) e−imχ . Jy = 0 0 0 . the extension from a local one must be unique. Of all groups of which the Lie algebras are homeomorphic. Therefore the representation and conjugate representation are equivalent in this case (see Excercise 3. however. one obtains the unitary representations Φ(g) = exp(i Φ(g)) of the full group G. For the transformations A in SU (2) and R in SO(3) these are commonly parametrized with Euler angles. ǫz } are 0 0 0 0 0 i 0 Jx = 0 0 −i . If the group is simply connected. or equivalently the Hilbert space can be written as a direct product space of spaces V (j) .

In this section we consider the Poincar´ group.26) (3. (ii) |Λ00 | ≥ 1. From this property.22) (Note that xT is a row vector). Writing x as a column vector and the metric tensor in matrix form. det Λ = −1 is called improper). G = 10 (3. which as a matrix equation with (Λ)µν = Λµν and (G)µν = gµν gives (ΛT )ρµ (G)µν (Λ)νσ = (G)ρσ . (3. (det Λ = +1 is called proper. We considered the generators and looked for representations in ﬁnite dimensional spaces. 0 0 g00 g01 g02 g03 1 0 g g11 g12 g13 0 −1 0 0 = .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 (3.3 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. Invariance of the length of a vector requires for the Lorentz transformations x′2 = gµν x′µ x′ν = gµν Λµρ xρ Λνσ xσ = x2 = gρσ xρ xσ or Λµρ gµν Λνσ = gρσ . e To derive some of the properties of the Lorentz group. σ/2 in a two-dimensional (spin 1/2) case. proof: (ΛT GΛ)00 = (G)00 = 1 → Λµ0 gµν Λν0 = 1 → (Λ00 )2 − i (Λi 0 )2 = 1.25) (3.24) L↑ + L↓ + L↑ − L↓ − proper orthochrone proper non-orthochrone improper orthochrone improper non-orthochrone . Thus one has the (pseudo-orthogonality) relation ΛT GΛ = G ⇔ GΛT G = Λ−1 ⇔ ΛGΛT = G.g.23) The proper tensor structure of the matrix element (Λ)µν is a tensor with one upper and one lower index. it is easy to derive some properties of the matrices Λ: (i) det(Λ) = ±1.Groups and their representations 18 3. e. explicitly x′µ = (Λ)µν xν + aµ or x′µ = Λµν xν + aµ . proof: det(ΛT GΛ) = det(G) → (det Λ)2 = 1. consisting of the Lorentz group and translations. it is convenient to use a vector notation for the points in Minkowski space. Denoted in terms of column vectors and 4-dimensional matrices one writes for the poincar´ transe formations x′ = Λ x + a.27) (3. 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.

28) 3′ 3 V 0 0 0 1 V 0 0 0 0 The parameter η runs from −∞ < η < ∞. J = (J 1 . z ) ≈ I + i ϕ J 3 . 0 0 +1 0 0 +1 0 0 ∈ L↑ (identity) I = (3. The commutator of two boosts in diﬀerent directions (e. 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). 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 . (3. Returning to the global group.31) 0 − 0 +1 0 0 0 0 +1 0 0 +1 0 0 −1 0 0 ∈ L↑ (space inversion) Is = (3. since I ∈ L↑ and + + det Λ and Λ0 change continuously along a path from the identity. z ) = exp(−i ηK 3 ). 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 . Using these explicit transformations. K 3 ). J j ] = i ǫijk J k . From the commutation relations one sees that the boosts (pure Lorentz transformations) do not form a group. K 2 . inﬁnitesimally given by ΛB (η. inﬁnitesimally given by ˆ ΛR (ϕ. The ﬁrst two sets of commutation relations exhibit the rotational behavior of J and K as vectors in E(3) under rotations. we have found the generators of rotations. K j ] = i ǫijk K k .30) 0 + 0 +1 0 0 0 0 +1 0 0 −1 0 0 +1 0 0 ∈ L↓ (time inversion) It = (3. [K i .g. K j ] = −i ǫijk J k . (3.Groups and their representations = i=1 (Λ0 )2 . i proof: use ΛT GΛ = G and ΛGΛT = G. J 3 ).32) 0 − 0 −1 0 0 0 0 −1 . z ) ≈ ˆ ˆ I − i η K 3 . In L↑ . which satisfy the commutation relations (check!) [J i . z ) = exp(i ϕ J 3 ). since the generators K do not form a closed algebra.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↑ . it is easy to ﬁnd the following typical examples from each of the four classes. Rotations around the z-axis are given by ΛR (ϕ. [J i . J 2 . K = (K 1 . and those of the boosts. one distinguishes rotations 0 + and boosts. This is the origin of the Thomas precession. 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. 3 i 2 i=1 (Λ0 ) 3 19 (iii) Boosts along the z-direction are given by ΛB (η. βγ = sinh η.

40) ΛB = I + i η 3 K 3 = I − i ω30 M 30 . inﬁnitesimally approximated by (I + ω. Of the four parts only L↑ forms a group. 3 30 with η = ω and K = M (e. Summarizing. One can easily convince oneself that i (ω)µν = ω µν = − ωαβ (M αβ )µν . we have for the (inﬁnitesimal) rotations (around z-axis) ΛR = I + i ϕ3 J 3 = I − i ω12 M 12 . Also this group can be divided into four parts.36) (3.37) The algebra of the generators of the Lorentz transformations can be obtained by an explicit calculation2 .33) These four transformations form the Vierer group (group of Klein). Thus the two vector operators J and K form under Lorentz transformations an antisymmetric tensor M µν .g. We now want to ﬁnd a covariant form of the six generators of the Lorentz transformations. explicitly reading x′µ = = (Λ)µν xν + aµ = ((I)µν + (ω)µν ) xν + ǫµ Λµν xν + aµ = (g µν + ω µν ) xν + ǫµ .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) (3.38) Explicitly. the Lorentz transformations form a (Lie) group (Λ1 Λ2 and Λ−1 are again Lorentz transformations. i i0 i i0 1 ijk M jk 2ǫ (3. three of which only involve spatial coordinates (rotations) and three others involving time components (boosts). we continue using covariance arguments.39) (e. (3. a). The e 2 These commutation relations are the same as those for the ’quantummechanics’ operators i(xµ ∂ ν − xν ∂ µ ) . 2 (M αβ )µν = i g αµ g βν − g α g βµ ν (3. In order to ﬁnd the commutation relations including the translation generators. This is a + normal subgroup and the factor group is the Vierer group.g. (3.34) The condition ΛT GΛ = G yields g ρµ + ω ρµ gρσ (g σν + ω σν ) = gµν =⇒ ωνµ + ωµν = 0. J 3 = M 12 ) and for the (inﬁnitesimal) boosts (along (3. P+ etc. 3. The extension to the Poincar´ group is e ↑ straightforward. Multiplying the proper orthochrone transformations with one of them gives all Lorentz transformations. ǫ). We will just require Poincar´ invariance for the generators themselves. We therefore ﬁnd (again) that there are six generators for the Lorentz group. inf (3.4 The generators of the Poincar´ group e ↑ The transformations belonging to P+ are denoted as (Λ. There are six generators.35) thus (ω)ij = −(ω)ji and (ω)0i = (ω)i0 . K = M ). which are obtained by writing the inﬁnitesimal parameters (ω)µν in terms of six antisymmetric matrices (M αβ )µν . with ϕi = − 1 ǫijk ω jk and J i = 2 z-axis). M ρσ ] = −i (g µρ M νσ − g νρ M µσ ) − i (g µσ M ρν − g νσ M ρµ ) . [M µν .

a)P µ U (Λ. [J i .47) (3.46) [K i . K j ] = i ǫijk K k . . the generators J 2 and J 3 in the case of the rotation group. For instance. They just state that M µν transforms as a tensor with two Lorentz indices. [J i . K j ] = −i ǫijk J k /c2 .48) . a) = Λ−1 µ P ν . P j ] = i ǫijk P k . . In order to label the states in an irreducible representation we construct a maximal set of commuting operators. .43) giving the following commutation relations by equating the coeﬃcients of ǫµ and ωµν . (3. . Of the generators of the Poincar´ group we choose ﬁrst of all the generators P µ . ν (3. check that U1 U2 corresponds with Λ1 Λ2 ). P ) in terms of the Hamiltonian and the three-momentum operators. (3. P ν ] [M µν . ǫ) = 1 − ωαβ M αβ + i ǫα P α . P ρ ] = 0. Λµν pν ⇒ ωµν pν ⇒ ωµν = = = pµ + ω µν pν = pµ 0 ǫµνρσ pρ sσ . a) = Λµν P ν or U † (Λ. = pµ |p. that commute e among themselves. because one then sees that by letting c → ∞ the commutation relations of the Galilei group. 2 2 (3. P j ] = [P i . a)P µ U † (Λ. (1 + iǫγ P γ − i i ωαβ M αβ )P µ (1 − iǫδ P δ + ωρσ M ρσ ) = P µ + ω µν P ν . H] = i P i . H] = 0. other states in that representation are obtained by the action of operators outside the maximal commuting set.38 could have been obtained in the same way. [P µ . To ﬁnd other generators that commute with P µ we look for Lorentz transformations that leave the four vector pµ invariant. known from non-relativistic quantum mechanics are obtained. writing the generator P = (H/c. These form a group called the little group associated with that four vector.44) (3. The generators of the translations are the (momentum) operators P µ . For instance. . The eigenvalues of these will be the four-momentum pµ of the state. [K i . The requirement that P µ transforms as a four-vector (for which we know the explicit behavior from the deﬁning four-dimensional representation). = −i (g µρ P ν − g νρ P µ ) . H] = [J i . P µ |p. leads to U (Λ. 3. [K i . [J i . as part of the set. the tensor M µν in terms of boosts cK i = M i0 and rotations J i = 1 ǫijk M jk . . Explicitly. 3. one obtains 2 [P i . We have here reinstated c. the generators M αβ no longer exclusively act in Minkowski space. Taking any one of the states in an irreducible representation.Groups and their representations 21 inﬁnitesimal form of any representation of the Poincar´ group (thus including the translations) is of e the form i U (I + ω. the generators J± in the case of the rotation group.45) Note that the commutation relations among the generators M µν in Eq. .42) (to decide on where the inverse is. P j ] = i δ ij H/c2 . (3. Inﬁnitesimally. These deﬁne states with speciﬁed quantum numbers that are eigenvalues of these generators.5 Representations of the Poincar´ group e where pµ is a set of four arbitrary real numbers. J j ] = i ǫijk J k .41) 2 At this point. (3.

Groups and their representations 22 where s is an arbitrary vector of which the length and the component along pσ are irrelevant. of which the components generate the little group of pµ or (for the third of the following relations) from explicit calculation [Mµν . . i exp − ωµν M µν 2 i exp − ǫµνρσ M µν pρ sσ 2 (3.50) where the Pauli-Lubanski operators W µ are given by 1 Wµ = − ǫµνρσ M νρ P σ . 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 . . 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. (3.49) (3. M ǫijk . Wν ] = −i (gµα Wν − gνα Wµ ) .51) The following properties follow from the fact that W µ is a four-vector by construction. 0. . These e operators are the Casimir operators of the algebra and can be used to deﬁne the representations of ↑ P+ . . From the algebra of the generators one ﬁnds that P 2 = Pµ P µ and W 2 = Wµ W µ (3. . . while the others have no physical signiﬁcance. Given the momentum four vector pµ we choose a tetrade consisting of three orthogonal spacelike unit vectors ni (p). 0.55) commute with all generators and therefore are invariants under Poincar´ transformations. the third case represents the vacuum. p0 > 0. The elements of the little group thus are U (Λ(p)) = = and U (Λ(p))|p. 2 exp (−isµ W µ ) |p. = = i exp − ǫµνρσ M µν pρ sσ |p. 2 (3. Wα ] = [Wµ . (3. .56) .54) This will enable us to pick a suitable commuting ’spin’ operator.53) (3. . Massive particles: p2 = M 2 > 0. which can thus be chosen spacelike.52) (3. . Pν ] = [Wµ .

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).60) and using the completeness relation nµ (p)nν (p) = − g µν − i i 2 i pµ pν M2 . 2 .e. Together with the four momentum states thus can be labeled as |M. 0. 0. M J) with components W i = (M/2)ǫijk M jk . More generally (fully covariantly) one can proceed by deﬁning S i (p) ≡ W i (p)/M and obtain [S i (p). p0 > 0. 0. . (1.. 0. show that W /M can be interpreted as the intrinsic spin. The commutation relations [W i (p).Groups and their representations 23 We can write W µ (p) = 3 W i (p) nµ (p) i i=1 (3. n1 (p). it is suﬃcient to choose a particular frame to investigate the coeﬃcients in Eq. 0). (3. s. 0).64) such that in an arbitrary frame where the four vectors p. 1.65) . . 3.61) one sees that S i (p) i =− W2 . W j (p)] = i M ǫijk W k (p). 0. 0. (3. M2 (3. Noting that S i (p) i 2 = 1 W µ (p)W ν (p) M2 niµ (p)niν (p).e. 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. 1 2 (3. −1) n1 (p0 ) ≡ n2 (p0 ) ≡ s(p0 ) ≡ (3. The explicit construction of the wave function can be done using the D-functions (analogous as the rotation functions). W i (p) = −W · ni (p)). i (3. ms . the S i (p) form the generators of an SU (2) subgroup that belongs to the maximal set of commuting operators.58) i. M2 (3. S j (p)] = −i nµ (p)nν (p) j i ǫµνρσ W ρ P σ = i ǫijk S k (p). . Having made a covariant decomposition. 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.57) (i. |p3 |). (0. 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 = 0.62) 1 Thus W 2 has the eigenvalues −M 2 s(s + 1) with s = 0. 1. p. (0. In that case the vectors ni (p) are just the space directions and W = (0.63) where E = p2 + M 2 and ms is the z-component of the spin −s ≤ ms ≤ +s (in steps of one).59) (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. 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). thus. U (Λ. |p. Thus 2 λ(p) = J ·p .Groups and their representations The above does actually include the massive case. W 2 and λ is derived from the general commutation relations for [Wµ . σ · RA a ≡ A(σ · a)A−1 ˆ (valid for any vector a).67) (3. The vacuum: pµ = 0. 3. i W 2 (p). A massless particle. which can take any integer or half-integer value. But if M = 0 one has a diﬀerent algebra (Eq. Exercises Exercise 3. is characterized by its momentum and the helicity. and gives [W 1 (p). (3. Wν ].69) We reproduce the algebra for the massive case with W 3 (p) = M λ(p). implying continuous values for the spin (actually for W i = M S i ). however. W 1 (p)] [λ(p).73) (3.1 (a) Derive from the relation that deﬁnes the mapping of SU (2) matrices A into SO(3) matrices RA in Eq.13. W µ (p) = λ(p) P µ .71) which is called the helicity. This algebra is (for instance) isomorphic to the one generated by rotations and translations in a 2-dimensional Euclidean plane (see exercises). Note that angular momentum does not contribute to helicity as L · p = 0. (3. W 2 (p)] [λ(p). W 2 (p)] = = = i M 2 λ(p). a)|0 = |0 . 3. Thus we have P 2 . |p| (3.67 has a vanishing right hand side). W 2 (3. −i W 1 (p). The algebra of W 1 . λ .72) One has P µ |0 = W µ |0 = 0 and the state is invariant under Lorentz transformations. The eigenvalues of W 1 and W 2 (in that case corresponding to the translations) thus can take any continuous values.70) as a commuting set with zero eigenvalues for P 2 and W 2 . 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 . The vector W µ (p) now can be expanded W µ (p) = W 1 (p) nµ (p) + W 2 (p) nµ (p) + λ(p) pµ + W s (p) sµ (p). We don’t know of any physical relevance.66) where W · p = 0 implies that W s (p) = −(M 2 /p · s) λ(p). 1 2 24 (3. where A = exp(i ϕ · σ/2) = exp(iϕ σ · n/2). This state is in physical applications nondegenerate and is denoted by |0 . P µ .68) (3.

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

Comment: extending this to more particles is a highly non-trivial procedure. = r × p + s. sj ] = [pi .6 Consider rotations and translations in a plane (two-dimensional Euclidean space). = p. [ri . H P J K = p 2 c 2 + m2 c4 . From either of these. H] + i dt ∂O . the others vanish. f (p)] = i ∇p f (p). Hint: for the Hamiltonian. p×s 1 (rH + Hr) − t p + . indeed assures d K /dt = 0.69 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. (b) In quantum mechanics one has the Ehrenfest relation i dO = [O. For a single free particle or for the free-moving center of mass system that is the quantity K = mr − tp. momentum and spin operators e satisfy the canonical commutation relations. although that in this case is a relation that is found already from the Ehrenfest relation for O = ri .7 (optional) One might wonder if it is actually possible to write down a set of operators that generate the Poincar´ e transformations. This is possible for a single particle. These do not matter in the non-relativistic limit (c → ∞).e. pj ] = iδ ij and [si . Exercise 3. Exercise 3. sj ] = 0. but it can be done. pj ] = [ri . consistent with the (canonical) commutation relations of a quantum theory. Use a 3-dimensional matrix representation or consider the generators of rotations and translations in the space of functions on the (x-y) plane. if you don’t want to do this in general. Do this by showing that the set of operators. and U (u) p U −1 (u) = p − m u. r2 ) inevitably leads to interaction terms in the boost operators. show ﬁrst the operator identity [r.67 to 3. i.. (a) In classical mechanics. sj ] = iǫijk sk . boost invariance leads to a conserved quantity. derive the commutation relations and show that the algebra is isomorphic to (i. = 2c2 H + mc2 satisfy the commutation relations of the Poincar´ group if the position. Determine the commutation relations for the corresponding operators Ki in quantum mechanics. the operator U (u) that gives U (u) r U −1 (u) = r − u t. [ri . ∂t Show that boost invariance for a free-moving particle (the commutation relations under (a) being valid). i. you might just check relations involving J or K by taking some (relevant) explicit components. there is a one to one mapping) the algebra in Eqs 3. (c) Give the unitary operator for boosts.e. . although the presence of an interaction term V (r1 .Groups and their representations 26 Exercise 3. in particular those with the momentum operator pj and the Hamiltonian H = mc2 + p2 /2m. rj ] = [pi .e. that’s why many-particle non-relativistic quantum mechanics is ’easy’. they describe r −→ r′ = r − u t while t′ = t.

where we restrict (for ﬁxed n) the parameters 0 ≤ ϕ ≤ 2π and 0 ≤ η < ∞. Aj ] = [B i .2) This shows that the Lie algebra of the Lorentz group is identical to that of SU (2) ⊗ SU (2). (4. similarly as the homeomorphism between SO(3) and SU (2).10) . B j ] = [Ai . 2 2 σ σ J = . 0. U (ϕ) H(η) (4. They precisely correspond to the two types of representations that we have seen before. 2 1 (J − i K). j ′ ). C): Type I (denoted M ): Type II (denoted M ): 27 J= σ σ . respectively.3) (4. 2 (4.8) ˆ ˆ ˆ with η = η n and ϕ = ϕn. K = −i . K = +i .9) (4. B j ] = i ǫijk Ak . K = i J (A = 0).5) = 1 (J + i K).Chapter 4 The Dirac equation 4.4) (4. C) can be written as a product of a unitary + matrix U and a hermitean matrix H. (j.1) (4. With this choice of parameter-spaces the plus and minus signs are actually relevant.1 The Lorentz group and SL(2. i ǫijk B k . 0) Type II : (0. Special cases are the following representations: Type I : (j. M = exp i ϕ·σ 2 1 exp ± η · σ 2 = U (ϕ) H(η) . becoming the deﬁning representations of SL(2. The matrices in SL(2. The group SL(2. it also follows directly that the Lorentz group is homeomorphic with the group SL(2.6) (4. (4. They will be labeled by two angular momenta corresponding to the A and B groups. C). 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 . j) K = −i J (B = 0). This tells us how to ﬁnd the representations of the group. It is simply connected and forms the covering group of L↑ . 2 2 (4.7) From the considerations above. C) is the group of complex 2 × 2 matrices with determinant one.

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. if there does not exist a unitary matrix S. ˜ Non-equivalence of M and M One might wonder if type I and type II representations are not equivalent. U ≡ (U † )−1 = U ) U (ϕ) = exp(i ϕ · σ/2) but diﬀerently under hermitean boosts (H † = H.13) 2M (E + M ) (exercise 4. U σ µ aµ U ˜ −1 H σ µ aµ H −1 = σ µ ΛB aµ H σ µ aµ H ˜ −1 (4. C) and. With E = M γ = M cosh(η) and p = M βγ n = M sinh(η) n. η → Hη. −σ). Also useful is the relation H 2 (p) = σ µ pµ /M = (E + σ · p)/M . This shows that M ∗ ≃ M . thus. are not covariant!) ˜ can be used to establish the homeomorphism between L↑ and SL(2. The following relations for + rotations and boosts are useful. (4. η → U η. Note that M ∗ = ǫ† M ǫ has its equivalent for the Lorentz transformations Λ(M ) and Λ(M ∗ ). 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). i.11) Considering ξ and η as spin states in the rest-frame. The Lorentz transformations Λ(M ) and Λ(M ). such that M = SM S −1 . ξ −→ Hξ. One has + −1 Λ(M ∗ ) = I2 Λ(M ) I2 .16) = σ µ Λ R aµ . (ǫ∗ = ǫ.1). H(η) = exp(−η · σ/2). C).14) satisfying Tr(σ µ σ ν ) = −2 gµν and Tr(σ µ σ ν ) = 2gµν = 2 δ µν (the matrices. (4. ˜ = σ µ Λ B aµ . In fact. 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.The Dirac equation 28 Let us investigate the deﬁning (two-dimensional) representations of SL(2. (4. U σ µ aµ U −1 = σ µ ΛR aµ .18 show that ±ǫξ ∗ transforms as a type-II (η) spinor. (4. σ µ ≡ (1. σ). C). Λ(ǫ) ∈ L+ . This is not true for the two-component spinors in SL(2. ǫ−1 = ǫ† = ǫT = −ǫ). namely H(η) = exp(η · σ/2).18) As already discussed for SU (2). however. One deﬁnes spinors ξ and η transforming similarly under unitary rotations (U † = U −1 . H ≡ (H † )−1 = H −1 ). ˜ The sets of four operators deﬁned by deﬁnitions σ µ ≡ (1. cannot be connected by a Lorentz transformation belonging to L↑ . one can use a boost to the frame with momentum ˆ ˆ p. thus. ¯ ¯ H ∗ = ǫ† Hǫ = ǫ−1 Hǫ.15) (4. C) or L↑ . while ±ǫη ∗ transforms as a type-I + spinor. 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. there exists the matrix ǫ = iσ 2 . ˜ (4.19) .12) ξ → U ξ. Eqs 4.e. (4.

must combine the repe resentations. We also have seen that the representations (0. (axial vector). −→ −p (vector). This provides the easiest way of seeing ˜ what is happening.23) u(p. −→ M (Λ) (4.22) u= η η M (Λ) 0 where M (Λ) = ǫM ∗ (Λ)ǫ−1 with ǫ given in Eq. C) are suitable for representing spin 1/2 particles. i. the aij elements of a tensor will not change sign. but by a transformation belonging ↑ ↑ to the class P− . 4. ) −→ ( . and the angular momentum in the rest frame is what we are familiar with as the spin of a particle. It has the same eﬀect as lowering the indices. the well-known two-component spinor for a spin 1/2 particle. The i angular momentum operators J are represented by σ i /2. etc. 2 2 (4. 0 χm uR H(p) .17. Within the Lorentz group. consider the four component spinor that transforms under a Lorentz transformation as 0 ξ ξ . 0). 0) (via operators J and K) one sees that under parity 2 2 1 1 (0. as we have seen in section 4. 1 ) and ( 1 .1.13. 4.The Dirac equation 29 where The matrix I2 does not belong to L↑ . i. From the deﬁnition of the representations (0. they can be connected. m) = η(0. ǫµνρσ will change sign. Examples (with between brackets the Euclidean assigments) are r t p H J λ(p) K −→ −λ(p) −→ −K −→ H −→ J −→ t (scalar). m) = χm . we have seen that in the rest frame W i (p)/M → J i . 0) of SL(2. e. Thus M ∗ and M are not equivalent within SL(2. For the spin 1/2 representations of the Poincar´ group including parity we. −a). (vector). Although these operators cannot be used to label the representations. that have the correct commutation relations. For a particle at rest only angular momentum is important and we can choose ξ(0. generators. a) and ˜ a = (a0 . C). −→ −r (vector). m) = χm uL H(p) 0 . The behavior is the same for classical quantities. but to L↑ . (4. + − 1 0 0 1 I2 = 0 0 0 0 0 0 −1 0 0 0 0 1 . Taking M (Λ) = H(p). (pseudoscalar).e. (scalar). we obtain for the two components of u which we will refer to as chiral right and chiral left components. the boost in Eq. they cannot be connected by a unitary 2 2 transformation. 1 ) and ( 1 .g.21) In nature parity turns (often) out to be a good quantum number for elementary particle states.e. therefore. The representing member of the class P− is the parity or space inversion operator Is . where a = (a0 . 2 ) and ( 2 .2 Spin 1/2 representations of the Lorentz group 1 1 Both representations (0. 0) are inequivalent.20) 4. = (4. A parity transformation changes aµ into aµ . but rather the operators W i (p) should be used.

(4. 4. 0 H 2 (p) uR uR . As in section 2 we can use Pµ = i∂µ as a representation for the momenta (translation operators) in function space.The Dirac equation 30 with H(p) = ¯ H(p) = 2M (E + M ) E+M −σ·p 2M (E + M ) E+M +σ·p . From this one obtains the Cliﬀord algebra for the Dirac matrices. which in general is a linear equation in pµ . {γ µ . The general requirements for the γ matrices are thus easily obtained. γ ν } = 2 g µν .28) which is a covariant (linear) ﬁrst order diﬀerential equation. Applying (iγ µ ∂µ + M ) from the left gives (4. To see this consider the equation for the hermitean † † conjugate spinor ψ † (noting that γ0 = γ0 and γi = −γi ). one must require that for ψ(x) the Klein-Gordon relation 2 + M 2 ψ(x) = 0 is valid for each component separately). This leads to the Dirac equation for ψ(x) = u(p)e−i p·x in coordinate space.32) suppressing on the RHS the identity matrix in Dirac space.33) deﬁne for a four vector a the contraction a = aµ γµ / . We will discuss another explicit realization of this algebra in the next section. ψ † −iγ 0 ∂0 +iγ i ∂i −M 1 We ← ← = 0.29) which is an explicit realization of the (momentum space) Dirac equation.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. γ } ∂µ ∂ν + M 2 ψ(x) = 0. (4.30) γ µ γ ν ∂µ ∂ν + M 2 ψ(x) = 0.27 is known as the Weyl representation. this can be rewritten 1 µ ν {γ . = H −1 (p). (4. p (4. (4.27) where γ µ are 4 × 4 matrices called the Dirac matrices1 . (iγ µ ∂µ − M ) ψ(x) = 0.7. (4. Since ∂µ ∂ν is symmetric. The explicit realization appearing in Eq. The ﬁrst indeed is solved. 2 (4. = (4. (pµ γ µ − M ) u(p) ≡ (/ − M ) u(p) = 0. It is of a form that we also played with in Exercise 2.31) To achieve also that the energy-momentum relation p2 = M 2 is satisﬁed for u(p). We ﬁrst want to investigate if the Dirac equation solves the problem with the negative densities and negative energies. .24) (4.25) It is straightforward to eliminate χm and obtain the following constraint on the components of u.

35) is indeed positive and j 0 can serve as a probability density. Relying on the Pauli exclusion principle for spin 1/2 particles. the one of the negative energy states. This can most easily be seen in the particle rest frame. We will see how this is properly implemented (also for bosons!) when quantizing ﬁelds. . 4. γ =ρ ⊗1= γ = iρ ⊗ σ = 0 −1 −σ k 0 2 We (4. γ ν } = 2 g µν . This problem was solved by Dirac through the introduction of a negative energy sea. 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.34) for the adjoint spinor ψ ≡ ψ † γ0 . opposite charge). This hole forms an antiparticle with the same mass.39) use ρi ⊗ σj with both ρ and σ being the standard 2 × 2 Pauli matrices. From the equations for ψ and ψ one immediately sees that j µ = ψγ µ ψ is a conserved probability current (exercise!).36) (4.1: The Dirac sea of negative energy states and antiparticles.The Dirac equation 31 E empty levels M -M filled levels E particle M -M hole Figure 4. (4. remains. (4.37) Using the explicit form of γ in the Weyl representation (see Eq. The Dirac sea forms the vacuum.3 General representations of γ matrices and Dirac spinors {γ µ . . one sees that there are two positive and two negative eigenvalues. but with properties such that it can be annihilated by the particle (e. 4.38) 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 . (4. The second problem encountered before. Multiplying with γ 0 on the right and pulling it through one ﬁnds a covariant equation ψ(x) iγ µ ∂µ +M ← = 0. where γ 0 p0 ψ = M ψ 0 −→ Eψ = M γ 0 ψ.g. The density j 0 = ψγ 0 ψ = ψ † ψ (4. From the vacuum a particle can be removed. E = +M (twice) and E = −M (twice).27).

(γ5 )S.R. 0 (4. = ρ ⊗ 1 = 1 0 i ǫµνρσ µ ν ρ σ γ γ γ γ . For instance in the Weyl representation (but valid generally). Another matrix which is often used is γ5 deﬁned as γ5 = i γ 0 γ 1 γ 2 γ 3 = −i γ0 γ1 γ2 γ3 = It satisﬁes {γ5 . γ 0 ]. =ρ ⊗1= 0 −1 3 (4.47) (4. γ 0 = ρ1 ⊗ 1 = 1 0 γµ ψ Diﬀerent representations can be related to each other by unitary transformations.The Dirac equation 32 The Weyl (or chiral) representation: 0 1 .44) and an adjoint spinor deﬁned by ψ = ψ † γ0 . For example. γ σ ]. ν The latter assures Lorentz invariance of the Dirac equation (see Exercise 4.R.42) We note that the explicit matrix taking us from the Weyl representation to the standard representation.9 and 4. (γµ )S. = S(γµ )W. it is easy to see that PR/L = 1 (1 ± γ5 ) 2 1 0 . γ µ } = 0 and explicitly one has 0 1 1 . −→ Sγµ S −1 . 0 σ3 2 4 1 −iσ 3 0 = − i [γ 3 . K 3 = M 30 = 0 iσ 3 2 4 or in general one has the transformation i ψ → Lψ = exp − ωρσ S ρσ ψ. We note that ψ → Lψ and ψ → ψ L−1 . which in the case of the Weyl representation are just the upper and lower components. 1 −1 2 For all representations one has † γµ = γ0 γµ γ0 . S −1 .49) i 1 with −i S ρσ = 2 σ ρσ = − 4 [γ ρ .R.43) S=√ .48) are projection operators. 4! (4.46) (4. the rotation and boost generators in Weyl representation in Eqs 4. while L−1 γ µ L = Λµ γ ν . γ 2 ]. that project out chiral right/left states.R. 0 γ k = −iρ2 ⊗ σ k = k σ −σ k .41) (4. is 1 1 1 (4. Lorentz invariance The Lorentz transformations can also be written in terms of Dirac matrices.40) (4. 2 (4.3) .10 are 1 σ3 0 i J 3 = M 12 = = [γ 1 .45) (γ5 )W. −→ Sψ. (4.

which is again a solution of the Dirac equation.56) (i/ − M ) ψ c (x) = 0.R. but it does. One has ψ c = −ψ T C −1 .58) where ηc is an arbitrary (unobservable) phase. ∂ with the solution ψ c (x) = ηc Cψ (x). that brings ψ → ψ c . −x) and via a few manipulations obtain again the ˜ Dirac equation (i/ − M ) ψ p (x) = 0.R. ∂ (4.54) e. is a left-handed spinor. It is easy to convince oneself that if ψ(x) is a solution of the Dirac equation.51) The existence of positive and negative energy solutions implies another symmetry in the Dirac equation. −iσ 2 0 −ǫ = .52) η ξ one can apply space inversion.6). T (4.57) (4. parity. x) → x = (t. = 0 ǫ iσ 2 (4. ψ= (4. As with parity we look for a transformation.g.g. (i/ − M ) ψ(x) = 0. This symmetry does not change the spin 1/2 nature. (or W. γ5 ] = 0. usually to be taken unity. ψ= (4.53) (4. T (4.) and all representations connected via a real (up to an overall phase) matrix S. 0 −ǫ 0 0 −ǫ 0 . such that T Cγµ C −1 = −γµ . reverse the charge of the particle. in Weyl representation we ﬁnd in Dirac space T C ξ −ǫ η ∗ . (C)W.50) Charge conjugation but with ψ p (x) ≡ ηp γ0 ψ(˜) (Exercise 4. (C)S. x = (t.The Dirac equation 33 Parity There are a number of symmetries in the Dirac equation. Some properties of C are C −1 = C † and C T = −C.55) (4. In S. = = 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. The latter implies that the conjugate of a right-handed spinor. ∂ (4.R. C is real and one has C −1 = C † = C T = −C c and [C. called charge conjugation. for instance. e.R.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 . Explicitly. ψR . In any representation a matrix C exist. 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.

¯ (4. s) v = . s) = Ep + M σ ·p Ep +M χs . v(p. s)u(p. s)¯(p. s)u(p. s) .66) An arbitrary spin 1/2 ﬁeld can be expanded in the independent solutions.62) where χs are two independent (s = ±) two-component spinors. ⇔ (/ + M ) v(p) = 0.4 Plane wave solutions Looking for positive energy solutions ∝ exp(−iEt) one ﬁnds two solutions. s′ ) = −2M δss′ .64) (4.70) = Ps = ˆ 0 1∓σ·s 2 2 . ∂ −iσ · ∇ −i ∂t − M (4.69) = − s In order to project out spin states. ¯ ′ ′ Explicit solutions in the standard representation are χs . s′ ) = 2M δss′ . p (4. in which γ 0 is diagonal (see discussion of negative energy states in section 4. the spin polarization vector in the rest frame is the starting ˆ point. s ) = v (p. p −σ · p (Ep − M ) Ep + M σ ·p ¯ Ep +M χs χs ¯ For a free massive particle. The explicit Dirac equation in the standard representation reads ∂ iσ · ∇ i ∂t − M (4.61) σ·p −(Ep + M ) There are also two negative energy solutions.1). The ¯ ¯ solutions are normalized to u(p. s) e−i k·x b(k. ψ(x) = u±s (p) e−i p·x .63) u(p.The Dirac equation 34 4. where u satisﬁes −σ · p Ep − M u(p) = 0. Choosing χ = −ǫχ∗ (the equivalent spin ¯ 1/2 conjugate representation). s) = u(p. s). It is easy to check that 1 1±σ·s 1 ± γ5 / s ˆ 0 (4. 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. s)v(p. where v satisﬁes σ·p −(Ep + M ) v(p) = 0. (4. s)¯(p. s ) = 0. the best representation to describe particles at rest is the standard representation. s) ei k·x d∗ (k. s). s)u(p.67) It is straightforward to ﬁnd projection operators for the positive and negative energy states P+ P− = s p /+M u(p. v (p. ¯ ¯ † ′ u (p. In that frame is a spacelike unit vector sµ = (0. s′ ) = 2Ep δss′ .65) (4. with E = Ep = + p2 + M 2 .60) ψ(x) = 0. ψ(x) = v ±s (p) ei p·x . s) u = . ⇔ (/ − M ) u(p) = 0. the spinors satisfy C uT (p. In an arbitrary frame one has s · p = 0 and s(p) can e.68) (4. s) = v(p.. s) = u(p. 2M 2M −/ + M p v(p. s) + v(k. (2π)3 2Ek (4.g. s)v(p. be obtained by a Lorentz transformation. s)v(p. s ) = v † (p. 2M 2M (4. s) and C v T (p. Note that the spinors in the negative energy modes (antiparticles) could be two diﬀerent spinors.

One ﬁnds for the helicity states in Weyl representation: √ √ 0√ E +M + E −M √ 1 √ 1 0√ 1 . λ = −1/2) = √ − E+M √ 0 0 E+M (4.74) . v(p. −1/2) = √ (4. helicity states. −p3 −p1 − i p2 1 v(p. γ5 /ψ = 0 but also p p p /γ5 ψ = −γ5 /ψ = 0. one has in standard representation: 0 E+M E+M 1 1 0 . etc. (4. The ratio of the components is (directly or inversely) proportional to √ √ E+M − E−M 2M M ≪E M √ √ √ . p3 E +M E+M 1 2 3 p + ip −p 1 2 p − ip 3 1 −p v(p. λ) = v(p. Note that for solutions of the massless Dirac equation /ψ = 0. . 1 u(p. λ). u(p. We note that for high energy ¯ ¯ the positive helicity (λ = +1/2) is in essence righthanded. Therefore. v(p. λ = −1/2) = u(p. 4. +1/2) = √ 0 E +M E+M .73) Also for helicity states C uT (p.72) ˆ Helicity states (p along z ) in Standard representation are: √ √ 0 E+M E +M . √ 0 u(p. = √ =⇒ 2E E+M + E−M ( E + M + E − M )2 which vanishes in the ultra-relativistic limit E ≫ M or in the massless case. (4. 2 2 0 √ 2 √ 2 − E+M − E−M 0 .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 ). λ = + 1 ) = √ + E +M − E −M . This means that in the solution space for massless fermions the chirality states. u(p. Explicit examples of spinors are useful to illustrate spin eigenstates. λ = +1/2) = v(p. In principle massless fermions can be described by twocomponent spinors. λ) and C uT (p. p ψR/L ≡ PR/L ψ are independent solutions. while the negative helicity (λ = −1/2) is in essence lefthanded. By writing the helicity states in Weyl representation it is easy to project out righthanded (upper components) and lefthanded (lower components). λ = + 2 ) = √ E +M − E −M 2 0√ 2 2 √ 0 E +M + E −M √ √ 0√ √ − E +M − E −M 1 1 E+M − E−M 0 √ . λ = − 1 ) = √ √ v(p. λ = +1/2) = E−M √ 0 − E −M 0 √ √ 0 − E−M − E −M 0 . λ = − 1 ) = √ E + M − E − M . For instance with the z-axis as spin quantization axis.48 replace the spin projection operators which are not deﬁned (by lack of a rest frame). λ) = u(p. chirality. −1/2) = √ E + M −(E + M ) 0 . +1/2) = √ u(p.71) p − i p2 . The chirality projection operators in Eq.

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

ν µ µ Look at the inﬁnitesimal forms Λµ = gν + ων and L = 1 − (i/2)ωρσ S ρσ to obtain ν [γ µ . −x) is also a solution of the Dirac equation. by proving that H 2 (p) = exp (η · σ) = (E + σ · p)/M = σ µ pµ /M. Show that the equality holds. (4.g.77) Exercises Excercise 4. ˜ Exercise 4. γν } = 2gµν that γµ / γ µ a = Tr(//) = ab µn µ1 Tr(γ · · · γ ) = −2 /. i S ρσ ] = g ρµ γ σ − g σµ 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 . by letting them act on the four basis spinors u(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. x → x. to the Dirac equation and use this to show that the spinor ψ p (x) = ˜ γ0 ψ(˜).5 Show starting from the relation {γµ . Do this e. s)¯(p. s).The Dirac equation 37 The trace is linear and can be split up in parts containing traces with up to four gamma-matrices. of which only the traces of four and two gamma-matrices are nonzero. 4 2 Exercise 4. H(p) = exp η·σ 2 = M +E+σ·p .g.6 Apply space-inversion. Excercise 4. where x = (t.13. e. s) u p /+M = 2M 2M both sides are projection operators.1 Prove Eq. γ σ ] = σ ρσ . 1 i −i S ρσ = − [γ ρ . x ˜ .3 Imposing Lorentz invariance on the Dirac equation can be achieved by requiring L−1 γ µ L = Λµ γ ν . 2M (E + M ) ˆ ˆ where E = M γ = M cosh(η) and p = M βγ n = M sinh(η) n.4 Show that in P+ ≡ s=±1/2 u(p. 4. a 4 a · b. s) and v(p. 0 if n is odd. Exercise 4.

5 and M → ∞. Do you understand why? The calculation of the life-time is completed in chapter 11.57 MeV. you need to determine the corresponding conﬁnement condition for ψ. Calculate for now just |M |2 . M R = 1. Show that the ratio is |M (π − → e− νe )|2 ¯ ≈ 54. Show that this n condition is suﬃcient to guarantee that nµ j µ = 0.8 A negatively charged pion (π − ) decays in a lepton ℓ (ℓ can be an electron or a muon) and a corresponding anti-neutrino (¯ℓ ). we want to conﬁne this solution to a spherical cavity with radius R. n It turns out that there is actually a whole family of conditions that provide conﬁnement.e. Note: To prove this. |M (π − → µ− νµ )|2 ¯ Actually if the lepton would be massless. Excercise 4. 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.511 MeV). is a stationary solution of the (free) Dirac equation. Note: the outcome shows that the decay into muons is much more likely that the decay into electrons (mπ = 139. t) = N k ˆ i E+M σ · r j1 (kr)χm −i Et e . The conditions for ψ and ψ imply that nµ j µ = ψ/ ψ = 0. the lepton and neutrino have spin 1/2 and are ν described by momentum-space spinors. namely i / ψ = eiαγ5 ψ. We will not pursue this chiral freedom at this point and take α = 0. r).7 (a) Show that (in standard representation) j0 (kr)χm ψ(x. A condition for conﬁning the fermion in the cavity is i / ψ = ψ (where n is the surface normal). 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. . which is the most important factor needed to ﬁnd the decay width (inverse lifetime).The Dirac equation 38 Exercise 4. The pion has spin 0. the decay would even be forbidden.66 MeV. 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 ). Calculate the value of k for M R = 0. there is no current ﬂowing through the surface of the cavity. n (c) Apply the conﬁnement condition to ﬁnd the value of k for the lowest eigenmode in the spherical ˆ cavity. For such a cavity nµ = (0.8 × 10−6 . mµ = 105. i. (d) Plot for these cases the vector densities ψ † ψ and the scalar densities ψψ as a function of r/R. ν p Note: how such parametrizations work will be explained in chapter 8. (b) Next. me = 0.

0).Chapter 5 Vector ﬁelds and Maxwell equations 5.6) where Aµ is the four-vector potential (5. M2 (5. with ∂µ V µ = 0 (Lorentz condition). 39 (5. These can describe spin 1 states in the rest-frame. These are solutions of the Klein-Gordon equation with in addition a condition. (2 + M 2 )Vµ = 0.2 we have discussed the vectors ǫλ (with λ = 0 and ±1) as basis states of the spin 1 representation. µ (2π)3 2E µ (5.1) The Lorentz condition implies that k µ ǫµ (k) = 0. −B 1 0 3 With the electromagnetic ﬁeld tensor (5.± (λ) d3 k ǫ(λ) (k) e−ik·x c(k. The vectorﬁeld is therefore (λ) suited to describe spin 1.± 5. ǫλ ).5) −E B2 .1 Fields for spin 1 In section 3. λ) + ǫ(λ)∗ (k) eik·x c∗ (k. where k = k0 = (M. In the restframe. namely ǫµ (k0 ) = (0. A). A real vector ﬁeld Vµ (x) then can be expanded in modes as Vµ (x) = λ=0. (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.4) λ=0.3) Since the vectors ǫµ (k) together with the momentum k µ form a complete set of four-vectors. there are then indeed three independent possibilities.2) (5.7) . ±(1) are obtained by boosting the restframe vectors. just as the two-component Dirac spinors did for spin 1/2. In an arbitrary frame ǫµ (k) with λ = 0. λ) . one has ǫ(λ) (k) ǫ(λ)∗ (k) = −gµν + µ ν kµ kν . The covariant generalization is to describe them with Vµ (x) = ǫµ (k) e±ik·x .

the antisymmetry of F µν implies that ˜ ∂µ F µν = 0 or ∂λ Fµν + ∂µ Fνλ + ∂ν Fλµ = 0.18) corresponding to the helicity states λ = ±1 of the massless photon ﬁeld. In that case one also has ∇·A = 0 or ki ǫi (k) = 0. transverse or Coulomb gauge).5. 40 (5. For electrodynamics one has furthermore the freedom of gauge transformations.12) (5. Furthermore. (5. 0).9) 1 ˜ where F µν ≡ − 2 ǫµνρσ Fρσ is the tensor where E → B and B → −E. j) the Maxwell equations read ∂µ F µν = j ν .15 the d’Alembertian is replaced by 2 + M 2 . 5. . for M = 0. it does not eliminate the freedom of gauge transformations but they are now restricted to Aµ → Aµ + ∂µ χ with 2χ = 0. if in Eq. ∓1.14) of which the solutions give the Li´nard-Wiechert potentials. ˜ x Note that via parity transformation one obtains another solution. (5. One possible choice is the gauge A0 = 0 (5. The equation in vacuum.e. Note that the incorporation of the Lorentz condition can also be done for the case of the massive vector ﬁeld. As remarked above. ∂µ Aµ = 0. 0. only two independent vectors remain ǫ(±) µ (k) = (0.16) (5.17) (radiation. Under a gauge transformation Aµ −→ Aµ + ∂µ χ. i.8) (5. Fµν −→ Fµν + (∂µ ∂ν χ − ∂ν ∂µ χ) = Fµν . in accordance with what we found in section 3. −i. AP (x) = Aµ (˜). (5. shows that the electromagnetic ﬁelds correspond to the case of a massless spin 1 ﬁeld. 2Aµ = 0. in which case one has 2Aµ = jµ . there is still a gauge freedom left (Aµ → Aµ + ∂µ χ with 2χ = 0). It states the absence of magnetic charges and Faraday’s law.15) (5. (5. |k|). it will vanish at all times. (5. The Lorenz condition can actually be incorporated by changing the equations of motion to 2Aµ + (λ − 1)∂µ (∂ν Aν ) = jµ . Therefore if k µ = (|k|. The equations of motion for the ﬁelds Aµ become 2Aµ − ∂µ (∂ν Aν ) = jµ . while the fact µ C ∗ that the Minkowski space is real implies that Aµ (x) = Aµ (x) = Aµ (x). 0. Although the Lorenz condition is a constraint.Vector ﬁelds and Maxwell equations and the (conserved) current j µ = (ρ. moree over. The gauge freedom can ﬁrst of all be used to impose the Lorentz condition for the electromagnetic ﬁeld.11) This equation is not aﬀected by a gauge transformation. Taking the divergence of this equiation. if ∂µ Aµ = 0 for large times |t|.10) the electric and magnetic ﬁelds are unchanged.13) Hence. one ﬁnds for a conserved current and λ = 0 2∂µ Aµ = 0.

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

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

8 and 5.Vector ﬁelds and Maxwell equations 43 Exercises Exercise 5. ∂t ∇ · B = 0.9 explicitly give the equations ∇ · E = ρ. ∇×B =j+ ∇×E+ ∂E .1 Show that Eqs. 5. . ∂t ∂B = 0.

3) (6. t1 (6. recall classical mechanics with the action t2 S= t1 dt L(x. x) ˙ (6. ∂x ˙ (6.4) 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 . thus t′ = t + δτ.8) 44 . ˙ p= ∂L .1 Euler-Lagrange equations In the previous chapter we have seen the equations for scalar ﬁelds (Klein-Gordon equation). x (t) = x(t) + δx(t). and the total change x′ (t′ ) = x(t) + ∆x(t). (6. ∂x (6. These equations can be obtained from a lagrangian using the action principle. x) = K − V = ˙ 1 mx2 − V (x). d dt ∂L ∂x ˙ = ∂L .2) The principle of minimal action looks for a stationary action under variations in the coordinates and time. Dirac ﬁelds (Dirac equation) and massless vector ﬁelds (Maxwell equations) and corresponding to these ﬁelds conserved currents describing the probability and probability current. As an example.Chapter 6 Classical lagrangian ﬁeld theory 6. ˙ 2 (6. The requirement δS = 0 with ﬁxed boundaries x(t1 ) = x1 and x(t2 ) = ˙ x2 leads to t2 ′ (6.7) The quantity ∂L/∂ x plays a special role and is known as the canonical momentum.5) with ∆x(t) = δx(t) + x(t) δτ .1) and as an example the lagrangian L(x.6) The ﬁrst term leads to the Euler-Lagrange equations.

∂µ φ(x)) = R d4 x L (φ(x). (R3 . = φ(x) + δφ(x) (6. The resulting variation of the action is δS = R µ ′ = xµ + δxµ . indicated with δF [φ]/δφ should pose no problems. ∂µ φ′ .17) Furthermore the variations δφ and δ∂µ φ contribute to δS. Variations in the action can come from the coordinates or the ﬁelds.6 can be rewritten as δS = .14) (6. .t2 ) d3 x . x′ ) − d4 x L (φ. . ∂µ φ(x)). + ∂L ∆x − H δτ ∂x ˙ t2 . . ∂µ φ. but which in 3 dimensions includes conserved quantities related to rotations and boosts. ˙ x The second term in Eq. indicated as x′µ φ (x) or combined φ′ (x′ ) = φ(x) + ∆φ(x). H is a conserved quantity.2.t1 ) (6. Consider a lagrangian density L which depends on these functions.18) 1 Taking a functional derivative.10) One sees that invariance under time translations requires that H(t1 ) = H(t2 ).11) λ=0 of which p and H are the simplest examples related to space and time translations. x). ∂µ φ(x).16) The change in variables x → x′ in the integration volume involves a surface variation of the form dσµ L δxµ . . In classical ﬁeld theory one proceeds in complete analogy but using functions depending on space and time (classical ﬁelds. i. dσµ . . gives rise to a conserved quantity (at least if H is invariant!) Q=p dx dλ −H dt dλ . which in classical mechanics vanishes at the boundary) does not play a role. ∆x = (dx/dλ)λ=0 δλ and δt = (dt/dλ)λ=0 δλ. . with ∆φ(x) = δφ(x) + (∂µ φ)δx . . = ∂R (R3 . .Classical lagrangian ﬁeld theory 45 For the lagrangian speciﬁed above this leads directly to Newton’s law p = m¨ = −∂V /∂x. − d3 x . We will come back to it in a bit more formal way in section 9. L (φ(x). . x) and an action t2 S[φ] = t1 dt L = dt d3 x L (φ(x). 6.13) (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 . ∂R Note for the speciﬁc choice of the surface for constant times t1 and t2 . their derivatives and possibly on the position. t2 ).12) Here R indicates a space-time volume bounded by (R3 . x) ≡ p x − L . In general any continuous transformation.9) which is done because the ﬁrst term (multiplying ∆x. The hamiltonian H is deﬁned by H(p. also indicated by ∂R (a more general volume in four-dimensional space-time with some boundary ∂R can also be considered). t1 (6. δ(∂µ φ) (6. λ=0 (6. (6. . ˙ (6. R (6. think for instance of a temperature or density distribution or of an electromagnetic ﬁeld).e. t1 ) and R3 .15) d4 x′ L (φ′ .

δL δL ∂µ = .24) (6.26) 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. (6. 2 ∗ (6. leading to the Euler-Lagrange equations. (6.28) where the second line is not symmetric but in the action only diﬀers from the symmetric version by a surface term (partial integration).29) .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. (2 + M )φ (x) = 0. For the complex scalar ﬁeld one conventionally uses L = = ∂µ φ∗ ∂ µ φ − M 2 φ∗ φ −φ∗ ∂µ ∂ µ + M 2 φ.19) δ(∂µ φ) δφ 6. The scalar ﬁeld It is straightforward to derive the equations of motion for a real scalar ﬁeld φ from the lagrangian densities. leading to (2 + M 2 )φ(x) = 0. (6. Using the variations in ψ (in the symmetric form). 2 (6.2 Lagrangians for spin 0. the second term is irrelevant. One easily obtains (2 + M 2 )φ(x) = 0.27) (6.Classical lagrangian ﬁeld theory 46 With for the situation of classical ﬁelds all variations of the ﬁelds and coordinates at the surface vanishing. The integrand of the ﬁrst term must vanish. δL δ(∂µ ψ) δL δψ one obtains immediately i / −M ∂ → i = − γ µψ 2 i → = ∂ / ψ − M ψ.20) (6.21) which diﬀer only by surface terms. the Dirac ﬁeld (spin 1/2) and the vector ﬁeld (spin 1) can be found.25) (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). ∂ (6.23) (6. 2 ψ = 0. L = 1 1 ∂µ φ∂ µ φ − M 2 φ2 2 2 1 µ = − φ ∂µ ∂ + M 2 φ.

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. The Majorana case is in fact more general. but note the factor 1/2 as compared to the Dirac lagrangian.31 is given by LM (Dirac) = −M ξ † η + η † ξ . that the lagrangian separates into two independent parts for M = 0. ∗ 0 ⇒ Υ ≡ −ǫ η ΥL = ψL = η η ∗ for which η0 = ǫ η0 .32) There exists another possibility to write down a mass term with only one kind of ﬁelds. The reasons for Grassmann variables will become clear in the next chapter.31) showing e.38) (6. 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 . (6.30) It is often useful to link to the two-component spinors ξ and η which we started with in chapter 4. 1 M η † ǫη ∗ − M ∗ η T ǫη .Classical lagrangian ﬁeld theory 47 and similarly from the variation with respect to ψ ψ i / +M ∂ ← = 0. (6. since a lagrangian with both Dirac and Majorana mass terms can be rewritten as the sum of two Majorana lagrangians after redeﬁning the ﬁelds (See e.35) Since the kinetic term in L separates naturally in left and right parts (or ξ and η).37) This gives an expression with a mass term that is actually of the same form as the Dirac mass term. ψL ≡ (ψL )c = C ψL = 0 (6. the mass term in the Dirac lagrangian 6.7). . 2 αβ = −βα.and lefthanded ones.g. We note that of which the left part coincides . which comes because we in essence use ’real’ spinors. ∂ ∂ 2 2 (6. Υi / Υ − ∂ 4 2 T −ǫ η ∗ c = ΥR .34) (6. Peshkin and Schroeder or Exercise 12. Using the twospinors ξ and η. 2 (6. and thus (β ∗ α)∗ = α∗ β = −βα∗ .39) 2 This term is written down with η being an anticommuting Grassmann number for which (αβ)∗ = β ∗ α∗ = −α∗ β ∗ . In that case one ﬁnds trivially L = ↔ ↔ 1 1 ψR i / ψR + ψL i / ψL − M (ψR ψL + ψL ψR ). (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 .g. namely2 LM (Majorana) = + (6.33) We now introduce a ’real’ (four-component) spinor satisfying Υc = Υ with ψL .36) (6. or equivalently separate the ﬁeld into right.

Q= d3 x J 0 (x. the presence of a symmetry that leaves the lagrangian invariant.3 Symmetries and conserved (Noether) currents Next.44) δL ∆φ + Θµν (x)δxν . implying ∂µ F µν + M 2 V ν = 0 and ∂µ V µ = 0.. requires the presence of a conserved quantity. Returning to δS the surface term is rewritten to δS = R d4 x . These do not aﬀect the dynamics and will give rise to conserved currents. .+ R 4 δL δL ∆φ − ∂ ν φ − L g µν δxν δ(∂µ φ) δ(∂µ φ) δL ∆φ − Θµν (x)δxν .Classical lagrangian ﬁeld theory 48 where Fµν = ∂µ Vν − ∂ν Vµ . t). (6. (6. In particular when the lagrangian is invariant under symmetries.42) Θµν = The variation δS thus can be expressed as δS with J µ (x) = − = δL ∂ ν φ − L g µν . t).18. δ(∂µ φ) (6. which in a consistent picture precisely generate the symmetries.45) J µ (x) = − δ(∂µ φ) dλ λ=0 dλ λ=0 Thus considering ∂R = σ2 . these conserved quantities become operators. ∂µ J µ (x) = 0. . δ(∂µ φ) For continuous transformations. δ(∂µ φ) (6.41) (6. 6. As an example for the classical ﬁeld.σ1 with σ = (R3 .43) F (σ1 ) − F (σ2 ) = dσµ J µ (x) = ∂R R d4 x ∂µ J µ (x). (6. which is the space-integral over the zero-component of a conserved current.47) . (6. one can consider the variations at the initial or ﬁnal surface.40) 6. (6. Q(t1 ) = Q(t2 ). discussed in the next chapter. In this case.. 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. ∆φ = (dφ/dλ)λ=0 δλ and δxµ = (dxµ /dλ)λ=0 δλ. consider U (1) transformations of the Dirac ﬁeld proportional to the charge e.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 . we consider the second term in Eq. ψ(x) → ei eΛ ψ(x) or ∆ψ(x) = i eΛ ψ(x). + ∂R dσµ dσµ ∂R ≡ where d x .

The integral over the zero-component.3) j µ = i e φ∗ ∂µ φ.51) (6.55) Note that the coordinate transformations can be written in a form similar to that for the ﬁelds (matrices in internal space).48) For the complex scalar ﬁeld the U (1) transformations leave the lagrangian invariant and lead to the conserved current (see Exercise 6.49) These currents are conserved as discussed already in chapter 2.53) These are the energy and momentum. 2 (6. In the next section we will see the quantity show up as the charge operator. . Therefore there are four conserved currents Θµν (ν labeling the currents!) and four conserved quantities Pν = dσµ Θµν = d3 x Θ0ν . = −(∂µ φ)δxµ = −ǫµ ∂µ φ(x).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. Lorentz transformations In this case the transformation of the coordinates and ﬁelds are written as δxµ ∆φi (x) = = ω µν xν . including e the space-time translations and the Lorentz transformations.Classical lagrangian ﬁeld theory From the lagrangian for the Dirac ﬁeld.50) (6. From Noether’s theorem one sees that the current Θµν ǫν is conserved. ↔ (6.4 Space-time symmetries One kind of symmetries that leave the lagrangian invariant are the Poincar´ transformations. (6. Translations µ Under translations (generated by Pop ) we have δxµ ∆φ(x) δφ(x) = ǫµ . (6. Summarizing. (6. For quantized ﬁelds this will become the expressions of the hamiltonian and the momentum operators in terms of the ﬁelds. t) is the conserved charge (Q = 0). e. = 0. (6. 6. (ω µν antisymmetric) 1 ωρσ (−i S ρσ )ij φj (x).g. 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.54) (6. 1 (6. we obtain (since δxµ = 0). H = d3 x Θ00 (x).57) For the ﬁelds the behavior under Lorentz transformations has been the subject of previous chapters.56) δxµ = ωρσ (aρσ )µν xν 2 with (aρσ )µν = aρσ µν = g ρµ g σν − g σµ g ρν . ˙ Q = d3 x j 0 (x.

Classical lagrangian ﬁeld theory 50 • Scalar ﬁeld φ: −i Sρσ = 0.e. γσ ] This result for the Dirac ﬁeld has been discussed in chapter 3 (see also Exercise 4.3 we have seen global transformations or gauge transformations of the ﬁrst kind. • Vector ﬁeld Aµ : −i Sρσ = aρσ .70) . 6.61) A ﬁnal note concernes the symmetry properties of Θµν .65) M µρσ = T µρ xσ − T µσ xρ .66) In section 6. where H ρµν is the quantity appearing in M ρµν .6) T µν = T νµ . Deﬁning T µν = Θµν − ∂ρ Gρµν . the angle of the transformation depends on the space-time point x. 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 . one has a tensor that satisﬁes (exercise 6.64) (6.59) (6.67) i. (6.g. e.58) (6. independent of x. φ(x). In some applications (speciﬁcally coupling to gravity) it is advantageous to have an equivalent current that is symmetric. • Dirac ﬁeld ψ: −i Sρσ = −(1/4)[γρ . is not invariant under local gauge transformations.3). The lagrangians which we have considered sofar are invariant under global gauge transformations and corresponding to this there exist a conserved Noether current.62) with Gρµν = (H ρµν + H µνρ + H νµρ )/2.68) (6. Any lagrangian containing derivatives.5 (Abelian) gauge theories φ(x) → ei eΛ φ(x).69) i eΛ(x) φ (x). however. in which the U (1) transformation involves an angle e Λ. In general this tensor is not symmetric. (6.63) µν = ∂µ Θ . (6. (6. 2 (6. ∂µ φ(x) + i e ∂µ Λ(x) e (6. Gauge transformations of the second kind or local gauge transformations are transformations of the type φ(x) → ei eΛ(x) φ(x).60) Therefore there are six conserved currents M µρσ labeled by ρ and σ (antisymmetric) and corresponding to it there exist conserved quantities M ρσ = d3 x M 0ρσ . (6. φ(x) φ (x) ∂µ φ(x) ∗ → e → e → ei eΛ(x) φ(x). ∂µ T µν (6. −i eΛ(x) ∗ i eΛ(x) (6.

73) (6.74) 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 . The equation of motion for the fermion follow from δL δ(∂µ ψ) δL δψ = = i − γ µ ψ. 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.71) Thus the covariant derivative has the correct transformation behavior provided Aµ → Aµ − ∂µ Λ.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).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 . 1 L (φ. ∂ 2 4 (6. As an example consider the Dirac lagrangian.72) (6. (6. 2 i ∂ / ψ − M ψ − e/ ψ A 2 giving the Dirac equation in an electromagnetic ﬁeld. (i/ − M ) ψ = (i/ − e/ − M ) ψ = 0.75) The ﬁeld φ is used here in a general sense standing for any possible ﬁeld. 4 (6. i L = ψγ µ ∂µ ψ − (∂µ ψ)γ µ ψ − M ψψ. 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. (6. Dµ φ(x) ≡ (∂µ + i eAµ (x))φ(x). Dµ φ) − Fµν F µν . 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 . −e j µ Aµ = −e ρφ + e j · A.78) . ∂µ φ) =⇒ L (φ. D ∂ A (6.Classical lagrangian ﬁeld theory 51 where the last term spoils gauge invariance. For this purpose it is necessary to introduce a vector ﬁeld Aµ . (6. A solution is the one known as minimal substitution in which the derivative ∂µ is replaced by a covariant derivative Dµ which satisﬁes Dµ φ(x) → ei eΛ(x) Dµ φ(x).

−eψγ ν ψ. ∂µ F µν = j ν . ∂µ F µν = j ν .e. giving the Maxwell equation coupling to the electromagnetic current. [iγ µ (∂µ − i e Aµ ) − M ]ψ = 0.1 (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 .4 Given the Dirac equation for a (negatively charged) spin-1/2 particle in an external electromagnetic T ﬁeld. where j µ = e ψγ µ ψ. give the equation which is satisﬁed by ψ c = Cψ . ∂µ j µ = 0 (Note that the addition of a total derivative to the lagrangian density does not modify the equations of motion). ↔ Exercise 6. Exercise 6.79) Exercises Exercise 6.2 (a) Show that the homogeneous Maxwell equations can be derived from the lagrangian density 1 L = − Fµν F µν . Exercise 6. (c) Show that the interaction term is invariant under gauge transformations only if the current jµ is conserved.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.Classical lagrangian ﬁeld theory 52 For the photon the equations of motion follow from δL δ(∂µ Aν ) δL δAν = = −F µν . . i. 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. and deduce what is the charge of the antiparticle as compared to a particle. (6. 4 (b) What is the form of the interaction term involving a current jµ and the ﬁeld Aµ that will give the inhomogeneous Maxwell equations. φ → eiΛ φ.

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. t). in order to show that Θρσ − Θσρ = ∂µ H µρσ . . e 1 (−i∇ + eA)2 + σ · B − eA0 ψu (r. 2M 2M with Enr ≡ E − M ≪ M . Hints: it may be useful to realize that Gµρσ is antisymmetric in ﬁrst two indices.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 .4. use ∂µ T µν = 0 and ∂µ M µρσ = 0.6 (optional) Prove the properties of the tensor T µν in section 6. This relation is known as the Pauli equation. Comment: this relation is known as the Pauli equation. 2me Exercise 6. Hint ≡ −gs µe · B = −gs Qe s · B. A) (see also exercise 6. ﬁnally note that it is suﬃcient to show the last equation for dσµ M µρσ .Classical lagrangian ﬁeld theory 53 Exercise 6. t) = Enr ψu (r.5 Show that the Dirac equation for an electron with charge −e in an external electromagnetic ﬁeld Aµ = (A0 .

starting from the lagrangian L(q.11) found through Noether’s theorem. labeled by the position. (7.5) ∆ x ∆3 x 54 . The bilinear product for the conserved quantities.2) H(p. [q. p]P = 1. ˙ L(q. t). q) also considered in the previous chapter. q) = ˙ 1 mq 2 − V (q). H]P dt and dO = [O. q) ≈ (dO/dλ)λ=0 δλ under symmetry transformations one has dO = [O. p)) = ˙ ˙ 2m where the (canonical) momentum p = ∂L/∂ q. Actually the conserved quantities satisfy a similar algebra as the Lie algebras for continuous symmetries that they are obtained from. dq The basic Poisson bracket is actually the one between q and its canonical momentum p.Chapter 7 Quantization of ﬁelds 7. This is nothing else as representing the classical Poisson algebra as operators in the Hilbert space. there are conserved ˙ ˙ quantities Q (Eq. q(q. dq dp dq dp (7.1) The Hamiltonian (also corresponding to a conserved quantity because of time translation invariance) is given by p2 + V (q). namely [q. ˙ 2 (7. Further. Q]P . however. 1 qx (t) = 3 d3 x φ(x. (7. 6. q) = p q − L(q. (7. An immediate generalization for ﬁelds can be obtained by considering them as coordinates. in our example p = mq. Quantization of the theory is achieved by imposing canonical commutation relations between q and p. is the Poisson bracket [A. p] = i. p]P .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. dλ Examples are dO = [O.1 Canonical quantization We will ﬁrst recall the example of classical mechanics for one coordinate q(t).3) For variations δO(p. B]P = dB dA dA dB − .

t)] = 0. This implies speciﬁc transformation properties in the Hilbert space for the quantum ﬁelds φ. • Poincar´ invariance e The full variation of the ﬁelds. ˙ δ φ(x) (7. φ(y. ∂ qx ˙ ∂ qx ˙ (7.12) (7. (7. discussed in the section 6.15) with furthermore the relations [φ(x. for the scalar ﬁeld theory. Sometimes it is useful to keep in mind that. t). Φ(f ) = d4 x φ(x) f (x). qx .11) (7. 2 2 2 (7. As an example. t). δ∂0 φ) 1 1 1 = Θ00 (x) = (∂0 φ)2 + (∇φ)2 + M 2 φ2 . ∂µ φ(x)) = ˙ d3 x L (φ. t)] = [Π(x. ˙ where Πx (t) is obtained from the continuous ﬁeld Π(x) ≡ δL /δ(∂0 φ).13) Π(x) H (x) For the quantization procedure we can formulate a number of basic axioms of quantum ﬁeld theory. t).14) In fact. px (t) = ∂L ∂Lx = ∆3 x = ∆3 x Πx (t). (7. ∇φ) (7.10) x ∆3 x [Πx qx − L x ] ˙ ∆3 x Hx (7.Quantization of ﬁelds 55 etc. ∆φ. Π(y. Π(y. i U † (Λ. for symmetry transformations.6) ∆3 x L x (qx . The lagrangian is given by L = = x In order to construct the hamiltonian it is necessary to ﬁnd the canonical momenta. formally.2.9) Note that this indeed corresponds to the zero-zero component (Θ00 ) of the conserved energy-momentum stress tensor Θµν . • Canonical commutation relations The quantization of the theory is achieved by imposing the canonical quantization condition [qx (t). the ﬁelds φ(x) can be considered as regular operators in the Hilbert space after smearing with a test function f .16) . albeit with ’sharp’ functions. a) = Rj (Λ−1 ) φj (Λ−1 x − a).7) d3 x L (φ(x). (7. The hamiltonian then is H= x px qx − L = ˙ = x where H (x) = δL ˙ φ(x) − L (x) = Θ00 (x).8) (7. py (t)] = i δxy or for the ﬁelds φ(x) and Π(x) the socalled equal time commutation relations [φ(x. the discretization procedure above is an explicit example. φ. 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. The essential conditions to consistently quantize a theory then are the following. t)] = i δ 3 (x − y). a) φi (x)U (Λ. qx+∆x ) .

1 Q = √ (a + a† ) 2ω and P = −i ω (a − a† ) 2 (7. The measurements cannot inﬂuence each other or [Φ(f ).23) The hamiltonian in this case can be expressed in the number operator N = a† a. [φ(x). (7.19) 7. and [N. a] = −a.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.21) Writing Q and P in terms of creation (a† ) and annihilation (a) operators. [a.17) (Sµν )i φj (x). Q] = [P. (7.24) It is straightforward to ﬁnd the commutation relations between N and a and a† . In simpliﬁed form the hamiltonian is given by H= 1 2 1 2 2 P + ω Q . i(xµ ∂ν − xν ∂µ )φ (x) + i (7. Microscopic causality implies local commutativity. a† ] = 1. a† ] = 0.25) . 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. φ(y)] = 0 if (x − y)2 < 0. P ] 2 (7. j (7.Quantization of ﬁelds 56 or if U (Λ. P ] = i.4. Φ(g)] = 0. Mµν ] = i i∂µ φi (x). (7. a† ] = a† . sometimes referred to as second quantization. ω a† a + 2 2 ω i − [Q. P ] = 0 (7. 2 2 (7. • Causality Operators Φ(f ) and Φ(g) for which the test functions are space-like separated can be measured simultaneously (macroscopic causality). Pµ ] = [φ (x). 2 i ωµν (S µν )i .20) where the coordinate Q and the momentum P satisfy the canonical commutation relations [Q.18) with S µν given in 6. a) = i Rj (Λ−1 ) = 1 + iǫµ P µ − 1− i ωµν M µν . a] = [a† . 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. H = = ω ω P P Q−i √ Q+i√ 2 2 2ω 2ω 1 1 =ω N+ . [Q. [N.

which represents the ’number of particles’ with momentum k. From the normalizations one obtains a |n = √ n + 1 |n + 1 and a|n = n |n − 1 . e. φ(x) = (2π)3 2Ek where the Fourier coeﬃcients a(k) and a† (k) are now operators. a† (k ′ )] = (2π)3 2Ek δ 3 (k − k′ ) [a(k). † √ a and a act as raising and lowering operators. the real scalar ﬁeld φ(x) becomes d3 k (7. and we see that a state |0 must exist for which N |0 = a|0 = 0.28) (7.e. φ(x )]x0 =x′0 = [Π(x). which we have split into positive and negative energy pieces with (complex) coeﬃcients a(k) and a∗ (k) multiplying them.3 The real scalar ﬁeld We have expanded the (classical) ﬁeld in plane wave solutions.29) It is straightforward to prove that the equal time commutation relations between φ(x) and Π(x′ ) are equivalent with ’harmonic oscillator .32) (7. a(k ′ )] = [a† (k).30) [φ(x). is equivalent with [a(k).4 for the classical ﬁeld) a(k) a† (k ′ ) = = d3 x ei k·x i ∂0 φ(x).26) a(k) e−i k·x + a† (k) ei k·x . 7.g.33) .Quantization of ﬁelds 57 Deﬁning states |n as eigenstates of N with eigenvalue n. i. [φ(x). (7. H = = = d3 x 1 1 1 (∂0 φ)2 + (∇φ)2 + M 2 φ2 2 2 2 d3 k Ek † a (k)a(k) + a(k)a† (k) (2π)3 2Ek 2 d3 k Ek N (k) + Evac . Π(x′ )]x0 =x′0 = i δ 3 (x − x′ ) ′ ′ and (7. The quantization of the ﬁeld is achieved by quantizing the coeﬃcients in the Fourier expansion. Π(x )]x0 =x′0 = 0. The canonical momentum becomes −i d3 k ˙ Π(x) = φ(x) = a(k) e−i k·x − a† (k) ei k·x .31) The hamiltonian can be rewritten in terms of a number operator N (k) = N (k) = a† (k)a(k).27) 2 (2π)3 It is easy to check that these equations can be inverted (see Exercise 2. d3 x φ(x) i ∂0 e−i k ·x ↔ ′ ↔ (7.like’ commutation relations between a(k) and a† (k ′ ). Note that we will often write a(k) or a† (k). (n − 1) a|n † i.e. N |n = n|n one ﬁnds N a† |n N a|n = = (n + 1) a† |n . (2π)3 2Ek (7. a† (k ′ )] = 0. but one needs to realize that in that case k 0 = Ek = k2 + M 2 . 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. and (7.

.45) (7.44) p|φ− (x)|0 = ei p·x . |0 .42) (7. . For the purpose of normal ordering it is convenient to decompose the ﬁeld in positive and negative frequency parts. k 0 = Ek = k2 + M 2 ). d3 k |k k|. (2π)3 2Ek (7. This term will be adressed below.40) This procedure is known as normal ordering. . |k = a† (k)|0 . 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).39) The problem with the vacuum or zero-point energy. The rest of the states are then obtained from the groundstate via the creation operator. φ(x) φ+ (x) φ− (x) = = = φ+ (x) + φ− (x).e. The 1-particle wave function e−ip·x is obtained via 0|φ(x)|p p|φ(x)|0 = = 0|φ+ (x)|p = d3 k 0|a(k)a† (p)|0 e−i k·x = e−i p·x (2π)3 2Ek (7. which amounts to redeﬁning H as H= d3 x : H (x) := d3 k Ek N (k). a(k)|0 = 0. (2π)3 2Ek (7.41) (7. is solved by subtracting it as an (inﬁnite) constant. .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).35) where the vacuum contribution disappears because of rotational symmetry. in ﬁeld theory also referred to as vacuum energy Evac = 1 V 2 d3 k Ek . (2π)3 (7. which now contains an inﬁnite number of oscillators. d3 k a(k) e−i k·x .37) (7. writing all annihilation operators to the right of the creation operators. deﬁning particle states |k = |k (with positive energy. = normalized as (a† (k1 ))n1 (a† (k2 ))n2 √ √ . (2π)3 2Ek d3 k a† (k) ei k·x . .36) and satisfying the completeness condition 1 = k|k ′ = (2π)3 2Ek δ 3 (k − k′ ).Quantization of ﬁelds 58 where the necessity to commute a(k)a† (k) (as in the case of the quantum mechanics case) leads to a zero-point energy. (2π)3 2Ek (7. assuring that the vacuum (by deﬁnition) has eigenvalue 0!.46) (7. n1 ! n2 ! (7. (2π)3 2Ek (7. and multiparticle states |(k1 )n1 (k2 )n2 .34) where V = (2π)3 δ 3 (0) is the space-volume. i. 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).38) d4 k 2π δ(k 2 − M 2 ) θ(k 0 ) |k k| = (2π)4 (7.

x) = 0 and hence ∆(x) = 0 for x2 < 0. a† (k ′ )] 3 2E ′ (2π) k [φ− (x). The last item to be checked for the scalar ﬁeld are the causality condition.51) (7.55) d4 k ǫ(k 0 ) 2π δ(k 2 − M 2 ) e−i k·x .53) t=0 .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. (7. (iv) The equal time commutation relations follow from ∂ ∆(x) ∂t (v) For M = 0.18). (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. that they satisfy the required commutation relations in Eqs (7.56) = −δ 3 (x). (iii) ∆(0.49) (7.e. d3 k = (2π)3 2Ek = (7. φ− (y)] ≡ i∆+ (x − y). ∆(x) = − ǫ(x0 ) δ(x2 ).47) ′ d3 k ′ e−i k·x+i k ·y [a(k).52) The result for the invariant commutator function is [φ(x).17) and (7. 2π (7.50) or as integrals over d4 k.48) (7. φ+ (y)] d3 k e−i k·(x−y) . φ(y)] consider µν [φ+ (x). (2π)3 2Ek ≡ i∆− (x − y). In order to calculate [φ(x). = − (2π)3 2Ek (7. i∆+ (x) i∆− (x) = = = d4 k θ(k 0 ) 2π δ(k 2 − M 2 ) e−i k·x .54) = i∆(x − y) = i (∆+ (x − y) + ∆− (x − y)) . (ii) ∆(x) is a solution of the homogeneous Klein-Gordon equation. i. (2π)4 (7. (7. d3 k ei k·(x−y) = −i∆+ (y − x). 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 ).

(2π)3 2Ek (7.66) . we will consider it as a repetition of the quantization procedure.60) = i φ∗ ∂µ φ. 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) . b† (k ′ )] = (2π)3 2Ek δ 3 (k − k′ ). a† (k ′ )] = [b(k).59) (7.58) and satisfy the equal time commutation relation (only nonzero ones) [φ(x).57) (7. which is equivalent to the relations (only nonzero ones) [a(k). ↔ (7. = ∂{µ φ∗ ∂ν} φ − L gµν .61) (7.Quantization of ﬁelds 60 7. 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 . (7.64) The commutator of φ and φ† is as for the real ﬁeld given by [φ(x). φ† (y)] = i∆(x − y). ∂0 φ† (y)]x0 =y0 = i δ 3 (x − y). 2 From the lagrangian density (7.63) i.4 The complex scalar ﬁeld In spite of the similarity with the case of the real ﬁeld. (2π)3 2Ek (7. 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) . Also the charge operator requires normal ordering (in order to give the vacuum eigenvalue zero).65) 7. (2π)3 2Ek d3 k b(k) e−i k·x + a† (k) ei k·x .e.62) (7. extending it with the charge operator and the introduction of particle and antiparticle operators.5 The Dirac ﬁeld L = ↔ i ψγ µ ∂µ ψ − M ψψ. (2π)3 2Ek (7. particles (created by a† ) and antiparticles (created by b† ) with the same momentum contribute equally to the energy.

e. † {ψi (x).67) i ψ / ψ − M ψψ gµν . (7. s)¯(k. s)e−i k·x + d† (k. s)ei k·x . s)¯(k. i.74) (7. {b(k.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) : . (2π)3 2Ek (7. d† (k ′ . ˙ δ ψ(x) ↔ i Θ00 (x) = − ψγ i ∂i ψ + M ψψ 2 = iψγ 0 ∂0 ψ = iψ † ∂0 ψ. s) ei k·(x−y) x0 =y 0 . .77) Note that achieving normal ordering.69) (7. ψj (y)}x0 =y0 = d3 k (2π)3 2Ek ui (k. s)u† (k.79) Also for the ﬁeld and the canonical conjugate momentum anticommutation relations are considered. s) e−i k·(x−y) j s † + vi (k. u v (2π)3 2Ek (7. s′ )} = {d(k.Quantization of ﬁelds 61 the conserved density and energy-momentum currents are easily obtained. The solution is the introduction of anticommutation relations. (2π)3 2Ek (7. ↔ i ψγµ ∂ν ψ − 2 (7. jµ Θµν = = ψγµ ψ. s)vj (k. interchanging creation and annihilation operators. 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) . ∂ 2 ↔ (7.70) where the last line is obtained by using the Dirac equation.75) (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)v(k.78) (7. (2π)3 2Ek d3 k b† (k. s). s)e−i k·x .68) The canonical momentum and the hamiltonian are given by Π(x) H (x) = = δL = i ψ † (x). s′ )} = (2π)3 2Ek δ 3 (k − k′ ) δss′ . s). b† (k ′ . (7. s)u(k.71) (7. s)ei k·x + d(k. The quantized ﬁelds are written ψ(x) = s d3 k b(k.73) (7.

Quantization of ﬁelds 62 Using the positive and negative energy projection operators discussed in section 4. ∂ (7. When we would have started with commutation relations for the ﬁeld ψ and the canonical momentum.80) 3 = δ 3 (x − y) δij . but has the problem of being noncovariant.88) . It is possible to continue in a covariant way with the lagrangian 1 λ L = − Fµν F µν − (∂µ Aµ )2 . one has † {ψi (x). as the vanishing of Π0 induces a constraint.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. 7. ψ j (y)] = (i/x + M )ij i∆1 (x − y). For the scalar combination {ψi (x).6 The electromagnetic ﬁeld 1 L = − Fµν F µν .87) If one wants to impose canonical commutation relations ∂µ Aµ = 0 cannot hold as an operator identity. 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. ˙ δ Ai (7. ∂ (7.84) (7. ψ j (y)} at arbitrary times one has {ψi (x). 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. but we must restrict ourselves to the weaker condition B|∂µ Aµ |A = 0. −λ(∂ρ Aρ ). 4 2 This gives the equations of motion discussed before. 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). ˙ δ A0 δL = F i0 = E i .82) where i∆1 = i∆+ −i∆− .86) (7.81) where i∆ = i∆+ + i∆− is the same invariant commutator function as encountered before. 4 Π0 Πi δL = 0.85) which reﬂects the gauge freedom. (7. we would have obtained [ψi (x).

Aν (y)]x0 =y0 = i gµν δ 3 (x − y).e. 6.11. (7. and [ℓi .90) where Πν = F ν0 − λ g ν0 (∂ρ Aρ ) and we have furthermore [Aµ (x). These problems are solved by the Lorentz constraint between physical states given above. λ)ei k·x . q) that dO/dλ = [O. pj ]P . ′ (7. λ).e. k 3 ) this reads |k 3 |c(0) (k) − k 3 c(3) (k) |A = 0 c(0) (k) ∓ c(3) (k) |A = 0. ˙ ˙ .Quantization of ﬁelds 63 for physical states |A and |B . In fact the commutation relations imply ˙ [Aµ (x). λ′ )] = −g λλ 2Ek (2π)3 δ 3 (k − k′ ). c† (k ′ . [ℓi . The canonical equal time commutation relations are [Aµ (x).3) satisfy the conditions of a Lie algebra. µ λ=0 (7. and are equivalent with [c(k. 0.92) 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. 0. λ) − c† (k. (7.89) with four independent vectors ǫµ . where Q is the conserved quantity in Eq. where Aµ + + is the part of the vector ﬁeld containing the annihilation operators.95) Exercises Exercise 7. Q]P . λ)|A = 0. Πν (y)]x0 =y0 = i gµν δ 3 (x − y). for which it is suﬃcient that ∂µ Aµ |A = 0. j = 1. rj ]P . It gives 3 k µ ǫ(λ) (k)c(k. i.94) Choosing k µ = (|k 3 |. 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. Πν (y)] = 0. λ)c(k. pj ]P .91) (7. λ)e−i k·x + c† (k. containing a time-like photon. µ λ=0 (7. 2. 0)c(k. The quantized ﬁeld is expanded as Aµ (x) = d3 k (2π)3 2Ek (λ) 3 ǫ(λ) (k) c(k. a longitudinal photon and two transverse photons. what are [ri . 0) . (d) Use Poisson brackets to calculate q and p. (b) Proof for a quantity O(p. (c) Calculate the Poisson brackets for the coordinates x and the standard conserved quantities p and ℓ.1 (optional) If you haven’t seen Poisson brackets in classical mechanics you should (a) Proof that Poisson brackets (Eq. (7. which reproduce the Hamilton equations. Aν (y)] = [Πµ (x).93) which does exhibit problems with the time-like photon. one time-like photon by itself is not allowed! This solves the problems with the normalization and the negative energies. 7. i.

. ∆hom (Λx) = ∆hom (x) if f is an invariant function. (e) Show that ∆R (x) = 0 if x0 < 0. (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. Is the subtraction of ’zero point’ contributions essential in this check? µ Pµ φ(x) ei a µ Pµ = φ(x − a) Exercise 7. (Similarly one can show that for the advanced Green’s function ∆A (x) = 0 if x0 > 0). (b) Check the above commutation relation [φ(x). Pµ ] = i∂µ φ(x).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. f (Λk) = f (k) for Lorentz transformations Λ. (2 + M 2 ) ∆inhom (x) = −δ 4 (x). Pµ ] for the (real) scalar ﬁeld using the expressions for the ﬁelds and momentum operator in terms of creation and annihilation operators.3 (a) Show that the correct implimentation of symmetries in the Hilbert space. we can write for ∆R an integral where the k 0 integration remains along the real axis. It is invariant. Give also the expression for ∆C . (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.Quantization of ﬁelds 64 Exercise 7. e−i a requires [φ(x). ∆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 + M 2 ) ∆hom (x) = 0.2 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.

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

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

x It is straightforward to apply this to the explicit ﬁeld operator ψ(x) using γ0 u(k. −λ) e−ik·x − d† (k.4) (8. .12) and for the vector ﬁeld −1 Aµ (x) −→ Pop Aµ (x) Pop = −Aµ (˜). m). (i˜ µ ∂µ − M ) ψ(˜) = 0. λ)v(k. −λ)u(k. m) = = ˜ u(k. if helicity λ is used instead of the z-component of the spin m. x Therefore γ0 ψ(˜) is again a solution of the Dirac equation and we have x ψ p (x) = γ0 ψ(˜).2 Charge conjugation We have already seen the particle-antiparticle symmetry with under what we will call charge conjugation the behavior T (8. λ)v(k.e. λ)γ0 v(k. one can also consider the scalar ﬁeld and vector ﬁelds. −λ) eik·x 3 2E (2π) ˜ k d3 k ˜ ˜ ηP b(k. iγ µ† ∂µ − M ψ(˜) = 0. λ) Pop = ηP b(k. In the same way as the Fermion ﬁeld. λ)γ0 u(k. −λ) eik·x (2π)3 2Ek ˜ d3 k ˜ ˜ ˜ ˜ ηP b(k. m). the above operation reverses the sign of λ).9) (8.10) i. −λ) e−ik·x − d† (k.13) ψ(x) −→ ψ c (x) = ηC C ψ (x). λ)u(k. γ x (8.8) 3 = λ = λ = λ From this one sees immediately that −1 ˜ Pop b(k. λ) eik·˜ (2π)3 2Ek d3 k ˜ ˜ ˜ ˜ ηP b(k.7) = λ d k x x ηP b(k.5) (8. λ) e−ik·x − d† (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). m) γ0 v(k.11) (8. The result is ψ p (x) −1 = Pop ψ(x) Pop = ηP γ0 ψ(˜) x (8. 8. λ) P −1 = −η ∗ d(k. choosing ηP is real (ηP = ±1) particle and antiparticle have opposite parity. −λ). ˜ −v(k. −λ)v(k. ˜ Pop d(k. For the scalar ﬁeld we have seen −1 φ(x) −→ Pop φ(x) Pop = ηP φ(˜). x (8. λ) eik·x . λ) e−ik·˜ + d† (k. op P (8. x The latter behavior of the vector ﬁeld will be discussed further below. λ)u(k.6) (check this for the standard representation.3) (8. x µ (iγ ∂µ − M ) γ0 ψ(˜) = 0. (2π)3 2Ek (8.

u(k. to ﬁnd ψ t (−˜) that is a solution of the Dirac equation.3 Time reversal xµ = (t. 8. x (8. (i˜ µ∗ ∂µ − M ) ψ(−˜) = 0. x x (8. λ)u(k. r). as discussed in section 4). (8.22) = Aψ|Aφ ∗ is antilinear if A(λ|φ + µ|ψ ) = λ∗ A|φ + µ∗ A|ψ . λ)C uT (k. m). r) −→ −˜µ ≡ −xµ = (−t.20) The time reversal operator transforms We will again consider the transformation properties for a fermion ﬁeld ψ(x). (2π)3 2Ek (8. m) ¯ = = v(k. . we start with the complex conjugated x Dirac equation for ψ. ˜ x −(iγ µ )∗ ∂µ − M ψ(−˜) = 0. λ) e−ik·x + b† (k. λ) e−ik·x + b† (k. Therefore ψ c (x) = = λ −1 ¯ Cop ψ(x) Cop = ηC C ψ T (x) (8.Discrete symmetries 68 the latter being also a solution of the Dirac equation.21) where A is a 4 × 4 matrix acting in the spinor space. λ) eik·x . i.14) (8.18) (8. it is antilinear1 .e. antilinear operator is anti-unitary if A† = A−1 . writing −1 ψ(x) −→ Top ψ(x)Top = ηT Aψ(−˜) ≡ ψ t (−˜). (8.16) d k ηC d(k.15) (where one must be aware of the choice of spinors made in the expansion.19) Cop d(k. λ)v(k. Norm conservation requires Top to be anti-unitary2 . λ) eik·x ¯ ¯ (2π)3 2Ek d3 k ηC d(k. λ) Cop = ηC d(k. m). x (8. x µ (iγ ∂µ − M ) γ5 Cψ(−˜) = 0. λ). The action on the spinors (using C = iγ 2 γ 0 = −iρ1 σ 2 in standard representation) gives C uT (k. m) ¯ C v T (k. λ) −1 Cop = ∗ ηC b(k. γ x i(γ5 C)−1 γ µ γ5 C∂µ − M ψ(−˜) = 0. x 1A 2 An −iC −1 γ µ C∂µ − M ψ(−˜) = 0. For a quantized ﬁeld one has −1 ∗ −1 Top fk (x)bk Top = fk (x)Top bk Top . One has Aφ|Aψ = φ|ψ ∗ = ψ|A† Aφ = ψ|φ . The same relations hold for helicity states. λ)C v T (k. x iγ µT ∂µ − M ψ(−˜) = 0. ((iγ µ )∗ ∂µ − M ) ψ(x) = 0.17) 3 = λ This shows that −1 Cop b(k. The (time-reversed) Dirac equation becomes. Top |φ = φt | = (|φt )∗ . As time reversal will transform ’bra’ into ’ket’. λ).

−p. there are 16 independent of these bilinear combinations. Therefore one has A|k = A|b† (k)|0 = A|T † T b† (k) T † T |0 ∗ = At |b† (k)|0 ∗ ˜ t = k|At . −λ |¯. 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) eik·x + d† (k. λ) e−ik·x (2π)3 2Ek (8. λ) e−ik·x 3 2E (2π) k d3 k ˜ ˜ ηT b(k. λ) e−ik·x 3 2E (2π) k ˜ d3 k ˜ ˜ ˜ ˜ ηT b(k. λ| a. ˜ Top d(k.33) (8.4 Bi-linear combinations In quantities such as currents and lagrangians often bilinear combinations of spinor ﬁelds are encountered. λ a P |a. λ).30) (8. λ| ¯ C |¯. λ)iγ5 Cv(k. ˜ = 0|b(k)|A 8.24) (8.29) In table 2 the behavior of particle states under the various transformations has been summarized. −p. Since there are 16 independent 4 × 4 matrices. λ) i γ5 C v(k.27) 3 = λ = λ = λ From this one obtains −1 ˜ Top b(k. λ) eik·˜ + d† (k. a state |a. −p. p. λ)v ∗ (k. λ). −λ a T a. λ)v ∗ (k. λ) ψ t (x) −1 Top ψ(x) Top = i ηT γ5 Cψ(−˜) x = = ˜ u∗ (k. op T (8. 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(−˜).31) (8. λ) T −1 = η ∗ d(k.34) . ∗ ˜ v (k. The result is = = λ (8. p. λ)u∗ (k. λ)iγ5 Cu(k.28) (8. λ) e−ik·˜ (2π)3 2Ek d3 k ˜ ˜ ˜ ˜ ηT b(k.Discrete symmetries 69 Table 8. p.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)v ∗ (k. (8. p. −p.25) (check this for the standard representation). x (8. λ).32) (8. λ) Top = ηT b(k.2: The transformation properties of physical states for particles (a) and antiparticles (¯). λ). λ a |a. λ)u∗ (k. λ |¯. 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. λ) eik·x + d† (k. λ) eik·x + d† (k. λ)u∗ (k.

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

e. µ p ˜ (8. • Parity p′ |V µ (x)|p = ei q·x u(p′ )Γµ (p′ . p)u(p) ¯ † † = p′ |Pop Pop V µ (x)Pop Pop |p ˜x = p′ |Vµ (˜)|˜ = ei q ·˜ u(˜′ )Γµ (˜′ . the Gordon decomposition u(p′)γ µ u(p) = ¯ 1 u(p′ ) [(p′ + p)µ + iσ µν qν ] u(p). • Hermiticity p′ |V µ (0)|p = p′ |V µ† (0)|p = u(p′ )Γµ (p′ . p) = γ0 Γµ (˜′ . p)(iγ5 C)u(p). p′ )u(p′ ) = u† (p′ )Γµ† (p. ¯ µ p ˜ Therefore time reversal invariance implies for the structure of Γµ (p′ . p ˜ • Time reversal p′ |V µ (x)|p = ei q·x u(p′ )Γµ (p′ . p)γ0 u(p). e. p). ¯ 2M (8.45) Exercises Exercise 8. p) that Γµ (p′ . or relations based on hermiticity of the operator or P . p′ )γ0 = Γµ (p′ . p) = (i γ5 C)Γ∗ (˜′ . Therefore hermiticity implies for the structure of Γµ (p′ .g. or relations that follow from the equations of motion. p)γ0 . p′ )γ0 u(p). p)u(p) ¯ † † = p′ |Top Top V µ (x)Top Top |p ∗ (8. p)(i γ5 C). p) that Γµ (p′ .g. γ5 σµν = (1/2)ǫµνρσ γ ρ γ σ . p′ )u(p′ )) u = ∗ † u† (p)γ0 Γµ (p.44) = p′ |Vµ (−˜)|˜ ˜ x p ∗ ˜x = ei q ·˜ [¯(˜′ )]∗ Γ∗ (˜′ .42) where q = p′ − p. p)u(˜) ˜ x p ¯ p p ˜ p (8.43) = ei q·x u(p′ )γ0 Γµ (˜′ . C and T invariance. p)u∗ (˜) up p µ p ˜ = ei q·x u(p′ ) (iγ5 C)Γ∗ (˜′ . p) that γ0 Γµ† (p.Discrete symmetries 71 but that are eliminated because of relations between Dirac matrices. ¯ p ˜ Therefore parity invariance implies for the structure of Γµ (p′ . p)u(p) ¯ = p|V µ (0)|p′ ∗ = (¯(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 ) + ′ .

2 (a) Calculate the current matrix element (using the explicit fermion spinors) in the Breit-frame in which q = (0. Exercise 8. −|q|/2). p = (E. L int = ejµ Aµ the charge is given by e GE (0) and the magnetic moment by e GM (0)/2M . (c) Show that hermiticity of the current requires that F1 en F2 are real. 0. 0. p′ = (E. 4M 2 = F1 + F2 . 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 . Exercise 8. (b) Show that in the interaction with the electromagnetic ﬁeld.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 .3 (a) Show that because of parity conservation no term of the form γ5 can appear in the current matrix element. (c) Show that if a term of this form would exist. 0. |q|). (b) Show that such a term is also not allowed by time-reversal symmetry. 0. it would correspond to an electric dipole moment d = −e F3 (0)/2M (Interaction term d σ · E). GE GM = F1 − Q2 F2 . σµν q ν F3 (q 2 ) 2M . 0.

2) (ii) Heisenberg picture. QH (t)|q. and the operators are time-dependent. t0 ) AH (t0 ) U (t. i ∂ |ψS ∂t ∂ i AS ∂t = H |ψS . in which the operators are time-independent. AH (t) = U −1 (t. AS (t) = AS and the states o are time dependent.1) The time evolution from t0 → t of a quantum mechanical system is generated by the Hamiltonian. t ′ |q . i ∂ |ψH ∂t ∂ i AH ∂t ≡ = 0. (9.4) ∂ U (t. t . H]. |ψH (t) = |ψH . t0 ) = H U (t. t ≡ q|q. t′ .Chapter 9 Path integrals and quantum mechanics 9. t0 ) = e−i(t−t0 )H or i Two situations can be distinguished: (i) Schr¨dinger picture.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. t0 )|ψS (t0 ) . t0 ) ∂t (9. (9. (9. Consider the two (time-independent) Heisenberg states: |q. t′ ≡ q ′ |q ′ . (9. ≡ 0.3) (9. |ψS (t) = U (t.1 Time evolution as path integral U (t. t0 ). t ′ QH (t′ )|q ′ . in which the states are time-independent.7) 73 . [AH .5) (9.

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|. qj )]) . .12) where the transformation between coordinate and momentum space involves q|p = ei p.16) this.q . use the Campbell-Baker-Hausdorﬀ formula.11) (9. (9.15) 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 . t and QH (t) = QS and study the evolution of the system by calculating the quantum mechanical overlap amplitude q ′ . The purpose of this is to calculate the evolution for an inﬁnitesimal time interval. t′ |q. ˙ 2π 1 1 1 [A. t = q ′ |e−i H(t −t) |q . t′ |q. B]] + [[A. Then qj+1 . n |q (9.Path integrals and quantum mechanics 74 and Schr¨dinger states o |q |q ′ QS |q ≡ q|q . 2π (9. which is a hamiltonian function in which the operators are replaced by real-numbered variables. t = = q ′ |e−i n H∆τ |q = q ′ | e−i H∆τ dq1 . tj+1 |qj .qj ) e 2π dpj exp (i ∆τ [pj qj − H(pj . eA eB = eC with C = A+B+ (9. ′ (9.14) . (9. QS |q ′ ≡ q ′ |q ′ . . B] + [A.8) Choose t as the starting point with |q = |q.qj )∆τ e e 2π dpj i pj (qj+1 −qj )−i ∆τ H(pj . B].13) At least for a simple hamiltonian such as consisting of a kinetic energy term and a local potential. 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 ..10) dqn−1 q ′ |e−i H∆τ |qn−1 qn−1 | . tj+1 |qj . H(P. . qj ). Combining the two terms gives qj+1 . B] + . 2 12 12 (9. Q) expressed in terms of the operators P and Q. with the correction1 being of order (∆τ )2 . |q1 q1 |e−i H∆τ |q . tj = = = 1 For dpj i pj (qj+1 −qj ) −i H(pj .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 ′ . [A.. 2π (9. The hamiltonian is an operator H = H(P. 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 . .

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

23) Having deﬁned the Gaussian integral.21) again a multidimensional integral.25) (see Exercises). y) = ∗ m. (9. .n dx fm (x)fn (x) = n = n dαn 1 exp − α2 N 2 n √ 2π ≡ 1. 2 (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.28) Dα Dα∗ exp (−α∗ α − α∗ ω − ω ∗ α) = exp (ω ∗ ω) . with Dα F [α] = n dαn N F (αn ) (9.27) (9. Using method 1 one writes 1 Dα exp − α2 2 = x dαx N exp − 1 2 ∆x α2 x x = x dαx 1 exp − ∆x α2 x N 2 1 N 2π ∆x ≡ 1. Consider the gaussian functional as an example.24) (9.26) As an important application of translation invariance we mention the identities 1 Dα exp − α2 − ωα 2 = exp 1 2 ω . α(x) = n αn fn and K(x. A useful property of functional integration is the translation invariance. (9. N (9. Note that procedure (i) is an example of the more general procedure under (ii). 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. the following integrals can be derived for a real symmetric or complex hermitean kernel K.n fm (x) Kmn fn (y) and consider F [α] = F (αn ) as a multivariable function.22) = x The last equality is obtained by deﬁning the right measure (normalization N ) in the integration. Physical answers will usually come out as the ratio of two functional integrals and are thus independent of the chosen measure. In both cases N is an appropriately choosen normalization constant in such a way that the integral is ﬁnite.Path integrals and quantum mechanics 76 (ii) Write α in terms of a sum of orthonormal basis functions. Dα F [α] = Dα F [α + β].

35) For applications to fermion ﬁelds. F [θ.e. δα(x) δα(x) δ αKβ = dy K(x.30) (9. i. δα(x) or in discretized form δ 1 ∂ ∂ with . It is now easy to check that Dθ∗ Dθ exp (−θ∗ K θ) = det K. αy = δxy . we need to consider Grassmann-valued functions involving anticommuting Grassmann variables.32) (9. n θn fn (x) dx θ∗ (x) θ(x) . Examples of functional diﬀerentiation are (9. which vanish. y) θ(y) .38) . the Gaussian-type functionals.31) δ δ α = dy α(y) = dy δ(x − y) = 1. e.Path integrals and quantum mechanics 77 Functional diﬀerentiation is deﬁned as δ . δα(x) 2 2 δ exp (−α β) = −β(x) exp (−α β) .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. θ∗ ] = exp (−θ∗ K θ) ≡ exp − Integration for Grassmann variables is deﬁned as dθ 1 = 0 and dθ θ = 1. δα(x) (9.33) (9. products of the same pair appear. (9.29) (9. α(y) = δ(x − y). αn = δmn . however. y) β(y) δα(x) δ 1 1 exp − α2 = −α(x) exp − α2 .g. θη = −ηθ.34) (9. −→ fn (x) δα(x) ∂αn ∂αm n Note that δ(x − y) = n ∗ fn (x) fn (y). dx dy θ∗ (x) K(x. (9. In the expansion of the exponential (from second to third line). θ∗ ] = exp (−θ∗ θ) ≡ exp − F [θ. θ2 = 0. −→ δα(x) ∆x ∂αx ∂αx δ ∂ ∂ ∗ with . (9.37) We note that from the ﬁrst to the second line one needs to realize that a pair of diﬀerent Grassmann variables behaves as ordinary complex variables. In principle the deﬁnitions of functionals is the same.

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

t is related to the Schr¨dinger state |q by |q. q) → L(q. t = ei Ht |q . say. We can again rewrite the time-ordered products as functional integrals as derived in the previous section.65) J dq ′ dq Q′ . ˙ The combination of exponentials in the last step is allowed because we are simply dealing with classical quantities (not operators!). Consider. T ′ |Q. t|Q. t′ |q. T = n J t′ = = Dq exp i t dτ [L(q. L(q.64) (9. t |T exp −i p exp −i 2π t′ ′ ′ t′ t dτ HI (Q(τ )) |q. making the electron) and the absorption of an electron (think of a detector). t′ q ′ .66) .e. T ′ |q ′ . n (9. T < t < t′ < T ′ .1: Physical picture of vacuum to vacuum amplitude in presence of a source term This is not a surprising result. for example x|n x|n = φn (x) = ei kn x = e −ω 2 x2 /2 (9. t t′ V =0 Dq D dτ HI (q) exp i t t′ t dτ [pq − H0 (p. ˙ J (9. q . t |q. q)] ˙ (9. which in turn are embedded between an early time T and a future time T ′ . t|n = q|n e−iEn t with q|n the time-independent wave function.62) p Dq D exp i 2π t dτ [pq − H(p. for g. one has q ′ . Considering the source to be present between times t and t′ . t Q′ . the creation of an electron (think of a radio-tube. φn (q.s. it is possible to switch on an interaction. q) + J(τ )q] . T = n q. furthermore. for the physical states o q. i. T with q. Before and after these processes there is only the vacuum or ground-state |0 . q)] . i. t = = ′ ′ V = q . T . 9. t q. harmonic oscillator.e. The Heisenberg state |q. t′ |q. t) φ∗ (Q) e+i En T . ˙ ˙ physically pictured as. q) + J(t) · q. a set |n of physical eigenstates. t|Q.Path integrals and quantum mechanics source experiment detector 81 T L t L+Jq t’ L T’ Figure 9.5 The generating functional for time ordered products By introducing a source-term.63) for plane waves. t|n n|Q.

J (9. t|Q. in which case eiEn T → e−En ·∞ and the term eiE0 T is dominant.70) for the generating functional is in fact ill-deﬁned. Better is the use ˜ of imaginary time t = −it. Since we have (neglecting multiplicative factors). 9. The importance of Z[J] is that the time ordered product of operators can be obtained from it. t′ ) q ′ . t 0 0out |0in J . Q(tn )|0 . T J lim −i E0 (T ′ −T ) φ (Q′ ) φ∗ (Q) 0 0 T ′ → −i∞ e T → i∞ dq ′ dq φ∗ (q ′ .67) = = φ0 (q. t). eiE0 T φ∗ (Q) 0 We deﬁne the generating functional Z[J] as Z[J] = Q′ . T = φ0 (q. such that t′ → −i∞ t → i∞ ↔ ↔ ˜ t → −∞.69) The factor that has been divided out in the ﬁrst line of this equation is a numerical factor depending on the boundaries of the space-time volume (T and T ′ ). J=0 (9. dq dτ + J(τ )q(τ ) . t′ |q. T ′ |Q.70) one immediately ﬁnds δ n Z[J] δJ(t1 ) . The generating functional precisely represents the vacuum to vacuum amplitude from initial (’in’) to ﬁnal(’out’) situation in the presence of a source. t) (9. = (i)n 0|T Q(t1 ) .68) (9. . . δJ(tn ) hence the name generating functional. ˜ t′ → ∞.2: Analytic continuation of boundaries T and T ′ We can project out the groundstate by an analytic continuation in the time.Path integrals and quantum mechanics t 82 T t t’ T’ Figure 9. thus T →i∞ lim q. . T′ Z[J] = lim T ′ → −i∞ T → i∞ Dq exp i T dτ L q. . T → i ∞ and T ′ → −i ∞. (9.71) 9. .6 Euclidean formulation The above expression (Eq.

As discussed in the previous section the interesting quantities are obtained from functional diﬀerentiation with respect to the sources. .73) d˜ LE q. . . . The diﬀerentiations in Z[J] and ZE [J] are related. δ n Z[J] 1 Z[J] δJ(t1 ) . for Grassmann valued functions. In all of these cases the determinant of the (hermitean) operator K is deﬁned as det K ≡ p Kp .76) which is a positive deﬁnite quantity. 1 = √ . Exercises Exercise 9.Path integrals and quantum mechanics 83 In terms of the imaginary time we can write the Euclidean generating functional. ZE [J] = = where LE q.n fm (x) Kmn fn (y). δJ(tn ) = (i)n J=0 1 δ n ZE [J] . ˜ ˜ ZE [J] δJ(t1 ) .g.2 Check the examples of functional derivation for real-. ensuring convergence for the functional integral ZE [J]. δJ(tn ) J = 0 ˜ t = it (9.77) where the expressions have been divided by Z[J] and ZE [J] in order to get rid of dependence on multiplicative factors. i dq . τ dq ˜ dt ≡ −L q. Dα Dα∗ exp (−α∗ K α) = for complex-valued functions and Dθ∗ Dθ exp (−θ∗ K θ) = det K. e. when L q. det K Exercise 9. (9. . dq dt = 1 2 1 2 dq dt 2 Dq exp Dq exp − ∞ d˜ L q. (9. + V (q). where Kp are the eigenvalues of the matrix Kmn in K(x.72) (9. complex. dq ˜ dt = 1 2 dq ˜ dt dq ˜ dt 2 − V (q). y) = ∗ m. det K 1 .74) − V (q) 2 (9. i τ −∞ ∞ −∞ dq d˜ τ + J(˜)q(˜) τ τ dq d˜ τ − J(˜)q(˜) τ τ .and Grassmann-valued functions.1 Convince yourself by choosing an appropriate (orthonormal) set of functions that 1 Dα exp − α K α 2 for real-valued functions.75) = − LE q. ˜ dt (9.

80) Dα∗ Dα exp (−α∗ K α − α∗ ω − ω ∗ α) = 1 exp ω ∗ K −1 ω . 2 (9. y) K −1 (y.3 Prove the following relations 1 Dα exp − α K α − αω 2 1 = √ exp det K 1 ω K −1 ω . .79) (9.Path integrals and quantum mechanics 84 Exercise 9. det K Dθ∗ Dθ exp (−θ∗ K θ − θ∗ η − η ∗ θ) = det K exp η ∗ K −1 η . z) = δ(x − z).78) (9. where dy K(x.

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. ∂µ φ) + Jφ] (10. Π) (10. .4) J=0 . xn ). . φ(x. x′ . (10. ∂µ φ) − Jφ] (10. . x. 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 ) . . δJ(xn ) 85 ≡ G(n) (x1 .2) = ∞ −∞ d4 xE [LE (φ. .e. t = = Dφ DΠ exp i t σ′ ˙ dτ d3 x Π(x)φ(x) − H (φ.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′ |φ.Chapter 10 Feynman diagrams for scattering amplitudes 10. . i.3) The Euclidean formulation is as before.

These Green functions appear in the expansion Z[J] = n in n! d4 x1 .4) one distinguishes the retarded Green’s function ∆R (x) = d4 k e−i k·x . = 0 2 2 2 k2 − M 2 (k ) − E 2 (k ) − k − M (10. (10. . . J(xn ).3) one ﬁnds 1 Z0 [J] = exp − i 2 d4 x d4 y J(x) ∆F (x − y) J(y) ≡ exp (i W0 [J]) .7) where we made the choice Z0 [0] = 1 and ∂µ ∂ µ + M 2 ∆F (x − y) = −δ 4 (x − y). . Furthermore. . xn ) = x2 xn .5) Z0 [J] for the free scalar ﬁeld For the (free) scalar ﬁeld lagrangian L = ← 1 1 ∂µ φ∂ µ φ − M 2 φ2 .Feynman diagrams for scattering amplitudes 86 for which we introduce the picture x1 G(n) (x1 . (10.10) J=0 In order to determine ∆F . . d4 xn G(n) (x1 .6) the generating functional is (using ∂µ = − ∂µ ) given by Z0 [J] = Dφ exp −i where Kxy = (∂µ ∂ µ + M 2 ) δ 4 (x − y).11) Depending on the path choosen in the complex k 0 plane (see exercise 7. . with Z0 [0] = 1. 2 2 → (10. . xn ) J(x1 ) . . As discussed in the previous section (exercise 9. . ∆F (the socalled Feynman propagator) is the Green’s function of the Klein-Gordon equation.13) .8) 1 d4 x φ (∂µ ∂ µ + M 2 ) φ − Jφ 2 K (10. . which can be written as ∆(x) = ˜ with (k 2 − M 2 )∆(k) = 1 or ˜ ∆(k) = 1 1 1 = 0 2 . 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 (10.12) d4 k −i k·x ˜ e ∆(k) (2π)4 (10.9) i. . (10. consider the general solution of ∂µ ∂ µ + M 2 ∆(x) = −δ 4 (x). .e. (2π)4 (k 0 + iǫ)2 − E 2 (10.

the second is to explicitly consider 0|T φ(x1 ) φ(x2 )|0 . φ− (0)] = (2π)3 2E d3 k e−i k·x = [φ− (x). (2π)4 k 2 − M 2 (10. (10.14) They are solutions of the inhomogeneous equation.1 for countours) µ (∂E )2 − M 2 (i ∆F (x − y)) = −δ 4 (x − y) 0 kE =∞ (10. (2π)4 k 2 − M 2 where the latter contour can be deformed to the contour for ∆C . the generating functional can be written 1 µ µ Z0 [J] = Dφ exp d4 xE (10.20) = exp − 2 where or (see ﬁg. φ(0)] = i∆+ (x) + i∆− (x) d3 k e−i k·x = [φ+ (x). . (2π)4 (k 0 − iǫ)2 − E 2 (10. The ﬁrst possibility is to consider the well-deﬁned Euclidean formulation. 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. t) = 0 for t < 0. t) = 0 for t > 0.e. 10.21) ∆F (x) = = −i 0 kE =−∞ d4 kE e−i kE ·xE 2 (2π)4 kE + M 2 (10. in the (well-deﬁned) Euclidean formulation starting with LE = 2 (∂E φ)2 + 2 M 2 φ2 µ 2 2 Jφ with the equation of motion (∂E ) − M φ(x) = −J(x). (10.23) k0 =i∞ k0 =−i∞ d4 k e−i k·x . It is straightforward (closing contours in the appropriate half of the complex plane) to prove for instance that i ∆C (x) = θ(x0 ) i ∆+ (x) − θ(−x0 ) i ∆− (x). φ+ (0)] = (2π)3 2E are also shown in exercise (7.19) φ ∂E ∂E − M 2 φ + Jφ 2 1 d4 xE d4 yE J(x) (−i ∆F (x − y)) J(y) . (2π)4 k 2 − M 2 (10. 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 . ∆F = ∆C . The contours for i ∆(x) i ∆+ (x) i ∆− (x) = [φ(x). µ 1 1 • Firstly.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.22) (10.16) C where C is a closed contour in the complex k 0 -plane. ∆(x) = − d4 k e−i k·x .18) Thus we see various solutions. i. the advanced Green’s function ∆A (x) = satisfying ∆A (x.17) d4 k e−i k·x .4). 4 k 2 − M 2 + iǫ (2π) (10.Feynman diagrams for scattering amplitudes 87 satisfying ∆R (x.

φ+ (y)]|0 = θ(x0 − y 0 ) i ∆+ (x − y) − θ(y 0 − x0 ) i ∆− (x − y) (10. x2 . φ− (y)]|0 − θ(y 0 − x0 ) 0|[φ− (x). We ﬁnd (see also section 7. = θ(x0 − y 0 ) 0|[φ+ (x).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).1: Contour in the k 0 plane for the Feynman propagator in Euclidean and Minkowski space • As the second possibility. Neglecting multiplicative factors or equivalently assuming that Z0 [0] = 1. because this is the only quantity entering Z0 [J]. x3 .24) Knowing the explicit form of Z0 [J] it is straightforward to calculate the 4-points Green’s function. . δJ(x4 ) 2 J=0 = − [∆F (x1 − x2 ) ∆F (x3 − x4 ) + ∆F (x1 − x3 ) ∆F (x2 − x4 ) + ∆F (x1 − x4 ) ∆F (x2 − x3 )] . (4) (2) It should be clear that G0 can be expressed in terms of G0 = i ∆F . or diagrammatically (4) (10.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. . x4 ) = 0|T φ(x1 ) φ(x2 ) φ(x3 ) φ(x4 )|0 1 δ4 exp − i J ∆F J = (−i)4 δJ(x1 ) .25) x1 x2 x3 = x4 + + . the same result as above. 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. we have G0 (x1 .

x2 ) = i ∆F (x1 − x2 ) = G(4.27) When interactions are present. x3 .31) G(2. i. (10.32) (10.26) Dφ exp i = exp i d4 z LI 1 δ i δJ(z) d4 x d4 y J(x) ∆F (x − y) J(y) (10. x2 . Consider as an example the interaction LI (φ) = − g 4 φ 4! in the scalar ﬁeld theory discussed sofar. x2 . (2) (0) (10. with propagators (i ∆F ) being connected to vertices (i LI ) according to the above expression for the functional Z[J]. (10. (10.Feynman diagrams for scattering amplitudes 89 10.2 Generating functionals for interacting scalar ﬁelds d4 x [L0 (φ) + LI (φ) + Jφ] 1 δ i δJ(z) d4 x [L0 (φ) + Jφ] (10.e.34) Introducing a vertex point to which four propagators are connected. x4 ) = G0 (x1 .30) and in zeroth order G(0. L (φ) = L0 (φ) + LI (φ). To zeroth order in the coupling one has 0 1 Z (g ) [J] = Z0 [J] = exp − i 2 J ∆F J .g ) (x1 . x4 ) = To ﬁrst order in g one has Z (g ) [J] = = 1 0 0 x1 + + x2 (10. ≡ −i g d4 z (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 .g 0 ) = G0 = 1. Z[J] = = Dφ exp i exp i d4 z LI (10. x2 ) = G0 (x1 .35) z . the generating functional can be written.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 .g ) (x1 . x3 .29) This expression will be the one from which Feynman rules will be derived.

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

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

in| (choose ϕ0 = 0). (2) (10. the vacuum expectation value of the ﬁeld φ(x) is zero. (10. in = |p1 . in|p′ . out|α. in|S|p′ . . 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 . pn .e. out . p′ . Proof: 0. we have W0 [J] = − 1 2 d4 x d4 y J(x) ∆F (x − y) J(y). in = S|α. x4 10. . out| and S † |α. in|S = 0. (10. then (n) (n) Gsc = Gc for n ≤ 3. x1 G(4. For the free scalar theory. considered explicitly. (3) S is unitary (it conserves the scalar product from initial to ﬁnal state). in|S = β. out|p′ . (10. e. . out = δαβ ↔ SS † = 1.Feynman diagrams for scattering amplitudes 92 i. out| = eiϕ0 0. out . in = p. x2 .g. out = p|p′ . . in into ﬁnal state particle states (out-states) |β. in = |α. out (suppressing except momenta all other 1 m quantum numbers).44) J=0 For the interacting theory. this will be translated to the action on ﬁelds.45) x2 as the only surviving diagram (see exercises). φout and the interpolating ﬁeld φ(x). Also for free ﬁeld (Heisenberg) operators a distinction is made between φin and φout . Sβα ≡ β.43) implying as expected as the only nonzero connected Green’s function G(2) (x1 − x2 ) = (−i) c δ 2 W [J] δJ(x1 ) δJ(x2 ) = G0 (x1 − x2 ) = i ∆F (x1 − x2 ). implying the absence of tadpoles. thus β. in = p. out|α. out = |p′ . in = p. In line with the consideration of the generating functional representing the vacuum to vacuum amplitude we consider ﬁelds φin . Next. translation invariance). x4 ) = c 1 x3 z (10. (2) The one-particle state is invariant (conservation of energy and momentum.g ) (x1 .47) . in = β. in|SS † |α. one ﬁnds for the connected 4-point Green’s function at order g. out| ⇔ |α. out|p′ . . . x3 . Proof: α. in|S = α. in ⇔ β. ∂µ ∂ µ + M 2 φin (x) = 0.3 Interactions and the S-matrix The S-matrix The S-matrix transforms initial state free particle states (in-states) |α. p.46) The properties of the S-matrix are (1) The vacuum is invariant or |S00 | = 1.

The relation between S and Z[J] To establish this relation. φin (x) = d3 k ∗ a(k) fk (x) + a† (k) fk (x) . 2 ∂µ ∂ µ + M0 φ(x) = J(x). (10. S is invariant under e Poincar´ transformations: U (Λ. β. φ(y)] = Z [φin (x).52) where Z is a constant. (10. 2 ˜ ∂µ ∂ µ + M 2 φ(x) = J(x) + (M 2 − M0 )φ(x) = J(x) √ ˜ φ(x) = Zφin (x) − d4 y ∆R (x − y) J(y) (10.49) The ﬁeld φ can be expressed in φin/out using retarded or advanced Green’s functions. which in a later stage (renormalization) will become more important. out|φout = β.54) .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. 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.g. 9. i. z) φ(z) d4 y d4 z ∆A (x − y) K(y. in|Sφout Finally we check that as expected S does not spoil Poincar´ invariance.e. 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.50) = φ(x) = √ Zφin (x) − √ Zφout (x) − d4 y d4 z ∆R (x − y) K(y. e.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 ). implies the strong (operator) convergence φ(x) → Zφin (x). a) = S. (10. The convergence therefore must be weakened to t→−∞ √ (10. this can actually not be used as it would imply [φ(x). a)SU −1 (Λ. a causality condition that can be proven to imply the absence of interactions. √ Although the above. The time evolution operator (see Eq. (2π)3 2E (10.and out-ﬁelds is: φin (x) = S φout (x)S −1 . Proof: β. the source term Jφ in the lagrangian density is treated in the interaction picture. z) φ(z).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).51) (10. φin (y)] = iZ ∆(x − y).62) is then T′ UT [J] T′ = T exp i T d4 x J(x)φ(x) . as it stands. The relation between S-matrix and in. in|φin S → φin S = S φout (x).

with the use of the Baker-Campbell-Hausdorﬀ formula e−B A e+B = A + [A. φ− f (10.61) e φ+ f where the normal ordered expression : e φ f : is used. φ(y)] f (y) e e .63) . one has [A. t i Ut∞ [J] φ(x)U−∞ [J] (10. y) δ δJ(y) : F [J]. Thus applied to the ﬁeld φ(x) = φ+ (x) + φ− (x). δJ(z) (10.55) 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).Feynman diagrams for scattering amplitudes 94 and satisﬁes the property T ′ δUT [J] t = i UtT [J] φ(x)UT [J]. S U [J]] = √ Z δ S U [J].60) e d4 y φ+ (y) f (y) d4 z φ− (z) f (z) d4 z φ+ (z) f (z) d4 y [φ(x). (10. B]eB . we can. express it in terms of advanced and retarded Green’s functions. z) . δJ(z) (10. just as we did for φ(x). eB ] = [A.59) (10. B] and [A. z) δU [J] δJ(z) δU [J] d4 y d4 z ∆A (x − y) K(y. B] is a c-number). : =i d4 y ∆(x − y) f (y) : e d4 z φ(z) f (z) :. 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. z) d4 y d4 z ∆(x − y)K(y. (10. z) δU [J] .57) i [φin (x).62) where F [J] is some (arbitrary) functional. δJ(x) ∞ or for U [J] ≡ U−∞ [J] ′ (10. [A.e.58) In order to ﬁnd a solution to this equation note that. φ(x). 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. C] are c-numbers. B + C]e e . which is equal to the expression e in which the creation operators are placed left of annihilation operators. : e d4 z φ(z) f (z) d4 x φ− (x) f (x) B C B C (10.56) Since δU [J]/δJ(x) satisﬁes the same equations as φ(x) and we know the limits. e e ] = [A. B] (for the case that [A. 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]. e = i. Thus S U [J] =: exp 1 √ Z d4 x d4 y φin (x) K(x. φ(x).

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

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. Note that in calculating amputated Green’s functions external lines are neglected. which cancels the factor 1/2 in the quadratic piece. (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. If problems arise go back to the deﬁning expression for the generating function in 10. = −i g . to be precise iLI vertices in momentum space are introduced. or calculating full Green’s functions external lines are treated as propagators.66 and 10.27. For the symmetry factor consider the examples (given in G. but overall momentum conservation at a vertex is understood. For the interaction terms in the lagrangian. = −i g (multiplied with 4! corresponding to the allowed number of permutations of identical particles). ’t Hooft and M. the factor 2 corresponds to the two-points Green’s function having two identical ends). 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. 3! 4! The vertices are: = −i f Consider ﬁrst the lowest order self-energy diagram.69) LI (φ) = − φ3 − φ4 .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. 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 10.27. At these vertices each line can be assigned a momentum. Diagrammar) in the case of the interaction terms g f (10. vertices and external particle wave functions in that diagram.

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

in e− e− → e− e− scattering (Møller scattering) the lowest order contribution is (see next section for e-e-γ vertex). .. (10. (Rule 6) Each closed fermion loop gets a sign −1.72) (10.e. e.g.. η).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. − 1 2 dx dy dx′ dy ′ η(x)S(x − y)η(y) η(x′ )S(x′ − y ′ )η(y ′ ). η] = Dψ Dψ exp i d4 z LI exp − d4 x ψ (i/ − M ) ψ + LI (ψ) + ηψ + ψη ∂ 1 δ 1 δ .. The two-points Green’s function 0|T φ(x) φ(y)|0 contains a fermionic loop contribution. 4 2 − M 2 + iǫ (2π) p and the (oriented) propagator in Feynman diagrams involving fermions is i p j (10. namely (Rule 5) Feynman diagrams which only diﬀer by exchanging identical fermions in initial or ﬁnal state have a relative minus sign. 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.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 arises from the quadratic term in exp i d4 z LI . η]. 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. 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 . = + . but the anticommutating properties of Grassmann variables imply some additional minus sign in Feynman diagrams.71) = exp i where iSF is the Feynman propagator for fermions.

writing down the propagators till the beginning of the ¯ line (incoming fermion u or outgoing antifermion v) has been reached. 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 . 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. 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 .Feynman diagrams for scattering amplitudes 99 The result is which contains an extra minus sign as compared to a bosonic loop. + 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. λ 1 ∂ (10. 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. The wave functions for fermions are given by ∝ g 2 S(z − z ′ ) S(z ′ − z) = −g 2 iS(z − z ′ ) iS(z ′ − z). 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 . 1 k 2 − M 2 + iǫ λ (k 2 − M 2 + iǫ)(k 2 − λ M 2 + iǫ) leads to . 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 . Particular choices of λ are λ = 1 (Feynman propagator or Feynman gauge) and λ = ∞ (Landau gauge). 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.

s3 )γ µ u(k. s)γν u(k ′ . s2 ).s′ ′ ′ = 2 kµ kν + 2 kν kµ + q 2 gµν 1 Tr [(/ ′ + m)γµ (/ + m)γν ] k k 2 ′ ′ = 2 kµ kν + kν kµ − gµν (k · k ′ − m2 ) (10. 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 . s1 ) u e2 u(p′ .78) = 2 α2 2 s + u2 + 4t(M 2 + m2 ) − 2(M 2 + m2 )2 . s′ ) ¯ u s. s1 ) ¯ −igµν u(p′ . where u ¯ L(m) µν = = 1 2 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. q 4 µν [using that (¯(k ′ )γµ u(k))∗ = u(k)γµ u(k ′ )]. 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 ′ .Feynman diagrams for scattering amplitudes 100 10.76) 1 . s4 )(−ie)γ ν u(p. t2 (10. s4 )γµ u(p. s2 )|2 ¯ q2 (10. s)¯(k.s2 ) which can be written as (see also chapter 4) |M |2 = = 1 4s |¯(k ′ . s3 )(−ie)γ µ u(k. s′ )γµ u(k.75) Note that the q µ q ν term in the photon propagator are irrelevant because the photon couples to a conserved current. To lowest order (∝ α = e2 /4π) only one diagram contributes.s2 .79) .s3 .5 Some examples eµ scattering The ﬁrst example is the electromagnetic scattering of an electron and a muon.s4 e4 (m) µν (M) L L . ¯ q2 (10.s4 ) and averaged over spins in the initial state (1/2 × 1/2 × s1 .

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

+ pn ). . 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). t and u. . (b) Show (by using x1 as shift-varible) that this implies that n i=1 d4 xi ei pi ·xi G(n) (x1 . . . xn ). . Show that i the amplitude is symmetric under the interchange of t ↔ u. . . Pictorially this implies.2 (a) Show that the translation properties of the ﬁelds and the vacuum imply G(n) (x1 + a. hence we can write n i=1 d4 xi ei pi ·xi G(n) (x1 . . . . . or if we in general would consider all momenta as incoming. xn ) ≡ (2π)4 δ 4 (p1 + . . but the incoming and outgoing momenta are identical. . . for instance. .38.3 Show that the combinatorial factors found using the rules given in section 10. It is of the form −i M = A1 − A2 . . . .3 reproduce for the diagrams precisely the combinatorial factors that appear in Eqs 10. . pn ). Express the contributions in invariants s. which means overall momentum conservation in Green functions in momentum space. e(p1 ) + e(p2 ) → e(p′ ) + e(p′ ) in lowest order in α. . . their sum is zero. |M |2 = T11 + T22 − T12 − T21 . Exercise 10. .37 and 10. . Exercise 10. in the amplitude (Tij = A∗ Aj ). . . xn ) ∝ (2π)4 δ 4 (p1 + . pn ) only has n − 1 independent variables in it.Feynman diagrams for scattering amplitudes 102 Exercise 10. + pn ) G(n) (p1 . . 1 2 (b) Calculate the quadratic pieces and interference terms. Note that the function G(n) (p1 . . xn + a) = G(n) (x1 . .4 (a) Write down the Feynman diagrams contributing to electron-electron scattering. .

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

(11.10) The function λ(s. m2 . m2 . p1 . 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. trf Ea (threshold) = 1 2 mb ( i mi )2 − m2 − m2 a b . 4π s 4π 8π s 4π (11. 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.11) (11. m2 ) a b . m2 . Prove that |q| = cm Ea cm Eb (s − m2 − m2 )2 − 4 m2 m2 a a b b = 4s s + m2 − m2 a b √ = . m2 ) a b . Also in this case one can express the energy and momentum in the invariants. for instance.14) One can. p1 .9) (11.12) pb = (mb . m2 ) is a function symmetric in its three arguments. 4s (11. ptrf ). m2 ) = 4(pa · pb )2 − 4p2 p2 . 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 √ = . In that case trf pa = (Ea . which in the speciﬁc case a b also can be expressed as λ(s. m2 . 0). 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. 2 s s − m2 + m2 a b √ .15) Explicit calculation of the reduced two-body phase space element gives dR(s. m2 .13) (11. mb and s). m2 ). p2 ) = = where λ12 denotes λ(s. 2mb (11. a (11. = 2 s λ(s.16) .8) (11. 2mb λ(s. Prove that trf Ea = |ptrf | = a s − m2 − m2 a b .Scattering theory 104 It is straightforward to prove that the unknowns in the particular system can be expressed in the invariants (ma .

pc .Scattering theory 105 Kinematics of 2 → 2 scattering processes π− + p The simplest scattering process is 2 particles in and 2 particles out. E .24) Using the relation between t and cos θcm it is straightforward to express dΩcm in dt. 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 . in which one particle is detected. usually many particles are produced.23) which are not independent as s + t + u = m2 + m2 + m2 + m2 . dt = 2 qq ′ d cos θcm and obtain for the two-body phase space element √ q ′ dΩcm λcd dΩcm √ = (11. py ). At high energies. usually is a 2-particle state. Writing E = mT cosh y. Consider the 1-particle inclusive case.26) = 16π q s 8π λab Kinematics of inclusive hard scattering processes In high energy (hard) scattering processes. tmax min λcd /4s.. The variable s is always larger than a c b d the minimal value (ma + mb )2 . pT = (px . where the ﬁrst is referred to as elastic channel and the set of all other channels as the inelastic channels. An often used set of invariants are the Mandelstam variables. in exclusive measurements all particles are detected.18) (11. there is usually a preferred direction.25) dR(s. however.. = (11.27) with m2 = m2 − p2 = m2 + p2 + p2 .20) → π0 + n → π+ + π− + n → . (11. In inclusive measurements no particles are detected in the ﬁnal state.21) (11.4). Examples appear in → π− + p (11. s = (pa + pb )2 = (pc + pd )2 t = (pa − pc ) = (pb − pd ) u = (pa − pd )2 = (pb − pc )2 2 2 (11. a c with q = λab /4s and q ′ = being 0 or 180 degrees. Instead of the scattering angle.17) (11. Of course there are not only 2-particle channels. The various possibilities are referred to as diﬀerent reaction channels. 2s (11. . pz = mT sinh y.22) (11. The initial state. for instance the momenta of incoming (colliding) hadrons and it is useful to use for the produced particle rapidity as a variable.19) (11. A speciﬁc reaction channel starts contributing at the threshold value (Eq. 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 . pd ) = 4π s 4π 8π s 4π dt dt √ √ . H1 + H2 → h + X. while the ﬁnal state often arises from a series of 2-particle processes combined with the decay of an intermediate particle (resonance). Consider the process a + b → c + d. the rapidity y is deﬁned T T x y y= 1 ln 2 E + pz E − pz = ln E + pz mT = tanh−1 pz . 11.

For the t-channel process (a¯ → ¯ one has c bd) st = (pa + pc )2 = t. For large energies and not too small scattering angles (θ ≫ 1/γ) y is approximately equal to the pseudo-rapidity η.2 Crossing symmetry In the previous chapter. u) with s = (p1 + p2 )2 = (p3 + p4 )2 .Scattering theory 106 It is convenient because under a boost (along z) with velocity β it changes as y → y − tanh−1 β. uu = s) = Mab→cd (s. t-channel and u-channel processes respectively.30) (11. c bd Mad→c¯(su = u. ¯ tt = (pa − p¯)2 = s. u = (p1 − p4 )2 = (p2 − p3 )2 . u). b ut = (pa − pd )2 = u. The one-particle phase space in terms of these variables becomes d3 p dy d|pT |2 dφ dy d2 pT |p| dη d|pT |2 dφ = = = . (11. t. ¯ while for the u-channel process (ad → c¯ one has b) su = (pa + pd )2 = u. This is known as crossing symmetry. u). ut = u) = Mab→cd (s. which means that rapidity distributions dN/dy maintain their shape.31) . t. t = (p1 − p3 )2 = (p2 − p4 )2 . 11. (11.29) 3 2E 3 2 (2π) 2 (2π) 16π 2π E 16π 2 2π with at high energies the factor |p|/E ≃ 1. ¯ tu = (pa − pc )2 = t. 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. These variables are precisely the Mandelstam variables for the s-channel process (ab → cd). 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. b Analiticity of the ﬁeld theoretically calculated result implies Ma¯→¯ (st = t. ¯ b (11. η = − ln(tan(θ/2)) = tanh−1 (cos θ). tt = s. uu = (pa − p¯)2 = s. t.28) which (only involving angles) is easier to determine. With the momenta deﬁned as in the ﬁgures above one has for all these processes the same amplitude M (s. tu = s.

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

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. T · V ρa (11.Scattering theory 108 Although this at ﬁrst sight does not look covariant. as expected.T d4 x = V · T (2π)4 δ 4 (Pi − Pf ). For the (proper) decay width one thus has 1 N Γ0 = . pn ). . (11. the amount of decaying particles is proportional to the number of particles with proportionality factor the em decay width Γ. Using ρtrf = a (0) (0) (0) lab lab lab ρa · γa = ρa · Ea /ma (where ρa is the rest frame density) and va = plab /Ea we have a trf ρa ρb va = (0) ρa ρb 2 λab 4 ma mb (0) (0) or with ρa = 2 ma . Microscopically one has Ndecay = Na · Γ T or Γ= N 1 . .e.T V. it is.35) This quantity is not covariant. (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.36) (11. N and T · V are covariant.38) (in which we can calculate −iMf i using Feynman diagrams). . 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. . N 1 .39) One of the δ functions can be rewritten as T · V (remember the normalization of plane waves). σ= √ 2 λab T · V (11. From the solution N (t) = N (0) e−Γ t one knows that the decay time τ = 1/Γ. (11.33) Decay of particles For the decay of particle a one has macroscopically dN = −Γ N. the number of scattered or decayed particles is given by 2 N = (2π)4 δ 4 (Pi − Pf ) iMf i dR(p1 . 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 .34) i.

θ) = −8π s ℓ (in analogy with the expansion for f (E. 8π s 2π 8π s .43) dΩcm = π M (s. q n ). θ)|2 . 16π 2 s (11.46) Partial wave expansion Often it is useful to make a partial wave expansion for the amplitude M (s. . t) λab 4π dt.45) Since the amplitudes also depend on all momenta the full result for two-particle intermediate states is (in CM. 2 λab and for instance for two particles dσ = q ′ M (s. θcm ) √ q 8π s 2 2 (11.42) 2−body decay The diﬀerential cross section (ﬁnal state not integrated over) is given by 1 dσ = √ |Mf i |2 dR(s. for which the √ amplitudes squared have been calculated in the previous chapter. ˆ ˆ Inserted in the unitarity condition for M . . (M † )f n (2π)4 δ 4 (Pi − Pn ) Mni . (11. 11. see 11.41) (11. p1 . pn ) |M |2 q 32π 2 m2 dΩ |M |2 . (11. (11. .4 Unitarity condition (S † )f n Sni = δf i The unitarity of the S-matrix.Scattering theory 109 (Using normalized wave packets these somewhat ill-deﬁned manipulations can be made more rigorous). . . . respectively. p1 . The result is N = |Mf i |2 dR(s. .e. . . pn ). i. q n ) q n qn √ Mni (q i . (11. The amplitude −M /8π s is the one to be compared with the quantum mechanical scattering amplitude f (E. δf n + i(2π)4 δ 4 (Pf − Pn ) (M † )f n or −i Mf i − (M † )f i = − n δni − i(2π)4 δ 4 (Pi − Pn ) Mni = δf i . The sign diﬀerence comes from the (conventional) sign in relation between S and quantummechanical and relativistic scattering amplitude. θ). for which one has dσ/dΩ = |f (E. √ (2ℓ + 1)Mℓ (s) Pℓ (cos θ). θ) in quantum mechanics. . . i M† M √ − √ 8π s 8π s = fi n dΩn ∗ Mnf qn Mni √ √ .25) −i Mf i − (M † )f i = − ∗ dΩ(ˆn )Mnf (q f .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 .47) M (s.44) This can be used to get the full expression for dσ/dt for eµ and e+ e− scattering. note the sign and cos θ = qi · qf ). θ) or M (q i . implies for the scattering matrix M . . (11. p1 . q f ). pn ).

Using σ= dΩ q ′ M (s.48) If only one channel is present this simpliﬁes to ∗ ∗ −i (Mℓ − Mℓ ) = 2 qMℓ Mℓ . ℓ (11.49) or Im Mℓ = q |Mℓ |2 .54) ℓ .52) From the imaginary part of M (s. 2i q 2i q (11. (11.53) The diﬀerence is the inelastic cross section. parametrized as Sℓ (s) = ηℓ (s) exp(2i δℓ (s)). n (11. (11. q ˆ i. σel = 4π ℓ (2ℓ + 1)|Mℓ (s)|2 (2ℓ + 1) ℓ = 4π q2 ηℓ e 2i δℓ − 1 2i 2 . † −i (Mℓ )f i − (Mℓ )f i = 2 † (Mℓ )f n qn (Mℓ )ni . σinel = π q2 2 (2ℓ + 1) (1 − ηℓ ).50) where Sℓ (s) satisﬁes |Sℓ (s)| = 1 and δℓ (s) is the phase shift.51) and for the case that this is the only channel (purely elastic scattering. 0). θcm ) √ q 8π s 2 .e.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 . q ˆ Pℓ (ˆ · q ′ ) = q ˆ (ℓ) m 4π Y (ℓ) (ˆ) Ym (ˆ′ ) q (ℓ)∗ q 2ℓ + 1 m and the orthogonality of the Ym functions to prove that RHS = 2 n ℓ † (2ℓ + 1)Pℓ (ˆi · qf ) (Mℓ )f n qn (Mℓ )ni . (11. Show that σT = = 4π q 2π q2 (2ℓ + 1) Im Mℓ (s) ℓ ℓ (2ℓ + 1) (1 − ηℓ cos 2δℓ ). which allows writing Mℓ (s) = Sℓ (s) − 1 e2i δℓ (s) − 1 = . in combination with the partial wave expansion for the amplitudes M and the orthogonality of the Legendre polynomials immediately gives for the elastic channel. the total cross section can be determined. η = 1) the result σel = 4π q2 (2ℓ + 1) sin2 δℓ . In general a given channel has |Sℓ (s)| ≤ 1. (11.

We note that unitarity is generally broken in a ﬁnite order calculation. . 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.Scattering theory 111 Note that the total cross section is maximal in the case of full absorption.5 Unstable particles i k 2 − M 2 + iǫ For a stable particle the propagator is i∆(k) = (11. This is what we will do to discuss decay widths. η = 0. ∝ U (t. in which case.55) (note that we have disregarded spin).57) This shows that Γ is the width of the resonance. . 0) ∝ e−i(E−iΓ/2)t . The prescription for the pole structure. For an unstable particle one expects that U (t. The product of amplitudes is of order e4 . it is straightforward to write down the partial wave amplitude. pn )MRn (2π)4 δ 4 (k − Pn ) MnR = 2M n Γn = 2M Γ. For the amplitude in a scattering process going through a resonance. The quantity Γ is precisely the width for unstable particles. 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. (11. the time-evolution operator.58) . 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 . 11.56) (again disregarding spin). (q Mℓ )ij (s) = −M Γi Γj . such that |U (t.e. but still real. . i. This is achieved with a propagator i∆R (k) = i k2 − M2 + iM Γ (11. For instance the tree-level calculation of Feynman diagrams at order e2 was real. The unitarity condition then states 2 Im ∆−1 (k) = R n † dR(p1 . s − M 2 + iM Γ (11. 0). σel = σinel . i. 0)|2 = e−Γt . one has poles at k 0 = ±(Ek − iǫ) where Ek = + k2 + M 2 guarantees the correct behavior. which is given by a sum of the partial widths into the diﬀerent channels. To check unitarity we need the amplitude at order e4 .e. however. . 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 .

At resonance the cross section σT (π + p) is about 210 mb.2 GeV.88 MeV. dt s(s − 4m2 ) with f (t. one can reconstruct the resonance shape for diﬀerent numbers of neutrino species. Note that because of the presence of a background the phase shift at resonance may actually be shifted. This implies that the resonance has spin J = 3/2 (decaying in a P-wave (ℓ = 1) pion-nucleon state) and isospin I = 3/2 (the latter under the assumption that isospin is conserved for the strong interactions). The phase shift in the resonating channel near the resonance is given by tan δℓ (s) = MΓ . it is furthermore easy to prove that the cross section is given by σel = M 2 Γ2 4π (2ℓ + 1) .Scattering theory 112 (Prove this using the unitarity condition for partial waves). The half-width of the resonance is M Γ. i. 2 q (s − M 2 )2 + M 2 Γ2 (11. This characteristic shape of a resonance is called the Breit-Wigner shape. Its mass is M = 3096. M2 − s (11. • The J/ψ resonance in e+ e− scattering. Three neutrinos explain the resonance shape. From this one sees that a resonance has the same shape in all channels but diﬀerent strength. This is a narrow resonance discovered in 1974. Γ = 2. t)} . Its mass is M = 1232 MeV. u) = 1 1 2 (s + u2 ) + 4m2 (t − m2 ) t2 2 2 1 1 ( s − m2 )( s − 3m2 ) tu 2 2 . u) + g(t. t) + g(u. Knowing that each neutrino contributes about 160 MeV (see next chapter). the full width is Γ = 88 keV. Limiting ourselves to a resonance in one channel.1 Show that the cross section for electron-electron scattering (exercise 10. the partial width into e+ e− is Γee = 5.26 keV.49 GeV. The cross section σT (π − p) also shows a resonance with the same width with a value of about 70 mb. u) = g(t.4) can be written as 4πα2 dσ = {f (t.60) showing that the phase shift at resonance rises through δ = π/2 with a ’velocity’ ∂δ/∂s = 1/M Γ. its width Γ = 120 MeV. Three famous resonances are: • The ∆-resonance seen in pion-nucleon scattering. u) + f (u. which is attributed to neutrinos. Essentially this resonance can decay into quark-antiquark pairs or into pairs of charged leptons.59) reaching the unitarity limit for s = M 2 . All these decays can be seen and leave an ’invisible’ width of 498 MeV. Exercises Exercise 11. • The Z 0 resonance in e+ e− scattering with M = 91.e. where furthermore σinel = 0. The cross section at resonance is about 30 nb. a fast change in the phase shift for a narrow resonance.

a quantity which we will ¯ ¯ encounter in the next chapter (section 12.4 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 .999 877 Γ and Γ(π − → e− νe ) = 0.8. ρcm ρcm a b 1 cm . s − M∆ ∆ ∆ 2 calculate Re N (M∆ ) . we have here already used unitarity to ﬁnd a ﬁnite imaginary part that is proportional to the total decay width. Show that unitarity can be used one step further to ﬁx also the real part of the residue of the pole in the amplitude.Scattering theory 113 Exercise 11. cm |va − vb | in the center of mass system for two colliding particles a and b. Compare the results with the experimental π − lifetime τ = 2. writing N (s) M1 (s) = 2 + iM Γ .000 123 Γ.6 × 10−8 s and the relative decay widths Γ(π − → µ− νµ ) = 0. Compare the result with the factor trf 1/ρa ρb va that we used in the derivation of the basic expression for cross sections.2).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. As discussed for unstable particles. Thus. Exercise 11.2 Calculate the ﬂux factor. 4.

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

(12. (12. should be independent of such transformations. The relation in Eq. (12. However.e. 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.14) (12. (12. requiring that the ’connection’ Aµ (x) is such that Dµ → S(x)D µ S −1 (x).16) which in the presence of the ’connection’ Aµ diﬀers from the total change between x ψ(x) and x+dx ψ(x+ dx). Dν ] ψ(x) 1 − ig dσ µν Gµν ψ(x). 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). ds which is solved by dψ(s) ds = dxµ .19) This gives rise to a (path dependent) phase in each point. if two diﬀerent paths to the same point give diﬀerent phases the eﬀects can be observed. In principle such a phase in a given point is not observable. i. What is this physical eﬀect by which the ’connection’ Aµ can be observed? For this consider the phase around a closed loop. x+dx ψ(x + dx) = 1 + dxµ Dµ x+dx ψ(x).The standard model the basis for x G and x+dx 117 G.e. i. but expresses them with respect to diﬀerent bases.17) The three equations given so far immediately give Dµ = ∂µ − ig Aµ (x). -dx -dy µ µ dyµ µ dx For a vector. x+dx ψ(x) = = 1 + ig dxµ Aa (x)Ta µ 1 + ig dx Aµ (x) µ x x ψ(x) (12. x x+dx ψ(x + dx) = (1 + dxµ ∂µ ) ψ(x).18) or considering a path xµ (s) from a ﬁxed origin (0) to point x. 12. = . which connects two identical vectors. A ’constant’ vector that only rotates because of the arbitrary deﬁnitions of local frames satisﬁes Dµ ψ(x) = 0. We note that local gauge invariance requires that we can modify all local systems with a (local) unitary transformation S(x). 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 .16. ds s dxµ ds′ P exp ig A (s′ ) ψ(0).15) ψ(x). ∂µ − ig Aµ (x) ψ(x) = 0 dxµ ∂µ − ig Aµ (x(s)) ψ(x) = 0.

2 Spontaneous symmetry breaking In this section we consider the situation that the groundstate of a physical system is degenerate. 12.The standard model 118 where Gµν = (i/g)[Dµ . i. which means that by an appropriate transformation S(x) it can be gauged away (see the example of Aharonov-Bohm eﬀect).2: The symmetry-breaking ’potential’ in the lagrangian for the case that m2 < 0.2. If it is zero one has dxµ Aµ (x) = 0. As there can be one and only one groundstate. usually being (slightly) prejudiced by impurities. If the temperature is high enough the spins are oriented randomly and the (macroscopic) ground state is spherically symmetric. external magnetic ﬁelds. 2 2 4 −V (φ) The potential V (φ) is shown in ﬁg. the groundstate itself appears degenerate. Only if this quantity is nonzero a physically observable eﬀect of Aµ exists.20) with a lagrangian density of the form L = 1 1 1 ∂µ φ · ∂ µ φ − m2 φ · φ − λ(φ · φ)2 . This characterizes spontaneous symmetry breaking. Dν ]. we can disregard those ’perturbations’ and look at the ideal situation. φ= 2 φ3 (12. 12. 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). this means that there is more than one possibility for the groundstate.g. In this case there are many possible groundstates (determined by a ﬁxed direction in space). Consider as an example a ferromagnet with an interaction hamiltonian of the form H=− Jij Si · Sj . e. If the temperature is below a certain critical temperature (T < Tc ) the kinetic energy is no longer dominant and the above hamiltonian prefers a lowest energy conﬁguration in which all spins are parallel. i>j which is rotationally invariant. a theory for a scalar degree of freedom (a scalar ﬁeld) having three (real) components. in reality a not perfectly symmetric situation. φ1 φ . and equivalently Aµ can be considered as a pure gauge eﬀect. Classically the (time-independent) ground state is found for V(φ ) F |φ| Figure 12. Nevertheless. Nature will choose one.e. In geometric language the eﬀect on parallel transport of a vector depends on the ’curvature’ Gµν . .

In this situation one speaks of spontaneous symmetry breaking. 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 . say 0 0 . (12.22) 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.e. The η massless particles are called Goldstone bosons. λ such that 0|ϕ1 |0 = 0|ϕ2 |0 = 0|η|0 = 0. They are known as the Weyl mode or the Goldstone mode: Weyl mode. The situation now is the following.21) Stability of the action requires the classical groundstate ϕc to have a well-deﬁned value (which can be nonzero). i. Any constant ﬁeld with ’length’ F is a possible groundstate. In order to see the consequences for the particle spectrum of the theory we construct the lagrangian in terms of the ﬁelds ϕ1 . A quantum ﬁeld theory has only one nondegenerate groundstate |0 . φ = ϕc + φquantum where for the (operator-valued) coeﬃcients in the quantum ﬁeld one wants 0|c† = c|0 = 0. The ﬁeld along the third axis plays a special role because of the choice of the vacuum expectation value. In this mode the lagrangian and the vacuum are both invariant under a set of symmetry transformations generated by Qa . for the vacuum Qa |0 = 0. . ϕc = 0|φ |0 = (12. It is only invariant under rotations around the 3-axis. etc. (12.) terms constitute interaction terms. It is appropriate to redeﬁne the ﬁeld as ϕ1 φ = ϕ2 . has less symmetry. 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. It is suﬃcient to do this to second order in the ﬁelds as the higher (cubic. (12. In the case of degeneracy. . The original lagrangian contained an SO(3) invariance under (length conserving) rotations among the three ﬁelds.24) 1 2 2 4 1 1 1 = (∂µ ϕ1 )2 + (∂µ ϕ2 )2 + (∂µ η)2 + m2 η 2 + . .23) F +η the latter only forming a minimum for m2 < 0. while the lagrangian including the nonzero groundstate expectation value chosen by nature. The classical groundstate appears degenerate. therefore a choice must be made. so one has 0|φquantum |0 = 0 and hence 0|φ|0 = ϕc . The presence of a nonzero value for the classical groundstate value of the ﬁeld will have an eﬀect when the ﬁeld is quantized.The standard model 119 a constant ﬁeld (∇φ = 0) and the condition ∂V ∂φ ϕc =0 −→ ϕc · ϕc = 0 or ϕc · ϕc = − m2 ≡ F 2.25) 2 2 2 Therefore there are 2 massless scalar particles. In this case the spectrum is described . ϕ2 and η. Writing the ﬁeld φ as a sum of a classical and a quantum ﬁeld. while the quadratic terms must correspond with non-negative masses.

Taking the derivative. A bit more formal.30) which in the case that we include only two ﬂavors form an SU (2)V symmetry (isospin symmetry) generated by the Pauli matrices.e. M = . i. In this mode the lagrangian is invariant but Qa |0 = 0 for a number of generators. [Qa .28) where ψ is extended to a vector in ﬂavor space and M is a diagonal matrix in ﬂavor space. T = τ /2.26) for all the states labeled by a corresponding to ’broken’ generators.. a ∂ µ Jµ = 0. (12. a and a′ are degenerate states.27) If the transformations in the lagrangian give rise to a symmetry the Noether currents are conserved. down. there must be a nonzero expectation value 0|Jµ (x)|π (k) . (12. π Example: chiral symmetry An example of spontaneous symmetry breaking is chiral symmetry breaking in QCD. As the a generators for a symmetry are precisely the zero-components of a conserved current Jµ (x) integrated a a over space. denoted |π a (k) . Known examples are rotational symmetry and the fact that the spectrum shows multiplets labeled by angular momentum ℓ (with members labeled by m). irrespective of the fact if they annihilate the vacuum.e. ψu mu md ψ = ψd . Goldstone mode. if the generators Qa generate a symmetry. a 0|∂ µ Jµ (x)|π b (k) = fπ k 2 ei k·x δab = fπ m2 a ei k·x δab . . ψ −→ ei α·T ψ. Including a sum over the diﬀerent ﬂavors (up. T ] ψ. etc. suggesting the stronger relation ∂ µ Jµ (x) = fπ m2 a π a (x). = ψ(i/ − M )ψ. The generators Qa (in that case the rotation operators Lz . π (12. . Neglecting at this point the local color symmetry.) one can write L = f ψ f (i/ − Mf )ψf .31) . strange. it is easy to see that one gets ∂µ V µ = i ψ [M.29) (Note that each of the entries in the vector for ψ in the above equation is still a 4-component Dirac spinor). V µ = ψγ µ T ψ. . 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. called ﬂavors. a massless Goldstone boson for each ’broken’ generator. Using the Dirac equation. This lagrangian density then is invariant under unitary (vector) transformations in the ﬂavor space. This means that they are operators that create states from the vacuum. the lagrangian for the quarks consists of the free Dirac lagrangian for each of the types of quarks. ∂ ∂ (12. and one must have mπa = 0. H] = 0. The conserved currents corresponding to this symmetry transformation are found directly using Noether’s theorem (see chapter 6).32) (12.The standard model 120 by degenerate representations of the symmetry group. 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′ . Note that for the ﬁelds π a (x) one would have a the relation 0|π a (x)|π a (k) = ei k·x .e. Lx and Ly or instead of the latter two L+ and L− ) are used to label the multiplet members or transform them into one another. . i. i. (12.

In the real world.e. i. Therefore realization in the Weyl mode would require that all particles come double with positive and negative parity.33) which in the case of two ﬂavors form SU (2)A transformations generated by the Pauli matrices. This symmetry is realized in the Weyl mode with the spectrum of QCD showing an almost perfect isospin symmetry.35 is nonzero. nature has choosen the path SU (2)L ⊗ SU (2)R =⇒ SU (2)V . (12.27. both are very small. How can we see this. socalled axial vector transformations. This situation.e. From group theory (Schur’s theorem) one knows that the latter can only be true. M = m · 1. Therefore the world of up and down quarks describing pions. including in Dirac space a γ5 matrix. 12.e. This symmetry. We know that mesons and baryons (such as the nucleons) have a well-deﬁned parity that is conserved. but the fact that the symmetry is not perfect and the right hand side of Eq. much smaller than any of the other mesons or baryons. . SU (2)V (isospin) symmetry is good if the up and down quark masses are identical. ψ −→ ei α·T γ5 ψ. That the massless theory has this symmetry can also be seen by writing it down for the socalled lefthanded and righthanded fermions. is what happens in the real world. This combined symmetry is what one calls chiral symmetry. as the quantity ∂µ Aµ (x). T γ5 = τ γ5 /2. From the number of broken generators it is clear that one expects three massless Goldstone bosons. proton and neutron. a doublet (isospin 1/2) of nucleons. but also a triplet (isospin 1) of pions. (12. it is easy to see that one gets ∂µ Aµ = i ψ {M. i. T } γ5 ψ. Note that these transformations also work on the spinor indices. etc. is by nature not realized in the Weyl mode. (leaving out the ﬂavor structure) the same as ψγ5 ψ. The currents corresponding to this symmetry transformation are again found using Noether’s theorem. etc. with almost degenerate masses (Mp = 938. ∂ ∂ (12. 12. 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. in terms of which the Dirac lagrangian 2 density looks like L = i ψL / ψL + i ψR / ψR − ψR M ψL − ψL M ψR . behaves as a pseudoscalar particle (spin zero.e. chiral symmetry is not perfect.3 MeV/c2 and Mn = 939. stated equivalently. if in ﬂavor space M is proportional to the unit matrix. the lagrangian density is invariant under left (L) and right (R) rotations independently. while the groundstate is only invariant under isospin rotations (R = L).e. (12. Still the basic fact that the generators acting on the vacuum give a nonzero result (i. for which the ﬁeld (according to the discussion above) has the same behavior under parity.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. The Goldstone bosons of QCD are the pions for which fπ = 93 MeV and which have a mass of mπ ≈ 138 MeV/c2 . ψR/L = 1 (1 ± γ5 )ψ. parity would not play a role in the world. T ] = 0.34) Using the Dirac equation. Aµ = ψγ µ T γ5 ψ.35) In this case ∂µ Aµ = 0 will be true if the quarks have zero mass. The chiral ﬁelds ψR and ψL are transformed into each other under parity. which is approximately true for the up and down quarks. or. gives also rise to a nonzero mass for the Goldstone bosons according to Eq. e.6 MeV/c2 ). I. fπ = 0 remains. nucleons and atomic nuclei has not only an isospin or vector symmetry SU (2)V but also an axial vector symmetry SU (2)A .The standard model 121 Furthermore ∂µ V µ = 0 ⇐⇒ [M. parity minus).g. There exists another set of symmetry transformations. however. i. which is equivalent 2 to the V-A symmetry. where the quark masses are not completely zero.

Considering the same example with rotational symmetry (SO(3)) as for spontaneous symmetry breaking of a scalar ﬁeld (Higgs ﬁeld) with three components. thus 9 independent degrees of freedom.38) ∂µ Wν − ∂ν Wµ + g Wµ × Wν . Therefore we can always rotate the ﬁeld φ into the 3-direction in order to simplify the calculation. The latter is a spin 0 particle (real scalar ﬁeld) called a Higgs particle. 0 0 0 = . We have the possibility to perform local gauge transformations.40) a Dµ φ = ∂µ φ − ig Wµ La φ. made into a gauge theory. (12. while the number of massive gauge bosons coincides with the number of ’broken’ generators. in this case three-dimensional) representation reads (La )ij = −i ǫaij . is made. Dµ φ = Gµν = ∂µ φ + g Wµ × φ.37) (12. (12. . two massive bosons (Wµ and Wµ with MW = gF ) and a massive scalar particle (η with m2 = η 2 −2 m . 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).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. . having two independent spin components) and three scalar ﬁelds (one degree of freedom each). one sees that the ﬁelds Wµ and Gµν also can be represented as three-component ﬁelds. After symmetry breaking the same number (as expected) comes out. 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 + . like a photon. The symmetry is broken in the same way as before and the same choice for the vacuum.39) (12.41) 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 . 2 massive vector ﬁelds or spin 1 bosons (2 × 3) and one scalar ﬁeld (1). Initially there are 3 massless gauge ﬁelds (each. . i.e. One may wonder about the degrees of freedom.42) 2 from which one reads oﬀ that the particle content of the theory consists of one massless gauge boson 3 1 2 (Wµ ). but one has 1 massless gauge ﬁeld (2). .The standard model 122 12. as in this case there are no massless Goldstone bosons. (12. 0 ϕc = 0|φ |0 = 0 . 0 φ= (12. 1 1 L = − Gµν · Gµν + Dµ φ · Dµ φ − V (φ). again 9 degrees of freedom. The diﬀerence comes when we reparametrize the ﬁeld φ. 4 2 where Since the explicit (adjoint.

U(1)Y U(1)Y (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. R −→ e−i β R.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. i.43) and One introduces a (weak) SU (2)W symmetry under which eR forms a singlet. left. explicitly Dµ L Dµ R i i g Wµ · τ L + g ′ Bµ L. the leptons (elektrons. 2 (12.48) (12. the quarks (up. explicitly L −→ e−i β/2 L. etc. For the neutrino only a lefthanded particle (and righthanded antiparticle) exist. The procedure. under which the lefthanded and righthanded particles with YW (L) = −1 and YW (R) = −2 transform with eiβYW /2 . that gives a good description of the physical world. By introducing gauge ﬁelds and breaking the symmetry a more complex lagrangian is obtained. electromagnetic and weak forces. neutrinos. (12. ∂ ∂ ∂ (12.and righthanded particles.The standard model 123 12.51) . = ∂µ L − (12. muon µ− or tau τ − ). muons. however.e. τ ). explicitly L R SU(2)W SU(2)W −→ ei α·τ /2 L.) and the gauge bosons responsible for the strong.45) (12.50) (12. etc. 1 As they are massless.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.). while the lefthanded particles form a doublet.47) and YW is considered as an operator that generates a U (1)Y symmetry. This is written as 3 Q = TW + YW . R. down. 2 2 = ∂µ R + i g ′ Bµ R. its corresponding neutrino (νe . Thus the lagrangian density is L (f ) ∂ ∂ = i L/ L + i R/ R. implies certain nontrivial relations between masses and mixing angles that can be tested experimentally and sofar are in excellent agreement with experiment. (12. νµ and ντ ) and their antiparticles. ψR/L = 2 (1 ± γ5 )ψ decouple.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. In the standard model one starts with a very simple basic lagrangian for (massless) fermions which exhibits more symmetry than observed in nature.44) which has an SU (2)W symmetry under transformations eiα·TW . The lagrangian for the leptons consists of three families each containing an elementary fermion (electron e− . Thus L (f ) = i eR / eR + i eL / eL + i νeL / νeL + (µ.

Putting this in leads to L (f ) =L (f 1) +L (f 2) .53) φ= √ (θ4 − iθ3 ) φ 2 with TW = 1/2 and YW = 1 is introduced. (12. 2 and 3 may be eliminated (the necessary SU (2)W rotation is precisely e−iθ(x)·τ ). (12. h) (12. (12.54) = (Dµ φ)† (Dµ φ) −m2 φ† φ − λ (φ† φ)2 .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. .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 . . this leads to the lagrangian L (h1) . with the following lagrangian density consisting of a symmetry breaking piece and a coupling to the fermions. L where L (h1) (h) =L (h1) +L (h2) . a complex Higgs ﬁeld + 1 φ √2 (θ2 + iθ1 ) 0 = 1 (12. which assures with the choice of YW for the Higgs ﬁeld that Q generates the remaining U (1) symmetry. leading to the parametrization 1 0 φ(x) = √ (12.The standard model 124 3 3 where Wµ is a triplet of gauge bosons with TW = 1. −V (φ) L and (h2) = −Ge (LφR + Rφ L). .59) . .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. + 8 (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. † i i g Wµ · τ − g ′ Bµ )φ. + 8 g2 v2 1 + − (Wµ )2 + (Wµ )2 (∂µ h)2 + m2 h2 + 2 8 (g 2 + g ′2 ) v 2 (Zµ )2 + . 4 4 In order to break the symmetry to the symmetry of the physical world.58) (12.57) = 1 g2 v2 1 2 (Wµ )2 + (Wµ )2 (∂µ h)2 + m2 h2 + 2 8 v2 2 3 gWµ − g ′ Bµ + . For the classical ﬁeld the choice θ4 = v is made. Using local gauge invariance θi for i = 1. the U (1)Q symmetry (generated by the charge operator). .

g ′ /g = tan θW . 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 . .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 . 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 + . 2 3 g Wµ − g ′ Bµ 3 ≡ cos θW Wµ − sin θW Bµ .62) g 2 + g ′2 3 ≡ sin θW Wµ + cos θW Bµ . 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 . we can read oﬀ e = g sin θW = g ′ cos θW .The standard model 125 where the quadratically appearing gauge ﬁelds that are furthermore eigenstates of the charge operator are ± Wµ = = = Zµ Aµ 1 1 2 √ Wµ ± i Wµ .61) (12. 2 (12.60) (12.66) From the coupling to the photon.69) . 2 + g ′2 g 3 g ′ Wµ + g Bµ (12. 2 Zµ (12. .65) The weak mixing angle is related to the ratio of coupling constants.67) The coupling of electrons or muons to their respective neutrinos. (12.64) (12. 2θ 2θ 4 cos W cos W 0. σ (νµ γ ρ µL ) 2 2 (eL γ νe ) 2 k + MW (12.63) (12. 4 2 MW g2 v2 = . (12.

The standard model 126 the good old four-point interaction introduced by Fermi to explain the weak interactions.40 GeV.73) (12. i. leptons with CV = −1/2 + 2 sin2 θW ≈ −0. 2 The value of α deviates from the known value α ≈ 1/137 because of higher order contributions. g ′ and v determine a number of experimentally measurable quantities. (12. For each neutrino species (with CV = 1/2 and CA = 1/2 one expects Γ(νν) ≈ 155 MeV. 2 2 2 (12.77) (12.g. giving rise to a running coupling constant after renormalization of the ﬁeld theory.05 and CA = −1/2 we calculate Γ(e+ e− ) ≈ 78. −5 (12. muon or tau. g2 MZ 2 2 (CV + CA ). The masses of the fermions and the coupling to the Higgs particle are contained in L (h2) . me /v is extremely small.71) GeV −2 = = = = . Γinvisible = 480 MeV one sees that three families of (light) neutrinos are allowed.72) (12. 91. from the MW mass is about 250 GeV. . as appear for instance in the decay of the neutron or the decay of the Λ (quark content uds). 1.e. the mass of the electron comes from the spontaneous symmetry breaking but is not predicted (it is in the coupling Ge ). 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. The coupling to the Higgs particle is weak as the value for v calculated e.231 2. v (12.70) = 2 = 2 sin2 θ 8 MW 2 v2 8 MW 2 W In this way the parameters g. (12.e.19 GeV.5 MeV (exp. one has the relation g2 e2 1 GF √ = . (12.78) From this coupling it is straightforward to calculate the partial width for Z 0 into a fermion-antifermion pair. one obtains L (h2) Ge v Ge = − √ (eL eR + eR eL ) − √ (eL eR + eR eL ) h 2 2 me = −me ee − eeh.80) First. Comparing this with the total width into (invisible!) channels.74) (12.76) with CV CA 3 = TW − 2 sin2 θW Q. (12. With the choosen vacuum expectation value for the Higgs ﬁeld. Finally we want to say something about the weak properties of the quarks. 80.166 4 × 10 0. such as2 α = e2 /4π GF sin2 θW MW MZ ≈ 1/127.75) 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 . i.79) Γ(Z 0 → f f ) = 48π cos2 θW For the electron. 3 = TW . Γe ≈ Γµ ≈ Γτ ≈ 83 MeV).

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

83) where Vu and Vd are unitary matrices. each of these containing the three families. The imaginary part i η gives rise to CP violation in decays of K and B-mesons (containing s and b quarks. A ≈ 0.5 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. i. † Λu Λ† = Vu G2 Vu . Thus one has † † † =⇒ −DL Vd Md Wd DR − DR Wd Md Vd† DL − UL Vu Mu Wu UR − UR Wu Mu Vu UL (12. ms and mb ). Note that we need the conjugate Higgs ﬁeld to get a U (1)Y singlet in the case of the charge +2/3 quarks. The Higgs ﬁeld is still limited to one complex doublet. d d (12.82 and ρ ≈ 0.The standard model 128 using the socalled Wolfenstein + O(λ4 ) with λ ≈ 0. 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 . (12.84) with Gu and Gd being real and positive and Wu and Wd being diﬀerent unitary matrices.g. for which we need the appropriate weak isospin doublet 0∗ 1 φ c = √ v +h .22 and η ≈ 0.85) √ √ with Mu = Gu v/ 2 (diagonal matrix containing mu .86) the unitary CKM-matrix introduced above in an ad hoc way. The Λu and Λd are complex matrices in the 3×3 family and Λd Λ† = Vd G2 Vd† .q) = −QL φΛd DR − DR Λ† φ† QL − QL φc Λu UR − UR Λ† φc† QL u d (12. the weak mass eigenstates are L (h2. 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. UR = uR cR tR . respectively). the three righthanded quarks with charge +2/3 in UR and the three righthanded quarks with charges −1/3 in DR .e.34. e. u u where we include now the three lefthanded quark doublets in QL .227. . 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 .q) ′ weak † mass mass DL = DL = Vu Vd DL = VCKM DL .82) space. allowing us to write † Λu = Vu Gu Wu † and Λd = Vd Gd Wd . mc and mt ) and Md = Gd v/ 2 (diagonal matrix containing md . φ = −φ− 0 2 For the (squared) complex matrices we can ﬁnd positive eigenvalues.

there is no family mixing for massless neutrinos. The mass ﬁelds ER † mass † mass = We ER .88) √ mass with Me = Ge v/ 2 the diagonal mass matrix with masses me . e N L= L EL (12. The W -current then becomes mass mass E L γ µ NL = E L γ µ NL .ℓ) =⇒ −Me EL Ve We ER − ER We Ve† EL . mµ and mτ .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 .89) with three righthanded neutrinos added to the previous case. and obtain L (h2. The lepton sector (massive Dirac neutrinos) In principle a massive Dirac neutrino could be accounted for by a lagrangian of the type L (h2. † † Again we continue as before now with matrices Λe = Ve Ge We and Λn = Vn Gn Wn .90) mass † mass mass † mass † We note that there are mass ﬁelds ER = We ER . EL = Ve† EL . For the (massless) neutrino ﬁelds we just can redeﬁne ﬁelds into NL = Ve† NL .5×10−3 eV2 . since the weak current is the only place where they show up. In this case. EL = Ve EL . NL = Vn NL and NR = Wn NR µ mass γ µ V † V N mass . i.ℓ) = −LφΛe ER − ER Λ† φ† L. we ﬁnd massless neutrinos. UPMNS a parametrization that in principle also could have been used for quarks. while θ23 then is determined by atmospheric neutrino oscillations requiring |∆m2 | ≈ 2.87) is a weak doublet containing the three families of neutrinos (NL ) and charged leptons (EL ) and ER is † a three-family weak singlet. (12.92) . 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. it is particularly useful because θ12 is essentially determined by solar neutrino oscillations requiring ∆m2 ≈ 8 × 10−5 12 eV2 (convention m2 > m1 ). = − 1/6 2/3 0 1/2 1/2 0 0 1 0 1/3 1/2 − 1/6 (12.ℓ) † † † =⇒ −EL Ve Me We ER − ER We Me Ve† EL − NL Vn Mn Wn NR − NR Wn Mn Vn NL . decoupling from all known interactions. one can write Λe = Ve Ge We and we ﬁnd † L (h2. Working with the mass eigenstates and the weak current becomes EL γ NL = EL e n L weak for the charged leptons we see that the weak eigenstates for the neutrinos are NL = Ve† NL with the relation to the mass eigenstates for the lefthanded neutrinos given by † ′ weak mass mass NL = NL = Ve† Vn NL = UPMNS NL . (12. 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 .e. (12.91) † with UPMNS = Vn Ve known as the Pontecorvo-Maki-Nakagawa-Sakata mixing matrix.The standard model 129 The lepton sector (massless neutrinos) For a lepton sector with a lagrangian density of the form L in which (h2. (12. As before.ℓ) = −LφΛe ER − ER Λ† φ† L − Lφc Λn NR − NR Λ† φc† L e n (12.

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

104) ↔ 1 1 Υi i / Υi − Mi Υi Υi ∂ 4 2 (12. (12. 2 with real masses Mi .2 Derive for the vector and axial vector currents. 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 .1 Consider the case of the Weyl mode for symmetries. 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′ . For the situation ML = 0 and MR ≫ MD (taking MD real) one 2 ﬁnds M1 ≈ MD /MR and M2 ≈ MR .e.105) for i = 1.101) 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 . a and a′ are degenerate states. 2 U (M M † ) U † = M0 . (12. related via c Υ1R = U nL . Exercises Exercise 12. V µ = ψγ µ T ψ and Aµ = ψγ µ γ5 T ψ ∂µ V µ ∂µ A µ = = i ψ [M. . This is a mixing problem with a symmetric (complex) mass matrix leading to two (real) mass eigenstates. Υ2R nR (12. Prove that if the generators Qa generate a symmetry. i.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)) .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 .e. Exercise 12.The standard model 131 taking ML and MR real and non-negative. T ] ψ. i ψ {M. i.103) for each of which one ﬁnds the lagrangians L = and we are left with two decoupled Majorana ﬁelds Υ1L = U ∗ nc . (12. [Qa . Exercise 12. L Υ2L nR Υ1 and Υ2 . which implies U ∗ M † U † = U ∗ M ∗ U † = M0 and a ’normal’ diagonalization of the (hermitean) matrix M 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 } γ5 ψ. H] = 0.

3 TW . (b) Calculate the width to electron-positron pair. Exercise 12. 2 2 The masses are mt ≈ 175 GeV. Note that you can ﬁnd the answer without explicit construction of all interaction terms in the lagrangian. hhh. . Mf M2 and ghf f = . The mass of the Z 0 is MZ = 91 GeV. Neglect the masses of fermions and use the fact that the sum over polarizations is 3 pµ pν . and the width to a pair of neutrino’s. Γ(Z 0 → νe νe ).4 Calculate the lifetime τ = 1/Γ for the top quark (t) assuming that the dominant decay mode is t → b + W +. Γ(Z 0 → e+ e− ). to be precise one has for bosons (b) or fermions (f).) are proportional to the mass squared for bosons or mass for fermions of the particles. 0 ¯ Γ(Z 0 → f f ) = MZ g2 48π cos2 θW f f CV 2 + CA 2 . Exercise 12. at least up to a high (which?) order in λ. ZZh.The standard model 132 with CV CA = = 3 TW − 2Q sin2 θW . mb ≈ 5 GeV and MW ≈ 80 GeV. the weak mixing angle is given by ¯ sin2 θW = 0. In the standard model this coupling is described by the vertex −i g µ √ (γ − γ µ γ5 ) . etc.5 Show that the couplings to the Higgs (for W + W − h.231. e+ e− h. Write down the matrix element squared (averaged over initial spins and summed over ﬁnal spins) for the decay of the Z 0 . ǫ(λ) (p)ǫ(λ)∗ (p) = −gµν + µ ν M2 λ=1 ¯ to calculate the width Γ(Z → f f ). Exercise 12.6 Check that the use of the Wolfenstein parametrization in the CKM matrix indeed gives a unitary matrix. Exercise 12. ghbb = b v v where v is the vacuum expectation value of the Higgs ﬁeld.7 In this exercise two limits are investigated for the two-Majorana case.

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