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

These notes Section 1.3 2. M. M.4 2.E.2 4.3 3.1.2 2. 5. 1996. 1985. Cambridge University Press. Weinberg. The quantum theory of ﬁelds.3.7 2. 1980.3 3. 4. C. 3. Peskin & Schroeder and Srednicki. Diagrammatica.3 6. S.2 2.2 3.4 6. Cambridge University Press.1 3.1 4.3 2.4 3. Peskin and D. but are rather compact.8 3. Quantum Field Theory.5 4. L. Veltman.3 4.1 2.3 References As most directly related books to these notes. 1.H. Ryder. Addison-Wesly. Quantum Field Theory.2 6. Cambridge University Press.4 3.4 3. 6.2.1 2.1 3.5 2.2 3. I: Foundations.3 2. Quantum Field Theory. The notes contain all essential information. 2.3 3.2 2. 3.V. Itzykson and J.3 3.4 2. Zuber.-B.2 3. 2007.3 3. 3. An introduction to Quantum Field Theory. I refer to the book of Srednicki [1] and Ryder [2]. Vol. 1994. Cambridge University Press. M.1 1. II: Modern Applications.5 2.1 6.2 36 22 . Vol. Corresponding chapters in books of Ryder.2 6. Schroeder.2 3.5 5.3 2.2 3. Cambridge University Press.1 5. 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). Other text books of Quantum Field Theory that are useful are given in refs [3-6].1 2 2 3. McGraw-Hill. Srednicki.3 2. 1995.5 Ryder Peskin & Schroeder Srednicki 1 1 1 1 3 2. 2.2 1. 1995.4 4.

3 7.4 12.6 4.5 4.4 These notes Section 7.2 11.4 3.5 9.5 4.7 4.8 6. 6.2 8.5 .3 2.3 6.1 10.3 20.3 8.8 5 4.4 12.1 7. 4. 4. 6.3 12.6 3.4 2.1 12.3 8.7 5.5 7.1. 2. 8.2. 6.7.6 3. 6.4 4.3 10. 2.3 3.1 9.6 8.2 10.6 10.5 Ryder 4 4 4 4 4 4 Peskin & Schroeder 2.4 10.1 8.1 5.1.2.1 20.5 6 7 6 6.3.5 9.2 8.2 4.4.1.5 7. 20. 4.2.4.3 6 8 9 Srednicki 3 3 3 37 5.5 6.4.3.1 11.6 6.4 9.10 8.5 2.4 7.6 3.2 5.2 9.1 9.4 8.1 20.2.5 11.5.3 11.1 20.3 7. 9.2 12. 6. 9. 4.3.6 9.3 9.2 7.

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

582 119 15 (56) × 10−22 MeV s.6) . (1. In the framework of classical mechanics as well as quantum mechanics the position of a particle is a well-deﬁned concept and the position coordinates can be used as dynamical variables in the description of the particles and their interactions. (1. (1. It is given by pop ψ(r) = −i ∇ψ(r). ≡ h/2π = 1. quantum eﬀects do not play a role anymore and one is in the classical domain. They satisfy the well-known (canonical) operator commutation relations [ri . Corresponding position and momentum operators do not commute. (1. pj ] = i δij .1 Quantum ﬁeld theory In quantum ﬁeld theory the theories of quantum mechanics and special relativity are united.3) The uncertainty principle tells us that in this representation the momenta cannot be fully determined. In quantum mechanics. ˜ ψ(p) = i d3 r exp − p · r ψ(r). usually given divided by 2π. In this coordinate representation the position operators r op simply acts as rop ψ(r) = r ψ(r). being eigenvalues of the position operators.4) where δij is the Kronecker δ function.1) In the limit that the action S is much larger than .5) One can also choose a representation in which the momenta of the particles are the dynamical variables.2) In the limit that v ≪ c one reaches the non-relativistic domain.Chapter 1 Introduction 1. 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 . In quantum mechanics a special role is played by Planck’s constant h. S ≫ . ˜ The corresponding states are |p and the wave functions ψ(p) = p||ψ are the Fourier transforms of the coordinate space wave functions. In special relativity a special role is played by the velocity of light c. 1 (1. Indeed. the position can in principle be determined at any time with any accuracy.054 571 68 (18) × 10−34 J s = 6. (1. One can talk about states |r and the wave function ψ(r) = r||ψ .

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.

The real. x1 .νn can be conν structed. Many other four vectors and tensors transforming like T ′ µ1 .21) (1. x3 ). Relations relating tensor expressions.. The invariant distance between two points x and y in Minkowski space is determined from the length dsµ = (x − y)µ . one must distinguish tensors with upper or lower indices and one can have mixed tensors. x) = (x0 .µn = Λµ1ν1 . When 3-dimensional space is considered as part of Minkowski space. The scalar product of two diﬀerent vectors x and y is denoted x · y = xµ y ν gµν = xµ yµ = xµ y µ = x0 y 0 − x · y. x2 . x2 = x · x = xµ xν nµ · nν = xµ xν gµν . No diﬀerence is made between upper or lower.. are . The (invariant) lengths often have special names.. One can distinguish: • ds2 > 0 (timelike intervals). Unlike in Euclidean space.3) with basis nµ (µ = ˆ 0.1. independent of a coordinate system. we will use upper indices for the three-vectors. denoted as V ′µ = Λµν V ν .2. • ds2 < 0 (spacelike intervals). −x). Useful relations are ǫijk ǫℓmn ǫijk ǫimn ǫijk ǫijl = = = δiℓ δjℓ δkℓ 2 δkl . • ds2 = 0 (lightlike or null intervals). We note that for Euclidean spaces (with a positive deﬁnite metric) vectors and tensors there is only one type of indices. Because of the diﬀerent signs occurring in gµν . such as eigentime τ for the position vector τ 2 ≡ x2 = t2 − x2 . xµ = gµν xν ..23) δjm δkn − δjn δkm . (1. Vectors are denoted x = xµ nµ . in which case zi = ǫijk xj yk . Λµn n T ν1 . it is convenient to distinguish lower indices from upper indices.25) Within Minkowski space the real. . In special relativity we start with a four-dimensional real vector space E(1. (1. and are given by (x0 . The length (squared) of a vector is obtained from the scalar ˆ product. linear transformations that leave the length of a vector invariant are called (homogeneous) Lorentz transformations. So we could have used all upper indices in the above equations. xi are the three space components.3). . 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 . where the coordinate t = x0 is referred to as the time component. 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. x2 . One has x2 = xµ xµ = t2 − x2 . ˆ ˆ (1.24) The quantity gµν ≡ nµ · nν is the metric tensor. x1 . The lower indices are constructed in the following way. the points lie on the lightcone and they can be connected by a light signal. The Poincar´ transformations include e e Lorentz transformations and translations. δim δjm δkm δin δjn δkn . For four-vectors in Minkowski space we will use the notation with upper indices and write x = (t.22) (1. x3 ) = (t. 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 .Introduction 5 that can be used in the cross product of two vectors z = x × y. given by g00 = −g11 = −g22 = −g33 = 1 (the ˆ ˆ other components are zero). however. These transformations do change the components of a vector. the invariant length or distance (squared) is not positive deﬁnite.26) (1. linear transformations that do not change the length of a four-vector are called the Lorentz transformations. In Minkowski space.

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 .35) ρ ǫµνρσ ǫµν σ σ ǫµνρσ ǫµνρ ′ ′ = −2 g ρρ g σσ − g ρσ g σρ = −6g σσ . = ∂ . (1.∇ . deﬁned by ∂2 − ∇2 . . (1. not aﬀected by Lorentz transformations. but more important has the same eﬀect on lefthandside and righthandside.29) 2 = ∂ µ ∂µ = ∂t2 The value of the antisymmetric tensor ǫµνρσ is determined in the same way as for ǫijk .36) Exercises Exercise 1. Examples are the scalar products above but also relations like pµ = m dxµ /dτ for the momentum four vector. ∂1 . Each of these permutations leads to a minus sign. Note that in this equation one has on left. a∞ = 4πǫ0 2 /me e2 . g0 = g1 = g2 = g3 = 1.Introduction 6 called covariant. where m is called the mass of the system. ∂2 . ∂t (1.1 (a) In the the Hydrogen atom (quantum system) the scale is set by the Bohr radius. starting from ǫ0123 = 1. explicitly written as (p0 . (1.and righthandside a four vector because τ is a scalar quantity! The equation with t = x0 instead of τ simply would not make sense! The momentum four vector. for instance. It is an invariant tensor. 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. gµ = gµρ g ρν and 0 1 2 3 is in essence a ’Kronecker delta’.31) ǫµνρσ ǫµν ′ ′ ρσ ′ = − (1.34) (1.27) It is easy to convince oneself of the nature of the indices in the above equation. (1. Relate this quantity to the electron Compton wavelength −e via the dimensionless λ ﬁne structure constant α. (1. = −24. because one has (1. ∂t ∂x ∂y ∂z ν ∂µ xν = gµ .28) ν ν Note that gµ with one upper and lower index. The derivative ∂µ is deﬁned ∂µ = ∂/∂xµ and we have a four vector ∂ with components (∂0 . ′ ′ . . . constructed via the metric tensor itself. p). For the contraction of a vector with the antisymmetric tensor one often uses the shorthand notation ǫABCD = ǫµνρσ Aµ Bν Cρ Dσ . The length squared of ∂ is the d’Alembertian operator. is timelike with invariant length (squared) p2 = p · p = pµ pµ = E 2 − p2 = m2 .33) (1. p) = (E.30) (Note that there are diﬀerent conventions around and sometimes the opposite sign is used).32) . ∂3 ) = ∂ ∂ ∂ ∂ . 0123 0123 ′ ′ ǫ µνρσ ǫµνρσ The ﬁrst identity.

a3 ) are the expansion coeﬃcients using the basis vectors n0 .Introduction 7 (b) Relate the classical radius of the electron (a relativistic concept). e. a2 . This demonstrates how a careful use of units can save a lot of work. Compare it with the proton mass and use Eq. ˆ ˆ ˆ ˆ These are n1 = (0. α and me c2 = 0. a1 . ν. 0) ˆ n2 = (0. Exercise 1. 1. (d) Use the value of the gravitational constant GN / c5 = 6. ǫ0 . 0. Also for the coordinates [a− .4 Lightcone coordinates for a four vector a (which we will denote with square brackets as [a− . a2 ] we can ﬁnd basis vectors n− . Use that to give the ˆ ˆµ ˆ components of the basis vectors n+ and n− . n2 . ρ. 0. c.3 Prove the following relation ǫµνρσ g αβ = ǫανρσ g µβ + ǫµαρσ g νβ + ǫµνασ g ρβ + ǫµνρα g σβ . 1) ˆ .2 Prove the identity A × (B × C) = (A · C) B . (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). 2. aT ]) are deﬁned through √ a± ≡ (a0 ± a3 )/ 2.511 MeV. Note that ˆ ˆ ˆ ˆ a · nα = aα and derive from that the components nα of the four vector nα . 0.71 × 10−39 GeV−2 to construct a mass Mpl (Planck mass). One does not need to know .(A · B) C using the properties of the tensor ǫijk given in section 1. (a) Express the scalar product a · b in lightcone coordinates and deduce from this the values of g++ . n3 . me . g−− . (b) The coordinates (a0 . a2 ] or [a− . a+ . n1 .13 to give its actual value in kg. re = e2 /4πǫ0 me c2 to the Compton wavelength. Also construct and calculate the Planck length Lpl . but only appropriate combinations. a1 . 0. How are a+ and a− related to a+ and a− .3. n2 . . . [Note: what is the MKS unit for V/T?] Exercise 1. n1 . a1 . 1. a+ . 0) ˆ n0 = (1. 1. g+− and g−+ . ˆ ˆ . σ} is a permutation of {0. Exercise 1. 0. 3} the index α can only be equal to one of the indices in ǫµνρσ . which is the Compton wavelength for the Planck mass. a+ . (c) Calculate the Compton wavelength of the electron and the quantities under (a) and (b) using the value of c. by a simple few-line reasoning [For instance: If {µ. 0. n+ . 1. 0) ˆ n3 = (0. ]. .

E = p2 .3 are not covariant.7) k 0 = ± k2 + M 2 = ±Ek .2) Acting on the wave function one ﬁnds for a free particle.Chapter 2 Relativistic wave equations 2. (2. Although it is straightforward to ﬁnd the solutions of this equation. (2. p). 8 .1 The Klein-Gordon equation In this chapter. ∂t2 (2. (2. 2 + M 2 φ(r. The Schr¨dinger equation in quantum mechanics is the operator equation o corresponding to the non-relativistic expression for the energy.5) with (k 0 )2 = k2 + M 2 . t) = 0. t) = ∂2 − ∇2 + M 2 φ(r. ∂t 2M (2. namely plane waves characterized by a wave number k. ∂t p −→ pop = −i∇. t) = exp(−i k 0 t + i k · r). t) = − ψ(r. the interpretation of this equation as a single-particle equation in which φ is a complex wave function poses problems because the energy spectrum is not bounded from below and the probability is not positive deﬁnite. one obtains (2. Thus a covariant equation can be obtained by starting with the (covariant) equation for the invariant length of the four vector (E.1 and 2. written as pµ −→ i∂µ is covariant (the same in every frame of reference). we just want to play a bit with covariant equations and study their behavior under Lorentz transformations. Substitution of operators gives the Klein-Gordon (KG) equation for a real or complex function φ.6) The energy spectrum is not bounded from below: considering the above stationary plane wave solutions. 2M (2. t).4) where M is the particle mass. i ∂ ∇2 ψ(r.1) under the substitution (in coordinate representation) E −→ Eop = i ∂ . But the replacement 2.3) Equations 2. φk (r. p2 = pµ pµ = E 2 − p2 = M 2 .2.

15) .5). At that point the dependence on position r and time t is just a dependence on numbers/parameters on which the operator depends. k) + ei k·x φ(−Ek . (2. we want the most general solution. (2π)3 2Ek (2. there are solutions with negative energy. relativistically the density is not a scalar quantity. This leads to the existence of negative densities. This continuity equation can be written down o covariantly using the components (ρ. −k) ≡ b∗ (k) one has φ(x) = d3 k e−i k·x a(k) + ei k·x b∗ (k) = φ+ (x) + φ− (x). The appropriate current corresponding to the KG equation (see Excercise 2. For this we note that an arbitrary solution for the ﬁeld φ can always be written as a superposition of plane waves.10) ∂t which follows directly from the Schr¨dinger equation. (2.3). And. The integration over k-modes clearly is covariant and restricted to the ‘mass’-shell (as required by Eq.14) ˜ ˜ Introducing φ(Ek .11) Therefore.2 Mode expansion of solutions of the KG equation Before quantizing ﬁelds. ∂µ j µ = 0. but rather the zero component of a four vector.2) is j µ = i φ∗ ∂ µ φ or (ρ. having the KG equation as a space-time symmetric (classical) equation. −k) . φ(x) = d4 k ˜ 2π δ(k 2 − M 2 ) e−i k·x φ(k) (2π)4 (2. however. j) = ↔ i φ∗ ∂0 φ. −i φ∗ ∇ φ . Then.8) (2. The KG equation. j) of the four-current j. 2M 2M (2. there are no longer fundamental objections to mix up space and time. ∂ρ = −∇ · j. 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.9) They satisfy the continuity equation.13) ˜ with (in principle complex) coeﬃcients φ(k). φ(x) = d3 k (2π)3 2Ek ˜ ˜ e−i k·x φ(Ek . rather than as a wave function. As we will see later both problems are related and have to do with the existence of particles and antiparticles.e.12) It is easy to see that this current is conserved if φ (and φ∗ ) satisfy the KG equation. for which we need the interpretation of φ itself as an operator. is a second order equation and φ and ∂φ/∂t can be ﬁxed arbitrarily at a given time. 2. (2. which is what Lorentz transformations do.Relativistic wave equations 9 i. k) ≡ a(k) and φ(−Ek . just as the dependence on time was in ordinary quantum mechanics. 2. It is possible to rewrite it as an integration over positive energies only but this gives two terms (use the result of exercise 2. ↔ ↔ (2. This operator has all possible solutions in it multiplied with creation (and annihilation) operators. it is simply a matter of being careful to ﬁnd a consistent (covariant) theory.

φP (x) ≡ φ(˜) (P for parity). They are referred to as mode and anti-mode amplitudes (or because of their origin positive and negative energy modes). (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 (2.21) 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) . x ∂µ ∂ µ + M 2 φ(˜) = 0.15 one has elimated k 0 and in both equations k · x = Ek t − k · x. after restricting the energies to be positive).23) . −x) ≡ (˜µ ).6 We will explicitly discuss the example of a discrete symmetry.e.20) one has for the mode amplitudes aP (k) = a(−k) and bP (k) = b(−k). changing the sign of the spatial coordinates. x) → (t. Performing some Lorentz transformation x → x′ = Λx. for which we consider space inversion. In terms of modes one has φC (x) = φ∗ (x) = ≡ d3 k e−i k·x b(k) + ei k·x a∗ (k) (2π)3 2Ek d3 k e−i k·x aC (k) + ei k·x bC∗ (k) . 2. (2. The choice of a and b∗ allows an easier distinction between the cases that φ is real (a = b) or complex (a and b are independent amplitudes). (2π)3 2Ek (2.3 Symmetries of the Klein-Gordon equation We arrived at the Klein-Gordon equation by constructing a covariant operator (∂µ ∂ µ + M 2 ) acting on a complex function φ. one thus must have that the function φ → φ′ such that φ′ (x′ ) = φ(x) or φ′ (x) = φ(Λ−1 x).18) (2.22) showing that with each solution there is a corresponding charge conjugated solution φC (x) = φ∗ (x).16) The consequence of this is discussed in Exercise 2.14 and 2.Relativistic wave equations 10 In Eqs 2. x Transforming everywhere in the KG equation x → x one obtains ˜ ˜ ˜ x ∂µ ∂ µ + M 2 φ(˜) = 0. which implies (xµ ) = (t. It is trivial to see that ∂µ ∂ µ + M 2 φ∗ (x) = 0.19) implying that for each solution φ(x) there exists a corresponding solution with the same energy. i. (2. (2π)3 2Ek (2. Another symmetry is found by complex conjugating the KG equation.17) (2. The coeﬃcients a(k) and b∗ (k) are the amplitudes of the two independent solutions (two. This shows how parity transforms k-modes into −k modes. Since a · b = a · ˜ it is easy to see that ˜ b.

t) ≡ d3 x φ(x. ˙ and thus for any normalized solution the full ‘charge’. S V d3 x j 0 (x).1 Show that for a conserved current (∂µ j µ = 0) the charge in a ﬁnite volume. that j µ = i φ∗ ∂ µ φ is a conserved current Note: A ∂µ B ≡ A∂µ B − (∂µ A)B). d3 k F (Ek .Relativistic wave equations 11 i. k). Use these to show that ˜ ˜ a(k) = ei Ek t i ∂0 φ(k. letting V → ∞. Note that a(k) and b(k) are independent of t. Q = 0. t) exp(−i k · x) and its time ˜ derivative i∂0 φ(k. t) for the mode expansion in Eq. 1 Use ↔ ↔ the following property of delta functions δ (f (x)) = zeros xn 1 δ(x − xn ). is conserved. k) = (2π)4 k2 + M 2 .e. Exercises Exercise 2. For the real ﬁeld one has aC (k) = a(k).3 Show that1 d4 k 2π δ(k 2 − M 2 ) θ(k 0 ) F (k 0 .15. t).2 Show that if φ and φ∗ are solutions of the KG equation. QV ≡ satisﬁes ˙ QV = − ds · j. 2. (2π)3 2Ek where Ek = Exercise 2. Exercise 2. ˜ b(k) = ei Ek t i ∂0 φ∗ (−k. t). for the mode amplitudes aC (k) = b(k) and bC (k) = a(k). This shows how charge conjugation transforms ’particle’ modes into ’antiparticle’ modes and vice versa. t) = ei Ek t (i∂0 + Ek ) φ(k. |f ′ (xn )| . ↔ ↔ Exercise 2.4 ˜ Write down the 3-dimensional Fourier transform φ(k.

Exercise 2. also using the result in (a). We will see a∗ (k) → a† (k) and a(k) → aop (k) with aop and a† op op being annihilation and creation operators (as known from a harmonic oscillator) and the same for the coeﬃcients b(k). 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). From this one concludes that α and β must be matrix-valued and ψ must be a multi-component wave function. j 0 = ψ†ψ and j = ψ † αψ. Note that a{µ bν} indicates symmetrization. i (a) Show that relativistic invariance.15. and β 2 = I. i.Relativistic wave equations 12 Exercise 2. ∂ ψ(r. 1 2 (b) Express now the full charge QV (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). d3 x (∂ {0 φ)∗ (∂ i} φ). (b) Show that one has current conservation for the current 2 We denote the commutator as [A. (c) Similarly. ∂t where ψ is a wave function and the nature of α and β is left open for the moment. in terms of the a(k) and b(k). {αi . 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. express the quantities E(t) = P i (t) = d3 x (∂0 φ)∗ (∂0 φ) + ∇φ∗ · ∇φ + M 2 φ∗ φ .6 Write down the mode expansion for the Lorentz transformed scalar ﬁeld φ′ (x) and show that it implies that the Lorentz transformed modes satisfy a′ (Λk) = a(k).e. . 2. We will encounter these quantities later as energy and momentum. a{µ bν} ≡ aµ bν + aν bµ . making sure that each component of ψ satisﬁes the KleinGordon equation (in terms of diﬀerential operators ∂µ ∂ µ +m2 = 0) requires the anticommutation relation2 {αi . β} = 0. 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 . αj } = 2 δ ij I. t) = (−i α · ∇ + m β) ψ(r. Exercise 2.5 (a) As an introduction to parts (b) and (c).1) for a complex scalar ﬁeld current (exercise 2. B} = AB + BA.7 To solve the problem with positive and negative energies and get a positive deﬁnite density.2) in terms of the a(k) and b(k) using the expansion in Eq. t).

in particular each component of these spinors will satisfy this equation. 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. i.3) ϕ=0 13 . and we must study the representations of the Lorentz group. it has 3 components that we know the behavior of under rotations. R(π.e. the symmetry group describing rotations is embedded in the Lorentz group. 3.1) for a rotation around the z-axis or shorthand V ′ = R(ϕ. n) ≡ R(π. Particles with spin then will be described by certain spinors. 0 ≤ ϕ ≤ π. Locally this parameter-space is 3-dimensional and correspondingly one has three generators. 0 0 (3. z ) = 1 + i δϕ Lz with as generator ˆ 1 ∂R(ϕ.1 The rotation group and SU (2) The rotation groups SO(3) and SU (2) are examples of Lie groups. we want to investigate if there exist objects other than just a scalar (real or complex) ﬁeld φ. n) where n is a unit vector and ˆ ˆ ϕ is the rotation angle. we also want it to hold for particles with spin. e.g.e. Since the KG equation expresses just the relativistic relation between energy and momentum. In this section we will investigate the requirements imposed by Poincar´ e invariance. e. provided we identify the antipodes. two-component ﬁelds in analogy to the two-component wave functions used to include spin in a quantum mechanical description of an electron in the atom. −n). however. The parameter-space of SO(3) is a ˆ ˆ ˆ sphere with radius π. In particular.g. The KG equation will actually remain valid. z )V .Chapter 3 Groups and their representations The simple systems that we want to describe in a relativistically invariant way are free particles with spin.2) (3. that is groups characterized by a ﬁnite number of real parameters. z ) Lz = i ∂ϕ 0 = i 0 −i 0 0 0 . In a relativistic theory. electrons. i. In quantum mechanics spin is described by a vector. in which the parameter space forms locally a Euclidean space. 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. Any rotation can be uniquely written as R(ϕ. For an inﬁnitesimal rotation around the z-axis one has ˆ R(δϕ.

Ly and Lz . ˆ R(ϕ. ] that satisﬁes • ∀ x. 0 (3. . y]] = 0 (Jacobi identity). 2 (3. z]] + [y. With matrix commutation this algebra satisﬁes the requirements for a Lie algebra. π) (n.Groups and their representations 14 (n. z ) = lim N →∞ that are generated by i 0 .7) = a0 0 −1 i 0 1 0 0 1 • [x. [y. which reﬂects itself in the noncommutation of the generators.1: the parameter spaces of SO(3) (left) and SU (2) (right). thus is as the surface i=0 of a sphere in 4 Euclidean dimensions. x]). Lx = Ly = 0 0 0 i 0 −i 0 ϕ ˆ .and y-axes 0 0 0 0 0 0 0 −i . namely that there exists a bilinear product [. [x. π ) π -sphere (antipodes identified) 2 π -sphere (surface identified) Figure 3. x] = 0 (thus [x. (3. x]] + [z. Lj ] = i ǫijk Lk . These generators form a threedimensional Lie algebra SO(3). the rotations in SO(3) can be generated from inﬁnitesimal rotations that can be expressed in terms of a basis of three generators Lx .5) Rotations in general do not commute. These matrices can be deﬁned as acting on 2-component spinors (χ → Aχ) or equivalently as acting on 2 × 2 matrices (B → ABA† ). with real a’s and 3 (ai )2 = 1.z N N or (Lk )ij = −i ǫijk . for rotations around z for instance. (3. It is straightforward to check that any (ﬁnite) rotation can be obtained from a combination of inﬁnitesimal rotations. In the same way we can consider rotations around the x. One way of viewing the parameter space. [z. ˆ therefore. 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 (3.2 π) (-n. • [x. They satisfy the commutation relations [Li . Writing a0 = cos(ϕ/2) and a = n sin(ϕ/2) we have1 A = A(ϕ. y ∈ A ⇒ [x. y] = −[y. Locally this is a 3-dimensional Euclidean space and SU (2). we turn to the group SU (2) of special (det A = 1) unitary (A† = A−1 ) 2 × 2 matrices. n) = 1 cos ˆ 1 Note ϕ ϕ ˆ + i (σ · n) sin 2 2 = exp i ϕ ˆ σ·n . and thus (σ · n)2 = 1. y] ∈ A. is a 3-dimensional Lie-group.6) Summarizing.4) R . Next.8) ˆ that (σ · a)(σ · b) = a · b + i σ · (a × b).

The next issue is to ﬁnd the appropriate symmetry operators in the Hilbert space. i. n) π ≤ ϕ ≤ 2π. For instance in a function space translations. it is suﬃcient that the generators commute with H. (3. 2 2 2 (3. ˆ ˆ A(ϕ. The existence of symmetries means that there are operators g belonging to a symmetry group G that commute with the Hamiltonian.e. 2! . Suppose we have a system described by a Hamiltonian H. In the full parameter space. with ˆ J ·n ≡ ˆ 1 ∂A(ϕ. H] = 0 for g ∈ G.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. the SU (2) → SO(3) mapping is a 2 : 1 mapping where both A = ±1 are mapped into R = I. 3.12) One. for ﬁxed n. 2 (3. thus. sometimes in more than one way (but this will be discussed later). in particular that AB → RAB = RA RB ).2 Representations of symmetry groups The presence of symmetries simpliﬁes the description of a physical system and is at the heart of physics. since any ﬁnite transformation can be constructed from the inﬁnitesimal ones. . φ′ (r) = φ(r + a) = φ(r) + a · ∇ φ + 1 (a · ∇)2 φ + . n! 2! n=0 ∞ (3. n) ≈ 1 + i ϕ J · n.10) σ ˆ · n.Groups and their representations 15 where we have used the (for operators new!) deﬁnition eA ≡ An 1 = 1 + A + A2 + .9) The parameter-space of SU (2). n) ˆ −→ R(2π − ϕ. . n) −→ R(ϕ. n) i ∂ϕ = ϕ=0 (3. There exists a corresponding mapping of the groups given by µ: SU (2) −→ SO(3) ˆ ˆ A(ϕ. . They satisfy σi σj σk = i ǫijk .e. (3. σy /2 and σz /2 form the basis of the Lie-algebra SU (2). one has a Lie algebra isomorphism that is linear and preserves the bilinear product. . [g. immediately sees that the Lie algebras are identical.14) For a Lie group.e. also can be considered as a ﬁlled 3-sphere. i. i.15) [g.11) Thus σx /2. however. thus. SU (2) ≃ SO(3). . H] = 0 for g ∈ G. . . (3. the above mapping corresponds to the trivial mapping of the Lie algebras. The inﬁnitesimal generators of SU (2) are ˆ obtained by considering inﬁnitesimal transformations. now with radius 2π and with all points at the surface identiﬁed (see ﬁgure). Near the identity.

m = j(j + 1) − m(m ± 1)|j. Casimir operators commute with all the generators and the eigenvalue of J 2 can be used to label the representation (j). but it is at the heart of being able to construct a suitable Hilbert space. it is suﬃcient to consider the representations Φ of the Lie-algebra G. ϕ + α) = U (α. pop and ℓop work in Hilbert space and their commutators indeed follow the required commutation relations as symmetry operators. ˆ U (α. The latter is suﬃcient to get the commutation relations of the (quantum) rotation operators. . Note that the same requirements would apply to classical mechanics. These are mappings from G into the same ﬁnite dimensional vector space (its dimension is the dimension of the representation). The other generators Jx and Jy (or J± ≡ Jx ± iJy ) then transform between the states in V (j) . Jz |j.Groups and their representations 16 are generated by the derivative acting on the functions. Within the (2j + 1)-dimensional representation space V (j) one can label the eigenstates |j. m . p) (we will come back to this also in Chapter 7). just starting oﬀ with the (basic) canonical commutation relations [ri . more precise φ(r + a) = U (a)φ(r) where U (a) = exp +ia pop . m ± 1 with 2j + 1 being integer and m = j. Similarly one has the time translation operator φ(t + τ ) = U (τ )φ(t). J = 0 0 + 0 0 . z )φ(r. . indeed. The above set of operators H. which preserve the Lie-algebra structure. J+ = 0 1 . for j = 1/2: 0 1/2 Jz = 0 −1/2 0 0 −1 . i. . . j − 1. In particular the quantum operators ℓop obey in quantum mechanics commutation relations [ℓi . These are mappings of G into a ﬁnite dimensional vector space. while J± |j. does in the nontrivial behavior of Poisson brackets of quantities A(x. m . . U (τ ) = exp (−iτ H) . 3. the commutation relations. z ) = exp (+iα ℓz ) . ℓj ] = i ǫijk ℓk that are identical to those in Eq. j − 1. The ﬁnite dimensional vector space just represents new degrees of freedom. p) and B(x. . preserving the group structure. J+ = 0 J− = 0 . the Poisson bracket operation satisﬁes all properties of a Lie algebra. J− = 2 √ 0 0 0 0 2 0 .18) (3.6. m = j(j + 1)|j. one looks among the generators for a maximal commuting set of operators.17) (3. 2 .16) with ℓz = −i∂/∂ϕ = −i(x ∂/∂y − y ∂/∂x) or ℓop = r op × pop . 0 for j = 1: 1 0 Jz = 0 0 0 0 √ 0 0 2 √ 0 √ 0 0 . ˆ (3. ϕ). with pop = −i∇. Indeed.e.g. θ. −j. for the function space. In many cases one can simply construct a suitable Hilbert space or part of it by considering the representations Φ of a group G. pj ] = i δij . Explicit representations using the basis states |j. Somehow the nontrivial structure of rotations should show up and it. −j one has for j = 0: Jz = 0 . e. m = j. . θ. m with eigenvalues of Jz . m with m-values running from the heighest to the lowest. φ(r. J− = 0 1 0 0 0 . m = m|j. From the algebra one derives J 2 |j. In order to ﬁnd local representations Φ of a Lie-group G. for rotations the operator Jz and the (quadratic Casimir) operator J 2 . with H = i ∂/∂t and the rotation operator. 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). . To get explicit representations. It may seem trivial.

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

it is easy to derive some properties of the matrices Λ: (i) det(Λ) = ±1. which as a matrix equation with (Λ)µν = Λµν and (G)µν = gµν gives (ΛT )ρµ (G)µν (Λ)νσ = (G)ρσ . det Λ = −1 is called improper).22) (Note that xT is a row vector). consisting of the Lorentz group and translations. (ii) |Λ00 | ≥ 1. In this section we consider the Poincar´ group.24) L↑ + L↓ + L↑ − L↓ − proper orthochrone proper non-orthochrone improper orthochrone improper non-orthochrone . Denoted in terms of column vectors and 4-dimensional matrices one writes for the poincar´ transe formations x′ = Λ x + a. From this property. We considered the generators and looked for representations in ﬁnite dimensional spaces.23) The proper tensor structure of the matrix element (Λ)µν is a tensor with one upper and one lower index. G = 10 (3.25) (3.Groups and their representations 18 3.27) (3. 0 0 g00 g01 g02 g03 1 0 g g11 g12 g13 0 −1 0 0 = . (3. proof: det(ΛT GΛ) = det(G) → (det Λ)2 = 1. 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. proof: (ΛT GΛ)00 = (G)00 = 1 → Λµ0 gµν Λν0 = 1 → (Λ00 )2 − i (Λi 0 )2 = 1. e To derive some of the properties of the Lorentz group. it is convenient to use a vector notation for the points in Minkowski space. 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.g.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. Invariance of the length of a vector requires for the Lorentz transformations x′2 = gµν x′µ x′ν = gµν Λµρ xρ Λνσ xσ = x2 = gρσ xρ xσ or Λµρ gµν Λνσ = gρσ . e. explicitly x′µ = (Λ)µν xν + aµ or x′µ = Λµν xν + aµ .26) (3. (det Λ = +1 is called proper.3 The Lorentz group In the previous section spin has been introduced as a representation of the rotation group SU (2) without worrying much about the rest of the symmetries of the world. Writing x as a column vector and the metric tensor in matrix form.

From the commutation relations one sees that the boosts (pure Lorentz transformations) do not form a group. we have found the generators of rotations.30) 0 + 0 +1 0 0 0 0 +1 0 0 −1 0 0 +1 0 0 ∈ L↓ (time inversion) It = (3. [K i . z ) ≈ I + i ϕ J 3 . Rotations around the z-axis are given by ΛR (ϕ. 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. z ) ≈ ˆ ˆ I − i η K 3 . 3 i 2 i=1 (Λ0 ) 3 19 (iii) Boosts along the z-direction are given by ΛB (η. Thus 0′ 0 V cosh η 0 0 sinh η V 0 0 0 i 1′ 1 V V 0 0 0 0 0 1 0 0 2′ = 2 −→ K3 = V V 0 0 0 0 .32) 0 − 0 −1 0 0 0 0 −1 . J 2 . inﬁnitesimally given by ΛB (η. 0 0 +1 0 0 +1 0 0 ∈ L↑ (identity) I = (3. The ﬁrst two sets of commutation relations exhibit the rotational behavior of J and K as vectors in E(3) under rotations. z ) = exp(i ϕ J 3 ). J = (J 1 . and those of the boosts. J j ] = i ǫijk J k . K j ] = −i ǫijk J k . Thus ˆ 0′ 0 0 0 V 0 V 1 0 0 0 0 1′ 1 V 0 cos ϕ sin ϕ 0 V 0 0 −i 0 2′ = 2 −→ J3 = V 0 − sin ϕ cos ϕ 0 V 0 i 0 0 . (3. the diﬀerence of ﬁrst performing a boost in the y-direction and thereafter in the x-direction and the boosts in reversed order) contains a rotation (in the example around the z-axis). βγ = sinh η.28) 3′ 3 V 0 0 0 1 V 0 0 0 0 The parameter η runs from −∞ < η < ∞. K j ] = i ǫijk K k .29) 0 0 1 0 3′ 3 V sinh η 0 0 cosh η V i 0 0 0 Note that Lorentz transformations generated from the identity must belong to L↑ . Returning to the global group. In L↑ . since I ∈ L↑ and + + det Λ and Λ0 change continuously along a path from the identity.g. The commutator of two boosts in diﬀerent directions (e. K = (K 1 . [J i .Groups and their representations = i=1 (Λ0 )2 . it is easy to ﬁnd the following typical examples from each of the four classes. one distinguishes rotations 0 + and boosts.31) 0 − 0 +1 0 0 0 0 +1 0 0 +1 0 0 −1 0 0 ∈ L↑ (space inversion) Is = (3. inﬁnitesimally given by ˆ ΛR (ϕ. since the generators K do not form a closed algebra. J 3 ). Using these explicit transformations. K 3 ). This is the origin of the Thomas precession. K 2 . which satisfy the commutation relations (check!) [J i . i proof: use ΛT GΛ = G and ΛGΛT = G. (3. z ) = exp(−i ηK 3 ).

we have for the (inﬁnitesimal) rotations (around z-axis) ΛR = I + i ϕ3 J 3 = I − i ω12 M 12 . Multiplying the proper orthochrone transformations with one of them gives all Lorentz transformations. In order to ﬁnd the commutation relations including the translation generators.37) The algebra of the generators of the Lorentz transformations can be obtained by an explicit calculation2 .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. Also this group can be divided into four parts. i i0 i i0 1 ijk M jk 2ǫ (3. Thus the two vector operators J and K form under Lorentz transformations an antisymmetric tensor M µν . There are six generators. explicitly reading x′µ = = (Λ)µν xν + aµ = ((I)µν + (ω)µν ) xν + ǫµ Λµν xν + aµ = (g µν + ω µν ) xν + ǫµ . One can easily convince oneself that i (ω)µν = ω µν = − ωαβ (M αβ )µν .g. with ϕi = − 1 ǫijk ω jk and J i = 2 z-axis). the Lorentz transformations form a (Lie) group (Λ1 Λ2 and Λ−1 are again Lorentz transformations. 3. we continue using covariance arguments. inf (3. J 3 = M 12 ) and for the (inﬁnitesimal) boosts (along (3.36) (3. Of the four parts only L↑ forms a group.39) (e. K = M ).34) The condition ΛT GΛ = G yields g ρµ + ω ρµ gρσ (g σν + ω σν ) = gµν =⇒ ωνµ + ωµν = 0. inﬁnitesimally approximated by (I + ω. ǫ). [M µν . The extension to the Poincar´ group is e ↑ straightforward. 2 (M αβ )µν = i g αµ g βν − g α g βµ ν (3. a).g.38) Explicitly.35) thus (ω)ij = −(ω)ji and (ω)0i = (ω)i0 . (3.40) ΛB = I + i η 3 K 3 = I − i ω30 M 30 . We therefore ﬁnd (again) that there are six generators for the Lorentz group. Summarizing. (3. P+ etc. which are obtained by writing the inﬁnitesimal parameters (ω)µν in terms of six antisymmetric matrices (M αβ )µν .4 The generators of the Poincar´ group e ↑ The transformations belonging to P+ are denoted as (Λ.33) These four transformations form the Vierer group (group of Klein). M ρσ ] = −i (g µρ M νσ − g νρ M µσ ) − i (g µσ M ρν − g νσ M ρµ ) . three of which only involve spatial coordinates (rotations) and three others involving time components (boosts). 3 30 with η = ω and K = M (e. We will just require Poincar´ invariance for the generators themselves. The e 2 These commutation relations are the same as those for the ’quantummechanics’ operators i(xµ ∂ ν − xν ∂ µ ) . We now want to ﬁnd a covariant form of the six generators of the Lorentz transformations. This is a + normal subgroup and the factor group is the Vierer group.

5 Representations of the Poincar´ group e where pµ is a set of four arbitrary real numbers. = −i (g µρ P ν − g νρ P µ ) . P ρ ] = 0.38 could have been obtained in the same way. that commute e among themselves. (3. These form a group called the little group associated with that four vector. [K i . . = pµ |p. a) = Λµν P ν or U † (Λ. [P µ . [J i . the generators J 2 and J 3 in the case of the rotation group. leads to U (Λ. Inﬁnitesimally. P j ] = i δ ij H/c2 . H] = [J i . 3. We have here reinstated c. one obtains 2 [P i . K j ] = −i ǫijk J k /c2 .43) giving the following commutation relations by equating the coeﬃcients of ǫµ and ωµν . The requirement that P µ transforms as a four-vector (for which we know the explicit behavior from the deﬁning four-dimensional representation). K j ] = i ǫijk K k . P j ] = [P i . Taking any one of the states in an irreducible representation. P ) in terms of the Hamiltonian and the three-momentum operators. ǫ) = 1 − ωαβ M αβ + i ǫα P α . Of the generators of the Poincar´ group we choose ﬁrst of all the generators P µ . a)P µ U † (Λ.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. writing the generator P = (H/c. In order to label the states in an irreducible representation we construct a maximal set of commuting operators. Λµν pν ⇒ ωµν pν ⇒ ωµν = = = pµ + ω µν pν = pµ 0 ǫµνρσ pρ sσ . check that U1 U2 corresponds with Λ1 Λ2 ). [J i . [J i . . other states in that representation are obtained by the action of operators outside the maximal commuting set. a) = Λ−1 µ P ν . . H] = 0.45) Note that the commutation relations among the generators M µν in Eq. . These deﬁne states with speciﬁed quantum numbers that are eigenvalues of these generators. For instance. . (3. (3. known from non-relativistic quantum mechanics are obtained. (1 + iǫγ P γ − i i ωαβ M αβ )P µ (1 − iǫδ P δ + ωρσ M ρσ ) = P µ + ω µν P ν . the generators M αβ no longer exclusively act in Minkowski space. [K i . (3. P j ] = i ǫijk P k . P µ |p. the generators J± in the case of the rotation group. 3.44) (3. a)P µ U (Λ. as part of the set. H] = i P i .42) (to decide on where the inverse is.46) [K i .41) 2 At this point. because one then sees that by letting c → ∞ the commutation relations of the Galilei group. To ﬁnd other generators that commute with P µ we look for Lorentz transformations that leave the four vector pµ invariant. They just state that M µν transforms as a tensor with two Lorentz indices.48) . The generators of the translations are the (momentum) operators P µ . P ν ] [M µν . the tensor M µν in terms of boosts cK i = M i0 and rotations J i = 1 ǫijk M jk . 2 2 (3. . Explicitly. The eigenvalues of these will be the four-momentum pµ of the state.47) (3. ν (3. . J j ] = i ǫijk J k . For instance.

of which the components generate the little group of pµ or (for the third of the following relations) from explicit calculation [Mµν .49) (3. Wα ] = [Wµ . . From the algebra of the generators one ﬁnds that P 2 = Pµ P µ and W 2 = Wµ W µ (3. . (3. (3. i exp − ωµν M µν 2 i exp − ǫµνρσ M µν pρ sσ 2 (3. . 2 exp (−isµ W µ ) |p. satisfying gµν nµ (p)nν (p) = j i nµ (p) pµ i ǫµνρσ pµ nν nρ nσ i j k = = −δij . . M ǫijk . . while the others have no physical signiﬁcance. Wν ] = −i (gµα Wν − gνα Wµ ) . The elements of the little group thus are U (Λ(p)) = = and U (Λ(p))|p.56) . p0 > 0. 0.51) The following properties follow from the fact that W µ is a four-vector by construction. which can thus be chosen spacelike.52) (3. . 2 (3. 0. Pν ] = [Wµ . . . 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.53) (3.Groups and their representations 22 where s is an arbitrary vector of which the length and the component along pσ are irrelevant. = = i exp − ǫµνρσ M µν pρ sσ |p. i ǫµνρσ W ρ P σ .54) This will enable us to pick a suitable commuting ’spin’ operator.55) commute with all generators and therefore are invariants under Poincar´ transformations. Given the momentum four vector pµ we choose a tetrade consisting of three orthogonal spacelike unit vectors ni (p). Massive particles: p2 = M 2 > 0. . These e operators are the Casimir operators of the algebra and can be used to deﬁne the representations of ↑ P+ . the third case represents the vacuum.50) where the Pauli-Lubanski operators W µ are given by 1 Wµ = − ǫµνρσ M νρ P σ .

.62) 1 Thus W 2 has the eigenvalues −M 2 s(s + 1) with s = 0.e. 1. (3. 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.e.57 The best insight in the meaning of the operators W µ is to sit in the rest frame of the particle and put P µ = (M. Noting that S i (p) i 2 = 1 W µ (p)W ν (p) M2 niµ (p)niν (p). 0. S j (p)] = −i nµ (p)nν (p) j i ǫµνρσ W ρ P σ = i ǫijk S k (p).64) such that in an arbitrary frame where the four vectors p. .60) and using the completeness relation nµ (p)nν (p) = − g µν − i i 2 i pµ pν M2 . (3. The commutation relations [W i (p). show that W /M can be interpreted as the intrinsic spin. n1 (p). 2 . (0.57) (i. ms . (1.65) . W i (p) = −W · ni (p)).63) where E = p2 + M 2 and ms is the z-component of the spin −s ≤ ms ≤ +s (in steps of one). 0. Together with the four momentum states thus can be labeled as |M.61) one sees that S i (p) i =− W2 . 1 2 (3. 0). M2 (3. M2 (3. 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. . M J) with components W i = (M/2)ǫijk M jk . 0). −1) n1 (p0 ) ≡ n2 (p0 ) ≡ s(p0 ) ≡ (3.Groups and their representations 23 We can write W µ (p) = 3 W i (p) nµ (p) i i=1 (3. |p3 |). The explicit construction of the wave function can be done using the D-functions (analogous as the rotation functions). it is suﬃcient to choose a particular frame to investigate the coeﬃcients in Eq. 0. s. In this case a set of four independent vectors is chosen starting with a reference frame in which pµ = pµ 0 in the following way: p0 = (p0 . 0. i (3. 0). 1. 1. In that case the vectors ni (p) are just the space directions and W = (0.59) (3. 0. 3. More generally (fully covariantly) one can proceed by deﬁning S i (p) ≡ W i (p)/M and obtain [S i (p).. p0 > 0.58) i. 0. (0. the S i (p) form the generators of an SU (2) subgroup that belongs to the maximal set of commuting operators. p. Having made a covariant decomposition. W j (p)] = i M ǫijk W k (p).

a)|0 = |0 .1 (a) Derive from the relation that deﬁnes the mapping of SU (2) matrices A into SO(3) matrices RA in Eq. Wν ]. λ .67) (3. The algebra of W 1 .73) (3. (3.Groups and their representations The above does actually include the massive case. But if M = 0 one has a diﬀerent algebra (Eq.70) as a commuting set with zero eigenvalues for P 2 and W 2 . Note that angular momentum does not contribute to helicity as L · p = 0. W 1 (p)] [λ(p). |p. σ · RA a ≡ A(σ · a)A−1 ˆ (valid for any vector a). however.13. and gives [W 1 (p). is characterized by its momentum and the helicity. W 2 (p)] [λ(p). 3. Thus 2 λ(p) = J ·p . This state is in physical applications nondegenerate and is denoted by |0 . The eigenvalues of W 1 and W 2 (in that case corresponding to the translations) thus can take any continuous values.68) (3. which can take any integer or half-integer value. thus. U (Λ. W µ (p) = λ(p) P µ . (3. The meaning of λ(p) is most easily seen by comparing for a momentum eigenstate W 0 (p) = λ(p) |p| with W 0 (p) = − 1 ǫ0ijk Mij pk = J · p.72) One has P µ |0 = W µ |0 = 0 and the state is invariant under Lorentz transformations. P µ . W 2 (3. 3. Exercises Exercise 3. Thus we have P 2 . The vacuum: pµ = 0.69) We reproduce the algebra for the massive case with W 3 (p) = M λ(p). A massless particle.66) where W · p = 0 implies that W s (p) = −(M 2 /p · s) λ(p).67 has a vanishing right hand side). where A = exp(i ϕ · σ/2) = exp(iϕ σ · n/2). W 2 and λ is derived from the general commutation relations for [Wµ . 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 . 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). W 2 (p)] = = = i M 2 λ(p). implying continuous values for the spin (actually for W i = M S i ). |p| (3. This algebra is (for instance) isomorphic to the one generated by rotations and translations in a 2-dimensional Euclidean plane (see exercises). i W 2 (p). −i W 1 (p).71) which is called the helicity. 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). 1 2 24 (3. We don’t know of any physical relevance.

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

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

similarly as the homeomorphism between SO(3) and SU (2). j) K = −i J (B = 0). With this choice of parameter-spaces the plus and minus signs are actually relevant. C) can be written as a product of a unitary + matrix U and a hermitean matrix H.Chapter 4 The Dirac equation 4.10) . K = i J (A = 0). It is simply connected and forms the covering group of L↑ . i ǫijk B k .9) (4.8) ˆ ˆ ˆ with η = η n and ϕ = ϕn. 2 1 (J − i K). C) Instead of the generators J and K of the homogeneous Lorentz transformations we can use the (hermitean) combinations A = B which satisfy the commutation relations [Ai .3) (4. where we restrict (for ﬁxed n) the parameters 0 ≤ ϕ ≤ 2π and 0 ≤ η < ∞. K = −i .5) = 1 (J + i K). 2 2 σ σ J = . 2 (4. it also follows directly that the Lorentz group is homeomorphic with the group SL(2.7) From the considerations above. respectively. U (ϕ) H(η) (4. M = exp i ϕ·σ 2 1 exp ± η · σ 2 = U (ϕ) H(η) . The matrices in SL(2. C) is the group of complex 2 × 2 matrices with determinant one. They precisely correspond to the two types of representations that we have seen before.1 The Lorentz group and SL(2. K = +i . (j. B j ] = i ǫijk Ak . The group SL(2. becoming the deﬁning representations of SL(2.1) (4. 0.4) (4. Special cases are the following representations: Type I : (j. C). This tells us how to ﬁnd the representations of the group. j ′ ). B j ] = [Ai .2) This shows that the Lie algebra of the Lorentz group is identical to that of SU (2) ⊗ SU (2). C): Type I (denoted M ): Type II (denoted M ): 27 J= σ σ . They will be labeled by two angular momenta corresponding to the A and B groups. 0) Type II : (0. (4.6) (4. (4. 2 2 (4. Aj ] = [B i .

ξ −→ Hξ. U σ µ aµ U ˜ −1 H σ µ aµ H −1 = σ µ ΛB aµ H σ µ aµ H ˜ −1 (4. thus.18 show that ±ǫξ ∗ transforms as a type-II (η) spinor.18) As already discussed for SU (2). This shows that M ∗ ≃ M . η → Hη. are not covariant!) ˜ can be used to establish the homeomorphism between L↑ and SL(2. This is not true for the two-component spinors in SL(2.15) (4. σ).19) . namely H(η) = exp(η · σ/2). ˜ = σ µ Λ B aµ . 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. ǫ−1 = ǫ† = ǫT = −ǫ).e. (4. (ǫ∗ = ǫ. U σ µ aµ U −1 = σ µ ΛR aµ . Eqs 4.12) ξ → U ξ.1). there exists the matrix ǫ = iσ 2 . One has + −1 Λ(M ∗ ) = I2 Λ(M ) I2 . In fact. (4. Note that M ∗ = ǫ† M ǫ has its equivalent for the Lorentz transformations Λ(M ) and Λ(M ∗ ). C). 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). Λ(ǫ) ∈ L+ .11) Considering ξ and η as spin states in the rest-frame. C) or L↑ . cannot be connected by a Lorentz transformation belonging to L↑ . if there does not exist a unitary matrix S.13) 2M (E + M ) (exercise 4. (4. Also useful is the relation H 2 (p) = σ µ pµ /M = (E + σ · p)/M . η → U η. i.The Dirac equation 28 Let us investigate the deﬁning (two-dimensional) representations of SL(2. 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. With E = M γ = M cosh(η) and p = M βγ n = M sinh(η) n. such that M = SM S −1 .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. σ µ ≡ (1. The following relations for + rotations and boosts are useful. ˜ (4.16) = σ µ Λ R aµ . H ≡ (H † )−1 = H −1 ). thus. C) and. One deﬁnes spinors ξ and η transforming similarly under unitary rotations (U † = U −1 . (4. ˜ The sets of four operators deﬁned by deﬁnitions σ µ ≡ (1.14) satisfying Tr(σ µ σ ν ) = −2 gµν and Tr(σ µ σ ν ) = 2gµν = 2 δ µν (the matrices. while ±ǫη ∗ transforms as a type-I + spinor. however. The Lorentz transformations Λ(M ) and Λ(M ). H(η) = exp(−η · σ/2). U ≡ (U † )−1 = U ) U (ϕ) = exp(i ϕ · σ/2) but diﬀerently under hermitean boosts (H † = H. C). ¯ ¯ H ∗ = ǫ† Hǫ = ǫ−1 Hǫ. ˜ Non-equivalence of M and M One might wonder if type I and type II representations are not equivalent. −σ). one can use a boost to the frame with momentum ˆ ˆ p. (4.

Taking M (Λ) = H(p). we have seen that in the rest frame W i (p)/M → J i . 4. 1 ) and ( 1 . C) are suitable for representing spin 1/2 particles. they cannot be connected by a unitary 2 2 transformation. (axial vector). e. From the deﬁnition of the representations (0. For a particle at rest only angular momentum is important and we can choose ξ(0. Within the Lorentz group.g.2 Spin 1/2 representations of the Lorentz group 1 1 Both representations (0. consider the four component spinor that transforms under a Lorentz transformation as 0 ξ ξ . where a = (a0 . the well-known two-component spinor for a spin 1/2 particle. 4. −a).e. 2 2 (4.17. = (4. 0) of SL(2. 0) are inequivalent. 0 χm uR H(p) . the boost in Eq.13.1. but to L↑ . must combine the repe resentations. 0) (via operators J and K) one sees that under parity 2 2 1 1 (0.22) u= η η M (Λ) 0 where M (Λ) = ǫM ∗ (Λ)ǫ−1 with ǫ given in Eq. Thus M ∗ and M are not equivalent within SL(2. 2 ) and ( 2 . (4.20) 4. For the spin 1/2 representations of the Poincar´ group including parity we. but rather the operators W i (p) should be used. and the angular momentum in the rest frame is what we are familiar with as the spin of a particle. as we have seen in section 4. The i angular momentum operators J are represented by σ i /2. We also have seen that the representations (0. they can be connected. we obtain for the two components of u which we will refer to as chiral right and chiral left components. This provides the easiest way of seeing ˜ what is happening. the aij elements of a tensor will not change sign. The behavior is the same for classical quantities. a) and ˜ a = (a0 . + − 1 0 0 1 I2 = 0 0 0 0 0 0 −1 0 0 0 0 1 .e.21) In nature parity turns (often) out to be a good quantum number for elementary particle states. therefore. m) = χm uL H(p) 0 . −→ −p (vector).The Dirac equation 29 where The matrix I2 does not belong to L↑ . m) = η(0. generators. The representing member of the class P− is the parity or space inversion operator Is . m) = χm . (pseudoscalar). etc. A parity transformation changes aµ into aµ . i. 0). but by a transformation belonging ↑ ↑ to the class P− . Although these operators cannot be used to label the representations.23) u(p. (vector). ǫµνρσ will change sign. Examples (with between brackets the Euclidean assigments) are r t p H J λ(p) K −→ −λ(p) −→ −K −→ H −→ J −→ t (scalar). that have the correct commutation relations. 1 ) and ( 1 . (scalar). i. −→ −r (vector). It has the same eﬀect as lowering the indices. −→ M (Λ) (4. ) −→ ( . C).

2 (4. 0 H 2 (p) uR uR .The Dirac equation 30 with H(p) = ¯ H(p) = 2M (E + M ) E+M −σ·p 2M (E + M ) E+M +σ·p . (4. From this one obtains the Cliﬀord algebra for the Dirac matrices.27 is known as the Weyl representation. We ﬁrst want to investigate if the Dirac equation solves the problem with the negative densities and negative energies. ψ † −iγ 0 ∂0 +iγ i ∂i −M 1 We ← ← = 0. Applying (iγ µ ∂µ + M ) from the left gives (4.24) (4. (pµ γ µ − M ) u(p) ≡ (/ − M ) u(p) = 0. To see this consider the equation for the hermitean † † conjugate spinor ψ † (noting that γ0 = γ0 and γi = −γi ).25) It is straightforward to eliminate χm and obtain the following constraint on the components of u. γ } ∂µ ∂ν + M 2 ψ(x) = 0. = (4. The ﬁrst indeed is solved.32) suppressing on the RHS the identity matrix in Dirac space. one must require that for ψ(x) the Klein-Gordon relation 2 + M 2 ψ(x) = 0 is valid for each component separately). The general requirements for the γ matrices are thus easily obtained. . which in general is a linear equation in pµ . (4. Since ∂µ ∂ν is symmetric. We will discuss another explicit realization of this algebra in the next section. It is of a form that we also played with in Exercise 2.31) To achieve also that the energy-momentum relation p2 = M 2 is satisﬁed for u(p). this can be rewritten 1 µ ν {γ .33) deﬁne for a four vector a the contraction a = aµ γµ / .26) −2 uL uL H (p) 0 or explicitly in the socalled Weyl representation −M E+σ·p E−σ·p −M uR uL = 0. {γ µ .28) which is a covariant (linear) ﬁrst order diﬀerential equation. (4.27) where γ µ are 4 × 4 matrices called the Dirac matrices1 . p (4. (4. This leads to the Dirac equation for ψ(x) = u(p)e−i p·x in coordinate space. (iγ µ ∂µ − M ) ψ(x) = 0. The explicit realization appearing in Eq.7. = H −1 (p). (4. γ ν } = 2 g µν . As in section 2 we can use Pµ = i∂µ as a representation for the momenta (translation operators) in function space.29) which is an explicit realization of the (momentum space) Dirac equation.30) γ µ γ ν ∂µ ∂ν + M 2 ψ(x) = 0. 4.

where γ 0 p0 ψ = M ψ 0 −→ Eψ = M γ 0 ψ. remains. (4. The second problem encountered before. E = +M (twice) and E = −M (twice). opposite charge). γ ν } = 2 g µν .1: The Dirac sea of negative energy states and antiparticles.37) Using the explicit form of γ in the Weyl representation (see Eq. .36) (4. (4.35) is indeed positive and j 0 can serve as a probability density. Relying on the Pauli exclusion principle for spin 1/2 particles. (4.34) for the adjoint spinor ψ ≡ ψ † γ0 . but with properties such that it can be annihilated by the particle (e. Multiplying with γ 0 on the right and pulling it through one ﬁnds a covariant equation ψ(x) iγ µ ∂µ +M ← = 0. 4. . We will see how this is properly implemented (also for bosons!) when quantizing ﬁelds. Dirac supposed that the negative energy states were already completely ﬁlled and the exclusion principle prevents any more particles to enter the sea of negative energy states. The Dirac sea forms the vacuum. the one of the negative energy states.39) use ρi ⊗ σj with both ρ and σ being the standard 2 × 2 Pauli matrices. 4.The Dirac equation 31 E empty levels M -M filled levels E particle M -M hole Figure 4. This hole forms an antiparticle with the same mass.g.3 General representations of γ matrices and Dirac spinors {γ µ . From the vacuum a particle can be removed. one sees that there are two positive and two negative eigenvalues.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 . From the equations for ψ and ψ one immediately sees that j µ = ψγ µ ψ is a conserved probability current (exercise!). The density j 0 = ψγ 0 ψ = ψ † ψ (4. This problem was solved by Dirac through the introduction of a negative energy sea. γ =ρ ⊗1= γ = iρ ⊗ σ = 0 −1 −σ k 0 2 We (4. This can most easily be seen in the particle rest frame.27).

9 and 4. 0 (4. Lorentz invariance The Lorentz transformations can also be written in terms of Dirac matrices.49) i 1 with −i S ρσ = 2 σ ρσ = − 4 [γ ρ .47) (4. We note that ψ → Lψ and ψ → ψ L−1 . (4. it is easy to see that PR/L = 1 (1 ± γ5 ) 2 1 0 . while L−1 γ µ L = Λµ γ ν .R.43) S=√ . which in the case of the Weyl representation are just the upper and lower components.41) (4. γ 2 ].10 are 1 σ3 0 i J 3 = M 12 = = [γ 1 . −→ Sγµ S −1 . 0 γ k = −iρ2 ⊗ σ k = k σ −σ k . For instance in the Weyl representation (but valid generally). γ 0 ]. S −1 . K 3 = M 30 = 0 iσ 3 2 4 or in general one has the transformation i ψ → Lψ = exp − ωρσ S ρσ ψ.3) .42) We note that the explicit matrix taking us from the Weyl representation to the standard representation. γ µ } = 0 and explicitly one has 0 1 1 .45) (γ5 )W. that project out chiral right/left states. γ 0 = ρ1 ⊗ 1 = 1 0 γµ ψ Diﬀerent representations can be related to each other by unitary transformations. =ρ ⊗1= 0 −1 3 (4.46) (4. = S(γµ )W. the rotation and boost generators in Weyl representation in Eqs 4.The Dirac equation 32 The Weyl (or chiral) representation: 0 1 . (γ5 )S.R.48) are projection operators. γ σ ].44) and an adjoint spinor deﬁned by ψ = ψ † γ0 .R.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 . 4! (4.40) (4. For example. 0 σ3 2 4 1 −iσ 3 0 = − i [γ 3 . 2 (4. ν The latter assures Lorentz invariance of the Dirac equation (see Exercise 4. (γµ )S. is 1 1 1 (4. 1 −1 2 For all representations one has † γµ = γ0 γµ γ0 . = ρ ⊗ 1 = 1 0 i ǫµνρσ µ ν ρ σ γ γ γ γ . −→ Sψ.

parity. x = (t.The Dirac equation 33 Parity There are a number of symmetries in the Dirac equation. ∂ (4.R. but it does. Explicitly. e. ∂ (4. that brings ψ → ψ c . (C)S. −x) and via a few manipulations obtain again the ˜ Dirac equation (i/ − M ) ψ p (x) = 0.R.55) (4. T (4. ψR . 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) = 0.50) Charge conjugation but with ψ p (x) ≡ ηp γ0 ψ(˜) (Exercise 4. usually to be taken unity. = = 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. called charge conjugation. x) → x = (t.53) (4. is a left-handed spinor. which is again a solution of the Dirac equation.R. As with parity we look for a transformation. (i/ − M ) ψ(x) = 0. ∂ with the solution ψ c (x) = ηc Cψ (x).) and all representations connected via a real (up to an overall phase) matrix S.R.g. In any representation a matrix C exist.59) −→ ψ c = C ψ = η ǫ ξ∗ . = 0 ǫ iσ 2 (4. This symmetry does not change the spin 1/2 nature.51) The existence of positive and negative energy solutions implies another symmetry in the Dirac equation. such that T Cγµ C −1 = −γµ . reverse the charge of the particle.56) (i/ − M ) ψ c (x) = 0. Some properties of C are C −1 = C † and C T = −C.52) η ξ one can apply space inversion.58) where ηc is an arbitrary (unobservable) phase. 0 −ǫ 0 0 −ǫ 0 . Note that we have (as expected) explicitly in Weyl x representation in Dirac space P ξ −→ ψ p = γ 0 ψ = η . (or W.g. C is real and one has C −1 = C † = C T = −C c and [C. γ5 ] = 0. −iσ 2 0 −ǫ = .54) e. The latter implies that the conjugate of a right-handed spinor. for instance.57) (4. In S. It is easy to convince oneself that if ψ(x) is a solution of the Dirac equation.6). ψ= (4. (C)W. One has ψ c = −ψ T C −1 . ψ= (4. T (4. in Weyl representation we ﬁnd in Dirac space T C ξ −ǫ η ∗ .

s ) = v (p. s)v(p. where v satisﬁes σ·p −(Ep + M ) v(p) = 0. the spin polarization vector in the rest frame is the starting ˆ point.63) u(p. p −σ · p (Ep − M ) Ep + M σ ·p ¯ Ep +M χs χs ¯ For a free massive particle. p (4.65) (4. In that frame is a spacelike unit vector sµ = (0. s) u = . s) v = . v (p. s′ ) = 2M δss′ . s ) = 0. The ¯ ¯ solutions are normalized to u(p. ψ(x) = v ±s (p) ei p·x . 2M 2M (4. After separating positive and negative energy solutions as done in the case of the scalar ﬁeld one has ψ(x) = s d3 k u(k. s). s)u(p. The explicit Dirac equation in the standard representation reads ∂ iσ · ∇ i ∂t − M (4. s)¯(p. s) = Ep + M σ ·p Ep +M χs . It is easy to check that 1 1±σ·s 1 ± γ5 / s ˆ 0 (4. in which γ 0 is diagonal (see discussion of negative energy states in section 4. be obtained by a Lorentz transformation. with E = Ep = + p2 + M 2 . s) . the best representation to describe particles at rest is the standard representation. ¯ (4.. ψ(x) = u±s (p) e−i p·x . s)¯(p. s ) = v † (p. where u satisﬁes −σ · p Ep − M u(p) = 0. s). ⇔ (/ − M ) u(p) = 0. s)v(p.4 Plane wave solutions Looking for positive energy solutions ∝ exp(−iEt) one ﬁnds two solutions.64) (4. s′ ) = 2Ep δss′ . 2M 2M −/ + M p v(p. s) = u(p. (2π)3 2Ek (4. s′ ) = −2M δss′ .1). s) ei k·x d∗ (k.The Dirac equation 34 4. s)u(p.g.67) It is straightforward to ﬁnd projection operators for the positive and negative energy states P+ P− = s p /+M u(p. (4. ¯ ¯ † ′ u (p. In an arbitrary frame one has s · p = 0 and s(p) can e. s) + v(k. ⇔ (/ + M ) v(p) = 0. s) and C v T (p.66) An arbitrary spin 1/2 ﬁeld can be expanded in the independent solutions.70) = Ps = ˆ 0 1∓σ·s 2 2 .62) where χs are two independent (s = ±) two-component spinors.69) = − s In order to project out spin states. s) e−i k·x b(k. ¯ ′ ′ Explicit solutions in the standard representation are χs .60) ψ(x) = 0. s)v(p. Choosing χ = −ǫχ∗ (the equivalent spin ¯ 1/2 conjugate representation). ∂ −iσ · ∇ −i ∂t − M (4. Note that the spinors in the negative energy modes (antiparticles) could be two diﬀerent spinors. s) = v(p.61) σ·p −(Ep + M ) There are also two negative energy solutions. the spinors satisfy C uT (p.68) (4. v(p. s) = u(p. s)u(p.

etc. (4. λ = − 1 ) = √ √ v(p. . We note that for high energy ¯ ¯ the positive helicity (λ = +1/2) is in essence righthanded. Explicit examples of spinors are useful to illustrate spin eigenstates. u(p. one has in standard representation: 0 E+M E+M 1 1 0 .73) Also for helicity states C uT (p. p ψR/L ≡ PR/L ψ are independent solutions. helicity states. λ = −1/2) = u(p. Therefore. λ) = u(p. λ = + 2 ) = √ E +M − E −M 2 0√ 2 2 √ 0 E +M + E −M √ √ 0√ √ − E +M − E −M 1 1 E+M − E−M 0 √ . For instance with the z-axis as spin quantization axis. +1/2) = √ u(p. This means that in the solution space for massless fermions the chirality states. λ = +1/2) = v(p. λ). λ) and C uT (p. λ = +1/2) = E−M √ 0 − E −M 0 √ √ 0 − E−M − E −M 0 . 1 u(p. One ﬁnds for the helicity states in Weyl representation: √ √ 0√ E +M + E −M √ 1 √ 1 0√ 1 . √ 0 u(p. while the negative helicity (λ = −1/2) is in essence lefthanded.72) ˆ Helicity states (p along z ) in Standard representation are: √ √ 0 E+M E +M . The chirality projection operators in Eq. λ = + 1 ) = √ + E +M − E −M . λ = −1/2) = √ − E+M √ 0 0 E+M (4.71) p − i p2 . λ = − 1 ) = √ E + M − E − M . v(p. Note that for solutions of the massless Dirac equation /ψ = 0.The Dirac equation 35 (the last equality in the restframe and in standard representation) projects out spin ±1/2 states (check ˆ ˆ this in the restframe for s = z ). 4. In principle massless fermions can be described by twocomponent spinors. 2 2 0 √ 2 √ 2 − E+M − E−M 0 . λ) = v(p. By writing the helicity states in Weyl representation it is easy to project out righthanded (upper components) and lefthanded (lower components). chirality. (4. = √ =⇒ 2E E+M + E−M ( E + M + E − M )2 which vanishes in the ultra-relativistic limit E ≫ M or in the massless case. v(p.48 replace the spin projection operators which are not deﬁned (by lack of a rest frame). −1/2) = √ E + M −(E + M ) 0 . −1/2) = √ (4. −p3 −p1 − i p2 1 v(p. u(p. γ5 /ψ = 0 but also p p p /γ5 ψ = −γ5 /ψ = 0. The ratio of the components is (directly or inversely) proportional to √ √ E+M − E−M 2M M ≪E M √ √ √ . +1/2) = √ 0 E +M E+M . p3 E +M E+M 1 2 3 p + ip −p 1 2 p − ip 3 1 −p v(p.74) .

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

77) Exercises Excercise 4. s) and v(p. Do this e. 4 2 Exercise 4. H(p) = exp η·σ 2 = M +E+σ·p . γ σ ] = σ ρσ . e. ν µ µ Look at the inﬁnitesimal forms Λµ = gν + ων and L = 1 − (i/2)ωρσ S ρσ to obtain ν [γ µ . i S ρσ ] = g ρµ γ σ − g σµ γ ρ .13.g. 0 if n is odd. Exercise 4. by letting them act on the four basis spinors u(p. 2M (E + M ) ˆ ˆ where E = M γ = M cosh(η) and p = M βγ n = M sinh(η) n. to the Dirac equation and use this to show that the spinor ψ p (x) = ˜ γ0 ψ(˜). 4. s). a 4 a · b. ˜ Exercise 4. by proving that H 2 (p) = exp (η · σ) = (E + σ · p)/M = σ µ pµ /M. s)¯(p.g. −x) is also a solution of the Dirac equation. Show that the equality holds. s) u p /+M = 2M 2M both sides are projection operators. of which only the traces of four and two gamma-matrices are nonzero.2 Show that j µ = ψγ µ ψ is a conserved current if ψ en ψ are solutions of the Dirac equation Exercise 4.5 Show starting from the relation {γµ .6 Apply space-inversion. 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 . x ˜ .1 Prove Eq. Excercise 4. γν } = 2gµν that γµ / γ µ a = Tr(//) = ab µn µ1 Tr(γ · · · γ ) = −2 /. 1 i −i S ρσ = − [γ ρ . (4. where x = (t.4 Show that in P+ ≡ s=±1/2 u(p.The Dirac equation 37 The trace is linear and can be split up in parts containing traces with up to four gamma-matrices. x → x.3 Imposing Lorentz invariance on the Dirac equation can be achieved by requiring L−1 γ µ L = Λµ γ ν .

Note: To prove this.66 MeV. (d) Plot for these cases the vector densities ψ † ψ and the scalar densities ψψ as a function of r/R. there is no current ﬂowing through the surface of the cavity. |M (π − → µ− νµ )|2 ¯ Actually if the lepton would be massless.5 and M → ∞. Show that the ratio is |M (π − → e− νe )|2 ¯ ≈ 54.57 MeV. n (c) Apply the conﬁnement condition to ﬁnd the value of k for the lowest eigenmode in the spherical ˆ cavity. ν p Note: how such parametrizations work will be explained in chapter 8. n It turns out that there is actually a whole family of conditions that provide conﬁnement. Excercise 4. A condition for conﬁning the fermion in the cavity is i / ψ = ψ (where n is the surface normal). the lepton and neutrino have spin 1/2 and are ν described by momentum-space spinors.The Dirac equation 38 Exercise 4. Show that this n condition is suﬃcient to guarantee that nµ j µ = 0. The pion has spin 0. t) = N k ˆ i E+M σ · r j1 (kr)χm −i Et e . In this solution the wave number k = |k| √ = E 2 − M 2 and χm a two-component spinor while j0 and j1 are spherical Bessel functions.511 MeV). namely i / ψ = eiαγ5 ψ. Calculate for now just |M |2 . We will not pursue this chiral freedom at this point and take α = 0. is a stationary solution of the (free) Dirac equation. we want to conﬁne this solution to a spherical cavity with radius R.8 × 10−6 . me = 0. you need to determine the corresponding conﬁnement condition for ψ. Do you understand why? The calculation of the life-time is completed in chapter 11. . mµ = 105. (b) Next. Calculate the value of k for M R = 0. The conditions for ψ and ψ imply that nµ j µ = ψ/ ψ = 0. r).e. M R = 1.8 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 ). Note: the outcome shows that the decay into muons is much more likely that the decay into electrons (mπ = 139. i. which is the most important factor needed to ﬁnd the decay width (inverse lifetime).7 (a) Show that (in standard representation) j0 (kr)χm ψ(x. the decay would even be forbidden. For such a cavity nµ = (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.

ǫλ ). (2 + M 2 )Vµ = 0. λ) + ǫ(λ)∗ (k) eik·x c∗ (k.5) −E B2 . In the restframe. there are then indeed three independent possibilities. λ) . −B 1 0 3 With the electromagnetic ﬁeld tensor (5.1 Fields for spin 1 In section 3.2 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. ±(1) are obtained by boosting the restframe vectors. A). A real vector ﬁeld Vµ (x) then can be expanded in modes as Vµ (x) = λ=0. 39 (5. where k = k0 = (M. These can describe spin 1 states in the rest-frame.2) (5.2 we have discussed the vectors ǫλ (with λ = 0 and ±1) as basis states of the spin 1 representation. 0). These are solutions of the Klein-Gordon equation with in addition a condition. with ∂µ V µ = 0 (Lorentz condition). just as the two-component Dirac spinors did for spin 1/2.3) Since the vectors ǫµ (k) together with the momentum k µ form a complete set of four-vectors. one has ǫ(λ) (k) ǫ(λ)∗ (k) = −gµν + µ ν kµ kν . M2 (5. The vectorﬁeld is therefore (λ) suited to describe spin 1.1) 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 . namely ǫµ (k0 ) = (0. (5.4) λ=0. µ (2π)3 2E µ (5.7) .± 5. In an arbitrary frame ǫµ (k) with λ = 0.± (λ) d3 k ǫ(λ) (k) e−ik·x c(k.Chapter 5 Vector ﬁelds and Maxwell equations 5.6) where Aµ is the four-vector potential (5.

˜ x Note that via parity transformation one obtains another solution. 0. The Lorenz condition can actually be incorporated by changing the equations of motion to 2Aµ + (λ − 1)∂µ (∂ν Aν ) = jµ . one ﬁnds for a conserved current and λ = 0 2∂µ Aµ = 0.16) (5.8) (5. Taking the divergence of this equiation. transverse or Coulomb gauge). |k|). The equations of motion for the ﬁelds Aµ become 2Aµ − ∂µ (∂ν Aν ) = jµ .e. if ∂µ Aµ = 0 for large times |t|. Under a gauge transformation Aµ −→ Aµ + ∂µ χ. (5.12) (5. −i. it will vanish at all times. moree over.15) (5. In that case one also has ∇·A = 0 or ki ǫi (k) = 0. (5. 0. if in Eq. 2Aµ = 0. (5. . it does not eliminate the freedom of gauge transformations but they are now restricted to Aµ → Aµ + ∂µ χ with 2χ = 0. (5.11) This equation is not aﬀected by a gauge transformation. there is still a gauge freedom left (Aµ → Aµ + ∂µ χ with 2χ = 0). (5. For electrodynamics one has furthermore the freedom of gauge transformations.17) (radiation.13) Hence. Note that the incorporation of the Lorentz condition can also be done for the case of the massive vector ﬁeld. in which case one has 2Aµ = jµ . 5. for M = 0. while the fact µ C ∗ that the Minkowski space is real implies that Aµ (x) = Aµ (x) = Aµ (x).14) of which the solutions give the Li´nard-Wiechert potentials.10) the electric and magnetic ﬁelds are unchanged.18) corresponding to the helicity states λ = ±1 of the massless photon ﬁeld. Therefore if k µ = (|k|. Furthermore. only two independent vectors remain ǫ(±) µ (k) = (0. It states the absence of magnetic charges and Faraday’s law.Vector ﬁelds and Maxwell equations and the (conserved) current j µ = (ρ.5. j) the Maxwell equations read ∂µ F µν = j ν . 40 (5. As remarked above. AP (x) = Aµ (˜).15 the d’Alembertian is replaced by 2 + M 2 . shows that the electromagnetic ﬁelds correspond to the case of a massless spin 1 ﬁeld. in accordance with what we found in section 3. 0). i. the antisymmetry of F µν implies that ˜ ∂µ F µν = 0 or ∂λ Fµν + ∂µ Fνλ + ∂ν Fλµ = 0. One possible choice is the gauge A0 = 0 (5. The equation in vacuum. ∓1. Although the Lorenz condition is a constraint. ∂µ Aµ = 0. The gauge freedom can ﬁrst of all be used to impose the Lorentz condition for the electromagnetic ﬁeld.9) 1 ˜ where F µν ≡ − 2 ǫµνρσ Fρσ is the tensor where E → B and B → −E. Fµν −→ Fµν + (∂µ ∂ν χ − ∂ν ∂µ χ) = Fµν .

0) and ϕ = (− sin 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 . A= By Bx Br ϕ= − ˆ . but A = 0 (hence there is structure in the vacuum).3 The electromagnetic ﬁeld and topology∗ The electric and magnetic ﬁelds E and B as appearing for instance in the Lorentz force on a moving charge. ϕ= − ˆ 2r 2r2 2r2 where r = (cos ϕ. may look a bit akward. λ L − λ λ or ∆x = (L− λ/d)δ.0 .Vector ﬁelds and Maxwell equations 41 L ∆x d a (infinitely) long solenoid Figure 5. The situation. but where E = B = 0. . 2 2 2 BR2 BR2 y BR2 x A= .1. This occurs in the region outside the solenoid.0 . This causes an interference pattern as a function of ∆x. . nothing more than a manisfes1 tation of a nontrivial vacuum (a la the Dirac sea for fermions). To be precise. ˆ B = 0.19) are gauge invariant. where Aµ = 0. Nevertheless the signiﬁcance of Aµ is shown in the Aharonov-Bohm experiment. The phase diﬀerence between the electrons travelling two diﬀerent paths is given by δ = 2π a a ∆x d = −≈ .1: The Aharonov-Bohm experiment 5. 0). This phase diﬀerence is actually observed. illustrated in ﬁg.20) e (∇ × A) · ds = B · ds = where Φ is the ﬂux through the solenoid. F = e E + ev × B (5. Inside solenoid: Outside solenoid: B = B z. cos ϕ. In this experiment an observable phase diﬀerence is measured for electrons moving through ﬁeld-free space (E = B = 0 but Aµ = 0). nevertheless. 5. In the presence of an electromagnetic ﬁeld Aµ an additional phase diﬀerence is observed. . The energy density 2 (E 2 + B 2 ) = 0 (hence a vacuum). It is however. sin ϕ. = 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. (5. while the electromagnetic ﬁeld Aµ is not.

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

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

These equations can be obtained from a lagrangian using the action principle. ∂x (6.3) (6.2) The principle of minimal action looks for a stationary action under variations in the coordinates and time. ∂x ˙ (6.1) and as an example the lagrangian L(x. t1 (6.6) The ﬁrst term leads to the Euler-Lagrange equations.1 Euler-Lagrange equations In the previous chapter we have seen the equations for scalar ﬁelds (Klein-Gordon equation). ˙ 2 (6. ˙ p= ∂L .8) 44 .5) with ∆x(t) = δx(t) + x(t) δτ . x) = K − V = ˙ 1 mx2 − V (x). recall classical mechanics with the action t2 S= t1 dt L(x. and the total change x′ (t′ ) = x(t) + ∆x(t). x) ˙ (6.Chapter 6 Classical lagrangian ﬁeld theory 6. Dirac ﬁelds (Dirac equation) and massless vector ﬁelds (Maxwell equations) and corresponding to these ﬁelds conserved currents describing the probability and probability current. (6. d dt ∂L ∂x ˙ = ∂L .7) The quantity ∂L/∂ x plays a special role and is known as the canonical momentum.4) δS = t1 t2 dt dt t1 ∂L ∂L δx + δ x + L δτ ˙ ∂x ∂x ˙ d ∂L − ∂x dt ∂L ∂x ˙ δx + t2 t1 = ∂L δx + L δτ ∂x ˙ t2 . x (t) = x(t) + δx(t). thus t′ = t + δτ. The requirement δS = 0 with ﬁxed boundaries x(t1 ) = x1 and x(t2 ) = ˙ x2 leads to t2 ′ (6. As an example.

Variations in the action can come from the coordinates or the ﬁelds.11) λ=0 of which p and H are the simplest examples related to space and time translations. dσµ . t2 ). ˙ x The second term in Eq. indicated as x′µ φ (x) or combined φ′ (x′ ) = φ(x) + ∆φ(x). We will come back to it in a bit more formal way in section 9. also indicated by ∂R (a more general volume in four-dimensional space-time with some boundary ∂R can also be considered).13) (6. which in classical mechanics vanishes at the boundary) does not play a role. ∂µ φ(x)) = R d4 x L (φ(x). x) and an action t2 S[φ] = t1 dt L = dt d3 x L (φ(x).9) which is done because the ﬁrst term (multiplying ∆x.e. their derivatives and possibly on the position.t1 ) (6. Consider a lagrangian density L which depends on these functions. . ∂µ φ(x).16) The change in variables x → x′ in the integration volume involves a surface variation of the form dσµ L δxµ . . x). indicated with δF [φ]/δφ should pose no problems. R (6.17) Furthermore the variations δφ and δ∂µ φ contribute to δS. In general any continuous transformation. − d3 x . . ∂µ φ′ . .14) (6. ∆x = (dx/dλ)λ=0 δλ and δt = (dt/dλ)λ=0 δλ.Classical lagrangian ﬁeld theory 45 For the lagrangian speciﬁed above this leads directly to Newton’s law p = m¨ = −∂V /∂x. but which in 3 dimensions includes conserved quantities related to rotations and boosts. .12) Here R indicates a space-time volume bounded by (R3 . ˙ (6. (6. x) ≡ p x − L . 6.t2 ) d3 x . . L (φ(x). think for instance of a temperature or density distribution or of an electromagnetic ﬁeld).10) One sees that invariance under time translations requires that H(t1 ) = H(t2 ). i. H is a conserved quantity. ∂µ φ(x)). .18) 1 Taking a functional derivative. t1 ) and R3 . ∂µ φ. In classical ﬁeld theory one proceeds in complete analogy but using functions depending on space and time (classical ﬁelds. = ∂R (R3 . . The resulting variation of the action is δS = R µ ′ = xµ + δxµ .6 can be rewritten as δS = . = φ(x) + δφ(x) (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 . . λ=0 (6. t1 (6. + ∂L ∆x − H δτ ∂x ˙ t2 . The hamiltonian H is deﬁned by H(p. δ(∂µ φ) (6. gives rise to a conserved quantity (at least if H is invariant!) Q=p dx dλ −H dt dλ . ∂R Note for the speciﬁc choice of the surface for constant times t1 and t2 . . with ∆φ(x) = δφ(x) + (∂µ φ)δx .15) d4 x′ L (φ′ . x′ ) − d4 x L (φ. (R3 .2.

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 δL ∂µ = . The integrand of the ﬁrst term must vanish.2 Lagrangians for spin 0. ∂ (6. For the complex scalar ﬁeld one conventionally uses L = = ∂µ φ∗ ∂ µ φ − M 2 φ∗ φ −φ∗ ∂µ ∂ µ + M 2 φ. the Dirac ﬁeld (spin 1/2) and the vector ﬁeld (spin 1) can be found.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. (6.Classical lagrangian ﬁeld theory 46 With for the situation of classical ﬁelds all variations of the ﬁelds and coordinates at the surface vanishing. (6.23) (6. δL δ(∂µ ψ) δL δψ one obtains immediately i / −M ∂ → i = − γ µψ 2 i → = ∂ / ψ − M ψ.21) which diﬀer only by surface terms. the second term is irrelevant. (6.20) (6. 2 ∗ (6. leading to the Euler-Lagrange equations.24) (6. One easily obtains (2 + M 2 )φ(x) = 0.26) 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. (2 + M )φ (x) = 0.29) . 2 ψ = 0.28) where the second line is not symmetric but in the action only diﬀers from the symmetric version by a surface term (partial integration). The scalar ﬁeld It is straightforward to derive the equations of motion for a real scalar ﬁeld φ from the lagrangian densities. leading to (2 + M 2 )φ(x) = 0. 2 (6. L = 1 1 ∂µ φ∂ µ φ − M 2 φ2 2 2 1 µ = − φ ∂µ ∂ + M 2 φ.19) δ(∂µ φ) δφ 6.27) (6.

33) We now introduce a ’real’ (four-component) spinor satisfying Υc = Υ with ψL . The reasons for Grassmann variables will become clear in the next chapter. 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.7).Classical lagrangian ﬁeld theory 47 and similarly from the variation with respect to ψ ψ i / +M ∂ ← = 0.35) Since the kinetic term in L separates naturally in left and right parts (or ξ and η). (6. or equivalently separate the ﬁeld into right. which comes because we in essence use ’real’ spinors. 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 . and thus (β ∗ α)∗ = α∗ β = −βα∗ . ψL ≡ (ψL )c = C ψL = 0 (6.38) (6. Peshkin and Schroeder or Exercise 12. . 2 (6. 1 M η † ǫη ∗ − M ∗ η T ǫη . 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. Using the twospinors ξ and η. We note that of which the left part coincides .g.30) It is often useful to link to the two-component spinors ξ and η which we started with in chapter 4. (6. (6. ∗ 0 ⇒ Υ ≡ −ǫ η ΥL = ψL = η η ∗ for which η0 = ǫ η0 .37) This gives an expression with a mass term that is actually of the same form as the Dirac mass term. that the lagrangian separates into two independent parts for M = 0. ∂ ∂ 2 2 (6.and lefthanded ones.g. Υi / Υ − ∂ 4 2 T −ǫ η ∗ c = ΥR . the mass term in the Dirac lagrangian 6. The Majorana case is in fact more general. 2 αβ = −βα.31) showing e. but note the factor 1/2 as compared to the Dirac lagrangian. namely2 LM (Majorana) = + (6. In that case one ﬁnds trivially L = ↔ ↔ 1 1 ψR i / ψR + ψL i / ψL − M (ψR ψL + ψL ψR ).31 is given by LM (Dirac) = −M ξ † η + η † ξ .32) There exists another possibility to write down a mass term with only one kind of ﬁelds. L = = ↔ 1 1 c c M ψL ψL + M ∗ ψL ψL ψL i / ψL − ∂ 2 2 ↔ 1 1 M ΥR ΥL + M ∗ ΥL ΥR .39) 2 This term is written down with η being an anticommuting Grassmann number for which (αβ)∗ = β ∗ α∗ = −α∗ β ∗ .34) (6.36) (6.

+ R 4 δL δL ∆φ − ∂ ν φ − L g µν δxν δ(∂µ φ) δ(∂µ φ) δL ∆φ − Θµν (x)δxν .18. In particular when the lagrangian is invariant under symmetries. δ(∂µ φ) (6. δ(∂µ φ) For continuous transformations. (6. t). which is the space-integral over the zero-component of a conserved current.. (6.46) For quantum ﬁelds (φ operator!) things become more subtle as it it is not possible to specify ˙ for instance φ and φ on the surface σ1 . These do not aﬀect the dynamics and will give rise to conserved currents. ψ(x) → ei eΛ ψ(x) or ∆ψ(x) = i eΛ ψ(x). implying ∂µ F µν + M 2 V ν = 0 and ∂µ V µ = 0.Classical lagrangian ﬁeld theory 48 where Fµν = ∂µ Vν − ∂ν Vµ . one can consider the variations at the initial or ﬁnal surface. ∆φ = (dφ/dλ)λ=0 δλ and δxµ = (dxµ /dλ)λ=0 δλ. (6.. Returning to δS the surface term is rewritten to δS = R d4 x . we get (omitting the parameter δλ) dxν dφ δL + Θµν (x) . . As an example for the classical ﬁeld. Q= d3 x J 0 (x.σ1 with σ = (R3 . δ(∂µ φ) (6. (6.42) Θµν = The variation δS thus can be expressed as δS with J µ (x) = − = δL ∂ ν φ − L g µν . which in a consistent picture precisely generate the symmetries. (6. consider U (1) transformations of the Dirac ﬁeld proportional to the charge e.40) 6.44) δL ∆φ + Θµν (x)δxν .45) J µ (x) = − δ(∂µ φ) dλ λ=0 dλ λ=0 Thus considering ∂R = σ2 . Q(t1 ) = Q(t2 ). requires the presence of a conserved quantity. we consider the second term in Eq. the presence of a symmetry that leaves the lagrangian invariant. In this case. these conserved quantities become operators. t).41) (6.3 Symmetries and conserved (Noether) currents Next. discussed in the next chapter. ∂µ J µ (x) = 0. 6.47) . .43) F (σ1 ) − F (σ2 ) = dσµ J µ (x) = ∂R R d4 x ∂µ J µ (x). + ∂R dσµ dσµ ∂R ≡ where d x . one immediately obtains δL = −∂ µ V ν + ∂ ν V µ − λ g µν (∂ρ V ρ ) = −F µν − λ g µν (∂ρ V ρ ) δ(∂µ Vν ) leading to the equations of motion ∂µ F µν + M 2 V ν + λ ∂ ν (∂ · V ) = 2 + M 2 V ν − (1 − λ)∂ ν (∂ · V ) = 0.

2 (6.g. (ω µν antisymmetric) 1 ωρσ (−i S ρσ )ij φj (x). we obtain (since δxµ = 0). including e the space-time translations and the Lorentz transformations.4 Space-time symmetries One kind of symmetries that leave the lagrangian invariant are the Poincar´ transformations. ↔ (6. For quantized ﬁelds this will become the expressions of the hamiltonian and the momentum operators in terms of the ﬁelds.3) j µ = i e φ∗ ∂µ φ.56) δxµ = ωρσ (aρσ )µν xν 2 with (aρσ )µν = aρσ µν = g ρµ g σν − g σµ g ρν . 1 (6. Translations µ Under translations (generated by Pop ) we have δxµ ∆φ(x) δφ(x) = ǫµ . . ˙ Q = d3 x j 0 (x.Classical lagrangian ﬁeld theory From the lagrangian for the Dirac ﬁeld.50) (6. Summarizing.55) Note that the coordinate transformations can be written in a form similar to that for the ﬁelds (matrices in internal space). t) is the conserved charge (Q = 0). The integral over the zero-component.49) These currents are conserved as discussed already in chapter 2. In the next section we will see the quantity show up as the charge operator. (6. Therefore there are four conserved currents Θµν (ν labeling the currents!) and four conserved quantities Pν = dσµ Θµν = d3 x Θ0ν . 6.52) The behavior of the ﬁeld under translations (δφ) is governed by the translational behavior of the argument in such a way that ∆φ = 0.57) For the ﬁelds the behavior under Lorentz transformations has been the subject of previous chapters.51) (6. Lorentz transformations In this case the transformation of the coordinates and ﬁelds are written as δxµ ∆φi (x) = = ω µν xν .53) These are the energy and momentum. = 0. (6.54) (6. omitting the parameter Λ jµ = − δL δL (i e ψi ) (−i e ψi ) − δ(∂µ ψi ) δ(∂µ ψ i ) 49 = e ψγ µ ψ. H = d3 x Θ00 (x). = −(∂µ φ)δxµ = −ǫµ ∂µ φ(x).48) For the complex scalar ﬁeld the U (1) transformations leave the lagrangian invariant and lead to the conserved current (see Exercise 6. (6. From Noether’s theorem one sees that the current Θµν ǫν is conserved. e. (6.

59) (6. (6.66) In section 6. In general this tensor is not symmetric.g. one has a tensor that satisﬁes (exercise 6.67) i. where H ρµν is the quantity appearing in M ρµν .5 (Abelian) gauge theories φ(x) → ei eΛ φ(x). φ(x). Any lagrangian containing derivatives.62) with Gρµν = (H ρµν + H µνρ + H νµρ )/2. Gauge transformations of the second kind or local gauge transformations are transformations of the type φ(x) → ei eΛ(x) φ(x). γσ ] This result for the Dirac ﬁeld has been discussed in chapter 3 (see also Exercise 4.3 we have seen global transformations or gauge transformations of the ﬁrst kind.61) A ﬁnal note concernes the symmetry properties of Θµν . ∂µ φ(x) + i e ∂µ Λ(x) e (6. • Vector ﬁeld Aµ : −i Sρσ = aρσ .3). φ(x) φ (x) ∂µ φ(x) ∗ → e → e → ei eΛ(x) φ(x).68) (6. (6. The current deduced from Noether’s theorem is Jµ = = = = − δL ∆φi + Θµν δxν δ(∂µ φi ) δL 1 ωρσ − (−i S ρσ )ij φj + Θµρ xσ − Θµσ xρ 2 δ(∂µ φi ) 1 ωρσ {H µρσ + Θµρ xσ − Θµσ xρ } 2 1 ωρσ M µρσ .58) (6.e. In some applications (speciﬁcally coupling to gravity) it is advantageous to have an equivalent current that is symmetric. ∂µ T µν (6.65) M µρσ = T µρ xσ − T µσ xρ . −i eΛ(x) ∗ i eΛ(x) (6.69) i eΛ(x) φ (x).60) Therefore there are six conserved currents M µρσ labeled by ρ and σ (antisymmetric) and corresponding to it there exist conserved quantities M ρσ = d3 x M 0ρσ . is not invariant under local gauge transformations.63) µν = ∂µ Θ . (6. Deﬁning T µν = Θµν − ∂ρ Gρµν .70) . 2 (6. e.Classical lagrangian ﬁeld theory 50 • Scalar ﬁeld φ: −i Sρσ = 0. independent of x. The lagrangians which we have considered sofar are invariant under global gauge transformations and corresponding to this there exist a conserved Noether current. in which the U (1) transformation involves an angle e Λ. (6.6) T µν = T νµ .64) (6. (6. 6. the angle of the transformation depends on the space-time point x. • Dirac ﬁeld ψ: −i Sρσ = −(1/4)[γρ . however.

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

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

Hint ≡ −gs µe · B = −gs Qe s · B. A) (see also exercise 6.4.Classical lagrangian ﬁeld theory 53 Exercise 6.6 (optional) Prove the properties of the tensor T µν in section 6.5 Show that the Dirac equation for an electron with charge −e in an external electromagnetic ﬁeld Aµ = (A0 . e 1 (−i∇ + eA)2 + σ · B − eA0 ψu (r. This relation is known as the Pauli equation. t) = Enr ψu (r. It shows that for a spin 1/2 particle the g-factor in the coupling of spin to a magnetic ﬁeld takes the value gs = 2. t). use ∂µ T µν = 0 and ∂µ M µρσ = 0. . Hints: it may be useful to realize that Gµρσ is antisymmetric in ﬁrst two indices. ﬁnally note that it is suﬃcient to show the last equation for dσµ M µρσ . Comment: this relation is known as the Pauli equation. 2M 2M with Enr ≡ E − M ≪ M .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 . in order to show that Θρσ − Θσρ = ∂µ H µρσ . 2me Exercise 6.

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

∂ qx ˙ ∂ qx ˙ (7.Quantization of ﬁelds 55 etc. Φ(f ) = d4 x φ(x) f (x). we have L (x) = 1 1 ∂µ φ∂ µ φ − M 2 φ2 2 2 1 1 1 2 (∂0 φ) − (∇φ)2 − M 2 φ2 = 2 2 2 δL = = ∂0 φ. The essential conditions to consistently quantize a theory then are the following. py (t)] = i δxy or for the ﬁelds φ(x) and Π(x) the socalled equal time commutation relations [φ(x. The lagrangian is given by L = = x In order to construct the hamiltonian it is necessary to ﬁnd the canonical momenta. t).12) (7. This implies speciﬁc transformation properties in the Hilbert space for the quantum ﬁelds φ. ∆φ. • Canonical commutation relations The quantization of the theory is achieved by imposing the canonical quantization condition [qx (t).2.13) Π(x) H (x) For the quantization procedure we can formulate a number of basic axioms of quantum ﬁeld theory.9) Note that this indeed corresponds to the zero-zero component (Θ00 ) of the conserved energy-momentum stress tensor Θµν . qx+∆x ) . • Poincar´ invariance e The full variation of the ﬁelds. ∂µ φ(x)) = ˙ d3 x L (φ.11) (7.6) ∆3 x L x (qx . t).15) with furthermore the relations [φ(x. As an example. formally. (7. for the scalar ﬁeld theory. albeit with ’sharp’ functions. discussed in the section 6.16) . t)] = 0. a) = Rj (Λ−1 ) φj (Λ−1 x − a). (7. φ. Π(y.14) In fact. ∇φ) (7.10) x ∆3 x [Πx qx − L x ] ˙ ∆3 x Hx (7. the ﬁelds φ(x) can be considered as regular operators in the Hilbert space after smearing with a test function f . Sometimes it is useful to keep in mind that. t). i U † (Λ. ˙ where Πx (t) is obtained from the continuous ﬁeld Π(x) ≡ δL /δ(∂0 φ). ˙ δ φ(x) (7. δ∂0 φ) 1 1 1 = Θ00 (x) = (∂0 φ)2 + (∇φ)2 + M 2 φ2 .7) d3 x L (φ(x).8) (7. Π(y. a) φi (x)U (Λ. (7. The hamiltonian then is H= x px qx − L = ˙ = x where H (x) = δL ˙ φ(x) − L (x) = Θ00 (x). qx . the discretization procedure above is an explicit example. for symmetry transformations. t)] = i δ 3 (x − y). t)] = [Π(x. px (t) = ∂L ∂Lx = ∆3 x = ∆3 x Πx (t). 2 2 2 (7. φ(y.

and [N. 2 2 (7. • Causality Operators Φ(f ) and Φ(g) for which the test functions are space-like separated can be measured simultaneously (macroscopic causality).23) The hamiltonian in this case can be expressed in the number operator N = a† a. Microscopic causality implies local commutativity. 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.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.20) where the coordinate Q and the momentum P satisfy the canonical commutation relations [Q. a† ] = 1. φ(y)] = 0 if (x − y)2 < 0.Quantization of ﬁelds 56 or if U (Λ.24) It is straightforward to ﬁnd the commutation relations between N and a and a† .22) it is straightforward to check that the commutation relations between Q and P are equivalent with the commutation relations [a. j (7. [φ(x). a] = [a† . 2 i ωµν (S µν )i . Φ(g)] = 0. H = = ω ω P P Q−i √ Q+i√ 2 2 2ω 2ω 1 1 =ω N+ . Q] = [P.18) with S µν given in 6.21) Writing Q and P in terms of creation (a† ) and annihilation (a) operators. The measurements cannot inﬂuence each other or [Φ(f ). ω a† a + 2 2 ω i − [Q. (7. In simpliﬁed form the hamiltonian is given by H= 1 2 1 2 2 P + ω Q .25) .19) 7. a] = −a.17) (Sµν )i φj (x). (7. sometimes referred to as second quantization. Mµν ] = i i∂µ φi (x). Pµ ] = [φ (x). a† ] = a† . (7. P ] 2 (7.4. j 2 one has [φi (x). i(xµ ∂ν − xν ∂µ )φ (x) + i (7. a) = i Rj (Λ−1 ) = 1 + iǫµ P µ − 1− i ωµν M µν . [a. 1 Q = √ (a + a† ) 2ω and P = −i ω (a − a† ) 2 (7. a† ] = 0. P ] = 0 (7. [Q. [N. P ] = i.

e. 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 . i. a† (k ′ )] = (2π)3 2Ek δ 3 (k − k′ ) [a(k).30) [φ(x). Π(x′ )]x0 =x′0 = i δ 3 (x − x′ ) ′ ′ and (7. a(k ′ )] = [a† (k).26) a(k) e−i k·x + a† (k) ei k·x .29) It is straightforward to prove that the equal time commutation relations between φ(x) and Π(x′ ) are equivalent with ’harmonic oscillator . The canonical momentum becomes −i d3 k ˙ Π(x) = φ(x) = a(k) e−i k·x − a† (k) ei k·x . Π(x )]x0 =x′0 = 0. (7.4 for the classical ﬁeld) a(k) a† (k ′ ) = = d3 x ei k·x i ∂0 φ(x). (2π)3 2Ek (7. Note that we will often write a(k) or a† (k).3 The real scalar ﬁeld We have expanded the (classical) ﬁeld in plane wave solutions.28) (7. the real scalar ﬁeld φ(x) becomes d3 k (7.Quantization of ﬁelds 57 Deﬁning states |n as eigenstates of N with eigenvalue n. φ(x )]x0 =x′0 = [Π(x). a† (k ′ )] = 0.33) . † √ a and a act as raising and lowering operators. φ(x) = (2π)3 2Ek where the Fourier coeﬃcients a(k) and a† (k) are now operators.like’ commutation relations between a(k) and a† (k ′ ). (n − 1) a|n † i. From the normalizations one obtains a |n = √ n + 1 |n + 1 and a|n = n |n − 1 .e. 7.g. which we have split into positive and negative energy pieces with (complex) coeﬃcients a(k) and a∗ (k) multiplying them. e. H = = = d3 x 1 1 1 (∂0 φ)2 + (∇φ)2 + M 2 φ2 2 2 2 d3 k Ek † a (k)a(k) + a(k)a† (k) (2π)3 2Ek 2 d3 k Ek N (k) + Evac . The 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. and (7. is equivalent with [a(k). N |n = n|n one ﬁnds N a† |n N a|n = = (n + 1) a† |n .31) The hamiltonian can be rewritten in terms of a number operator N (k) = N (k) = a† (k)a(k). [φ(x).27) 2 (2π)3 It is easy to check that these equations can be inverted (see Exercise 2. d3 x φ(x) i ∂0 e−i k ·x ↔ ′ ↔ (7. which represents the ’number of particles’ with momentum k.32) (7.

= normalized as (a† (k1 ))n1 (a† (k2 ))n2 √ √ . a(k)|0 = 0. which amounts to redeﬁning H as H= d3 x : H (x) := d3 k Ek N (k).40) This procedure is known as normal ordering.36) and satisfying the completeness condition 1 = k|k ′ = (2π)3 2Ek δ 3 (k − k′ ). d3 k a(k) e−i k·x . . i.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.41) (7. (2π)3 2Ek (7.34) where V = (2π)3 δ 3 (0) is the space-volume.37) (7. is solved by subtracting it as an (inﬁnite) constant.42) (7. Just as in the case of the harmonic oscillator it is essential (axiom) that there exists a ground state |0 that is annihilated by a(k). k 0 = Ek = k2 + M 2 ).38) d4 k 2π δ(k 2 − M 2 ) θ(k 0 ) |k k| = (2π)4 (7. φ(x) φ+ (x) φ− (x) = = = φ+ (x) + φ− (x).44) p|φ− (x)|0 = ei p·x . This term will be adressed below. (2π)3 2Ek (7. The 1-particle wave function e−ip·x is obtained via 0|φ(x)|p p|φ(x)|0 = = 0|φ+ (x)|p = d3 k 0|a(k)a† (p)|0 e−i k·x = e−i p·x (2π)3 2Ek (7. assuring that the vacuum (by deﬁnition) has eigenvalue 0!. . . The rest of the states are then obtained from the groundstate via the creation operator.39) The problem with the vacuum or zero-point energy. d3 k |k k|. writing all annihilation operators to the right of the creation operators. and multiparticle states |(k1 )n1 (k2 )n2 . in ﬁeld theory also referred to as vacuum energy Evac = 1 V 2 d3 k Ek . For the momentum operator one has Pi = d3 x Θ0i (x) = 1 2 d3 x ∂ {0 φ ∂ i} φ = d3 k k i N (k). . (2π)3 (7.e. deﬁning particle states |k = |k (with positive energy.45) (7. For the purpose of normal ordering it is convenient to decompose the ﬁeld in positive and negative frequency parts.35) where the vacuum contribution disappears because of rotational symmetry. (2π)3 2Ek (7. (2π)3 2Ek d3 k a† (k) ei k·x . n1 ! n2 ! (7. (2π)3 2Ek (7. |0 . .46) (7.43) The normal ordered product can be expressed as : φ(x)φ(y) := φ+ (x)φ+ (y) + φ− (x)φ+ (y) + φ− (y)φ+ (x) + φ− (x)φ− (y). which now contains an inﬁnite number of oscillators. |k = a† (k)|0 .

(2π)3 2Ek ≡ i∆− (x − y).Quantization of ﬁelds 59 In order to ensure the consistency of the theory it is necessary to check that the operators P ν and M obtained from the conserved currents Θρσ and M µρσ are indeed the generators of the Poincar´ e group. i∆+ (x) i∆− (x) = = = d4 k θ(k 0 ) 2π δ(k 2 − M 2 ) e−i k·x . In order to calculate [φ(x). d3 k ei k·(x−y) = −i∆+ (y − x). x) = 0 and hence ∆(x) = 0 for x2 < 0. (2π)4 (7. 2π (7. a† (k ′ )] 3 2E ′ (2π) k [φ− (x). φ+ (y)] d3 k e−i k·(x−y) .48) (7. The last item to be checked for the scalar ﬁeld are the causality condition.17) and (7. (iv) The equal time commutation relations follow from ∂ ∆(x) ∂t (v) For M = 0.51) (7. (2π)4 d4 k − θ(k 0 ) 2π δ(k 2 − M 2 ) ei k·x (2π)4 d4 k − θ(−k 0 ) 2π δ(k 2 − M 2 ) e−i k·x = −i∆+ (−x) (2π)4 (7. φ(y)] consider µν [φ+ (x).49) (7. (7. φ(y)] which has the properties (i) i∆(x) = i∆+ (x) + i∆− (x) can be expressed as i∆(x) = where ǫ(k 0 ) = θ(k 0 ) − θ(−k 0 ). (ii) ∆(x) is a solution of the homogeneous Klein-Gordon equation.e. (iii) ∆(0.52) The result for the invariant commutator function is [φ(x).18). φ− (y)] ≡ i∆+ (x − y). i. ∆(x) = − ǫ(x0 ) δ(x2 ). (7. that they satisfy the required commutation relations in Eqs (7.53) t=0 .56) = −δ 3 (x). = − (2π)3 2Ek (7.50) or as integrals over d4 k. d3 k = (2π)3 2Ek = (7.54) = i∆(x − y) = i (∆+ (x − y) + ∆− (x − y)) .55) d4 k ǫ(k 0 ) 2π δ(k 2 − M 2 ) e−i k·x .47) ′ d3 k ′ e−i k·x+i k ·y [a(k).

58) and satisfy the equal time commutation relation (only nonzero ones) [φ(x). Q = = = i d3 x : φ† ∂0 φ − ∂0 φ† φ(x) : d3 k : a† (k)a(k) − b(k)b† (k) : (2π)3 2Ek d3 k a† (k)a(k) − b† (k)b(k) . extending it with the charge operator and the introduction of particle and antiparticle operators. 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) .59) (7.66) .63) i. particles (created by a† ) and antiparticles (created by b† ) with the same momentum contribute equally to the energy.60) = i φ∗ ∂µ φ. (2π)3 2Ek (7. (2π)3 2Ek (7. Also the charge operator requires normal ordering (in order to give the vacuum eigenvalue zero). = ∂{µ φ∗ ∂ν} φ − L gµν . (7. The ﬁeld satisﬁes the Klein-Gordon equation and the density current (U (1) transformations) and the energy-momentum tensor are jµ Θµν The quantized ﬁelds are written as φ(x) = φ† (x) = d3 k a(k) e−i k·x + b† (k) ei k·x .Quantization of ﬁelds 60 7. ↔ (7.64) The commutator of φ and φ† is as for the real ﬁeld given by [φ(x). φ† (y)] = i∆(x − y).5 The Dirac ﬁeld L = ↔ i ψγ µ ∂µ ψ − M ψψ.61) (7. we will consider it as a repetition of the quantization procedure. (2π)3 2Ek (7.62) (7.4 The complex scalar ﬁeld In spite of the similarity with the case of the real ﬁeld. 2 From the lagrangian density (7.57) (7. which is equivalent to the relations (only nonzero ones) [a(k). a† (k ′ )] = [b(k). (2π)3 2Ek d3 k b(k) e−i k·x + a† (k) ei k·x . b† (k ′ )] = (2π)3 2Ek δ 3 (k − k′ ).e.65) 7. ∂0 φ† (y)]x0 =y0 = i δ 3 (x − y).

s)ei k·x . d† (k ′ .77) Note that achieving normal ordering. ∂ 2 ↔ (7. (2π)3 2Ek (7. {b(k. s)ei k·x + d(k.Quantization of ﬁelds 61 the conserved density and energy-momentum currents are easily obtained. s′ )} = (2π)3 2Ek δ 3 (k − k′ ) δss′ . s) e−i k·(x−y) j s † + vi (k.79) Also for the ﬁeld and the canonical conjugate momentum anticommutation relations are considered. (2π)3 2Ek d3 k b† (k. The solution is the introduction of anticommutation relations. b† (k ′ .68) The canonical momentum and the hamiltonian are given by Π(x) H (x) = = δL = i ψ † (x). The quantized ﬁelds are written ψ(x) = s d3 k b(k. interchanging creation and annihilation operators. then leads to additional minus signs and H Q = = d3 k Ek b† (k)b(k) + d† (k)d(k) (2π)3 2Ek d3 k b† (k)b(k) − d† (k)d(k) . s)e−i k·x . ↔ i ψγµ ∂ν ψ − 2 (7. . ψj (y)}x0 =y0 = d3 k (2π)3 2Ek ui (k.67) i ψ / ψ − M ψψ gµν . s). u v (2π)3 2Ek (7. s)¯(k. i. s)vj (k.69) (7. (2π)3 2Ek (7. jµ Θµν = = ψγµ ψ. s). † {ψi (x). s) ei k·(x−y) x0 =y 0 . ˙ δ ψ(x) ↔ i Θ00 (x) = − ψγ i ∂i ψ + M ψψ 2 = iψγ 0 ∂0 ψ = iψ † ∂0 ψ. s)e−i k·x + d† (k.76) which seems to cause problems as the antiparticles (d-quanta) contribute negatively to the energy and the charges of particles (b-quanta) and antiparticles (d-quanta) are the same.75) (7.74) (7. s)¯(k. s)v(k. (7. (7. s)u† (k. s)u(k.e. s′ )} = {d(k.73) (7.78) (7.72) ψ(x) = s In terms of the operators for the b and d quanta the hamiltonian and charge operators are (omitting mostly the spin summation in the rest of this section) H = = Q = = d3 x : ψ † (x) i∂0 ψ(x) : d3 k Ek : b† (k)b(k) − d(k)d† (k) : (2π)3 2Ek d3 x : ψ † ψ : d3 k : b† (k)b(k) + d(k)d† (k) : .71) (7.70) where the last line is obtained by using the Dirac equation.

81) where i∆ = i∆+ + i∆− is the same invariant commutator function as encountered before. ˙ δ A0 δL = F i0 = E i .82) where i∆1 = i∆+ −i∆− .Quantization of ﬁelds 62 Using the positive and negative energy projection operators discussed in section 4. ψ j (y)] = (i/x + M )ij i∆1 (x − y).6 The electromagnetic ﬁeld 1 L = − Fµν F µν . 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). ∂ (7.83) From the lagrangian density the canonical momenta are = = (7. it implies the Lorentz constraint and leads to the canonical momenta Π0 Π i = = E i. For the scalar combination {ψi (x). ψ j (y)} at arbitrary times one has {ψi (x). When we would have started with commutation relations for the ﬁeld ψ and the canonical momentum. 7. one has † {ψi (x). ∂ (7. 4 2 This gives the equations of motion discussed before. but has the problem of being noncovariant. we would have obtained [ψi (x).84) (7. 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. 4 Π0 Πi δL = 0.87) If one wants to impose canonical commutation relations ∂µ Aµ = 0 cannot hold as an operator identity.88) . −λ(∂ρ Aρ ).80) 3 = δ 3 (x − y) δij . but we must restrict ourselves to the weaker condition B|∂µ Aµ |A = 0.85) which reﬂects the gauge freedom. It is possible to continue in a covariant way with the lagrangian 1 λ L = − Fµν F µν − (∂µ Aµ )2 . 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).86) (7. (7. (7. as the vanishing of Π0 induces a constraint. ˙ δ Ai (7.

˙ ˙ . 2. a longitudinal photon and two transverse photons. q) that dO/dλ = [O. c† (k ′ . The canonical equal time commutation relations are [Aµ (x).Quantization of ﬁelds 63 for physical states |A and |B .3) satisfy the conditions of a Lie algebra. λ)e−i k·x + c† (k. 6. pj ]P . ′ (7. 0)c(k. 0. and are equivalent with [c(k. The quantized ﬁeld is expanded as Aµ (x) = d3 k (2π)3 2Ek (λ) 3 ǫ(λ) (k) c(k. Πν (y)]x0 =y0 = i gµν δ 3 (x − y).e. i. In fact the commutation relations imply ˙ [Aµ (x). i.11. and [ℓi . λ)ei k·x . where Aµ + + is the part of the vector ﬁeld containing the annihilation operators.1 (optional) If you haven’t seen Poisson brackets in classical mechanics you should (a) Proof that Poisson brackets (Eq. (7. It gives 3 k µ ǫ(λ) (k)c(k. k 3 ) this reads |k 3 |c(0) (k) − k 3 c(3) (k) |A = 0 c(0) (k) ∓ c(3) (k) |A = 0. λ)c(k. λ′ )] = −g λλ 2Ek (2π)3 δ 3 (k − k′ ). (c) Calculate the Poisson brackets for the coordinates x and the standard conserved quantities p and ℓ. µ λ=0 (7. pj ]P .90) where Πν = F ν0 − λ g ν0 (∂ρ Aρ ) and we have furthermore [Aµ (x). which reproduce the Hamilton equations.89) with four independent vectors ǫµ . λ) − c† (k. Aν (y)] = [Πµ (x). λ). These problems are solved by the Lorentz constraint between physical states given above.94) Choosing k µ = (|k 3 |. Q]P . µ λ=0 (7. (b) Proof for a quantity O(p. for which it is suﬃcient that ∂µ Aµ |A = 0.95) Exercises Exercise 7. Πν (y)] = 0. rj ]P . what are [ri .93) which does exhibit problems with the time-like photon. 0. Aν (y)]x0 =y0 = i gµν δ 3 (x − y). 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. containing a time-like photon. (7. (7. (d) Use Poisson brackets to calculate q and p. 7.92) Note that for the transverse states there are no problems with the normalization and the statistics of the states as −g ij = δ ij for i. λ)|A = 0. where Q is the conserved quantity in Eq.91) (7. 0) . j = 1. one time-like photon by itself is not allowed! This solves the problems with the normalization and the negative energies.e. [ℓi .

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

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

˜ Pop d(k. m). λ)v(k. op P (8. λ) eik·˜ (2π)3 2Ek d3 k ˜ ˜ ˜ ˜ ηP b(k.10) i. For the scalar ﬁeld we have seen −1 φ(x) −→ Pop φ(x) Pop = ηP φ(˜).12) and for the vector ﬁeld −1 Aµ (x) −→ Pop Aµ (x) Pop = −Aµ (˜). x (8. −λ)u(k. ˜ −v(k. (i˜ µ ∂µ − M ) ψ(˜) = 0. x µ (iγ ∂µ − M ) γ0 ψ(˜) = 0.5) (8. . −λ). x Therefore γ0 ψ(˜) is again a solution of the Dirac equation and we have x ψ p (x) = γ0 ψ(˜).11) (8. −λ) eik·x (2π)3 2Ek ˜ d3 k ˜ ˜ ˜ ˜ ηP b(k.13) ψ(x) −→ ψ c (x) = ηC C ψ (x).3) (8. γ x (8. x The latter behavior of the vector ﬁeld will be discussed further below. λ)v(k. −λ) e−ik·x − d† (k. 8. (2π)3 2Ek (8. iγ µ† ∂µ − M ψ(˜) = 0.6) (check this for the standard representation.9) (8. the above operation reverses the sign of λ). −λ) eik·x 3 2E (2π) ˜ k d3 k ˜ ˜ ηP b(k. x It is straightforward to apply this to the explicit ﬁeld operator ψ(x) using γ0 u(k. λ) eik·x . λ)γ0 u(k. −λ). λ) P −1 = −η ∗ d(k.7) = λ d k x x ηP b(k. choosing ηP is real (ηP = ±1) particle and antiparticle have opposite parity.4) (8. if helicity λ is used instead of the z-component of the spin m.8) 3 = λ = λ = λ From this one sees immediately that −1 ˜ Pop b(k. λ) e−ik·˜ + d† (k. λ)u(k.e. λ) Pop = ηP b(k. one can also consider the scalar ﬁeld and vector ﬁelds. λ) e−ik·x − d† (k. m) = = ˜ u(k. −λ) e−ik·x − d† (k. −λ)v(k. m) γ0 v(k. m). λ)u(k. λ)γ0 v(k.Discrete symmetries 67 After parity transforming x to x the Dirac equation becomes after some manipulations ˜ ˜ x iγ µ ∂µ − M ψ(˜) = 0.2 Charge conjugation We have already seen the particle-antiparticle symmetry with under what we will call charge conjugation the behavior T (8. The result is ψ p (x) −1 = Pop ψ(x) Pop = ηP γ0 ψ(˜) x (8. In the same way as the Fermion ﬁeld.

As time reversal will transform ’bra’ into ’ket’. to ﬁnd ψ t (−˜) that is a solution of the Dirac equation. For a quantized ﬁeld one has −1 ∗ −1 Top fk (x)bk Top = fk (x)Top bk Top . ˜ x −(iγ µ )∗ ∂µ − M ψ(−˜) = 0. x µ (iγ ∂µ − M ) γ5 Cψ(−˜) = 0. (2π)3 2Ek (8. Top |φ = φt | = (|φt )∗ . i. 8.22) = Aψ|Aφ ∗ is antilinear if A(λ|φ + µ|ψ ) = λ∗ A|φ + µ∗ A|ψ .16) d k ηC d(k. m). λ) eik·x . The same relations hold for helicity states. Therefore ψ c (x) = = λ −1 ¯ Cop ψ(x) Cop = ηC C ψ T (x) (8. we start with the complex conjugated x Dirac equation for ψ. λ). One has Aφ|Aψ = φ|ψ ∗ = ψ|A† Aφ = ψ|φ .3 Time reversal xµ = (t. λ)u(k. . m) ¯ C v T (k. as discussed in section 4). γ x i(γ5 C)−1 γ µ γ5 C∂µ − M ψ(−˜) = 0. ((iγ µ )∗ ∂µ − M ) ψ(x) = 0. m). r) −→ −˜µ ≡ −xµ = (−t.20) The time reversal operator transforms We will again consider the transformation properties for a fermion ﬁeld ψ(x). λ) e−ik·x + b† (k. antilinear operator is anti-unitary if A† = A−1 . it is antilinear1 . x (8. λ)C v T (k. x (8. λ) eik·x ¯ ¯ (2π)3 2Ek d3 k ηC d(k. x iγ µT ∂µ − M ψ(−˜) = 0. m) ¯ = = v(k. Norm conservation requires Top to be anti-unitary2 . The action on the spinors (using C = iγ 2 γ 0 = −iρ1 σ 2 in standard representation) gives C uT (k.e. The (time-reversed) Dirac equation becomes. writing −1 ψ(x) −→ Top ψ(x)Top = ηT Aψ(−˜) ≡ ψ t (−˜). (8. λ) −1 Cop = ∗ ηC b(k.17) 3 = λ This shows that −1 Cop b(k. λ) Cop = ηC d(k. (8. x 1A 2 An −iC −1 γ µ C∂µ − M ψ(−˜) = 0. λ)v(k. u(k. λ) e−ik·x + b† (k. (i˜ µ∗ ∂µ − M ) ψ(−˜) = 0.21) where A is a 4 × 4 matrix acting in the spinor space.18) (8. λ). r). x x (8.Discrete symmetries 68 the latter being also a solution of the Dirac equation.14) (8.15) (where one must be aware of the choice of spinors made in the expansion. λ)C uT (k.19) Cop d(k.

˜ = 0|b(k)|A 8.28) (8. λ| ¯ C |¯. (8. λ).23) In the standard representation iγ5 C = σ2 and it is straightforward to apply this to the explicit ﬁeld operator ψ(x) using i γ5 C u(k. λ)iγ5 Cu(k. −p. p.33) (8. λ |¯. p. λ) Top = ηT b(k. −p. λ)u∗ (k. −λ |¯.27) 3 = λ = λ = λ From this one obtains −1 ˜ Top b(k. λ a P |a. λ) eik·x + d† (k. λ)u∗ (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) (8.2: The transformation properties of physical states for particles (a) and antiparticles (¯). −p. λ) ψ t (x) −1 Top ψ(x) Top = i ηT γ5 Cψ(−˜) x = = ˜ u∗ (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(−˜).4 Bi-linear combinations In quantities such as currents and lagrangians often bilinear combinations of spinor ﬁelds are encountered. λ) e−ik·˜ (2π)3 2Ek d3 k ˜ ˜ ˜ ˜ ηT b(k. λ)v ∗ (k. −λ a T a. ˜ Top d(k. Therefore one has A|k = A|b† (k)|0 = A|T † T b† (k) T † T |0 ∗ = At |b† (k)|0 ∗ ˜ t = k|At . λ) e−ik·x (2π)3 2Ek (8. λ) T −1 = η ∗ d(k. λ)v ∗ (k. ∗ ˜ v (k. λ) e−ik·x 3 2E (2π) k d3 k ˜ ˜ ηT b(k. λ) e−ik·x 3 2E (2π) k ˜ d3 k ˜ ˜ ˜ ˜ ηT b(k.25) (check this for the standard representation). λ) eik·x + d† (k. The result is = = λ (8. λ)v ∗ (k. Since there are 16 independent 4 × 4 matrices. λ| a. λ) eik·˜ + d† (k.34) .26) d k x x ηT b(k. p. op T (8. λ)iγ5 Cv(k.32) (8. −p. x (8. a state |a. Note that applying an anti-unitary transformation such as Top one must take for the matrix element ˜ the complex conjugate.30) (8.31) (8.24) (8. p. there are 16 independent of these bilinear combinations. λ). λ)u∗ (k.Discrete symmetries 69 Table 8. λ) eik·x + d† (k. λ a |a. λ). λ) i γ5 C v(k.29) In table 2 the behavior of particle states under the various transformations has been summarized. λ).

(q 2 ). for the vector current V µ (x).Discrete symmetries 70 Table 8. s .40) where the coeﬃcients are functions of invariants constructed out of the momenta. 2M (8. F. (q 2 ). As the coupling of the photon ﬁeld to fermions is given by an interaction term in the ¯ lagrangian of the form : ψ(x)γ µ ψ(x) : Aµ (x) and behaves as a scalar one sees immediately that the ¯ photon ﬁeld Aµ (x) behaves in the same way as the vector combination ψ(x)γ µ ψ(x). p′ .37) (8. p)u(p) e−i (p−p )·x .39) where Γµ (p′ . in this case only q 2 = (p′ − p)2 . C. The 16 combinations of Dirac matrices appearing above are linearly independent. Note that the lagrangian density L (x) → L (−x) under Θ.35) The x-dependence can be accounted for straightforwardly using translation invariance..5 Form factors Currents play an important role in ﬁeld theory.41) 2M 2M . e. For instance for nucleons one has Γµ (p′ . Applying the results from the previous sections it is straightforward to determine the behavior of the combinations under P. ¯ In general the expectation value between momentum states can be more complicated. p′ |V µ (0)|p = u(p′ )γ µ u(p). p) = γµ F1 (q 2 ) + iσµν q ν F2 (q 2 ). σ µν . C. as well as under the combined operation Θ = PCT (see Table 3). ¯ V µ (x) =: ψ(x)γ µ ψ(x) :. γ µ . p) can be built from any combination of Dirac matrices (1. ¯ ′ ′ (8. and T. (8.3: The behavior of the independent bilinear combinations of fermi ﬁelds under P. γ5 γ µ or γ5 ).. F. 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 ). called form factors. 8. This implies p′ |V µ (x)|p = e−i (p−p )·x p′ |V µ (0)|p .36) (8. (8. p′ |V µ (x)|p = u(p′ )Γµ (p′ . In many applications the expectation values of currents are needed. of which the expectation value between momentum eigenstates can be simply found. momenta (pµ or p′µ ) or constant tensors (ǫµνρσ or gµν ).38) (8.g. As an example consider the vector current for a point fermion. 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 . γ ν ]. V µ (x) = ei Pop ·x V µ (0)e−i Pop ·x . s′ |V µ (x)|p.

p)u(p) ¯ = p|V µ (0)|p′ ∗ = (¯(p)Γµ (p. p) that γ0 Γµ† (p. e. • Parity p′ |V µ (x)|p = ei q·x u(p′ )Γµ (p′ . Therefore hermiticity implies for the structure of Γµ (p′ .42) where q = p′ − p. p) = γ0 Γµ (˜′ . p)γ0 .g.44) = p′ |Vµ (−˜)|˜ ˜ x p ∗ ˜x = ei q ·˜ [¯(˜′ )]∗ Γ∗ (˜′ .1 The matrix element of the electromagnetic current between nucleon states is written as < p′ |Jµ (x)|p >= e−i(p−p )·x u(p′ )Γµ u(p) ¯ Here Γµ iσµν q ν F2 (q 2 ) 2M pµ + pµ ′ = γµ H1 (q 2 ) − H2 (q 2 ) 2M = γµ F1 (q 2 ) + ′ . p)γ0 u(p). γ5 σµν = (1/2)ǫµνρσ γ ρ γ σ . p ˜ • Time reversal p′ |V µ (x)|p = ei q·x u(p′ )Γµ (p′ . p′ )γ0 = Γµ (p′ . p)(iγ5 C)u(p).g. ¯ µ p ˜ Therefore time reversal invariance implies for the structure of Γµ (p′ . µ p ˜ (8. p′ )u(p′ )) u = ∗ † u† (p)γ0 Γµ (p. p)u(p) ¯ † † = p′ |Pop Pop V µ (x)Pop Pop |p ˜x = p′ |Vµ (˜)|˜ = ei q ·˜ u(˜′ )Γµ (˜′ . p)u(p) ¯ † † = p′ |Top Top V µ (x)Top Top |p ∗ (8. p) that Γµ (p′ . p)u∗ (˜) up p µ p ˜ = ei q·x u(p′ ) (iγ5 C)Γ∗ (˜′ . or relations that follow from the equations of motion. p)u(˜) ˜ x p ¯ p p ˜ p (8. p)(i γ5 C).43) = ei q·x u(p′ )γ0 Γµ (˜′ . ¯ p ˜ Therefore parity invariance implies for the structure of Γµ (p′ . p) = (i γ5 C)Γ∗ (˜′ . p′ )u(p′ ) = u† (p′ )Γµ† (p. C and T invariance. p). p) that Γµ (p′ . the Gordon decomposition u(p′)γ µ u(p) = ¯ 1 u(p′ ) [(p′ + p)µ + iσ µν qν ] u(p). ¯ 2M (8.45) Exercises Exercise 8. • Hermiticity p′ |V µ (0)|p = p′ |V µ† (0)|p = u(p′ )Γµ (p′ .Discrete symmetries 71 but that are eliminated because of relations between Dirac matrices. or relations based on hermiticity of the operator or P . e. p′ )γ0 u(p).

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

t0 ).Chapter 9 Path integrals and quantum mechanics 9. AH (t) = U −1 (t. i ∂ |ψH ∂t ∂ i AH ∂t ≡ = 0.1 Time evolution as path integral U (t. t′ . t0 )|ψS (t0 ) . i ∂ |ψS ∂t ∂ i AS ∂t = H |ψS . t ′ QH (t′ )|q ′ . (9. ≡ 0. and the operators are time-dependent. t0 ) AH (t0 ) U (t. t . t0 ) ∂t (9.1) The time evolution from t0 → t of a quantum mechanical system is generated by the Hamiltonian.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.2) (ii) Heisenberg picture. (9. |ψS (t) = U (t. [AH . Consider the two (time-independent) Heisenberg states: |q. in which the states are time-independent. |ψH (t) = |ψH . (9. in which the operators are time-independent. QH (t)|q. t0 ) = e−i(t−t0 )H or i Two situations can be distinguished: (i) Schr¨dinger picture. t ′ |q . t ≡ q|q.5) (9. t′ ≡ q ′ |q ′ . (9. AS (t) = AS and the states o are time dependent.4) ∂ U (t. H].7) 73 .3) (9. t0 ) = H U (t.

. H(P. . t = = q ′ |e−i n H∆τ |q = q ′ | e−i H∆τ dq1 . t and QH (t) = QS and study the evolution of the system by calculating the quantum mechanical overlap amplitude q ′ .16) this. |q1 q1 |e−i H∆τ |q . qj ). 2π (9.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 .Path integrals and quantum mechanics 74 and Schr¨dinger states o |q |q ′ QS |q ≡ q|q . B]. (9.12) where the transformation between coordinate and momentum space involves q|p = ei p. ˙ 2π 1 1 1 [A.13) At least for a simple hamiltonian such as consisting of a kinetic energy term and a local potential.14) . [A. B] + .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 .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 ′ . ′ (9. 2π (9. B]] + [[A. The purpose of this is to calculate the evolution for an inﬁnitesimal time interval. Combining the two terms gives qj+1 . 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 .qj ) e 2π dpj exp (i ∆τ [pj qj − H(pj . 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|. Then qj+1 .11) (9. t′ |q. which is a hamiltonian function in which the operators are replaced by real-numbered variables. t′ |q. tj+1 |qj .qj )∆τ e e 2π dpj i pj (qj+1 −qj )−i ∆τ H(pj . eA eB = eC with C = A+B+ (9. tj = = = 1 For dpj i pj (qj+1 −qj ) −i H(pj . 2 12 12 (9. tj+1 |qj . (9. The hamiltonian is an operator H = H(P. qj )]) . QS |q ′ ≡ q ′ |q ′ . use the Campbell-Baker-Hausdorﬀ formula. B] + [A.8) Choose t as the starting point with |q = |q. t = q ′ |e−i H(t −t) |q . Q) expressed in terms of the operators P and Q. . . n |q (9. with the correction1 being of order (∆τ )2 . .10) dqn−1 q ′ |e−i H∆τ |qn−1 qn−1 | .

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

A useful property of functional integration is the translation invariance. Consider the gaussian functional as an example. 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.n fm (x) Kmn fn (y) and consider F [α] = F (αn ) as a multivariable function.28) Dα Dα∗ exp (−α∗ α − α∗ ω − ω ∗ α) = exp (ω ∗ ω) .22) = x The last equality is obtained by deﬁning the right measure (normalization N ) in the integration.n dx fm (x)fn (x) = n = n dαn 1 exp − α2 N 2 n √ 2π ≡ 1. with Dα F [α] = n dαn N F (αn ) (9. (9.23) Having deﬁned the Gaussian integral.24) (9. 1 = √ det K 1 Dα Dα∗ exp (−α∗ K α) = det K 1 Dα exp − α K α 2 (α real-valued) (α complex-valued) (9. Note that procedure (i) is an example of the more general procedure under (ii).21) again a multidimensional integral. Dα F [α] = Dα F [α + β].Path integrals and quantum mechanics 76 (ii) Write α in terms of a sum of orthonormal basis functions. 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 (9.27) (9. the following integrals can be derived for a real symmetric or complex hermitean kernel K. 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. α(x) = n αn fn and K(x. (9. In both cases N is an appropriately choosen normalization constant in such a way that the integral is ﬁnite. . y) = ∗ m.26) As an important application of translation invariance we mention the identities 1 Dα exp − α2 − ωα 2 = exp 1 2 ω . 2 (9.

θη = −ηθ.e. θ∗ ] = exp (−θ∗ K θ) ≡ exp − Integration for Grassmann variables is deﬁned as dθ 1 = 0 and dθ θ = 1. n θn fn (x) dx θ∗ (x) θ(x) . αy = δxy .38) . which vanish. θ∗ ] = exp (−θ∗ θ) ≡ exp − F [θ.34) (9. 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).36) where the coeﬃ- This gives for the Gaussian functional integral after expanding θ(x) = cients θn are Grassmann variables the result Dθ∗ Dθ exp (−θ∗ θ) = n ∗ dθn dθn ∗ dθn n exp − ∗ θn θn n = = n ∗ dθn exp (−θn θn ) ∗ ∗ dθn dθn (1 − θn θn ) = 1. i.29) (9. F [θ. the Gaussian-type functionals. δα(x) 2 2 δ exp (−α β) = −β(x) exp (−α β) .35) For applications to fermion ﬁelds. dx dy θ∗ (x) K(x. however.33) (9. δα(x) δα(x) δ αKβ = dy K(x. θ2 = 0. y) θ(y) .31) δ δ α = dy α(y) = dy δ(x − y) = 1.Path integrals and quantum mechanics 77 Functional diﬀerentiation is deﬁned as δ .32) (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. (9. α(y) = δ(x − y). e. (9. In the expansion of the exponential (from second to third line). αn = δmn . It is now easy to check that Dθ∗ Dθ exp (−θ∗ K θ) = det K.30) (9. −→ δα(x) ∆x ∂αx ∂αx δ ∂ ∂ ∗ with . we need to consider Grassmann-valued functions involving anticommuting Grassmann variables. In principle the deﬁnitions of functionals is the same. products of the same pair appear. δα(x) (9. (9. δα(x) or in discretized form δ 1 ∂ ∂ with .g. Examples of functional diﬀerentiation are (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

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

in which case eiEn T → e−En ·∞ and the term eiE0 T is dominant. T → i ∞ and T ′ → −i ∞.67) = = φ0 (q. T = φ0 (q. t′ |q. (9. dq dτ + J(τ )q(τ ) . such that t′ → −i∞ t → i∞ ↔ ↔ ˜ t → −∞. . t) (9. Better is the use ˜ of imaginary time t = −it. ˜ t′ → ∞. t′ ) q ′ . . Q(tn )|0 . thus T →i∞ lim q. . T ′ |Q. t). t|Q. The importance of Z[J] is that the time ordered product of operators can be obtained from it. . eiE0 T φ∗ (Q) 0 We deﬁne the generating functional Z[J] as Z[J] = Q′ . t 0 0out |0in J . = (i)n 0|T Q(t1 ) . J=0 (9.69) The factor that has been divided out in the ﬁrst line of this equation is a numerical factor depending on the boundaries of the space-time volume (T and T ′ ).2: Analytic continuation of boundaries T and T ′ We can project out the groundstate by an analytic continuation in the time.70) one immediately ﬁnds δ n Z[J] δJ(t1 ) . δJ(tn ) hence the name generating functional. 9. Since we have (neglecting multiplicative factors). The generating functional precisely represents the vacuum to vacuum amplitude from initial (’in’) to ﬁnal(’out’) situation in the presence of a source. .Path integrals and quantum mechanics t 82 T t t’ T’ Figure 9.71) 9.6 Euclidean formulation The above expression (Eq. T′ Z[J] = lim T ′ → −i∞ T → i∞ Dq exp i T dτ L q. J (9.70) for the generating functional is in fact ill-deﬁned.68) (9. T J lim −i E0 (T ′ −T ) φ (Q′ ) φ∗ (Q) 0 0 T ′ → −i∞ e T → i∞ dq ′ dq φ∗ (q ′ .

dq dt = 1 2 1 2 dq dt 2 Dq exp Dq exp − ∞ d˜ L q.and Grassmann-valued functions. ZE [J] = = where LE q. det K Exercise 9. ˜ dt (9. δ n Z[J] 1 Z[J] δJ(t1 ) . det K 1 . i τ −∞ ∞ −∞ dq d˜ τ + J(˜)q(˜) τ τ dq d˜ τ − J(˜)q(˜) τ τ .g. δJ(tn ) J = 0 ˜ t = it (9. . 1 = √ . for Grassmann valued functions. (9.75) = − LE q. complex.74) − V (q) 2 (9. ensuring convergence for the functional integral ZE [J].73) d˜ LE q. τ dq ˜ dt ≡ −L q. dq ˜ dt = 1 2 dq ˜ dt dq ˜ dt 2 − V (q). e.1 Convince yourself by choosing an appropriate (orthonormal) set of functions that 1 Dα exp − α K α 2 for real-valued functions. y) = ∗ m. i dq . Exercises Exercise 9. . . In all of these cases the determinant of the (hermitean) operator K is deﬁned as det K ≡ p Kp .72) (9. . + V (q).77) where the expressions have been divided by Z[J] and ZE [J] in order to get rid of dependence on multiplicative factors. Dα Dα∗ exp (−α∗ K α) = for complex-valued functions and Dθ∗ Dθ exp (−θ∗ K θ) = det K. (9.76) which is a positive deﬁnite quantity. δJ(tn ) = (i)n J=0 1 δ n ZE [J] . The diﬀerentiations in Z[J] and ZE [J] are related.n fm (x) Kmn fn (y). As discussed in the previous section the interesting quantities are obtained from functional diﬀerentiation with respect to the sources. when L q. ˜ ˜ ZE [J] δJ(t1 ) .2 Check the examples of functional derivation for real-. . where Kp are the eigenvalues of the matrix Kmn in K(x.Path integrals and quantum mechanics 83 In terms of the imaginary time we can write the Euclidean generating functional.

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

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

(10. . = 0 2 2 2 k2 − M 2 (k ) − E 2 (k ) − k − M (10.9) i.11) Depending on the path choosen in the complex k 0 plane (see exercise 7. . Furthermore. . ∆F (the socalled Feynman propagator) is the Green’s function of the Klein-Gordon equation.e. . xn ) = x2 xn . These Green functions appear in the expansion Z[J] = n in n! d4 x1 . . 2 2 → (10.7) where we made the choice Z0 [0] = 1 and ∂µ ∂ µ + M 2 ∆F (x − y) = −δ 4 (x − y). consider the general solution of ∂µ ∂ µ + M 2 ∆(x) = −δ 4 (x). . .13) .3) one ﬁnds 1 Z0 [J] = exp − i 2 d4 x d4 y J(x) ∆F (x − y) J(y) ≡ exp (i W0 [J]) .Feynman diagrams for scattering amplitudes 86 for which we introduce the picture x1 G(n) (x1 .6) the generating functional is (using ∂µ = − ∂µ ) given by Z0 [J] = Dφ exp −i where Kxy = (∂µ ∂ µ + M 2 ) δ 4 (x − y). d4 xn G(n) (x1 . (2π)4 (k 0 + iǫ)2 − E 2 (10.12) d4 k −i k·x ˜ e ∆(k) (2π)4 (10. . . with Z0 [0] = 1. (10.10) J=0 In order to determine ∆F . (10. .8) 1 d4 x φ (∂µ ∂ µ + M 2 ) φ − Jφ 2 K (10. 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. which can be written as ∆(x) = ˜ with (k 2 − M 2 )∆(k) = 1 or ˜ ∆(k) = 1 1 1 = 0 2 . . xn ) J(x1 ) . As discussed in the previous section (exercise 9.4) one distinguishes the retarded Green’s function ∆R (x) = d4 k e−i k·x . J(xn ). .5) Z0 [J] for the free scalar ﬁeld For the (free) scalar ﬁeld lagrangian L = ← 1 1 ∂µ φ∂ µ φ − M 2 φ2 .

∆(x) = − d4 k e−i k·x . ∆F = ∆C .4). the second is to explicitly consider 0|T φ(x1 ) φ(x2 )|0 .22) (10.Feynman diagrams for scattering amplitudes 87 satisfying ∆R (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 where the latter contour can be deformed to the contour for ∆C . t) = 0 for t < 0.16) C where C is a closed contour in the complex k 0 -plane. φ(0)] = i∆+ (x) + i∆− (x) d3 k e−i k·x = [φ+ (x). in the (well-deﬁned) Euclidean formulation starting with LE = 2 (∂E φ)2 + 2 M 2 φ2 µ 2 2 Jφ with the equation of motion (∂E ) − M φ(x) = −J(x). 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. φ− (0)] = (2π)3 2E d3 k e−i k·x = [φ− (x).23) k0 =i∞ k0 =−i∞ d4 k e−i k·x .14) They are solutions of the inhomogeneous equation.18) Thus we see various solutions. t) = 0 for t > 0. The ﬁrst possibility is to consider the well-deﬁned Euclidean formulation. i.e. (2π)4 k 2 − M 2 (10.1 for countours) µ (∂E )2 − M 2 (i ∆F (x − y)) = −δ 4 (x − y) 0 kE =∞ (10.15) e−i k·x d4 k . .21) ∆F (x) = = −i 0 kE =−∞ d4 kE e−i kE ·xE 2 (2π)4 kE + M 2 (10. (2π)4 (k 0 − iǫ)2 − E 2 (10. (2π)4 k 2 − M 2 (10.19) φ ∂E ∂E − M 2 φ + Jφ 2 1 d4 xE d4 yE J(x) (−i ∆F (x − y)) J(y) . the advanced Green’s function ∆A (x) = satisfying ∆A (x. the causal Green’s function ∆C (x) = and ﬁnally the Green’s function d4 k e−i k·x = 4 (k 0 )2 − E 2 + iǫ (2π) ∆(x) = PV d4 k e−i k·x . (10. µ 1 1 • Firstly.20) = exp − 2 where or (see ﬁg. the generating functional can be written 1 µ µ Z0 [J] = Dφ exp d4 xE (10. (10. 4 k 2 − M 2 + iǫ (2π) (10. 10. The contours for i ∆(x) i ∆+ (x) i ∆− (x) = [φ(x). φ+ (0)] = (2π)3 2E are also shown in exercise (7.17) d4 k e−i k·x . The solution of the homogeneous equation ∂µ ∂ µ + M 2 ∆(x) = 0 can also be written as an integral in k-space.

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

x4 ) = To ﬁrst order in g one has Z (g ) [J] = = 1 0 0 x1 + + x2 (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 . L (φ) = L0 (φ) + LI (φ). To zeroth order in the coupling one has 0 1 Z (g ) [J] = Z0 [J] = exp − i 2 J ∆F J . x3 . (10.Feynman diagrams for scattering amplitudes 89 10. (10.g 0 ) = G0 = 1.27) When interactions are present. x2 ) = G0 (x1 . (2) (0) (10.31) G(2. (10.30) and in zeroth order G(0. Z[J] = = Dφ exp i exp i d4 z LI (10.34) Introducing a vertex point to which four propagators are connected. x4 ) = G0 (x1 . ≡ −i g d4 z (10. x2 .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. x2 ) = i ∆F (x1 − x2 ) = G(4.32) (10. x3 .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. the generating functional can be written. with propagators (i ∆F ) being connected to vertices (i LI ) according to the above expression for the functional Z[J]. i.29) This expression will be the one from which Feynman rules will be derived. x2 . Consider as an example the interaction LI (φ) = − g 4 φ 4! in the scalar ﬁeld theory discussed sofar.g ) (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 .

. .36) J J from which one obtains G(0. . i.39) We have calculated the function up to ﬁrst order in the coupling constant for φ4 -theroy. . . . . This remains also true for the interacting case. . . J(xn ) c (10. (10.40) in−1 n! d4 x1 . . x2 ) = 1 z x1 x2 + 1 2 x1 z x2 . G(n) (x1 . δJ(xn ) . (10. . . . 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]. . . (10. .42) x1 G(n) (x1 . which also was the only Green’s function for which the diagram was connected. J(xn ). . . J=0 (10.g ) (x1 . xn ) J(x1 ) . xn ) J(x1 ) . d4 xn G(n) (x1 .38) The result for the 4-points Green’s function is left as an exercise. 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].g 1 ) = 1 8 z 1 8 . xn ) = (−i)n−1 c which are denoted δ n W [J] δJ(x1 ) . Z[J] = n in n! d4 x1 . . .41) in terms of socalled connected Green’s functions. Diagrammatically this exponent only contained the two-point function. . (10. . did not contain parts that could be written as products of simpler Green’s functions. . x2 ) = c (vacuum bubble) (tadpole) (connected propagator) . (10. In the free case we have seen that the exponent of Z[J] contained all essential information. . xn ) = x2 c xn Speciﬁc examples are G(0) = c G(1) (x1 ) = c G(2) (x1 . .Feynman diagrams for scattering amplitudes 90 one has J Z (g1 ) [J] = 1 8 1 − 4 J J 1 + 24 J Z0 [J].e. d4 xn G(n) (x1 .37) G(2. .

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

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

as it stands.49) The ﬁeld φ can be expressed in φin/out using retarded or advanced Green’s functions.62) is then T′ UT [J] T′ = T exp i T d4 x J(x)φ(x) . S is invariant under e Poincar´ transformations: U (Λ. this can actually not be used as it would imply [φ(x). 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) (10.52) where Z is a constant. (2π)3 2E (10. φ(y)] = Z [φin (x).51) (10. φin (y)] = iZ ∆(x − y).54) . z) φ(z) d4 y d4 z ∆A (x − y) K(y. The relation between S and Z[J] To establish this relation. a)SU −1 (Λ. in|φin S → φin S = S φout (x). 2 ∂µ ∂ µ + M0 φ(x) = J(x). a) = S. (10. out|φout = β. the source term Jφ in the lagrangian density is treated in the interaction picture. 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. (10. β. a causality condition that can be proven to imply the absence of interactions. φin (x) = d3 k ∗ a(k) fk (x) + a† (k) fk (x) . 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. z) φ(z).53) α|φf (t)β −→ Z α|φf (t)β in for normalizable states |α and |β and φf (t) ≡ d3 (x) f ∗ (x) i ∂ 0 φ(x) with f a normalizable solution of the Klein-Gordon equation (wave packet). (10. The convergence therefore must be weakened to t→−∞ √ (10. √ Although the above.e. which in a later stage (renormalization) will become more important.50) = φ(x) = √ Zφin (x) − √ Zφout (x) − d4 y d4 z ∆R (x − y) K(y. Proof: β. implies the strong (operator) convergence φ(x) → Zφin (x). 9.48) 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. The time evolution operator (see Eq. The relation between S-matrix and in. i.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 ).g. e. in|Sφout Finally we check that as expected S does not spoil Poincar´ invariance.and out-ﬁelds is: φin (x) = S φout (x)S −1 .

z) δU [J] δJ(z) δU [J] d4 y d4 z ∆A (x − y) K(y.62) where F [J] is some (arbitrary) functional.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). z) δU [J] . Noting that 0| : eA : |0 = 1 it follows that F [J] = 0|S U [J]|0 = 0|U [J]|0 = 0out |0in ∝ Z[J]. Thus applied to the ﬁeld φ(x) = φ+ (x) + φ− (x). S U [J]] = √ Z δ S U [J]. y) δ δJ(y) : F [J]. provided [A. φ− f (10.Feynman diagrams for scattering amplitudes 94 and satisﬁes the property T ′ δUT [J] t = i UtT [J] φ(x)UT [J]. δJ(z) (10.63) . eB ] = [A. B] is a c-number). δJ(x) ∞ or for U [J] ≡ U−∞ [J] ′ (10. we can. t i Ut∞ [J] φ(x)U−∞ [J] (10. φ(x). which is equal to the expression e in which the creation operators are placed left of annihilation operators.60) e d4 y φ+ (y) f (y) d4 z φ− (z) f (z) d4 z φ+ (z) f (z) d4 y [φ(x). (10. : e d4 z φ(z) f (z) d4 x φ− (x) f (x) B C B C (10. B] and [A. δJ(z) (10. (10. φ(y)] f (y) e e . one has [A.59) (10.e. just as we did for φ(x). B] (for the case that [A. φ(x). [A. δJ(z) Taking the diﬀerence between the two expressions √ Z (φout (x) U [J] − U [J] φin (x)) =i or d4 y d4 z (∆R (x − y) − ∆A (x − y)) K(y.56) Since δU [J]/δJ(x) satisﬁes the same equations as φ(x) and we know the limits. e = i. C] are c-numbers. : =i d4 y ∆(x − y) f (y) : e d4 z φ(z) f (z) :. B]eB .61) e φ+ f where the normal ordered expression : e φ f : is used. Thus S U [J] =: exp 1 √ Z d4 x d4 y φin (x) K(x. z) d4 y d4 z ∆(x − y)K(y. z) . with the use of the Baker-Campbell-Hausdorﬀ formula e−B A e+B = A + [A. 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. B + C]e e .58) In order to ﬁnd a solution to this equation note that.57) i [φin (x). e e ] = [A. express it in terms of advanced and retarded Green’s functions.

.68) 10. (λ) v(p). Sf i = δf i − i (2π)4 δ 4 (Pi − Pf ) Mf i . . ←− − → × G(n) (x′ . . . . . pn ) = i=1 d4 xi ei pi ·xi G(n) (x1 . (2π)3 2E n (10.e. n (2π)4 δ 4 (p1 + . . . . see exercise 10. (10. which is determined by the inverse of the operator found in the quadratic term in the lagrangian i. fp (x) . pn ). Next we introduce the Fourier transform (after extracting a momentum conserving delta function coming from translation invariance. c (n) (10. i.. . .65) we get the Lehman-Symanzik-Zimmerman (LSZ) reduction formula.4 Feynman rules The real scalar ﬁeld The procedure to obtain the matrix element is commonly summarized by a set of rules known as Feynman rules. S =: exp 1 √ Z d4 x d4 y φin (x) K(x. . . . Note that the Green’s function connecting the external point xi with the bubble is annihilated by K(x. . p′ . . iKx′ .2). xn ). ¯ ¯ and the amputated or one-particle irreducible (1PI) Green’s functions n −i (n) Γ(n) (p1 . Explicitly. fp′ (x′ ) . . u(p). . . i. . . |S|p. 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. 1 for scalar case. . .. .+ 1 √ Z ∗ d4 x′ . . k = k2 i − M 2 + iǫ . .64) Therefore.66) where Gc is the connected Green’s function.δ. . . . . y) δ δJ(y) : Z[J] Z[0] . . x.e. using that φin (x) = d3 k ∗ a(k) fk (x) + a† (k) fk (x) . = . pn ) = G (p1 . .. . d4 x . They start with the propagator (i∆F (k)) in momentum space. . . an S-matrix element between momentum eigenstates in initial and ﬁnal states is found by considering those source-connected Green’s functions (action of δ/δJ on Z[J]/Z[0]) where the external sources J(xi ) are replaced by the particle wave functions (which are the result of acting with φin (xi ) on momentum eigenstates). .e.Feynman diagrams for scattering amplitudes 95 while for J = 0 one has U [0] = 1.67) where ∆(p) is the Fourier transform of the (full) propagator ∆F (x). iKx′ = (2x′ + M 2 ) precisely annihilating an external propagator i∆F in the Green’s function.) iKx . . . . . y). J=0 (10.. + pn ) G(n) (p1 . . ∆(pj ) j=1 (10. . which is obtained considering only connected diagrams. Usually we are interested in the part of the S-matrix describing the scattering. . u(p) and v (p) for fermions and ǫµ (p) for vector ﬁelds). .

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. which cancels the factor 1/2 in the quadratic piece. = −i g . or calculating full Green’s functions external lines are treated as propagators. ’t Hooft and M. = −i g (multiplied with 4! corresponding to the allowed number of permutations of identical particles).27. vertices and external particle wave functions in that diagram. the factor 2 corresponds to the two-points Green’s function having two identical ends). Veltman. to be precise iLI vertices in momentum space are introduced. 3! 4! The vertices are: = −i f Consider ﬁrst the lowest order self-energy diagram.27. (Rule 3) Carry out the integration over all internal momenta (keeping track of momentum conservation at all vertices!) (Rule 4) Add a symmetry factor 1/S corresponding to permutation of internal lines and vertices (keeping external lines ﬁxed). For the symmetry factor consider the examples (given in G.69) LI (φ) = − φ3 − φ4 .66 and 10. but overall momentum conservation at a vertex is understood. If problems arise go back to the deﬁning expression for the generating function in 10. For the interaction terms in the lagrangian.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. Note that in calculating amputated Green’s functions external lines are neglected. Diagrammar) in the case of the interaction terms g f (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. 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. At these vertices each line can be assigned a momentum.

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

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

As an example some scattering processes in quantum electrodynamics will be discussed in the next section after the introduction of the Feynman rules for vector ﬁelds. 1 k 2 − M 2 + iǫ λ (k 2 − M 2 + iǫ)(k 2 − λ M 2 + iǫ) leads to . 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 . 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. The wave functions are given by µ k ǫµ (k) k µ incoming photon outgoing photon ǫ∗ (k) µ The vertex for the coupling of photon to the electron is given by µ = i j −i e(γµ )ji . Particular choices of λ are λ = 1 (Feynman propagator or Feynman gauge) and λ = ∞ (Landau gauge). λ 1 ∂ (10. 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.Feynman diagrams for scattering amplitudes 99 The result is which contains an extra minus sign as compared to a bosonic loop. The wave functions for fermions are given by ∝ g 2 S(z − z ′ ) S(z ′ − z) = −g 2 iS(z − z ′ ) iS(z ′ − z). Vector ﬁelds and Quantum Electrodynamics As the most important example of vector ﬁelds consider the lagrangian density for quantum electrodynamics (QED). + iǫ)2 where the kµ kν terms in the case where the photon couples to a conserved current (such as ψγ µ ψ) will not contribute. λ (∂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 .74) L = − Fµν F µν − (∂ · A)2 + ψ(i/ − M )ψ − e ψγ µ ψ Aµ .

t2 (10. s1 ) u e2 u(p′ . 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 . ¯ q2 (10.79) .s2 . s′ )γµ u(k. To lowest order (∝ α = e2 /4π) only one diagram contributes.s4 ) and averaged over spins in the initial state (1/2 × 1/2 × s1 . s3 )(−ie)γ µ u(k.Feynman diagrams for scattering amplitudes 100 10.s3 . s′ ) ¯ u s.s2 ) which can be written as (see also chapter 4) |M |2 = = 1 4s |¯(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′ ′ ′ = 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 ) (10.75) Note that the q µ q ν term in the photon propagator are irrelevant because the photon couples to a conserved current.76) 1 .78) = 2 α2 2 s + u2 + 4t(M 2 + m2 ) − 2(M 2 + m2 )2 . 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 ′ . s2 )|2 ¯ q2 (10. s1 ) ¯ −igµν u(p′ . q 4 µν [using that (¯(k ′ )γµ u(k))∗ = u(k)γµ u(k ′ )].s4 e4 (m) µν (M) L L .5 Some examples eµ scattering The ﬁrst example is the electromagnetic scattering of an electron and a muon. s)¯(k. s4 )γµ u(p. s2 ). s4 )(−ie)γ ν u(p. where u ¯ L(m) µν = = 1 2 u(k ′ . s)γν u(k ′ . s3 )γ µ u(k.

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

.3 reproduce for the diagrams precisely the combinatorial factors that appear in Eqs 10. + pn ) G(n) (p1 . .38. Pictorially this implies. in the amplitude (Tij = A∗ Aj ). Show that i the amplitude is symmetric under the interchange of t ↔ u. . . xn + a) = G(n) (x1 . Express the contributions in invariants s. . . . |M |2 = T11 + T22 − T12 − T21 . . xn ) ≡ (2π)4 δ 4 (p1 + . Exercise 10. It is of the form −i M = A1 − A2 . . . . . (b) Show (by using x1 as shift-varible) that this implies that n i=1 d4 xi ei pi ·xi G(n) (x1 . Exercise 10. their sum is zero. . pn ). . . xn ) ∝ (2π)4 δ 4 (p1 + . . .2 (a) Show that the translation properties of the ﬁelds and the vacuum imply G(n) (x1 + a. pn ) only has n − 1 independent variables in it.3 Show that the combinatorial factors found using the rules given in section 10. or if we in general would consider all momenta as incoming. 1 2 (b) Calculate the quadratic pieces and interference terms. Note that the function G(n) (p1 . which means overall momentum conservation in Green functions in momentum space. . e(p1 ) + e(p2 ) → e(p′ ) + e(p′ ) in lowest order in α. xn ). hence we can write n i=1 d4 xi ei pi ·xi G(n) (x1 .Feynman diagrams for scattering amplitudes 102 Exercise 10.37 and 10. . + pn ). for instance. . t and u. .4 (a) Write down the Feynman diagrams contributing to electron-electron scattering. . . . 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). but the incoming and outgoing momenta are identical. . . . . .

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

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 √ = . ptrf ).11) (11.9) (11. m2 ) a b . m2 .14) One can.8) (11. m2 ) a b . a (11. which in the speciﬁc case a b also can be expressed as λ(s.10) The function λ(s. use the ﬁrst relation and the abovementioned threshold value for s to calculate the threshold for a speciﬁc n-particle ﬁnal state in the target rest frame. Prove that trf Ea = |ptrf | = a s − m2 − m2 a b . mb and s). m2 ) = 4(pa · pb )2 − 4p2 p2 . 2mb λ(s. m2 . 0). (11.16) . In that case trf pa = (Ea . 4π s 4π 8π s 4π (11. for instance. a a b b The second frame considered explicitly is the socalled target rest frame in which one of the particles (called the target) is at rest. m2 .13) (11. 2mb (11. 2 s s − m2 + m2 a b √ . p2 ) = CM = = 1 (2π)2 1 (2π)2 1 (2π)2 d3 p1 d3 p2 4 δ (P − p1 − p2 ) 2E1 2E2 √ d3 q δ( s − E1 − E2 ) 4 E1 E2 q 2 d|q| √ dΩ(ˆ) q δ( s − E1 − E2 ) 4 E1 E2 which using |q| d|q| = (E1 E2 /(E1 + E2 )) d(E1 + E2 ) gives dR(s.15) Explicit calculation of the reduced two-body phase space element gives dR(s. p2 ) = = where λ12 denotes λ(s. p1 . m2 ). 4s (11. m2 . m2 ) is a function symmetric in its three arguments. m2 .Scattering theory 104 It is straightforward to prove that the unknowns in the particular system can be expressed in the invariants (ma .12) pb = (mb . = 2 s λ(s. trf Ea (threshold) = 1 2 mb ( i mi )2 − m2 − m2 a b . Also in this case one can express the energy and momentum in the invariants. p1 . Prove that |q| = cm Ea cm Eb (s − m2 − m2 )2 − 4 m2 m2 a a b b = 4s s + m2 − m2 a b √ = .

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 . pd ) = 4π s 4π 8π s 4π dt dt √ √ . there is usually a preferred direction.24) Using the relation between t and cos θcm it is straightforward to express dΩcm in dt. Instead of the scattering angle. in which one particle is detected..23) which are not independent as s + t + u = m2 + m2 + m2 + m2 .22) (11. At high energies. Examples appear in → π− + p (11. usually is a 2-particle state. 11. 2s (11. Writing E = mT cosh y. In inclusive measurements no particles are detected in the ﬁnal state. s = (pa + pb )2 = (pc + pd )2 t = (pa − pc ) = (pb − pd ) u = (pa − pd )2 = (pb − pc )2 2 2 (11. pc .25) dR(s.26) = 16π q s 8π λab Kinematics of inclusive hard scattering processes In high energy (hard) scattering processes. E . the rapidity y is deﬁned T T x y y= 1 ln 2 E + pz E − pz = ln E + pz mT = tanh−1 pz .4). (11..20) → π0 + n → π+ + π− + n → . A speciﬁc reaction channel starts contributing at the threshold value (Eq. Consider the process a + b → c + d. for instance the momenta of incoming (colliding) hadrons and it is useful to use for the produced particle rapidity as a variable. where the ﬁrst is referred to as elastic channel and the set of all other channels as the inelastic channels. Of course there are not only 2-particle channels. pT = (px . pz = mT sinh y. An often used set of invariants are the Mandelstam variables. while the ﬁnal state often arises from a series of 2-particle processes combined with the decay of an intermediate particle (resonance).17) (11. The variable s is always larger than a c b d the minimal value (ma + mb )2 . . usually many particles are produced.27) with m2 = m2 − p2 = m2 + p2 + p2 . Consider the 1-particle inclusive case. = (11. H1 + H2 → h + X.21) (11. py ). The minimum and maximum values for t correspond to θcm CM = m2 + m2 − 2 Ea Ec ± 2 qq ′ a c = m2 a + m2 c (s + m2 − m2 )(s + m2 − m2 ) a c b d − ± 2s √ λab λcd . in exclusive measurements all particles are detected.Scattering theory 105 Kinematics of 2 → 2 scattering processes π− + p The simplest scattering process is 2 particles in and 2 particles out.18) (11. dt = 2 qq ′ d cos θcm and obtain for the two-body phase space element √ q ′ dΩcm λcd dΩcm √ = (11. a c with q = λab /4s and q ′ = being 0 or 180 degrees. tmax min λcd /4s. The various possibilities are referred to as diﬀerent reaction channels. The initial state. however.19) (11.

which means that rapidity distributions dN/dy maintain their shape. u). The one-particle phase space in terms of these variables becomes d3 p dy d|pT |2 dφ dy d2 pT |p| dη d|pT |2 dφ = = = . c bd Mad→c¯(su = u. b Analiticity of the ﬁeld theoretically calculated result implies Ma¯→¯ (st = t.30) (11. tt = s. t = (p1 − p3 )2 = (p2 − p4 )2 . 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. t-channel and u-channel processes respectively. t. For large energies and not too small scattering angles (θ ≫ 1/γ) y is approximately equal to the pseudo-rapidity η. Given a two-to-two scattering process ab → cd one can relate the processes b p p a 2 1 d p 4 _ c −p p 1 3 d p 4 _ d −p p a 1 4 _ b −p 2 p 3 c ab cd a _ ac −p 2 _ _ b bd _ ad p 3 _ c cb They are referred to as s-channel. These variables are precisely the Mandelstam variables for the s-channel process (ab → cd). uu = (pa − p¯)2 = s. ¯ tu = (pa − pc )2 = t. 11. t.31) . u) with s = (p1 + p2 )2 = (p3 + p4 )2 .28) which (only involving angles) is easier to determine. (11. η = − ln(tan(θ/2)) = tanh−1 (cos θ). ut = u) = Mab→cd (s. ¯ while for the u-channel process (ad → c¯ one has b) su = (pa + pd )2 = u. (11.29) 3 2E 3 2 (2π) 2 (2π) 16π 2π E 16π 2 2π with at high energies the factor |p|/E ≃ 1. t. b ut = (pa − pd )2 = u.2 Crossing symmetry In the previous chapter. For the t-channel process (a¯ → ¯ one has c bd) st = (pa + pc )2 = t. This is known as crossing symmetry. With the momenta deﬁned as in the ﬁgures above one has for all these processes the same amplitude M (s. u). ¯ b (11. tu = s.Scattering theory 106 It is convenient because under a boost (along z) with velocity β it changes as y → y − tanh−1 β. u = (p1 − p4 )2 = (p2 − p3 )2 . uu = s) = Mab→cd (s. ¯ tt = (pa − p¯)2 = s.

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

For the (proper) decay width one thus has 1 N Γ0 = .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.39) One of the δ functions can be rewritten as T · V (remember the normalization of plane waves). (11. N and T · V are covariant. Using ρtrf = a (0) (0) (0) lab lab lab ρa · γa = ρa · Ea /ma (where ρa is the rest frame density) and va = plab /Ea we have a trf ρa ρb va = (0) ρa ρb 2 λab 4 ma mb (0) (0) or with ρa = 2 ma . .T d4 x = V · T (2π)4 δ 4 (Pi − Pf ). pn ). . σ= √ 2 λab T · V (11. Microscopically one has Ndecay = Na · Γ T or Γ= N 1 . . N 1 . the number of scattered or decayed particles is given by 2 N = (2π)4 δ 4 (Pi − Pf ) iMf i dR(p1 .T V. 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 .35) This quantity is not covariant.33) Decay of particles For the decay of particle a one has macroscopically dN = −Γ N.e. T · V ρa (11.36) (11. as expected. the amount of decaying particles is proportional to the number of particles with proportionality factor the em decay width Γ. (11. From the solution N (t) = N (0) e−Γ t one knows that the decay time τ = 1/Γ. it is. .34) i. dt (11. .Scattering theory 108 Although this at ﬁrst sight does not look covariant. 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). (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.

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

q ˆ i. Using σ= dΩ q ′ M (s. σel = 4π ℓ (2ℓ + 1)|Mℓ (s)|2 (2ℓ + 1) ℓ = 4π q2 ηℓ e 2i δℓ − 1 2i 2 . which allows writing Mℓ (s) = Sℓ (s) − 1 e2i δℓ (s) − 1 = . (11. ℓ (11. (11. θcm ) √ q 8π s 2 . Show that σT = = 4π q 2π q2 (2ℓ + 1) Im Mℓ (s) ℓ ℓ (2ℓ + 1) (1 − ηℓ cos 2δℓ ). 2i q 2i q (11. the total cross section can be determined. (11. in combination with the partial wave expansion for the amplitudes M and the orthogonality of the Legendre polynomials immediately gives for the elastic channel. parametrized as Sℓ (s) = ηℓ (s) exp(2i δℓ (s)).54) ℓ . † −i (Mℓ )f i − (Mℓ )f i = 2 † (Mℓ )f n qn (Mℓ )ni .e.53) The diﬀerence is the inelastic cross section.48) If only one channel is present this simpliﬁes to ∗ ∗ −i (Mℓ − Mℓ ) = 2 qMℓ Mℓ . η = 1) the result σel = 4π q2 (2ℓ + 1) sin2 δℓ .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 .49) or Im Mℓ = q |Mℓ |2 . In general a given channel has |Sℓ (s)| ≤ 1.50) where Sℓ (s) satisﬁes |Sℓ (s)| = 1 and δℓ (s) is the phase shift. σinel = π q2 2 (2ℓ + 1) (1 − ηℓ ). 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 .52) From the imaginary part of M (s. 0). n (11.51) and for the case that this is the only channel (purely elastic scattering. (11.

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

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

s − M∆ ∆ ∆ 2 calculate Re N (M∆ ) .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∆ .000 123 Γ. Compare the results with the experimental π − lifetime τ = 2. Calculate now fπ . cm |va − vb | in the center of mass system for two colliding particles a and b. writing N (s) M1 (s) = 2 + iM Γ . Thus. 4. a quantity which we will ¯ ¯ encounter in the next chapter (section 12. Compare the result with the factor trf 1/ρa ρb va that we used in the derivation of the basic expression for cross sections. ρcm ρcm a b 1 cm .6 × 10−8 s and the relative decay widths Γ(π − → µ− νµ ) = 0. we have here already used unitarity to ﬁnd a ﬁnite imaginary part that is proportional to the total decay width.Scattering theory 113 Exercise 11. As discussed for unstable particles. Show that unitarity can be used one step further to ﬁx also the real part of the residue of the pole in the amplitude. Exercise 11.2).3 Complete the calculation of the decay width and lifetime of the pion by including the appropriate phase space factors and using the calculation of the squared matrix element in Exc. Exercise 11.2 Calculate the ﬂux factor.8.999 877 Γ and Γ(π − → e− νe ) = 0.

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

ds s dxµ ds′ P exp ig A (s′ ) ψ(0). (12. = .16) which in the presence of the ’connection’ Aµ diﬀers from the total change between x ψ(x) and x+dx ψ(x+ dx). (12. if two diﬀerent paths to the same point give diﬀerent phases the eﬀects can be observed.e. -dx -dy µ µ dyµ µ dx For a vector. i.16. 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).14) (12.15) ψ(x).19) This gives rise to a (path dependent) phase in each point. 12. 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 . which connects two identical vectors.The standard model the basis for x G and x+dx 117 G. x+dx ψ(x + dx) = 1 + dxµ Dµ x+dx ψ(x). should be independent of such transformations. We note that local gauge invariance requires that we can modify all local systems with a (local) unitary transformation S(x). A ’constant’ vector that only rotates because of the arbitrary deﬁnitions of local frames satisﬁes Dµ ψ(x) = 0.e.17) The three equations given so far immediately give Dµ = ∂µ − ig Aµ (x). In principle such a phase in a given point is not observable. (12. The relation in Eq. but expresses them with respect to diﬀerent bases. requiring that the ’connection’ Aµ (x) is such that Dµ → S(x)D µ S −1 (x). x x+dx ψ(x + dx) = (1 + dxµ ∂µ ) ψ(x).18) or considering a path xµ (s) from a ﬁxed origin (0) to point x. However. ds which is solved by dψ(s) ds = dxµ . Dν ] ψ(x) 1 − ig dσ µν Gµν ψ(x). (12. i. x+dx ψ(x) = = 1 + ig dxµ Aa (x)Ta µ 1 + ig dx Aµ (x) µ x x ψ(x) (12. 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. What is this physical eﬀect by which the ’connection’ Aµ can be observed? For this consider the phase around a closed loop. ∂µ − ig Aµ (x) ψ(x) = 0 dxµ ∂µ − ig Aµ (x(s)) ψ(x) = 0.

this means that there is more than one possibility for the groundstate.The standard model 118 where Gµν = (i/g)[Dµ .2 Spontaneous symmetry breaking In this section we consider the situation that the groundstate of a physical system is degenerate. Nevertheless. Nature will choose one. This characterizes spontaneous symmetry breaking. usually being (slightly) prejudiced by impurities.g. In this case there are many possible groundstates (determined by a ﬁxed direction in space). e. which means that by an appropriate transformation S(x) it can be gauged away (see the example of Aharonov-Bohm eﬀect). the groundstate itself appears degenerate. 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).e. φ= 2 φ3 (12. In geometric language the eﬀect on parallel transport of a vector depends on the ’curvature’ Gµν . As there can be one and only one groundstate. a theory for a scalar degree of freedom (a scalar ﬁeld) having three (real) components. Dν ]. 12. . If it is zero one has dxµ Aµ (x) = 0. Only if this quantity is nonzero a physically observable eﬀect of Aµ exists.2. we can disregard those ’perturbations’ and look at the ideal situation. 12.2: The symmetry-breaking ’potential’ in the lagrangian for the case that m2 < 0.20) with a lagrangian density of the form L = 1 1 1 ∂µ φ · ∂ µ φ − m2 φ · φ − λ(φ · φ)2 . Classically the (time-independent) ground state is found for V(φ ) F |φ| Figure 12. and equivalently Aµ can be considered as a pure gauge eﬀect. Consider as an example a ferromagnet with an interaction hamiltonian of the form H=− Jij Si · Sj . in reality a not perfectly symmetric situation. i. If the temperature is high enough the spins are oriented randomly and the (macroscopic) ground state is spherically symmetric. external magnetic ﬁelds. φ1 φ . 2 2 4 −V (φ) The potential V (φ) is shown in ﬁg. If the temperature is below a certain critical temperature (T < Tc ) the kinetic energy is no longer dominant and the above hamiltonian prefers a lowest energy conﬁguration in which all spins are parallel. i>j which is rotationally invariant.

A quantum ﬁeld theory has only one nondegenerate groundstate |0 . therefore a choice must be made. i. The ﬁeld along the third axis plays a special role because of the choice of the vacuum expectation value.23) F +η the latter only forming a minimum for m2 < 0. (12.e. In order to see the consequences for the particle spectrum of the theory we construct the lagrangian in terms of the ﬁelds ϕ1 . In this mode the lagrangian and the vacuum are both invariant under a set of symmetry transformations generated by Qa . etc. The situation now is the following. for the vacuum Qa |0 = 0.22) F Realization of symmetries (model independent) In this section we want to discuss a bit more formal the two possible ways that a symmetry can be implemented. . It is appropriate to redeﬁne the ﬁeld as ϕ1 φ = ϕ2 .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. λ such that 0|ϕ1 |0 = 0|ϕ2 |0 = 0|η|0 = 0. . Writing the ﬁeld φ as a sum of a classical and a quantum ﬁeld. while the quadratic terms must correspond with non-negative masses.21) Stability of the action requires the classical groundstate ϕc to have a well-deﬁned value (which can be nonzero). while the lagrangian including the nonzero groundstate expectation value chosen by nature. (12. Any constant ﬁeld with ’length’ F is a possible groundstate. has less symmetry. In the case of degeneracy. 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. corresponding to the number of broken generators (in this case rotations around 1 and 2 axis) and 1 massive scalar particle with mass m2 = −2m2 . . The original lagrangian contained an SO(3) invariance under (length conserving) rotations among the three ﬁelds. The classical groundstate appears degenerate.25) 2 2 2 Therefore there are 2 massless scalar particles. They are known as the Weyl mode or the Goldstone mode: Weyl mode. In this case the spectrum is described .) terms constitute interaction terms. (12. ϕc = 0|φ |0 = (12. The η massless particles are called Goldstone bosons.24) 1 2 2 4 1 1 1 = (∂µ ϕ1 )2 + (∂µ ϕ2 )2 + (∂µ η)2 + m2 η 2 + . ϕ2 and η. φ = ϕc + φquantum where for the (operator-valued) coeﬃcients in the quantum ﬁeld one wants 0|c† = c|0 = 0. The result is 1 1 1 1 (∂µ ϕ1 )2 + (∂µ ϕ2 )2 + (∂µ η)2 − m2 (ϕ2 + ϕ2 ) L = 1 2 2 2 2 2 1 2 1 − m (F + η)2 − λ (ϕ2 + ϕ2 + F 2 + η 2 + 2 F η)2 (12. In this situation one speaks of spontaneous symmetry breaking. It is only invariant under rotations around the 3-axis. It is suﬃcient to do this to second order in the ﬁelds as the higher (cubic. so one has 0|φquantum |0 = 0 and hence 0|φ|0 = ϕc .

. A bit more formal. down. . V µ = ψγ µ T ψ.28) where ψ is extended to a vector in ﬂavor space and M is a diagonal matrix in ﬂavor space. This lagrangian density then is invariant under unitary (vector) transformations in the ﬂavor space. This means that they are operators that create states from the vacuum. H] = 0. As the a generators for a symmetry are precisely the zero-components of a conserved current Jµ (x) integrated a a over space.31) . In this mode the lagrangian is invariant but Qa |0 = 0 for a number of generators. Goldstone mode.) one can write L = f ψ f (i/ − Mf )ψf . if the generators Qa generate a symmetry. i. T = τ /2. (12. Taking the derivative.. strange. i. (12. Using the Dirac equation. a and a′ are degenerate states.30) which in the case that we include only two ﬂavors form an SU (2)V symmetry (isospin symmetry) generated by the Pauli matrices. T ] ψ. there must be a nonzero expectation value 0|Jµ (x)|π (k) . Note that for the ﬁelds π a (x) one would have a the relation 0|π a (x)|π a (k) = ei k·x . ∂ ∂ (12. irrespective of the fact if they annihilate the vacuum. a massless Goldstone boson for each ’broken’ generator. Neglecting at this point the local color symmetry. 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. 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. (12. suggesting the stronger relation ∂ µ Jµ (x) = fπ m2 a π a (x). Known examples are rotational symmetry and the fact that the spectrum shows multiplets labeled by angular momentum ℓ (with members labeled by m).29) (Note that each of the entries in the vector for ψ in the above equation is still a 4-component Dirac spinor). π (12. and one must have mπa = 0. = ψ(i/ − M )ψ. called ﬂavors. denoted |π a (k) .e.27) If the transformations in the lagrangian give rise to a symmetry the Noether currents are conserved. M = . etc.e. a ∂ µ Jµ = 0. i. The generators Qa (in that case the rotation operators Lz .e. [Qa . . The conserved currents corresponding to this symmetry transformation are found directly using Noether’s theorem (see chapter 6).32) (12.26) for all the states labeled by a corresponding to ’broken’ generators. 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′ . it is easy to see that one gets ∂µ V µ = i ψ [M.The standard model 120 by degenerate representations of the symmetry group. π Example: chiral symmetry An example of spontaneous symmetry breaking is chiral symmetry breaking in QCD. . the lagrangian for the quarks consists of the free Dirac lagrangian for each of the types of quarks. Including a sum over the diﬀerent ﬂavors (up. ψu mu md ψ = ψd . ψ −→ ei α·T ψ. a 0|∂ µ Jµ (x)|π b (k) = fπ k 2 ei k·x δab = fπ m2 a ei k·x δab .

This situation.e. parity would not play a role in the world. proton and neutron.g.e. In the real world. while the groundstate is only invariant under isospin rotations (R = L). stated equivalently. The Goldstone bosons of QCD are the pions for which fπ = 93 MeV and which have a mass of mπ ≈ 138 MeV/c2 . it is easy to see that one gets ∂µ Aµ = i ψ {M.33) which in the case of two ﬂavors form SU (2)A transformations generated by the Pauli matrices. where the quark masses are not completely zero.3 MeV/c2 and Mn = 939. fπ = 0 remains.34) Using the Dirac equation.e. for which the ﬁeld (according to the discussion above) has the same behavior under parity. ψR/L = 1 (1 ± γ5 )ψ. The chiral ﬁelds ψR and ψL are transformed into each other under parity. T ] = 0. both are very small. nature has choosen the path SU (2)L ⊗ SU (2)R =⇒ SU (2)V .35 is nonzero. behaves as a pseudoscalar particle (spin zero. This symmetry. (leaving out the ﬂavor structure) the same as ψγ5 ψ. etc. however. 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. including in Dirac space a γ5 matrix. This symmetry is realized in the Weyl mode with the spectrum of QCD showing an almost perfect isospin symmetry. etc. generated by TR/L = 1 (1 ± γ5 )T . but the fact that the symmetry is not perfect and the right hand side of Eq. From group theory (Schur’s theorem) one knows that the latter can only be true. There exists another set of symmetry transformations. a doublet (isospin 1/2) of nucleons.e. . Therefore realization in the Weyl mode would require that all particles come double with positive and negative parity. with almost degenerate masses (Mp = 938. The currents corresponding to this symmetry transformation are again found using Noether’s theorem. or. We know that mesons and baryons (such as the nucleons) have a well-deﬁned parity that is conserved.6 MeV/c2 ). socalled axial vector transformations. if in ﬂavor space M is proportional to the unit matrix.36) If the mass is zero the lagrangian is split into two disjunct parts for L and R showing that there is a direct product SU (2)L ⊗ SU (2)R symmetry. ∂ ∂ (12. 12. e. 12. Still the basic fact that the generators acting on the vacuum give a nonzero result (i. which is approximately true for the up and down quarks. chiral symmetry is not perfect. How can we see this. is by nature not realized in the Weyl mode. M = m · 1. parity minus). That the massless theory has this symmetry can also be seen by writing it down for the socalled lefthanded and righthanded fermions.e. but also a triplet (isospin 1) of pions. much smaller than any of the other mesons or baryons. i. i. i. the lagrangian density is invariant under left (L) and right (R) rotations independently. gives also rise to a nonzero mass for the Goldstone bosons according to Eq. T } γ5 ψ. I. as the quantity ∂µ Aµ (x). T γ5 = τ γ5 /2. nucleons and atomic nuclei has not only an isospin or vector symmetry SU (2)V but also an axial vector symmetry SU (2)A . Note that these transformations also work on the spinor indices. is what happens in the real world. (12. Therefore the world of up and down quarks describing pions. SU (2)V (isospin) symmetry is good if the up and down quark masses are identical.27. ψ −→ ei α·T γ5 ψ. From the number of broken generators it is clear that one expects three massless Goldstone bosons. (12.The standard model 121 Furthermore ∂µ V µ = 0 ⇐⇒ [M. Aµ = ψγ µ T γ5 ψ.35) In this case ∂µ Aµ = 0 will be true if the quarks have zero mass. This combined symmetry is what one calls chiral symmetry. which is equivalent 2 to the V-A symmetry. in terms of which the Dirac lagrangian 2 density looks like L = i ψL / ψL + i ψR / ψR − ψR M ψL − ψL M ψR . (12.

. Initially there are 3 massless gauge ﬁelds (each. One may wonder about the degrees of freedom. . but one has 1 massless gauge ﬁeld (2).37) (12.40) a Dµ φ = ∂µ φ − ig Wµ La φ. while the number of massive gauge bosons coincides with the number of ’broken’ generators. Therefore we can always rotate the ﬁeld φ into the 3-direction in order to simplify the calculation. like a photon. made into a gauge theory. one sees that the ﬁelds Wµ and Gµν also can be represented as three-component ﬁelds. again 9 degrees of freedom. The symmetry is broken in the same way as before and the same choice for the vacuum. as in this case there are no massless Goldstone bosons. We have the possibility to perform local gauge transformations.42) 2 from which one reads oﬀ that the particle content of the theory consists of one massless gauge boson 3 1 2 (Wµ ). The latter is a spin 0 particle (real scalar ﬁeld) called a Higgs particle. two massive bosons (Wµ and Wµ with MW = gF ) and a massive scalar particle (η with m2 = η 2 −2 m .38) ∂µ Wν − ∂ν Wµ + g Wµ × Wν . (12. 0 ϕc = 0|φ |0 = 0 . Considering the same example with rotational symmetry (SO(3)) as for spontaneous symmetry breaking of a scalar ﬁeld (Higgs ﬁeld) with three components. 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 + .e. 1 1 L = − Gµν · Gµν + Dµ φ · Dµ φ − V (φ). 0 φ= (12.The standard model 122 12. having two independent spin components) and three scalar ﬁelds (one degree of freedom each). . The diﬀerence comes when we reparametrize the ﬁeld φ.39) (12. (12. i. 4 2 where Since the explicit (adjoint. 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 . 2 massive vector ﬁelds or spin 1 bosons (2 × 3) and one scalar ﬁeld (1). After symmetry breaking the same number (as expected) comes out. . is made. Dµ φ = Gµν = ∂µ φ + g Wµ × φ.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 . 0 0 0 = .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. thus 9 independent degrees of freedom. (12.

R −→ e−i β R.50) (12. R. = ∂µ L − (12. Thus the lagrangian density is L (f ) ∂ ∂ = i L/ L + i R/ R.51) . implies certain nontrivial relations between masses and mixing angles that can be tested experimentally and sofar are in excellent agreement with experiment. The lagrangian for the leptons consists of three families each containing an elementary fermion (electron e− .48) (12. ψR/L = 2 (1 ± γ5 )ψ decouple. For the neutrino only a lefthanded particle (and righthanded antiparticle) exist. that gives a good description of the physical world. i. however. etc.e.4 The standard model SU (2)W ⊗ U (1)Y The symmetry ideas discussed before play an essential role in the standard model that describes the elementary particles. This is written as 3 Q = TW + YW . 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. The procedure.and righthanded particles. τ ). explicitly L R SU(2)W SU(2)W −→ ei α·τ /2 L. 2 (12.) and the gauge bosons responsible for the strong.47) and YW is considered as an operator that generates a U (1)Y symmetry. νµ and ντ ) and their antiparticles. By introducing gauge ﬁelds and breaking the symmetry a more complex lagrangian is obtained. the quarks (up.43) and One introduces a (weak) SU (2)W symmetry under which eR forms a singlet. down.45) (12. explicitly Dµ L Dµ R i i g Wµ · τ L + g ′ Bµ L.The standard model 123 12. muons. while the lefthanded particles form a doublet. ∂ ∂ ∂ (12. etc. U(1)Y U(1)Y (12. under which the lefthanded and righthanded particles with YW (L) = −1 and YW (R) = −2 transform with eiβYW /2 . explicitly L −→ e−i β/2 L. muon µ− or tau τ − ). neutrinos. electromagnetic and weak forces. 2 2 = ∂µ R + i g ′ Bµ R. (12.44) which has an SU (2)W symmetry under transformations eiα·TW .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 leptons (elektrons. 1 As they are massless. its corresponding neutrino (νe . left. In the standard model one starts with a very simple basic lagrangian for (massless) fermions which exhibits more symmetry than observed in nature. Thus L (f ) = i eR / eR + i eL / eL + i νeL / νeL + (µ.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. (12.).

the U (1)Q symmetry (generated by the charge operator). . L where L (h1) (h) =L (h1) +L (h2) . (12. a complex Higgs ﬁeld + 1 φ √2 (θ2 + iθ1 ) 0 = 1 (12.53) φ= √ (θ4 − iθ3 ) φ 2 with TW = 1/2 and YW = 1 is introduced.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. leading to the parametrization 1 0 φ(x) = √ (12. .56) v + h(x) 2 Dµ φ = (∂µ − and Dµ φ = = 1 √ 2 − ig 2 ∂µ h + 1 2 Wµ −iWµ √ (v + h) 2 3 gWµ −g′ Bµ i √ (v + 2 2 Up to cubic terms.58) (12.The standard model 124 3 3 where Wµ is a triplet of gauge bosons with TW = 1. Using local gauge invariance θi for i = 1. h) (12. (12.57) = 1 g2 v2 1 2 (Wµ )2 + (Wµ )2 (∂µ h)2 + m2 h2 + 2 8 v2 2 3 gWµ − g ′ Bµ + . 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. + 8 (12. . 2 and 3 may be eliminated (the necessary SU (2)W rotation is precisely e−iθ(x)·τ ). (12. † i i g Wµ · τ − g ′ Bµ )φ. with the following lagrangian density consisting of a symmetry breaking piece and a coupling to the fermions.59) .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 . 4 4 In order to break the symmetry to the symmetry of the physical world. which assures with the choice of YW for the Higgs ﬁeld that Q generates the remaining U (1) symmetry. + 8 g2 v2 1 + − (Wµ )2 + (Wµ )2 (∂µ h)2 + m2 h2 + 2 8 (g 2 + g ′2 ) v 2 (Zµ )2 + . . Putting this in leads to L (f ) =L (f 1) +L (f 2) . .54) = (Dµ φ)† (Dµ φ) −m2 φ† φ − λ (φ† φ)2 . For the classical ﬁeld the choice θ4 = v is made. −V (φ) L and (h2) = −Ge (LφR + Rφ L). this leads to the lagrangian L (h1) .

.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 . 4 2 MW g2 v2 = . 2 3 g Wµ − g ′ Bµ 3 ≡ cos θW Wµ − sin θW Bµ .69) . we can read oﬀ e = g sin θW = g ′ cos θW . 2θ 2θ 4 cos W cos W 0. 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 + .60) (12. (12. . and correspond to three massive particle ﬁelds (W ± and Z 0 ) and one massless ﬁeld (photon γ) with 2 MW 2 MZ 2 Mγ = = = g2 v2 .61) (12.64) (12.67) The coupling of electrons or muons to their respective neutrinos. (12.62) g 2 + g ′2 3 ≡ sin θW Wµ + cos θW Bµ .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 . σ (νµ γ ρ µL ) 2 2 (eL γ νe ) 2 k + MW (12. The coupling of the fermions to the physical gauge bosons are contained in L (f 1) giving L (f 1) = i eγ µ ∂µ e + i νe γ µ ∂µ νe − g sin θW eγ µ e Aµ g 1 1 + sin2 θW eR γ µ eR − cos 2θW eL γ µ eL + νe γ µ νe cos θW 2 2 g µ + µ − + √ νe γ eL Wµ + eL γ νe Wµ .63) (12. 2 Zµ (12. 2 (12. g ′ /g = tan θW .66) From the coupling to the photon. 2 + g ′2 g 3 g ′ Wµ + g Bµ (12.65) The weak mixing angle is related to the ratio of coupling constants.

muon or tau. 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. as appear for instance in the decay of the neutron or the decay of the Λ (quark content uds). −5 (12. . 91. The masses of the fermions and the coupling to the Higgs particle are contained in L (h2) .19 GeV.The standard model 126 the good old four-point interaction introduced by Fermi to explain the weak interactions. 3 = TW . i. 2 2 2 (12. For each neutrino species (with CV = 1/2 and CA = 1/2 one expects Γ(νν) ≈ 155 MeV.80) First.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 .71) GeV −2 = = = = . Finally we want to say something about the weak properties of the quarks. Comparing this with the total width into (invisible!) channels.74) (12. leptons with CV = −1/2 + 2 sin2 θW ≈ −0.78) From this coupling it is straightforward to calculate the partial width for Z 0 into a fermion-antifermion pair. 80. (12.g. (12. one has the relation g2 e2 1 GF √ = .5 MeV (exp.231 2. Γe ≈ Γµ ≈ Γτ ≈ 83 MeV). i.e. the mass of the electron comes from the spontaneous symmetry breaking but is not predicted (it is in the coupling Ge ).05 and CA = −1/2 we calculate Γ(e+ e− ) ≈ 78. 2 The value of α deviates from the known value α ≈ 1/137 because of higher order contributions. (12.72) (12. g2 MZ 2 2 (CV + CA ).79) Γ(Z 0 → f f ) = 48π cos2 θW For the electron.73) (12.70) = 2 = 2 sin2 θ 8 MW 2 v2 8 MW 2 W In this way the parameters g. from the MW mass is about 250 GeV.166 4 × 10 0. such as2 α = e2 /4π GF sin2 θW MW MZ ≈ 1/127.e. (12. g ′ and v determine a number of experimentally measurable quantities. giving rise to a running coupling constant after renormalization of the ﬁeld theory.77) (12. me /v is extremely small. v (12. 1. one obtains L (h2) Ge v Ge = − √ (eL eR + eR eL ) − √ (eL eR + eR eL ) h 2 2 me = −me ee − eeh. With the choosen vacuum expectation value for the Higgs ﬁeld.76) with CV CA 3 = TW − 2 sin2 θW Q. Γinvisible = 480 MeV one sees that three families of (light) neutrinos are allowed.40 GeV. The coupling to the Higgs particle is weak as the value for v calculated e.

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

φ = −φ− 0 2 For the (squared) complex matrices we can ﬁnd positive eigenvalues. d d (12.84) with Gu and Gd being real and positive and Wu and Wd being diﬀerent unitary matrices. each of these containing the three families. for which we need the appropriate weak isospin doublet 0∗ 1 φ c = √ v +h .The standard model 128 using the socalled Wolfenstein + O(λ4 ) with λ ≈ 0. The Higgs ﬁeld is still limited to one complex doublet.86) the unitary CKM-matrix introduced above in an ad hoc way. i. One then reads oﬀ that starting with the family basis as deﬁned via the left doublets that the mass eigenstates (and states coupling to the Higgs ﬁeld) involve the righthanded † mass † mass mass † mass states UR = Wu UR and DR = Wd DR and the lefthanded states UL = Vu UL and DL = † Vd DL . (12. UR = uR cR tR .34. 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 . † Λu Λ† = Vu G2 Vu .85) √ √ with Mu = Gu v/ 2 (diagonal matrix containing mu . 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. Thus one has † † † =⇒ −DL Vd Md Wd DR − DR Wd Md Vd† DL − UL Vu Mu Wu UR − UR Wu Mu Vu UL (12. ms and mb ). allowing us to write † Λu = Vu Gu Wu † and Λd = Vd Gd Wd . the three righthanded quarks with charge +2/3 in UR and the three righthanded quarks with charges −1/3 in DR . . the weak mass eigenstates are L (h2. A ≈ 0.83) where Vu and Vd are unitary matrices. respectively). The imaginary part i η gives rise to CP violation in decays of K and B-mesons (containing s and b quarks. e. The Λu and Λd are complex matrices in the 3×3 family and Λd Λ† = Vd G2 Vd† . (12.q) ′ weak † mass mass DL = DL = Vu Vd DL = VCKM DL .82) space. Note that we need the conjugate Higgs ﬁeld to get a U (1)Y singlet in the case of the charge +2/3 quarks. mc and mt ) and Md = Gd v/ 2 (diagonal matrix containing md .q) = −QL φΛd DR − DR Λ† φ† QL − QL φc Λu UR − UR Λ† φc† QL u d (12.22 and η ≈ 0.e.g.227.5 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 (h2.82 and ρ ≈ 0. u u where we include now the three lefthanded quark doublets in QL .

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

is to add in Eq.ν = − MR NR NR + MR NR NR .The standard model 130 The lepton sector (massive Majorana ﬁelds) An even simpler option than sterile righthanded Dirac neutrinos. however. Thus 2 Υ ′ = nc + nL . Υ′ and Υ′ (satisfying Υc = Υ and as discussed in chapter 6. couple the right. Absorption of phases in the states is not possible for Majorana neutrinos.99) . with for the righthanded sector a mass term MR . are equivalent to two decoupled Majorana neutrinos (see below).97) As the most general mass term in the lagrangian density we have L mass = − = which for MD = 0 is a pure Majorana lagrangian and for ML = MR = 0 and real MD represents the Dirac case. the massless and massive Majorana neutrinos.c. Actually. The mass matrix can be written as ML |MD | eiφ M = (12.98) (12. it is easy to see R that this incorporates the Dirac case by considering the lefthanded part of Υ′ and the righthanded 1 part of Υ′ as a Dirac spinor ψ.and righthanded species then just form a Dirac fermion. For the neutrino sector. Υc ≡ (ΥL )c = ΥR 1 2 L and Υc = ΥL ). one must for the neutrinos as well as charged leptons. (12. The see-saw mechanism Consider (for one family N = n) the most general Lorentz invariant mass term for two independent Majorana spinors. 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. one being massive. as above. 1 c ∗ c L mass.100) |MD | eiφ MR 1 1 ∗ ∗ M L nc nL + M L nL nc − M R nR nc + M R nc nR L R L R 2 2 1 1 ∗ ∗ − M D nc nc + M D nL nR − M D nR nL + M D nc nc L R R L 2 2 1 nc nR M L M D nL L c + h. 1 L Υ ′ = nR + nc . 2 although this option is not attractive as it violates the electroweak symmetry. hence the mixing matrix becomes iα /2 0 0 e 1 (12. We use here the primes starting with the weak eigenstates. If the Majorana mass MR ≫ MD one actually obtains in a natural way one Majorana neutrino with a very small mass. 2 R ψ = nR + nL . the left. For the leptons. For these light Majorana neutrinos one has.88 a Majorana mass term for the (lefthanded) neutrino mass eigenstates. 12. 1 ∗ c c (12. The way to circumvent this is to introduce as in the previous section righthanded neutrinos.95) 2 In order to have more than a completely decoupled sector.94) M L NL NL + M L NL NL . a unitary matrix relating them to the weak eigenstates. coupled by a Dirac mass term. α2 and δ). For the charged leptons there cannot exist a Majorana mass term as this would break the U(1) electromagnetic symmetry. Thus (disregarding family structure) one has two Majorana neutrinos. (12. This is called the see-saw mechanism (outlined below). − MD MR nR 2 (12.96) VPMNS = UPMNS K with K = 0 eiα2 /2 0 .ν =− containing three (CP-violating) phases (α1 .

The standard model 131 taking ML and MR real and non-negative.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 . 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′ . Prove that if the generators Qa generate a symmetry. which implies U ∗ M † U † = U ∗ M ∗ U † = M0 and a ’normal’ diagonalization of the (hermitean) matrix M M † . (12. i. i. a and a′ are degenerate states. H] = 0.e. i ψ {M. 2 U (M M † ) U † = M0 .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 . This is a mixing problem with a symmetric (complex) mass matrix leading to two (real) mass eigenstates. (12. related via c Υ1R = U nL . L Υ2L nR Υ1 and Υ2 .1 Consider the case of the Weyl mode for symmetries.105) for i = 1. Υ2R nR (12. For the situation ML = 0 and MR ≫ MD (taking MD real) one 2 ﬁnds M1 ≈ MD /MR and M2 ≈ MR . Exercise 12.e. [Qa . T ] ψ. Exercises Exercise 12. 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). V µ = ψγ µ T ψ and Aµ = ψγ µ γ5 T ψ ∂µ V µ ∂µ A µ = = i ψ [M. .104) ↔ 1 1 Υi i / Υi − Mi Υi Υi ∂ 4 2 (12. T } γ5 ψ. 2 with real masses Mi . 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 . (12. Exercise 12.2 Derive for the vector and axial vector currents.102) = 1 2 2 ML + MR + 2|MD |2 2 ± 2 2 2 2 (ML − MR )2 + 4|MD |2 (ML + MR + 2ML MR cos(2φ)) .103) for each of which one ﬁnds the lagrangians L = and we are left with two decoupled Majorana ﬁelds Υ1L = U ∗ nc .

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

. Given that neutrino masses are of the order of 0. Give the explicit expressions that relate ψ and ψ c with Υ1 and Υ2 .The standard model 133 (a) Calculate for the special choice ML = MR = 0 and MD real. which leads to the socalled see-saw mechanism. what is the mass MX if we take for MD the electroweak symmetry breaking scale v (about 250 GeV). (b) A more interesting situation is 0 = ML < |MD | ≪ MR . the mass eigenvalues and show that the mixing matrix is 1 1 1 √ U= i −i 2 which enables one to rewrite the Dirac ﬁeld in terms of Majorana spinors. Calculate the eigenvalues ML = 0 and MR = MX .05 eV.

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