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

5 Ryder Peskin & Schroeder Srednicki 1 1 1 1 3 2. Diagrammatica.3 6. Quantum Field Theory.2 3. These notes Section 1.3 3. Vol. L.8 3. 1. The quantum theory of ﬁelds.2 3. An introduction to Quantum Field Theory.2 3.2 2. C. 1985.4 3. Addison-Wesly. 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).1 4. The notes contain all essential information. 3.1 6.1.1 2.3 2.1 1.E.3 2.3 References As most directly related books to these notes.5 5. Cambridge University Press.2 2. 2.5 2. 1980.2.4 2. I refer to the book of Srednicki [1] and Ryder [2].V. M.2 36 22 . 4. 1996. 2. Veltman.2 3. Ryder. Quantum Field Theory.2 6. Weinberg.1 3. I: Foundations. Cambridge University Press.3 3. Itzykson and J.H. M. Quantum Field Theory. Schroeder. 3.2 4. S.4 3.4 2. but are rather compact.7 2.3 4. Cambridge University Press. Corresponding chapters in books of Ryder.3 3. 1995.3 2. Peskin and D.3 3. 5.1 2 2 3. 6. McGraw-Hill. Cambridge University Press.3 3.5 4. 2007.5 2. Vol.3 2.1 3.4 3. Zuber.1 5. 3. Other text books of Quantum Field Theory that are useful are given in refs [3-6]. 1995.4 4.2 3.1 2.2 6.3.-B. Cambridge University Press. M.3 2. Peskin & Schroeder and Srednicki.4 6. II: Modern Applications. Srednicki.2 2. 1994.2 1.

7. 4. 6.4 4. 8.5 4.2 9.3 9.4 12.5 7.1 11.3 6.2 7.4 These notes Section 7.1 20.4 10.3.2.8 6.2 4.1 9.6 4.3 11.1 5.4 3.7 5.1 10.5 6 7 6 6.6 3.5 7.3.6 8.1 20.6 3.6 3.5 2.2.8 5 4.5 9.5 9. 2. 6. 9. 6.6 6. 6. 4.3 2.1 12.4 8.2 5.7 4.4.1 20.3 3.5 Ryder 4 4 4 4 4 4 Peskin & Schroeder 2.3 6 8 9 Srednicki 3 3 3 37 5.2 8.3 12.5 11.5.3 7.4.2 11.3 10.6 9.1.4.5 . 2. 6.4 2.2.2 12. 9.3 8. 4.1.4 7.1.5 6.3 7.3. 20.5 4. 4.1 9.1 8.6 10.2 10.1 7.4 9.3 8.4 12.3 20.10 8.2.2 8.

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

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

Introduction

2

and ψ(r) =

d3 p exp (2π )3

i

˜ p · r ψ(p).

(1.7)

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

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

1.2

Units

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

Introduction

3

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

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

2 2

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

c

= 0.389 379 323 (67) GeV2 mbarn

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

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

−27 2

(1.12) (1.13)

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

2

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

Introduction

4

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

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

1.3

**Conventions for vectors and tensors
**

3

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

i=1

xi ei = xi ei , ˆ ˆ

(1.17)

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

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

. in which case zi = ǫijk xj yk . • ds2 < 0 (spacelike intervals). the points lie on the lightcone and they can be connected by a light signal.24) The quantity gµν ≡ nµ · nν is the metric tensor. . δim δjm δkm δin δjn δkn . linear transformations that leave the length of a vector invariant are called (homogeneous) Lorentz transformations. Because of the diﬀerent signs occurring in gµν .25) Within Minkowski space the real. −x). x2 . No diﬀerence is made between upper or lower. Unlike in Euclidean space. The real. x1 . Relations relating tensor expressions... Many other four vectors and tensors transforming like T ′ µ1 . 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 . denoted as V ′µ = Λµν V ν . The lower indices are constructed in the following way. however. The Poincar´ transformations include e e Lorentz transformations and translations. One can distinguish: • ds2 > 0 (timelike intervals). (1. such as eigentime τ for the position vector τ 2 ≡ x2 = t2 − x2 . For four-vectors in Minkowski space we will use the notation with upper indices and write x = (t. one must distinguish tensors with upper or lower indices and one can have mixed tensors. We note that for Euclidean spaces (with a positive deﬁnite metric) vectors and tensors there is only one type of indices. In special relativity we start with a four-dimensional real vector space E(1. In Minkowski space. The length (squared) of a vector is obtained from the scalar ˆ product.. 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. x2 = x · x = xµ xν nµ · nν = xµ xν gµν . When 3-dimensional space is considered as part of Minkowski space. independent of a coordinate system. linear transformations that do not change the length of a four-vector are called the Lorentz transformations.21) (1. The (invariant) lengths often have special names. • ds2 = 0 (lightlike or null intervals). x3 ). One has x2 = xµ xµ = t2 − x2 .Introduction 5 that can be used in the cross product of two vectors z = x × y. (1. Useful relations are ǫijk ǫℓmn ǫijk ǫimn ǫijk ǫijl = = = δiℓ δjℓ δkℓ 2 δkl .23) δjm δkn − δjn δkm . Vectors are denoted x = xµ nµ .26) (1.3).22) (1. 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 . x2 . . These transformations do change the components of a vector. ˆ ˆ (1.2. are . it is convenient to distinguish lower indices from upper indices. xi are the three space components.µn = Λµ1ν1 . So we could have used all upper indices in the above equations. Λµn n T ν1 .1. we will use upper indices for the three-vectors. The invariant distance between two points x and y in Minkowski space is determined from the length dsµ = (x − y)µ . given by g00 = −g11 = −g22 = −g33 = 1 (the ˆ ˆ other components are zero). x3 ) = (t. The transformations that leave invariant the distance ds2 = dt2 − (dx2 + dy 2 + dz 2 ) between two points are called inhomogeneous Lorentz transformations or Poincar´ transformations.νn can be conν structed. the invariant length or distance (squared) is not positive deﬁnite. x) = (x0 . x1 . where the coordinate t = x0 is referred to as the time component. and are given by (x0 .3) with basis nµ (µ = ˆ 0. xµ = gµν xν .

∂3 ) = ∂ ∂ ∂ ∂ . (1. Relate this quantity to the electron Compton wavelength −e via the dimensionless λ ﬁne structure constant α. ∂t ∂x ∂y ∂z ν ∂µ xν = gµ . constructed via the metric tensor itself.∇ . 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. . a∞ = 4πǫ0 2 /me e2 .31) ǫµνρσ ǫµν ′ ′ ρσ ′ = − (1.27) It is easy to convince oneself of the nature of the indices in the above equation.36) Exercises Exercise 1. ∂2 .33) (1.1 (a) In the the Hydrogen atom (quantum system) the scale is set by the Bohr radius.35) ρ ǫµνρσ ǫµν σ σ ǫµνρσ ǫµνρ ′ ′ = −2 g ρρ g σσ − g ρσ g σρ = −6g σσ . = ∂ . p) = (E.29) 2 = ∂ µ ∂µ = ∂t2 The value of the antisymmetric tensor ǫµνρσ is determined in the same way as for ǫijk . (1. (1. = −24. 0123 0123 ′ ′ ǫ µνρσ ǫµνρσ The ﬁrst identity. 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 . deﬁned by ∂2 − ∇2 . not aﬀected by Lorentz transformations. For the contraction of a vector with the antisymmetric tensor one often uses the shorthand notation ǫABCD = ǫµνρσ Aµ Bν Cρ Dσ . . but more important has the same eﬀect on lefthandside and righthandside. gµ = gµρ g ρν and 0 1 2 3 is in essence a ’Kronecker delta’.34) (1. . ∂t (1. Examples are the scalar products above but also relations like pµ = m dxµ /dτ for the momentum four vector. is timelike with invariant length (squared) p2 = p · p = pµ pµ = E 2 − p2 = m2 .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. ′ ′ . for instance. p). Each of these permutations leads to a minus sign. Note that in this equation one has on left. (1. starting from ǫ0123 = 1. ∂1 . The derivative ∂µ is deﬁned ∂µ = ∂/∂xµ and we have a four vector ∂ with components (∂0 .30) (Note that there are diﬀerent conventions around and sometimes the opposite sign is used). explicitly written as (p0 .Introduction 6 called covariant. where m is called the mass of the system. It is an invariant tensor. because one has (1. The length squared of ∂ is the d’Alembertian operator.32) . (1. g0 = g1 = g2 = g3 = 1.28) ν ν Note that gµ with one upper and lower index.

0. Note that ˆ ˆ ˆ ˆ a · nα = aα and derive from that the components nα of the four vector nα . 0) ˆ n2 = (0. n+ . 0. ˆ ˆ ˆ ˆ These are n1 = (0. One does not need to know . This demonstrates how a careful use of units can save a lot of work. e.Introduction 7 (b) Relate the classical radius of the electron (a relativistic concept). g+− and g−+ . a2 ] we can ﬁnd basis vectors n− . by a simple few-line reasoning [For instance: If {µ. (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). (a) Express the scalar product a · b in lightcone coordinates and deduce from this the values of g++ .511 MeV. a+ .2 Prove the identity A × (B × C) = (A · C) B . 1) ˆ . 1. (c) Calculate the Compton wavelength of the electron and the quantities under (a) and (b) using the value of c. 1. n1 . aT ]) are deﬁned through √ a± ≡ (a0 ± a3 )/ 2. ν. a1 . 0. σ} is a permutation of {0. Use that to give the ˆ ˆµ ˆ components of the basis vectors n+ and n− . 0. 1. 1. n2 . How are a+ and a− related to a+ and a− . Also for the coordinates [a− . . me . a3 ) are the expansion coeﬃcients using the basis vectors n0 . . but only appropriate combinations. a+ . 3} the index α can only be equal to one of the indices in ǫµνρσ . (d) Use the value of the gravitational constant GN / c5 = 6. [Note: what is the MKS unit for V/T?] Exercise 1. a+ . a2 . n1 . 0) ˆ n0 = (1.3 Prove the following relation ǫµνρσ g αβ = ǫανρσ g µβ + ǫµαρσ g νβ + ǫµνασ g ρβ + ǫµνρα g σβ . Exercise 1. α and me c2 = 0. c. ]. Compare it with the proton mass and use Eq. 0. a1 . n3 . .71 × 10−39 GeV−2 to construct a mass Mpl (Planck mass).4 Lightcone coordinates for a four vector a (which we will denote with square brackets as [a− . 0.3. Exercise 1. a2 ] or [a− . a1 . (b) The coordinates (a0 . g−− . n2 . re = e2 /4πǫ0 me c2 to the Compton wavelength.(A · B) C using the properties of the tensor ǫijk given in section 1. 2. ǫ0 . ρ.13 to give its actual value in kg. ˆ ˆ . Also construct and calculate the Planck length Lpl . 0) ˆ n3 = (0. which is the Compton wavelength for the Planck mass.

t) = exp(−i k 0 t + i k · r). t) = ∂2 − ∇2 + M 2 φ(r.4) where M is the particle mass. φk (r. written as pµ −→ i∂µ is covariant (the same in every frame of reference).2.7) k 0 = ± k2 + M 2 = ±Ek . (2. p2 = pµ pµ = E 2 − p2 = M 2 . we just want to play a bit with covariant equations and study their behavior under Lorentz transformations.Chapter 2 Relativistic wave equations 2. E = p2 . 2 + M 2 φ(r.3) Equations 2. ∂t 2M (2. Substitution of operators gives the Klein-Gordon (KG) equation for a real or complex function φ. the interpretation of this equation as a single-particle equation in which φ is a complex wave function poses problems because the energy spectrum is not bounded from below and the probability is not positive deﬁnite. (2.2) Acting on the wave function one ﬁnds for a free particle. 8 . p). Although it is straightforward to ﬁnd the solutions of this equation. t) = − ψ(r.1 and 2.6) The energy spectrum is not bounded from below: considering the above stationary plane wave solutions. (2. ∂t p −→ pop = −i∇. 2M (2. Thus a covariant equation can be obtained by starting with the (covariant) equation for the invariant length of the four vector (E. t) = 0. ∂t2 (2. i ∂ ∇2 ψ(r.1 The Klein-Gordon equation In this chapter. namely plane waves characterized by a wave number k. one obtains (2.5) with (k 0 )2 = k2 + M 2 .3 are not covariant. But the replacement 2. The Schr¨dinger equation in quantum mechanics is the operator equation o corresponding to the non-relativistic expression for the energy.1) under the substitution (in coordinate representation) E −→ Eop = i ∂ . t).

∂ρ = −∇ · j. This continuity equation can be written down o covariantly using the components (ρ. rather than as a wave function.8) (2. φ(x) = d3 k (2π)3 2Ek ˜ ˜ e−i k·x φ(Ek . The appropriate current corresponding to the KG equation (see Excercise 2. Then.10) ∂t which follows directly from the Schr¨dinger equation.e. we want the most general solution. (2. there are no longer fundamental objections to mix up space and time. there are solutions with negative energy.2) is j µ = i φ∗ ∂ µ φ or (ρ.12) It is easy to see that this current is conserved if φ (and φ∗ ) satisfy the KG equation. As we will see later both problems are related and have to do with the existence of particles and antiparticles. The integration over k-modes clearly is covariant and restricted to the ‘mass’-shell (as required by Eq. however. And. for which we need the interpretation of φ itself as an operator.11) Therefore. having the KG equation as a space-time symmetric (classical) equation. j) = ↔ i φ∗ ∂0 φ. (2. it is simply a matter of being careful to ﬁnd a consistent (covariant) theory. (2π)3 2Ek (2. −i φ∗ ∇ φ . For this we note that an arbitrary solution for the ﬁeld φ can always be written as a superposition of plane waves. k) ≡ a(k) and φ(−Ek .9) They satisfy the continuity equation. k) + ei k·x φ(−Ek . ↔ ↔ (2. 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.15) . relativistically the density is not a scalar quantity. 2.Relativistic wave equations 9 i.5). j) of the four-current j. The KG equation. This operator has all possible solutions in it multiplied with creation (and annihilation) operators. 2M 2M (2. It is possible to rewrite it as an integration over positive energies only but this gives two terms (use the result of exercise 2.13) ˜ with (in principle complex) coeﬃcients φ(k). This leads to the existence of negative densities. ∂µ j µ = 0.14) ˜ ˜ Introducing φ(Ek . which is what Lorentz transformations do. (2. is a second order equation and φ and ∂φ/∂t can be ﬁxed arbitrarily at a given time. At that point the dependence on position r and time t is just a dependence on numbers/parameters on which the operator depends. −k) .2 Mode expansion of solutions of the KG equation Before quantizing ﬁelds. but rather the zero component of a four vector.3). 2. just as the dependence on time was in ordinary quantum mechanics. −k) ≡ b∗ (k) one has φ(x) = d3 k e−i k·x a(k) + ei k·x b∗ (k) = φ+ (x) + φ− (x). φ(x) = d4 k ˜ 2π δ(k 2 − M 2 ) e−i k·x φ(k) (2π)4 (2.

Performing some Lorentz transformation x → x′ = Λx. −x) ≡ (˜µ ). They are referred to as mode and anti-mode amplitudes (or because of their origin positive and negative energy modes). φP (x) ≡ φ(˜) (P for parity).16) The consequence of this is discussed in Exercise 2. (2. changing the sign of the spatial coordinates.18) (2. The coeﬃcients a(k) and b∗ (k) are the amplitudes of the two independent solutions (two. one thus must have that the function φ → φ′ such that φ′ (x′ ) = φ(x) or φ′ (x) = φ(Λ−1 x).e. i. x Transforming everywhere in the KG equation x → x one obtains ˜ ˜ ˜ x ∂µ ∂ µ + M 2 φ(˜) = 0. This shows how parity transforms k-modes into −k modes. Another symmetry is found by complex conjugating the KG equation. Since a · b = a · ˜ it is easy to see that ˜ b.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. (2.22) showing that with each solution there is a corresponding charge conjugated solution φC (x) = φ∗ (x). 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) . 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) . x ∂µ ∂ µ + M 2 φ(˜) = 0. after restricting the energies to be positive). 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).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) . which implies (xµ ) = (t.23) . (2π)3 2Ek (2.20) one has for the mode amplitudes aP (k) = a(−k) and bP (k) = b(−k). (2π)3 2Ek (2. (2.Relativistic wave equations 10 In Eqs 2.6 We will explicitly discuss the example of a discrete symmetry.19) implying that for each solution φ(x) there exists a corresponding solution with the same energy. It is trivial to see that ∂µ ∂ µ + M 2 φ∗ (x) = 0.15 one has elimated k 0 and in both equations k · x = Ek t − k · x. (2π)3 2Ek (2. x) → (t. 2. for which we consider space inversion.17) (2.14 and 2.

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

5 (a) As an introduction to parts (b) and (c). Exercise 2. (c) Similarly. ∂t where ψ is a wave function and the nature of α and β is left open for the moment. and β 2 = I. Exercise 2. j 0 = ψ†ψ and j = ψ † αψ.15. B} = AB + BA. β} = 0. Note that a{µ bν} indicates symmetrization. i. i (a) Show that relativistic invariance. a{µ bν} ≡ aµ bν + aν bµ . 2.2) in terms of the a(k) and b(k) using the expansion in Eq. We will encounter these quantities later as energy and momentum. αj } = 2 δ ij I. an attempt to construct a ﬁrst order diﬀerential equation for the time evolution would be to write. 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). express the quantities E(t) = P i (t) = d3 x (∂0 φ)∗ (∂0 φ) + ∇φ∗ · ∇φ + M 2 φ∗ φ .e. From this one concludes that α and β must be matrix-valued and ψ must be a multi-component wave function.Relativistic wave equations 12 Exercise 2.7 To solve the problem with positive and negative energies and get a positive deﬁnite density. {αi . d3 x (∂ {0 φ)∗ (∂ i} φ).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. 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 . (b) Show that one has current conservation for the current 2 We denote the commutator as [A. One then ﬁnds that Q is expressed as an inﬁnite sum over number operators a† a (particles) and b† b (anti-particles). 1 2 (b) Express now the full charge QV (t) (exercise 2. 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 . t) = (−i α · ∇ + m β) ψ(r. also using the result in (a). We will see a∗ (k) → a† (k) and a(k) → aop (k) with aop and a† op op being annihilation and creation operators (as known from a harmonic oscillator) and the same for the coeﬃcients b(k).1) for a complex scalar ﬁeld current (exercise 2. t). in terms of the a(k) and b(k). B] = AB − BA and the anticommutator as {A. .

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

2 (3. with real a’s and 3 (ai )2 = 1. [z. It is straightforward to check that any (ﬁnite) rotation can be obtained from a combination of inﬁnitesimal rotations. With matrix commutation this algebra satisﬁes the requirements for a Lie algebra. ˆ therefore.Groups and their representations 14 (n. thus is as the surface i=0 of a sphere in 4 Euclidean dimensions.z N N or (Lk )ij = −i ǫijk . π ) π -sphere (antipodes identified) 2 π -sphere (surface identified) Figure 3. In the same way we can consider rotations around the x.2 π) (-n.and y-axes 0 0 0 0 0 0 0 −i . and thus (σ · n)2 = 1. y] = −[y. ˆ R(ϕ. n) = 1 cos ˆ 1 Note ϕ ϕ ˆ + i (σ · n) sin 2 2 = exp i ϕ ˆ σ·n .6) Summarizing. 0 (3.5) Rotations in general do not commute. x] = 0 (thus [x.8) ˆ that (σ · a)(σ · b) = a · b + i σ · (a × b). (3.1: the parameter spaces of SO(3) (left) and SU (2) (right). π) (n. y]] = 0 (Jacobi identity). [x. They satisfy the commutation relations [Li . y ∈ A ⇒ [x. namely that there exists a bilinear product [. we turn to the group SU (2) of special (det A = 1) unitary (A† = A−1 ) 2 × 2 matrices. for rotations around z for instance. • [x. These generators form a threedimensional Lie algebra SO(3). [y. .7) = a0 0 −1 i 0 1 0 0 1 • [x. x]] + [z. the rotations in SO(3) can be generated from inﬁnitesimal rotations that can be expressed in terms of a basis of three generators Lx . is a 3-dimensional Lie-group. x]). z]] + [y. z ) = lim N →∞ that are generated by i 0 . Locally this is a 3-dimensional Euclidean space and SU (2). Lx = Ly = 0 0 0 i 0 −i 0 ϕ ˆ . ] that satisﬁes • ∀ x. These matrices can be deﬁned as acting on 2-component spinors (χ → Aχ) or equivalently as acting on 2 × 2 matrices (B → ABA† ). which reﬂects itself in the noncommutation of the generators. One way of viewing the parameter space. Next. Lj ] = i ǫijk Lk . Ly and Lz . (3. 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. Writing a0 = cos(ϕ/2) and a = n sin(ϕ/2) we have1 A = A(ϕ.4) R . y] ∈ A.

now with radius 2π and with all points at the surface identiﬁed (see ﬁgure). σy /2 and σz /2 form the basis of the Lie-algebra SU (2). also can be considered as a ﬁlled 3-sphere. . sometimes in more than one way (but this will be discussed later). n) ≈ 1 + i ϕ J · n. n) i ∂ϕ = ϕ=0 (3. They satisfy σi σj σk = i ǫijk . n) π ≤ ϕ ≤ 2π. . i.12) One. ˆ ˆ A(ϕ. [g. one has a Lie algebra isomorphism that is linear and preserves the bilinear product.9) The parameter-space of SU (2). since any ﬁnite transformation can be constructed from the inﬁnitesimal ones.11) Thus σx /2. with ˆ J ·n ≡ ˆ 1 ∂A(ϕ. 2! . H] = 0 for g ∈ G. The next issue is to ﬁnd the appropriate symmetry operators in the Hilbert space. (3. . For instance in a function space translations. n) −→ R(ϕ.13) 0≤ϕ≤π The relation A(σ · a)A−1 = σ · RA a (valid for any vector a) can be used to establish the homomorphism (Check that it satisﬁes the requirements of a homomorphism. 2 (3. (3. thus. it is suﬃcient that the generators commute with H. n) ˆ −→ R(2π − ϕ. . thus. i. the SU (2) → SO(3) mapping is a 2 : 1 mapping where both A = ±1 are mapped into R = I. φ′ (r) = φ(r + a) = φ(r) + a · ∇ φ + 1 (a · ∇)2 φ + .e. n! 2! n=0 ∞ (3. .2 Representations of symmetry groups The presence of symmetries simpliﬁes the description of a physical system and is at the heart of physics.Groups and their representations 15 where we have used the (for operators new!) deﬁnition eA ≡ An 1 = 1 + A + A2 + . immediately sees that the Lie algebras are identical. i. (3.15) [g. H] = 0 for g ∈ G. In the full parameter space. There exists a corresponding mapping of the groups given by µ: SU (2) −→ SO(3) ˆ ˆ A(ϕ. The existence of symmetries means that there are operators g belonging to a symmetry group G that commute with the Hamiltonian. for ﬁxed n. in particular that AB → RAB = RA RB ). .e. 3.e. . Near the identity.10) σ ˆ · n. The inﬁnitesimal generators of SU (2) are ˆ obtained by considering inﬁnitesimal transformations. Suppose we have a system described by a Hamiltonian H. the above mapping corresponds to the trivial mapping of the Lie algebras.14) For a Lie group. 2 2 2 (3. however. SU (2) ≃ SO(3).

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

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

proof: det(ΛT GΛ) = det(G) → (det Λ)2 = 1.24) L↑ + L↓ + L↑ − L↓ − proper orthochrone proper non-orthochrone improper orthochrone improper non-orthochrone .22) (Note that xT is a row vector). which as a matrix equation with (Λ)µν = Λµν and (G)µν = gµν gives (ΛT )ρµ (G)µν (Λ)νσ = (G)ρσ . it is easy to derive some properties of the matrices Λ: (i) det(Λ) = ±1. e To derive some of the properties of the Lorentz group. consisting of the Lorentz group and translations.Groups and their representations 18 3. 0 0 g00 g01 g02 g03 1 0 g g11 g12 g13 0 −1 0 0 = . (det Λ = +1 is called proper. e. it is convenient to use a vector notation for the points in Minkowski space. 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 . 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. explicitly x′µ = (Λ)µν xν + aµ or x′µ = Λµν xν + aµ . In this section we consider the Poincar´ group. From this property. proof: (ΛT GΛ)00 = (G)00 = 1 → Λµ0 gµν Λν0 = 1 → (Λ00 )2 − i (Λi 0 )2 = 1. Thus one has the (pseudo-orthogonality) relation ΛT GΛ = G ⇔ GΛT G = Λ−1 ⇔ ΛGΛT = G. (3.26) (3. We considered the generators and looked for representations in ﬁnite dimensional spaces.g.27) (3. (ii) |Λ00 | ≥ 1. G = 10 (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. σ/2 in a two-dimensional (spin 1/2) case.23) The proper tensor structure of the matrix element (Λ)µν is a tensor with one upper and one lower index. Writing x as a column vector and the metric tensor in matrix form.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. Denoted in terms of column vectors and 4-dimensional matrices one writes for the poincar´ transe formations x′ = Λ x + a.25) (3. det Λ = −1 is called improper).

J 2 . J 3 ). The ﬁrst two sets of commutation relations exhibit the rotational behavior of J and K as vectors in E(3) under rotations. From the commutation relations one sees that the boosts (pure Lorentz transformations) do not form a group. Using these explicit transformations. since the generators K do not form a closed algebra. 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). one distinguishes rotations 0 + and boosts. K j ] = −i ǫijk J k . Thus ˆ 0′ 0 0 0 V 0 V 1 0 0 0 0 1′ 1 V 0 cos ϕ sin ϕ 0 V 0 0 −i 0 2′ = 2 −→ J3 = V 0 − sin ϕ cos ϕ 0 V 0 i 0 0 . K 3 ). The commutator of two boosts in diﬀerent directions (e. (3. and those of the boosts. K = (K 1 . K 2 . z ) = exp(−i ηK 3 ). since I ∈ L↑ and + + det Λ and Λ0 change continuously along a path from the identity. which satisfy the commutation relations (check!) [J i . 0 0 +1 0 0 +1 0 0 ∈ L↑ (identity) I = (3. inﬁnitesimally given by ΛB (η. z ) = exp(i ϕ J 3 ). Returning to the global group. J j ] = i ǫijk J k . βγ = sinh η. we have found the generators of rotations.30) 0 + 0 +1 0 0 0 0 +1 0 0 −1 0 0 +1 0 0 ∈ L↓ (time inversion) It = (3.29) 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↑ . (3. 3 i 2 i=1 (Λ0 ) 3 19 (iii) Boosts along the z-direction are given by ΛB (η. [J i .31) 0 − 0 +1 0 0 0 0 +1 0 0 +1 0 0 −1 0 0 ∈ L↑ (space inversion) Is = (3. 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.Groups and their representations = i=1 (Λ0 )2 . Rotations around the z-axis are given by ΛR (ϕ.g.28) 3′ 3 V 0 0 0 1 V 0 0 0 0 The parameter η runs from −∞ < η < ∞. 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 . This is the origin of the Thomas precession. z ) ≈ ˆ ˆ I − i η K 3 .32) 0 − 0 −1 0 0 0 0 −1 . it is easy to ﬁnd the following typical examples from each of the four classes. z ) ≈ I + i ϕ J 3 . K j ] = i ǫijk K k . i proof: use ΛT GΛ = G and ΛGΛT = G. [K i . J = (J 1 . inﬁnitesimally given by ˆ ΛR (ϕ. In L↑ .

40) ΛB = I + i η 3 K 3 = I − i ω30 M 30 . we continue using covariance arguments.38) Explicitly. ǫ).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. We now want to ﬁnd a covariant form of the six generators of the Lorentz transformations. The extension to the Poincar´ group is e ↑ straightforward. we have for the (inﬁnitesimal) rotations (around z-axis) ΛR = I + i ϕ3 J 3 = I − i ω12 M 12 .39) (e. This is a + normal subgroup and the factor group is the Vierer group. [M µν . One can easily convince oneself that i (ω)µν = ω µν = − ωαβ (M αβ )µν . P+ etc.4 The generators of the Poincar´ group e ↑ The transformations belonging to P+ are denoted as (Λ. with ϕi = − 1 ǫijk ω jk and J i = 2 z-axis). We will just require Poincar´ invariance for the generators themselves. In order to ﬁnd the commutation relations including the translation generators.35) thus (ω)ij = −(ω)ji and (ω)0i = (ω)i0 . K = M ). The e 2 These commutation relations are the same as those for the ’quantummechanics’ operators i(xµ ∂ ν − xν ∂ µ ) . Of the four parts only L↑ forms a group.33) These four transformations form the Vierer group (group of Klein). (3. 3. There are six generators. inf (3. Thus the two vector operators J and K form under Lorentz transformations an antisymmetric tensor M µν . We therefore ﬁnd (again) that there are six generators for the Lorentz group. 3 30 with η = ω and K = M (e. J 3 = M 12 ) and for the (inﬁnitesimal) boosts (along (3. the Lorentz transformations form a (Lie) group (Λ1 Λ2 and Λ−1 are again Lorentz transformations.34) The condition ΛT GΛ = G yields g ρµ + ω ρµ gρσ (g σν + ω σν ) = gµν =⇒ ωνµ + ωµν = 0.g. three of which only involve spatial coordinates (rotations) and three others involving time components (boosts).36) (3. i i0 i i0 1 ijk M jk 2ǫ (3. Summarizing. 2 (M αβ )µν = i g αµ g βν − g α g βµ ν (3. inﬁnitesimally approximated by (I + ω. (3.g. Multiplying the proper orthochrone transformations with one of them gives all Lorentz transformations.37) The algebra of the generators of the Lorentz transformations can be obtained by an explicit calculation2 . Also this group can be divided into four parts. a). explicitly reading x′µ = = (Λ)µν xν + aµ = ((I)µν + (ω)µν ) xν + ǫµ Λµν xν + aµ = (g µν + ω µν ) xν + ǫµ . M ρσ ] = −i (g µρ M νσ − g νρ M µσ ) − i (g µσ M ρν − g νσ M ρµ ) . which are obtained by writing the inﬁnitesimal parameters (ω)µν in terms of six antisymmetric matrices (M αβ )µν .

. The requirement that P µ transforms as a four-vector (for which we know the explicit behavior from the deﬁning four-dimensional representation). known from non-relativistic quantum mechanics are obtained. a)P µ U (Λ. H] = [J i . Λµν pν ⇒ ωµν pν ⇒ ωµν = = = pµ + ω µν pν = pµ 0 ǫµνρσ pρ sσ . = −i (g µρ P ν − g νρ P µ ) . ǫ) = 1 − ωαβ M αβ + i ǫα P α . .Groups and their representations 21 inﬁnitesimal form of any representation of the Poincar´ group (thus including the translations) is of e the form i U (I + ω. the generators J 2 and J 3 in the case of the rotation group. [K i .5 Representations of the Poincar´ group e where pµ is a set of four arbitrary real numbers. the generators J± in the case of the rotation group. To ﬁnd other generators that commute with P µ we look for Lorentz transformations that leave the four vector pµ invariant.43) giving the following commutation relations by equating the coeﬃcients of ǫµ and ωµν . the tensor M µν in terms of boosts cK i = M i0 and rotations J i = 1 ǫijk M jk . H] = i P i . P ) in terms of the Hamiltonian and the three-momentum operators. a) = Λ−1 µ P ν . because one then sees that by letting c → ∞ the commutation relations of the Galilei group. [K i . (3.42) (to decide on where the inverse is. one obtains 2 [P i . a)P µ U † (Λ. [J i . J j ] = i ǫijk J k . In order to label the states in an irreducible representation we construct a maximal set of commuting operators. (3. that commute e among themselves. . We have here reinstated c. They just state that M µν transforms as a tensor with two Lorentz indices.44) (3. [J i . (1 + iǫγ P γ − i i ωαβ M αβ )P µ (1 − iǫδ P δ + ωρσ M ρσ ) = P µ + ω µν P ν .38 could have been obtained in the same way. P ρ ] = 0. leads to U (Λ. other states in that representation are obtained by the action of operators outside the maximal commuting set. Explicitly. The eigenvalues of these will be the four-momentum pµ of the state. K j ] = −i ǫijk J k /c2 . Of the generators of the Poincar´ group we choose ﬁrst of all the generators P µ . as part of the set. [J i . writing the generator P = (H/c.45) Note that the commutation relations among the generators M µν in Eq. K j ] = i ǫijk K k . P ν ] [M µν . P j ] = [P i . = pµ |p.48) . The generators of the translations are the (momentum) operators P µ . . For instance. check that U1 U2 corresponds with Λ1 Λ2 ). These form a group called the little group associated with that four vector.47) (3. P j ] = i ǫijk P k . H] = 0. Taking any one of the states in an irreducible representation. (3.46) [K i . For instance. Inﬁnitesimally. 3. a) = Λµν P ν or U † (Λ. (3. These deﬁne states with speciﬁed quantum numbers that are eigenvalues of these generators.41) 2 At this point. P j ] = i δ ij H/c2 . 2 2 (3. the generators M αβ no longer exclusively act in Minkowski space. P µ |p. . ν (3. . 3. [P µ . .

. 0. p0 > 0. i exp − ωµν M µν 2 i exp − ǫµνρσ M µν pρ sσ 2 (3. 2 (3. . while the others have no physical signiﬁcance. = = i exp − ǫµνρσ M µν pρ sσ |p. i ǫµνρσ W ρ P σ . 2 exp (−isµ W µ ) |p.54) This will enable us to pick a suitable commuting ’spin’ operator. 0.50) where the Pauli-Lubanski operators W µ are given by 1 Wµ = − ǫµνρσ M νρ P σ . Given the momentum four vector pµ we choose a tetrade consisting of three orthogonal spacelike unit vectors ni (p). 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. .Groups and their representations 22 where s is an arbitrary vector of which the length and the component along pσ are irrelevant.49) (3. . .51) The following properties follow from the fact that W µ is a four-vector by construction. . From the algebra of the generators one ﬁnds that P 2 = Pµ P µ and W 2 = Wµ W µ (3.55) commute with all generators and therefore are invariants under Poincar´ transformations. (3. Wα ] = [Wµ . the third case represents the vacuum. Wν ] = −i (gµα Wν − gνα Wµ ) . The elements of the little group thus are U (Λ(p)) = = and U (Λ(p))|p. (3. These e operators are the Casimir operators of the algebra and can be used to deﬁne the representations of ↑ P+ .52) (3. Massive particles: p2 = M 2 > 0. Pν ] = [Wµ . . satisfying gµν nµ (p)nν (p) = j i nµ (p) pµ i ǫµνρσ pµ nν nρ nσ i j k = = −δij .53) (3. which can thus be chosen spacelike. . . M ǫijk .56) . of which the components generate the little group of pµ or (for the third of the following relations) from explicit calculation [Mµν .

p. 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. 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.Groups and their representations 23 We can write W µ (p) = 3 W i (p) nµ (p) i i=1 (3. 0. it is suﬃcient to choose a particular frame to investigate the coeﬃcients in Eq. .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. 0. show that W /M can be interpreted as the intrinsic spin. (0.57) (i.64) such that in an arbitrary frame where the four vectors p. 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 . p0 > 0. ms . M2 (3.61) one sees that S i (p) i =− W2 .65) .63) where E = p2 + M 2 and ms is the z-component of the spin −s ≤ ms ≤ +s (in steps of one). W i (p) = −W · ni (p)). −1) n1 (p0 ) ≡ n2 (p0 ) ≡ s(p0 ) ≡ (3. In that case the vectors ni (p) are just the space directions and W = (0. (3..62) 1 Thus W 2 has the eigenvalues −M 2 s(s + 1) with s = 0. Having made a covariant decomposition. 0. 0. 1. The explicit construction of the wave function can be done using the D-functions (analogous as the rotation functions). n1 (p). M2 (3.e. 2 . The commutation relations [W i (p). 1. (0. .e. i (3. S j (p)] = −i nµ (p)nν (p) j i ǫµνρσ W ρ P σ = i ǫijk S k (p). 0. W j (p)] = i M ǫijk W k (p). s. . Together with the four momentum states thus can be labeled as |M. 0. the S i (p) form the generators of an SU (2) subgroup that belongs to the maximal set of commuting operators.60) and using the completeness relation nµ (p)nν (p) = − g µν − i i 2 i pµ pν M2 . 1 2 (3. 0). 0). |p3 |).58) i. 3. 1. 0). (1.59) (3. More generally (fully covariantly) one can proceed by deﬁning S i (p) ≡ W i (p)/M and obtain [S i (p). M J) with components W i = (M/2)ǫijk M jk . (3. Noting that S i (p) i 2 = 1 W µ (p)W ν (p) M2 niµ (p)niν (p).

W 1 (p)] [λ(p). 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). This algebra is (for instance) isomorphic to the one generated by rotations and translations in a 2-dimensional Euclidean plane (see exercises). Exercises Exercise 3. |p| (3. The eigenvalues of W 1 and W 2 (in that case corresponding to the translations) thus can take any continuous values. −i W 1 (p). where A = exp(i ϕ · σ/2) = exp(iϕ σ · n/2). a)|0 = |0 . the action of RA b · RA a = = ˆ ˆ ˆ b · a cos ϕ + n · (b × a) sin ϕ + 2(b · n)(a · n) sin2 (ϕ/2) ˆ ˆ b · a + b · (ˆ × n) sin ϕ − 2 (b · a − (b · n)(a · n)) sin2 (ϕ/2) a .13. W 2 and λ is derived from the general commutation relations for [Wµ . is characterized by its momentum and the helicity. |p.69) We reproduce the algebra for the massive case with W 3 (p) = M λ(p). Thus 2 λ(p) = J ·p . The vacuum: pµ = 0. W 2 (3. But if M = 0 one has a diﬀerent algebra (Eq. U (Λ.73) (3. W µ (p) = λ(p) P µ . thus.1 (a) Derive from the relation that deﬁnes the mapping of SU (2) matrices A into SO(3) matrices RA in Eq. The algebra of W 1 . (3. P µ .Groups and their representations The above does actually include the massive case. Thus we have P 2 .71) which is called the helicity. 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. (3. W 2 (p)] = = = i M 2 λ(p). This state is in physical applications nondegenerate and is denoted by |0 . Wν ].72) One has P µ |0 = W µ |0 = 0 and the state is invariant under Lorentz transformations. 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). σ · RA a ≡ A(σ · a)A−1 ˆ (valid for any vector a). λ .66) where W · p = 0 implies that W s (p) = −(M 2 /p · s) λ(p). 3. i W 2 (p). implying continuous values for the spin (actually for W i = M S i ).67) (3.70) as a commuting set with zero eigenvalues for P 2 and W 2 . which can take any integer or half-integer value. W 2 (p)] [λ(p). and gives [W 1 (p). 3. Note that angular momentum does not contribute to helicity as L · p = 0. A massless particle. 1 2 24 (3.68) (3.67 has a vanishing right hand side). however. We don’t know of any physical relevance.

O] + . Is this easy to understand. triplet and anti-triplet representation. Now write down the three isospin 1 meson quark-antiquark (meson) states. i. The appropriate isospin doublet to be used for antiquarks then is (−d. what are the matrices J c in the Lie-algebra representation describing the spin rotations for two-component ’anti-spinors’. one has to explicitly include this rotation. the pions and the isospin 0 meson state. Note the diﬀerence with the (usual) spin 0 and spin 1 states composed from two spin 1/2 states. Tr(σi σj σk ) = 2i ǫijk .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. φ0 ) doublet representation for SU (2)weak . For a spin 1/2 antiparticle the representation A∗ is needed.2 Show that for a unitary operator U = exp[iαk Jk ] with real coeﬃcients αk . Show that if det(U ) = 1. Exercise 3. respectively). In both of these cases the symmetry is broken by the charges ¯¯ of the particles.Groups and their representations 25 for any vectors a and b. This is true for SU (2) but it is.e. show that the transformation for the operators is given by O′ = U O U −1 = O + i [J. Excercise 3.4 (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. not true for SU (3) (ﬂavor or color properties of particles know two distinct deﬁning 3-dimensional representation. . . and calculate U (a) p U −1 (a) . Tr(σi σj σk σl ) = 2 (δij δkl − δik δjl + δil δjk ) . (b) This establishes that for spin 1/2 one doesn’t need to talk about ’anti-spin’ for the antiparticle. Show actually that ǫ = 2 eiπJ (up to a phase). . Examples are the quark (u. (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). (b) Show that this transformation for the position operator in quantum mechanics indeed gives U (a) r U −1 (a) = exp (i a · p) r exp (−i a · p) = r + a. the (non-strange) η-meson state. d) doublet representation for SU (2)isospin or the Higgs (φ+ . Both particle and antiparticle have ’spin 1/2’. u). the trace of the operator Jk is zero (Convince yourself that det(eA ) = eTr A by diagonalizing A. for instance. Show that the Lie-algebra representations J and J c are equivalent. the operators Jk must be hermitean. Writing A∗ = exp(i α · J c ). show that one matrix ǫ exist such that J c = ǫ J ǫ† . Exercise 3. Note: Recall (or derive) that the Pauli matrices σ satisfy σi σj = δij + i ǫijk σk and thus Tr(σi σj ) = 2 δij . (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.

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

(4. C) can be written as a product of a unitary + matrix U and a hermitean matrix H. It is simply connected and forms the covering group of L↑ . j) K = −i J (B = 0). j ′ ). K = −i .Chapter 4 The Dirac equation 4. The group SL(2. This tells us how to ﬁnd the representations of the group. C) is the group of complex 2 × 2 matrices with determinant one.1) (4.4) (4. K = +i .3) (4. M = exp i ϕ·σ 2 1 exp ± η · σ 2 = U (ϕ) H(η) . 2 2 (4.10) . K = i J (A = 0). 0. similarly as the homeomorphism between SO(3) and SU (2).2) This shows that the Lie algebra of the Lorentz group is identical to that of SU (2) ⊗ SU (2). They will be labeled by two angular momenta corresponding to the A and B groups. Special cases are the following representations: Type I : (j.5) = 1 (J + i K).1 The Lorentz group and SL(2. U (ϕ) H(η) (4. With this choice of parameter-spaces the plus and minus signs are actually relevant. C). B j ] = [Ai . i ǫijk B k . 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 . becoming the deﬁning representations of SL(2. (j.7) From the considerations above. 2 2 σ σ J = . 0) Type II : (0. respectively. The matrices in SL(2. (4. B j ] = i ǫijk Ak . where we restrict (for ﬁxed n) the parameters 0 ≤ ϕ ≤ 2π and 0 ≤ η < ∞.6) (4. 2 (4. C): Type I (denoted M ): Type II (denoted M ): 27 J= σ σ .9) (4. 2 1 (J − i K). They precisely correspond to the two types of representations that we have seen before. it also follows directly that the Lorentz group is homeomorphic with the group SL(2.8) ˆ ˆ ˆ with η = η n and ϕ = ϕn. Aj ] = [B i .

−σ). thus. (4. With E = M γ = M cosh(η) and p = M βγ n = M sinh(η) n. C). U σ µ aµ U ˜ −1 H σ µ aµ H −1 = σ µ ΛB aµ H σ µ aµ H ˜ −1 (4.17) 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.11) Considering ξ and η as spin states in the rest-frame. Eqs 4.16) = σ µ Λ R aµ . H(η) = exp(−η · σ/2). C) and. One has + −1 Λ(M ∗ ) = I2 Λ(M ) I2 . σ). ǫ−1 = ǫ† = ǫT = −ǫ).1). the ﬁrst equation shows that the conjugate representation with spinors χ∗ transforming according to −σ∗ /2 is equivalent with the ordinary two-dimensional representation (±ǫχ∗ transforms according to σ/2). such that M = SM S −1 .18 show that ±ǫξ ∗ transforms as a type-II (η) spinor. ξ −→ Hξ.e.15) (4. σ µ ≡ (1.13) 2M (E + M ) (exercise 4. ˜ Non-equivalence of M and M One might wonder if type I and type II representations are not equivalent. i. This is not true for the two-component spinors in SL(2. Note that M ∗ = ǫ† M ǫ has its equivalent for the Lorentz transformations Λ(M ) and Λ(M ∗ ). C). thus. H ≡ (H † )−1 = H −1 ). C) or L↑ . One deﬁnes spinors ξ and η transforming similarly under unitary rotations (U † = U −1 .18) As already discussed for SU (2). however. 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. while ±ǫη ∗ transforms as a type-I + spinor. This shows that M ∗ ≃ M . In fact. U ≡ (U † )−1 = U ) U (ϕ) = exp(i ϕ · σ/2) but diﬀerently under hermitean boosts (H † = H. cannot be connected by a Lorentz transformation belonging to L↑ . η → U η. if there does not exist a unitary matrix S.19) . (4. η → Hη. ¯ ¯ H ∗ = ǫ† Hǫ = ǫ−1 Hǫ. Λ(ǫ) ∈ L+ .14) satisfying Tr(σ µ σ ν ) = −2 gµν and Tr(σ µ σ ν ) = 2gµν = 2 δ µν (the matrices. (4.12) ξ → U ξ. ˜ = σ µ Λ B aµ . there exists the matrix ǫ = iσ 2 . Also useful is the relation H 2 (p) = σ µ pµ /M = (E + σ · p)/M . ˜ The sets of four operators deﬁned by deﬁnitions σ µ ≡ (1. (ǫ∗ = ǫ. namely H(η) = exp(η · σ/2). one can use a boost to the frame with momentum ˆ ˆ p. The following relations for + rotations and boosts are useful. 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. ˜ (4. The Lorentz transformations Λ(M ) and Λ(M ). are not covariant!) ˜ can be used to establish the homeomorphism between L↑ and SL(2. (4. (4. U σ µ aµ U −1 = σ µ ΛR aµ .

m) = χm uL H(p) 0 . + − 1 0 0 1 I2 = 0 0 0 0 0 0 −1 0 0 0 0 1 .13. 0).e. 2 ) and ( 2 . ) −→ ( . (scalar). (4. C). we have seen that in the rest frame W i (p)/M → J i .23) u(p. Examples (with between brackets the Euclidean assigments) are r t p H J λ(p) K −→ −λ(p) −→ −K −→ H −→ J −→ t (scalar). ǫµνρσ will change sign. 0) are inequivalent. (axial vector). −→ −r (vector). generators. This provides the easiest way of seeing ˜ what is happening. 0) of SL(2. must combine the repe resentations. i. they cannot be connected by a unitary 2 2 transformation. and the angular momentum in the rest frame is what we are familiar with as the spin of a particle. The representing member of the class P− is the parity or space inversion operator Is . 4. For a particle at rest only angular momentum is important and we can choose ξ(0.22) u= η η M (Λ) 0 where M (Λ) = ǫM ∗ (Λ)ǫ−1 with ǫ given in Eq. 0) (via operators J and K) one sees that under parity 2 2 1 1 (0. = (4.e. as we have seen in section 4. m) = χm .The Dirac equation 29 where The matrix I2 does not belong to L↑ . therefore. they can be connected. the aij elements of a tensor will not change sign. −→ M (Λ) (4. C) are suitable for representing spin 1/2 particles. 0 χm uR H(p) . m) = η(0. but to L↑ . consider the four component spinor that transforms under a Lorentz transformation as 0 ξ ξ . 1 ) and ( 1 . etc. Within the Lorentz group. the well-known two-component spinor for a spin 1/2 particle. a) and ˜ a = (a0 .2 Spin 1/2 representations of the Lorentz group 1 1 Both representations (0. It has the same eﬀect as lowering the indices. (vector). For the spin 1/2 representations of the Poincar´ group including parity we. 1 ) and ( 1 . −a).21) In nature parity turns (often) out to be a good quantum number for elementary particle states. (pseudoscalar). We also have seen that the representations (0. −→ −p (vector). A parity transformation changes aµ into aµ . Although these operators cannot be used to label the representations.1. e. i. Taking M (Λ) = H(p). but by a transformation belonging ↑ ↑ to the class P− . From the deﬁnition of the representations (0. 4.g. Thus M ∗ and M are not equivalent within SL(2. the boost in Eq. that have the correct commutation relations. but rather the operators W i (p) should be used.20) 4. where a = (a0 .17. The i angular momentum operators J are represented by σ i /2. we obtain for the two components of u which we will refer to as chiral right and chiral left components. The behavior is the same for classical quantities. 2 2 (4.

. The explicit realization appearing in Eq. We will discuss another explicit realization of this algebra in the next section.30) γ µ γ ν ∂µ ∂ν + M 2 ψ(x) = 0. which in general is a linear equation in pµ . one must require that for ψ(x) the Klein-Gordon relation 2 + M 2 ψ(x) = 0 is valid for each component separately).27 is known as the Weyl representation. γ ν } = 2 g µν . (4.31) To achieve also that the energy-momentum relation p2 = M 2 is satisﬁed for u(p). ψ † −iγ 0 ∂0 +iγ i ∂i −M 1 We ← ← = 0. It is of a form that we also played with in Exercise 2. To see this consider the equation for the hermitean † † conjugate spinor ψ † (noting that γ0 = γ0 and γi = −γi ). (4. (iγ µ ∂µ − M ) ψ(x) = 0.The Dirac equation 30 with H(p) = ¯ H(p) = 2M (E + M ) E+M −σ·p 2M (E + M ) E+M +σ·p . (4. From this one obtains the Cliﬀord algebra for the Dirac matrices. (4. (pµ γ µ − M ) u(p) ≡ (/ − M ) u(p) = 0.24) (4. p (4.33) deﬁne for a four vector a the contraction a = aµ γµ / . We ﬁrst want to investigate if the Dirac equation solves the problem with the negative densities and negative energies. (4. 4.29) 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. Since ∂µ ∂ν is symmetric. γ } ∂µ ∂ν + M 2 ψ(x) = 0.28) which is a covariant (linear) ﬁrst order diﬀerential equation. = (4. 0 H 2 (p) uR uR . = H −1 (p). Applying (iγ µ ∂µ + M ) from the left gives (4.7.32) suppressing on the RHS the identity matrix in Dirac space. {γ µ . The ﬁrst indeed is solved. 2 (4.27) where γ µ are 4 × 4 matrices called the Dirac matrices1 . This leads to the Dirac equation for ψ(x) = u(p)e−i p·x in coordinate space. this can be rewritten 1 µ ν {γ .25) It is straightforward to eliminate χm and obtain the following constraint on the components of u. The general requirements for the γ matrices are thus easily obtained. As in section 2 we can use Pµ = i∂µ as a representation for the momenta (translation operators) in function space.

the one of the negative energy states. (4. 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. (4.36) (4. 4. The density j 0 = ψγ 0 ψ = ψ † ψ (4. remains. one sees that there are two positive and two negative eigenvalues.3 General representations of γ matrices and Dirac spinors {γ µ . The Dirac sea forms the vacuum. but with properties such that it can be annihilated by the particle (e.g. . . We will see how this is properly implemented (also for bosons!) when quantizing ﬁelds.The Dirac equation 31 E empty levels M -M filled levels E particle M -M hole Figure 4.1: The Dirac sea of negative energy states and antiparticles. This can most easily be seen in the particle rest frame.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 . E = +M (twice) and E = −M (twice). From the vacuum a particle can be removed. (4. This hole forms an antiparticle with the same mass. Relying on the Pauli exclusion principle for spin 1/2 particles. γ =ρ ⊗1= γ = iρ ⊗ σ = 0 −1 −σ k 0 2 We (4. From the equations for ψ and ψ one immediately sees that j µ = ψγ µ ψ is a conserved probability current (exercise!). where γ 0 p0 ψ = M ψ 0 −→ Eψ = M γ 0 ψ.34) for the adjoint spinor ψ ≡ ψ † γ0 .35) is indeed positive and j 0 can serve as a probability density.37) Using the explicit form of γ in the Weyl representation (see Eq. 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. This problem was solved by Dirac through the introduction of a negative energy sea. γ ν } = 2 g µν .39) use ρi ⊗ σj with both ρ and σ being the standard 2 × 2 Pauli matrices. opposite charge).27).

We note that ψ → Lψ and ψ → ψ L−1 .49) i 1 with −i S ρσ = 2 σ ρσ = − 4 [γ ρ .R.45) (γ5 )W. −→ Sψ. 1 −1 2 For all representations one has † γµ = γ0 γµ γ0 . −→ Sγµ S −1 .10 are 1 σ3 0 i J 3 = M 12 = = [γ 1 . 4! (4.The Dirac equation 32 The Weyl (or chiral) representation: 0 1 . 0 γ k = −iρ2 ⊗ σ k = k σ −σ k . γ 0 ]. which in the case of the Weyl representation are just the upper and lower components. is 1 1 1 (4. (γµ )S. γ σ ]. Lorentz invariance The Lorentz transformations can also be written in terms of Dirac matrices.R. ν The latter assures Lorentz invariance of the Dirac equation (see Exercise 4. S −1 . that project out chiral right/left states. γ 2 ]. For example. while L−1 γ µ L = Λµ γ ν . = S(γµ )W. it is easy to see that PR/L = 1 (1 ± γ5 ) 2 1 0 . (4. 2 (4. For instance in the Weyl representation (but valid generally).48) are projection operators.47) (4. = ρ ⊗ 1 = 1 0 i ǫµνρσ µ ν ρ σ γ γ γ γ . γ 0 = ρ1 ⊗ 1 = 1 0 γµ ψ Diﬀerent representations can be related to each other by unitary transformations. =ρ ⊗1= 0 −1 3 (4. 0 (4.3) .43) S=√ . (γ5 )S.46) (4.R. K 3 = M 30 = 0 iσ 3 2 4 or in general one has the transformation i ψ → Lψ = exp − ωρσ S ρσ ψ. γ µ } = 0 and explicitly one has 0 1 1 . 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 . the rotation and boost generators in Weyl representation in Eqs 4.42) We note that the explicit matrix taking us from the Weyl representation to the standard representation.40) (4.44) and an adjoint spinor deﬁned by ψ = ψ † γ0 .R. 0 σ3 2 4 1 −iσ 3 0 = − i [γ 3 .9 and 4.41) (4.

∂ (4. usually to be taken unity. that brings ψ → ψ c .g.R.R. (or W. Explicitly. for instance. γ5 ] = 0. The latter implies that the conjugate of a right-handed spinor. In S.59) −→ ψ c = C ψ = η ǫ ξ∗ .52) η ξ one can apply space inversion. This symmetry does not change the spin 1/2 nature. One has ψ c = −ψ T C −1 . ∂ (4. (C)W. ψ= (4. reverse the charge of the particle. x) → x = (t. −iσ 2 0 −ǫ = .g. Note that we have (as expected) explicitly in Weyl x representation in Dirac space P ξ −→ ψ p = γ 0 ψ = η .R. is a left-handed spinor.56) (i/ − M ) ψ c (x) = 0. but it does. (C)S. T (4.55) (4. (i/ − M ) ψ(x) = 0. It is easy to convince oneself that if ψ(x) is a solution of the Dirac equation. As with parity we look for a transformation.57) (4. ψ= (4.58) where ηc is an arbitrary (unobservable) phase. C is real and one has C −1 = C † = C T = −C c and [C. ∂ with the solution ψ c (x) = ηc Cψ (x).54) e. called charge conjugation. 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. = 0 ǫ iσ 2 (4. such that T Cγµ C −1 = −γµ .R.51) The existence of positive and negative energy solutions implies another symmetry in the Dirac equation.The Dirac equation 33 Parity There are a number of symmetries in the Dirac equation.50) Charge conjugation but with ψ p (x) ≡ ηp γ0 ψ(˜) (Exercise 4. T (4. which is again a solution of the Dirac equation.6). parity.53) (4.) and all representations connected via a real (up to an overall phase) matrix S. ψR . −x) and via a few manipulations obtain again the ˜ Dirac equation (i/ − M ) ψ p (x) = 0. 0 −ǫ 0 0 −ǫ 0 . x = (t. e. In any representation a matrix C exist. Some properties of C are C −1 = C † and C T = −C. = = 0 iγ γ = −iρ ⊗ σ = −iσ 2 −iσ 2 iγ 2 γ 0 = −iρ3 ⊗ σ 2 = 0 2 0 1 2 Thus we ﬁnd back the Dirac equation. in Weyl representation we ﬁnd in Dirac space T C ξ −ǫ η ∗ .

s) + v(k. In an arbitrary frame one has s · p = 0 and s(p) can e. 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.67) It is straightforward to ﬁnd projection operators for the positive and negative energy states P+ P− = s p /+M u(p. ψ(x) = u±s (p) e−i p·x . In that frame is a spacelike unit vector sµ = (0. s) e−i k·x b(k. the best representation to describe particles at rest is the standard representation. s) v = . s) = Ep + M σ ·p Ep +M χs . s). s) u = .63) u(p. the spin polarization vector in the rest frame is the starting ˆ point.g. 2M 2M −/ + M p v(p. s′ ) = 2Ep δss′ . s)v(p. s)u(p. ¯ ¯ † ′ u (p. ∂ −iσ · ∇ −i ∂t − M (4. where u satisﬁes −σ · p Ep − M u(p) = 0. s)u(p. It is easy to check that 1 1±σ·s 1 ± γ5 / s ˆ 0 (4.70) = Ps = ˆ 0 1∓σ·s 2 2 . s). ¯ (4. v (p.61) σ·p −(Ep + M ) There are also two negative energy solutions. s ) = v (p. in which γ 0 is diagonal (see discussion of negative energy states in section 4. Choosing χ = −ǫχ∗ (the equivalent spin ¯ 1/2 conjugate representation).4 Plane wave solutions Looking for positive energy solutions ∝ exp(−iEt) one ﬁnds two solutions. s′ ) = 2M δss′ . where v satisﬁes σ·p −(Ep + M ) v(p) = 0. ⇔ (/ − M ) u(p) = 0.1). s)v(p. s ) = 0. s) ei k·x d∗ (k.60) ψ(x) = 0. be obtained by a Lorentz transformation. ⇔ (/ + M ) v(p) = 0. s) and C v T (p.The Dirac equation 34 4.62) where χs are two independent (s = ±) two-component spinors. s) = u(p.. s)v(p. s)¯(p. s)u(p. with E = Ep = + p2 + M 2 . s) . (2π)3 2Ek (4. 2M 2M (4. ¯ ′ ′ Explicit solutions in the standard representation are χs .68) (4. the spinors satisfy C uT (p.66) An arbitrary spin 1/2 ﬁeld can be expanded in the independent solutions. p −σ · p (Ep − M ) Ep + M σ ·p ¯ Ep +M χs χs ¯ For a free massive particle. ψ(x) = v ±s (p) ei p·x .64) (4. s) = u(p. s) = v(p. v(p. Note that the spinors in the negative energy modes (antiparticles) could be two diﬀerent spinors. The ¯ ¯ solutions are normalized to u(p. s ) = v † (p. (4. s′ ) = −2M δss′ .69) = − s In order to project out spin states. s)¯(p. p (4.65) (4. The explicit Dirac equation in the standard representation reads ∂ iσ · ∇ i ∂t − M (4.

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

s)γ ν u(k ′ .) In order to show the use of the above relations.) (2) Tr(1) = 4. (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 . etc. namely the calculation of the quantity 1 ∗ (¯(k. s)(γ ν )kl ul (k ′ .76) ul (k ′ . s′ )(γ µ )ij uj (k. (7) Tr(γ5 γ µ γ ν ) = 0 or Tr(γ5 //) = 0. a / (3) γ µ γ ρ γ σ γµ = 4 g ρσ or γ µ //γµ = 4 a · b.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 ). Properties of traces of γ-matrices (1) Tr(γ µ1 · · · γ µn ) = 0 if n is odd. s)(γ ν )kl u u s. s). s′ )¯i (k ′ . u u (4. s′ )) . s′ ) ¯ u s. s′ )) (¯(k. s′ ). ab (8) Tr(γ5 γ µ γ ν γ ρ γ σ ) = −4i ǫµνρσ .g µρ g νσ + g µσ g νρ ). 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.75) Lµν (k. s)γ µ u(k ′ . ab (4) Tr(γ µ γ ν γ ρ γ σ ) = 4 S µνρσ . s)γ µ u(k ′ . s)¯k (k. u ¯ This leads with explicit Dirac indices to Lµν = = = = 1 2 1 2 ui (k ′ .s′ 1 ′ (/ + m)li (γ µ )ij (/ + m)jk (γ ν )kl k k 2 1 Tr [(/ ′ + m)γ µ (/ + m)γ ν ] . (Use (γ5 )2 = 1 and pull one γ5 through. (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 . more convenient ¯ to use the projection operators introduced earlier. (6) Tr(γ5 ) = 0.5 γ gymnastics and applications An overview of properties can be found in many text books. γν ] (9) σµν γ5 = γ5 σµν = ǫµνρσ γ ρ γ σ /2. k ′ ) = 2 ′ s.g µ1 µ3 Tr(γ µ2 γ µ4 · · · γ µn ) + · · · . (6) γ5 = i γ 0 γ 1 γ 2 γ 3 = −i ǫµνρσ γ µ γ ν γ ρ γ σ /4!. we will give one example. (8) σµν ≡ (i/2)[γµ . (5) Tr(γ µ1 · · · γ µn ) = g µ1 µ2 Tr(γ µ3 · · · γ µn ) . In principle one can take a representation and just calculate the quantity u(k. s)γ µ u(k ′ . For this we ﬁrst have to prove (do this) that (¯(k. (7) γµ γ5 = -γ5 γµ = i ǫµνρσ γ ν γ ρ γ σ /3!. s)¯k (k. ab (4) γ µ γ ν γ ρ γ σ γµ = −2 γ σ γ ρ γ ν . (9) Tr(γ µ1 · · · γ µ2n ) = Tr(γ µ2n · · · γ µ1 ). s′ )(γ µ )ij uj (k. k k 2 . (Use matrices C to transpose the matrices in the expression under the trace and use Tr AT = Tr A. (3) Tr(γ µ γ ν ) = 4 g µν or Tr(//) = 4 a · b.s′ ∗ (4. s′ )) = u(k ′ . It is. however. s′ )γ µ u(k.The Dirac equation 36 4.

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

there is no current ﬂowing through the surface of the cavity. ν p Note: how such parametrizations work will be explained in chapter 8. We will not pursue this chiral freedom at this point and take α = 0. mµ = 105. the decay would even be forbidden. 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.5 and M → ∞. we want to conﬁne this solution to a spherical cavity with radius R. r). Note: To prove this. n (c) Apply the conﬁnement condition to ﬁnd the value of k for the lowest eigenmode in the spherical ˆ cavity. Note: the outcome shows that the decay into muons is much more likely that the decay into electrons (mπ = 139.7 (a) Show that (in standard representation) j0 (kr)χm ψ(x. Excercise 4. namely i / ψ = eiαγ5 ψ. M R = 1. is a stationary solution of the (free) Dirac equation. The pion has spin 0. The conditions for ψ and ψ imply that nµ j µ = ψ/ ψ = 0. t) = N k ˆ i E+M σ · r j1 (kr)χm −i Et e . n It turns out that there is actually a whole family of conditions that provide conﬁnement. For such a cavity nµ = (0. |M (π − → µ− νµ )|2 ¯ Actually if the lepton would be massless.e. you need to determine the corresponding conﬁnement condition for ψ. 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 ). Show that the ratio is |M (π − → e− νe )|2 ¯ ≈ 54. which is the most important factor needed to ﬁnd the decay width (inverse lifetime). the lepton and neutrino have spin 1/2 and are ν described by momentum-space spinors. me = 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 ).511 MeV). (b) Next.66 MeV. Calculate for now just |M |2 . A condition for conﬁning the fermion in the cavity is i / ψ = ψ (where n is the surface normal). (d) Plot for these cases the vector densities ψ † ψ and the scalar densities ψψ as a function of r/R. 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. Do you understand why? The calculation of the life-time is completed in chapter 11. i. Calculate the value of k for M R = 0. Show that this n condition is suﬃcient to guarantee that nµ j µ = 0.57 MeV.8 × 10−6 .The Dirac equation 38 Exercise 4.

3) Since the vectors ǫµ (k) together with the momentum k µ form a complete set of four-vectors. −B 1 0 3 With the electromagnetic ﬁeld tensor (5. ±(1) are obtained by boosting the restframe vectors. λ) . (2 + M 2 )Vµ = 0. The covariant generalization is to describe them with Vµ (x) = ǫµ (k) e±ik·x . λ) + ǫ(λ)∗ (k) eik·x c∗ (k. The vectorﬁeld is therefore (λ) suited to describe spin 1. A).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.Chapter 5 Vector ﬁelds and Maxwell equations 5. 0).± (λ) d3 k ǫ(λ) (k) e−ik·x c(k. These can describe spin 1 states in the rest-frame. namely ǫµ (k0 ) = (0. In an arbitrary frame ǫµ (k) with λ = 0.7) . with ∂µ V µ = 0 (Lorentz condition).5) −E B2 .± 5. In the restframe.1 Fields for spin 1 In section 3. just as the two-component Dirac spinors did for spin 1/2. These are solutions of the Klein-Gordon equation with in addition a condition. where k = k0 = (M. ǫλ ). A real vector ﬁeld Vµ (x) then can be expanded in modes as Vµ (x) = λ=0. one has ǫ(λ) (k) ǫ(λ)∗ (k) = −gµν + µ ν kµ kν . 39 (5. M2 (5. (5.2) (5.2 we have discussed the vectors ǫλ (with λ = 0 and ±1) as basis states of the spin 1 representation.4) λ=0. µ (2π)3 2E µ (5.6) where Aµ is the four-vector potential (5. there are then indeed three independent possibilities.1) The Lorentz condition implies that k µ ǫµ (k) = 0.

As remarked above. 0. in which case one has 2Aµ = jµ . 5. −i.Vector ﬁelds and Maxwell equations and the (conserved) current j µ = (ρ.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 . transverse or Coulomb gauge). 0). . It states the absence of magnetic charges and Faraday’s law.18) corresponding to the helicity states λ = ±1 of the massless photon ﬁeld. for M = 0.13) Hence. i. One possible choice is the gauge A0 = 0 (5. the antisymmetry of F µν implies that ˜ ∂µ F µν = 0 or ∂λ Fµν + ∂µ Fνλ + ∂ν Fλµ = 0. 2Aµ = 0. in accordance with what we found in section 3. (5. (5. only two independent vectors remain ǫ(±) µ (k) = (0. The equation in vacuum. Although the Lorenz condition is a constraint. ∂µ Aµ = 0.e. The Lorenz condition can actually be incorporated by changing the equations of motion to 2Aµ + (λ − 1)∂µ (∂ν Aν ) = jµ .15 the d’Alembertian is replaced by 2 + M 2 . Under a gauge transformation Aµ −→ Aµ + ∂µ χ. (5. The gauge freedom can ﬁrst of all be used to impose the Lorentz condition for the electromagnetic ﬁeld. there is still a gauge freedom left (Aµ → Aµ + ∂µ χ with 2χ = 0).14) of which the solutions give the Li´nard-Wiechert potentials. The equations of motion for the ﬁelds Aµ become 2Aµ − ∂µ (∂ν Aν ) = jµ . ∓1. For electrodynamics one has furthermore the freedom of gauge transformations. one ﬁnds for a conserved current and λ = 0 2∂µ Aµ = 0. Taking the divergence of this equiation. AP (x) = Aµ (˜).10) the electric and magnetic ﬁelds are unchanged. (5.16) (5. Therefore if k µ = (|k|. 40 (5. ˜ x Note that via parity transformation one obtains another solution. it will vanish at all times. Note that the incorporation of the Lorentz condition can also be done for the case of the massive vector ﬁeld. 0. (5.11) This equation is not aﬀected by a gauge transformation. Furthermore. while the fact µ C ∗ that the Minkowski space is real implies that Aµ (x) = Aµ (x) = Aµ (x). if in Eq. shows that the electromagnetic ﬁelds correspond to the case of a massless spin 1 ﬁeld. |k|).12) (5. it does not eliminate the freedom of gauge transformations but they are now restricted to Aµ → Aµ + ∂µ χ with 2χ = 0.15) (5. Fµν −→ Fµν + (∂µ ∂ν χ − ∂ν ∂µ χ) = Fµν .8) (5.5. if ∂µ Aµ = 0 for large times |t|. moree over.17) (radiation. In that case one also has ∇·A = 0 or ki ǫi (k) = 0.

F = e E + ev × B (5. nothing more than a manisfes1 tation of a nontrivial vacuum (a la the Dirac sea for fermions).19) are gauge invariant. In this experiment an observable phase diﬀerence is measured for electrons moving through ﬁeld-free space (E = B = 0 but Aµ = 0). This phase diﬀerence is actually observed. Nevertheless the signiﬁcance of Aµ is shown in the Aharonov-Bohm experiment. In the presence of an electromagnetic ﬁeld Aµ an additional phase diﬀerence is observed. nevertheless. This causes an interference pattern as a function of ∆x. This occurs in the region outside the solenoid. 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 . illustrated in ﬁg. = 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.1: The Aharonov-Bohm experiment 5.0 . A= By Bx Br ϕ= − ˆ . cos ϕ. 0).Vector ﬁelds and Maxwell equations 41 L ∆x d a (infinitely) long solenoid Figure 5. To be precise. . but A = 0 (hence there is structure in the vacuum).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. sin ϕ. but where E = B = 0. 0) and ϕ = (− sin ϕ.1. . where Aµ = 0. The phase diﬀerence between the electrons travelling two diﬀerent paths is given by δ = 2π a a ∆x d = −≈ . The energy density 2 (E 2 + B 2 ) = 0 (hence a vacuum). The situation. λ L − λ λ or ∆x = (L− λ/d)δ. . Inside solenoid: Outside solenoid: B = B z. while the electromagnetic ﬁeld Aµ is not.0 . It is however. ϕ= − ˆ 2r 2r2 2r2 where r = (cos ϕ. ˆ B = 0. 2 2 2 BR2 BR2 y BR2 x A= . (5.20) e (∇ × A) · ds = B · ds = where Φ is the ﬂux through the solenoid. may look a bit akward. 5.

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

. ∇×B =j+ ∇×E+ ∂E .8 and 5. ∂t ∇ · B = 0.1 Show that Eqs. 5.Vector ﬁelds and Maxwell equations 43 Exercises Exercise 5.9 explicitly give the equations ∇ · E = ρ. ∂t ∂B = 0.

3) (6.6) The ﬁrst term leads to the Euler-Lagrange equations. d dt ∂L ∂x ˙ = ∂L . x) ˙ (6. thus t′ = t + δτ.5) with ∆x(t) = δx(t) + x(t) δτ . The requirement δS = 0 with ﬁxed boundaries x(t1 ) = x1 and x(t2 ) = ˙ x2 leads to t2 ′ (6.8) 44 . x) = K − V = ˙ 1 mx2 − V (x). x (t) = x(t) + δx(t).1) and as an example the lagrangian L(x. t1 (6. These equations can be obtained from a lagrangian using the action principle. Dirac ﬁelds (Dirac equation) and massless vector ﬁelds (Maxwell equations) and corresponding to these ﬁelds conserved currents describing the probability and probability current.7) The quantity ∂L/∂ x plays a special role and is known as the canonical momentum. and the total change x′ (t′ ) = x(t) + ∆x(t). ∂x ˙ (6. (6.Chapter 6 Classical lagrangian ﬁeld theory 6. ˙ 2 (6. ˙ p= ∂L . ∂x (6. recall classical mechanics with the action t2 S= t1 dt L(x.2) The principle of minimal action looks for a stationary action under variations in the coordinates and time.4) δS = t1 t2 dt dt t1 ∂L ∂L δx + δ x + L δτ ˙ ∂x ∂x ˙ d ∂L − ∂x dt ∂L ∂x ˙ δx + t2 t1 = ∂L δx + L δτ ∂x ˙ t2 .1 Euler-Lagrange equations In the previous chapter we have seen the equations for scalar ﬁelds (Klein-Gordon equation). As an example.

11) λ=0 of which p and H are the simplest examples related to space and time translations. − d3 x . (R3 .18) 1 Taking a functional derivative. . Variations in the action can come from the coordinates or the ﬁelds. H is a conserved quantity. (6. . The hamiltonian H is deﬁned by H(p. . ˙ x The second term in Eq. ∂µ φ.2. Consider a lagrangian density L which depends on these functions. ∆x = (dx/dλ)λ=0 δλ and δt = (dt/dλ)λ=0 δλ. + ∂L ∆x − H δτ ∂x ˙ t2 .e. indicated as x′µ φ (x) or combined φ′ (x′ ) = φ(x) + ∆φ(x).10) One sees that invariance under time translations requires that H(t1 ) = H(t2 ).17) Furthermore the variations δφ and δ∂µ φ contribute to δS. λ=0 (6. their derivatives and possibly on the position.t1 ) (6. δ(∂µ φ) (6. i. ∂µ φ(x). t2 ).t2 ) d3 x . R (6.12) Here R indicates a space-time volume bounded by (R3 . with ∆φ(x) = δφ(x) + (∂µ φ)δx . x). = φ(x) + δφ(x) (6. but which in 3 dimensions includes conserved quantities related to rotations and boosts. ∂µ φ′ . . ∂µ φ(x)) = R d4 x L (φ(x). . ∂R Note for the speciﬁc choice of the surface for constant times t1 and t2 . The resulting variation of the action is δS = R µ ′ = xµ + δxµ . dσµ . gives rise to a conserved quantity (at least if H is invariant!) Q=p dx dλ −H dt dλ . . In classical ﬁeld theory one proceeds in complete analogy but using functions depending on space and time (classical ﬁelds.16) The change in variables x → x′ in the integration volume involves a surface variation of the form dσµ L δxµ . t1 ) and R3 .6 can be rewritten as δS = .14) (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 . ∂µ φ(x)).9) which is done because the ﬁrst term (multiplying ∆x. ˙ (6. x) and an action t2 S[φ] = t1 dt L = dt d3 x L (φ(x). In general any continuous transformation. . L (φ(x).Classical lagrangian ﬁeld theory 45 For the lagrangian speciﬁed above this leads directly to Newton’s law p = m¨ = −∂V /∂x. . .15) d4 x′ L (φ′ . x′ ) − d4 x L (φ. We will come back to it in a bit more formal way in section 9. t1 (6. also indicated by ∂R (a more general volume in four-dimensional space-time with some boundary ∂R can also be considered). x) ≡ p x − L . indicated with δF [φ]/δφ should pose no problems. which in classical mechanics vanishes at the boundary) does not play a role. . = ∂R (R3 . think for instance of a temperature or density distribution or of an electromagnetic ﬁeld). 6.

1/2 and 1 ﬁelds By an appropriate choice of lagrangian density the equations of motion discussed in previous chapters for the scalar ﬁeld (spin 0). leading to (2 + M 2 )φ(x) = 0.22) which can be considered as the sum of the lagrangian densities for two real scalar ﬁelds φ1 and φ2 √ with φ = (φ1 + iφ2 )/ 2.Classical lagrangian ﬁeld theory 46 With for the situation of classical ﬁelds all variations of the ﬁelds and coordinates at the surface vanishing. The scalar ﬁeld It is straightforward to derive the equations of motion for a real scalar ﬁeld φ from the lagrangian densities. For the complex scalar ﬁeld one conventionally uses L = = ∂µ φ∗ ∂ µ φ − M 2 φ∗ φ −φ∗ ∂µ ∂ µ + M 2 φ. One easily obtains (2 + M 2 )φ(x) = 0.28) where the second line is not symmetric but in the action only diﬀers from the symmetric version by a surface term (partial integration).2 Lagrangians for spin 0. δL δ(∂µ ψ) δL δψ one obtains immediately i / −M ∂ → i = − γ µψ 2 i → = ∂ / ψ − M ψ.19) δ(∂µ φ) δφ 6.23) (6. δL δL ∂µ = .25) (6. ∂ (6. 2 ψ = 0. The integrand of the ﬁrst term must vanish.29) . (6. leading to the Euler-Lagrange equations. 2 (6. (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 ∗ (6.20) (6.24) (6. the Dirac ﬁeld (spin 1/2) and the vector ﬁeld (spin 1) can be found. (2 + M )φ (x) = 0.27) (6. the second term is irrelevant.21) which diﬀer only by surface terms. L = 1 1 ∂µ φ∂ µ φ − M 2 φ2 2 2 1 µ = − φ ∂µ ∂ + M 2 φ. Using the variations in ψ (in the symmetric form).

7).and lefthanded ones. We note that of which the left part coincides .30) It is often useful to link to the two-component spinors ξ and η which we started with in chapter 4. ∂ ∂ 2 2 (6.33) We now introduce a ’real’ (four-component) spinor satisfying Υc = Υ with ψL . Peshkin and Schroeder or Exercise 12. 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 .34) (6. namely2 LM (Majorana) = + (6. which comes because we in essence use ’real’ spinors. The Majorana case is in fact more general. (6.35) Since the kinetic term in L separates naturally in left and right parts (or ξ and η).g. or equivalently separate the ﬁeld into right.31 is given by LM (Dirac) = −M ξ † η + η † ξ . that the lagrangian separates into two independent parts for M = 0. Υi / Υ − ∂ 4 2 T −ǫ η ∗ c = ΥR .31) showing e.36) (6.38) (6. 2 αβ = −βα. Using the twospinors ξ and η. 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. In that case one ﬁnds trivially L = ↔ ↔ 1 1 ψR i / ψR + ψL i / ψL − M (ψR ψL + ψL ψR ).Classical lagrangian ﬁeld theory 47 and similarly from the variation with respect to ψ ψ i / +M ∂ ← = 0.39) 2 This term is written down with η being an anticommuting Grassmann number for which (αβ)∗ = β ∗ α∗ = −α∗ β ∗ . L = = ↔ 1 1 c c M ψL ψL + M ∗ ψL ψL ψL i / ψL − ∂ 2 2 ↔ 1 1 M ΥR ΥL + M ∗ ΥL ΥR . (6. but note the factor 1/2 as compared to the Dirac lagrangian.g. and thus (β ∗ α)∗ = α∗ β = −βα∗ . the mass term in the Dirac lagrangian 6. (6. 2 (6. 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 reasons for Grassmann variables will become clear in the next chapter. . ψL ≡ (ψL )c = C ψL = 0 (6. ∗ 0 ⇒ Υ ≡ −ǫ η ΥL = ψL = η η ∗ for which η0 = ǫ η0 . 1 M η † ǫη ∗ − M ∗ η T ǫη .37) This gives an expression with a mass term that is actually of the same form as the Dirac mass term.32) There exists another possibility to write down a mass term with only one kind of ﬁelds.

(6. 6. ∂µ J µ (x) = 0.40) 6. Q(t1 ) = Q(t2 ). . . ∆φ = (dφ/dλ)λ=0 δλ and δxµ = (dxµ /dλ)λ=0 δλ.Classical lagrangian ﬁeld theory 48 where Fµν = ∂µ Vν − ∂ν Vµ . δ(∂µ φ) (6. Q= d3 x J 0 (x.44) δL ∆φ + Θµν (x)δxν . δ(∂µ φ) (6. consider U (1) transformations of the Dirac ﬁeld proportional to the charge e.σ1 with σ = (R3 .. t). one immediately obtains δL = −∂ µ V ν + ∂ ν V µ − λ g µν (∂ρ V ρ ) = −F µν − λ g µν (∂ρ V ρ ) δ(∂µ Vν ) leading to the equations of motion ∂µ F µν + M 2 V ν + λ ∂ ν (∂ · V ) = 2 + M 2 V ν − (1 − λ)∂ ν (∂ · V ) = 0. t).+ R 4 δL δL ∆φ − ∂ ν φ − L g µν δxν δ(∂µ φ) δ(∂µ φ) δL ∆φ − Θµν (x)δxν . Returning to δS the surface term is rewritten to δS = R d4 x . one can consider the variations at the initial or ﬁnal surface. (6.3 Symmetries and conserved (Noether) currents Next.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 . requires the presence of a conserved quantity. the presence of a symmetry that leaves the lagrangian invariant. we get (omitting the parameter δλ) dxν dφ δL + Θµν (x) .41) (6. ψ(x) → ei eΛ ψ(x) or ∆ψ(x) = i eΛ ψ(x). discussed in the next chapter. These do not aﬀect the dynamics and will give rise to conserved currents. (6.42) Θµν = The variation δS thus can be expressed as δS with J µ (x) = − = δL ∂ ν φ − L g µν . implying ∂µ F µν + M 2 V ν = 0 and ∂µ V µ = 0. δ(∂µ φ) For continuous transformations. As an example for the classical ﬁeld. which in a consistent picture precisely generate the symmetries. we consider the second term in Eq. In this case. (6. these conserved quantities become operators.45) J µ (x) = − δ(∂µ φ) dλ λ=0 dλ λ=0 Thus considering ∂R = σ2 ..18.47) . In particular when the lagrangian is invariant under symmetries. which is the space-integral over the zero-component of a conserved current. (6.43) F (σ1 ) − F (σ2 ) = dσµ J µ (x) = ∂R R d4 x ∂µ J µ (x). + ∂R dσµ dσµ ∂R ≡ where d x .

˙ Q = d3 x j 0 (x. Therefore there are four conserved currents Θµν (ν labeling the currents!) and four conserved quantities Pν = dσµ Θµν = d3 x Θ0ν . (6. t) is the conserved charge (Q = 0).52) The behavior of the ﬁeld under translations (δφ) is governed by the translational behavior of the argument in such a way that ∆φ = 0.56) δxµ = ωρσ (aρσ )µν xν 2 with (aρσ )µν = aρσ µν = g ρµ g σν − g σµ g ρν .55) Note that the coordinate transformations can be written in a form similar to that for the ﬁelds (matrices in internal space). (6. 6. The integral over the zero-component.4 Space-time symmetries One kind of symmetries that leave the lagrangian invariant are the Poincar´ transformations.g.Classical lagrangian ﬁeld theory From the lagrangian for the Dirac ﬁeld.53) These are the energy and momentum. 1 (6.3) j µ = i e φ∗ ∂µ φ. . Lorentz transformations In this case the transformation of the coordinates and ﬁelds are written as δxµ ∆φi (x) = = ω µν xν . = −(∂µ φ)δxµ = −ǫµ ∂µ φ(x). Summarizing. From Noether’s theorem one sees that the current Θµν ǫν is conserved.51) (6. In the next section we will see the quantity show up as the charge operator.49) These currents are conserved as discussed already in chapter 2. e. H = d3 x Θ00 (x). (ω µν antisymmetric) 1 ωρσ (−i S ρσ )ij φj (x). = 0. ↔ (6. we obtain (since δxµ = 0). 2 (6.54) (6.48) For the complex scalar ﬁeld the U (1) transformations leave the lagrangian invariant and lead to the conserved current (see Exercise 6. including e the space-time translations and the Lorentz transformations.50) (6. (6. omitting the parameter Λ jµ = − δL δL (i e ψi ) (−i e ψi ) − δ(∂µ ψi ) δ(∂µ ψ i ) 49 = e ψγ µ ψ. For quantized ﬁelds this will become the expressions of the hamiltonian and the momentum operators in terms of the ﬁelds. (6. Translations µ Under translations (generated by Pop ) we have δxµ ∆φ(x) δφ(x) = ǫµ .57) For the ﬁelds the behavior under Lorentz transformations has been the subject of previous chapters.

(6. • Dirac ﬁeld ψ: −i Sρσ = −(1/4)[γρ .66) In section 6.70) .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 . φ(x). ∂µ φ(x) + i e ∂µ Λ(x) e (6.65) M µρσ = T µρ xσ − T µσ xρ . ∂µ T µν (6. independent of x. however. where H ρµν is the quantity appearing in M ρµν . in which the U (1) transformation involves an angle e Λ.61) A ﬁnal note concernes the symmetry properties of Θµν .68) (6.g. (6. In general this tensor is not symmetric. • Vector ﬁeld Aµ : −i Sρσ = aρσ .62) with Gρµν = (H ρµν + H µνρ + H νµρ )/2. (6.58) (6. is not invariant under local gauge transformations. the angle of the transformation depends on the space-time point x. The lagrangians which we have considered sofar are invariant under global gauge transformations and corresponding to this there exist a conserved Noether current.67) i. −i eΛ(x) ∗ i eΛ(x) (6. one has a tensor that satisﬁes (exercise 6. The current deduced from Noether’s theorem is Jµ = = = = − δL ∆φi + Θµν δxν δ(∂µ φi ) δL 1 ωρσ − (−i S ρσ )ij φj + Θµρ xσ − Θµσ xρ 2 δ(∂µ φi ) 1 ωρσ {H µρσ + Θµρ xσ − Θµσ xρ } 2 1 ωρσ M µρσ .3 we have seen global transformations or gauge transformations of the ﬁrst kind.3). 2 (6. (6. φ(x) φ (x) ∂µ φ(x) ∗ → e → e → ei eΛ(x) φ(x). Deﬁning T µν = Θµν − ∂ρ Gρµν . e. In some applications (speciﬁcally coupling to gravity) it is advantageous to have an equivalent current that is symmetric.e.69) i eΛ(x) φ (x). Any lagrangian containing derivatives.64) (6.6) T µν = T νµ .63) µν = ∂µ Θ . 6.Classical lagrangian ﬁeld theory 50 • Scalar ﬁeld φ: −i Sρσ = 0. γσ ] This result for the Dirac ﬁeld has been discussed in chapter 3 (see also Exercise 4.5 (Abelian) gauge theories φ(x) → ei eΛ φ(x).59) (6. Gauge transformations of the second kind or local gauge transformations are transformations of the type φ(x) → ei eΛ(x) φ(x). (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 .78) .72) (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). As an example consider the Dirac lagrangian. The equation of motion for the fermion follow from δL δ(∂µ ψ) δL δψ = = i − γ µ ψ. (6. (i/ − M ) ψ = (i/ − e/ − M ) ψ = 0. The required transformation for Dµ then demands Dµ φ(x) = = (∂µ + i eAµ (x))φ(x) ei eΛ ∂µ + i e(A′ + ∂µ Λ) φ µ → ei eΛ ∂µ φ + i e (∂µ Λ) ei eΛ φ + i eA′ ei eΛ φ µ ≡ ei eΛ (∂µ + i eAµ ) φ.Classical lagrangian ﬁeld theory 51 where the last term spoils gauge invariance. Dµ φ) − Fµν F µν . −e j µ Aµ = −e ρφ + e j · A. 1 L (φ.75) The ﬁeld φ is used here in a general sense standing for any possible ﬁeld.73) (6. (6.71) Thus the covariant derivative has the correct transformation behavior provided Aµ → Aµ − ∂µ Λ. 4 (6. (6. 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. Dµ φ(x) ≡ (∂µ + i eAµ (x))φ(x). ∂ 2 4 (6. 2 i ∂ / ψ − M ψ − e/ ψ A 2 giving the Dirac equation in an electromagnetic ﬁeld. 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 . D ∂ A (6.76) 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 . For this purpose it is necessary to introduce a vector ﬁeld Aµ .77) involving the interaction of the charge (ρ = j 0 ) and three-current density (j) with the electric potential (φ = A0 ) and the vector potential (A). ∂µ φ) =⇒ L (φ. i L = ψγ µ ∂µ ψ − (∂µ ψ)γ µ ψ − M ψψ.

i. φ → eiΛ φ.Classical lagrangian ﬁeld theory 52 For the photon the equations of motion follow from δL δ(∂µ Aν ) δL δAν = = −F µν .e. ∂µ F µν = j ν .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. ∂µ F µν = j ν . [iγ µ (∂µ − i e Aµ ) − M ]ψ = 0. giving the Maxwell equation coupling to the electromagnetic current.3 Show that the current jµ = i φ∗ ∂µ φ for a complex scalar ﬁeld is connected to a U (1) transformation on the ﬁelds.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. Exercise 6. 2 2 (b) Show that the Klein-Gordon equation for the complex scalar ﬁeld (considering φ and φ∗ as independent ﬁelds can be derived from the lagrangian density L = ∂µ φ∗ ∂ µ φ − M 2 φ∗ φ.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 . ∂µ j µ = 0 (Note that the addition of a total derivative to the lagrangian density does not modify the equations of motion). . 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. give the equation which is satisﬁed by ψ c = Cψ . and deduce what is the charge of the antiparticle as compared to a particle. −eψγ ν ψ.79) Exercises Exercise 6. Exercise 6. (6.

in order to show that Θρσ − Θσρ = ∂µ H µρσ . A) (see also exercise 6. e 1 (−i∇ + eA)2 + σ · B − eA0 ψu (r. Comment: this relation is known as the Pauli equation.6 (optional) Prove the properties of the tensor T µν in section 6. 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) = Enr ψu (r.Classical lagrangian ﬁeld theory 53 Exercise 6. Hints: it may be useful to realize that Gµρσ is antisymmetric in ﬁrst two indices. . Hint ≡ −gs µe · B = −gs Qe s · B.4) for a stationary solution in the non-relativistic (|Enr | ≪ M ) and weak-ﬁeld (|A0 | ≪ M ) limits yields in the standard representation for the ’upper (two) components’ ψu . This relation is known as the Pauli equation. ﬁnally note that it is suﬃcient to show the last equation for dσµ M µρσ . t). 2me Exercise 6. 2M 2M with Enr ≡ E − M ≪ M . use ∂µ T µν = 0 and ∂µ M µρσ = 0.4.5 Show that the Dirac equation for an electron with charge −e in an external electromagnetic ﬁeld Aµ = (A0 .

however.5) ∆ x ∆3 x 54 .3) For variations δO(p. dq The basic Poisson bracket is actually the one between q and its canonical momentum p. t).1) The Hamiltonian (also corresponding to a conserved quantity because of time translation invariance) is given by p2 + V (q). 1 qx (t) = 3 d3 x φ(x. Further. (7.1 Canonical quantization We will ﬁrst recall the example of classical mechanics for one coordinate q(t).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. dq dp dq dp (7. Quantization of the theory is achieved by imposing canonical commutation relations between q and p. in our example p = mq. This is nothing else as representing the classical Poisson algebra as operators in the Hilbert space. q) ≈ (dO/dλ)λ=0 δλ under symmetry transformations one has dO = [O. 6.2) H(p. An immediate generalization for ﬁelds can be obtained by considering them as coordinates. (7. ˙ L(q. q) = ˙ 1 mq 2 − V (q). Q]P . labeled by the position. p] = i.11) found through Noether’s theorem. starting from the lagrangian L(q.Chapter 7 Quantization of ﬁelds 7. B]P = dB dA dA dB − . q(q. H]P dt and dO = [O. namely [q. q) also considered in the previous chapter. Actually the conserved quantities satisfy a similar algebra as the Lie algebras for continuous symmetries that they are obtained from. (7. is the Poisson bracket [A. p)) = ˙ ˙ 2m where the (canonical) momentum p = ∂L/∂ q. [q. ˙ 2 (7. there are conserved ˙ ˙ quantities Q (Eq. dλ Examples are dO = [O. p]P . q) = p q − L(q. The bilinear product for the conserved quantities. p]P = 1.

9) Note that this indeed corresponds to the zero-zero component (Θ00 ) of the conserved energy-momentum stress tensor Θµν .Quantization of ﬁelds 55 etc. 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. discussed in the section 6. py (t)] = i δxy or for the ﬁelds φ(x) and Π(x) the socalled equal time commutation relations [φ(x. Sometimes it is useful to keep in mind that. formally. Π(y. Φ(f ) = d4 x φ(x) f (x). ∆φ.8) (7.11) (7. Π(y. qx . t). for the scalar ﬁeld theory.12) (7. The essential conditions to consistently quantize a theory then are the following. φ. t)] = 0.16) . 2 2 2 (7. t). px (t) = ∂L ∂Lx = ∆3 x = ∆3 x Πx (t). t)] = [Π(x. As an example. ˙ δ φ(x) (7. qx+∆x ) .2. ∂µ φ(x)) = ˙ d3 x L (φ. • Poincar´ invariance e The full variation of the ﬁelds. t)] = i δ 3 (x − y). a) = Rj (Λ−1 ) φj (Λ−1 x − a).15) with furthermore the relations [φ(x. ∇φ) (7.13) Π(x) H (x) For the quantization procedure we can formulate a number of basic axioms of quantum ﬁeld theory. The hamiltonian then is H= x px qx − L = ˙ = x where H (x) = δL ˙ φ(x) − L (x) = Θ00 (x). the discretization procedure above is an explicit example. (7. This implies speciﬁc transformation properties in the Hilbert space for the quantum ﬁelds φ. i U † (Λ. δ∂0 φ) 1 1 1 = Θ00 (x) = (∂0 φ)2 + (∇φ)2 + M 2 φ2 . (7. the ﬁelds φ(x) can be considered as regular operators in the Hilbert space after smearing with a test function f . t). φ(y. ∂ qx ˙ ∂ qx ˙ (7. for symmetry transformations. The lagrangian is given by L = = x In order to construct the hamiltonian it is necessary to ﬁnd the canonical momenta.6) ∆3 x L x (qx . albeit with ’sharp’ functions.10) x ∆3 x [Πx qx − L x ] ˙ ∆3 x Hx (7. ˙ where Πx (t) is obtained from the continuous ﬁeld Π(x) ≡ δL /δ(∂0 φ).14) In fact. • Canonical commutation relations The quantization of the theory is achieved by imposing the canonical quantization condition [qx (t).7) d3 x L (φ(x). (7. a) φi (x)U (Λ.

a† ] = a† . 2 2 (7. (7. (7. [φ(x). sometimes referred to as second quantization. and [N.23) The hamiltonian in this case can be expressed in the number operator N = a† a. [Q.24) It is straightforward to ﬁnd the commutation relations between N and a and 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. Q] = [P.20) where the coordinate Q and the momentum P satisfy the canonical commutation relations [Q. a] = −a.Quantization of ﬁelds 56 or if U (Λ. a† ] = 0. 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. j 2 one has [φi (x). 1 Q = √ (a + a† ) 2ω and P = −i ω (a − a† ) 2 (7. H = = ω ω P P Q−i √ Q+i√ 2 2 2ω 2ω 1 1 =ω N+ .19) 7. j (7. P ] = i. ω a† a + 2 2 ω i − [Q. a† ] = 1.18) with S µν given in 6. 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. P ] 2 (7. P ] = 0 (7. Pµ ] = [φ (x). a] = [a† . Φ(g)] = 0. • Causality Operators Φ(f ) and Φ(g) for which the test functions are space-like separated can be measured simultaneously (macroscopic causality). Microscopic causality implies local commutativity. Mµν ] = i i∂µ φi (x). 2 i ωµν (S µν )i .22) it is straightforward to check that the commutation relations between Q and P are equivalent with the commutation relations [a.4.17) (Sµν )i φj (x). a) = i Rj (Λ−1 ) = 1 + iǫµ P µ − 1− i ωµν M µν . The measurements cannot inﬂuence each other or [Φ(f ). [N. [a. φ(y)] = 0 if (x − y)2 < 0. (7.21) Writing Q and P in terms of creation (a† ) and annihilation (a) operators.25) .

a† (k ′ )] = 0.like’ commutation relations between a(k) and a† (k ′ ). (n − 1) a|n † i.29) It is straightforward to prove that the equal time commutation relations between φ(x) and Π(x′ ) are equivalent with ’harmonic oscillator . 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. but one needs to realize that in that case k 0 = Ek = k2 + M 2 .g. the real scalar ﬁeld φ(x) becomes d3 k (7. [φ(x).28) (7. 7.e. d3 x φ(x) i ∂0 e−i k ·x ↔ ′ ↔ (7. N |n = n|n one ﬁnds N a† |n N a|n = = (n + 1) a† |n . and we see that a state |0 must exist for which N |0 = a|0 = 0. From the normalizations one obtains a |n = √ n + 1 |n + 1 and a|n = n |n − 1 . φ(x) = (2π)3 2Ek where the Fourier coeﬃcients a(k) and a† (k) are now operators.4 for the classical ﬁeld) a(k) a† (k ′ ) = = d3 x ei k·x i ∂0 φ(x). The canonical momentum becomes −i d3 k ˙ Π(x) = φ(x) = a(k) e−i k·x − a† (k) ei k·x . i. is equivalent with [a(k).e.26) a(k) e−i k·x + a† (k) ei k·x .3 The real scalar ﬁeld We have expanded the (classical) ﬁeld in plane wave solutions.27) 2 (2π)3 It is easy to check that these equations can be inverted (see Exercise 2.31) The hamiltonian can be rewritten in terms of a number operator N (k) = N (k) = a† (k)a(k).33) . e. (2π)3 2Ek (7. which represents the ’number of particles’ with momentum k. a† (k ′ )] = (2π)3 2Ek δ 3 (k − k′ ) [a(k). which we have split into positive and negative energy pieces with (complex) coeﬃcients a(k) and a∗ (k) multiplying them. (7. Π(x′ )]x0 =x′0 = i δ 3 (x − x′ ) ′ ′ and (7.Quantization of ﬁelds 57 Deﬁning states |n as eigenstates of N with eigenvalue n. 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 . and (7.30) [φ(x). † √ a and a act as raising and lowering operators. Π(x )]x0 =x′0 = 0. Note that we will often write a(k) or a† (k).32) (7. The quantization of the ﬁeld is achieved by quantizing the coeﬃcients in the Fourier expansion. a(k ′ )] = [a† (k). φ(x )]x0 =x′0 = [Π(x).

= normalized as (a† (k1 ))n1 (a† (k2 ))n2 √ √ . .36) and satisfying the completeness condition 1 = k|k ′ = (2π)3 2Ek δ 3 (k − k′ ). which amounts to redeﬁning H as H= d3 x : H (x) := d3 k Ek N (k). . 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. (2π)3 (7.34) where V = (2π)3 δ 3 (0) is the space-volume.46) (7. assuring that the vacuum (by deﬁnition) has eigenvalue 0!. |k = a† (k)|0 .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. d3 k |k k|.e. d3 k a(k) e−i k·x . deﬁning particle states |k = |k (with positive energy. a(k)|0 = 0. (2π)3 2Ek (7.43) 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). and multiparticle states |(k1 )n1 (k2 )n2 . (2π)3 2Ek (7. . which now contains an inﬁnite number of oscillators. (2π)3 2Ek (7.45) (7. is solved by subtracting it as an (inﬁnite) constant. (2π)3 2Ek d3 k a† (k) ei k·x . . This term will be adressed below.42) (7.35) 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 . . writing all annihilation operators to the right of the creation operators. k 0 = Ek = k2 + M 2 ).38) d4 k 2π δ(k 2 − M 2 ) θ(k 0 ) |k k| = (2π)4 (7.37) (7. For the momentum operator one has Pi = d3 x Θ0i (x) = 1 2 d3 x ∂ {0 φ ∂ i} φ = d3 k k i N (k). (2π)3 2Ek (7. For the purpose of normal ordering it is convenient to decompose the ﬁeld in positive and negative frequency parts. The rest of the states are then obtained from the groundstate via the creation operator.40) This procedure is known as normal ordering.44) p|φ− (x)|0 = ei p·x .39) The problem with the vacuum or zero-point energy.41) (7. 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). n1 ! n2 ! (7. |0 . φ(x) φ+ (x) φ− (x) = = = φ+ (x) + φ− (x).

(7.52) The result for the invariant commutator function is [φ(x).50) or as integrals over d4 k. (2π)3 2Ek ≡ i∆− (x − y). (ii) ∆(x) is a solution of the homogeneous Klein-Gordon equation.54) = i∆(x − y) = i (∆+ (x − y) + ∆− (x − y)) . i∆+ (x) i∆− (x) = = = d4 k θ(k 0 ) 2π δ(k 2 − M 2 ) e−i k·x .Quantization of ﬁelds 59 In order to ensure the consistency of the theory it is necessary to check that the operators P ν and M obtained from the conserved currents Θρσ and M µρσ are indeed the generators of the Poincar´ e group. φ(y)] consider µν [φ+ (x).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. In order to calculate [φ(x).53) t=0 .e. a† (k ′ )] 3 2E ′ (2π) k [φ− (x). 2π (7.51) (7. i. (7. 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 ). d3 k ei k·(x−y) = −i∆+ (y − x).17) and (7.56) = −δ 3 (x). (2π)4 (7. φ+ (y)] d3 k e−i k·(x−y) . (iii) ∆(0. φ− (y)] ≡ i∆+ (x − y).49) (7.55) d4 k ǫ(k 0 ) 2π δ(k 2 − M 2 ) e−i k·x .47) ′ d3 k ′ e−i k·x+i k ·y [a(k). The last item to be checked for the scalar ﬁeld are the causality condition. that they satisfy the required commutation relations in Eqs (7. ∆(x) = − ǫ(x0 ) δ(x2 ). (iv) The equal time commutation relations follow from ∂ ∆(x) ∂t (v) For M = 0. d3 k = (2π)3 2Ek = (7.48) (7. x) = 0 and hence ∆(x) = 0 for x2 < 0. = − (2π)3 2Ek (7.

60) = i φ∗ ∂µ φ.62) (7.65) 7. = ∂{µ φ∗ ∂ν} φ − L gµν . (2π)3 2Ek (7. (2π)3 2Ek (7.66) .57) (7. (2π)3 2Ek d3 k b(k) e−i k·x + a† (k) ei k·x . 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) .58) and satisfy the equal time commutation relation (only nonzero ones) [φ(x).59) (7. (7. a† (k ′ )] = [b(k). (2π)3 2Ek (7. φ† (y)] = i∆(x − y).63) i.4 The complex scalar ﬁeld In spite of the similarity with the case of the real ﬁeld. 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 . which is equivalent to the relations (only nonzero ones) [a(k). particles (created by a† ) and antiparticles (created by b† ) with the same momentum contribute equally to the energy.e. extending it with the charge operator and the introduction of particle and antiparticle operators. we will consider it as a repetition of the quantization procedure.5 The Dirac ﬁeld L = ↔ i ψγ µ ∂µ ψ − M ψψ.61) (7. Also the charge operator requires normal ordering (in order to give the vacuum eigenvalue zero).Quantization of ﬁelds 60 7. ∂0 φ† (y)]x0 =y0 = i δ 3 (x − y). ↔ (7. b† (k ′ )] = (2π)3 2Ek δ 3 (k − k′ ). 2 From the lagrangian density (7.64) The commutator of φ and φ† is as for the real ﬁeld given by [φ(x). Q = = = i d3 x : φ† ∂0 φ − ∂0 φ† φ(x) : d3 k : a† (k)a(k) − b(k)b† (k) : (2π)3 2Ek d3 k a† (k)a(k) − b† (k)b(k) .

∂ 2 ↔ (7. s) e−i k·(x−y) j s † + vi (k. s)e−i k·x .68) The canonical momentum and the hamiltonian are given by Π(x) H (x) = = δL = i ψ † (x).73) (7. s)e−i k·x + d† (k.78) (7. d† (k ′ . (7. b† (k ′ .74) (7. then leads to additional minus signs and H Q = = d3 k Ek b† (k)b(k) + d† (k)d(k) (2π)3 2Ek d3 k b† (k)b(k) − d† (k)d(k) . The solution is the introduction of anticommutation relations. (7. u v (2π)3 2Ek (7.Quantization of ﬁelds 61 the conserved density and energy-momentum currents are easily obtained.79) Also for the ﬁeld and the canonical conjugate momentum anticommutation relations are considered.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) : . s)u(k. interchanging creation and annihilation operators. s)ei k·x + d(k.69) (7. (2π)3 2Ek d3 k b† (k. s)¯(k. i.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. . ˙ δ ψ(x) ↔ i Θ00 (x) = − ψγ i ∂i ψ + M ψψ 2 = iψγ 0 ∂0 ψ = iψ † ∂0 ψ.75) (7. (2π)3 2Ek (7.67) i ψ / ψ − M ψψ gµν . jµ Θµν = = ψγµ ψ. {b(k. s). † {ψi (x).77) Note that achieving normal ordering. s)ei k·x . s)¯(k.71) (7. (2π)3 2Ek (7. s′ )} = {d(k. s′ )} = (2π)3 2Ek δ 3 (k − k′ ) δss′ . s)v(k. s). s)vj (k. The quantized ﬁelds are written ψ(x) = s d3 k b(k. ψj (y)}x0 =y0 = d3 k (2π)3 2Ek ui (k. s) ei k·(x−y) x0 =y 0 .e. ↔ i ψγµ ∂ν ψ − 2 (7. s)u† (k.70) where the last line is obtained by using the Dirac equation.

˙ δ Ai (7. ˙ δ A0 δL = F i0 = E i .86) (7.Quantization of ﬁelds 62 Using the positive and negative energy projection operators discussed in section 4.83) From the lagrangian density the canonical momenta are = = (7.88) . 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). For the scalar combination {ψi (x).6 The electromagnetic ﬁeld 1 L = − Fµν F µν . as the vanishing of Π0 induces a constraint. we would have obtained [ψi (x). It is possible to continue in a covariant way with the lagrangian 1 λ L = − Fµν F µν − (∂µ Aµ )2 .87) If one wants to impose canonical commutation relations ∂µ Aµ = 0 cannot hold as an operator identity. ψ j (y)] = (i/x + M )ij i∆1 (x − y). (7. one has † {ψi (x). ψ j (y)} at arbitrary times one has {ψi (x).82) where i∆1 = i∆+ −i∆− .84) (7.81) where i∆ = i∆+ + i∆− is the same invariant commutator function as encountered before.80) 3 = δ 3 (x − y) δij .85) which reﬂects the gauge freedom. 7. ∂ (7. it implies the Lorentz constraint and leads to the canonical momenta Π0 Π i = = E i. (7. 4 Π0 Πi δL = 0. When we would have started with commutation relations for the ﬁeld ψ and the canonical momentum. but has the problem of being noncovariant. ∂ (7. 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). 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 ).

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. q) that dO/dλ = [O. µ λ=0 (7. rj ]P .93) which does exhibit problems with the time-like photon.11. where Q is the conserved quantity in Eq. pj ]P . The quantized ﬁeld is expanded as Aµ (x) = d3 k (2π)3 2Ek (λ) 3 ǫ(λ) (k) c(k. (c) Calculate the Poisson brackets for the coordinates x and the standard conserved quantities p and ℓ.91) (7. Aν (y)] = [Πµ (x).95) Exercises Exercise 7. The canonical equal time commutation relations are [Aµ (x). These problems are solved by the Lorentz constraint between physical states given above. [ℓi . 0)c(k. (7.89) with four independent vectors ǫµ . (7.e. what are [ri .90) where Πν = F ν0 − λ g ν0 (∂ρ Aρ ) and we have furthermore [Aµ (x). for which it is suﬃcient that ∂µ Aµ |A = 0. c† (k ′ . λ)c(k. 2. i. containing a time-like photon. λ) − c† (k. Πν (y)] = 0. λ)|A = 0. µ λ=0 (7. λ).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.1 (optional) If you haven’t seen Poisson brackets in classical mechanics you should (a) Proof that Poisson brackets (Eq. which reproduce the Hamilton equations. i. ′ (7. Aν (y)]x0 =y0 = i gµν δ 3 (x − y). and [ℓi . λ′ )] = −g λλ 2Ek (2π)3 δ 3 (k − k′ ). λ)ei k·x . Q]P . λ)e−i k·x + c† (k. ˙ ˙ . j = 1.94) Choosing k µ = (|k 3 |. (d) Use Poisson brackets to calculate q and p. Πν (y)]x0 =y0 = i gµν δ 3 (x − y). 0) . It gives 3 k µ ǫ(λ) (k)c(k. (b) Proof for a quantity O(p. In fact the commutation relations imply ˙ [Aµ (x). (7. pj ]P .Quantization of ﬁelds 63 for physical states |A and |B . and are equivalent with [c(k. 6. 7.3) satisfy the conditions of a Lie algebra. a longitudinal photon and two transverse photons. k 3 ) this reads |k 3 |c(0) (k) − k 3 c(3) (k) |A = 0 c(0) (k) ∓ c(3) (k) |A = 0. 0. 0.e. one time-like photon by itself is not allowed! This solves the problems with the normalization and the negative energies. where Aµ + + is the part of the vector ﬁeld containing the annihilation operators.

3 (a) Show that the correct implimentation of symmetries in the Hilbert space. . (b) Check the above commutation relation [φ(x). Pµ ] = i∂µ φ(x). (e) Show that ∆R (x) = 0 if x0 < 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. ∆hom (Λx) = ∆hom (x) if f is an invariant function. Is the subtraction of ’zero point’ contributions essential in this check? µ Pµ φ(x) ei a µ Pµ = φ(x − a) Exercise 7. we can write for ∆R an integral where the k 0 integration remains along the real axis.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 ) ∆hom (x) = 0.Quantization of ﬁelds 64 Exercise 7. (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. (Similarly one can show that for the advanced Green’s function ∆A (x) = 0 if x0 > 0). Give also the expression for ∆C . e−i a requires [φ(x). Pµ ] for the (real) scalar ﬁeld using the expressions for the ﬁelds and momentum operator in terms of creation and annihilation operators. It is invariant. (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. ∆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 ) ∆inhom (x) = −δ 4 (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 ). In that case one actually needs discontinuous functions f (k) such as the step function. (i) (Optional) Give similar expressions for the other Green’s functions under (d) in terms of the homogeneous solutions under (f). f (k) ∝ θ(k 0 ). (g) Show that the homogeneous solution ∆ in (f) satisﬁes ∆(0. (h) Show that the causal Green’s function ∆C (x) = θ(x0 ) ∆+ (x) − θ(−x0 ) ∆− (x).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).

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 . x (8. Both ψ p and ψ satisfy the Dirac equation. −r). time reversal (T) and charge conjugation (C). T. ˜ (8. 8. T and C for classical quantities is shown in the table 1.1 Parity xµ = (t.1: The behavior of classical quantities under P. The consequences of P. r) −→ xµ ≡ xµ = (t. writing −1 ψ(x) −→ Pop ψ(x)Pop = ηP Aψ(˜) ≡ ψ p (x). starting with the Dirac equation for ψ(x). (iγ µ ∂µ − M ) ψ(x) = 0. Table 8. parity (P).Chapter 8 Discrete symmetries In this chapter we discuss the discrete symmetries.1) The parity operator transforms We will consider the transformation properties for a fermion ﬁeld ψ(x).2) where ηP is the intrinsic parity of the ﬁeld and A is a 4 × 4 matrix acting in the spinor space. We can determine A.

. m). λ)γ0 u(k. −λ). −λ)u(k. (2π)3 2Ek (8. −λ) eik·x (2π)3 2Ek ˜ d3 k ˜ ˜ ˜ ˜ ηP b(k. one can also consider the scalar ﬁeld and vector ﬁelds. γ x (8.2 Charge conjugation We have already seen the particle-antiparticle symmetry with under what we will call charge conjugation the behavior T (8. x µ (iγ ∂µ − M ) γ0 ψ(˜) = 0. m) γ0 v(k. −λ)v(k.4) (8. the above operation reverses the sign of λ). λ) eik·˜ (2π)3 2Ek d3 k ˜ ˜ ˜ ˜ ηP b(k. λ) e−ik·˜ + d† (k. λ)γ0 v(k.6) (check this for the standard representation. −λ) eik·x 3 2E (2π) ˜ k d3 k ˜ ˜ ηP b(k.8) 3 = λ = λ = λ From this one sees immediately that −1 ˜ Pop b(k.11) (8. λ) eik·x . −λ) e−ik·x − d† (k. if helicity λ is used instead of the z-component of the spin m.e. −λ). iγ µ† ∂µ − M ψ(˜) = 0.13) ψ(x) −→ ψ c (x) = ηC C ψ (x). ˜ Pop d(k. λ)v(k. x The latter behavior of the vector ﬁeld will be discussed further below. λ)v(k. λ) Pop = ηP b(k. x It is straightforward to apply this to the explicit ﬁeld operator ψ(x) using γ0 u(k. −λ) e−ik·x − d† (k.7) = λ d k x x ηP b(k.3) (8.Discrete symmetries 67 After parity transforming x to x the Dirac equation becomes after some manipulations ˜ ˜ x iγ µ ∂µ − M ψ(˜) = 0. x Therefore γ0 ψ(˜) is again a solution of the Dirac equation and we have x ψ p (x) = γ0 ψ(˜).5) (8. In the same way as the Fermion ﬁeld.9) (8. λ)u(k. (i˜ µ ∂µ − M ) ψ(˜) = 0. λ)u(k. 8. λ) P −1 = −η ∗ d(k. m) = = ˜ u(k.12) and for the vector ﬁeld −1 Aµ (x) −→ Pop Aµ (x) Pop = −Aµ (˜). For the scalar ﬁeld we have seen −1 φ(x) −→ Pop φ(x) Pop = ηP φ(˜). choosing ηP is real (ηP = ±1) particle and antiparticle have opposite parity. m). λ) e−ik·x − d† (k. ˜ −v(k.10) i. op P (8. The result is ψ p (x) −1 = Pop ψ(x) Pop = ηP γ0 ψ(˜) x (8. x (8.

λ) eik·x .22) = Aψ|Aφ ∗ is antilinear if A(λ|φ + µ|ψ ) = λ∗ A|φ + µ∗ A|ψ .21) where A is a 4 × 4 matrix acting in the spinor space.18) (8. One has Aφ|Aψ = φ|ψ ∗ = ψ|A† Aφ = ψ|φ . λ). m). λ) e−ik·x + b† (k. . r). i. to ﬁnd ψ t (−˜) that is a solution of the Dirac equation. As time reversal will transform ’bra’ into ’ket’. 8. λ)C uT (k.16) d k ηC d(k. ((iγ µ )∗ ∂µ − M ) ψ(x) = 0.14) (8. λ)u(k. as discussed in section 4). m). we start with the complex conjugated x Dirac equation for ψ.17) 3 = λ This shows that −1 Cop b(k. ˜ x −(iγ µ )∗ ∂µ − M ψ(−˜) = 0. Norm conservation requires Top to be anti-unitary2 . For a quantized ﬁeld one has −1 ∗ −1 Top fk (x)bk Top = fk (x)Top bk Top . m) ¯ C v T (k.15) (where one must be aware of the choice of spinors made in the expansion. λ)C v T (k. λ) eik·x ¯ ¯ (2π)3 2Ek d3 k ηC d(k. γ x i(γ5 C)−1 γ µ γ5 C∂µ − M ψ(−˜) = 0.e.3 Time reversal xµ = (t. writing −1 ψ(x) −→ Top ψ(x)Top = ηT Aψ(−˜) ≡ ψ t (−˜). x 1A 2 An −iC −1 γ µ C∂µ − M ψ(−˜) = 0. Top |φ = φt | = (|φt )∗ . (8. (2π)3 2Ek (8. x (8.20) The time reversal operator transforms We will again consider the transformation properties for a fermion ﬁeld ψ(x). λ)v(k. λ) −1 Cop = ∗ ηC b(k. The action on the spinors (using C = iγ 2 γ 0 = −iρ1 σ 2 in standard representation) gives C uT (k. (8.Discrete symmetries 68 the latter being also a solution of the Dirac equation.19) Cop d(k. u(k. λ) e−ik·x + b† (k. antilinear operator is anti-unitary if A† = A−1 . m) ¯ = = v(k. x (8. The same relations hold for helicity states. Therefore ψ c (x) = = λ −1 ¯ Cop ψ(x) Cop = ηC C ψ T (x) (8. The (time-reversed) Dirac equation becomes. x µ (iγ ∂µ − M ) γ5 Cψ(−˜) = 0. x iγ µT ∂µ − M ψ(−˜) = 0. r) −→ −˜µ ≡ −xµ = (−t. λ). λ) Cop = ηC d(k. it is antilinear1 . (i˜ µ∗ ∂µ − M ) ψ(−˜) = 0. x x (8.

−p. there are 16 independent of these bilinear combinations.28) (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(−˜). a state |a. Therefore one has A|k = A|b† (k)|0 = A|T † T b† (k) T † T |0 ∗ = At |b† (k)|0 ∗ ˜ t = k|At .33) (8. −λ a T a. λ)v ∗ (k.30) (8.26) d k x x ηT b(k. λ) i γ5 C v(k. λ) eik·x + d† (k. λ a |a. λ). λ)iγ5 Cv(k. λ)u∗ (k.24) (8. λ)iγ5 Cu(k. λ) Top = ηT b(k. λ| a. −λ |¯. λ) eik·˜ + d† (k.27) 3 = λ = λ = λ From this one obtains −1 ˜ Top b(k. Since there are 16 independent 4 × 4 matrices. λ)u∗ (k. −p. λ). p. λ) eik·x + d† (k. 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)v ∗ (k.29) In table 2 the behavior of particle states under the various transformations has been summarized.25) (check this for the standard representation). −p. p. λ). λ| ¯ C |¯.34) .4 Bi-linear combinations In quantities such as currents and lagrangians often bilinear combinations of spinor ﬁelds are encountered. ∗ ˜ v (k. The result is = = λ (8.32) (8. ˜ Top d(k. −p. p. λ)u∗ (k.31) (8. p. ˜ = 0|b(k)|A 8. λ) T −1 = η ∗ d(k. λ a P |a. λ). λ)v ∗ (k. λ) e−ik·˜ (2π)3 2Ek d3 k ˜ ˜ ˜ ˜ ηT b(k. λ) e−ik·x 3 2E (2π) k ˜ d3 k ˜ ˜ ˜ ˜ ηT b(k.Discrete symmetries 69 Table 8. λ) e−ik·x (2π)3 2Ek (8. λ) eik·x + d† (k. λ) e−ik·x 3 2E (2π) k d3 k ˜ ˜ ηT b(k. λ |¯. op T (8. (8.2: The transformation properties of physical states for particles (a) and antiparticles (¯). Note that applying an anti-unitary transformation such as Top one must take for the matrix element ˜ the complex conjugate. x (8. λ) ψ t (x) −1 Top ψ(x) Top = i ηT γ5 Cψ(−˜) x = = ˜ u∗ (k.23) In the standard representation iγ5 C = σ2 and it is straightforward to apply this to the explicit ﬁeld operator ψ(x) using i γ5 C u(k.

¯ V µ (x) =: ψ(x)γ µ ψ(x) :. (8.3: The behavior of the independent bilinear combinations of fermi ﬁelds under P. as well as under the combined operation Θ = PCT (see Table 3). and T. γ5 γ µ or γ5 ). p′ |V µ (0)|p = u(p′ )γ µ u(p). (8. p)u(p) e−i (p−p )·x . In many applications the expectation values of currents are needed.g. F.39) where Γµ (p′ .36) (8. (q 2 ). γ µ . p′ .5 Form factors Currents play an important role in ﬁeld theory. F. Note that the lagrangian density L (x) → L (−x) under Θ. s′ |V µ (x)|p. For instance for nucleons one has Γµ (p′ . 8. C.. p′ |V µ (x)|p = u(p′ )Γµ (p′ . e.41) 2M 2M . for the vector current V µ (x). As an example consider the vector current for a point fermion.38) (8. p) = γµ F1 (q 2 ) + iσµν q ν F2 (q 2 ). ¯ ′ ′ (8.35) The x-dependence can be accounted for straightforwardly using translation invariance. The 16 combinations of Dirac matrices appearing above are linearly independent. of which the expectation value between momentum eigenstates can be simply found. (q 2 ). ¯ In general the expectation value between momentum states can be more complicated. This implies p′ |V µ (x)|p = e−i (p−p )·x p′ |V µ (0)|p . C. p) can be built from any combination of Dirac matrices (1. γ ν ]..Discrete symmetries 70 Table 8. momenta (pµ or p′µ ) or constant tensors (ǫµνρσ or gµν ).37) (8. Applying the results from the previous sections it is straightforward to determine the behavior of the combinations under P.40) where the coeﬃcients are functions of invariants constructed out of the momenta. 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 ). 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). in this case only q 2 = (p′ − p)2 . s . called form factors. 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 . 2M (8. σ µν .

• Parity p′ |V µ (x)|p = ei q·x u(p′ )Γµ (p′ . p) = (i γ5 C)Γ∗ (˜′ .g. or relations that follow from the equations of motion. C and T invariance. ¯ µ p ˜ Therefore time reversal invariance implies for the structure of Γµ (p′ . γ5 σµν = (1/2)ǫµνρσ γ ρ γ σ . p ˜ • Time reversal p′ |V µ (x)|p = ei q·x u(p′ )Γµ (p′ . p) that γ0 Γµ† (p. p) = γ0 Γµ (˜′ . p′ )u(p′ )) u = ∗ † u† (p)γ0 Γµ (p. p) that Γµ (p′ . p)γ0 u(p). or relations based on hermiticity of the operator or P . p). ¯ p ˜ Therefore parity invariance implies for the structure of Γµ (p′ .43) = ei q·x u(p′ )γ0 Γµ (˜′ . p)(i γ5 C). p)u(˜) ˜ x p ¯ p p ˜ p (8.44) = p′ |Vµ (−˜)|˜ ˜ x p ∗ ˜x = ei q ·˜ [¯(˜′ )]∗ Γ∗ (˜′ . p′ )u(p′ ) = u† (p′ )Γµ† (p. p)γ0 . ¯ 2M (8. e.g.45) Exercises Exercise 8. p)u(p) ¯ † † = p′ |Pop Pop V µ (x)Pop Pop |p ˜x = p′ |Vµ (˜)|˜ = ei q ·˜ u(˜′ )Γµ (˜′ . µ p ˜ (8. Therefore hermiticity implies for the structure of Γµ (p′ . p′ )γ0 u(p). p)u(p) ¯ † † = p′ |Top Top V µ (x)Top Top |p ∗ (8. p)u∗ (˜) up p µ p ˜ = ei q·x u(p′ ) (iγ5 C)Γ∗ (˜′ .Discrete symmetries 71 but that are eliminated because of relations between Dirac matrices. • Hermiticity p′ |V µ (0)|p = p′ |V µ† (0)|p = u(p′ )Γµ (p′ . p)(iγ5 C)u(p).1 The matrix element of the electromagnetic current between nucleon states is written as < p′ |Jµ (x)|p >= e−i(p−p )·x u(p′ )Γµ u(p) ¯ Here Γµ iσµν q ν F2 (q 2 ) 2M pµ + pµ ′ = γµ H1 (q 2 ) − H2 (q 2 ) 2M = γµ F1 (q 2 ) + ′ . p′ )γ0 = Γµ (p′ . the Gordon decomposition u(p′)γ µ u(p) = ¯ 1 u(p′ ) [(p′ + p)µ + iσ µν qν ] u(p). p)u(p) ¯ = p|V µ (0)|p′ ∗ = (¯(p)Γµ (p. e. p) that Γµ (p′ .42) where q = p′ − p.

(b) Show that in the interaction with the electromagnetic ﬁeld. |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 . 0. 0.2 (a) Calculate the current matrix element (using the explicit fermion spinors) in the Breit-frame in which q = (0. 0. (b) Show that such a term is also not allowed by time-reversal symmetry. p = (E. p′ = (E. −|q|/2). Exercise 8. 0. (c) Show that hermiticity of the current requires that F1 en F2 are real. GE GM = F1 − Q2 F2 . L int = ejµ Aµ the charge is given by e GE (0) and the magnetic moment by e GM (0)/2M . 0. Exercise 8.3 (a) Show that because of parity conservation no term of the form γ5 can appear in the current matrix element. 0. (c) Show that if a term of this form would exist.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 . it would correspond to an electric dipole moment d = −e F3 (0)/2M (Interaction term d σ · E). |q|). σµν q ν F3 (q 2 ) 2M . 4M 2 = F1 + F2 .

AH (t) = U −1 (t. i ∂ |ψH ∂t ∂ i AH ∂t ≡ = 0. t0 ) = H U (t. i ∂ |ψS ∂t ∂ i AS ∂t = H |ψS . (9.2) (ii) Heisenberg picture. t ≡ q|q. AS (t) = AS and the states o are time dependent.3) (9. QH (t)|q.1) The time evolution from t0 → t of a quantum mechanical system is generated by the Hamiltonian. t0 ). t0 )|ψS (t0 ) .7) 73 .5) (9. t0 ) AH (t0 ) U (t. |ψH (t) = |ψH . Consider the two (time-independent) Heisenberg states: |q. t0 ) ∂t (9. t′ .1 Time evolution as path integral U (t.Chapter 9 Path integrals and quantum mechanics 9. t ′ |q . and the operators are time-dependent. in which the states are time-independent. ≡ 0. H].4) ∂ U (t. t ′ QH (t′ )|q ′ . t0 ) = e−i(t−t0 )H or i Two situations can be distinguished: (i) Schr¨dinger picture. (9. [AH .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. t . |ψS (t) = U (t. t′ ≡ q ′ |q ′ . (9. (9. in which the operators are time-independent.

B]] + [[A. n |q (9.qj )∆τ e e 2π dpj i pj (qj+1 −qj )−i ∆τ H(pj . tj = = = 1 For dpj i pj (qj+1 −qj ) −i H(pj . t = = q ′ |e−i n H∆τ |q = q ′ | e−i H∆τ dq1 . t and QH (t) = QS and study the evolution of the system by calculating the quantum mechanical overlap amplitude q ′ . with the correction1 being of order (∆τ )2 . 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|. use the Campbell-Baker-Hausdorﬀ formula.10) dqn−1 q ′ |e−i H∆τ |qn−1 qn−1 | . 2 12 12 (9. B] + . Q) expressed in terms of the operators P and Q. Combining the two terms gives qj+1 . .Path integrals and quantum mechanics 74 and Schr¨dinger states o |q |q ′ QS |q ≡ q|q .8) Choose t as the starting point with |q = |q. t = q ′ |e−i H(t −t) |q . eA eB = eC with C = A+B+ (9. . qj )]) . QS |q ′ ≡ q ′ |q ′ . 2π (9. (9. 2π (9.13) At least for a simple hamiltonian such as consisting of a kinetic energy term and a local potential. . |q1 q1 |e−i H∆τ |q .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 . The purpose of this is to calculate the evolution for an inﬁnitesimal time interval. tj+1 |qj . [A. ′ (9. t′ |q. which is a hamiltonian function in which the operators are replaced by real-numbered variables.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 ′ .q .14) . t′ |q.qj ) e 2π dpj exp (i ∆τ [pj qj − H(pj . qj ). B]. B] + [A. Then qj+1 . 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 ... 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 . tj+1 |qj . H(P. The hamiltonian is an operator H = H(P.12) where the transformation between coordinate and momentum space involves q|p = ei p. .11) (9.16) this. (9. ˙ 2π 1 1 1 [A.

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

2 (9. (9.24) (9.28) Dα Dα∗ exp (−α∗ α − α∗ ω − ω ∗ α) = exp (ω ∗ ω) . A useful property of functional integration is the translation invariance. Using method 1 one writes 1 Dα exp − α2 2 = x dαx N exp − 1 2 ∆x α2 x x = x dαx 1 exp − ∆x α2 x N 2 1 N 2π ∆x ≡ 1. y) = ∗ m. Physical answers will usually come out as the ratio of two functional integrals and are thus independent of the chosen measure.n fm (x) Kmn fn (y) and consider F [α] = F (αn ) as a multivariable function. 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. Note that procedure (i) is an example of the more general procedure under (ii).n dx fm (x)fn (x) = n = n dαn 1 exp − α2 N 2 n √ 2π ≡ 1.22) = x The last equality is obtained by deﬁning the right measure (normalization N ) in the integration. .27) (9. Dα F [α] = Dα F [α + β]. 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. with Dα F [α] = n dαn N F (αn ) (9.25) (see Exercises).23) Having deﬁned the Gaussian integral. N (9. (9. the following integrals can be derived for a real symmetric or complex hermitean kernel K.21) again a multidimensional integral. α(x) = n αn fn and K(x.26) As an important application of translation invariance we mention the identities 1 Dα exp − α2 − ωα 2 = exp 1 2 ω . In both cases N is an appropriately choosen normalization constant in such a way that the integral is ﬁnite.Path integrals and quantum mechanics 76 (ii) Write α in terms of a sum of orthonormal basis functions. Consider the gaussian functional as an example.

n θn fn (x) dx θ∗ (x) θ(x) .g.38) . α(y) = δ(x − y). δα(x) 2 2 δ exp (−α β) = −β(x) exp (−α β) . Examples of functional diﬀerentiation are (9. (9.31) δ δ α = dy α(y) = dy δ(x − y) = 1.30) (9.Path integrals and quantum mechanics 77 Functional diﬀerentiation is deﬁned as δ . e. which vanish.32) (9. θη = −ηθ. It is now easy to check that Dθ∗ Dθ exp (−θ∗ K θ) = det K.35) For applications to fermion ﬁelds. we need to consider Grassmann-valued functions involving anticommuting Grassmann variables. −→ fn (x) δα(x) ∂αn ∂αm n Note that δ(x − y) = n ∗ fn (x) fn (y). y) β(y) δα(x) δ 1 1 exp − α2 = −α(x) exp − α2 . δα(x) or in discretized form δ 1 ∂ ∂ with .37) We note that from the ﬁrst to the second line one needs to realize that a pair of diﬀerent Grassmann variables behaves as ordinary complex variables. i. however. αy = δxy . (9. the Gaussian-type functionals.29) (9. −→ δα(x) ∆x ∂αx ∂αx δ ∂ ∂ ∗ with .33) (9. αn = δmn . δα(x) δα(x) δ αKβ = dy K(x. dx dy θ∗ (x) K(x. θ2 = 0.e. θ∗ ] = exp (−θ∗ K θ) ≡ exp − Integration for Grassmann variables is deﬁned as dθ 1 = 0 and dθ θ = 1. In principle the deﬁnitions of functionals is the same. In the expansion of the exponential (from second to third line). θ∗ ] = exp (−θ∗ θ) ≡ exp − F [θ. F [θ.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. δα(x) (9.34) (9. y) θ(y) . (9. products of the same pair appear.

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

n (9. q) + J(t) · q.66) .Path integrals and quantum mechanics source experiment detector 81 T L t L+Jq t’ L T’ Figure 9.62) p Dq D exp i 2π t dτ [pq − H(p. t′ |q. t′ q ′ . harmonic oscillator. t q. t = ei Ht |q . for the physical states o q. q) + J(τ )q] . We can again rewrite the time-ordered products as functional integrals as derived in the previous section. q)] ˙ (9. 9. T ′ |Q. which in turn are embedded between an early time T and a future time T ′ . t t′ V =0 Dq D dτ HI (q) exp i t t′ t dτ [pq − H0 (p. t|n n|Q. ˙ J (9. φn (q. the creation of an electron (think of a radio-tube. t|n = q|n e−iEn t with q|n the time-independent wave function. T = n J t′ = = Dq exp i t dτ [L(q.e. t |T exp −i p exp −i 2π t′ ′ ′ t′ t dτ HI (Q(τ )) |q. for example x|n x|n = φn (x) = ei kn x = e −ω 2 x2 /2 (9. T = n q. Before and after these processes there is only the vacuum or ground-state |0 . one has q ′ . T with q. Considering the source to be present between times t and t′ . it is possible to switch on an interaction. q)] . t Q′ . making the electron) and the absorption of an electron (think of a detector). i. T . T < t < t′ < T ′ .5 The generating functional for time ordered products By introducing a source-term. t = = ′ ′ V = q . q .63) for plane waves. t|Q. Consider. furthermore. t) φ∗ (Q) e+i En T . T ′ |q ′ . say.65) J dq ′ dq Q′ .e. t′ |q. t is related to the Schr¨dinger state |q by |q. i. The Heisenberg state |q. t |q. L(q. a set |n of physical eigenstates. t|Q.s. q) → L(q. ˙ The combination of exponentials in the last step is allowed because we are simply dealing with classical quantities (not operators!). ˙ ˙ physically pictured as.1: Physical picture of vacuum to vacuum amplitude in presence of a source term This is not a surprising result. for g.64) (9.

T → i ∞ and T ′ → −i ∞. t′ |q. T ′ |Q.71) 9.70) one immediately ﬁnds δ n Z[J] δJ(t1 ) . t) (9.68) (9. Better is the use ˜ of imaginary time t = −it. 9. ˜ t′ → ∞. t′ ) q ′ . J=0 (9. The generating functional precisely represents the vacuum to vacuum amplitude from initial (’in’) to ﬁnal(’out’) situation in the presence of a source.70) for the generating functional is in fact ill-deﬁned. such that t′ → −i∞ t → i∞ ↔ ↔ ˜ t → −∞. eiE0 T φ∗ (Q) 0 We deﬁne the generating functional Z[J] as Z[J] = Q′ . . . T′ Z[J] = lim T ′ → −i∞ T → i∞ Dq exp i T dτ L q. in which case eiEn T → e−En ·∞ and the term eiE0 T is dominant. The importance of Z[J] is that the time ordered product of operators can be obtained from it. Since we have (neglecting multiplicative factors). Q(tn )|0 . dq dτ + J(τ )q(τ ) .2: Analytic continuation of boundaries T and T ′ We can project out the groundstate by an analytic continuation in the time. . t 0 0out |0in J .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 ′ ). = (i)n 0|T Q(t1 ) .67) = = φ0 (q.Path integrals and quantum mechanics t 82 T t t’ T’ Figure 9. . . t|Q. T J lim −i E0 (T ′ −T ) φ (Q′ ) φ∗ (Q) 0 0 T ′ → −i∞ e T → i∞ dq ′ dq φ∗ (q ′ . thus T →i∞ lim q. (9. t). T = φ0 (q.6 Euclidean formulation The above expression (Eq. δJ(tn ) hence the name generating functional. J (9.

1 Convince yourself by choosing an appropriate (orthonormal) set of functions that 1 Dα exp − α K α 2 for real-valued functions. when L q.and Grassmann-valued functions. dq dt = 1 2 1 2 dq dt 2 Dq exp Dq exp − ∞ d˜ L q.76) which is a positive deﬁnite quantity. + V (q).74) − V (q) 2 (9. dq ˜ dt = 1 2 dq ˜ dt dq ˜ dt 2 − V (q). . ZE [J] = = where LE q. .77) where the expressions have been divided by Z[J] and ZE [J] in order to get rid of dependence on multiplicative factors.Path integrals and quantum mechanics 83 In terms of the imaginary time we can write the Euclidean generating functional. complex. δ n Z[J] 1 Z[J] δJ(t1 ) . i dq .72) (9.73) d˜ LE q.g. δJ(tn ) J = 0 ˜ t = it (9. In all of these cases the determinant of the (hermitean) operator K is deﬁned as det K ≡ p Kp . As discussed in the previous section the interesting quantities are obtained from functional diﬀerentiation with respect to the sources. ˜ ˜ ZE [J] δJ(t1 ) . . y) = ∗ m. ensuring convergence for the functional integral ZE [J]. i τ −∞ ∞ −∞ dq d˜ τ + J(˜)q(˜) τ τ dq d˜ τ − J(˜)q(˜) τ τ . (9. det K 1 . ˜ dt (9. 1 = √ . δJ(tn ) = (i)n J=0 1 δ n ZE [J] . e. det K Exercise 9.n fm (x) Kmn fn (y). . τ dq ˜ dt ≡ −L q. for Grassmann valued functions.75) = − LE q. . Exercises Exercise 9. (9. The diﬀerentiations in Z[J] and ZE [J] are related. Dα Dα∗ exp (−α∗ K α) = for complex-valued functions and Dθ∗ Dθ exp (−θ∗ K θ) = det K. where Kp are the eigenvalues of the matrix Kmn in K(x.2 Check the examples of functional derivation for real-.

z) = δ(x − z). det K Dθ∗ Dθ exp (−θ∗ K θ − θ∗ η − η ∗ θ) = det K exp η ∗ K −1 η . y) K −1 (y.79) (9.80) Dα∗ Dα exp (−α∗ K α − α∗ ω − ω ∗ α) = 1 exp ω ∗ K −1 ω .3 Prove the following relations 1 Dα exp − α K α − αω 2 1 = √ exp det K 1 ω K −1 ω . where dy K(x. 2 (9.78) (9.Path integrals and quantum mechanics 84 Exercise 9. .

∂µ φ) + Jφ] (10. δJ(xn ) 85 ≡ G(n) (x1 . Π) (10. φ(x. xn ). ∂µ φ) − Jφ] (10. φ(xn )|0 = (−i)n δ n Z[J] δJ(x1 ) .4) J=0 . t = = Dφ DΠ exp i t σ′ ˙ dτ d3 x Π(x)φ(x) − H (φ.e. .2) = ∞ −∞ d4 xE [LE (φ. 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 ) . . .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. x′ . .3) The Euclidean formulation is as before. . t′ |φ. t) is considered as a set of quantum operators φx (t) in the Heisenberg picture and t′ φ′ . x. (10.Chapter 10 Feynman diagrams for scattering amplitudes 10.1) 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. i. . . .

.e. .8) 1 d4 x φ (∂µ ∂ µ + M 2 ) φ − Jφ 2 K (10. . d4 xn G(n) (x1 .7) where we made the choice Z0 [0] = 1 and ∂µ ∂ µ + M 2 ∆F (x − y) = −δ 4 (x − y). .5) Z0 [J] for the free scalar ﬁeld For the (free) scalar ﬁeld lagrangian L = ← 1 1 ∂µ φ∂ µ φ − M 2 φ2 . . with Z0 [0] = 1.4) one distinguishes the retarded Green’s function ∆R (x) = d4 k e−i k·x . These Green functions appear in the expansion Z[J] = n in n! d4 x1 . J(xn ). . As discussed in the previous section (exercise 9. (10. .11) Depending on the path choosen in the complex k 0 plane (see exercise 7. 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 (10.10) J=0 In order to determine ∆F . consider the general solution of ∂µ ∂ µ + M 2 ∆(x) = −δ 4 (x).Feynman diagrams for scattering amplitudes 86 for which we introduce the picture x1 G(n) (x1 . . ∆F (the socalled Feynman propagator) is the Green’s function of the Klein-Gordon equation.9) i. . which can be written as ∆(x) = ˜ with (k 2 − M 2 )∆(k) = 1 or ˜ ∆(k) = 1 1 1 = 0 2 .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). 2 2 → (10.3) one ﬁnds 1 Z0 [J] = exp − i 2 d4 x d4 y J(x) ∆F (x − y) J(y) ≡ exp (i W0 [J]) .13) . = 0 2 2 2 k2 − M 2 (k ) − E 2 (k ) − k − M (10. . . Furthermore. (10. (10. .12) d4 k −i k·x ˜ e ∆(k) (2π)4 (10. xn ) = x2 xn . (2π)4 (k 0 + iǫ)2 − E 2 (10. xn ) J(x1 ) .

the second is to explicitly consider 0|T φ(x1 ) φ(x2 )|0 .4). in the (well-deﬁned) Euclidean formulation starting with LE = 2 (∂E φ)2 + 2 M 2 φ2 µ 2 2 Jφ with the equation of motion (∂E ) − M φ(x) = −J(x). (2π)4 k 2 − M 2 (10. t) = 0 for t < 0. ∆(x) = − d4 k e−i k·x . φ+ (0)] = (2π)3 2E are also shown in exercise (7.14) They are solutions of the inhomogeneous equation. The contours for i ∆(x) i ∆+ (x) i ∆− (x) = [φ(x). ∆F = ∆C .15) e−i k·x d4 k .1 for countours) µ (∂E )2 − M 2 (i ∆F (x − y)) = −δ 4 (x − y) 0 kE =∞ (10. t) = 0 for t > 0.18) Thus we see various solutions.Feynman diagrams for scattering amplitudes 87 satisfying ∆R (x. (2π)4 (k 0 − iǫ)2 − E 2 (10. i. 10. φ− (0)] = (2π)3 2E d3 k e−i k·x = [φ− (x). (2π)4 k 2 − M 2 (10. 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).17) d4 k e−i k·x .23) k0 =i∞ k0 =−i∞ d4 k e−i k·x . 4 k 2 − M 2 + iǫ (2π) (10. the generating functional can be written 1 µ µ Z0 [J] = Dφ exp d4 xE (10.19) φ ∂E ∂E − M 2 φ + Jφ 2 1 d4 xE d4 yE J(x) (−i ∆F (x − y)) J(y) . µ 1 1 • Firstly. The ﬁrst possibility is to consider the well-deﬁned Euclidean formulation. 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. 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 . . φ(0)] = i∆+ (x) + i∆− (x) d3 k e−i k·x = [φ+ (x).20) = exp − 2 where or (see ﬁg. (10. In order to see which is the appropriate Green’s function to be used in the generating functional Z0 [J] we will take two routes.21) ∆F (x) = = −i 0 kE =−∞ d4 kE e−i kE ·xE 2 (2π)4 kE + M 2 (10.e. the advanced Green’s function ∆A (x) = satisfying ∆A (x. (10. (2π)4 k 2 − M 2 where the latter contour can be deformed to the contour for ∆C .

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

35) z . (2) (0) (10. with propagators (i ∆F ) being connected to vertices (i LI ) according to the above expression for the functional Z[J]. Consider as an example the interaction LI (φ) = − g 4 φ 4! in the scalar ﬁeld theory discussed sofar.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. L (φ) = L0 (φ) + LI (φ).29) This expression will be the one from which Feynman rules will be derived.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) (10.34) Introducing a vertex point to which four propagators are connected. x2 ) = G0 (x1 . (10.g ) (x1 .Feynman diagrams for scattering amplitudes 89 10. (10. ≡ −i g d4 z (10. Z[J] = = Dφ exp i exp i d4 z LI (10. x2 .30) and in zeroth order G(0. x4 ) = G0 (x1 .g ) (x1 .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 . x3 .27) When interactions are present.e.g 0 ) = G0 = 1. x3 . x4 ) = To ﬁrst order in g one has Z (g ) [J] = = 1 0 0 x1 + + x2 (10. To zeroth order in the coupling one has 0 1 Z (g ) [J] = Z0 [J] = exp − i 2 J ∆F J . i. the generating functional can be written. (10.32) (10.28) 1 exp − i 2 = exp 1 2 d4 x d4 y exp i δ δ i ∆F (x − y) δφ(x) δφ(y) d4 z [LI (φ) + J(z)φ(z)] φ=0 . x2 . x2 ) = i ∆F (x1 − x2 ) = G(4.31) G(2.

. This remains also true for the interacting case. . . xn ) J(x1 ) .41) in terms of socalled connected Green’s functions. . (10. J(xn ) c (10.37) 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]. d4 xn G(n) (x1 . J(xn ). (10. did not contain parts that could be written as products of simpler Green’s functions.38) The result for the 4-points Green’s function is left as an exercise. x2 ) = 1 z x1 x2 + 1 2 x1 z x2 .39) We have calculated the function up to ﬁrst order in the coupling constant for φ4 -theroy. J=0 (10. . .36) J J from which one obtains G(0. . (10. Z[J] = n in n! d4 x1 . x2 ) = c (vacuum bubble) (tadpole) (connected propagator) . . xn ) J(x1 ) . .42) x1 G(n) (x1 . . .e. which also was the only Green’s function for which the diagram was connected. (10. . Diagrammatically this exponent only contained the two-point function. . . .g 1 ) = 1 8 z 1 8 . . d4 xn G(n) (x1 . xn ) = x2 c xn Speciﬁc examples are G(0) = c G(1) (x1 ) = c G(2) (x1 . i. . . In the free case we have seen that the exponent of Z[J] contained all essential information. .Feynman diagrams for scattering amplitudes 90 one has J Z (g1 ) [J] = 1 8 1 − 4 J J 1 + 24 J Z0 [J].40) in−1 n! d4 x1 . . (10. xn ) = (−i)n−1 c which are denoted δ n W [J] δJ(x1 ) . G(n) (x1 .g ) (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]. . δJ(xn ) . . . . .

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

in|S|p′ . Proof: α. in = p.47) .g ) (x1 . out = |p′ . x2 . p′ . φout and the interpolating ﬁeld φ(x). . in = β. considered explicitly. (3) S is unitary (it conserves the scalar product from initial to ﬁnal state). out|α. (2) The one-particle state is invariant (conservation of energy and momentum. in|p′ . x3 . (10. this will be translated to the action on ﬁelds. translation invariance). in = S|α. the vacuum expectation value of the ﬁeld φ(x) is zero. Also for free ﬁeld (Heisenberg) operators a distinction is made between φin and φout . in ⇔ β. For the free scalar theory. in = p.46) The properties of the S-matrix are (1) The vacuum is invariant or |S00 | = 1. p. . out = δαβ ↔ SS † = 1. out (suppressing except momenta all other 1 m quantum numbers).g. x4 ) = c 1 x3 z (10.43) implying as expected as the only nonzero connected Green’s function G(2) (x1 − x2 ) = (−i) c δ 2 W [J] δJ(x1 ) δJ(x2 ) = G0 (x1 − x2 ) = i ∆F (x1 − x2 ). (10. e. in|S = α. in|S = β. (10. then (n) (n) Gsc = Gc for n ≤ 3. in|SS † |α. out|p′ . thus β. out|p′ . implying the absence of tadpoles.45) x2 as the only surviving diagram (see exercises). pn . out| ⇔ |α. . Proof: 0. x4 10.3 Interactions and the S-matrix The S-matrix The S-matrix transforms initial state free particle states (in-states) |α. in = |p1 . . . in| (choose ϕ0 = 0). in = p. out| and S † |α.Feynman diagrams for scattering amplitudes 92 i. out = p|p′ . out . Sβα ≡ β. in|S = 0. ∂µ ∂ µ + M 2 φin (x) = 0. out . In line with the consideration of the generating functional representing the vacuum to vacuum amplitude we consider ﬁelds φin . in into ﬁnal state particle states (out-states) |β. . x1 G(4. we have W0 [J] = − 1 2 d4 x d4 y J(x) ∆F (x − y) J(y).e.44) J=0 For the interacting theory. out| = eiϕ0 0. (2) (10. in = |α. t = −∞ φin (x) t = +∞ φout (x) φ(x) where φin and φout transform under the Poincar´ group as scalar ﬁelds and satisfy the homogeneous e Klein-Gordon equation with the physical mass M . out|α. Next. one ﬁnds for the connected 4-point Green’s function at order g.

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

φ(x).60) e d4 y φ+ (y) f (y) d4 z φ− (z) f (z) d4 z φ+ (z) f (z) d4 y [φ(x). provided [A.62) where F [J] is some (arbitrary) functional. 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. C] are c-numbers. t i Ut∞ [J] φ(x)U−∞ [J] (10. which is equal to the expression e in which the creation operators are placed left of annihilation operators. z) d4 y d4 z ∆(x − y)K(y. 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. 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. δJ(z) (10. B]eB . : =i d4 y ∆(x − y) f (y) : e d4 z φ(z) f (z) :. eB ] = [A. δJ(z) (10. we can.59) (10. (10. : e d4 z φ(z) f (z) d4 x φ− (x) f (x) B C B C (10. e e ] = [A. one has [A.e. just as we did for φ(x). δJ(x) ∞ or for U [J] ≡ U−∞ [J] ′ (10. z) . z) δU [J] δJ(z) δU [J] d4 y d4 z ∆A (x − y) K(y. Thus applied to the ﬁeld φ(x) = φ+ (x) + φ− (x). B] (for the case that [A.56) Since δU [J]/δJ(x) satisﬁes the same equations as φ(x) and we know the limits.58) In order to ﬁnd a solution to this equation note that. S U [J]] = √ Z δ S U [J]. express it in terms of advanced and retarded Green’s functions.63) . B] and [A. φ(x).55) δU [J] δJ(x) = t→∞ t→−∞ −→ −→ √ i φ(x) U [J] → i Z φout (x) U [J] √ i U [J] φ(x) → i Z U [J] φin (x). y) δ δJ(y) : F [J]. Thus S U [J] =: exp 1 √ Z d4 x d4 y φin (x) K(x. z) δU [J] . B] is a c-number).61) e φ+ f where the normal ordered expression : e φ f : is used. with the use of the Baker-Campbell-Hausdorﬀ formula e−B A e+B = A + [A. B + C]e e .Feynman diagrams for scattering amplitudes 94 and satisﬁes the property T ′ δUT [J] t = i UtT [J] φ(x)UT [J]. [A. (10. φ− f (10. φ(y)] f (y) e e .57) i [φin (x).

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

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. or calculating full Green’s functions external lines are treated as propagators. Diagrammar) in the case of the interaction terms g f (10. but overall momentum conservation at a vertex is understood. the factor 2 corresponds to the two-points Green’s function having two identical ends). At these vertices each line can be assigned a momentum. If problems arise go back to the deﬁning expression for the generating function in 10. For the symmetry factor consider the examples (given in G. to be precise iLI vertices in momentum space are introduced. 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. vertices and external particle wave functions in that diagram. For the interaction terms in the lagrangian. which cancels the factor 1/2 in the quadratic piece. 3! 4! The vertices are: = −i f Consider ﬁrst the lowest order self-energy diagram. (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).27. 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. Veltman. = −i g (multiplied with 4! corresponding to the allowed number of permutations of identical particles). ’t Hooft and M. Note that in calculating amputated Green’s functions external lines are neglected.66 and 10.69) LI (φ) = − φ3 − φ4 . = −i g .27.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.

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

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

Vector ﬁelds and Quantum Electrodynamics As the most important example of vector ﬁelds consider the lagrangian density for quantum electrodynamics (QED). writing down the propagators till the beginning of the ¯ line (incoming fermion u or outgoing antifermion v) has been reached. 1 k 2 − M 2 + iǫ λ (k 2 − M 2 + iǫ)(k 2 − λ M 2 + iǫ) leads to . 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 1 ∂ (10. 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. λ (∂µ Aµ )2 2 1 For a massive vector boson inversion of the quadratic term including the Lorentz constraint ∝ the propagator kµ kν 1 −gµν + 1− i Dµν (k) = i . i p p i ui (p) ui (p) ¯ vi (p) ¯ incoming fermion outgoing fermion incoming antifermion outgoing antifermion i p p i vi (p) In writing down the expressions for Feynman diagrams one has to be aware that the wave functions and the propagators have more than one component. 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.74) L = − Fµν F µν − (∂ · A)2 + ψ(i/ − M )ψ − e ψγ µ ψ Aµ . + iǫ)2 where the kµ kν terms in the case where the photon couples to a conserved current (such as ψγ µ ψ) will not contribute. 4 2 In addition to the fermion propagator and fermion wave functions discussed in the previous section we have the photon propagator1 µ k ν = −i k2 gµν 1 − 1− + iǫ λ (k 2 kµ kν . Particular choices of λ are λ = 1 (Feynman propagator or Feynman gauge) and λ = ∞ (Landau gauge). 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 .Feynman diagrams for scattering amplitudes 99 The result is which contains an extra minus sign as compared to a bosonic loop.

s)¯(k. If we are interested in the scattering process of an unpolarized initial state and we are not interested in the spins in the ﬁnal state we need |M |2 summed over spins in the ﬁnal state ( s3 . s′ )γµ u(k. s4 )(−ie)γ ν u(p.5 Some examples eµ scattering The ﬁrst example is the electromagnetic scattering of an electron and a muon.s2 .s2 ) which can be written as (see also chapter 4) |M |2 = = 1 4s |¯(k ′ .79) . s3 )(−ie)γ µ u(k. where u ¯ L(m) µν = = 1 2 u(k ′ .s4 e4 (m) µν (M) L L .s4 ) and averaged over spins in the initial state (1/2 × 1/2 × s1 .Feynman diagrams for scattering amplitudes 100 10.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. To lowest order (∝ α = e2 /4π) only one diagram contributes. t2 (10.77) Combining Lµν and Lµν (M) one obtains L(m) Lµν (M) = 2 s2 + u2 + 4t(M 2 + m2 ) − 2(M 2 + m2 )2 µν and M 4π 2 (m) (10.s3 . s2 ). s′ ) ¯ u s. s2 )|2 ¯ q2 (10.76) 1 . s)γν u(k ′ . ¯ q2 (10.78) = 2 α2 2 s + u2 + 4t(M 2 + m2 ) − 2(M 2 + m2 )2 . 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 ′ . s1 ) u e2 u(p′ . s4 )γµ u(p. s3 )γ µ u(k. s1 ) ¯ −igµν u(p′ .75) Note that the q µ q ν term in the photon propagator are irrelevant because the photon couples to a conserved current.

Similarly. .s2 . s4 ) u e2 v (k ′ . s2 (10. s2 ) ¯ ¯ 1 2s u(p.1 (a) Give in diagrammatic notation the full Green functions G(4) (x1 . (b) Use the deﬁnition of the source-connected Green function Gsc and show that indeed sourceconnected diagrams survive. s2 )γµ u(k. . To lowest order (∝ α = e2 /4π) only one diagram contributes. Exercises Exercise 10. for instance Møller scattering (e− e− → e− e− ) and Bhabha scattering (e− e+ → e− e+ ) are related using crossing symmetry. The diagram. s1 )|2 ¯ s 1 . (4) (4) .Feynman diagrams for scattering amplitudes 101 The second example is the annihilation of an electron pair and creation of a muon pair.80) Note the similarity in the amplitudes for eµ scattering and e− e+ → µ− µ+ . s1 ) u(k.s4 = 1 1 e Tr(/ − m)γµ (/ ′ + m)γν k k Tr(/ + M )γ µ (/′ − M )γ ν p p s2 2 2 1 e4 1 ′ ′ = 4 2 kµ kν + kν kµ − gµν s pµ p′ν + pν p′µ − g µν s s 2 2 M 4π 2 4 and = 2 α2 2 t + u2 + 4s(M 2 + m2 ) − 2(M 2 + m2 )2 .s2 3 .s3 . This is known as crossing symmetry. the masses. 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. Basically the same diagram is calculated and the result is the same after the interchange of s ↔ t.s4 = 1 2s × v (k ′ .2. s1 )γν v(k ′ . s2 )γµ u(k. x4 ) for the interacting case in φ4 -theory to ﬁrst order in the coupling constant g as obtained from the full expression for Z[J] in section 10. (c) Do the same for the connected Green function Gc . . s3 )γ µ v(p′ . s3 )γ µ v(p′ . s4 )γ ν u(p. s4 ) v (p′ . s3 ) ¯ ¯ 1 .

. . . in the amplitude (Tij = A∗ Aj ).3 Show that the combinatorial factors found using the rules given in section 10.2 (a) Show that the translation properties of the ﬁelds and the vacuum imply G(n) (x1 + a. xn ) ≡ (2π)4 δ 4 (p1 + . .37 and 10. . . . . pn ). . . pn ) only has n − 1 independent variables in it. e(p1 ) + e(p2 ) → e(p′ ) + e(p′ ) in lowest order in α. hence we can write n i=1 d4 xi ei pi ·xi G(n) (x1 . . for instance. but the incoming and outgoing momenta are identical. . 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). . . . . Show that i the amplitude is symmetric under the interchange of t ↔ u. .3 reproduce for the diagrams precisely the combinatorial factors that appear in Eqs 10. Express the contributions in invariants s. Exercise 10. t and u.Feynman diagrams for scattering amplitudes 102 Exercise 10. |M |2 = T11 + T22 − T12 − T21 . xn + a) = G(n) (x1 . . . Pictorially this implies.4 (a) Write down the Feynman diagrams contributing to electron-electron scattering. . . Note that the function G(n) (p1 . Exercise 10. xn ). (b) Show (by using x1 as shift-varible) that this implies that n i=1 d4 xi ei pi ·xi G(n) (x1 . their sum is zero. + pn ) G(n) (p1 .38. . . or if we in general would consider all momenta as incoming. . . . 1 2 (b) Calculate the quadratic pieces and interference terms. xn ) ∝ (2π)4 δ 4 (p1 + . + pn ). . . which means overall momentum conservation in Green functions in momentum space. It is of the form −i M = A1 − A2 . .

6) (11. q). . . Here s is the invariant mass of the n-particle system. . (2π)3 2Ei i=1 (11. and the total momentum four-vector P = pa + pb . . A physical state has positive energy. . This is generalized to the multi-particle phase space dR(p1 . pn ) = (2π)4 δ 4 (P − pi ) dR(p1 . In that case cm pa = (Ea . + pn )2 . . The ﬁrst is the CM system. For two particles. . The phase space is determined by the weight factors assigned to each state in the summation or integration over states.7) pb = cm (Eb . .1 kinematics in scattering processes Phase space The 1-particle state is denoted |p . . Its square root. pb ) satisfying p2 = m2 and p2 = m2 . To be speciﬁc let us consider two frequently used frames. s is for obvious reasons known as the center of mass (CM) energy. In the CM frame the threshold value for s obviously is n 2 sthreshold = i=1 mi . −q).4) For two particle states |pa . pn ) = and the reduced phase space element by dR(s. . i n d3 pi . p) which satisﬁes p2 = E 2 −p2 = m2 .e. . It is determined by the energy-momentum four vector p = (E. (11.2) (11. . pa ) and pb = (Eb . a a b b the quantity s = P 2 = (pa + pb )2 . . for instance for determining the threshold energy for the production of a ﬁnal state 1 + 2 + . p1 . (11.1) = (2π)3 2E (2π)4 (proven in Chapter 2). . pb we start with the four vectors pa = (Ea . i. It is a useful quantity. s = (p1 + . . + n. (11.5) √ is referred to as the invariant mass squared. the 1particle phase space is d3 p d4 p θ(p0 ) (2π)δ(p2 − m2 ).Chapter 11 Scattering theory 11.3) which is useful because the total 4-momentum of the ﬁnal state usually is ﬁxed by overall momentum conservation. 103 . (11. pn ). .

9) (11. ptrf ).15) Explicit calculation of the reduced two-body phase space element gives dR(s. m2 . for instance. 2mb (11. mb and s).16) . 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. m2 .13) (11. 4π s 4π 8π s 4π (11. which in the speciﬁc case a b also can be expressed as λ(s. m2 ) a b . trf Ea (threshold) = 1 2 mb ( i mi )2 − m2 − m2 a b . 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.8) (11. p1 . = 2 s λ(s.11) (11. p1 .12) pb = (mb . a (11. Also in this case one can express the energy and momentum in the invariants. 4s (11.14) One can. m2 . m2 . m2 ) = 4(pa · pb )2 − 4p2 p2 . 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. 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 . In that case trf pa = (Ea . m2 ). 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 √ = . 0). Prove that |q| = cm Ea cm Eb (s − m2 − m2 )2 − 4 m2 m2 a a b b = 4s s + m2 − m2 a b √ = . 2mb λ(s. p2 ) = = where λ12 denotes λ(s. (11. m2 ) a b . m2 ) is a function symmetric in its three arguments. m2 .10) The function λ(s. 2 s s − m2 + m2 a b √ .

Consider the 1-particle inclusive case. in which one particle is detected. Writing E = mT cosh y. pd ) = 4π s 4π 8π s 4π dt dt √ √ . Of course there are not only 2-particle channels. The initial state. The variable s is always larger than a c b d the minimal value (ma + mb )2 . a c with q = λab /4s and q ′ = being 0 or 180 degrees. At high energies. 11.19) (11. An often used set of invariants are the Mandelstam variables.25) dR(s. . usually many particles are produced. s = (pa + pb )2 = (pc + pd )2 t = (pa − pc ) = (pb − pd ) u = (pa − pd )2 = (pb − pc )2 2 2 (11.. Consider the process a + b → c + d. In inclusive measurements no particles are detected in the ﬁnal state. pz = mT sinh y. usually is a 2-particle state. 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 . py ).24) Using the relation between t and cos θcm it is straightforward to express dΩcm in dt.. Instead of the scattering angle. where the ﬁrst is referred to as elastic channel and the set of all other channels as the inelastic channels.Scattering theory 105 Kinematics of 2 → 2 scattering processes π− + p The simplest scattering process is 2 particles in and 2 particles out. 2s (11. tmax min λcd /4s. there is usually a preferred direction. however. (11.4).20) → π0 + n → π+ + π− + n → . in exclusive measurements all particles are detected.23) which are not independent as s + t + u = m2 + m2 + m2 + m2 . = (11. E .27) with m2 = m2 − p2 = m2 + p2 + p2 . 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 .26) = 16π q s 8π λab Kinematics of inclusive hard scattering processes In high energy (hard) scattering processes.17) (11. pT = (px . Examples appear in → π− + p (11.18) (11. A speciﬁc reaction channel starts contributing at the threshold value (Eq.22) (11. for instance the momenta of incoming (colliding) hadrons and it is useful to use for the produced particle rapidity as a variable. while the ﬁnal state often arises from a series of 2-particle processes combined with the decay of an intermediate particle (resonance).21) (11. pc . the rapidity y is deﬁned T T x y y= 1 ln 2 E + pz E − pz = ln E + pz mT = tanh−1 pz . dt = 2 qq ′ d cos θcm and obtain for the two-body phase space element √ q ′ dΩcm λcd dΩcm √ = (11. H1 + H2 → h + X. The various possibilities are referred to as diﬀerent reaction channels.

¯ b (11. u = (p1 − p4 )2 = (p2 − p3 )2 .28) which (only involving angles) is easier to determine. tu = s. 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 ut = (pa − pd )2 = u. which means that rapidity distributions dN/dy maintain their shape. ¯ while for the u-channel process (ad → c¯ one has b) su = (pa + pd )2 = u. 11. t. c bd Mad→c¯(su = u. u). Given a two-to-two scattering process ab → cd one can relate the processes b p p a 2 1 d p 4 _ c −p p 1 3 d p 4 _ d −p p a 1 4 _ b −p 2 p 3 c ab cd a _ ac −p 2 _ _ b bd _ ad p 3 _ c cb They are referred to as s-channel. ut = u) = Mab→cd (s. For large energies and not too small scattering angles (θ ≫ 1/γ) y is approximately equal to the pseudo-rapidity η. tt = s.29) 3 2E 3 2 (2π) 2 (2π) 16π 2π E 16π 2 2π with at high energies the factor |p|/E ≃ 1. t-channel and u-channel processes respectively. u). b Analiticity of the ﬁeld theoretically calculated result implies Ma¯→¯ (st = t.31) . This is known as crossing symmetry. u) with s = (p1 + p2 )2 = (p3 + p4 )2 . With the momenta deﬁned as in the ﬁgures above one has for all these processes the same amplitude M (s. These variables are precisely the Mandelstam variables for the s-channel process (ab → cd).30) (11. η = − ln(tan(θ/2)) = tanh−1 (cos θ).Scattering theory 106 It is convenient because under a boost (along z) with velocity β it changes as y → y − tanh−1 β. ¯ tt = (pa − p¯)2 = s.2 Crossing symmetry In the previous chapter. t = (p1 − p3 )2 = (p2 − p4 )2 . t. uu = (pa − p¯)2 = s. ¯ tu = (pa − pc )2 = t. For the t-channel process (a¯ → ¯ one has c bd) st = (pa + pc )2 = t. (11. (11. t. 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 = = = . uu = s) = Mab→cd (s.

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

Scattering theory 108 Although this at ﬁrst sight does not look covariant. T · V ρa (11. The decay time for moving particles τ is related to the decay time in the rest frame of that particle (the proper decay time τ0 ) by τ = γ τ0 . it is. . 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. σ= √ 2 λab T · V (11. as expected. (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.e. (11.33) Decay of particles For the decay of particle a one has macroscopically dN = −Γ N.35) This quantity is not covariant.34) i. For the (proper) decay width one thus has 1 N Γ0 = . N 1 . the amount of decaying particles is proportional to the number of particles with proportionality factor the em decay width Γ. pn ).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. the number of scattered or decayed particles is given by 2 N = (2π)4 δ 4 (Pi − Pf ) iMf i dR(p1 . N and T · V are covariant.T V. 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 .38) (in which we can calculate −iMf i using Feynman diagrams). From the solution N (t) = N (0) e−Γ t one knows that the decay time τ = 1/Γ.39) One of the δ functions can be rewritten as T · V (remember the normalization of plane waves). . . . (11. . dt (11.T d4 x = V · T (2π)4 δ 4 (Pi − Pf ).36) (11.

θ)|2 . The sign diﬀerence comes from the (conventional) sign in relation between S and quantummechanical and relativistic scattering amplitude. √ (2ℓ + 1)Mℓ (s) Pℓ (cos θ). . The amplitude −M /8π s is the one to be compared with the quantum mechanical scattering amplitude f (E. θ) = −8π s ℓ (in analogy with the expansion for f (E. . see 11. (11. . implies for the scattering matrix M . for which one has dσ/dΩ = |f (E.43) dΩcm = π M (s.45) Since the amplitudes also depend on all momenta the full result for two-particle intermediate states is (in CM. note the sign and cos θ = qi · qf ). . pn ) |M |2 q 32π 2 m2 dΩ |M |2 . q n ). . . 11. 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 . p1 . 2 λab and for instance for two particles dσ = q ′ M (s. θcm ) √ q 8π s 2 2 (11. . pn ). (11. . respectively. for which the √ amplitudes squared have been calculated in the previous chapter. p1 . (11. q f ). . t) λab 4π dt. . . θ) in quantum mechanics. q n ) q n qn √ Mni (q i . i M† M √ − √ 8π s 8π s = fi n dΩn ∗ Mnf qn Mni √ √ .4 Unitarity condition (S † )f n Sni = δf i The unitarity of the S-matrix. (11.46) Partial wave expansion Often it is useful to make a partial wave expansion for the amplitude M (s.41) (11. p1 . θ).44) This can be used to get the full expression for dσ/dt for eµ and e+ e− scattering. The result is N = |Mf i |2 dR(s. 8π s 2π 8π s . pn ). 16π 2 s (11. (M † )f n (2π)4 δ 4 (Pi − Pn ) Mni . ˆ ˆ Inserted in the unitarity condition for M .Scattering theory 109 (Using normalized wave packets these somewhat ill-deﬁned manipulations can be made more rigorous). .e. θ) or M (q i . i.42) 2−body decay The diﬀerential cross section (ﬁnal state not integrated over) is given by 1 dσ = √ |Mf i |2 dR(s.40) T ·V Combining this with the expressions for the width or the cross section one obtains for the decay width Γ = 1 2m = dR(m2 . (11.47) M (s.25) −i Mf i − (M † )f i = − ∗ dΩ(ˆn )Mnf (q f .

48) If only one channel is present this simpliﬁes to ∗ ∗ −i (Mℓ − Mℓ ) = 2 qMℓ Mℓ . ℓ (11. (11.50) where Sℓ (s) satisﬁes |Sℓ (s)| = 1 and δℓ (s) is the phase shift. (11.Scattering theory 110 we obtain LHS = −i while for the RHS use is made of ℓ † (2ℓ + 1) Pℓ (ˆi · qf ) (Mℓ )f i − (Mℓ )f i .e. In general a given channel has |Sℓ (s)| ≤ 1. the total cross section can be determined. n (11. Show that σT = = 4π q 2π q2 (2ℓ + 1) Im Mℓ (s) ℓ ℓ (2ℓ + 1) (1 − ηℓ cos 2δℓ ). in combination with the partial wave expansion for the amplitudes M and the orthogonality of the Legendre polynomials immediately gives for the elastic channel. (11.53) The diﬀerence is the inelastic cross section. † −i (Mℓ )f i − (Mℓ )f i = 2 † (Mℓ )f n qn (Mℓ )ni .54) ℓ . σinel = π q2 2 (2ℓ + 1) (1 − ηℓ ). parametrized as Sℓ (s) = ηℓ (s) exp(2i δℓ (s)).49) or Im Mℓ = q |Mℓ |2 . 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 . which allows writing Mℓ (s) = Sℓ (s) − 1 e2i δℓ (s) − 1 = . θcm ) √ q 8π s 2 . 0). q ˆ i. Using σ= dΩ q ′ M (s. (11.52) From the imaginary part of M (s. η = 1) the result σel = 4π q2 (2ℓ + 1) sin2 δℓ . σel = 4π ℓ (2ℓ + 1)|Mℓ (s)|2 (2ℓ + 1) ℓ = 4π q2 ηℓ e 2i δℓ − 1 2i 2 . 2i q 2i q (11.51) and for the case that this is the only channel (purely elastic scattering.

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

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 . Its mass is M = 1232 MeV. The half-width of the resonance is M Γ. the partial width into e+ e− is Γee = 5.Scattering theory 112 (Prove this using the unitarity condition for partial waves). u) + g(t. All these decays can be seen and leave an ’invisible’ width of 498 MeV. u) = 1 1 2 (s + u2 ) + 4m2 (t − m2 ) t2 2 2 1 1 ( s − m2 )( s − 3m2 ) tu 2 2 . This characteristic shape of a resonance is called the Breit-Wigner shape. Three neutrinos explain the resonance shape. • The Z 0 resonance in e+ e− scattering with M = 91. it is furthermore easy to prove that the cross section is given by σel = M 2 Γ2 4π (2ℓ + 1) .49 GeV. u) + f (u. Three famous resonances are: • The ∆-resonance seen in pion-nucleon scattering. Note that because of the presence of a background the phase shift at resonance may actually be shifted. Knowing that each neutrino contributes about 160 MeV (see next chapter). Its mass is M = 3096. the full width is Γ = 88 keV. • The J/ψ resonance in e+ e− scattering. dt s(s − 4m2 ) with f (t. Essentially this resonance can decay into quark-antiquark pairs or into pairs of charged leptons. Exercises Exercise 11. This is a narrow resonance discovered in 1974.26 keV. 2 q (s − M 2 )2 + M 2 Γ2 (11.1 Show that the cross section for electron-electron scattering (exercise 10. u) = g(t.88 MeV.4) can be written as 4πα2 dσ = {f (t. Limiting ourselves to a resonance in one channel. its width Γ = 120 MeV. The cross section σT (π − p) also shows a resonance with the same width with a value of about 70 mb.2 GeV. t) + g(u. t)} . one can reconstruct the resonance shape for diﬀerent numbers of neutrino species. At resonance the cross section σT (π + p) is about 210 mb. Γ = 2. where furthermore σinel = 0. From this one sees that a resonance has the same shape in all channels but diﬀerent strength. a fast change in the phase shift for a narrow resonance. M2 − s (11.60) showing that the phase shift at resonance rises through δ = π/2 with a ’velocity’ ∂δ/∂s = 1/M Γ. The cross section at resonance is about 30 nb.59) reaching the unitarity limit for s = M 2 . i. which is attributed to neutrinos.e.

Thus. a quantity which we will ¯ ¯ encounter in the next chapter (section 12. Compare the results with the experimental π − lifetime τ = 2. we have here already used unitarity to ﬁnd a ﬁnite imaginary part that is proportional to the total decay width.000 123 Γ. ρcm ρcm a b 1 cm . 4.4 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∆ . Compare the result with the factor trf 1/ρa ρb va that we used in the derivation of the basic expression for cross sections. Exercise 11.999 877 Γ and Γ(π − → e− νe ) = 0. Exercise 11. Show that unitarity can be used one step further to ﬁx also the real part of the residue of the pole in the amplitude. writing N (s) M1 (s) = 2 + iM Γ .2 Calculate the ﬂux factor.Scattering theory 113 Exercise 11. cm |va − vb | in the center of mass system for two colliding particles a and b. s − M∆ ∆ ∆ 2 calculate Re N (M∆ ) . Calculate now fπ .6 × 10−8 s and the relative decay widths Γ(π − → µ− νµ ) = 0.3 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.2).8.

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. x+dx ψ(x + dx) = 1 + dxµ Dµ x+dx 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). What is this physical eﬀect by which the ’connection’ Aµ can be observed? For this consider the phase around a closed loop.14) (12.19) This gives rise to a (path dependent) phase in each point.The standard model the basis for x G and x+dx 117 G.17) The three equations given so far immediately give Dµ = ∂µ − ig Aµ (x). 12.15) ψ(x). ds s dxµ ds′ P exp ig A (s′ ) ψ(0). which connects two identical vectors. (12. = . (12. should be independent of such transformations. ds which is solved by dψ(s) ds = dxµ . Dν ] ψ(x) 1 − ig dσ µν Gµν ψ(x).e.16) which in the presence of the ’connection’ Aµ diﬀers from the total change between x ψ(x) and x+dx ψ(x+ dx). However. but expresses them with respect to diﬀerent bases. the real change thus is given by ψ(x) = (1 − dy ρ Dρ )(1 − dxσ Dσ )(1 + dy ν Dν )(1 + dxµ Dµ )ψ(x) = 1 + dxµ dy ν [Dµ . requiring that the ’connection’ Aµ (x) is such that Dµ → S(x)D µ S −1 (x). x+dx ψ(x) = = 1 + ig dxµ Aa (x)Ta µ 1 + ig dx Aµ (x) µ x x ψ(x) (12. if two diﬀerent paths to the same point give diﬀerent phases the eﬀects can be observed. i. -dx -dy µ µ dyµ µ dx For a vector.e. The relation in Eq. 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. i. We note that local gauge invariance requires that we can modify all local systems with a (local) unitary transformation S(x). (12. In principle such a phase in a given point is not observable.16. ∂µ − ig Aµ (x) ψ(x) = 0 dxµ ∂µ − ig Aµ (x(s)) ψ(x) = 0. A ’constant’ vector that only rotates because of the arbitrary deﬁnitions of local frames satisﬁes Dµ ψ(x) = 0.18) or considering a path xµ (s) from a ﬁxed origin (0) to point x. x x+dx ψ(x + dx) = (1 + dxµ ∂µ ) ψ(x).

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

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

Including a sum over the diﬀerent ﬂavors (up. V µ = ψγ µ T ψ. Neglecting at this point the local color symmetry. if the generators Qa generate a symmetry. (12. The generators Qa (in that case the rotation operators Lz . irrespective of the fact if they annihilate the vacuum. [Qa . This means that they are operators that create states from the vacuum. .e. i. etc. . 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. The conserved currents corresponding to this symmetry transformation are found directly using Noether’s theorem (see chapter 6). there must be a nonzero expectation value 0|Jµ (x)|π (k) . i. .30) which in the case that we include only two ﬂavors form an SU (2)V symmetry (isospin symmetry) generated by the Pauli matrices. suggesting the stronger relation ∂ µ Jµ (x) = fπ m2 a π a (x).The standard model 120 by degenerate representations of the symmetry group.31) . T = τ /2. H] = 0. Note that for the ﬁelds π a (x) one would have a the relation 0|π a (x)|π a (k) = ei k·x .) one can write L = f ψ f (i/ − Mf )ψf . 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′ .32) (12.29) (Note that each of the entries in the vector for ψ in the above equation is still a 4-component Dirac spinor).27) If the transformations in the lagrangian give rise to a symmetry the Noether currents are conserved. Taking the derivative.e. a massless Goldstone boson for each ’broken’ generator. . 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.e. a ∂ µ Jµ = 0. strange. In this mode the lagrangian is invariant but Qa |0 = 0 for a number of generators. i. a 0|∂ µ Jµ (x)|π b (k) = fπ k 2 ei k·x δab = fπ m2 a ei k·x δab . denoted |π a (k) . a and a′ are degenerate states. (12.28) where ψ is extended to a vector in ﬂavor space and M is a diagonal matrix in ﬂavor space. ψu mu md ψ = ψd . = ψ(i/ − M )ψ. Known examples are rotational symmetry and the fact that the spectrum shows multiplets labeled by angular momentum ℓ (with members labeled by m). This lagrangian density then is invariant under unitary (vector) transformations in the ﬂavor space. down. (12. As the a generators for a symmetry are precisely the zero-components of a conserved current Jµ (x) integrated a a over space. Goldstone mode. ψ −→ ei α·T ψ. π Example: chiral symmetry An example of spontaneous symmetry breaking is chiral symmetry breaking in QCD. M = . and one must have mπa = 0. π (12. T ] ψ.26) for all the states labeled by a corresponding to ’broken’ generators. it is easy to see that one gets ∂µ V µ = i ψ [M. Using the Dirac equation. A bit more formal. the lagrangian for the quarks consists of the free Dirac lagrangian for each of the types of quarks. ∂ ∂ (12.. called ﬂavors.

. T ] = 0. ψ −→ ei α·T γ5 ψ. but also a triplet (isospin 1) of pions. e. both are very small. etc. Therefore realization in the Weyl mode would require that all particles come double with positive and negative parity. 12. behaves as a pseudoscalar particle (spin zero. In the real world. the lagrangian density is invariant under left (L) and right (R) rotations independently. for which the ﬁeld (according to the discussion above) has the same behavior under parity. From group theory (Schur’s theorem) one knows that the latter can only be true. From the number of broken generators it is clear that one expects three massless Goldstone bosons. however. 12. M = m · 1. fπ = 0 remains. That the massless theory has this symmetry can also be seen by writing it down for the socalled lefthanded and righthanded fermions. it is easy to see that one gets ∂µ Aµ = i ψ {M. T γ5 = τ γ5 /2. (12.34) Using the Dirac equation. as the quantity ∂µ Aµ (x). which is approximately true for the up and down quarks. We know that mesons and baryons (such as the nucleons) have a well-deﬁned parity that is conserved. This symmetry is realized in the Weyl mode with the spectrum of QCD showing an almost perfect isospin symmetry. ∂ ∂ (12. nature has choosen the path SU (2)L ⊗ SU (2)R =⇒ SU (2)V . i. Therefore the world of up and down quarks describing pions.27. proton and neutron. a doublet (isospin 1/2) of nucleons. (leaving out the ﬂavor structure) the same as ψγ5 ψ. Note that these transformations also work on the spinor indices. Aµ = ψγ µ T γ5 ψ. T } γ5 ψ.3 MeV/c2 and Mn = 939.The standard model 121 Furthermore ∂µ V µ = 0 ⇐⇒ [M. How can we see this. much smaller than any of the other mesons or baryons. if in ﬂavor space M is proportional to the unit matrix. 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 . i.e. or. which is equivalent 2 to the V-A symmetry. chiral symmetry is not perfect. is what happens in the real world. including in Dirac space a γ5 matrix. with almost degenerate masses (Mp = 938. (12. generated by TR/L = 1 (1 ± γ5 )T . I. is by nature not realized in the Weyl mode. The Goldstone bosons of QCD are the pions for which fπ = 93 MeV and which have a mass of mπ ≈ 138 MeV/c2 .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. SU (2)V (isospin) symmetry is good if the up and down quark masses are identical. The currents corresponding to this symmetry transformation are again found using Noether’s theorem.6 MeV/c2 ). This symmetry. (12. i. but the fact that the symmetry is not perfect and the right hand side of Eq.e. parity minus). parity would not play a role in the world.35) In this case ∂µ Aµ = 0 will be true if the quarks have zero mass. stated equivalently.e. Still the basic fact that the generators acting on the vacuum give a nonzero result (i.35 is nonzero. The chiral ﬁelds ψR and ψL are transformed into each other under parity.g. ψR/L = 1 (1 ± γ5 )ψ. etc. nucleons and atomic nuclei has not only an isospin or vector symmetry SU (2)V but also an axial vector symmetry SU (2)A . This situation. gives also rise to a nonzero mass for the Goldstone bosons according to Eq. socalled axial vector transformations. There exists another set of symmetry transformations. where the quark masses are not completely zero. while the groundstate is only invariant under isospin rotations (R = L).e. This combined symmetry is what one calls chiral symmetry.33) which in the case of two ﬂavors form SU (2)A transformations generated by the Pauli matrices.e.

as in this case there are no massless Goldstone bosons. having two independent spin components) and three scalar ﬁelds (one degree of freedom each).37) (12. 0 ϕc = 0|φ |0 = 0 . 0 0 0 = . . 1 1 L = − Gµν · Gµν + Dµ φ · Dµ φ − V (φ). 2 massive vector ﬁelds or spin 1 bosons (2 × 3) and one scalar ﬁeld (1). 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 + . 4 2 where Since the explicit (adjoint. We have the possibility to perform local gauge transformations. in this case three-dimensional) representation reads (La )ij = −i ǫaij .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 .40) a Dµ φ = ∂µ φ − ig Wµ La φ. Dµ φ = Gµν = ∂µ φ + g Wµ × φ. again 9 degrees of freedom. but one has 1 massless gauge ﬁeld (2). thus 9 independent degrees of freedom.38) ∂µ Wν − ∂ν Wµ + g Wµ × Wν . 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). After symmetry breaking the same number (as expected) comes out. 0 φ= (12. (12.42) 2 from which one reads oﬀ that the particle content of the theory consists of one massless gauge boson 3 1 2 (Wµ ). The diﬀerence comes when we reparametrize the ﬁeld φ. like a photon. Initially there are 3 massless gauge ﬁelds (each. made into a gauge theory. while the number of massive gauge bosons coincides with the number of ’broken’ generators. (12. one sees that the ﬁelds Wµ and Gµν also can be represented as three-component ﬁelds. Therefore we can always rotate the ﬁeld φ into the 3-direction in order to simplify the calculation. Considering the same example with rotational symmetry (SO(3)) as for spontaneous symmetry breaking of a scalar ﬁeld (Higgs ﬁeld) with three components. One may wonder about the degrees of freedom. . .The standard model 122 12. The latter is a spin 0 particle (real scalar ﬁeld) called a Higgs particle. . The symmetry is broken in the same way as before and the same choice for the vacuum.39) (12. two massive bosons (Wµ and Wµ with MW = gF ) and a massive scalar particle (η with m2 = η 2 −2 m .e. (12. i. is made.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.

43) and One introduces a (weak) SU (2)W symmetry under which eR forms a singlet.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. νµ and ντ ) and their antiparticles. Thus L (f ) = i eR / eR + i eL / eL + i νeL / νeL + (µ. In the standard model one starts with a very simple basic lagrangian for (massless) fermions which exhibits more symmetry than observed in nature. τ ).45) (12.) and the gauge bosons responsible for the strong. R. ∂ ∂ ∂ (12. The procedure.44) which has an SU (2)W symmetry under transformations eiα·TW . etc. muons. the quarks (up. = ∂µ L − (12. 1 As they are massless. while the lefthanded particles form a doublet.50) (12. implies certain nontrivial relations between masses and mixing angles that can be tested experimentally and sofar are in excellent agreement with experiment. By introducing gauge ﬁelds and breaking the symmetry a more complex lagrangian is obtained. explicitly L R SU(2)W SU(2)W −→ ei α·τ /2 L.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.47) and YW is considered as an operator that generates a U (1)Y symmetry. U(1)Y U(1)Y (12. left.and righthanded particles. 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. i. electromagnetic and weak forces.48) (12. however. R −→ e−i β R. down. The lagrangian for the leptons consists of three families each containing an elementary fermion (electron e− . explicitly L −→ e−i β/2 L. its corresponding neutrino (νe . For the neutrino only a lefthanded particle (and righthanded antiparticle) exist.).The standard model 123 12. This is written as 3 Q = TW + YW . that gives a good description of the physical world.51) . under which the lefthanded and righthanded particles with YW (L) = −1 and YW (R) = −2 transform with eiβYW /2 . (12. (12. neutrinos.49) Next the SU (2)W ⊗ U (1)Y symmetry is made into a local symmetry introducing gauge ﬁelds Wµ and Bµ in the covariant derivative Dµ = ∂µ − i g Wµ · TW − i g ′ Bµ YW /2. the leptons (elektrons. muon µ− or tau τ − ).e. etc. 2 2 = ∂µ R + i g ′ Bµ R. Thus the lagrangian density is L (f ) ∂ ∂ = i L/ L + i R/ R. ψR/L = 2 (1 ± γ5 )ψ decouple. 2 (12. explicitly Dµ L Dµ R i i g Wµ · τ L + g ′ Bµ L.

Using local gauge invariance θi for i = 1.52) 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 . a complex Higgs ﬁeld + 1 φ √2 (θ2 + iθ1 ) 0 = 1 (12. . (12. h) (12. 2 and 3 may be eliminated (the necessary SU (2)W rotation is precisely e−iθ(x)·τ ). (12.The standard model 124 3 3 where Wµ is a triplet of gauge bosons with TW = 1.54) = (Dµ φ)† (Dµ φ) −m2 φ† φ − λ (φ† φ)2 .58) (12. −V (φ) L and (h2) = −Ge (LφR + Rφ L).59) . which assures with the choice of YW for the Higgs ﬁeld that Q generates the remaining U (1) symmetry. For the classical ﬁeld the choice θ4 = v is made. leading to the parametrization 1 0 φ(x) = √ (12. (12. .57) = 1 g2 v2 1 2 (Wµ )2 + (Wµ )2 (∂µ h)2 + m2 h2 + 2 8 v2 2 3 gWµ − g ′ Bµ + .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 /2.53) φ= √ (θ4 − iθ3 ) φ 2 with TW = 1/2 and YW = 1 is introduced. with the following lagrangian density consisting of a symmetry breaking piece and a coupling to the fermions. this leads to the lagrangian L (h1) .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. † i i g Wµ · τ − g ′ Bµ )φ. . 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. . 4 4 In order to break the symmetry to the symmetry of the physical world. Putting this in leads to L (f ) =L (f 1) +L (f 2) . . the U (1)Q symmetry (generated by the charge operator). + 8 g2 v2 1 + − (Wµ )2 + (Wµ )2 (∂µ h)2 + m2 h2 + 2 8 (g 2 + g ′2 ) v 2 (Zµ )2 + . + 8 (12. L where L (h1) (h) =L (h1) +L (h2) .

2θ 2θ 4 cos W cos W 0.63) (12.64) (12. (12.The standard model 125 where the quadratically appearing gauge ﬁelds that are furthermore eigenstates of the charge operator are ± Wµ = = = Zµ Aµ 1 1 2 √ Wµ ± i Wµ . 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 . σ (νµ γ ρ µL ) 2 2 (eL γ νe ) 2 k + MW (12. .69) .68) g2 ρ 5 2 (νµ γρ (1 − γ5 )µ) (eγ (1 − γ )νe ) 8 MW (µ)† (jL )ρ (e) (jL )ρ GF (µ)† (e) ≡ i √ (jL )ρ (jL )ρ .60) (12. 2 3 g Wµ − g ′ Bµ 3 ≡ cos θW Wµ − sin θW Bµ . 2 + g ′2 g 3 g ′ Wµ + g Bµ (12.61) (12. g ′ /g = tan θW . 4 2 MW g2 v2 = .67) The coupling of electrons or muons to their respective neutrinos.66) From the coupling to the photon.65) The weak mixing angle is related to the ratio of coupling constants. . 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 .62) g 2 + g ′2 3 ≡ sin θW Wµ + cos θW Bµ . 2 (12. we can read oﬀ e = g sin θW = g ′ cos θW . (12. 2 Zµ (12. 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 + .

g. leptons with CV = −1/2 + 2 sin2 θW ≈ −0. one has the relation g2 e2 1 GF √ = .80) First. Finally we want to say something about the weak properties of the quarks.70) = 2 = 2 sin2 θ 8 MW 2 v2 8 MW 2 W In this way the parameters g. 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.73) (12. 91.77) (12.40 GeV.76) with CV CA 3 = TW − 2 sin2 θW Q. one obtains L (h2) Ge v Ge = − √ (eL eR + eR eL ) − √ (eL eR + eR eL ) h 2 2 me = −me ee − eeh. Γe ≈ Γµ ≈ Γτ ≈ 83 MeV). g ′ and v determine a number of experimentally measurable quantities.e. −5 (12.166 4 × 10 0. 80. the mass of the electron comes from the spontaneous symmetry breaking but is not predicted (it is in the coupling Ge ).71) GeV −2 = = = = .05 and CA = −1/2 we calculate Γ(e+ e− ) ≈ 78. me /v is extremely small. 2 The value of α deviates from the known value α ≈ 1/137 because of higher order contributions. v (12. 3 = TW . such as2 α = e2 /4π GF sin2 θW MW MZ ≈ 1/127. With the choosen vacuum expectation value for the Higgs ﬁeld. 1. i.231 2. Comparing this with the total width into (invisible!) channels. (12. 2 2 2 (12.72) (12.19 GeV. g2 MZ 2 2 (CV + CA ).74) (12.79) Γ(Z 0 → f f ) = 48π cos2 θW For the electron. Γinvisible = 480 MeV one sees that three families of (light) neutrinos are allowed. . muon or tau.78) From this coupling it is straightforward to calculate the partial width for Z 0 into a fermion-antifermion pair. (12. from the MW mass is about 250 GeV. giving rise to a running coupling constant after renormalization of the ﬁeld theory. (12. i. The coupling to the Higgs particle is weak as the value for v calculated e. For each neutrino species (with CV = 1/2 and CA = 1/2 one expects Γ(νν) ≈ 155 MeV.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 . The masses of the fermions and the coupling to the Higgs particle are contained in L (h2) .e. as appear for instance in the decay of the neutron or the decay of the Λ (quark content uds). (12.The standard model 126 the good old four-point interaction introduced by Fermi to explain the weak interactions.5 MeV (exp.

We will not discuss this any further in this chapter. t c u ′ . 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. leptons. ′ ′ 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. The pattern is actually intriguing. The electric charge Q is then 3 ﬁxed. 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. which can account for the (observed) CP violation of the weak interactions. their antiparticles and the electroweak gauge bosons as appearing in each family. for which SU (5) or SO(10) are actually nice candidates. ¯ 3 This mixing allows all quarks with TW = −1/2 to decay into an up quark. this requires particular YW -TW assignments to get the charges right. Decay of B-mesons containing b-quarks allow estimate of Vub . suggesting an underlying larger unifying symmetry group. The . In principle one complex phase is allowed in the most general form of the CKM matrix.81) n −→ pe− νe ⇐⇒ d −→ ue− νe . Λ −→ pe− νe ¯ ⇐⇒ s −→ ue− νe . etc. Q = TW + YW /2 and constant along speciﬁc diagonals as indicated in the ﬁgure. i. This is only true if the mixing matrix is at least three-dimensional.3: Appropriate YW and TW assignments of quarks. but with diﬀerent strength.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. 12.3. 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. CP violation requires three generations.

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 ). The Λu and Λd are complex matrices in the 3×3 family and Λd Λ† = Vd G2 Vd† . the weak mass eigenstates are L (h2.q) ′ weak † mass mass DL = DL = Vu Vd DL = VCKM DL .86) the unitary CKM-matrix introduced above in an ad hoc way. .q) = −QL φΛd DR − DR Λ† φ† QL − QL φc Λu UR − UR Λ† φc† QL u d (12.227. 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 . 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. 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 . e.34. φ = −φ− 0 2 For the (squared) complex matrices we can ﬁnd positive eigenvalues. Note that we need the conjugate Higgs ﬁeld to get a U (1)Y singlet in the case of the charge +2/3 quarks. u u where we include now the three lefthanded quark doublets in QL .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.e.82 and ρ ≈ 0. the three righthanded quarks with charge +2/3 in UR and the three righthanded quarks with charges −1/3 in DR . for which we need the appropriate weak isospin doublet 0∗ 1 φ c = √ v +h . mc and mt ) and Md = Gd v/ 2 (diagonal matrix containing md . UR = uR cR tR . (12.83) where Vu and Vd are unitary matrices.82) space. each of these containing the three families.85) √ √ with Mu = Gu v/ 2 (diagonal matrix containing mu . A ≈ 0.The standard model 128 using the socalled Wolfenstein + O(λ4 ) with λ ≈ 0. i.g. d d (12. respectively). The Higgs ﬁeld is still limited to one complex doublet.22 and η ≈ 0. allowing us to write † Λu = Vu Gu Wu † and Λd = Vd Gd Wd . (12. † Λu Λ† = Vu G2 Vu . The imaginary part i η gives rise to CP violation in decays of K and B-mesons (containing s and b quarks.84) with Gu and Gd being real and positive and Wu and Wd being diﬀerent unitary matrices.

there is no family mixing for massless neutrinos.92) . Working with the mass eigenstates and the weak current becomes EL γ NL = EL e n L weak for the charged leptons we see that the weak eigenstates for the neutrinos are NL = Ve† NL with the relation to the mass eigenstates for the lefthanded neutrinos given by † ′ weak mass mass NL = NL = Ve† Vn NL = UPMNS NL .ℓ) = −LφΛe ER − ER Λ† φ† L − Lφc Λn NR − NR Λ† φc† L e n (12.91) † with UPMNS = Vn Ve known as the Pontecorvo-Maki-Nakagawa-Sakata mixing matrix. (12. UPMNS a parametrization that in principle also could have been used for quarks. we ﬁnd massless neutrinos.e.90) mass † mass mass † mass † We note that there are mass ﬁelds ER = We ER .ℓ) = −LφΛe ER − ER Λ† φ† L.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.87) is a weak doublet containing the three families of neutrinos (NL ) and charged leptons (EL ) and ER is † a three-family weak singlet. one can write Λe = Ve Ge We and we ﬁnd † L (h2. The mass ﬁelds ER † mass † mass = We ER . mµ and mτ . = − 1/6 2/3 0 1/2 1/2 0 0 1 0 1/3 1/2 − 1/6 (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 . and obtain L (h2. i. e N L= L EL (12.88) √ mass with Me = Ge v/ 2 the diagonal mass matrix with masses me . while θ23 then is determined by atmospheric neutrino oscillations requiring |∆m2 | ≈ 2.ℓ) =⇒ −Me EL Ve We ER − ER We Ve† EL .ℓ) † † † =⇒ −EL Ve Me We ER − ER We Me Ve† EL − NL Vn Mn Wn NR − NR Wn Mn Vn NL . The W -current then becomes mass mass E L γ µ NL = E L γ µ NL . The lepton sector (massive Dirac neutrinos) In principle a massive Dirac neutrino could be accounted for by a lagrangian of the type L (h2. (12. it is particularly useful because θ12 is essentially determined by solar neutrino oscillations requiring ∆m2 ≈ 8 × 10−5 12 eV2 (convention m2 > m1 ). 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. (12.5×10−3 eV2 .The standard model 129 The lepton sector (massless neutrinos) For a lepton sector with a lagrangian density of the form L in which (h2. As before. In this case. For the (massless) neutrino ﬁelds we just can redeﬁne ﬁelds into NL = Ve† NL . EL = Ve† EL . since the weak current is the only place where they show up. † † Again we continue as before now with matrices Λe = Ve Ge We and Λn = Vn Gn Wn . (12. decoupling from all known interactions. NL = Vn NL and NR = Wn NR µ mass γ µ V † V N mass . EL = Ve EL .

This is called the see-saw mechanism (outlined below).100) |MD | eiφ MR 1 1 ∗ ∗ M L nc nL + M L nL nc − M R nR nc + M R nc nR L R L R 2 2 1 1 ∗ ∗ − M D nc nc + M D nL nR − M D nR nL + M D nc nc L R R L 2 2 1 nc nR M L M D nL L c + h. one must for the neutrinos as well as charged leptons. 0 0 1 L mass.and lefthanded species through Dirac mass terms coming from the coupling to the Higgs sector as in the previous section.98) (12. the left.99) .c. the massless and massive Majorana neutrinos.ν = − MR NR NR + MR NR NR . 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. The way to circumvent this is to introduce as in the previous section righthanded neutrinos. couple the right. Thus 2 Υ ′ = nc + nL . 12. as above. For the charged leptons there cannot exist a Majorana mass term as this would break the U(1) electromagnetic symmetry. 1 ∗ c c (12. Υ′ and Υ′ (satisfying Υc = Υ and as discussed in chapter 6. 1 c ∗ c L mass. (12. Absorption of phases in the states is not possible for Majorana neutrinos. 2 although this option is not attractive as it violates the electroweak symmetry. 1 L Υ ′ = nR + nc .ν =− containing three (CP-violating) phases (α1 . are equivalent to two decoupled Majorana neutrinos (see below).88 a Majorana mass term for the (lefthanded) neutrino mass eigenstates.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 . The see-saw mechanism Consider (for one family N = n) the most general Lorentz invariant mass term for two independent Majorana spinors. The mass matrix can be written as ML |MD | eiφ M = (12. a unitary matrix relating them to the weak eigenstates. For these light Majorana neutrinos one has. Actually. with for the righthanded sector a mass term MR . For the neutrino sector. coupled by a Dirac mass term. α2 and δ). (12. If the Majorana mass MR ≫ MD one actually obtains in a natural way one Majorana neutrino with a very small mass. hence the mixing matrix becomes iα /2 0 0 e 1 (12. Υc ≡ (ΥL )c = ΥR 1 2 L and Υc = ΥL ). − MD MR nR 2 (12.95) 2 In order to have more than a completely decoupled sector. one being massive. is to add in Eq. We use here the primes starting with the weak eigenstates.and righthanded species then just form a Dirac fermion. 2 R ψ = nR + nL .96) VPMNS = UPMNS K with K = 0 eiα2 /2 0 .The standard model 130 The lepton sector (massive Majorana ﬁelds) An even simpler option than sterile righthanded Dirac neutrinos. however. For the leptons. Thus (disregarding family structure) one has two Majorana neutrinos.

Exercises Exercise 12.1 Consider the case of the Weyl mode for symmetries. V µ = ψγ µ T ψ and Aµ = ψγ µ γ5 T ψ ∂µ V µ ∂µ A µ = = i ψ [M. 2 with real masses Mi . H] = 0.102) = 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)) . This choice is possible without loss of generality because the phases can be absorbed into Υ′ and Υ′ (real must be replaced by hermitean if one includes 1 2 families). T ] ψ.e.e. (12. (12.105) for i = 1. a and a′ are degenerate states.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 . For the situation ML = 0 and MR ≫ MD (taking MD real) one 2 ﬁnds M1 ≈ MD /MR and M2 ≈ MR . i ψ {M. i. This is a mixing problem with a symmetric (complex) mass matrix leading to two (real) mass eigenstates. i. Υ2R nR (12.The standard model 131 taking ML and MR real and non-negative. 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 . Exercise 12. T } γ5 ψ.2 Derive for the vector and axial vector currents. related via c Υ1R = U nL .103) for each of which one ﬁnds the lagrangians L = and we are left with two decoupled Majorana ﬁelds Υ1L = U ∗ nc . Prove that if the generators Qa generate a symmetry. 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 † . 2 U (M M † ) U † = M0 . .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 . (12. Exercise 12. [Qa . 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′ .104) ↔ 1 1 Υi i / Υi − Mi Υi Υi ∂ 4 2 (12.

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

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