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

BooksAudiobooksComicsSheet Music### Categories

### Categories

### Categories

Editors' Picks Books

Hand-picked favorites from

our editors

our editors

Editors' Picks Audiobooks

Hand-picked favorites from

our editors

our editors

Editors' Picks Comics

Hand-picked favorites from

our editors

our editors

Editors' Picks Sheet Music

Hand-picked favorites from

our editors

our editors

Top Books

What's trending, bestsellers,

award-winners & more

award-winners & more

Top Audiobooks

What's trending, bestsellers,

award-winners & more

award-winners & more

Top Comics

What's trending, bestsellers,

award-winners & more

award-winners & more

Top Sheet Music

What's trending, bestsellers,

award-winners & more

award-winners & more

Welcome to Scribd! Start your free trial and access books, documents and more.Find out more

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

Vol. Other text books of Quantum Field Theory that are useful are given in refs [3-6].1 2 2 3.5 2. 1985. 3.2 3. Schroeder. 1995.2 2.4 3.2 36 22 . Diagrammatica.E.H.4 6. 3.3 3. Quantum Field Theory.3 3. The notes contain all essential information.3 2. Cambridge University Press. 2.4 2.4 3. 1996.5 5. M. C.3 2. 1980. Veltman. but are rather compact. An introduction to Quantum Field Theory. S. Quantum Field Theory.3 4.2 3. Cambridge University Press.5 2. Corresponding chapters in books of Ryder.2 2.7 2.3 2.-B.4 3. Addison-Wesly.1 3.1 6. The quantum theory of ﬁelds.1. 1995. L.3 6. These notes Section 1. 1.3 2.2 6.4 4.2 3.1 1.3. II: Modern Applications. Peskin and D. Quantum Field Theory.5 4.3 3. Weinberg.1 4.8 3.1 2. Srednicki.2 3.1 2. 2.2 3. 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). 6.1 5.2 1.3 References As most directly related books to these notes.2. Ryder. 1994. I: Foundations. Cambridge University Press. M.3 3. 4. Zuber. Vol.2 6. Peskin & Schroeder and Srednicki. McGraw-Hill. I refer to the book of Srednicki [1] and Ryder [2]. Cambridge University Press.3 3.4 2.2 4.V. 5. 3.1 3. Itzykson and J.3 2.5 Ryder Peskin & Schroeder Srednicki 1 1 1 1 3 2.2 2. M. 2007. Cambridge University Press.

2 7.6 8. 4.6 4.2 11. 6. 9.6 9.4 3.8 5 4.4 7.4 4. 20.3 2.3 7.2.7. 4.5 11.4 12.5 7.3 10.6 3.4 These notes Section 7.1.2 8.1 5.4.3 6 8 9 Srednicki 3 3 3 37 5.7 4. 9.2 8.2 9.5 4.6 3.1 7.5 9.5 7. 8.5 Ryder 4 4 4 4 4 4 Peskin & Schroeder 2.2 12.6 6.2 10.3.8 6.7 5.3 20.1 20.3 12.4.1 9.1 20. 4.1 8.3 6.2.5 2.6 3.5. 2.5 .2.6 10.3 3.1.4 8.1 12.3.1 9.3.4 10.5 4.4 2.4 12.3 9. 2.4.3 7.3 8. 4.5 9. 6. 6. 6.4 9.1.3 11.2 5. 6.2.2 4.10 8.1 11.5 6.1 10.1 20.3 8.5 6 7 6 6.

5 Corrections 2011/2012) • Extension of Exercise 3. • Slight rewriting of discussion of Majorana mass term in section 6.3 • First two sentences of Chapter 5 have been modiﬁed.2 .

3) The uncertainty principle tells us that in this representation the momenta cannot be fully determined. It is given by pop ψ(r) = −i ∇ψ(r). In quantum mechanics. usually given divided by 2π. Corresponding position and momentum operators do not commute. In the framework of classical mechanics as well as quantum mechanics the position of a particle is a well-deﬁned concept and the position coordinates can be used as dynamical variables in the description of the particles and their interactions. quantum eﬀects do not play a role anymore and one is in the classical domain.054 571 68 (18) × 10−34 J s = 6.1 Quantum ﬁeld theory In quantum ﬁeld theory the theories of quantum mechanics and special relativity are united.5) One can also choose a representation in which the momenta of the particles are the dynamical variables. In special relativity a special role is played by the velocity of light c.6) . One can talk about states |r and the wave function ψ(r) = r||ψ . 1 (1. ˜ The corresponding states are |p and the wave functions ψ(p) = p||ψ are the Fourier transforms of the coordinate space wave functions.2) In the limit that v ≪ c one reaches the non-relativistic domain. S ≫ .4) where δij is the Kronecker δ function. In quantum mechanics a special role is played by Planck’s constant h. ≡ h/2π = 1. (1. being eigenvalues of the position operators. the action of the momentum operator in the coordinate representation is not as simple as the position operator. c = 299 792 458 m s−1 .Chapter 1 Introduction 1.1) In the limit that the action S is much larger than . (1.582 119 15 (56) × 10−22 MeV s. (1. (1. pj ] = i δij . In this coordinate representation the position operators r op simply acts as rop ψ(r) = r ψ(r). ˜ ψ(p) = i d3 r exp − p · r ψ(r). Indeed. They satisfy the well-known (canonical) operator commutation relations [ri . (1. the position can in principle be determined at any time with any accuracy.

Introduction

2

and ψ(r) =

d3 p exp (2π )3

i

˜ p · r ψ(p).

(1.7)

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

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

1.2

Units

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

Introduction

3

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

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

2 2

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

c

= 0.389 379 323 (67) GeV2 mbarn

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

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

−27 2

(1.12) (1.13)

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

2

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

Introduction

4

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

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

1.3

**Conventions for vectors and tensors
**

3

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

i=1

xi ei = xi ei , ˆ ˆ

(1.17)

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

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

So we could have used all upper indices in the above equations. (1.µn = Λµ1ν1 .3). x1 . The (invariant) lengths often have special names. • ds2 < 0 (spacelike intervals).1. These transformations do change the components of a vector. • ds2 = 0 (lightlike or null intervals). such as eigentime τ for the position vector τ 2 ≡ x2 = t2 − x2 .3) with basis nµ (µ = ˆ 0. in which case zi = ǫijk xj yk . one must distinguish tensors with upper or lower indices and one can have mixed tensors. x2 = x · x = xµ xν nµ · nν = xµ xν gµν .Introduction 5 that can be used in the cross product of two vectors z = x × y. independent of a coordinate system. x2 . The lower indices are constructed in the following way. In special relativity we start with a four-dimensional real vector space E(1. we will use upper indices for the three-vectors. (1. The scalar product of two diﬀerent vectors x and y is denoted x · y = xµ y ν gµν = xµ yµ = xµ y µ = x0 y 0 − x · y. The length (squared) of a vector is obtained from the scalar ˆ product. Useful relations are ǫijk ǫℓmn ǫijk ǫimn ǫijk ǫijl = = = δiℓ δjℓ δkℓ 2 δkl . x1 . linear transformations that do not change the length of a four-vector are called the Lorentz transformations. x3 ) = (t. The Poincar´ transformations include e e Lorentz transformations and translations. 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 . δim δjm δkm δin δjn δkn . ...νn can be conν structed. Relations relating tensor expressions. it is convenient to distinguish lower indices from upper indices. Vectors are denoted x = xµ nµ . given by g00 = −g11 = −g22 = −g33 = 1 (the ˆ ˆ other components are zero).21) (1.23) δjm δkn − δjn δkm . ˆ ˆ (1. Because of the diﬀerent signs occurring in gµν . We note that for Euclidean spaces (with a positive deﬁnite metric) vectors and tensors there is only one type of indices. and are given by (x0 . xµ = gµν xν . One can distinguish: • ds2 > 0 (timelike intervals). xi are the three space components. x3 ). linear transformations that leave the length of a vector invariant are called (homogeneous) Lorentz transformations. One has x2 = xµ xµ = t2 − x2 .. where the coordinate t = x0 is referred to as the time component. For four-vectors in Minkowski space we will use the notation with upper indices and write x = (t. Λµn n T ν1 . The real. 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.26) (1. No diﬀerence is made between upper or lower. are . however. x) = (x0 . x2 .24) The quantity gµν ≡ nµ · nν is the metric tensor. Unlike in Euclidean space. In Minkowski space.22) (1. denoted as V ′µ = Λµν V ν . the invariant length or distance (squared) is not positive deﬁnite.2. 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 . When 3-dimensional space is considered as part of Minkowski space. the points lie on the lightcone and they can be connected by a light signal. . −x). The invariant distance between two points x and y in Minkowski space is determined from the length dsµ = (x − y)µ .. Many other four vectors and tensors transforming like T ′ µ1 .25) Within Minkowski space the real.

(1. ∂2 . because one has (1. (1. The product of two epsilon tensors is given by g µµ ′ g νµ ′ g ρµ ′ g σµ g νν ′ g ρν ′ g σν ′ ′ ′ ǫµνρσ ǫµ ν ′ ′ ′ ρ σ′ = − g µν ′ g νν ′ g ρν ′ g σν g νρ ′ g ρρ ′ g σρ ′ ′ ′ g µρ ′ g νρ ′ g ρρ ′ g σρ g νσ ′ g ρσ ′ g σσ ′ ′ ′ g µσ ′ g νσ ′ g ρσ ′ g σσ .33) (1. p) = (E. starting from ǫ0123 = 1. . It is an invariant tensor. is timelike with invariant length (squared) p2 = p · p = pµ pµ = E 2 − p2 = m2 .Introduction 6 called covariant. g0 = g1 = g2 = g3 = 1. 0123 0123 ′ ′ ǫ µνρσ ǫµνρσ The ﬁrst identity. Relate this quantity to the electron Compton wavelength −e via the dimensionless λ ﬁne structure constant α. deﬁned by ∂2 − ∇2 . ′ ′ . ∂t ∂x ∂y ∂z ν ∂µ xν = gµ .36) Exercises Exercise 1. a∞ = 4πǫ0 2 /me e2 . 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. ∂1 .31) ǫµνρσ ǫµν ′ ′ ρσ ′ = − (1. where m is called the mass of the system. For the contraction of a vector with the antisymmetric tensor one often uses the shorthand notation ǫABCD = ǫµνρσ Aµ Bν Cρ Dσ . ∂t (1. Note that in this equation one has on left.29) 2 = ∂ µ ∂µ = ∂t2 The value of the antisymmetric tensor ǫµνρσ is determined in the same way as for ǫijk .30) (Note that there are diﬀerent conventions around and sometimes the opposite sign is used). Examples are the scalar products above but also relations like pµ = m dxµ /dτ for the momentum four vector.35) ρ ǫµνρσ ǫµν σ σ ǫµνρσ ǫµνρ ′ ′ = −2 g ρρ g σσ − g ρσ g σρ = −6g σσ . (1.27) It is easy to convince oneself of the nature of the indices in the above equation. ∂3 ) = ∂ ∂ ∂ ∂ .1 (a) In the the Hydrogen atom (quantum system) the scale is set by the Bohr radius. = ∂ . constructed via the metric tensor itself. .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.28) ν ν Note that gµ with one upper and lower index. 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. The length squared of ∂ is the d’Alembertian operator.32) . not aﬀected by Lorentz transformations. (1. The derivative ∂µ is deﬁned ∂µ = ∂/∂xµ and we have a four vector ∂ with components (∂0 . Each of these permutations leads to a minus sign. (1. explicitly written as (p0 .∇ . . for instance. p). = −24.

a1 . [Note: what is the MKS unit for V/T?] Exercise 1. Use that to give the ˆ ˆµ ˆ components of the basis vectors n+ and n− . σ} is a permutation of {0. 1) ˆ . Note that ˆ ˆ ˆ ˆ a · nα = aα and derive from that the components nα of the four vector nα . Also construct and calculate the Planck length Lpl . ˆ ˆ . n2 . but only appropriate combinations. 0. c. n2 .2 Prove the identity A × (B × C) = (A · C) B . which is the Compton wavelength for the Planck mass. a2 . 3} the index α can only be equal to one of the indices in ǫµνρσ . by a simple few-line reasoning [For instance: If {µ. 0. Compare it with the proton mass and use Eq. . a2 ] we can ﬁnd basis vectors n− . How are a+ and a− related to a+ and a− . 0) ˆ n2 = (0. . ǫ0 . g+− and g−+ . 1. g−− . 0) ˆ n0 = (1. re = e2 /4πǫ0 me c2 to the Compton wavelength. Exercise 1.4 Lightcone coordinates for a four vector a (which we will denote with square brackets as [a− . a1 . 0. ˆ ˆ ˆ ˆ These are n1 = (0. e. 1. a+ . 0. 0. n+ . 0) ˆ n3 = (0. a1 . (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). me . (a) Express the scalar product a · b in lightcone coordinates and deduce from this the values of g++ . (b) The coordinates (a0 . (d) Use the value of the gravitational constant GN / c5 = 6.3. This demonstrates how a careful use of units can save a lot of work. aT ]) are deﬁned through √ a± ≡ (a0 ± a3 )/ 2. 2. ].3 Prove the following relation ǫµνρσ g αβ = ǫανρσ g µβ + ǫµαρσ g νβ + ǫµνασ g ρβ + ǫµνρα g σβ . One does not need to know . . a+ . n3 . 1. 0. α and me c2 = 0. n1 .13 to give its actual value in kg. ν. a2 ] or [a− . (c) Calculate the Compton wavelength of the electron and the quantities under (a) and (b) using the value of c. 1.511 MeV.71 × 10−39 GeV−2 to construct a mass Mpl (Planck mass).Introduction 7 (b) Relate the classical radius of the electron (a relativistic concept). ρ.(A · B) C using the properties of the tensor ǫijk given in section 1. a3 ) are the expansion coeﬃcients using the basis vectors n0 . Exercise 1. Also for the coordinates [a− . a+ . n1 .

3 are not covariant.7) k 0 = ± k2 + M 2 = ±Ek . φk (r. Thus a covariant equation can be obtained by starting with the (covariant) equation for the invariant length of the four vector (E.Chapter 2 Relativistic wave equations 2. one obtains (2.2) Acting on the wave function one ﬁnds for a free particle. (2. t) = ∂2 − ∇2 + M 2 φ(r.1 The Klein-Gordon equation In this chapter. ∂t p −→ pop = −i∇. 2 + M 2 φ(r. 2M (2.4) where M is the particle mass. (2. E = p2 . i ∂ ∇2 ψ(r. (2. The Schr¨dinger equation in quantum mechanics is the operator equation o corresponding to the non-relativistic expression for the energy.6) The energy spectrum is not bounded from below: considering the above stationary plane wave solutions. 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. p). written as pµ −→ i∂µ is covariant (the same in every frame of reference). namely plane waves characterized by a wave number k. we just want to play a bit with covariant equations and study their behavior under Lorentz transformations.3) Equations 2. t) = exp(−i k 0 t + i k · r).1 and 2. Although it is straightforward to ﬁnd the solutions of this equation. t) = − ψ(r.2. t). t) = 0.1) under the substitution (in coordinate representation) E −→ Eop = i ∂ . p2 = pµ pµ = E 2 − p2 = M 2 . ∂t 2M (2. But the replacement 2.5) with (k 0 )2 = k2 + M 2 . 8 . Substitution of operators gives the Klein-Gordon (KG) equation for a real or complex function φ. ∂t2 (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.2 Mode expansion of solutions of the KG equation Before quantizing ﬁelds. −i φ∗ ∇ φ .8) (2. relativistically the density is not a scalar quantity. For this we note that an arbitrary solution for the ﬁeld φ can always be written as a superposition of plane waves. The integration over k-modes clearly is covariant and restricted to the ‘mass’-shell (as required by Eq. for which we need the interpretation of φ itself as an operator. This leads to the existence of negative densities.14) ˜ ˜ Introducing φ(Ek . just as the dependence on time was in ordinary quantum mechanics. ↔ ↔ (2. we want the most general solution.15) .9) They satisfy the continuity equation. (2π)3 2Ek (2. −k) ≡ b∗ (k) one has φ(x) = d3 k e−i k·x a(k) + ei k·x b∗ (k) = φ+ (x) + φ− (x). 2M 2M (2. rather than as a wave function.12) It is easy to see that this current is conserved if φ (and φ∗ ) satisfy the KG equation. −k) . (2. there are no longer fundamental objections to mix up space and time.2) is j µ = i φ∗ ∂ µ φ or (ρ. φ(x) = d4 k ˜ 2π δ(k 2 − M 2 ) e−i k·x φ(k) (2π)4 (2. The KG equation. 2. k) + ei k·x φ(−Ek .13) ˜ with (in principle complex) coeﬃcients φ(k). It is possible to rewrite it as an integration over positive energies only but this gives two terms (use the result of exercise 2. there are solutions with negative energy. (2.5). As we will see later both problems are related and have to do with the existence of particles and antiparticles. it is simply a matter of being careful to ﬁnd a consistent (covariant) theory. 2. (2.3). but rather the zero component of a four vector. 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.11) Therefore. φ(x) = d3 k (2π)3 2Ek ˜ ˜ e−i k·x φ(Ek . Then. ∂µ j µ = 0.10) ∂t which follows directly from the Schr¨dinger equation. j) of the four-current j. And. however.e. This operator has all possible solutions in it multiplied with creation (and annihilation) operators.Relativistic wave equations 9 i. which is what Lorentz transformations do. ∂ρ = −∇ · j. j) = ↔ i φ∗ ∂0 φ. The appropriate current corresponding to the KG equation (see Excercise 2. This continuity equation can be written down o covariantly using the components (ρ. k) ≡ a(k) and φ(−Ek . having the KG equation as a space-time symmetric (classical) equation.

14 and 2.20) one has for the mode amplitudes aP (k) = a(−k) and bP (k) = b(−k). 2. It is easy to show that x φP (x) = φ(˜) x = = = or since one can deﬁne φP (x) ≡ d3 k e−i k·x aP (k) + ei k·x bP ∗ (k) . (2π)3 2Ek (2.e. This shows how parity transforms k-modes into −k modes. Performing some Lorentz transformation x → x′ = Λx.23) . Another symmetry is found by complex conjugating the KG equation. It is trivial to see that ∂µ ∂ µ + M 2 φ∗ (x) = 0.3 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. i. one thus must have that the function φ → φ′ such that φ′ (x′ ) = φ(x) or φ′ (x) = φ(Λ−1 x). which implies (xµ ) = (t.22) showing that with each solution there is a corresponding charge conjugated solution φC (x) = φ∗ (x). In terms of modes one has φC (x) = φ∗ (x) = ≡ d3 k e−i k·x b(k) + ei k·x a∗ (k) (2π)3 2Ek d3 k e−i k·x aC (k) + ei k·x bC∗ (k) . for which we consider space inversion. φP (x) ≡ φ(˜) (P for parity). −x) ≡ (˜µ ). x ∂µ ∂ µ + M 2 φ(˜) = 0.19) implying that for each solution φ(x) there exists a corresponding solution with the same energy.15 one has elimated k 0 and in both equations k · x = Ek t − k · x. (2. x Transforming everywhere in the KG equation x → x one obtains ˜ ˜ ˜ x ∂µ ∂ µ + M 2 φ(˜) = 0.21) 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) . Since a · b = a · ˜ it is easy to see that ˜ b.Relativistic wave equations 10 In Eqs 2.18) (2. (2π)3 2Ek (2. (2. The coeﬃcients a(k) and b∗ (k) are the amplitudes of the two independent solutions (two.16) The consequence of this is discussed in Exercise 2.17) (2.6 We will explicitly discuss the example of a discrete symmetry. x) → (t. (2π)3 2Ek (2. They are referred to as mode and anti-mode amplitudes (or because of their origin positive and negative energy modes). 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). changing the sign of the spatial coordinates. (2.

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

Exercise 2. One then ﬁnds that Q is expressed as an inﬁnite sum over number operators a† a (particles) and b† b (anti-particles). i (a) Show that relativistic invariance. We will encounter these quantities later as energy and momentum. ∂t where ψ is a wave function and the nature of α and β is left open for the moment.5 (a) As an introduction to parts (b) and (c). d3 x (∂ {0 φ)∗ (∂ i} φ).15.Relativistic wave equations 12 Exercise 2. 2. 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).7 To solve the problem with positive and negative energies and get a positive deﬁnite density. in terms of the a(k) and b(k). (c) Similarly.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). Exercise 2. . B} = AB + BA. express the quantities E(t) = P i (t) = d3 x (∂0 φ)∗ (∂0 φ) + ∇φ∗ · ∇φ + M 2 φ∗ φ . an attempt to construct a ﬁrst order diﬀerential equation for the time evolution would be to write. B] = AB − BA and the anticommutator as {A. ∂ ψ(r. j 0 = ψ†ψ and j = ψ † αψ. 1 2 (b) Express now the full charge QV (t) (exercise 2. αj } = 2 δ ij I. t). and β 2 = I. (b) Show that one has current conservation for the current 2 We denote the commutator as [A. show that for two ﬁelds φ1 (with coeﬃcients a1 and b∗ ) 1 and φ2 (with coeﬃcients a2 and b∗ ) one has 2 d3 x φ∗ (x) φ2 (x) 1 = d3 k 2 (2π)3 4Ek a∗ (k)a2 (k) + b1 (k)b∗ (k) 1 2 + a∗ (k)b∗ (−k) e2iEk t + b1 (k)a2 (−k) e−2iEk t .2) in terms of the a(k) and b(k) using the expansion in Eq. t) = (−i α · ∇ + m β) ψ(r. From this one concludes that α and β must be matrix-valued and ψ must be a multi-component wave function. β} = 0. also using the result in (a). Note that a{µ bν} indicates symmetrization. {αi . 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 . i.e. a{µ bν} ≡ aµ bν + aν bµ . 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.

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

the rotations in SO(3) can be generated from inﬁnitesimal rotations that can be expressed in terms of a basis of three generators Lx . (3.8) ˆ that (σ · a)(σ · b) = a · b + i σ · (a × b).6) Summarizing. (3. y] ∈ A.z N N or (Lk )ij = −i ǫijk .2 π) (-n. x]). ] that satisﬁes • ∀ x. With matrix commutation this algebra satisﬁes the requirements for a Lie algebra. . is a 3-dimensional Lie-group. 2 (3. for rotations around z for instance. which reﬂects itself in the noncommutation of the generators.and y-axes 0 0 0 0 0 0 0 −i . One way of viewing the parameter space. π ) π -sphere (antipodes identified) 2 π -sphere (surface identified) Figure 3. thus is as the surface i=0 of a sphere in 4 Euclidean dimensions. y] = −[y. Locally this is a 3-dimensional Euclidean space and SU (2). [y. x]] + [z. 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 (3.1: the parameter spaces of SO(3) (left) and SU (2) (right). ˆ therefore. Ly and Lz . It is straightforward to check that any (ﬁnite) rotation can be obtained from a combination of inﬁnitesimal rotations. namely that there exists a bilinear product [.5) Rotations in general do not commute. They satisfy the commutation relations [Li . we turn to the group SU (2) of special (det A = 1) unitary (A† = A−1 ) 2 × 2 matrices. ˆ R(ϕ. π) (n. • [x. x] = 0 (thus [x. Lj ] = i ǫijk Lk . 0 (3.4) R . with real a’s and 3 (ai )2 = 1. [x.7) = a0 0 −1 i 0 1 0 0 1 • [x. These matrices can be deﬁned as acting on 2-component spinors (χ → Aχ) or equivalently as acting on 2 × 2 matrices (B → ABA† ). y]] = 0 (Jacobi identity). n) = 1 cos ˆ 1 Note ϕ ϕ ˆ + i (σ · n) sin 2 2 = exp i ϕ ˆ σ·n . In the same way we can consider rotations around the x. Lx = Ly = 0 0 0 i 0 −i 0 ϕ ˆ . z]] + [y. These generators form a threedimensional Lie algebra SO(3). Writing a0 = cos(ϕ/2) and a = n sin(ϕ/2) we have1 A = A(ϕ. [z. y ∈ A ⇒ [x. and thus (σ · n)2 = 1. Next.Groups and their representations 14 (n. z ) = lim N →∞ that are generated by i 0 .

For instance in a function space translations. H] = 0 for g ∈ G. (3. n) −→ R(ϕ. In the full parameter space. φ′ (r) = φ(r + a) = φ(r) + a · ∇ φ + 1 (a · ∇)2 φ + . since any ﬁnite transformation can be constructed from the inﬁnitesimal ones.Groups and their representations 15 where we have used the (for operators new!) deﬁnition eA ≡ An 1 = 1 + A + A2 + . .e. immediately sees that the Lie algebras are identical.12) One. the above mapping corresponds to the trivial mapping of the Lie algebras. i. H] = 0 for g ∈ G. n) i ∂ϕ = ϕ=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. .11) Thus σx /2. . n) π ≤ ϕ ≤ 2π. n) ≈ 1 + i ϕ J · n. [g. . SU (2) ≃ SO(3). They satisfy σi σj σk = i ǫijk .15) [g. The next issue is to ﬁnd the appropriate symmetry operators in the Hilbert space. 2 2 2 (3. 2! .10) σ ˆ · n. The inﬁnitesimal generators of SU (2) are ˆ obtained by considering inﬁnitesimal transformations. thus. i.e. one has a Lie algebra isomorphism that is linear and preserves the bilinear product. it is suﬃcient that the generators commute with H. now with radius 2π and with all points at the surface identiﬁed (see ﬁgure).e. . There exists a corresponding mapping of the groups given by µ: SU (2) −→ SO(3) ˆ ˆ A(ϕ. the SU (2) → SO(3) mapping is a 2 : 1 mapping where both A = ±1 are mapped into R = I.14) For a Lie group. thus. (3. 3.13) 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. sometimes in more than one way (but this will be discussed later). σy /2 and σz /2 form the basis of the Lie-algebra SU (2). .9) The parameter-space of SU (2). The existence of symmetries means that there are operators g belonging to a symmetry group G that commute with the Hamiltonian. Near the identity. i. . n) ˆ −→ R(2π − ϕ. in particular that AB → RAB = RA RB ). however. n! 2! n=0 ∞ (3. 2 (3. Suppose we have a system described by a Hamiltonian H. with ˆ J ·n ≡ ˆ 1 ∂A(ϕ. for ﬁxed n. also can be considered as a ﬁlled 3-sphere. (3. ˆ ˆ A(ϕ.

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

however. Therefore the representation and conjugate representation are equivalent in this case (see Excercise 3. 2 From the (hermitean) representation Φ(g) of G. Given a representation. then ǫχ → ǫσχ = ǫσǫ−1 ǫχ = −σ∗ ǫχ. the extension from a local one must be unique. For a decent (welldeﬁned) global representation. Explicitly. if χ → σχ. however. χ) ≡ e−iϕJz e−iθJy e−iχJz . This turns out to be the case for all half-integer representations of the Lie algebra. 0 ≡ ǫ0 ≡ ǫz . i. If a transformation U acts on χ. J = 0 0 + 0 0 0 0 0 0 0 0 √ 2 0 √ 0 . For SO(3). the possibility thus exist that some extensions will not be well-deﬁned representations. any point in the parameter space can be reached in a unique way and any local (inﬁnitesimal) representation can be extended to a global one. The jump U → U ∗ implies for the generators σ/2 → −σ ∗ /2.20) By moving through the parameter space the D-functions can be extended to global functions for all allowed angles. Consider the j = 1/2 representation of SU(2). Of all groups of which the Lie algebras are homeomorphic. all states in the representation space have the same energy. . ǫz } are 0 0 0 0 0 i 0 Jx = 0 0 −i . the conjugate representation is not a new one. For those global representations. one can look at the conjugate representation. ǫy . Jy = 0 0 0 . If rotations leave H invariant. however. contractable and paths that run from a point at the surface to its antipode. θ. For an element in the group G the corresponding point in the parameter space can be reached in two ways. The matrix elements of these unitary representations are known as D-functions. dm′ m (θ) ≈ δm′ m − iθ(J 2 )m′ m . UE (ϕ. The group SO(3). however. For j = 1 another commonly used representation starts with three Cartesian basis states ǫ related to the previous basis via 1 |1. 2 |1.19) (3.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 . This works for SU (2). If the group is simply connected. θ. e. Jz = i 0 i 0 −i 0 0 0 −i 0 0 0 0 0 . the simply connected group is called the (universal) covering group. the conjugate transformation U ∗ acts on χ∗ . 1 ≡ ǫ1 ≡ − √ (ǫx + i ǫy ) . 1 |1. χ) = eim ϕ dm′ m (θ) e−imχ . that is any closed curve in the parameter space can be contracted to a point. SU (2) is the covering group of SO(3). to be precise the states ǫχ transform via −σ ∗ .e. There exist two diﬀerent types of paths. one obtains the unitary representations Φ(g) = exp(i Φ(g)) of the full group G. J− = 2 0 0 0 0 √ 3 0 0 0 0 √ 2 0 0 0 0 . m = Dm′ m (ϕ. For the transformations A in SU (2) and R in SO(3) these are commonly parametrized with Euler angles. θ. (j) (j) ′ The spin matrices for that Cartesian basis {ǫx . −1 ≡ ǫ−1 ≡ √ (ǫx − i ǫy ) .g. the topological structure of the group is important. χ)|j. (j) (3. For SU(2). is not simply connected. j. Inﬁnitesimally (around the identity) the D-functions for SU (2) and SO(3) are the same. Because there exists a matrix ǫ such that −σ ∗ = ǫ σǫ−1 one immediately sees that appropriate linear combinations of conjugate states. or equivalently the Hilbert space can be written as a direct product space of spaces V (j) . m′ |UE (ϕ.3).

25) (3. (det Λ = +1 is called proper. it is convenient to use a vector notation for the points in Minkowski space.22) (Note that xT is a row vector).23) The proper tensor structure of the matrix element (Λ)µν is a tensor with one upper and one lower index. proof: (ΛT GΛ)00 = (G)00 = 1 → Λµ0 gµν Λν0 = 1 → (Λ00 )2 − i (Λi 0 )2 = 1. 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 . 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 . σ/2 in a two-dimensional (spin 1/2) case. Thus one has the (pseudo-orthogonality) relation ΛT GΛ = G ⇔ GΛT G = Λ−1 ⇔ ΛGΛT = G. 0 0 g00 g01 g02 g03 1 0 g g11 g12 g13 0 −1 0 0 = . In this section we consider the Poincar´ group. 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. consisting of the Lorentz group and translations. which as a matrix equation with (Λ)µν = Λµν and (G)µν = gµν gives (ΛT )ρµ (G)µν (Λ)νσ = (G)ρσ . e To derive some of the properties of the Lorentz group. Denoted in terms of column vectors and 4-dimensional matrices one writes for the poincar´ transe formations x′ = Λ x + a.27) (3. (ii) |Λ00 | ≥ 1. We considered the generators and looked for representations in ﬁnite dimensional spaces.21) g 20 g21 g22 g23 0 0 −1 0 g30 g31 g32 g33 0 0 0 −1 the scalar product can be written as x2 = xT G x (3.3 The Lorentz group In the previous section spin has been introduced as a representation of the rotation group SU (2) without worrying much about the rest of the symmetries of the world.Groups and their representations 18 3. (3. From this property. explicitly x′µ = (Λ)µν xν + aµ or x′µ = Λµν xν + aµ . e. G = 10 (3. det Λ = −1 is called improper). it is easy to derive some properties of the matrices Λ: (i) det(Λ) = ±1. Writing x as a column vector and the metric tensor in matrix form.26) (3.g.

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

g. M ρσ ] = −i (g µρ M νσ − g νρ M µσ ) − i (g µσ M ρν − g νσ M ρµ ) . The extension to the Poincar´ group is e ↑ straightforward.35) thus (ω)ij = −(ω)ji and (ω)0i = (ω)i0 . (3. 3 30 with η = ω and K = M (e. which are obtained by writing the inﬁnitesimal parameters (ω)µν in terms of six antisymmetric matrices (M αβ )µν .38) Explicitly. with ϕi = − 1 ǫijk ω jk and J i = 2 z-axis). Multiplying the proper orthochrone transformations with one of them gives all Lorentz transformations. We will just require Poincar´ invariance for the generators themselves. ǫ). explicitly reading x′µ = = (Λ)µν xν + aµ = ((I)µν + (ω)µν ) xν + ǫµ Λµν xν + aµ = (g µν + ω µν ) xν + ǫµ . inﬁnitesimally approximated by (I + ω.33) These four transformations form the Vierer group (group of Klein). three of which only involve spatial coordinates (rotations) and three others involving time components (boosts). We therefore ﬁnd (again) that there are six generators for the Lorentz group. One can easily convince oneself that i (ω)µν = ω µν = − ωαβ (M αβ )µν . K = M ). Also this group can be divided into four parts. In order to ﬁnd the commutation relations including the translation generators.34) The condition ΛT GΛ = G yields g ρµ + ω ρµ gρσ (g σν + ω σν ) = gµν =⇒ ωνµ + ωµν = 0.4 The generators of the Poincar´ group e ↑ The transformations belonging to P+ are denoted as (Λ. inf (3. Of the four parts only L↑ forms a group.39) (e.Groups and their representations 0 0 −1 0 0 −1 0 0 = 0 0 −1 0 0 0 0 −1 ∈ L↓ + 20 Is It (space-time inversion) (3. J 3 = M 12 ) and for the (inﬁnitesimal) boosts (along (3. There are six generators. i i0 i i0 1 ijk M jk 2ǫ (3. (3. the Lorentz transformations form a (Lie) group (Λ1 Λ2 and Λ−1 are again Lorentz transformations. Thus the two vector operators J and K form under Lorentz transformations an antisymmetric tensor M µν .g. The e 2 These commutation relations are the same as those for the ’quantummechanics’ operators i(xµ ∂ ν − xν ∂ µ ) . a). Summarizing.37) The algebra of the generators of the Lorentz transformations can be obtained by an explicit calculation2 . This is a + normal subgroup and the factor group is the Vierer group. [M µν . 3. we have for the (inﬁnitesimal) rotations (around z-axis) ΛR = I + i ϕ3 J 3 = I − i ω12 M 12 . we continue using covariance arguments.40) ΛB = I + i η 3 K 3 = I − i ω30 M 30 .36) (3. 2 (M αβ )µν = i g αµ g βν − g α g βµ ν (3. We now want to ﬁnd a covariant form of the six generators of the Lorentz transformations. P+ etc.

= −i (g µρ P ν − g νρ P µ ) . H] = i P i . Of the generators of the Poincar´ group we choose ﬁrst of all the generators P µ . . writing the generator P = (H/c. P ) in terms of the Hamiltonian and the three-momentum operators.43) giving the following commutation relations by equating the coeﬃcients of ǫµ and ωµν . . H] = 0. ǫ) = 1 − ωαβ M αβ + i ǫα P α .48) . check that U1 U2 corresponds with Λ1 Λ2 ). . a)P µ U † (Λ. (3. a)P µ U (Λ.41) 2 At this point. . known from non-relativistic quantum mechanics are obtained. For instance. K j ] = i ǫijk K k . = pµ |p. [J i . To ﬁnd other generators that commute with P µ we look for Lorentz transformations that leave the four vector pµ invariant. K j ] = −i ǫijk J k /c2 . These form a group called the little group associated with that four vector. [P µ . . (3. P ν ] [M µν . as part of the set. 3. leads to U (Λ. J j ] = i ǫijk J k . [J i . Explicitly. [K i . . one obtains 2 [P i . We have here reinstated c. The requirement that P µ transforms as a four-vector (for which we know the explicit behavior from the deﬁning four-dimensional representation). (3. P µ |p.45) Note that the commutation relations among the generators M µν in Eq. other states in that representation are obtained by the action of operators outside the maximal commuting set. the generators J± in the case of the rotation group.47) (3. 2 2 (3. a) = Λ−1 µ P ν . the generators M αβ no longer exclusively act in Minkowski space. These deﬁne states with speciﬁed quantum numbers that are eigenvalues of these generators.44) (3. the generators J 2 and J 3 in the case of the rotation group. The eigenvalues of these will be the four-momentum pµ of the state. ν (3. [J i .5 Representations of the Poincar´ group e where pµ is a set of four arbitrary real numbers. 3. For instance. because one then sees that by letting c → ∞ the commutation relations of the Galilei group. (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. P j ] = [P i . P j ] = i ǫijk P k .46) [K i .42) (to decide on where the inverse is. that commute e among themselves. Inﬁnitesimally. (3. They just state that M µν transforms as a tensor with two Lorentz indices. [K i . Λµν pν ⇒ ωµν pν ⇒ ωµν = = = pµ + ω µν pν = pµ 0 ǫµνρσ pρ sσ . a) = Λµν P ν or U † (Λ. In order to label the states in an irreducible representation we construct a maximal set of commuting operators. The generators of the translations are the (momentum) operators P µ . P j ] = i δ ij H/c2 . . Taking any one of the states in an irreducible representation. the tensor M µν in terms of boosts cK i = M i0 and rotations J i = 1 ǫijk M jk .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. H] = [J i .

0.Groups and their representations 22 where s is an arbitrary vector of which the length and the component along pσ are irrelevant. .56) . Wν ] = −i (gµα Wν − gνα Wµ ) . 2 exp (−isµ W µ ) |p. . 2 (3. for which we distinguish the following cases p 2 = m2 > 0 p2 = 0 pµ ≡ 0 p 2 = m2 > 0 p2 = 0 p2 < 0 p0 > 0 p0 > 0 p0 < 0 p0 < 0 Only the ﬁrst two cases correspond to physical states.49) (3.54) This will enable us to pick a suitable commuting ’spin’ operator.53) (3. . Given the momentum four vector pµ we choose a tetrade consisting of three orthogonal spacelike unit vectors ni (p).51) The following properties follow from the fact that W µ is a four-vector by construction. . These e operators are the Casimir operators of the algebra and can be used to deﬁne the representations of ↑ P+ . = = i exp − ǫµνρσ M µν pρ sσ |p. .50) where the Pauli-Lubanski operators W µ are given by 1 Wµ = − ǫµνρσ M νρ P σ . . Pν ] = [Wµ . . the third case represents the vacuum.52) (3. Wα ] = [Wµ . From the algebra of the generators one ﬁnds that P 2 = Pµ P µ and W 2 = Wµ W µ (3. i ǫµνρσ W ρ P σ . (3. of which the components generate the little group of pµ or (for the third of the following relations) from explicit calculation [Mµν . Massive particles: p2 = M 2 > 0. 0.55) commute with all generators and therefore are invariants under Poincar´ transformations. . M ǫijk . . satisfying gµν nµ (p)nν (p) = j i nµ (p) pµ i ǫµνρσ pµ nν nρ nσ i j k = = −δij . while the others have no physical signiﬁcance. which can thus be chosen spacelike. The elements of the little group thus are U (Λ(p)) = = and U (Λ(p))|p. p0 > 0. (3. i exp − ωµν M µν 2 i exp − ǫµνρσ M µν pρ sσ 2 (3.

n1 (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. The explicit construction of the wave function can be done using the D-functions (analogous as the rotation functions). (0. More generally (fully covariantly) one can proceed by deﬁning S i (p) ≡ W i (p)/M and obtain [S i (p). 0. W i (p) = −W · ni (p)). . Together with the four momentum states thus can be labeled as |M. Having made a covariant decomposition.e. 0). S j (p)] = −i nµ (p)nν (p) j i ǫµνρσ W ρ P σ = i ǫijk S k (p). M J) with components W i = (M/2)ǫijk M jk . it is suﬃcient to choose a particular frame to investigate the coeﬃcients in Eq. M2 (3.65) . 0. (3. 2 . 0. 0. The commutation relations [W i (p). Noting that S i (p) i 2 = 1 W µ (p)W ν (p) M2 niµ (p)niν (p). ms . 1.64) such that in an arbitrary frame where the four vectors p. 1.e. p0 > 0. i (3. 0). 3.61) one sees that S i (p) i =− W2 .58) i. 1.59) (3. (3. s. (1.60) and using the completeness relation nµ (p)nν (p) = − g µν − i i 2 i pµ pν M2 . We will not do this but construct them as solutions of a wave equation to be discussed explicitly in chapter 4 Massless particles: p2 = M 2 = 0.57) (i.62) 1 Thus W 2 has the eigenvalues −M 2 s(s + 1) with s = 0. 0). 0. −1) n1 (p0 ) ≡ n2 (p0 ) ≡ s(p0 ) ≡ (3. In this case a set of four independent vectors is chosen starting with a reference frame in which pµ = pµ 0 in the following way: p0 = (p0 .63) where E = p2 + M 2 and ms is the z-component of the spin −s ≤ ms ≤ +s (in steps of one). |p3 |).. . In that case the vectors ni (p) are just the space directions and W = (0. 1 2 (3. . W j (p)] = i M ǫijk W k (p). show that W /M can be interpreted as the intrinsic spin. 0.Groups and their representations 23 We can write W µ (p) = 3 W i (p) nµ (p) i i=1 (3. the S i (p) form the generators of an SU (2) subgroup that belongs to the maximal set of commuting operators. p.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. M2 (3. (0.

A massless particle. 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). This algebra is (for instance) isomorphic to the one generated by rotations and translations in a 2-dimensional Euclidean plane (see exercises). |p.71) which is called the helicity.68) (3. W 1 (p)] [λ(p). W 2 (p)] [λ(p). |p| (3. 1 2 24 (3. We don’t know of any physical relevance.13. W µ (p) = λ(p) P µ . P µ . which can take any integer or half-integer value. W 2 (p)] = = = i M 2 λ(p). thus. σ · RA a ≡ A(σ · a)A−1 ˆ (valid for any vector a). The algebra of W 1 . where A = exp(i ϕ · σ/2) = exp(iϕ σ · n/2). 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. Wν ].Groups and their representations The above does actually include the massive case. a)|0 = |0 . The vector W µ (p) now can be expanded W µ (p) = W 1 (p) nµ (p) + W 2 (p) nµ (p) + λ(p) pµ + W s (p) sµ (p).66) where W · p = 0 implies that W s (p) = −(M 2 /p · s) λ(p). 3. 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 . i W 2 (p). W 2 (3.67) (3. Thus we have P 2 .1 (a) Derive from the relation that deﬁnes the mapping of SU (2) matrices A into SO(3) matrices RA in Eq. But if M = 0 one has a diﬀerent algebra (Eq. (3. is characterized by its momentum and the helicity.70) as a commuting set with zero eigenvalues for P 2 and W 2 . This state is in physical applications nondegenerate and is denoted by |0 . (3.67 has a vanishing right hand side). and gives [W 1 (p). −i W 1 (p). Thus 2 λ(p) = J ·p . 3. Note that angular momentum does not contribute to helicity as L · p = 0. The vacuum: pµ = 0.72) One has P µ |0 = W µ |0 = 0 and the state is invariant under Lorentz transformations. λ . W 2 and λ is derived from the general commutation relations for [Wµ . The eigenvalues of W 1 and W 2 (in that case corresponding to the translations) thus can take any continuous values. implying continuous values for the spin (actually for W i = M S i ). Exercises Exercise 3. however.69) We reproduce the algebra for the massive case with W 3 (p) = M λ(p). U (Λ.73) (3.

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

rj ] = [pi . you might just check relations involving J or K by taking some (relevant) explicit components. that’s why many-particle non-relativistic quantum mechanics is ’easy’.7 (optional) One might wonder if it is actually possible to write down a set of operators that generate the Poincar´ e transformations. = r × p + s. and U (u) p U −1 (u) = p − m u.Groups and their representations 26 Exercise 3. Exercise 3. f (p)] = i ∇p f (p). i. ∂t Show that boost invariance for a free-moving particle (the commutation relations under (a) being valid). show ﬁrst the operator identity [r. p×s 1 (rH + Hr) − t p + . if you don’t want to do this in general.5 In classical mechanics (Gallilean invariance) boosts describe invariance of physical laws when viewed from a moving frame. there is a one to one mapping) the algebra in Eqs 3. = 2c2 H + mc2 satisfy the commutation relations of the Poincar´ group if the position. (c) Give the unitary operator for boosts. the others vanish. (a) In classical mechanics. Use a 3-dimensional matrix representation or consider the generators of rotations and translations in the space of functions on the (x-y) plane. boost invariance leads to a conserved quantity. From either of these. Do this by showing that the set of operators.e. pj ] = iδ ij and [si . sj ] = iǫijk sk .e. in particular those with the momentum operator pj and the Hamiltonian H = mc2 + p2 /2m.e. derive the commutation relations and show that the algebra is isomorphic to (i. although the presence of an interaction term V (r1 . the operator U (u) that gives U (u) r U −1 (u) = r − u t. pj ] = [ri . momentum and spin operators e satisfy the canonical commutation relations. H P J K = p 2 c 2 + m2 c4 . consistent with the (canonical) commutation relations of a quantum theory. indeed assures d K /dt = 0. For a single free particle or for the free-moving center of mass system that is the quantity K = mr − tp. This is possible for a single particle.67 to 3.6 Consider rotations and translations in a plane (two-dimensional Euclidean space). Hint: for the Hamiltonian.. but it can be done. Determine the commutation relations for the corresponding operators Ki in quantum mechanics. i. H] + i dt ∂O . sj ] = [pi . These do not matter in the non-relativistic limit (c → ∞). r2 ) inevitably leads to interaction terms in the boost operators. Comment: extending this to more particles is a highly non-trivial procedure. Exercise 3. they describe r −→ r′ = r − u t while t′ = t. (b) In quantum mechanics one has the Ehrenfest relation i dO = [O. sj ] = 0. = p.69 of the Pauli-Lubanski operators for massless particles. [ri . although that in this case is a relation that is found already from the Ehrenfest relation for O = ri . . [ri .

2 1 (J − i K). The group SL(2. Aj ] = [B i . C): Type I (denoted M ): Type II (denoted M ): 27 J= σ σ . similarly as the homeomorphism between SO(3) and SU (2). It is simply connected and forms the covering group of L↑ . C). where we restrict (for ﬁxed n) the parameters 0 ≤ ϕ ≤ 2π and 0 ≤ η < ∞. They precisely correspond to the two types of representations that we have seen before. B j ] = [Ai . C) is the group of complex 2 × 2 matrices with determinant one. M = exp i ϕ·σ 2 1 exp ± η · σ 2 = U (ϕ) H(η) . j ′ ). 0) Type II : (0.6) (4. 0. The matrices in SL(2.5) = 1 (J + i K). becoming the deﬁning representations of SL(2.1 The Lorentz group and SL(2. 2 2 σ σ J = . j) K = −i J (B = 0).3) (4. With this choice of parameter-spaces the plus and minus signs are actually relevant.9) (4.Chapter 4 The Dirac equation 4. K = i J (A = 0).7) From the considerations above. C) can be written as a product of a unitary + matrix U and a hermitean matrix H.4) (4.10) . respectively.2) This shows that the Lie algebra of the Lorentz group is identical to that of SU (2) ⊗ SU (2). B j ] = i ǫijk Ak .1) (4. it also follows directly that the Lorentz group is homeomorphic with the group SL(2. K = −i . (4. 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 . 2 2 (4. (j. (4. Special cases are the following representations: Type I : (j. They will be labeled by two angular momenta corresponding to the A and B groups. U (ϕ) H(η) (4. This tells us how to ﬁnd the representations of the group.8) ˆ ˆ ˆ with η = η n and ϕ = ϕn. K = +i . i ǫijk B k . 2 (4.

σ µ ≡ (1. ˜ The sets of four operators deﬁned by deﬁnitions σ µ ≡ (1. H ≡ (H † )−1 = H −1 ). one can use a boost to the frame with momentum ˆ ˆ p. such that M = SM S −1 . σ).12) ξ → U ξ.15) (4. however.11) Considering ξ and η as spin states in the rest-frame. thus. C) or L↑ . η → U η. i.18) As already discussed for SU (2). They are connected via Λ(M ∗ ) = Λ−1 (ǫ)Λ(M )Λ(ǫ) within L↑ because iσ 2 = U (πˆ ) showing that y + ↑ ∗ ǫ ∈ SL(2. cannot be connected by a Lorentz transformation belonging to L↑ .1). (4. (ǫ∗ = ǫ. C). (4. One has + −1 Λ(M ∗ ) = I2 Λ(M ) I2 . Note that M ∗ = ǫ† M ǫ has its equivalent for the Lorentz transformations Λ(M ) and Λ(M ∗ ). ˜ Non-equivalence of M and M One might wonder if type I and type II representations are not equivalent. The Lorentz transformations Λ(M ) and Λ(M ).14) satisfying Tr(σ µ σ ν ) = −2 gµν and Tr(σ µ σ ν ) = 2gµν = 2 δ µν (the matrices. ˜ (4. The following relations for + rotations and boosts are useful.19) .e. ˜ = σ µ Λ B aµ . This shows that M ∗ ≃ M . Λ(ǫ) ∈ L+ . thus. while ±ǫη ∗ transforms as a type-I + spinor. 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).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. (4. (4. U ≡ (U † )−1 = U ) U (ϕ) = exp(i ϕ · σ/2) but diﬀerently under hermitean boosts (H † = H. −σ). U σ µ aµ U ˜ −1 H σ µ aµ H −1 = σ µ ΛB aµ H σ µ aµ H ˜ −1 (4. C) and. η → Hη. Also useful is the relation H 2 (p) = σ µ pµ /M = (E + σ · p)/M . ¯ ¯ H ∗ = ǫ† Hǫ = ǫ−1 Hǫ. Eqs 4. (4.13) 2M (E + M ) (exercise 4. U σ µ aµ U −1 = σ µ ΛR aµ . namely H(η) = exp(η · σ/2). ξ −→ Hξ. With E = M γ = M cosh(η) and p = M βγ n = M sinh(η) n. there exists the matrix ǫ = iσ 2 .16) = σ µ Λ R aµ . H(η) = exp(−η · σ/2). ǫ−1 = ǫ† = ǫT = −ǫ). One deﬁnes spinors ξ and η transforming similarly under unitary rotations (U † = U −1 . In fact. C). if there does not exist a unitary matrix S. 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. are not covariant!) ˜ can be used to establish the homeomorphism between L↑ and SL(2.18 show that ±ǫξ ∗ transforms as a type-II (η) spinor. This is not true for the two-component spinors in SL(2.The Dirac equation 28 Let us investigate the deﬁning (two-dimensional) representations of SL(2.

m) = η(0.17. For the spin 1/2 representations of the Poincar´ group including parity we. but rather the operators W i (p) should be used. ) −→ ( . Within the Lorentz group.g. Examples (with between brackets the Euclidean assigments) are r t p H J λ(p) K −→ −λ(p) −→ −K −→ H −→ J −→ t (scalar). m) = χm uL H(p) 0 . must combine the repe resentations. = (4. 0) are inequivalent. etc.1.e. but by a transformation belonging ↑ ↑ to the class P− . A parity transformation changes aµ into aµ . 4. (scalar). m) = χm . −a). 0) of SL(2. 4. The representing member of the class P− is the parity or space inversion operator Is . we obtain for the two components of u which we will refer to as chiral right and chiral left components. C) are suitable for representing spin 1/2 particles.23) u(p.The Dirac equation 29 where The matrix I2 does not belong to L↑ . C). a) and ˜ a = (a0 . Thus M ∗ and M are not equivalent within SL(2. 0). e. 1 ) and ( 1 .21) In nature parity turns (often) out to be a good quantum number for elementary particle states. i. they cannot be connected by a unitary 2 2 transformation. 0 χm uR H(p) . 0) (via operators J and K) one sees that under parity 2 2 1 1 (0.e. the aij elements of a tensor will not change sign. the boost in Eq. where a = (a0 . Taking M (Λ) = H(p).20) 4.22) u= η η M (Λ) 0 where M (Λ) = ǫM ∗ (Λ)ǫ−1 with ǫ given in Eq. and the angular momentum in the rest frame is what we are familiar with as the spin of a particle. We also have seen that the representations (0. (4. consider the four component spinor that transforms under a Lorentz transformation as 0 ξ ξ . 2 ) and ( 2 . that have the correct commutation relations. i.2 Spin 1/2 representations of the Lorentz group 1 1 Both representations (0. (vector). −→ −p (vector). we have seen that in the rest frame W i (p)/M → J i . the well-known two-component spinor for a spin 1/2 particle. therefore. From the deﬁnition of the representations (0. (pseudoscalar). −→ M (Λ) (4. 2 2 (4. 1 ) and ( 1 . −→ −r (vector). The i angular momentum operators J are represented by σ i /2. generators. For a particle at rest only angular momentum is important and we can choose ξ(0. This provides the easiest way of seeing ˜ what is happening.13. The behavior is the same for classical quantities. they can be connected. (axial vector). It has the same eﬀect as lowering the indices. as we have seen in section 4. + − 1 0 0 1 I2 = 0 0 0 0 0 0 −1 0 0 0 0 1 . ǫµνρσ will change sign. but to L↑ . Although these operators cannot be used to label the representations.

= (4.27 is known as the Weyl representation. p (4. We will discuss another explicit realization of this algebra in the next section. (4.30) γ µ γ ν ∂µ ∂ν + M 2 ψ(x) = 0.27) where γ µ are 4 × 4 matrices called the Dirac matrices1 . The general requirements for the γ matrices are thus easily obtained. (iγ µ ∂µ − M ) ψ(x) = 0. 0 H 2 (p) uR uR . γ } ∂µ ∂ν + M 2 ψ(x) = 0. The ﬁrst indeed is solved. = H −1 (p). 2 (4.26) −2 uL uL H (p) 0 or explicitly in the socalled Weyl representation −M E+σ·p E−σ·p −M uR uL = 0. (4.7. The explicit realization appearing in Eq. (4. From this one obtains the Cliﬀord algebra for the Dirac matrices. Since ∂µ ∂ν is symmetric. {γ µ . this can be rewritten 1 µ ν {γ .The Dirac equation 30 with H(p) = ¯ H(p) = 2M (E + M ) E+M −σ·p 2M (E + M ) E+M +σ·p . one must require that for ψ(x) the Klein-Gordon relation 2 + M 2 ψ(x) = 0 is valid for each component separately). (4.32) suppressing on the RHS the identity matrix in Dirac space.29) which is an explicit realization of the (momentum space) Dirac equation. 4. To see this consider the equation for the hermitean † † conjugate spinor ψ † (noting that γ0 = γ0 and γi = −γi ).33) deﬁne for a four vector a the contraction a = aµ γµ / . . which in general is a linear equation in pµ . We ﬁrst want to investigate if the Dirac equation solves the problem with the negative densities and negative energies.28) which is a covariant (linear) ﬁrst order diﬀerential equation. γ ν } = 2 g µν . Applying (iγ µ ∂µ + M ) from the left gives (4. It is of a form that we also played with in Exercise 2. (pµ γ µ − M ) u(p) ≡ (/ − M ) u(p) = 0.25) It is straightforward to eliminate χm and obtain the following constraint on the components of u. ψ † −iγ 0 ∂0 +iγ i ∂i −M 1 We ← ← = 0. As in section 2 we can use Pµ = i∂µ as a representation for the momenta (translation operators) in function space. This leads to the Dirac equation for ψ(x) = u(p)e−i p·x in coordinate space.24) (4. (4.31) To achieve also that the energy-momentum relation p2 = M 2 is satisﬁed for u(p).

We will see how this is properly implemented (also for bosons!) when quantizing ﬁelds. one sees that there are two positive and two negative eigenvalues. E = +M (twice) and E = −M (twice). . From the equations for ψ and ψ one immediately sees that j µ = ψγ µ ψ is a conserved probability current (exercise!). . remains. Relying on the Pauli exclusion principle for spin 1/2 particles.3 General representations of γ matrices and Dirac spinors {γ µ . (4. the one of the negative energy states. The density j 0 = ψγ 0 ψ = ψ † ψ (4. opposite charge).g.37) Using the explicit form of γ in the Weyl representation (see Eq. but with properties such that it can be annihilated by the particle (e. From the vacuum a particle can be removed.The Dirac equation 31 E empty levels M -M filled levels E particle M -M hole Figure 4. γ ν } = 2 g µν . 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.34) for the adjoint spinor ψ ≡ ψ † γ0 .35) is indeed positive and j 0 can serve as a probability density. This can most easily be seen in the particle rest frame. This hole forms an antiparticle with the same mass.1: The Dirac sea of negative energy states and antiparticles.27).39) use ρi ⊗ σj with both ρ and σ being the standard 2 × 2 Pauli matrices. (4. This problem was solved by Dirac through the introduction of a negative energy sea. (4. γ =ρ ⊗1= γ = iρ ⊗ σ = 0 −1 −σ k 0 2 We (4.38) The general algebra for the Dirac matrices is Two often used explicit representations are the following2 : The standard representation: σk 1 0 0 0 3 k 2 k . 4. The second problem encountered before.36) (4. The Dirac sea forms the vacuum. where γ 0 p0 ψ = M ψ 0 −→ Eψ = M γ 0 ψ. Multiplying with γ 0 on the right and pulling it through one ﬁnds a covariant equation ψ(x) iγ µ ∂µ +M ← = 0.

Lorentz invariance The Lorentz transformations can also be written in terms of Dirac matrices.R.44) and an adjoint spinor deﬁned by ψ = ψ † γ0 . For instance in the Weyl representation (but valid generally). 0 (4.3) .10 are 1 σ3 0 i J 3 = M 12 = = [γ 1 .48) are projection operators. that project out chiral right/left states. = S(γµ )W. 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 2 ].43) S=√ . For example. = ρ ⊗ 1 = 1 0 i ǫµνρσ µ ν ρ σ γ γ γ γ . 0 γ k = −iρ2 ⊗ σ k = k σ −σ k .9 and 4. 0 σ3 2 4 1 −iσ 3 0 = − i [γ 3 .47) (4. 2 (4. γ 0 ]. =ρ ⊗1= 0 −1 3 (4. 1 −1 2 For all representations one has † γµ = γ0 γµ γ0 . 4! (4.45) (γ5 )W. γ σ ]. S −1 . while L−1 γ µ L = Λµ γ ν .R. −→ Sψ.42) We note that the explicit matrix taking us from the Weyl representation to the standard representation.R. 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 . which in the case of the Weyl representation are just the upper and lower components.40) (4.The Dirac equation 32 The Weyl (or chiral) representation: 0 1 .46) (4. (γµ )S. is 1 1 1 (4. −→ Sγµ S −1 .R. ν The latter assures Lorentz invariance of the Dirac equation (see Exercise 4. it is easy to see that PR/L = 1 (1 ± γ5 ) 2 1 0 . the rotation and boost generators in Weyl representation in Eqs 4.41) (4. (γ5 )S.49) i 1 with −i S ρσ = 2 σ ρσ = − 4 [γ ρ . γ µ } = 0 and explicitly one has 0 1 1 . γ 0 = ρ1 ⊗ 1 = 1 0 γµ ψ Diﬀerent representations can be related to each other by unitary transformations. We note that ψ → Lψ and ψ → ψ L−1 . (4.

(C)W. that brings ψ → ψ 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 any representation a matrix C exist. usually to be taken unity. in Weyl representation we ﬁnd in Dirac space T C ξ −ǫ η ∗ .53) (4.R. called charge conjugation. (C)S.58) where ηc is an arbitrary (unobservable) phase. is a left-handed spinor.6). ψ= (4. 0 −ǫ 0 0 −ǫ 0 . x = (t. One has ψ c = −ψ T C −1 . parity.54) e. Some properties of C are C −1 = C † and C T = −C. C is real and one has C −1 = C † = C T = −C c and [C. for instance. The latter implies that the conjugate of a right-handed spinor.) and all representations connected via a real (up to an overall phase) matrix S. (or W. −x) and via a few manipulations obtain again the ˜ Dirac equation (i/ − M ) ψ p (x) = 0. such that T Cγµ C −1 = −γµ . ∂ (4. T (4.57) (4. which is again a solution of the Dirac equation.56) (i/ − M ) ψ c (x) = 0. As with parity we look for a transformation. reverse the charge of the particle. γ5 ] = 0.The Dirac equation 33 Parity There are a number of symmetries in the Dirac equation. ψR . ψ= (4. = 0 ǫ iσ 2 (4. It is easy to convince oneself that if ψ(x) is a solution of the Dirac equation.R.59) −→ ψ c = C ψ = η ǫ ξ∗ .R.R. but it does. ∂ (4.55) (4.51) The existence of positive and negative energy solutions implies another symmetry in the Dirac equation.50) Charge conjugation but with ψ p (x) ≡ ηp γ0 ψ(˜) (Exercise 4. Note that we have (as expected) explicitly in Weyl x representation in Dirac space P ξ −→ ψ p = γ 0 ψ = η . T (4. Explicitly. In S. (i/ − M ) ψ(x) = 0. 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. This symmetry does not change the spin 1/2 nature.g.g. ∂ with the solution ψ c (x) = ηc Cψ (x). −iσ 2 0 −ǫ = .52) η ξ one can apply space inversion. x) → x = (t. e.

s)v(p. s). s′ ) = 2M δss′ . s) = u(p.The Dirac equation 34 4.61) σ·p −(Ep + M ) There are also two negative energy solutions. ¯ ¯ † ′ u (p.1). s) + v(k. the spinors satisfy C uT (p. ¯ ′ ′ Explicit solutions in the standard representation are χs . s)¯(p. with E = Ep = + p2 + M 2 .60) ψ(x) = 0. 2M 2M (4. s′ ) = −2M δss′ . The explicit Dirac equation in the standard representation reads ∂ iσ · ∇ i ∂t − M (4. s)v(p. p (4. In that frame is a spacelike unit vector sµ = (0. s) v = . ⇔ (/ − M ) u(p) = 0. the best representation to describe particles at rest is the standard representation.g. where u satisﬁes −σ · p Ep − M u(p) = 0. s) = v(p. s) . s) ei k·x d∗ (k. s) u = . s ) = 0.62) where χs are two independent (s = ±) two-component spinors. p −σ · p (Ep − M ) Ep + M σ ·p ¯ Ep +M χs χs ¯ For a free massive particle.70) = Ps = ˆ 0 1∓σ·s 2 2 . s) = u(p. s)v(p. s)u(p.64) (4. s)u(p.. s)¯(p. Choosing χ = −ǫχ∗ (the equivalent spin ¯ 1/2 conjugate representation). s). It is easy to check that 1 1±σ·s 1 ± γ5 / s ˆ 0 (4. The ¯ ¯ solutions are normalized to u(p. be obtained by a Lorentz transformation.69) = − s In order to project out spin states. the spin polarization vector in the rest frame is the starting ˆ point. s) e−i k·x b(k. v (p. in which γ 0 is diagonal (see discussion of negative energy states in section 4. s)u(p. where v satisﬁes σ·p −(Ep + M ) v(p) = 0. Note that the spinors in the negative energy modes (antiparticles) could be two diﬀerent spinors. ψ(x) = v ±s (p) ei p·x . ⇔ (/ + M ) v(p) = 0. s′ ) = 2Ep δss′ .65) (4. v(p.4 Plane wave solutions Looking for positive energy solutions ∝ exp(−iEt) one ﬁnds two solutions. s) and C v T (p. ψ(x) = u±s (p) e−i p·x . (2π)3 2Ek (4. (4.67) It is straightforward to ﬁnd projection operators for the positive and negative energy states P+ P− = s p /+M u(p. s) = Ep + M σ ·p Ep +M χs . In an arbitrary frame one has s · p = 0 and s(p) can e. ¯ (4.68) (4. ∂ −iσ · ∇ −i ∂t − M (4.66) An arbitrary spin 1/2 ﬁeld can be expanded in the independent solutions. s ) = v (p. 2M 2M −/ + M p v(p. s ) = v † (p.63) u(p. 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.

Therefore. Explicit examples of spinors are useful to illustrate spin eigenstates.48 replace the spin projection operators which are not deﬁned (by lack of a rest frame). . By writing the helicity states in Weyl representation it is easy to project out righthanded (upper components) and lefthanded (lower components). −p3 −p1 − i p2 1 v(p. p3 E +M E+M 1 2 3 p + ip −p 1 2 p − ip 3 1 −p v(p. helicity states. u(p. In principle massless fermions can be described by twocomponent spinors. λ = +1/2) = v(p. For instance with the z-axis as spin quantization axis. while the negative helicity (λ = −1/2) is in essence lefthanded. λ) = v(p. chirality. √ 0 u(p. One ﬁnds for the helicity states in Weyl representation: √ √ 0√ E +M + E −M √ 1 √ 1 0√ 1 . v(p. λ = − 1 ) = √ E + M − E − M . +1/2) = √ u(p.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 ). (4. p ψR/L ≡ PR/L ψ are independent solutions. one has in standard representation: 0 E+M E+M 1 1 0 . λ = − 1 ) = √ √ v(p.73) Also for helicity states C uT (p. λ = −1/2) = √ − E+M √ 0 0 E+M (4. This means that in the solution space for massless fermions the chirality states. λ) = u(p.71) p − i p2 . Note that for solutions of the massless Dirac equation /ψ = 0.72) ˆ Helicity states (p along z ) in Standard representation are: √ √ 0 E+M E +M . λ) and C uT (p. 2 2 0 √ 2 √ 2 − E+M − E−M 0 . λ = +1/2) = E−M √ 0 − E −M 0 √ √ 0 − E−M − E −M 0 . 1 u(p. −1/2) = √ (4. λ = + 1 ) = √ + E +M − E −M . λ). λ = −1/2) = u(p. The ratio of the components is (directly or inversely) proportional to √ √ E+M − E−M 2M M ≪E M √ √ √ . = √ =⇒ 2E E+M + E−M ( E + M + E − M )2 which vanishes in the ultra-relativistic limit E ≫ M or in the massless case. γ5 /ψ = 0 but also p p p /γ5 ψ = −γ5 /ψ = 0. λ = + 2 ) = √ E +M − E −M 2 0√ 2 2 √ 0 E +M + E −M √ √ 0√ √ − E +M − E −M 1 1 E+M − E−M 0 √ . We note that for high energy ¯ ¯ the positive helicity (λ = +1/2) is in essence righthanded. −1/2) = √ E + M −(E + M ) 0 . 4. u(p. (4.74) . +1/2) = √ 0 E +M E+M . etc. The chirality projection operators in Eq. v(p.

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

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

Do you understand why? The calculation of the life-time is completed in chapter 11. the lepton and neutrino have spin 1/2 and are ν described by momentum-space spinors.8 × 10−6 . |M (π − → µ− νµ )|2 ¯ Actually if the lepton would be massless.5 and M → ∞. you need to determine the corresponding conﬁnement condition for ψ. mµ = 105. ν p Note: how such parametrizations work will be explained in chapter 8. 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. is a stationary solution of the (free) Dirac equation. (b) Next. Show that the ratio is |M (π − → e− νe )|2 ¯ ≈ 54. i. For such a cavity nµ = (0. Calculate for now just |M |2 . we want to conﬁne this solution to a spherical cavity with radius R. n (c) Apply the conﬁnement condition to ﬁnd the value of k for the lowest eigenmode in the spherical ˆ cavity. Excercise 4. t) = N k ˆ i E+M σ · r j1 (kr)χm −i Et e .66 MeV. Calculate the value of k for M R = 0. Note: use ﬁrst the appropriate equation (which?) that is obeyed by each individual component of the Dirac equation to show that the upper two components are allowed solutions. Note: the outcome shows that the decay into muons is much more likely that the decay into electrons (mπ = 139. 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. r). .57 MeV. n It turns out that there is actually a whole family of conditions that provide conﬁnement. me = 0. The pion has spin 0. Note: To prove this.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 ). 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 ).511 MeV). M R = 1. the decay would even be forbidden. there is no current ﬂowing through the surface of the cavity. The conditions for ψ and ψ imply that nµ j µ = ψ/ ψ = 0. We will not pursue this chiral freedom at this point and take α = 0. which is the most important factor needed to ﬁnd the decay width (inverse lifetime).e. namely i / ψ = eiαγ5 ψ.7 (a) Show that (in standard representation) j0 (kr)χm ψ(x. Show that this n condition is suﬃcient to guarantee that nµ j µ = 0.The Dirac equation 38 Exercise 4.

M2 (5. These can describe spin 1 states in the rest-frame. just as the two-component Dirac spinors did for spin 1/2.2 we have discussed the vectors ǫλ (with λ = 0 and ±1) as basis states of the spin 1 representation. These are solutions of the Klein-Gordon equation with in addition a condition. In the restframe. λ) . A real vector ﬁeld Vµ (x) then can be expanded in modes as Vµ (x) = λ=0.Chapter 5 Vector ﬁelds and Maxwell equations 5.± 5.4) λ=0. µ (2π)3 2E µ (5. λ) + ǫ(λ)∗ (k) eik·x c∗ (k.1) The Lorentz condition implies that k µ ǫµ (k) = 0. The covariant generalization is to describe them with Vµ (x) = ǫµ (k) e±ik·x .7) .± (λ) d3 k ǫ(λ) (k) e−ik·x c(k.1 Fields for spin 1 In section 3. ±(1) are obtained by boosting the restframe vectors. there are then indeed three independent possibilities. where k = k0 = (M. 39 (5.2) (5. A). ǫλ ). (5. 0). In an arbitrary frame ǫµ (k) with λ = 0. namely ǫµ (k0 ) = (0. one has ǫ(λ) (k) ǫ(λ)∗ (k) = −gµν + µ ν kµ kν . (2 + M 2 )Vµ = 0.3) Since the vectors ǫµ (k) together with the momentum k µ form a complete set of four-vectors.5) −E B2 . The vectorﬁeld is therefore (λ) suited to describe spin 1.6) where Aµ is the four-vector potential (5. −B 1 0 3 With the electromagnetic ﬁeld tensor (5. with ∂µ V µ = 0 (Lorentz condition).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.

The equations of motion for the ﬁelds Aµ become 2Aµ − ∂µ (∂ν Aν ) = jµ .5. Therefore if k µ = (|k|. ˜ x Note that via parity transformation one obtains another solution.13) Hence. 40 (5. 2Aµ = 0. for M = 0. (5. (5. 0. Taking the divergence of this equiation.17) (radiation. ∂µ Aµ = 0. while the fact µ C ∗ that the Minkowski space is real implies that Aµ (x) = Aµ (x) = Aµ (x). −i. it does not eliminate the freedom of gauge transformations but they are now restricted to Aµ → Aµ + ∂µ χ with 2χ = 0.15 the d’Alembertian is replaced by 2 + M 2 .11) This equation is not aﬀected by a gauge transformation.8) (5. (5.e. Fµν −→ Fµν + (∂µ ∂ν χ − ∂ν ∂µ χ) = Fµν . Furthermore. if ∂µ Aµ = 0 for large times |t|. The gauge freedom can ﬁrst of all be used to impose the Lorentz condition for the electromagnetic ﬁeld. Note that the incorporation of the Lorentz condition can also be done for the case of the massive vector ﬁeld. In that case one also has ∇·A = 0 or ki ǫi (k) = 0. (5. there is still a gauge freedom left (Aµ → Aµ + ∂µ χ with 2χ = 0). in accordance with what we found in section 3.16) (5. only two independent vectors remain ǫ(±) µ (k) = (0. The Lorenz condition can actually be incorporated by changing the equations of motion to 2Aµ + (λ − 1)∂µ (∂ν Aν ) = jµ . 0. |k|). AP (x) = Aµ (˜).18) corresponding to the helicity states λ = ±1 of the massless photon ﬁeld. it will vanish at all times. transverse or Coulomb gauge). i. As remarked above.9) 1 ˜ where F µν ≡ − 2 ǫµνρσ Fρσ is the tensor where E → B and B → −E. One possible choice is the gauge A0 = 0 (5. . one ﬁnds for a conserved current and λ = 0 2∂µ Aµ = 0. 0).14) of which the solutions give the Li´nard-Wiechert potentials.10) the electric and magnetic ﬁelds are unchanged. ∓1. shows that the electromagnetic ﬁelds correspond to the case of a massless spin 1 ﬁeld. j) the Maxwell equations read ∂µ F µν = j ν . moree over. 5. Although the Lorenz condition is a constraint. Under a gauge transformation Aµ −→ Aµ + ∂µ χ. if in Eq. For electrodynamics one has furthermore the freedom of gauge transformations.15) (5.Vector ﬁelds and Maxwell equations and the (conserved) current j µ = (ρ. The equation in vacuum.12) (5. in which case one has 2Aµ = jµ . (5. the antisymmetry of F µν implies that ˜ ∂µ F µν = 0 or ∂λ Fµν + ∂µ Fνλ + ∂ν Fλµ = 0. It states the absence of magnetic charges and Faraday’s law.

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

Now back to the Aharonov-Bohm experiment. Therefore the gauge transformation must be uniquely deﬁned in the space one is working in.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. minimizing the space occupied by B ﬁelds. 5. If this would be be happening in empty space one would be in trouble. forming a simple (connected) space. it can be obtained from the situation A = 0 by a gauge transformation. Consider a superconductor.21. 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. however. A = ∇χ with χ= BR2 ϕ 2 (5.2: The topology of the space in the Aharonov-Bohm experiment.e. This situation that A can be written as ∇χ is always true when ∇ × A = 0. illustrating also that the group structure of the mapping of U (1) into this space is that of the group of integers Indeed when one considers the A-ﬁeld in the example of the Aharonov-Bohm eﬀect outside the solenoid. The diﬀerence here is that we are working in a space with a ’defect’ (the inﬁnitely long solenoid). giving no observable phase2 . ϕ=2π The function χ in Eq. The observable phase going around in general is ∆δ = e A · dr = e ∇χ · dr = e χ|ϕ=0 . 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. i. the relevant charge turns out to be 2e because the electrons appear in Cooper pairs .21) (ϕ being the azimuthal angle). Therefore space outside the solenoid doesn’t care that χ is multi-valued. has been employed by Dirac in constructing magnetic monopoles in electrodynamics 2 Actually. Below the critical temperature the magnetic ﬁeld is squeezed out of the superconducting material. The AharonovBohm experiment shows a realization of this possibility. In such a space loops with diﬀerent winding number (i. the number of times they go around the defect) cannot be continuously deformed into one another. is multivalued as ϕ = 0 and ϕ = 2π are the same point in space. nonobservable phases ϕ = 2π n. implying that it must be single-valued1 . organizing itself in tiny ﬂux tubes (Abrikosov strings). Another situation in which the topology of space can be used is in the case of superconductors. 1 The occurrence of nontrivial possibilities.e.

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

These equations can be obtained from a lagrangian using the action principle. x) ˙ (6.Chapter 6 Classical lagrangian ﬁeld theory 6.5) with ∆x(t) = δx(t) + x(t) δτ . x) = K − V = ˙ 1 mx2 − V (x). and the total change x′ (t′ ) = x(t) + ∆x(t). (6. recall classical mechanics with the action t2 S= t1 dt L(x. ∂x ˙ (6.3) (6. x (t) = x(t) + δx(t). t1 (6. Dirac ﬁelds (Dirac equation) and massless vector ﬁelds (Maxwell equations) and corresponding to these ﬁelds conserved currents describing the probability and probability current. thus t′ = t + δτ. ˙ 2 (6.1 Euler-Lagrange equations In the previous chapter we have seen the equations for scalar ﬁelds (Klein-Gordon equation).8) 44 . The requirement δS = 0 with ﬁxed boundaries x(t1 ) = x1 and x(t2 ) = ˙ x2 leads to t2 ′ (6.4) 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 .7) The quantity ∂L/∂ x plays a special role and is known as the canonical momentum. d dt ∂L ∂x ˙ = ∂L . ∂x (6. As an example.2) The principle of minimal action looks for a stationary action under variations in the coordinates and time.6) The ﬁrst term leads to the Euler-Lagrange equations.1) and as an example the lagrangian L(x. ˙ p= ∂L .

also indicated by ∂R (a more general volume in four-dimensional space-time with some boundary ∂R can also be considered). indicated with δF [φ]/δφ should pose no problems. ∂µ φ′ .16) The change in variables x → x′ in the integration volume involves a surface variation of the form dσµ L δxµ . gives rise to a conserved quantity (at least if H is invariant!) Q=p dx dλ −H dt dλ .14) (6.Classical lagrangian ﬁeld theory 45 For the lagrangian speciﬁed above this leads directly to Newton’s law p = m¨ = −∂V /∂x. ∂µ φ(x)). δ(∂µ φ) (6. t1 ) and R3 .t1 ) (6. which in classical mechanics vanishes at the boundary) does not play a role. . In general any continuous transformation. ∂µ φ.e. x).9) which is done because the ﬁrst term (multiplying ∆x.12) Here R indicates a space-time volume bounded by (R3 . ∂µ φ(x)) = R d4 x L (φ(x).15) d4 x′ L (φ′ . t1 (6. indicated as x′µ φ (x) or combined φ′ (x′ ) = φ(x) + ∆φ(x). x) and an action t2 S[φ] = t1 dt L = dt d3 x L (φ(x). but which in 3 dimensions includes conserved quantities related to rotations and boosts. ˙ x The second term in Eq. The resulting variation of the action is δS = R µ ′ = xµ + δxµ . x′ ) − d4 x L (φ. t2 ). ˙ (6. ∆x = (dx/dλ)λ=0 δλ and δt = (dt/dλ)λ=0 δλ. R (6. 6. The hamiltonian H is deﬁned by H(p. . . . x) ≡ p x − L . . . (R3 .10) One sees that invariance under time translations requires that H(t1 ) = H(t2 ). . + ∂L ∆x − H δτ ∂x ˙ t2 .11) λ=0 of which p and H are the simplest examples related to space and time translations.17) Furthermore the variations δφ and δ∂µ φ contribute to δS. their derivatives and possibly on the position. λ=0 (6. . ∂R Note for the speciﬁc choice of the surface for constant times t1 and t2 .2. dσµ .6 can be rewritten as δS = . In classical ﬁeld theory one proceeds in complete analogy but using functions depending on space and time (classical ﬁelds. . − d3 x .13) (6. . (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 . i. = φ(x) + δφ(x) (6. with ∆φ(x) = δφ(x) + (∂µ φ)δx . L (φ(x).t2 ) d3 x . = ∂R (R3 . We will come back to it in a bit more formal way in section 9.18) 1 Taking a functional derivative. Consider a lagrangian density L which depends on these functions. think for instance of a temperature or density distribution or of an electromagnetic ﬁeld). ∂µ φ(x). H is a conserved quantity. Variations in the action can come from the coordinates or the ﬁelds.

One easily obtains (2 + M 2 )φ(x) = 0. (2 + M )φ (x) = 0. The integrand of the ﬁrst term must vanish. L = 1 1 ∂µ φ∂ µ φ − M 2 φ2 2 2 1 µ = − φ ∂µ ∂ + M 2 φ. (6.28) where the second line is not symmetric but in the action only diﬀers from the symmetric version by a surface term (partial integration).19) δ(∂µ φ) δφ 6.29) . the second term is irrelevant. 2 ψ = 0. leading to the Euler-Lagrange equations.22) which can be considered as the sum of the lagrangian densities for two real scalar ﬁelds φ1 and φ2 √ with φ = (φ1 + iφ2 )/ 2.26) 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. δL δL ∂µ = .Classical lagrangian ﬁeld theory 46 With for the situation of classical ﬁelds all variations of the ﬁelds and coordinates at the surface vanishing.27) (6. 2 ∗ (6.23) (6. leading to (2 + M 2 )φ(x) = 0. 2 (6. (6. For the complex scalar ﬁeld one conventionally uses L = = ∂µ φ∗ ∂ µ φ − M 2 φ∗ φ −φ∗ ∂µ ∂ µ + M 2 φ.24) (6. (6.21) which diﬀer only by surface terms.20) (6. the Dirac ﬁeld (spin 1/2) and the vector ﬁeld (spin 1) can be found. The scalar ﬁeld It is straightforward to derive the equations of motion for a real scalar ﬁeld φ from the lagrangian densities.2 Lagrangians for spin 0. Using the variations in ψ (in the symmetric form).25) (6. 1/2 and 1 ﬁelds By an appropriate choice of lagrangian density the equations of motion discussed in previous chapters for the scalar ﬁeld (spin 0). δL δ(∂µ ψ) δL δψ one obtains immediately i / −M ∂ → i = − γ µψ 2 i → = ∂ / ψ − M ψ.

(6.30) It is often useful to link to the two-component spinors ξ and η which we started with in chapter 4. Peshkin and Schroeder or Exercise 12.34) (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. Using the twospinors ξ and η. the mass term in the Dirac lagrangian 6. or equivalently separate the ﬁeld into right. that the lagrangian separates into two independent parts for M = 0. L = = ↔ 1 1 c c M ψL ψL + M ∗ ψL ψL ψL i / ψL − ∂ 2 2 ↔ 1 1 M ΥR ΥL + M ∗ ΥL ΥR . but note the factor 1/2 as compared to the Dirac lagrangian.33) We now introduce a ’real’ (four-component) spinor satisfying Υc = Υ with ψL . 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. 2 (6. Vector ﬁeld From the lagrangian density L 1 1 λ 2 = − Fµν F µν + M 2 Vµ V µ − (∂µ V µ ) 4 2 2 1 µ = V (∂ 2 + M 2 )gµν − (1 − λ)∂µ ∂ν V ν . The reasons for Grassmann variables will become clear in the next chapter. ∂ ∂ 2 2 (6.36) (6.g.31 is given by LM (Dirac) = −M ξ † η + η † ξ .g.32) There exists another possibility to write down a mass term with only one kind of ﬁelds. .37) This gives an expression with a mass term that is actually of the same form as the Dirac mass term.7).and lefthanded ones. Υi / Υ − ∂ 4 2 T −ǫ η ∗ c = ΥR .38) (6. We note that of which the left part coincides . which comes because we in essence use ’real’ spinors. (6.39) 2 This term is written down with η being an anticommuting Grassmann number for which (αβ)∗ = β ∗ α∗ = −α∗ β ∗ . In that case one ﬁnds trivially L = ↔ ↔ 1 1 ψR i / ψR + ψL i / ψL − M (ψR ψL + ψL ψR ).35) Since the kinetic term in L separates naturally in left and right parts (or ξ and η). namely2 LM (Majorana) = + (6. ∗ 0 ⇒ Υ ≡ −ǫ η ΥL = ψL = η η ∗ for which η0 = ǫ η0 .Classical lagrangian ﬁeld theory 47 and similarly from the variation with respect to ψ ψ i / +M ∂ ← = 0. (6. The Majorana case is in fact more general. 1 M η † ǫη ∗ − M ∗ η T ǫη . and thus (β ∗ α)∗ = α∗ β = −βα∗ . 2 αβ = −βα. ψL ≡ (ψL )c = C ψL = 0 (6.31) showing e.

ψ(x) → ei eΛ ψ(x) or ∆ψ(x) = i eΛ ψ(x). . ∂µ J µ (x) = 0. δ(∂µ φ) (6.. δ(∂µ φ) For continuous transformations. Q= d3 x J 0 (x.44) δL ∆φ + Θµν (x)δxν . (6. these conserved quantities become operators.18.+ 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 .41) (6. Q(t1 ) = Q(t2 ). t).40) 6. discussed in the next chapter.47) . one can consider the variations at the initial or ﬁnal surface. requires the presence of a conserved quantity.45) J µ (x) = − δ(∂µ φ) dλ λ=0 dλ λ=0 Thus considering ∂R = σ2 . we get (omitting the parameter δλ) dxν dφ δL + Θµν (x) .Classical lagrangian ﬁeld theory 48 where Fµν = ∂µ Vν − ∂ν Vµ .43) F (σ1 ) − F (σ2 ) = dσµ J µ (x) = ∂R R d4 x ∂µ J µ (x). consider U (1) transformations of the Dirac ﬁeld proportional to the charge e. . ∆φ = (dφ/dλ)λ=0 δλ and δxµ = (dxµ /dλ)λ=0 δλ. implying ∂µ F µν + M 2 V ν = 0 and ∂µ V µ = 0. (6. which in a consistent picture precisely generate the symmetries. which is the space-integral over the zero-component of a conserved current.. the presence of a symmetry that leaves the lagrangian invariant.3 Symmetries and conserved (Noether) currents Next. (6.42) Θµν = The variation δS thus can be expressed as δS with J µ (x) = − = δL ∂ ν φ − L g µν . t).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 . δ(∂µ φ) (6. we consider the second term in Eq. + ∂R dσµ dσµ ∂R ≡ where d x . These do not aﬀect the dynamics and will give rise to conserved currents. In this case.σ1 with σ = (R3 . As an example for the classical ﬁeld. (6. (6. 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. 6. In particular when the lagrangian is invariant under symmetries.

From Noether’s theorem one sees that the current Θµν ǫν is conserved. = 0. (ω µν antisymmetric) 1 ωρσ (−i S ρσ )ij φj (x). (6. H = d3 x Θ00 (x). ↔ (6. (6.53) These are the energy and momentum. For quantized ﬁelds this will become the expressions of the hamiltonian and the momentum operators in terms of the ﬁelds.57) For the ﬁelds the behavior under Lorentz transformations has been the subject of previous chapters.50) (6. = −(∂µ φ)δxµ = −ǫµ ∂µ φ(x).4 Space-time symmetries One kind of symmetries that leave the lagrangian invariant are the Poincar´ transformations. e. Therefore there are four conserved currents Θµν (ν labeling the currents!) and four conserved quantities Pν = dσµ Θµν = d3 x Θ0ν . we obtain (since δxµ = 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 .48) For the complex scalar ﬁeld the U (1) transformations leave the lagrangian invariant and lead to the conserved current (see Exercise 6. (6. 6.49) These currents are conserved as discussed already in chapter 2. 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.51) (6.54) (6. (6. Translations µ Under translations (generated by Pop ) we have δxµ ∆φ(x) δφ(x) = ǫµ . ˙ Q = d3 x j 0 (x. In the next section we will see the quantity show up as the charge operator.Classical lagrangian ﬁeld theory From the lagrangian for the Dirac ﬁeld.3) j µ = i e φ∗ ∂µ φ.55) Note that the coordinate transformations can be written in a form similar to that for the ﬁelds (matrices in internal space). 2 (6.g. 1 (6. Summarizing. t) is the conserved charge (Q = 0). The integral over the zero-component. including e the space-time translations and the Lorentz transformations. Lorentz transformations In this case the transformation of the coordinates and ﬁelds are written as δxµ ∆φi (x) = = ω µν xν .

(6. independent of x.3 we have seen global transformations or gauge transformations of the ﬁrst kind. • Dirac ﬁeld ψ: −i Sρσ = −(1/4)[γρ . in which the U (1) transformation involves an angle e Λ. Deﬁning T µν = Θµν − ∂ρ Gρµν . 2 (6. (6.61) A ﬁnal note concernes the symmetry properties of Θµν . (6. the angle of the transformation depends on the space-time point x.58) (6. ∂µ T µν (6.68) (6. −i eΛ(x) ∗ i eΛ(x) (6.6) T µν = T νµ . ∂µ φ(x) + i e ∂µ Λ(x) e (6.67) i. one has a tensor that satisﬁes (exercise 6. however.62) with Gρµν = (H ρµν + H µνρ + H νµρ )/2. • Vector ﬁeld Aµ : −i Sρσ = aρσ .63) µν = ∂µ Θ .e. where H ρµν is the quantity appearing in M ρµν . Any lagrangian containing derivatives. The lagrangians which we have considered sofar are invariant under global gauge transformations and corresponding to this there exist a conserved Noether current. γσ ] This result for the Dirac ﬁeld has been discussed in chapter 3 (see also Exercise 4. φ(x) φ (x) ∂µ φ(x) ∗ → e → e → ei eΛ(x) φ(x). 6.70) . 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 . In some applications (speciﬁcally coupling to gravity) it is advantageous to have an equivalent current that is symmetric. e.3).69) i eΛ(x) φ (x).Classical lagrangian ﬁeld theory 50 • Scalar ﬁeld φ: −i Sρσ = 0.64) (6. φ(x).60) Therefore there are six conserved currents M µρσ labeled by ρ and σ (antisymmetric) and corresponding to it there exist conserved quantities M ρσ = d3 x M 0ρσ .65) M µρσ = T µρ xσ − T µσ xρ . is not invariant under local gauge transformations.59) (6.5 (Abelian) gauge theories φ(x) → ei eΛ φ(x). (6. In general this tensor is not symmetric. Gauge transformations of the second kind or local gauge transformations are transformations of the type φ(x) → ei eΛ(x) φ(x).66) In section 6. (6.g.

Dµ φ(x) ≡ (∂µ + i eAµ (x))φ(x). As an example consider the Dirac lagrangian. 1 L (φ.78) . (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). i L = ψγ µ ∂µ ψ − (∂µ ψ)γ µ ψ − M ψψ. −e j µ Aµ = −e ρφ + e j · A. 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 . (6. The required transformation for Dµ then demands Dµ φ(x) = = (∂µ + i eAµ (x))φ(x) ei eΛ ∂µ + i e(A′ + ∂µ Λ) φ µ → ei eΛ ∂µ φ + i e (∂µ Λ) ei eΛ φ + i eA′ ei eΛ φ µ ≡ ei eΛ (∂µ + i eAµ ) φ. (6. Dµ φ) − Fµν F µν . 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. 4 (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 . D ∂ A (6.77) 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). For this purpose it is necessary to introduce a vector ﬁeld Aµ . ∂µ φ) =⇒ L (φ.73) (6.74) the behavior which we have encountered before as a gauge freedom for massless vector ﬁelds with the (free) lagrangian density L = −(1/4)Fµν F µν . (i/ − M ) ψ = (i/ − e/ − M ) ψ = 0. 2 i ∂ / ψ − M ψ − e/ ψ A 2 giving the Dirac equation in an electromagnetic ﬁeld. The equation of motion for the fermion follow from δL δ(∂µ ψ) δL δψ = = i − γ µ ψ.72) (6.71) Thus the covariant derivative has the correct transformation behavior provided Aµ → Aµ − ∂µ Λ. ∂ 2 4 (6.75) The ﬁeld φ is used here in a general sense standing for any possible ﬁeld.Classical lagrangian ﬁeld theory 51 where the last term spoils gauge invariance.

φ → eiΛ φ.3 Show that the current jµ = i φ∗ ∂µ φ for a complex scalar ﬁeld is connected to a U (1) transformation on the ﬁelds. give the equation which is satisﬁed by ψ c = Cψ . i. ∂µ F µν = j ν . ∂µ F µν = j ν .Classical lagrangian ﬁeld theory 52 For the photon the equations of motion follow from δL δ(∂µ Aν ) δL δAν = = −F µν . (6.79) Exercises Exercise 6. (c) Show that the interaction term is invariant under gauge transformations only if the current jµ is conserved.2 (a) Show that the homogeneous Maxwell equations can be derived from the lagrangian density 1 L = − Fµν F µν .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 .e. where j µ = e ψγ µ ψ. −eψγ ν ψ. Exercise 6. and deduce what is the charge of the antiparticle as compared to a particle. [iγ µ (∂µ − i e Aµ ) − M ]ψ = 0. 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. 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. ↔ Exercise 6. Exercise 6. ∂µ j µ = 0 (Note that the addition of a total derivative to the lagrangian density does not modify the equations of motion). . giving the Maxwell equation coupling to the electromagnetic current.4 Given the Dirac equation for a (negatively charged) spin-1/2 particle in an external electromagnetic T ﬁeld.

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

in our example p = mq. is the Poisson bracket [A. This is nothing else as representing the classical Poisson algebra as operators in the Hilbert space. dq The basic Poisson bracket is actually the one between q and its canonical momentum p. q(q.1 Canonical quantization We will ﬁrst recall the example of classical mechanics for one coordinate q(t). t).2) H(p.4) with a possible realization as operators in the Hilbert space of (coordinate space) wave functions through qop ψ(q) = qψ(q) and pop ψ(q) = −idψ/dq. The bilinear product for the conserved quantities. Quantization of the theory is achieved by imposing canonical commutation relations between q and p. there are conserved ˙ ˙ quantities Q (Eq. 1 qx (t) = 3 d3 x φ(x. q) also considered in the previous chapter. however. (7. starting from the lagrangian L(q. (7. ˙ L(q. ˙ 2 (7. namely [q.3) For variations δO(p.1) The Hamiltonian (also corresponding to a conserved quantity because of time translation invariance) is given by p2 + V (q). dq dp dq dp (7. An immediate generalization for ﬁelds can be obtained by considering them as coordinates. Q]P . dλ Examples are dO = [O. Actually the conserved quantities satisfy a similar algebra as the Lie algebras for continuous symmetries that they are obtained from. Further. p)) = ˙ ˙ 2m where the (canonical) momentum p = ∂L/∂ q. p] = i. p]P . labeled by the position.5) ∆ x ∆3 x 54 . B]P = dB dA dA dB − . p]P = 1. (7.Chapter 7 Quantization of ﬁelds 7. q) = p q − L(q. 6. q) ≈ (dO/dλ)λ=0 δλ under symmetry transformations one has dO = [O.11) found through Noether’s theorem. q) = ˙ 1 mq 2 − V (q). [q. H]P dt and dO = [O.

(7. t). t)] = 0. formally. qx . for the scalar ﬁeld theory. As an example. The hamiltonian then is H= x px qx − L = ˙ = x where H (x) = δL ˙ φ(x) − L (x) = Θ00 (x).12) (7.15) with furthermore the relations [φ(x. t)] = [Π(x. ∆φ. we have L (x) = 1 1 ∂µ φ∂ µ φ − M 2 φ2 2 2 1 1 1 2 (∂0 φ) − (∇φ)2 − M 2 φ2 = 2 2 2 δL = = ∂0 φ. ∂µ φ(x)) = ˙ d3 x L (φ. for symmetry transformations. Π(y. δ∂0 φ) 1 1 1 = Θ00 (x) = (∂0 φ)2 + (∇φ)2 + M 2 φ2 . a) = Rj (Λ−1 ) φj (Λ−1 x − a). py (t)] = i δxy or for the ﬁelds φ(x) and Π(x) the socalled equal time commutation relations [φ(x.Quantization of ﬁelds 55 etc.6) ∆3 x L x (qx . • Canonical commutation relations The quantization of the theory is achieved by imposing the canonical quantization condition [qx (t). a) φi (x)U (Λ. t). • Poincar´ invariance e The full variation of the ﬁelds.11) (7. the ﬁelds φ(x) can be considered as regular operators in the Hilbert space after smearing with a test function f .9) Note that this indeed corresponds to the zero-zero component (Θ00 ) of the conserved energy-momentum stress tensor Θµν . Φ(f ) = d4 x φ(x) f (x). discussed in the section 6. ˙ where Πx (t) is obtained from the continuous ﬁeld Π(x) ≡ δL /δ(∂0 φ). t)] = i δ 3 (x − y). φ. The lagrangian is given by L = = x In order to construct the hamiltonian it is necessary to ﬁnd the canonical momenta. The essential conditions to consistently quantize a theory then are the following.13) Π(x) H (x) For the quantization procedure we can formulate a number of basic axioms of quantum ﬁeld theory. 2 2 2 (7. ˙ δ φ(x) (7. (7.8) (7. (7. t).10) x ∆3 x [Πx qx − L x ] ˙ ∆3 x Hx (7. This implies speciﬁc transformation properties in the Hilbert space for the quantum ﬁelds φ. Π(y. ∇φ) (7. ∂ qx ˙ ∂ qx ˙ (7. albeit with ’sharp’ functions. the discretization procedure above is an explicit example. px (t) = ∂L ∂Lx = ∆3 x = ∆3 x Πx (t). φ(y.14) In fact. Sometimes it is useful to keep in mind that.2.16) . qx+∆x ) . i U † (Λ.7) d3 x L (φ(x).

21) Writing Q and P in terms of creation (a† ) and annihilation (a) operators. 2 2 (7. Pµ ] = [φ (x). 2 i ωµν (S µν )i . [φ(x). Mµν ] = i i∂µ φi (x). In simpliﬁed form the hamiltonian is given by H= 1 2 1 2 2 P + ω Q .17) (Sµν )i φj (x).23) The hamiltonian in this case can be expressed in the number operator N = a† a. φ(y)] = 0 if (x − y)2 < 0.Quantization of ﬁelds 56 or if U (Λ. and [N. (7. (7. i(xµ ∂ν − xν ∂µ )φ (x) + i (7. a† ] = 1. [a. • Causality Operators Φ(f ) and Φ(g) for which the test functions are space-like separated can be measured simultaneously (macroscopic causality). P ] = i. The measurements cannot inﬂuence each other or [Φ(f ). a) = i Rj (Λ−1 ) = 1 + iǫµ P µ − 1− i ωµν M µν . a] = [a† .19) 7. Φ(g)] = 0. sometimes referred to as second quantization. H = = ω ω P P Q−i √ Q+i√ 2 2 2ω 2ω 1 1 =ω N+ . 1 Q = √ (a + a† ) 2ω and P = −i ω (a − a† ) 2 (7.25) . j (7. Microscopic causality implies local commutativity. P ] = 0 (7.4. a] = −a. Q] = [P. [Q. (7. a† ] = a† . This must be valid for the – now operators – generators P µ and M µν that in the previous chapter have been found in terms of the ﬁelds through Noether’s theorem.20) where the coordinate Q and the momentum P satisfy the canonical commutation relations [Q. j 2 one has [φi (x). [N. ω a† a + 2 2 ω i − [Q.18) with S µν given in 6.22) it is straightforward to check that the commutation relations between Q and P are equivalent with the commutation relations [a. P ] 2 (7. a† ] = 0.2 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.24) It is straightforward to ﬁnd the commutation relations between N and a and a† .

(n − 1) a|n † i.4 for the classical ﬁeld) a(k) a† (k ′ ) = = d3 x ei k·x i ∂0 φ(x). a† (k ′ )] = 0. φ(x )]x0 =x′0 = [Π(x). [φ(x).e.32) (7. From the normalizations one obtains a |n = √ n + 1 |n + 1 and a|n = n |n − 1 . Π(x )]x0 =x′0 = 0. The quantization of the ﬁeld is achieved by quantizing the coeﬃcients in the Fourier expansion. In this way one has found for the harmonic oscillator the spectrum of eigenstates |n (with n a non-negative integer) with En = (n + 1/2)ω.26) a(k) e−i k·x + a† (k) ei k·x .e. Π(x′ )]x0 =x′0 = i δ 3 (x − x′ ) ′ ′ and (7.31) The hamiltonian can be rewritten in terms of a number operator N (k) = N (k) = a† (k)a(k). † √ a and a act as raising and lowering operators. N |n = n|n one ﬁnds N a† |n N a|n = = (n + 1) a† |n . φ(x) = (2π)3 2Ek where the Fourier coeﬃcients a(k) and a† (k) are now operators. and (7.Quantization of ﬁelds 57 Deﬁning states |n as eigenstates of N with eigenvalue n. i. which we have split into positive and negative energy pieces with (complex) coeﬃcients a(k) and a∗ (k) multiplying them.33) . d3 x φ(x) i ∂0 e−i k ·x ↔ ′ ↔ (7. H = = = d3 x 1 1 1 (∂0 φ)2 + (∇φ)2 + M 2 φ2 2 2 2 d3 k Ek † a (k)a(k) + a(k)a† (k) (2π)3 2Ek 2 d3 k Ek N (k) + Evac .27) 2 (2π)3 It is easy to check that these equations can be inverted (see Exercise 2. 7. the real scalar ﬁeld φ(x) becomes d3 k (7. The canonical momentum becomes −i d3 k ˙ Π(x) = φ(x) = a(k) e−i k·x − a† (k) ei k·x . and we see that a state |0 must exist for which N |0 = a|0 = 0. but one needs to realize that in that case k 0 = Ek = k2 + M 2 . which represents the ’number of particles’ with momentum k.30) [φ(x). is equivalent with [a(k). a(k ′ )] = [a† (k).29) It is straightforward to prove that the equal time commutation relations between φ(x) and Π(x′ ) are equivalent with ’harmonic oscillator .28) (7.g. (7. Note that we will often write a(k) or a† (k).like’ commutation relations between a(k) and a† (k ′ ). (2π)3 2Ek (7. a† (k ′ )] = (2π)3 2Ek δ 3 (k − k′ ) [a(k).3 The real scalar ﬁeld We have expanded the (classical) ﬁeld in plane wave solutions. e.

e. writing all annihilation operators to the right of the creation operators. . .35) where the vacuum contribution disappears because of rotational symmetry. in ﬁeld theory also referred to as vacuum energy Evac = 1 V 2 d3 k Ek . (2π)3 2Ek (7. φ(x) φ+ (x) φ− (x) = = = φ+ (x) + φ− (x). and multiparticle states |(k1 )n1 (k2 )n2 .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. is solved by subtracting it as an (inﬁnite) constant. (2π)3 (7. |0 . i. assuring that the vacuum (by deﬁnition) has eigenvalue 0!.34) where V = (2π)3 δ 3 (0) is the space-volume.37) (7. |k = a† (k)|0 .39) The problem with the vacuum or zero-point energy. d3 k |k k|. Just as in the case of the harmonic oscillator it is essential (axiom) that there exists a ground state |0 that is annihilated by a(k). (2π)3 2Ek d3 k a† (k) ei k·x . (2π)3 2Ek (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).44) p|φ− (x)|0 = ei p·x . = normalized as (a† (k1 ))n1 (a† (k2 ))n2 √ √ . which amounts to redeﬁning H as H= d3 x : H (x) := d3 k Ek N (k). n1 ! n2 ! (7.46) (7. deﬁning particle states |k = |k (with positive energy.42) (7. .38) d4 k 2π δ(k 2 − M 2 ) θ(k 0 ) |k k| = (2π)4 (7. . This term will be adressed below.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).41) (7. The rest of the states are then obtained from the groundstate via the creation operator. (2π)3 2Ek (7. . d3 k a(k) e−i k·x .45) (7. For the purpose of normal ordering it is convenient to decompose the ﬁeld in positive and negative frequency parts. which now contains an inﬁnite number of oscillators. 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. a(k)|0 = 0. k 0 = Ek = k2 + M 2 ).36) and satisfying the completeness condition 1 = k|k ′ = (2π)3 2Ek δ 3 (k − k′ ). (2π)3 2Ek (7.40) This procedure is known as normal ordering.

d3 k ei k·(x−y) = −i∆+ (y − x). = − (2π)3 2Ek (7. The last item to be checked for the scalar ﬁeld are the causality condition.48) (7.50) or as integrals over d4 k.55) d4 k ǫ(k 0 ) 2π δ(k 2 − M 2 ) e−i k·x . φ(y)] which has the properties (i) i∆(x) = i∆+ (x) + i∆− (x) can be expressed as i∆(x) = where ǫ(k 0 ) = θ(k 0 ) − θ(−k 0 ).17) and (7. that they satisfy the required commutation relations in Eqs (7. (ii) ∆(x) is a solution of the homogeneous Klein-Gordon equation. (iii) ∆(0. a† (k ′ )] 3 2E ′ (2π) k [φ− (x).51) (7. d3 k = (2π)3 2Ek = (7. φ+ (y)] d3 k e−i k·(x−y) . i∆+ (x) i∆− (x) = = = d4 k θ(k 0 ) 2π δ(k 2 − M 2 ) e−i k·x .e. (2π)4 (7. (7.Quantization of ﬁelds 59 In order to ensure the consistency of the theory it is necessary to check that the operators P ν and M obtained from the conserved currents Θρσ and M µρσ are indeed the generators of the Poincar´ e group. (7. (2π)3 2Ek ≡ i∆− (x − y). ∆(x) = − ǫ(x0 ) δ(x2 ).54) = i∆(x − y) = i (∆+ (x − y) + ∆− (x − y)) .56) = −δ 3 (x).52) The result for the invariant commutator function is [φ(x). x) = 0 and hence ∆(x) = 0 for x2 < 0.53) t=0 . φ− (y)] ≡ i∆+ (x − y). 2π (7. In order to calculate [φ(x). (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. (iv) The equal time commutation relations follow from ∂ ∆(x) ∂t (v) For M = 0. i.47) ′ d3 k ′ e−i k·x+i k ·y [a(k). φ(y)] consider µν [φ+ (x).49) (7.18).

e. 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 .64) The commutator of φ and φ† is as for the real ﬁeld given by [φ(x). 2 From the lagrangian density (7. ↔ (7. which is equivalent to the relations (only nonzero ones) [a(k). (2π)3 2Ek (7. (2π)3 2Ek d3 k b(k) e−i k·x + a† (k) ei k·x .66) .59) (7.Quantization of ﬁelds 60 7.61) (7.60) = i φ∗ ∂µ φ. b† (k ′ )] = (2π)3 2Ek δ 3 (k − k′ ).4 The complex scalar ﬁeld In spite of the similarity with the case of the real ﬁeld. = ∂{µ φ∗ ∂ν} φ − L gµν . Also the charge operator requires normal ordering (in order to give the vacuum eigenvalue zero).57) (7.5 The Dirac ﬁeld L = ↔ i ψγ µ ∂µ ψ − M ψψ. 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) .65) 7. The hamiltonian is as before given by the normal ordered expression H = = = d3 x : Θ00 (x) : d3 k Ek : a† (k)a(k) + b(k)b† (k) : (2π)3 2Ek d3 k Ek a† (k)a(k) + b† (k)b(k) .62) (7. particles (created by a† ) and antiparticles (created by b† ) with the same momentum contribute equally to the energy. (2π)3 2Ek (7. ∂0 φ† (y)]x0 =y0 = i δ 3 (x − y). φ† (y)] = i∆(x − y). (2π)3 2Ek (7.58) and satisfy the equal time commutation relation (only nonzero ones) [φ(x). a† (k ′ )] = [b(k). 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. (7.63) i.

s)vj (k. (2π)3 2Ek (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) . s)v(k.75) (7. s′ )} = (2π)3 2Ek δ 3 (k − k′ ) δss′ .71) (7. (2π)3 2Ek d3 k b† (k. s)¯(k.72) ψ(x) = s In terms of the operators for the b and d quanta the hamiltonian and charge operators are (omitting mostly the spin summation in the rest of this section) H = = Q = = d3 x : ψ † (x) i∂0 ψ(x) : d3 k Ek : b† (k)b(k) − d(k)d† (k) : (2π)3 2Ek d3 x : ψ † ψ : d3 k : b† (k)b(k) + d(k)d† (k) : .78) (7. The quantized ﬁelds are written ψ(x) = s d3 k b(k. d† (k ′ .Quantization of ﬁelds 61 the conserved density and energy-momentum currents are easily obtained. ˙ δ ψ(x) ↔ i Θ00 (x) = − ψγ i ∂i ψ + M ψψ 2 = iψγ 0 ∂0 ψ = iψ † ∂0 ψ. (7. s)¯(k. (7. b† (k ′ .67) i ψ / ψ − M ψψ gµν . † {ψi (x). {b(k. s) ei k·(x−y) x0 =y 0 .69) (7.73) (7. jµ Θµν = = ψγµ ψ. s)u† (k. s)e−i k·x . s′ )} = {d(k. ↔ i ψγµ ∂ν ψ − 2 (7.76) which seems to cause problems as the antiparticles (d-quanta) contribute negatively to the energy and the charges of particles (b-quanta) and antiparticles (d-quanta) are the same. ∂ 2 ↔ (7. interchanging creation and annihilation operators. s)ei k·x .70) where the last line is obtained by using the Dirac equation.e. i. s)ei k·x + d(k. ψj (y)}x0 =y0 = d3 k (2π)3 2Ek ui (k.68) The canonical momentum and the hamiltonian are given by Π(x) H (x) = = δL = i ψ † (x). The solution is the introduction of anticommutation relations. s)e−i k·x + d† (k. (2π)3 2Ek (7. s) e−i k·(x−y) j s † + vi (k.74) (7. s).79) Also for the ﬁeld and the canonical conjugate momentum anticommutation relations are considered. u v (2π)3 2Ek (7. .77) Note that achieving normal ordering. s). s)u(k.

83) From the lagrangian density the canonical momenta are = = (7. ψ j (y)} = d3 k (/ + M )ij e−i k·(x−y) + (/ − M )ij ei k·(x−y) k k (2π)3 2Ek d3 k e−i k·(x−y) + −ei k·(x−y) = (i/x + M )ij ∂ (2π)3 2Ek = (i/x + M )ij i∆(x − y). 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. (7.84) (7. ψ j (y)] = (i/x + M )ij i∆1 (x − y).86) (7.Quantization of ﬁelds 62 Using the positive and negative energy projection operators discussed in section 4. ˙ δ Ai (7.88) . ˙ δ A0 δL = F i0 = E i . When we would have started with commutation relations for the ﬁeld ψ and the canonical momentum.87) If one wants to impose canonical commutation relations ∂µ Aµ = 0 cannot hold as an operator identity.81) where i∆ = i∆+ + i∆− is the same invariant commutator function as encountered before. For the scalar combination {ψi (x).6 The electromagnetic ﬁeld 1 L = − Fµν F µν .80) 3 = δ 3 (x − y) δij . (7. but we must restrict ourselves to the weaker condition B|∂µ Aµ |A = 0. it implies the Lorentz constraint and leads to the canonical momenta Π0 Π i = = E i. ∂ (7.85) which reﬂects the gauge freedom. 4 Π0 Πi δL = 0. we would have obtained [ψi (x). 4 2 This gives the equations of motion discussed before. −λ(∂ρ Aρ ). which however has wrong causality properties! Therefore the relation between spin and statistics is required to get micro-causality (which is also known as the spin statistics theorem). ∂ (7.82) where i∆1 = i∆+ −i∆− . but has the problem of being noncovariant. 7. ψ j (y)} at arbitrary times one has {ψi (x). one has † {ψi (x). as the vanishing of Π0 induces a constraint. It is possible to continue in a covariant way with the lagrangian 1 λ L = − Fµν F µν − (∂µ Aµ )2 .

It gives 3 k µ ǫ(λ) (k)c(k. a longitudinal photon and two transverse photons. 2. and are equivalent with [c(k. Aν (y)] = [Πµ (x). where Aµ + + is the part of the vector ﬁeld containing the annihilation operators. pj ]P . 0. one time-like photon by itself is not allowed! This solves the problems with the normalization and the negative energies. λ′ )] = −g λλ 2Ek (2π)3 δ 3 (k − k′ ). i. for which it is suﬃcient that ∂µ Aµ |A = 0. In fact the commutation relations imply ˙ [Aµ (x). 7. ˙ ˙ . The quantized ﬁeld is expanded as Aµ (x) = d3 k (2π)3 2Ek (λ) 3 ǫ(λ) (k) c(k. µ λ=0 (7. where Q is the conserved quantity in Eq. (7. The hamiltonian in terms of the creation and annihilation operators is (after normal ordering) given by H= d3 k Ek (2π)3 2Ek 3 λ=1 c† (k. λ)c(k.e. Πν (y)] = 0. 0) . λ)ei k·x . (b) Proof for a quantity O(p. λ)|A = 0. The canonical equal time commutation relations are [Aµ (x).94) Choosing k µ = (|k 3 |.Quantization of ﬁelds 63 for physical states |A and |B . j = 1.90) where Πν = F ν0 − λ g ν0 (∂ρ Aρ ) and we have furthermore [Aµ (x). [ℓi . (d) Use Poisson brackets to calculate q and p.11. Aν (y)]x0 =y0 = i gµν δ 3 (x − y). and [ℓi . λ) − c† (k. c† (k ′ . λ). containing a time-like photon. λ)e−i k·x + c† (k. i. rj ]P . These problems are solved by the Lorentz constraint between physical states given above. (7.89) with four independent vectors ǫµ . Q]P . 0. what are [ri . µ λ=0 (7.93) which does exhibit problems with the time-like photon. ′ (7.3) satisfy the conditions of a Lie algebra. q) that dO/dλ = [O.1 (optional) If you haven’t seen Poisson brackets in classical mechanics you should (a) Proof that Poisson brackets (Eq. 6.e.95) Exercises Exercise 7. pj ]P .91) (7. 0)c(k.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. k 3 ) this reads |k 3 |c(0) (k) − k 3 c(3) (k) |A = 0 c(0) (k) ∓ c(3) (k) |A = 0. (c) Calculate the Poisson brackets for the coordinates x and the standard conserved quantities p and ℓ. Πν (y)]x0 =y0 = i gµν δ 3 (x − y). which reproduce the Hamilton equations. (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. Pµ ] for the (real) scalar ﬁeld using the expressions for the ﬁelds and momentum operator in terms of creation and annihilation operators. f (Λk) = f (k) for Lorentz transformations Λ. (e) Show that ∆R (x) = 0 if x0 < 0. (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.3 (a) Show that the correct implimentation of symmetries in the Hilbert space. Give also the expression for ∆C . (Similarly one can show that for the advanced Green’s function ∆A (x) = 0 if x0 > 0). e−i a requires [φ(x). (2 + M 2 ) ∆inhom (x) = −δ 4 (x). .Quantization of ﬁelds 64 Exercise 7.4 (Green’s functions) (a) Show that ∆hom (x) = 1 (2π)4 d4 k e−ik·x δ(k 2 − M 2 )f (k) is a solution of the homogeneous Klein-Gordon equation. (2 + M 2 ) ∆hom (x) = 0. (b) Check the above commutation relation [φ(x).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. Pµ ] = i∂µ φ(x). Is the subtraction of ’zero point’ contributions essential in this check? µ Pµ φ(x) ei a µ Pµ = φ(x − a) Exercise 7. ∆hom (Λx) = ∆hom (x) if f is an invariant function. It is invariant. we can write for ∆R an integral where the k 0 integration remains along the real axis. ∆R = 1 (2π)4 d4 k e−ikx k2 1 − M 2 + iǫk 0 (with the implicit prescription to take ǫ → 0 after integration).

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

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

if helicity λ is used instead of the z-component of the spin m. x Therefore γ0 ψ(˜) is again a solution of the Dirac equation and we have x ψ p (x) = γ0 ψ(˜).2 Charge conjugation We have already seen the particle-antiparticle symmetry with under what we will call charge conjugation the behavior T (8.7) = λ d k x x ηP b(k.6) (check this for the standard representation.9) (8. −λ) e−ik·x − d† (k. The result is ψ p (x) −1 = Pop ψ(x) Pop = ηP γ0 ψ(˜) x (8. ˜ −v(k.13) ψ(x) −→ ψ c (x) = ηC C ψ (x). m) = = ˜ u(k. the above operation reverses the sign of λ). −λ)u(k. choosing ηP is real (ηP = ±1) particle and antiparticle have opposite parity. λ)v(k.4) (8.12) and for the vector ﬁeld −1 Aµ (x) −→ Pop Aµ (x) Pop = −Aµ (˜). −λ) eik·x (2π)3 2Ek ˜ d3 k ˜ ˜ ˜ ˜ ηP b(k.Discrete symmetries 67 After parity transforming x to x the Dirac equation becomes after some manipulations ˜ ˜ x iγ µ ∂µ − M ψ(˜) = 0. λ)γ0 v(k. −λ) e−ik·x − d† (k. x (8. −λ). γ x (8. −λ)v(k. λ) P −1 = −η ∗ d(k.10) i. λ)u(k. ˜ Pop d(k. λ) e−ik·x − d† (k. (i˜ µ ∂µ − M ) ψ(˜) = 0.11) (8. λ) eik·x . op P (8. (2π)3 2Ek (8.e. −λ). . In the same way as the Fermion ﬁeld.5) (8. λ)u(k. λ) eik·˜ (2π)3 2Ek d3 k ˜ ˜ ˜ ˜ ηP b(k. m). m) γ0 v(k. iγ µ† ∂µ − M ψ(˜) = 0. −λ) eik·x 3 2E (2π) ˜ k d3 k ˜ ˜ ηP b(k. x The latter behavior of the vector ﬁeld will be discussed further below. For the scalar ﬁeld we have seen −1 φ(x) −→ Pop φ(x) Pop = ηP φ(˜).8) 3 = λ = λ = λ From this one sees immediately that −1 ˜ Pop b(k. m).3) (8. x µ (iγ ∂µ − M ) γ0 ψ(˜) = 0. λ)γ0 u(k. 8. λ)v(k. λ) e−ik·˜ + d† (k. λ) Pop = ηP b(k. x It is straightforward to apply this to the explicit ﬁeld operator ψ(x) using γ0 u(k. one can also consider the scalar ﬁeld and vector ﬁelds.

x 1A 2 An −iC −1 γ µ C∂µ − M ψ(−˜) = 0. x x (8. One has Aφ|Aψ = φ|ψ ∗ = ψ|A† Aφ = ψ|φ . (8. m). as discussed in section 4). The action on the spinors (using C = iγ 2 γ 0 = −iρ1 σ 2 in standard representation) gives C uT (k.Discrete symmetries 68 the latter being also a solution of the Dirac equation.21) where A is a 4 × 4 matrix acting in the spinor space. x (8. Top |φ = φt | = (|φt )∗ . λ). (2π)3 2Ek (8.22) = Aψ|Aφ ∗ is antilinear if A(λ|φ + µ|ψ ) = λ∗ A|φ + µ∗ A|ψ . γ x i(γ5 C)−1 γ µ γ5 C∂µ − M ψ(−˜) = 0. . writing −1 ψ(x) −→ Top ψ(x)Top = ηT Aψ(−˜) ≡ ψ t (−˜). The same relations hold for helicity states. Norm conservation requires Top to be anti-unitary2 . m) ¯ C v T (k.18) (8. x iγ µT ∂µ − M ψ(−˜) = 0.3 Time reversal xµ = (t.20) The time reversal operator transforms We will again consider the transformation properties for a fermion ﬁeld ψ(x). λ)C uT (k. 8. The (time-reversed) Dirac equation becomes. λ) e−ik·x + b† (k.17) 3 = λ This shows that −1 Cop b(k. (i˜ µ∗ ∂µ − M ) ψ(−˜) = 0. m) ¯ = = v(k. to ﬁnd ψ t (−˜) that is a solution of the Dirac equation. i. m). (8. r). λ)C v T (k. For a quantized ﬁeld one has −1 ∗ −1 Top fk (x)bk Top = fk (x)Top bk Top . u(k. λ) Cop = ηC d(k. λ). λ) −1 Cop = ∗ ηC b(k. λ)u(k. it is antilinear1 . x (8.e.16) d k ηC d(k. we start with the complex conjugated x Dirac equation for ψ. Therefore ψ c (x) = = λ −1 ¯ Cop ψ(x) Cop = ηC C ψ T (x) (8. λ) eik·x . λ)v(k. antilinear operator is anti-unitary if A† = A−1 . x µ (iγ ∂µ − M ) γ5 Cψ(−˜) = 0. λ) e−ik·x + b† (k.15) (where one must be aware of the choice of spinors made in the expansion.14) (8.19) Cop d(k. ((iγ µ )∗ ∂µ − M ) ψ(x) = 0. λ) eik·x ¯ ¯ (2π)3 2Ek d3 k ηC d(k. ˜ x −(iγ µ )∗ ∂µ − M ψ(−˜) = 0. As time reversal will transform ’bra’ into ’ket’. r) −→ −˜µ ≡ −xµ = (−t.

λ)v ∗ (k.23) In the standard representation iγ5 C = σ2 and it is straightforward to apply this to the explicit ﬁeld operator ψ(x) using i γ5 C u(k. ∗ ˜ v (k. λ| ¯ C |¯. λ)u∗ (k. λ) eik·˜ + d† (k. λ) i γ5 C v(k. p. −λ |¯.4 Bi-linear combinations In quantities such as currents and lagrangians often bilinear combinations of spinor ﬁelds are encountered.2: The transformation properties of physical states for particles (a) and antiparticles (¯). x (8. λ) e−ik·x (2π)3 2Ek (8. Therefore one has A|k = A|b† (k)|0 = A|T † T b† (k) T † T |0 ∗ = At |b† (k)|0 ∗ ˜ t = k|At . λ). λ a P |a.30) (8. −p. (8. op T (8. λ) T −1 = η ∗ d(k. p. λ)u∗ (k.28) (8. λ| a. λ |¯.26) d k x x ηT b(k. λ) eik·x + d† (k.33) (8. p.32) (8. λ) e−ik·˜ (2π)3 2Ek d3 k ˜ ˜ ˜ ˜ ηT b(k. λ) e−ik·x 3 2E (2π) k d3 k ˜ ˜ ηT b(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. −p.34) . Since there are 16 independent 4 × 4 matrices. −p. a state |a.Discrete symmetries 69 Table 8. λ) Top = ηT b(k. λ Therefore γ5 Cψ(−˜) is again a solution of the (ordinary) Dirac equation and we can choose (phase is x convention) ψ t (x) = i γ5 C ψ(−˜). λ). λ). there are 16 independent of these bilinear combinations. Note that applying an anti-unitary transformation such as Top one must take for the matrix element ˜ the complex conjugate.27) 3 = λ = λ = λ From this one obtains −1 ˜ Top b(k. λ). λ)iγ5 Cv(k. λ)v ∗ (k. λ)u∗ (k. −λ a T a.31) (8. ˜ Top d(k.25) (check this for the standard representation).29) In table 2 the behavior of particle states under the various transformations has been summarized. λ) e−ik·x 3 2E (2π) k ˜ d3 k ˜ ˜ ˜ ˜ ηT b(k. ˜ = 0|b(k)|A 8. p. −p. The result is = = λ (8.24) (8. λ) eik·x + d† (k. λ a |a. λ) eik·x + d† (k. λ)v ∗ (k. λ) ψ t (x) −1 Top ψ(x) Top = i ηT γ5 Cψ(−˜) x = = ˜ u∗ (k. λ)iγ5 Cu(k.

In many applications the expectation values of currents are needed. p′ |V µ (x)|p = u(p′ )Γµ (p′ . There are several other terms that also have the correct tensorial behavior such as (pµ + p′µ ) γ5 σ µν qν q µ F3 (q 2 ). ¯ V µ (x) =: ψ(x)γ µ ψ(x) :.Discrete symmetries 70 Table 8. For instance for nucleons one has Γµ (p′ . called form factors. This implies p′ |V µ (x)|p = e−i (p−p )·x p′ |V µ (0)|p . as well as under the combined operation Θ = PCT (see Table 3). for the vector current V µ (x). (8. γ ν ]. Applying the results from the previous sections it is straightforward to determine the behavior of the combinations under P. The 16 combinations of Dirac matrices appearing above are linearly independent.35) The x-dependence can be accounted for straightforwardly using translation invariance. Note that the lagrangian density L (x) → L (−x) under Θ.g.. 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). γ µ . γ5 γ µ or γ5 ). (q 2 ). s′ |V µ (x)|p.39) where Γµ (p′ . p′ |V µ (0)|p = u(p′ )γ µ u(p). ¯ ′ ′ (8. (q 2 ). C. As an example consider the vector current for a point fermion. ¯ In general the expectation value between momentum states can be more complicated.38) (8.37) (8. F.. σ µν . p) = γµ F1 (q 2 ) + iσµν q ν F2 (q 2 ). 8. p)u(p) e−i (p−p )·x .41) 2M 2M . e. p) can be built from any combination of Dirac matrices (1. of which the expectation value between momentum eigenstates can be simply found. p′ . in this case only q 2 = (p′ − p)2 . V µ (x) = ei Pop ·x V µ (0)e−i Pop ·x . 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 .40) where the coeﬃcients are functions of invariants constructed out of the momenta. C. s .36) (8. F. momenta (pµ or p′µ ) or constant tensors (ǫµνρσ or gµν ). and T.5 Form factors Currents play an important role in ﬁeld theory. (8.3: The behavior of the independent bilinear combinations of fermi ﬁelds under P. 2M (8.

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

|q|). GE GM = F1 − Q2 F2 . p = (E. 4M 2 = F1 + F2 . (b) Show that such a term is also not allowed by time-reversal symmetry. (b) Show that in the interaction with the electromagnetic ﬁeld. p′ = (E. −|q|/2). Exercise 8.Discrete symmetries 72 (a) Use the Dirac equation to prove the Gordon-decomposition u(p′)γ µ u(p) = ¯ 1 u(p′ )[(p′ + p)µ + iσ µν (p′ − p)ν ]u(p) ¯ 2M (b) Give the relation between Hi en Fi . 0. (c) Show that if a term of this form would exist. 0.2 (a) Calculate the current matrix element (using the explicit fermion spinors) in the Breit-frame in which q = (0.3 (a) Show that because of parity conservation no term of the form γ5 can appear in the current matrix element. |q|/2) with E 2 = M 2 + |q|2 /4 and express them in terms of the Sachs form factors depending on Q2 = −q 2 . Exercise 8. 0. 0. (c) Show that hermiticity of the current requires that F1 en F2 are real. σµν q ν F3 (q 2 ) 2M . 0. L int = ejµ Aµ the charge is given by e GE (0) and the magnetic moment by e GM (0)/2M . 0. it would correspond to an electric dipole moment d = −e F3 (0)/2M (Interaction term d σ · E).

QH (t)|q. |ψS (t) = U (t.Chapter 9 Path integrals and quantum mechanics 9. t0 ) = e−i(t−t0 )H or i Two situations can be distinguished: (i) Schr¨dinger picture. t′ ≡ q ′ |q ′ . and the operators are time-dependent. AS (t) = AS and the states o are time dependent. H]. t . Consider the two (time-independent) Heisenberg states: |q. (9.5) (9. i ∂ |ψS ∂t ∂ i AS ∂t = H |ψS .4) ∂ U (t. [AH . ≡ 0.3) (9. |ψH (t) = |ψH . t′ . AH (t) = U −1 (t.7) 73 .2) (ii) Heisenberg picture. (9. t0 ) AH (t0 ) U (t. t ′ |q . t ≡ q|q.1 Time evolution as path integral U (t. (9. t0 ). in which the states are time-independent.1) The time evolution from t0 → t of a quantum mechanical system is generated by the Hamiltonian. t0 ) ∂t (9. t0 ) = H U (t. (9. in which the operators are time-independent. t0 )|ψS (t0 ) .6) Of these the Heisenberg picture is most appropriate for quantum ﬁeld theory since the ﬁeld operators do depend on the position and one would like to have position and time on the same footing. i ∂ |ψH ∂t ∂ i AH ∂t ≡ = 0. t ′ QH (t′ )|q ′ .

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

q1 .Path integrals and quantum mechanics 75 or for the full interval q ′ . F3 [α. . y) α(y) . while Dα F [α] = x dαx N F (αx ) (9. Before proceeding we also give the straightforward extension to more than one degree of freedom. What one is after is the meaning of Dα F [α]. . .e. (9. .e. Examples are F1 [α] = α ≡ F2 [α] = α2 ≡ dx α(x). . dx dy α(x) K(x. t′ |q1 . y) β(y). . qN ) ˙ . ˙ where Dq and Dp indicate functional integrals. t′ |q. dx α2 (x). qN . qN . . etc. β] = α K β ≡ 1 F4 [α] = exp − α2 2 F5 [α. ′ ′ q1 . . .18) t ′ = n=1 Dqn D pn 2π exp i t dτ n pn qn − H(p1 . 9.2 Functional integrals In this section I want to give a fairly heuristic discussion of functional integrals.17) = Dq D dτ [pq − H(p. t N (9. pN . . F [α] ∈ R. Two working approaches for functional integration are the following: (i) the heuristic division of the space on which α acts into cells. i. . q)] ˙ (9. dx dy α∗ (x) K(x. i.20) becomes a multidimensional integral. ≡ exp − 1 2 dx α2 (x) . t = τ dq(τ ) dp(τ ) exp i 2π p exp i 2π t′ t t′ t ∆τ [pq − H(p. y) in the above examples can be (hermitean) operators acting on the functions. 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. α∗ ] = exp (−α∗ K α) ≡ exp − The kernels K(x. . including diﬀerential operators. α(x) → αx and F [α] = F (αx ) changes into a multivariable function. q)] .19) where F [α] is a functional that represents a mapping from a function space F of (real) functions α (R → R) into the real numbers R. .

27) (9. .24) (9. with Dα F [α] = n dαn N F (αn ) (9.22) = x The last equality is obtained by deﬁning the right measure (normalization N ) in the integration. A useful property of functional integration is the translation invariance.n fm (x) Kmn fn (y) and consider F [α] = F (αn ) as a multivariable function. Note that procedure (i) is an example of the more general procedure under (ii). (9.23) Having deﬁned the Gaussian integral. Consider the gaussian functional as an example.28) Dα Dα∗ exp (−α∗ α − α∗ ω − ω ∗ α) = exp (ω ∗ ω) . 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. In both cases N is an appropriately choosen normalization constant in such a way that the integral is ﬁnite. Dα F [α] = Dα F [α + β]. y) = ∗ m. Physical answers will usually come out as the ratio of two functional integrals and are thus independent of the chosen measure.25) (see Exercises).n dx fm (x)fn (x) = n = n dαn 1 exp − α2 N 2 n √ 2π ≡ 1.Path integrals and quantum mechanics 76 (ii) Write α in terms of a sum of orthonormal basis functions. 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. 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 ≡ 1.26) As an important application of translation invariance we mention the identities 1 Dα exp − α2 − ωα 2 = exp 1 2 ω . (9. N (9.21) again a multidimensional integral. 2 (9. the following integrals can be derived for a real symmetric or complex hermitean kernel K. α(x) = n αn fn and K(x.

n θn fn (x) dx θ∗ (x) θ(x) .38) .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.e. δα(x) (9. It is now easy to check that Dθ∗ Dθ exp (−θ∗ K θ) = det K.34) (9.37) We note that from the ﬁrst to the second line one needs to realize that a pair of diﬀerent Grassmann variables behaves as ordinary complex variables.32) (9. (9. −→ δα(x) ∆x ∂αx ∂αx δ ∂ ∂ ∗ with . α(y) = δ(x − y).31) δ δ α = dy α(y) = dy δ(x − y) = 1. (9. θ∗ ] = exp (−θ∗ θ) ≡ exp − F [θ. θη = −ηθ. αy = δxy .Path integrals and quantum mechanics 77 Functional diﬀerentiation is deﬁned as δ . θ∗ ] = exp (−θ∗ K θ) ≡ exp − Integration for Grassmann variables is deﬁned as dθ 1 = 0 and dθ θ = 1. δα(x) 2 2 δ exp (−α β) = −β(x) exp (−α β) . i. products of the same pair appear. e. y) β(y) δα(x) δ 1 1 exp − α2 = −α(x) exp − α2 . −→ fn (x) δα(x) ∂αn ∂αm n Note that δ(x − y) = n ∗ fn (x) fn (y). In the expansion of the exponential (from second to third line). dx dy θ∗ (x) K(x. F [θ.33) (9. y) θ(y) .29) (9. Examples of functional diﬀerentiation are (9.35) For applications to fermion ﬁelds.g. which vanish. δα(x) δα(x) δ αKβ = dy K(x. θ2 = 0. In principle the deﬁnitions of functionals is the same. αn = δmn . the Gaussian-type functionals. (9. we need to consider Grassmann-valued functions involving anticommuting Grassmann variables. δα(x) or in discretized form δ 1 ∂ ∂ with . however.30) (9.

Path integrals and quantum mechanics

78

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

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

= δ(x − y),

(9.39)

(9.40) (9.41)

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

9.3

Time ordered products of operators and path integrals

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

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

q ′ , t′ |q, t =

Dq exp i

t

dτ L(q, q) , ˙

(9.44)

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

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

t′

Dq exp i

t i 2

dτ Leﬀ (q, q) , ˙

(9.45)

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

i

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

δ(0) ln f (q).

dqi q ′ , t′ |qn , tn . . . (9.46)

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

t′ t′ t

dτ [pq − H(p, q)] ˙

(9.47) (9.48)

Dq q(s) exp i

t

dτ L(q, q) . ˙

Path integrals and quantum mechanics

79

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

t t′

(9.49)

dτ L(q, q) . ˙

(9.50)

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

9.4

An application: time-dependent perturbation theory

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

′

(9.51) (9.52) (9.53)

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

t′ ∞ n=0 t′

= HI (t) ψI (t), = [AI , H0 ],

(9.54) (9.55)

(9.56) (9.57)

dτ HI (τ ) U (τ, t),

t

(9.58)

U (n) (t′ , t),

(9.59)

1−i (−i)

t t′ t

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

**Path integrals and quantum mechanics
**

t′ t t′ t

80

U

(2)

(t , t) = 1 − i = (−i) . . .

′

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

t′

U

(n)

(t , t) = (−i)

t

′

dτ HI (τ ) U (n−1) (τ, t)

t′ τ1 τn−1

= (−i)n

t

dτ1

t t′ t′

dτ2 . . .

t t′

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

t

=

(−i)n n!

dτ1

t t

dτ2 . . .

(9.60)

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

t′ τ1

dτ1

t t

dτ2 HI (τ1 ) HI (τ2 )

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

t′ τ1

dτ1

t t

dτ2 HI (τ1 ) HI (τ2 )

t′ t′

+

1 2

dτ2

t τ2

dτ1 HI (τ1 ) HI (τ2 ),

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

t′ t′

dτ1

t t

dτ2 HI (τ1 ) HI (τ2 )θ(τ1 − τ2 )

t′

+

1 2

t′

dτ2

t t

dτ1 HI (τ1 ) HI (τ2 )θ(τ1 − τ2 ),

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

t′ t′

dτ1

t t

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

the desired result.

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

∞ t′ t′ t′

dτ1

t t′ t

dτ2 . . .

t

dτn T HI (τ1 ) . . . HI (τn ) (9.61)

≡ T exp −i

dτ HI (τ ) .

t

t q. t Q′ . for the physical states o q. i. ˙ ˙ physically pictured as. it is possible to switch on an interaction. harmonic oscillator. T . T ′ |Q. ˙ J (9. t|n n|Q. φn (q. say. 9. t|Q. t) φ∗ (Q) e+i En T . a set |n of physical eigenstates. The Heisenberg state |q. the creation of an electron (think of a radio-tube. t |q. i. T with q. q)] .Path integrals and quantum mechanics source experiment detector 81 T L t L+Jq t’ L T’ Figure 9. T = n q.62) p Dq D exp i 2π t dτ [pq − H(p. L(q.1: Physical picture of vacuum to vacuum amplitude in presence of a source term This is not a surprising result. q)] ˙ (9. for g. ˙ The combination of exponentials in the last step is allowed because we are simply dealing with classical quantities (not operators!).5 The generating functional for time ordered products By introducing a source-term. t = = ′ ′ V = q . t′ |q. We can again rewrite the time-ordered products as functional integrals as derived in the previous section.63) for plane waves.64) (9.e. t is related to the Schr¨dinger state |q by |q. T = n J t′ = = Dq exp i t dτ [L(q. for example x|n x|n = φn (x) = ei kn x = e −ω 2 x2 /2 (9.66) . Before and after these processes there is only the vacuum or ground-state |0 . n (9. q) + J(τ )q] . t = ei Ht |q . which in turn are embedded between an early time T and a future time T ′ . t′ |q. T < t < t′ < T ′ . q) + J(t) · q. t′ q ′ . one has q ′ . T ′ |q ′ . t|n = q|n e−iEn t with q|n the time-independent wave function. Consider. t|Q. q .e. Considering the source to be present between times t and t′ . making the electron) and the absorption of an electron (think of a detector). t |T exp −i p exp −i 2π t′ ′ ′ t′ t dτ HI (Q(τ )) |q.s. t t′ V =0 Dq D dτ HI (q) exp i t t′ t dτ [pq − H0 (p. q) → L(q.65) J dq ′ dq Q′ . furthermore.

t) (9. t′ ) q ′ . T → i ∞ and T ′ → −i ∞. The importance of Z[J] is that the time ordered product of operators can be obtained from it.71) 9. . T J lim −i E0 (T ′ −T ) φ (Q′ ) φ∗ (Q) 0 0 T ′ → −i∞ e T → i∞ dq ′ dq φ∗ (q ′ . such that t′ → −i∞ t → i∞ ↔ ↔ ˜ t → −∞. Q(tn )|0 . T′ Z[J] = lim T ′ → −i∞ T → i∞ Dq exp i T dτ L q.70) one immediately ﬁnds δ n Z[J] δJ(t1 ) . T ′ |Q. The generating functional precisely represents the vacuum to vacuum amplitude from initial (’in’) to ﬁnal(’out’) situation in the presence of a source. . thus T →i∞ lim q. J (9. eiE0 T φ∗ (Q) 0 We deﬁne the generating functional Z[J] as Z[J] = Q′ . . Since we have (neglecting multiplicative factors). (9. δJ(tn ) hence the name generating functional. T = φ0 (q. ˜ t′ → ∞. Better is the use ˜ of imaginary time t = −it. t 0 0out |0in J . 9.Path integrals and quantum mechanics t 82 T t t’ T’ Figure 9. J=0 (9. = (i)n 0|T Q(t1 ) . in which case eiEn T → e−En ·∞ and the term eiE0 T is dominant. t). t|Q. .68) (9.70) for the generating functional is in fact ill-deﬁned. .67) = = φ0 (q. t′ |q. dq dτ + J(τ )q(τ ) .6 Euclidean formulation The above expression (Eq.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 ′ ).2: Analytic continuation of boundaries T and T ′ We can project out the groundstate by an analytic continuation in the time.

y) = ∗ m. ensuring convergence for the functional integral ZE [J]. ˜ ˜ ZE [J] δJ(t1 ) .n fm (x) Kmn fn (y).1 Convince yourself by choosing an appropriate (orthonormal) set of functions that 1 Dα exp − α K α 2 for real-valued functions. δ n Z[J] 1 Z[J] δJ(t1 ) . As discussed in the previous section the interesting quantities are obtained from functional diﬀerentiation with respect to the sources. i τ −∞ ∞ −∞ dq d˜ τ + J(˜)q(˜) τ τ dq d˜ τ − J(˜)q(˜) τ τ . . det K 1 . In all of these cases the determinant of the (hermitean) operator K is deﬁned as det K ≡ p Kp . ˜ dt (9. for Grassmann valued functions.Path integrals and quantum mechanics 83 In terms of the imaginary time we can write the Euclidean generating functional.75) = − LE q. Exercises Exercise 9. δJ(tn ) = (i)n J=0 1 δ n ZE [J] . δJ(tn ) J = 0 ˜ t = it (9. + V (q). i dq .74) − V (q) 2 (9.2 Check the examples of functional derivation for real-.72) (9.76) which is a positive deﬁnite quantity. τ dq ˜ dt ≡ −L q. . .and Grassmann-valued functions.73) d˜ LE q. dq dt = 1 2 1 2 dq dt 2 Dq exp Dq exp − ∞ d˜ L q. . . 1 = √ . (9. where Kp are the eigenvalues of the matrix Kmn in K(x. dq ˜ dt = 1 2 dq ˜ dt dq ˜ dt 2 − V (q). (9. The diﬀerentiations in Z[J] and ZE [J] are related. 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. e. Dα Dα∗ exp (−α∗ K α) = for complex-valued functions and Dθ∗ Dθ exp (−θ∗ K θ) = det K. det K Exercise 9.g. complex. when L q.

det K Dθ∗ Dθ exp (−θ∗ K θ − θ∗ η − η ∗ θ) = det K exp η ∗ K −1 η . z) = δ(x − z). where dy K(x. y) K −1 (y.80) Dα∗ Dα exp (−α∗ K α − α∗ ω − ω ∗ α) = 1 exp ω ∗ K −1 ω .Path integrals and quantum mechanics 84 Exercise 9.79) (9.3 Prove the following relations 1 Dα exp − α K α − αω 2 1 = √ exp det K 1 ω K −1 ω . 2 (9.78) (9. .

. t) is considered as a set of quantum operators φx (t) in the Heisenberg picture and t′ φ′ . i. . . ∂µ φ) − Jφ] (10. . . t = = Dφ DΠ exp i t σ′ ˙ dτ d3 x Π(x)φ(x) − H (φ. (10.3) The Euclidean formulation is as before. xn ).e. .4) J=0 . δJ(xn ) 85 ≡ G(n) (x1 . .2) = ∞ −∞ d4 xE [LE (φ.1) Dφ exp i σ d4 x L (x) and T′ Z[J] = lim T → i∞ T ′ → −i∞ Dφ exp − Dφ exp i T d4 x [L (φ. x′ . φ(xn )|0 = (−i)n δ n Z[J] δJ(x1 ) . ∂µ φ) + Jφ] (10. t′ |φ. x. implying at the level of four vectors in coordinate and momentum space (when k · x ≡ kE · xE ) for instance x4 = i x0 = i t x · x = (x0 )2 − d4 xE = i d4 x L = 1 2 3 i 2 i=1 (x ) k 4 = −i k 0 =− 4 µ 2 µ=1 (x ) k · k = (k 0 )2 − d4 kE = −i d4 k 3 i 2 i=1 (k ) =− 4 µ 2 µ=1 (k ) ∂µ φ∂ µ φ − 1 2 M 2 φ2 LE = 1 2 µ (∂E φ)2 + 1 2 M 2 φ2 Furthermore we ﬁnd the n-point Green functions 0|T φ(x1 ) . Π) (10. φ(x.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. .Chapter 10 Feynman diagrams for scattering amplitudes 10.

xn ) = x2 xn . These Green functions appear in the expansion Z[J] = n in n! d4 x1 . . xn ) J(x1 ) . ∆F (the socalled Feynman propagator) is the Green’s function of the Klein-Gordon equation. . one immediately sees that i∆F (x1 − x2 ) = δ 2 Z0 [J] (−i)2 Z0 [0] δJ(x1 ) δJ(x2 ) = G0 (x1 − x2 ) ≡ (2) x1 x2 (10. .12) d4 k −i k·x ˜ e ∆(k) (2π)4 (10. . which can be written as ∆(x) = ˜ with (k 2 − M 2 )∆(k) = 1 or ˜ ∆(k) = 1 1 1 = 0 2 .Feynman diagrams for scattering amplitudes 86 for which we introduce the picture x1 G(n) (x1 . (10.4) one distinguishes the retarded Green’s function ∆R (x) = d4 k e−i k·x . . . 2 2 → (10. .5) Z0 [J] for the free scalar ﬁeld For the (free) scalar ﬁeld lagrangian L = ← 1 1 ∂µ φ∂ µ φ − M 2 φ2 . (2π)4 (k 0 + iǫ)2 − E 2 (10.9) i.e. with Z0 [0] = 1.8) 1 d4 x φ (∂µ ∂ µ + M 2 ) φ − Jφ 2 K (10. (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]) . . J(xn ). .7) where we made the choice Z0 [0] = 1 and ∂µ ∂ µ + M 2 ∆F (x − y) = −δ 4 (x − y).10) J=0 In order to determine ∆F . . d4 xn G(n) (x1 . (10. Furthermore.6) the generating functional is (using ∂µ = − ∂µ ) given by Z0 [J] = Dφ exp −i where Kxy = (∂µ ∂ µ + M 2 ) δ 4 (x − y).11) Depending on the path choosen in the complex k 0 plane (see exercise 7. consider the general solution of ∂µ ∂ µ + M 2 ∆(x) = −δ 4 (x).13) . . As discussed in the previous section (exercise 9. = 0 2 2 2 k2 − M 2 (k ) − E 2 (k ) − k − M (10. .

t) = 0 for t < 0.19) φ ∂E ∂E − M 2 φ + Jφ 2 1 d4 xE d4 yE J(x) (−i ∆F (x − y)) J(y) . 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).4).14) They are solutions of the inhomogeneous equation.15) e−i k·x d4 k . the second is to explicitly consider 0|T φ(x1 ) φ(x2 )|0 .16) C where C is a closed contour in the complex k 0 -plane. The contours for i ∆(x) i ∆+ (x) i ∆− (x) = [φ(x).Feynman diagrams for scattering amplitudes 87 satisfying ∆R (x. (10. The ﬁrst possibility is to consider the well-deﬁned Euclidean formulation. φ− (0)] = (2π)3 2E d3 k e−i k·x = [φ− (x). 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.18) Thus we see various solutions. µ 1 1 • Firstly.17) d4 k e−i k·x . (2π)4 (k 0 − iǫ)2 − E 2 (10.1 for countours) µ (∂E )2 − M 2 (i ∆F (x − y)) = −δ 4 (x − y) 0 kE =∞ (10. the generating functional can be written 1 µ µ Z0 [J] = Dφ exp d4 xE (10. (2π)4 k 2 − M 2 (10.23) k0 =i∞ k0 =−i∞ d4 k e−i k·x . (2π)4 k 2 − M 2 where the latter contour can be deformed to the contour for ∆C . ∆F = ∆C .20) = exp − 2 where or (see ﬁg. 4 k 2 − M 2 + iǫ (2π) (10.e. i. the advanced Green’s function ∆A (x) = satisfying ∆A (x.22) (10. The solution of the homogeneous equation ∂µ ∂ µ + M 2 ∆(x) = 0 can also be written as an integral in k-space. t) = 0 for t > 0. φ(0)] = i∆+ (x) + i∆− (x) d3 k e−i k·x = [φ+ (x). It is straightforward (closing contours in the appropriate half of the complex plane) to prove for instance that i ∆C (x) = θ(x0 ) i ∆+ (x) − θ(−x0 ) i ∆− (x). (2π)4 k 2 − M 2 (10. the causal Green’s function ∆C (x) = and ﬁnally the Green’s function d4 k e−i k·x = 4 (k 0 )2 − E 2 + iǫ (2π) ∆(x) = PV d4 k e−i k·x .21) ∆F (x) = = −i 0 kE =−∞ d4 kE e−i kE ·xE 2 (2π)4 kE + M 2 (10. φ+ (0)] = (2π)3 2E are also shown in exercise (7. (10. ∆(x) = − d4 k e−i k·x . . 10.

φ− (y)]|0 − θ(y 0 − x0 ) 0|[φ− (x).24) Knowing the explicit form of Z0 [J] it is straightforward to calculate the 4-points Green’s function. x2 . because this is the only quantity entering Z0 [J]. . we have G0 (x1 . x3 .1: Contour in the k 0 plane for the Feynman propagator in Euclidean and Minkowski space • As the second possibility. δJ(x4 ) 2 J=0 = − [∆F (x1 − x2 ) ∆F (x3 − x4 ) + ∆F (x1 − x3 ) ∆F (x2 − x4 ) + ∆F (x1 − x4 ) ∆F (x2 − x3 )] . x4 ) = 0|T φ(x1 ) φ(x2 ) φ(x3 ) φ(x4 )|0 1 δ4 exp − i J ∆F J = (−i)4 δJ(x1 ) . φ+ (y)]|0 = θ(x0 − y 0 ) i ∆+ (x − y) − θ(y 0 − x0 ) i ∆− (x − y) (10. we 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. . Neglecting multiplicative factors or equivalently assuming that Z0 [0] = 1. 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. the same result as above.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). = θ(x0 − y 0 ) 0|[φ+ (x). (4) (2) It should be clear that G0 can be expressed in terms of G0 = i ∆F . or diagrammatically (4) (10.25) x1 x2 x3 = x4 + + .

with propagators (i ∆F ) being connected to vertices (i LI ) according to the above expression for the functional Z[J]. (10. ≡ −i g d4 z (10. x3 . (2) (0) (10. Consider as an example the interaction LI (φ) = − g 4 φ 4! in the scalar ﬁeld theory discussed sofar.Feynman diagrams for scattering amplitudes 89 10. (10. x2 ) = i ∆F (x1 − x2 ) = G(4. x2 ) = G0 (x1 .g ) (x1 .e.26) Dφ exp i = exp i d4 z LI 1 δ i δJ(z) d4 x d4 y J(x) ∆F (x − y) J(y) (10. i. the generating functional can be written.30) and in zeroth order G(0.31) G(2. L (φ) = L0 (φ) + LI (φ).g ) (x1 . (10. x3 . x4 ) = To ﬁrst order in g one has Z (g ) [J] = = 1 0 0 x1 + + x2 (10. x2 . x4 ) = G0 (x1 .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 . Z[J] = = Dφ exp i exp i d4 z LI (10.29) This expression will be the one from which Feynman rules will be derived.34) Introducing a vertex point to which four propagators are connected.2 Generating functionals for interacting scalar ﬁelds d4 x [L0 (φ) + LI (φ) + Jφ] 1 δ i δJ(z) d4 x [L0 (φ) + Jφ] (10.32) (10.35) z .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 . To zeroth order in the coupling one has 0 1 Z (g ) [J] = Z0 [J] = exp − i 2 J ∆F J .g 0 ) = G0 = 1.27) When interactions are present. x2 .

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

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

out|p′ . φout and the interpolating ﬁeld φ(x). translation invariance). ∂µ ∂ µ + M 2 φin (x) = 0.47) . in|S = β. out| = eiϕ0 0. out| ⇔ |α.45) x2 as the only surviving diagram (see exercises). out = δαβ ↔ SS † = 1. . then (n) (n) Gsc = Gc for n ≤ 3. we have W0 [J] = − 1 2 d4 x d4 y J(x) ∆F (x − y) J(y). in = S|α. Next. out|α.g ) (x1 . in = p. out .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 ). out|α. implying the absence of tadpoles. the vacuum expectation value of the ﬁeld φ(x) is zero. in = p. in ⇔ β. p′ . . this will be translated to the action on ﬁelds.44) J=0 For the interacting theory. (10. out = p|p′ . p. x1 G(4. x4 ) = c 1 x3 z (10. . Proof: α. one ﬁnds for the connected 4-point Green’s function at order g. in| (choose ϕ0 = 0).3 Interactions and the S-matrix The S-matrix The S-matrix transforms initial state free particle states (in-states) |α. in|SS † |α. For the free scalar theory. in|p′ . in into ﬁnal state particle states (out-states) |β. e. in|S = 0. out = |p′ . .46) The properties of the S-matrix are (1) The vacuum is invariant or |S00 | = 1.Feynman diagrams for scattering amplitudes 92 i. Proof: 0. (10. out|p′ . in = p. thus β. out . pn . x3 . (10. out (suppressing except momenta all other 1 m quantum numbers). in|S|p′ .g. in = |α. . . (2) The one-particle state is invariant (conservation of energy and momentum. (3) S is unitary (it conserves the scalar product from initial to ﬁnal state). x2 . in = β. Sβα ≡ β. In line with the consideration of the generating functional representing the vacuum to vacuum amplitude we consider ﬁelds φin . x4 10. 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 .e. (2) (10. in = |p1 . out| and S † |α. considered explicitly. in|S = α. Also for free ﬁeld (Heisenberg) operators a distinction is made between φin and φout .

out|φout = β. this can actually not be used as it would imply [φ(x). The relation between S and Z[J] To establish this relation. The relation between S-matrix and in. which in a later stage (renormalization) will become more important. i. φ(y)] = Z [φin (x).and out-ﬁelds is: φin (x) = S φout (x)S −1 . 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.51) (10. as it stands. implies the strong (operator) convergence φ(x) → Zφin (x). z) φ(z). a)SU −1 (Λ. φin (y)] = iZ ∆(x − y). Proof: β. a causality condition that can be proven to imply the absence of interactions. S is invariant under e Poincar´ transformations: U (Λ. The convergence therefore must be weakened to t→−∞ √ (10. z) φ(z) d4 y d4 z ∆A (x − y) K(y.54) .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) . β. e.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). 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) (10.50) = φ(x) = √ Zφin (x) − √ Zφout (x) − d4 y d4 z ∆R (x − y) K(y. √ Although the above.49) The ﬁeld φ can be expressed in φin/out using retarded or advanced Green’s functions. (10. in|φin S → φin S = S φout (x). 2 ∂µ ∂ µ + M0 φ(x) = J(x). in|Sφout Finally we check that as expected S does not spoil Poincar´ invariance. (2π)3 2E (10. (10. 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.52) where Z is a constant.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.g. φin (x) = d3 k ∗ a(k) fk (x) + a† (k) fk (x) . 9.e. the source term Jφ in the lagrangian density is treated in the interaction picture. The time evolution operator (see Eq. a) = S. (10.

e = i. S U [J]] = √ Z δ S U [J]. y) δ δJ(y) : F [J]. we can. z) δU [J] δJ(z) δU [J] d4 y d4 z ∆A (x − y) K(y. φ(y)] f (y) e e . Thus applied to the ﬁeld φ(x) = φ+ (x) + φ− (x). B + C]e e . B] and [A. z) δU [J] . 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. φ(x). C] are c-numbers. (10. δJ(z) (10.57) i [φin (x).61) e φ+ f where the normal ordered expression : e φ f : is used. just as we did for 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). Thus S U [J] =: exp 1 √ Z d4 x d4 y φin (x) K(x. express it in terms of advanced and retarded Green’s functions. δJ(z) Taking the diﬀerence between the two expressions √ Z (φout (x) U [J] − U [J] φin (x)) =i or d4 y d4 z (∆R (x − y) − ∆A (x − y)) K(y. e e ] = [A. Noting that 0| : eA : |0 = 1 it follows that F [J] = 0|S U [J]|0 = 0|U [J]|0 = 0out |0in ∝ Z[J].e. one has [A. [A. B] is a c-number). φ(x). provided [A. which is equal to the expression e in which the creation operators are placed left of annihilation operators. z) .59) (10. t i Ut∞ [J] φ(x)U−∞ [J] (10. B]eB . : e d4 z φ(z) f (z) d4 x φ− (x) f (x) B C B C (10.56) Since δU [J]/δJ(x) satisﬁes the same equations as φ(x) and we know the limits. z) d4 y d4 z ∆(x − y)K(y. eB ] = [A.58) In order to ﬁnd a solution to this equation note that.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).62) where F [J] is some (arbitrary) functional. δJ(x) ∞ or for U [J] ≡ U−∞ [J] ′ (10.63) . (10. φ− f (10.Feynman diagrams for scattering amplitudes 94 and satisﬁes the property T ′ δUT [J] t = i UtT [J] φ(x)UT [J]. B] (for the case that [A. δJ(z) (10. with the use of the Baker-Campbell-Hausdorﬀ formula e−B A e+B = A + [A. : =i d4 y ∆(x − y) f (y) : e d4 z φ(z) f (z) :.

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

Note that in calculating amputated Green’s functions external lines are neglected. = −i g (multiplied with 4! corresponding to the allowed number of permutations of identical particles). Veltman. ’t Hooft and M. Diagrammar) in the case of the interaction terms g f (10.Feynman diagrams for scattering amplitudes 96 (in fact the inverse of the operator found in the quadratic term is for real scalar ﬁelds also multiplied by a factor 2. the factor 2 corresponds to the two-points Green’s function having two identical ends).69) LI (φ) = − φ3 − φ4 . or calculating full Green’s functions external lines are treated as propagators. 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. which cancels the factor 1/2 in the quadratic piece. For the symmetry factor consider the examples (given in G. (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). If problems arise go back to the deﬁning expression for the generating function in 10. 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.66 and 10.27. but overall momentum conservation at a vertex is understood. for example up to order g 2 the scattering of two neutral spin 0 particles (real scalar ﬁeld) is described by + + + (Rule 2) The contribution of each diagram is obtained by multiplying the contributions from propagators.27. vertices and external particle wave functions in that diagram. = −i g . to be precise iLI vertices in momentum space are introduced. At these vertices each line can be assigned a momentum. 3! 4! The vertices are: = −i f Consider ﬁrst the lowest order self-energy diagram. For the interaction terms in the lagrangian.

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

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

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

s4 )γµ 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. where u ¯ L(m) µν = = 1 2 u(k ′ . s3 )γ µ u(k. s2 ).s2 .s′ ′ ′ = 2 kµ kν + 2 kν kµ + q 2 gµν 1 Tr [(/ ′ + m)γµ (/ + m)γν ] k k 2 ′ ′ = 2 kµ kν + kν kµ − gµν (k · k ′ − m2 ) (10. s1 ) ¯ −igµν u(p′ . s′ )γµ u(k. q 4 µν [using that (¯(k ′ )γµ u(k))∗ = u(k)γµ u(k ′ )].Feynman diagrams for scattering amplitudes 100 10.s4 ) and averaged over spins in the initial state (1/2 × 1/2 × s1 .5 Some examples eµ scattering The ﬁrst example is the electromagnetic scattering of an electron and a muon.79) .76) 1 . s)¯(k. s4 )(−ie)γ ν u(p.78) = 2 α2 2 s + u2 + 4t(M 2 + m2 ) − 2(M 2 + m2 )2 . If we are interested in the scattering process of an unpolarized initial state and we are not interested in the spins in the ﬁnal state we need |M |2 summed over spins in the ﬁnal state ( s3 . s1 ) u e2 u(p′ . ¯ q2 (10.s3 . t2 (10. To lowest order (∝ α = e2 /4π) only one diagram contributes. 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 ′ .77) Combining Lµν and Lµν (M) one obtains L(m) Lµν (M) = 2 s2 + u2 + 4t(M 2 + m2 ) − 2(M 2 + m2 )2 µν and M 4π 2 (m) (10. s)γν u(k ′ . s2 )|2 ¯ q2 (10. s′ ) ¯ u s.s2 ) which can be written as (see also chapter 4) |M |2 = = 1 4s |¯(k ′ . s3 )(−ie)γ µ u(k.s4 e4 (m) µν (M) L L .

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

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

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

m2 . mb and s).8) (11. Prove that trf Ea = |ptrf | = a s − m2 − m2 a b .11) (11. 4s (11. for instance. In that case trf pa = (Ea . = 2 s λ(s. Also in this case one can express the energy and momentum in the invariants. m2 ) a b .12) pb = (mb . p1 .Scattering theory 104 It is straightforward to prove that the unknowns in the particular system can be expressed in the invariants (ma . m2 . 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. which in the speciﬁc case a b also can be expressed as λ(s. m2 ) is a function symmetric in its three arguments. 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. 0). m2 .14) One can. 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. 2mb λ(s.13) (11. m2 ) a b .9) (11.10) The function λ(s. 4π s 4π 8π s 4π (11. Prove that |q| = cm Ea cm Eb (s − m2 − m2 )2 − 4 m2 m2 a a b b = 4s s + m2 − m2 a b √ = . m2 ). p2 ) = = where λ12 denotes λ(s. m2 . 2mb (11. m2 ) = 4(pa · pb )2 − 4p2 p2 . trf Ea (threshold) = 1 2 mb ( i mi )2 − m2 − m2 a b . 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 √ = . p1 .15) Explicit calculation of the reduced two-body phase space element gives dR(s.16) . (11. ptrf ). m2 . 2 s s − m2 + m2 a b √ . a (11.

21) (11. 11.19) (11. usually many particles are produced. The initial state. Of course there are not only 2-particle channels. 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 . where the ﬁrst is referred to as elastic channel and the set of all other channels as the inelastic channels.. Writing E = mT cosh y.27) with m2 = m2 − p2 = m2 + p2 + p2 .17) (11. in exclusive measurements all particles are detected. A speciﬁc reaction channel starts contributing at the threshold value (Eq. however. dt = 2 qq ′ d cos θcm and obtain for the two-body phase space element √ q ′ dΩcm λcd dΩcm √ = (11. for instance the momenta of incoming (colliding) hadrons and it is useful to use for the produced particle rapidity as a variable. Consider the process a + b → c + d. (11.25) dR(s.18) (11.26) = 16π q s 8π λab Kinematics of inclusive hard scattering processes In high energy (hard) scattering processes. pz = mT sinh y. py ).4). The variable s is always larger than a c b d the minimal value (ma + mb )2 . An often used set of invariants are the Mandelstam variables. 2s (11. The various possibilities are referred to as diﬀerent reaction channels. At high energies. = (11. s = (pa + pb )2 = (pc + pd )2 t = (pa − pc ) = (pb − pd ) u = (pa − pd )2 = (pb − pc )2 2 2 (11.22) (11. Consider the 1-particle inclusive case. pc . 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 . pT = (px . Examples appear in → π− + p (11.24) Using the relation between t and cos θcm it is straightforward to express dΩcm in dt. in which one particle is detected.23) which are not independent as s + t + u = m2 + m2 + m2 + m2 . H1 + H2 → h + X. usually is a 2-particle state. tmax min λcd /4s. there is usually a preferred direction. a c with q = λab /4s and q ′ = being 0 or 180 degrees. E .20) → π0 + n → π+ + π− + n → . Instead of the scattering angle.Scattering theory 105 Kinematics of 2 → 2 scattering processes π− + p The simplest scattering process is 2 particles in and 2 particles out. while the ﬁnal state often arises from a series of 2-particle processes combined with the decay of an intermediate particle (resonance).. pd ) = 4π s 4π 8π s 4π dt dt √ √ . In inclusive measurements no particles are detected in the ﬁnal state. the rapidity y is deﬁned T T x y y= 1 ln 2 E + pz E − pz = ln E + pz mT = tanh−1 pz . .

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

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

(11. .T V. .Scattering theory 108 Although this at ﬁrst sight does not look covariant. dt (11. Microscopically one has Ndecay = Na · Γ T or Γ= N 1 .34) i.e. (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.T d4 x = V · T (2π)4 δ 4 (Pi − Pf ). the amount of decaying particles is proportional to the number of particles with proportionality factor the em decay width Γ. N and T · V are covariant. it is.35) This quantity is not covariant. σ= √ 2 λab T · V (11. For this we employ in essence Fermi’s Golden rule stating that when the S-matrix element is written as Sf i = δf i − (2π)4 δ 4 (Pi − Pf ) iMf i (11.38) (in which we can calculate −iMf i using Feynman diagrams). 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 .33) Decay of particles For the decay of particle a one has macroscopically dN = −Γ N. From the solution N (t) = N (0) e−Γ t one knows that the decay time τ = 1/Γ. For the (proper) decay width one thus has 1 N Γ0 = . . the number of scattered or decayed particles is given by 2 N = (2π)4 δ 4 (Pi − Pf ) iMf i dR(p1 . pn ). as expected. (11.37) 2 ma T · V Fermi’s Golden Rule In both the scattering cross section and the decay constant the quantity N/T V appears. N 1 .36) (11. . T · V ρa (11. 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 . .39) One of the δ functions can be rewritten as T · V (remember the normalization of plane waves).

41) (11. . for which the √ amplitudes squared have been calculated in the previous chapter. pn ) |M |2 q 32π 2 m2 dΩ |M |2 .43) dΩcm = π M (s. √ (2ℓ + 1)Mℓ (s) Pℓ (cos θ). i. pn ). . .44) This can be used to get the full expression for dσ/dt for eµ and e+ e− scattering. θ) or M (q i . . respectively. θ) = −8π s ℓ (in analogy with the expansion for f (E. 8π s 2π 8π s . implies for the scattering matrix M . . i M† M √ − √ 8π s 8π s = fi n dΩn ∗ Mnf qn Mni √ √ . 11. (11. θ)|2 . q n ) q n qn √ Mni (q i . p1 . . 16π 2 s (11. .40) T ·V Combining this with the expressions for the width or the cross section one obtains for the decay width Γ = 1 2m = dR(m2 .47) M (s. θ) in quantum mechanics.Scattering theory 109 (Using normalized wave packets these somewhat ill-deﬁned manipulations can be made more rigorous). q f ). pn ). (11.46) Partial wave expansion Often it is useful to make a partial wave expansion for the amplitude M (s. δf n + i(2π)4 δ 4 (Pf − Pn ) (M † )f n or −i Mf i − (M † )f i = − n δni − i(2π)4 δ 4 (Pi − Pn ) Mni = δf i . .4 Unitarity condition (S † )f n Sni = δf i The unitarity of the S-matrix. (11. (11.e. . . t) λab 4π dt. θ).25) −i Mf i − (M † )f i = − ∗ dΩ(ˆn )Mnf (q f . q n ).42) 2−body decay The diﬀerential cross section (ﬁnal state not integrated over) is given by 1 dσ = √ |Mf i |2 dR(s. . p1 . (11.45) Since the amplitudes also depend on all momenta the full result for two-particle intermediate states is (in CM. see 11. The amplitude −M /8π s is the one to be compared with the quantum mechanical scattering amplitude f (E. ˆ ˆ Inserted in the unitarity condition for M . . 2 λab and for instance for two particles dσ = q ′ M (s. θcm ) √ q 8π s 2 2 (11. p1 . note the sign and cos θ = qi · qf ). (M † )f n (2π)4 δ 4 (Pi − Pn ) Mni . The result is N = |Mf i |2 dR(s. for which one has dσ/dΩ = |f (E. The sign diﬀerence comes from the (conventional) sign in relation between S and quantummechanical and relativistic scattering amplitude.

2i q 2i q (11. q ˆ Pℓ (ˆ · q ′ ) = q ˆ (ℓ) m 4π Y (ℓ) (ˆ) Ym (ˆ′ ) q (ℓ)∗ q 2ℓ + 1 m and the orthogonality of the Ym functions to prove that RHS = 2 n ℓ † (2ℓ + 1)Pℓ (ˆi · qf ) (Mℓ )f n qn (Mℓ )ni . (11. θcm ) √ q 8π s 2 . (11. Show that σT = = 4π q 2π q2 (2ℓ + 1) Im Mℓ (s) ℓ ℓ (2ℓ + 1) (1 − ηℓ cos 2δℓ ).50) where Sℓ (s) satisﬁes |Sℓ (s)| = 1 and δℓ (s) is the phase shift. Using σ= dΩ q ′ M (s.e. q ˆ i. 0). n (11. (11. which allows writing Mℓ (s) = Sℓ (s) − 1 e2i δℓ (s) − 1 = .Scattering theory 110 we obtain LHS = −i while for the RHS use is made of ℓ † (2ℓ + 1) Pℓ (ˆi · qf ) (Mℓ )f i − (Mℓ )f i . the total cross section can be determined. In general a given channel has |Sℓ (s)| ≤ 1.49) or Im Mℓ = q |Mℓ |2 .48) If only one channel is present this simpliﬁes to ∗ ∗ −i (Mℓ − Mℓ ) = 2 qMℓ Mℓ .51) and for the case that this is the only channel (purely elastic scattering. σinel = π q2 2 (2ℓ + 1) (1 − ηℓ ).53) The diﬀerence is the inelastic cross section. η = 1) the result σel = 4π q2 (2ℓ + 1) sin2 δℓ .52) From the imaginary part of M (s. in combination with the partial wave expansion for the amplitudes M and the orthogonality of the Legendre polynomials immediately gives for the elastic channel. † −i (Mℓ )f i − (Mℓ )f i = 2 † (Mℓ )f n qn (Mℓ )ni . ℓ (11. σel = 4π ℓ (2ℓ + 1)|Mℓ (s)|2 (2ℓ + 1) ℓ = 4π q2 ηℓ e 2i δℓ − 1 2i 2 . parametrized as Sℓ (s) = ηℓ (s) exp(2i δℓ (s)). (11.54) ℓ .

it is straightforward to write down the partial wave amplitude. i.56) (again disregarding spin). The quantity Γ is precisely the width for unstable particles. . . 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 . the time-evolution operator. but still real. For an unstable particle one expects that U (t.e. To check unitarity we need the amplitude at order e4 . This is achieved with a propagator i∆R (k) = i k2 − M2 + iM Γ (11.Scattering theory 111 Note that the total cross section is maximal in the case of full absorption. however. in which case. one has poles at k 0 = ±(Ek − iǫ) where Ek = + k2 + M 2 guarantees the correct behavior. The product of amplitudes is of order e4 . 0)|2 = e−Γt . pn )MRn (2π)4 δ 4 (k − Pn ) MnR = 2M n Γn = 2M Γ. i. (11. σel = σinel . speciﬁcally one has for t > 0 that the Fourier transform is dk e 0 −ik0 t ∆(k) ∝ e−ik t dk (k 0 − Ek + iǫ)(k 0 + Ek − iǫ) 0 0 (t>0) ∝ e−iEk t . The prescription for the pole structure. s − M 2 + iM Γ (11. 0) ∝ e−i(E−iΓ/2)t . 0).57) This shows that Γ is the width of the resonance. . such that |U (t. 11.55) (note that we have disregarded spin). The unitarity condition then states 2 Im ∆−1 (k) = R n † dR(p1 . For instance the tree-level calculation of Feynman diagrams at order e2 was real. which is given by a sum of the partial widths into the diﬀerent channels. We note that unitarity is generally broken in a ﬁnite order calculation.e. ∝ U (t. For the amplitude in a scattering process going through a resonance. 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.5 Unstable particles i k 2 − M 2 + iǫ For a stable particle the propagator is i∆(k) = (11. 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 is what we will do to discuss decay widths.58) . η = 0. . (q Mℓ )ij (s) = −M Γi Γj .

M2 − s (11. u) + f (u. The phase shift in the resonating channel near the resonance is given by tan δℓ (s) = MΓ .60) showing that the phase shift at resonance rises through δ = π/2 with a ’velocity’ ∂δ/∂s = 1/M Γ. The cross section σT (π − p) also shows a resonance with the same width with a value of about 70 mb. Limiting ourselves to a resonance in one channel. t) + g(u. which is attributed to neutrinos. dt s(s − 4m2 ) with f (t.Scattering theory 112 (Prove this using the unitarity condition for partial waves). its width Γ = 120 MeV. Exercises Exercise 11. 2 q (s − M 2 )2 + M 2 Γ2 (11.1 Show that the cross section for electron-electron scattering (exercise 10. All these decays can be seen and leave an ’invisible’ width of 498 MeV.4) can be written as 4πα2 dσ = {f (t. Its mass is M = 1232 MeV. • The Z 0 resonance in e+ e− scattering with M = 91. i. Note that because of the presence of a background the phase shift at resonance may actually be shifted.88 MeV. This is a narrow resonance discovered in 1974. it is furthermore easy to prove that the cross section is given by σel = M 2 Γ2 4π (2ℓ + 1) .49 GeV. Γ = 2. The cross section at resonance is about 30 nb. Its mass is M = 3096. u) = 1 1 2 (s + u2 ) + 4m2 (t − m2 ) t2 2 2 1 1 ( s − m2 )( s − 3m2 ) tu 2 2 . At resonance the cross section σT (π + p) is about 210 mb. Three neutrinos explain the resonance shape. 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).59) reaching the unitarity limit for s = M 2 .26 keV. one can reconstruct the resonance shape for diﬀerent numbers of neutrino species. t)} .2 GeV. where furthermore σinel = 0. This characteristic shape of a resonance is called the Breit-Wigner shape. From this one sees that a resonance has the same shape in all channels but diﬀerent strength.e. • The J/ψ resonance in e+ e− scattering. Knowing that each neutrino contributes about 160 MeV (see next chapter). the partial width into e+ e− is Γee = 5. u) + g(t. Essentially this resonance can decay into quark-antiquark pairs or into pairs of charged leptons. the full width is Γ = 88 keV. Three famous resonances are: • The ∆-resonance seen in pion-nucleon scattering. u) = g(t. The half-width of the resonance is M Γ. a fast change in the phase shift for a narrow resonance.

Compare the result with the factor trf 1/ρa ρb va that we used in the derivation of the basic expression for cross sections.2 Calculate the ﬂux factor. As discussed for unstable particles. writing N (s) M1 (s) = 2 + iM Γ . cm |va − vb | in the center of mass system for two colliding particles a and b. Show that unitarity can be used one step further to ﬁx also the real part of the residue of the pole in the amplitude.Scattering theory 113 Exercise 11.999 877 Γ and Γ(π − → e− νe ) = 0. ρcm ρcm a b 1 cm .4 The ∆ resonance in N π scattering appears in just one partial wave (a P-wave) as a pole at the position 2 s = s0 = M∆ − iM∆ Γ∆ . Exercise 11. Thus.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. Compare the results with the experimental π − lifetime τ = 2. Calculate now fπ . s − M∆ ∆ ∆ 2 calculate Re N (M∆ ) . 4. a quantity which we will ¯ ¯ encounter in the next chapter (section 12.8.2).000 123 Γ.6 × 10−8 s and the relative decay widths Γ(π − → µ− νµ ) = 0. Exercise 11. we have here already used unitarity to ﬁnd a ﬁnite imaginary part that is proportional to the total decay width.

Chapter 12

**The standard model
**

12.1 Non-abelian gauge theories

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

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

a

La

φ(x) = (1 + i θa La ) φ(x)

inf

(12.2)

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

inf

(12.3)

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

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

The standard model

115

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

′a a c Wµ = Wµ − cabc θb Wµ +

i (∂µ U (θ)) U −1 (θ), g

(12.7)

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

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

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

(12.8)

(12.9) (12.10)

The gauge-invariant lagrangian density is now constructed as

Gµν → U (θ) Gµν U −1 (θ).

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

(12.11)

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

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

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

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

**The standard model
**

xψ x+dx

116

(x)

ψ(x)

x+dx ψ(x+dx)

**x-space internal space
**

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

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

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

**A geometric picture of gauge theories
**

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

1 The

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

λ7 =

i

−i

1 1 λ8 = √ 3

−2

The standard model the basis for x G and x+dx 117 G. A ’constant’ vector that only rotates because of the arbitrary deﬁnitions of local frames satisﬁes Dµ ψ(x) = 0. x x+dx ψ(x + dx) = (1 + dxµ ∂µ ) ψ(x). We note that local gauge invariance requires that we can modify all local systems with a (local) unitary transformation S(x). requiring that the ’connection’ Aµ (x) is such that Dµ → S(x)D µ S −1 (x).16.16) which in the presence of the ’connection’ Aµ diﬀers from the total change between x ψ(x) and x+dx ψ(x+ dx).e. i. ds which is solved by dψ(s) ds = dxµ . (12. should be independent of such transformations.18) or considering a path xµ (s) from a ﬁxed origin (0) to point x. What is this physical eﬀect by which the ’connection’ Aµ can be observed? For this consider the phase around a closed loop. x+dx ψ(x) = = 1 + ig dxµ Aa (x)Ta µ 1 + ig dx Aµ (x) µ x x ψ(x) (12. which connects two identical vectors. 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 .15) ψ(x). if two diﬀerent paths to the same point give diﬀerent phases the eﬀects can be observed. However. 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). The relation in Eq.14) (12. ∂µ − ig Aµ (x) ψ(x) = 0 dxµ ∂µ − ig Aµ (x(s)) ψ(x) = 0. In principle such a phase in a given point is not observable. (12.e. (12. but expresses them with respect to diﬀerent bases. = .19) This gives rise to a (path dependent) phase in each point. -dx -dy µ µ dyµ µ dx For a vector. Dν ] ψ(x) 1 − ig dσ µν Gµν ψ(x).17) The three equations given so far immediately give Dµ = ∂µ − ig Aµ (x). 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. ds s dxµ ds′ P exp ig A (s′ ) ψ(0). x+dx ψ(x + dx) = 1 + dxµ Dµ x+dx ψ(x). 12. (12. i.

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

(12. ϕ2 and η. 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 .23) F +η the latter only forming a minimum for m2 < 0. The classical groundstate appears degenerate. (12. It is suﬃcient to do this to second order in the ﬁelds as the higher (cubic.25) 2 2 2 Therefore there are 2 massless scalar particles. for the vacuum Qa |0 = 0. In order to see the consequences for the particle spectrum of the theory we construct the lagrangian in terms of the ﬁelds ϕ1 . In the case of degeneracy. The η massless particles are called Goldstone bosons. etc. 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. In this situation one speaks of spontaneous symmetry breaking. It is appropriate to redeﬁne the ﬁeld as ϕ1 φ = ϕ2 . i.e. A quantum ﬁeld theory has only one nondegenerate groundstate |0 . say 0 0 . The presence of a nonzero value for the classical groundstate value of the ﬁeld will have an eﬀect when the ﬁeld is quantized.) terms constitute interaction terms.21) Stability of the action requires the classical groundstate ϕc to have a well-deﬁned value (which can be nonzero). has less symmetry. In this mode the lagrangian and the vacuum are both invariant under a set of symmetry transformations generated by Qa .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. φ = ϕc + φquantum where for the (operator-valued) coeﬃcients in the quantum ﬁeld one wants 0|c† = c|0 = 0. ϕc = 0|φ |0 = (12. They are known as the Weyl mode or the Goldstone mode: Weyl mode. The original lagrangian contained an SO(3) invariance under (length conserving) rotations among the three ﬁelds. Any constant ﬁeld with ’length’ F is a possible groundstate. while the lagrangian including the nonzero groundstate expectation value chosen by nature. Writing the ﬁeld φ as a sum of a classical and a quantum ﬁeld. while the quadratic terms must correspond with non-negative masses. It is only invariant under rotations around the 3-axis. . The ﬁeld along the third axis plays a special role because of the choice of the vacuum expectation value. . . 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. In this case the spectrum is described .24) 1 2 2 4 1 1 1 = (∂µ ϕ1 )2 + (∂µ ϕ2 )2 + (∂µ η)2 + m2 η 2 + .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. so one has 0|φquantum |0 = 0 and hence 0|φ|0 = ϕc . (12.

H] = 0. and one must have mπa = 0.30) which in the case that we include only two ﬂavors form an SU (2)V symmetry (isospin symmetry) generated by the Pauli matrices. the lagrangian for the quarks consists of the free Dirac lagrangian for each of the types of quarks. 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. . . denoted |π a (k) . i. and |a and |a′ belong to the same multiplet (there is a Qa such that |a′ = Qa |a ) then H|a = Ea |a implies that H|a′ = Ea |a′ . .) one can write L = f ψ f (i/ − Mf )ψf . (12. it is easy to see that one gets ∂µ V µ = i ψ [M. A bit more formal. T = τ /2. This lagrangian density then is invariant under unitary (vector) transformations in the ﬂavor space.28) where ψ is extended to a vector in ﬂavor space and M is a diagonal matrix in ﬂavor space.e. suggesting the stronger relation ∂ µ Jµ (x) = fπ m2 a π a (x). Note that for the ﬁelds π a (x) one would have a the relation 0|π a (x)|π a (k) = ei k·x . V µ = ψγ µ T ψ. a ∂ µ Jµ = 0. [Qa .e. Taking the derivative. down.31) . Neglecting at this point the local color symmetry. ψu mu md ψ = ψd . In this mode the lagrangian is invariant but Qa |0 = 0 for a number of generators.. irrespective of the fact if they annihilate the vacuum. a and a′ are degenerate states. M = . . ∂ ∂ (12. i. This means that they are operators that create states from the vacuum.29) (Note that each of the entries in the vector for ψ in the above equation is still a 4-component Dirac spinor). Goldstone mode. Including a sum over the diﬀerent ﬂavors (up. a massless Goldstone boson for each ’broken’ generator.32) (12.27) If the transformations in the lagrangian give rise to a symmetry the Noether currents are conserved. T ] ψ. a 0|∂ µ Jµ (x)|π b (k) = fπ k 2 ei k·x δab = fπ m2 a ei k·x δab .26) for all the states labeled by a corresponding to ’broken’ generators. strange. π Example: chiral symmetry An example of spontaneous symmetry breaking is chiral symmetry breaking in QCD. As the a generators for a symmetry are precisely the zero-components of a conserved current Jµ (x) integrated a a over space. = ψ(i/ − M )ψ. i.e. there must be a nonzero expectation value 0|Jµ (x)|π (k) . π (12. (12. etc. 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.The standard model 120 by degenerate representations of the symmetry group. Known examples are rotational symmetry and the fact that the spectrum shows multiplets labeled by angular momentum ℓ (with members labeled by m). The conserved currents corresponding to this symmetry transformation are found directly using Noether’s theorem (see chapter 6). if the generators Qa generate a symmetry. The generators Qa (in that case the rotation operators Lz . (12. ψ −→ ei α·T ψ. called ﬂavors. Using the Dirac equation.

is by nature not realized in the Weyl mode. proton and neutron. Still the basic fact that the generators acting on the vacuum give a nonzero result (i. etc.e. including in Dirac space a γ5 matrix. (12. 12. nature has choosen the path SU (2)L ⊗ SU (2)R =⇒ SU (2)V . much smaller than any of the other mesons or baryons. both are very small. That the massless theory has this symmetry can also be seen by writing it down for the socalled lefthanded and righthanded fermions. Therefore realization in the Weyl mode would require that all particles come double with positive and negative parity. In the real world. etc. The chiral ﬁelds ψR and ψL are transformed into each other under parity. which is equivalent 2 to the V-A symmetry.3 MeV/c2 and Mn = 939.e. with almost degenerate masses (Mp = 938. for which the ﬁeld (according to the discussion above) has the same behavior under parity. as the quantity ∂µ Aµ (x). the lagrangian density is invariant under left (L) and right (R) rotations independently. but the fact that the symmetry is not perfect and the right hand side of Eq. Aµ = ψγ µ T γ5 ψ. chiral symmetry is not perfect. (leaving out the ﬂavor structure) the same as ψγ5 ψ. M = m · 1. while the groundstate is only invariant under isospin rotations (R = L).The standard model 121 Furthermore ∂µ V µ = 0 ⇐⇒ [M. parity minus). How can we see this. The currents corresponding to this symmetry transformation are again found using Noether’s theorem. . T } γ5 ψ. i. generated by TR/L = 1 (1 ± γ5 )T . i. or. This symmetry. The Goldstone bosons of QCD are the pions for which fπ = 93 MeV and which have a mass of mπ ≈ 138 MeV/c2 .e. is what happens in the real world. in terms of which the Dirac lagrangian 2 density looks like L = i ψL / ψL + i ψR / ψR − ψR M ψL − ψL M ψR . Note that these transformations also work on the spinor indices.27. From group theory (Schur’s theorem) one knows that the latter can only be true. however.35) In this case ∂µ Aµ = 0 will be true if the quarks have zero mass. fπ = 0 remains.6 MeV/c2 ). e. There exists another set of symmetry transformations. which is approximately true for the up and down quarks.e. gives also rise to a nonzero mass for the Goldstone bosons according to Eq. parity would not play a role in the world. if in ﬂavor space M is proportional to the unit matrix. We know that mesons and baryons (such as the nucleons) have a well-deﬁned parity that is conserved. 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. but also a triplet (isospin 1) of pions. socalled axial vector transformations. stated equivalently. This symmetry is realized in the Weyl mode with the spectrum of QCD showing an almost perfect isospin symmetry. it is easy to see that one gets ∂µ Aµ = i ψ {M. From the number of broken generators it is clear that one expects three massless Goldstone bosons.33) which in the case of two ﬂavors form SU (2)A transformations generated by the Pauli matrices.g. (12. This situation. nucleons and atomic nuclei has not only an isospin or vector symmetry SU (2)V but also an axial vector symmetry SU (2)A . behaves as a pseudoscalar particle (spin zero. (12. I.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.34) Using the Dirac equation. T ] = 0. a doublet (isospin 1/2) of nucleons. Therefore the world of up and down quarks describing pions. where the quark masses are not completely zero. i. ψ −→ ei α·T γ5 ψ. SU (2)V (isospin) symmetry is good if the up and down quark masses are identical. ψR/L = 1 (1 ± γ5 )ψ. This combined symmetry is what one calls chiral symmetry. 12. T γ5 = τ γ5 /2. ∂ ∂ (12.35 is nonzero.e.

. Therefore we can always rotate the ﬁeld φ into the 3-direction in order to simplify the calculation.39) (12. (12.41) φ3 F +η Explicitly one then has 2 2 gF Wµ + g Wµ η 1 1 −gF Wµ − g Wµ η Dµ φ = ∂µ φ + g Wµ × φ = ∂µ η . having two independent spin components) and three scalar ﬁelds (one degree of freedom each). Considering the same example with rotational symmetry (SO(3)) as for spontaneous symmetry breaking of a scalar ﬁeld (Higgs ﬁeld) with three components. again 9 degrees of freedom.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. 1 1 L = − Gµν · Gµν + Dµ φ · Dµ φ − V (φ). 0 φ= (12. 4 2 where Since the explicit (adjoint.e. 0 0 0 = . Dµ φ = Gµν = ∂µ φ + g Wµ × φ. two massive bosons (Wµ and Wµ with MW = gF ) and a massive scalar particle (η with m2 = η 2 −2 m . made into a gauge theory. i. Initially there are 3 massless gauge ﬁelds (each.42) 2 from which one reads oﬀ that the particle content of the theory consists of one massless gauge boson 3 1 2 (Wµ ). 0 ϕc = 0|φ |0 = 0 . One may wonder about the degrees of freedom. .38) ∂µ Wν − ∂ν Wµ + g Wµ × Wν . . but one has 1 massless gauge ﬁeld (2).The standard model 122 12. one sees that the ﬁelds Wµ and Gµν also can be represented as three-component ﬁelds. 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). in this case three-dimensional) representation reads (La )ij = −i ǫaij . thus 9 independent degrees of freedom. The diﬀerence comes when we reparametrize the ﬁeld φ.40) a Dµ φ = ∂µ φ − ig Wµ La φ. After symmetry breaking the same number (as expected) comes out. while the number of massive gauge bosons coincides with the number of ’broken’ generators. . as in this case there are no massless Goldstone bosons. The latter is a spin 0 particle (real scalar ﬁeld) called a Higgs particle. 2 massive vector ﬁelds or spin 1 bosons (2 × 3) and one scalar ﬁeld (1). like a photon. is made.37) (12. F which gives for the lagrangian density up to quadratic terms L 1 1 λ 1 = − Gµν · Gµν + Dµ φ · Dµ φ − m2 φ · φ − (φ · φ)2 4 2 2 4 1 2 2 1 1 2 = − (∂µ Wν − ∂ν Wµ ) · (∂ µ W ν − ∂ ν W µ ) − g F Wµ W µ 1 + Wµ W µ 2 4 2 1 + (∂µ η)2 + m2 η 2 + . We have the possibility to perform local gauge transformations. (12. The symmetry is broken in the same way as before and the same choice for the vacuum. (12.

44) which has an SU (2)W symmetry under transformations eiα·TW . νµ and ντ ) and their antiparticles. while the lefthanded particles form a doublet. etc. The lagrangian for the leptons consists of three families each containing an elementary fermion (electron e− .49) Next the SU (2)W ⊗ U (1)Y symmetry is made into a local symmetry introducing gauge ﬁelds Wµ and Bµ in the covariant derivative Dµ = ∂µ − i g Wµ · TW − i g ′ Bµ YW /2.e.47) and YW is considered as an operator that generates a U (1)Y symmetry. τ ). i.50) (12. explicitly L R SU(2)W SU(2)W −→ ei α·τ /2 L. In the standard model one starts with a very simple basic lagrangian for (massless) fermions which exhibits more symmetry than observed in nature. 2 (12. R −→ e−i β R.and righthanded particles.45) (12. For the neutrino only a lefthanded particle (and righthanded antiparticle) exist.51) . left. 0 1 +1/2 L νe 3 − = with TW = and TW = L= −1/2 eL L 2 R = R − = eR 3 with TW = 0 and TW = 0. explicitly Dµ L Dµ R i i g Wµ · τ L + g ′ Bµ L. etc. ψR/L = 2 (1 ± γ5 )ψ decouple. the leptons (elektrons. muon µ− or tau τ − ). however. = ∂µ L − (12. electromagnetic and weak forces.) and the gauge bosons responsible for the strong. muons. Thus L (f ) = i eR / eR + i eL / eL + i νeL / νeL + (µ.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. the quarks (up. U(1)Y U(1)Y (12. Thus the lagrangian density is L (f ) ∂ ∂ = i L/ L + i R/ R.43) and One introduces a (weak) SU (2)W symmetry under which eR forms a singlet. (12. 1 As they are massless. ∂ ∂ ∂ (12. By introducing gauge ﬁelds and breaking the symmetry a more complex lagrangian is obtained. neutrinos.The standard model 123 12. 2 2 = ∂µ R + i g ′ Bµ R. implies certain nontrivial relations between masses and mixing angles that can be tested experimentally and sofar are in excellent agreement with experiment. R. (12.4 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. under which the lefthanded and righthanded particles with YW (L) = −1 and YW (R) = −2 transform with eiβYW /2 . The procedure. This is written as 3 Q = TW + YW .48) (12.). its corresponding neutrino (νe . that gives a good description of the physical world. down. explicitly L −→ e−i β/2 L.

h) (12. with the following lagrangian density consisting of a symmetry breaking piece and a coupling to the fermions. (12. 4 4 In order to break the symmetry to the symmetry of the physical world. . L where L (h1) (h) =L (h1) +L (h2) . the U (1)Q symmetry (generated by the charge operator). For the classical ﬁeld the choice θ4 = v is made. leading to the parametrization 1 0 φ(x) = √ (12.58) (12. † i i g Wµ · τ − g ′ Bµ )φ. this leads to the lagrangian L (h1) . Using local gauge invariance θi for i = 1.55) 2 2 The Higgs potential V (φ) is choosen such that it gives rise to spontaneous symmetry breaking with ϕ† ϕ = −m2 /2λ ≡ v 2 /2. (12. . Putting this in leads to L (f ) =L (f 1) +L (f 2) . . . which assures with the choice of YW for the Higgs ﬁeld that Q generates the remaining U (1) symmetry. .59) . a complex Higgs ﬁeld + 1 φ √2 (θ2 + iθ1 ) 0 = 1 (12.The standard model 124 3 3 where Wµ is a triplet of gauge bosons with TW = 1. + 8 (12.52) 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 .57) = 1 g2 v2 1 2 (Wµ )2 + (Wµ )2 (∂µ h)2 + m2 h2 + 2 8 v2 2 3 gWµ − g ′ Bµ + . 2 and 3 may be eliminated (the necessary SU (2)W rotation is precisely e−iθ(x)·τ ). + 8 g2 v2 1 + − (Wµ )2 + (Wµ )2 (∂µ h)2 + m2 h2 + 2 8 (g 2 + g ′2 ) v 2 (Zµ )2 + .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.53) φ= √ (θ4 − iθ3 ) φ 2 with TW = 1/2 and YW = 1 is introduced. (12. TW = ±1 or 0 and YW = 0 (thus Q = TW ) and 3 Bµ is a singlet under SU (2)W (TW = TW = 0) and also has YW = 0.54) = (Dµ φ)† (Dµ φ) −m2 φ† φ − λ (φ† φ)2 . −V (φ) L and (h2) = −Ge (LφR + Rφ L).

60) (12. . 4 2 MW g2 v2 = . The coupling of the fermions to the physical gauge bosons are contained in L (f 1) giving L (f 1) = i eγ µ ∂µ e + i νe γ µ ∂µ νe − g sin θW eγ µ e Aµ g 1 1 + sin2 θW eR γ µ eR − cos 2θW eL γ µ eL + νe γ µ νe cos θW 2 2 g µ + µ − + √ νe γ eL Wµ + eL γ νe Wµ .62) g 2 + g ′2 3 ≡ sin θW Wµ + cos θW Bµ . 2 + g ′2 g 3 g ′ Wµ + g Bµ (12.68) 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 .The standard model 125 where the quadratically appearing gauge ﬁelds that are furthermore eigenstates of the charge operator are ± Wµ = = = Zµ Aµ 1 1 2 √ Wµ ± i Wµ .66) From the coupling to the photon. we can read oﬀ e = g sin θW = g ′ cos θW .69) .61) (12. g ′ /g = tan θW . for instance in the amplitude for the decay of the muon νµ µ− W − e− νµ = µ− e− νe νe is given by −i M = − ≈ i g2 −i gρσ + .65) The weak mixing angle is related to the ratio of coupling constants.64) (12. .67) The coupling of electrons or muons to their respective neutrinos. 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 . 2θ 2θ 4 cos W cos W 0. (12. (12. 2 (12.63) (12. σ (νµ γ ρ µL ) 2 2 (eL γ νe ) 2 k + MW (12. 2 3 g Wµ − g ′ Bµ 3 ≡ cos θW Wµ − sin θW Bµ . 2 Zµ (12.

40 GeV. 2 The value of α deviates from the known value α ≈ 1/137 because of higher order contributions. muon or tau. . Comparing this with the total width into (invisible!) channels. (12.78) From this coupling it is straightforward to calculate the partial width for Z 0 into a fermion-antifermion pair.231 2. 3 = TW . g ′ and v determine a number of experimentally measurable quantities.e.70) = 2 = 2 sin2 θ 8 MW 2 v2 8 MW 2 W In this way the parameters g.19 GeV.5 MeV (exp. −5 (12.e. Γinvisible = 480 MeV one sees that three families of (light) neutrinos are allowed. 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. Γe ≈ Γµ ≈ Γτ ≈ 83 MeV). the mass of the electron comes from the spontaneous symmetry breaking but is not predicted (it is in the coupling Ge ). giving rise to a running coupling constant after renormalization of the ﬁeld theory. as appear for instance in the decay of the neutron or the decay of the Λ (quark content uds).79) Γ(Z 0 → f f ) = 48π cos2 θW For the electron.166 4 × 10 0.80) First. (12. 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. one obtains L (h2) Ge v Ge = − √ (eL eR + eR eL ) − √ (eL eR + eR eL ) h 2 2 me = −me ee − eeh. i.71) GeV −2 = = = = .05 and CA = −1/2 we calculate Γ(e+ e− ) ≈ 78. me /v is extremely small.76) with CV CA 3 = TW − 2 sin2 θW Q. v (12. (12.72) (12. g2 MZ 2 2 (CV + CA ). 2 2 2 (12. 91.The standard model 126 the good old four-point interaction introduced by Fermi to explain the weak interactions. one has the relation g2 e2 1 GF √ = . With the choosen vacuum expectation value for the Higgs ﬁeld.77) (12.g. i. from the MW mass is about 250 GeV. (12. 80. The masses of the fermions and the coupling to the Higgs particle are contained in L (h2) .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 . such as2 α = e2 /4π GF sin2 θW MW MZ ≈ 1/127. leptons with CV = −1/2 + 2 sin2 θW ≈ −0. Finally we want to say something about the weak properties of the quarks.74) (12.73) (12. 1.

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

q) = −QL φΛd DR − DR Λ† φ† QL − QL φc Λu UR − UR Λ† φc† QL u d (12. e.q) ′ weak † mass mass DL = DL = Vu Vd DL = VCKM DL . . The Λu and Λd are complex matrices in the 3×3 family and Λd Λ† = Vd G2 Vd† .84) with Gu and Gd being real and positive and Wu and Wd being diﬀerent unitary matrices. for which we need the appropriate weak isospin doublet 0∗ 1 φ c = √ v +h . i.85) √ √ with Mu = Gu v/ 2 (diagonal matrix containing mu . the three righthanded quarks with charge +2/3 in UR and the three righthanded quarks with charges −1/3 in DR . Note that we need the conjugate Higgs ﬁeld to get a U (1)Y singlet in the case of the charge +2/3 quarks. Thus one has † † † =⇒ −DL Vd Md Wd DR − DR Wd Md Vd† DL − UL Vu Mu Wu UR − UR Wu Mu Vu UL (12.86) the unitary CKM-matrix introduced above in an ad hoc way. ms and mb ). u u where we include now the three lefthanded quark doublets in QL . φ = −φ− 0 2 For the (squared) complex matrices we can ﬁnd positive eigenvalues. UR = uR cR tR .The standard model 128 using the socalled Wolfenstein + O(λ4 ) with λ ≈ 0. (12. The imaginary part i η gives rise to CP violation in decays of K and B-mesons (containing s and b quarks.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. 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 12.82 and ρ ≈ 0.34. d d (12.82) space. respectively).83) where Vu and Vd are unitary matrices.22 and η ≈ 0. One then reads oﬀ that starting with the family basis as deﬁned via the left doublets that the mass eigenstates (and states coupling to the Higgs ﬁeld) involve the righthanded † mass † mass mass † mass states UR = Wu UR and DR = Wd DR and the lefthanded states UL = Vu UL and DL = † Vd DL .227.e. allowing us to write † Λu = Vu Gu Wu † and Λd = Vd Gd Wd . † Λu Λ† = Vu G2 Vu . the weak mass eigenstates are L (h2. (12. mc and mt ) and Md = Gd v/ 2 (diagonal matrix containing md . 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 .g. A ≈ 0. The Higgs ﬁeld is still limited to one complex doublet. each of these containing the three families.

and obtain L (h2. we ﬁnd massless neutrinos.91) † with UPMNS = Vn Ve known as the Pontecorvo-Maki-Nakagawa-Sakata mixing matrix.ℓ) = −LφΛe ER − ER Λ† φ† L − Lφc Λn NR − NR Λ† φc† L e n (12. since the weak current is the only place where they show up.5×10−3 eV2 . there is no family mixing for massless neutrinos.ℓ) = −LφΛe ER − ER Λ† φ† L. EL = Ve† EL . (12.ℓ) † † † =⇒ −EL Ve Me We ER − ER We Me Ve† EL − NL Vn Mn Wn NR − NR Wn Mn Vn NL . 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. UPMNS a parametrization that in principle also could have been used for quarks. (12. it is particularly useful because θ12 is essentially determined by solar neutrino oscillations requiring ∆m2 ≈ 8 × 10−5 12 eV2 (convention m2 > m1 ). 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 . In this case.92) . For the (massless) neutrino ﬁelds we just can redeﬁne ﬁelds into NL = Ve† NL .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 . (12. The W -current then becomes mass mass E L γ µ NL = E L γ µ NL .ℓ) =⇒ −Me EL Ve We ER − ER We Ve† EL . † † Again we continue as before now with matrices Λe = Ve Ge We and Λn = Vn Gn Wn . 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 . (12. e N L= L EL (12. = − 1/6 2/3 0 1/2 1/2 0 0 1 0 1/3 1/2 − 1/6 (12.89) with three righthanded neutrinos added to the previous case. The lepton sector (massive Dirac neutrinos) In principle a massive Dirac neutrino could be accounted for by a lagrangian of the type L (h2.87) is a weak doublet containing the three families of neutrinos (NL ) and charged leptons (EL ) and ER is † a three-family weak singlet.e.88) √ mass with Me = Ge v/ 2 the diagonal mass matrix with masses me . The mass ﬁelds ER † mass † mass = We ER .The standard model 129 The lepton sector (massless neutrinos) For a lepton sector with a lagrangian density of the form L in which (h2. one can write Λe = Ve Ge We and we ﬁnd † L (h2. decoupling from all known interactions. mµ and mτ . NL = Vn NL and NR = Wn NR µ mass γ µ V † V N mass . while θ23 then is determined by atmospheric neutrino oscillations requiring |∆m2 | ≈ 2. EL = Ve EL .90) mass † mass mass † mass † We note that there are mass ﬁelds ER = We ER . As before. i.

α2 and δ).88 a Majorana mass term for the (lefthanded) neutrino mass eigenstates. one being massive. For the leptons.ν = − MR NR NR + MR NR NR . is to add in Eq. Actually. 1 c ∗ c L mass. Υ′ and Υ′ (satisfying Υc = Υ and as discussed in chapter 6. 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. however. the massless and massive Majorana neutrinos. For the neutrino sector. The mass matrix can be written as ML |MD | eiφ M = (12. 1 L Υ ′ = nR + nc . are equivalent to two decoupled Majorana neutrinos (see below). Absorption of phases in the states is not possible for Majorana neutrinos.98) (12. a unitary matrix relating them to the weak eigenstates.The standard model 130 The lepton sector (massive Majorana ﬁelds) An even simpler option than sterile righthanded Dirac neutrinos.ν =− containing three (CP-violating) phases (α1 .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. the left. 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. − MD MR nR 2 (12. For the charged leptons there cannot exist a Majorana mass term as this would break the U(1) electromagnetic symmetry. coupled by a Dirac mass term.96) VPMNS = UPMNS K with K = 0 eiα2 /2 0 . hence the mixing matrix becomes iα /2 0 0 e 1 (12. one must for the neutrinos as well as charged leptons. Υc ≡ (ΥL )c = ΥR 1 2 L and Υc = ΥL ).and righthanded species then just form a Dirac fermion. couple the right. For these light Majorana neutrinos one has. If the Majorana mass MR ≫ MD one actually obtains in a natural way one Majorana neutrino with a very small mass. as above.c.95) 2 In order to have more than a completely decoupled sector. with for the righthanded sector a mass term MR . 1 ∗ c c (12. Thus 2 Υ ′ = nc + nL . 2 R ψ = nR + nL . 12. The way to circumvent this is to introduce as in the previous section righthanded neutrinos.97) As the most general mass term in the lagrangian density we have L mass = − = which for MD = 0 is a pure Majorana lagrangian and for ML = MR = 0 and real MD represents the Dirac case. The see-saw mechanism Consider (for one family N = n) the most general Lorentz invariant mass term for two independent Majorana spinors. 2 although this option is not attractive as it violates the electroweak symmetry. This is called the see-saw mechanism (outlined below). Thus (disregarding family structure) one has two Majorana neutrinos.94) M L NL NL + M L NL NL .99) . (12. (12. We use here the primes starting with the weak eigenstates.

For the situation ML = 0 and MR ≫ MD (taking MD real) one 2 ﬁnds M1 ≈ MD /MR and M2 ≈ MR . Exercises Exercise 12. Exercise 12.2 Derive for the vector and axial vector currents. Υ2R nR (12. T ] ψ. V µ = ψγ µ T ψ and Aµ = ψγ µ γ5 T ψ ∂µ V µ ∂µ A µ = = i ψ [M. and |a and |a′ belong to the same multiplet (there is a Qa such that |a′ = Qa |a ) then H|a = Ea |a implies that H|a′ = Ea |a′ . This choice is possible without loss of generality because the phases can be absorbed into Υ′ and Υ′ (real must be replaced by hermitean if one includes 1 2 families). T } γ5 ψ. i ψ {M. This is a mixing problem with a symmetric (complex) mass matrix leading to two (real) mass eigenstates. i.103) for each of which one ﬁnds the lagrangians L = and we are left with two decoupled Majorana ﬁelds Υ1L = U ∗ nc .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)) . 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 † . a and a′ are degenerate states. . i. (12. 2 U (M M † ) U † = M0 .1 Consider the case of the Weyl mode for symmetries. 2 with real masses Mi . Exercise 12.e.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 . H] = 0. (12. related via c Υ1R = U nL . 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 . [Qa .104) ↔ 1 1 Υi i / Υi − Mi Υi Υi ∂ 4 2 (12.105) for i = 1.The standard model 131 taking ML and MR real and non-negative.e.101) Thus one obtains from 2 ML + |MD |2 † MM = |MD | ML e+iφ + MR e−iφ the eigenvalues 2 M1/2 |MD | ML e−iφ + MR e+iφ 2 MR + |MD |2 . (12.

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

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

relativistic quantum mechanics

relativistic quantum mechanics

Are you sure?

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

We've moved you to where you read on your other device.

Get the full title to continue

Get the full title to continue listening from where you left off, or restart the preview.

scribd