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

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

1980. L.3 2. Diagrammatica.2 36 22 .V. M. Cambridge University Press.E. These notes Section 1. Zuber. Srednicki.2 3.1 1.2. 4. Peskin and D.1 6.3 3. I: Foundations.2 6.2 6. 1996. Schroeder.4 3.2 3. C. Vol. Peskin & Schroeder and Srednicki. 3.1 4. 1994.1 5. Corresponding chapters in books of Ryder. I refer to the book of Srednicki [1] and Ryder [2].5 5. Itzykson and J.H. 1985.3. 2. M.3 3. 6.3 2.1 2 2 3.5 2.3 4.1. An introduction to Quantum Field Theory. 1995.3 3.1 3. Cambridge University Press. 3.3 6. Quantum Field Theory. M.4 4. S. 3.2 4. Vol. The quantum theory of ﬁelds.5 2. Quantum Field Theory. Cambridge University Press.2 1.5 Ryder Peskin & Schroeder Srednicki 1 1 1 1 3 2.7 2.2 2.5 4. McGraw-Hill. 1995. Weinberg. but are rather compact. The notes contain all essential information.4 3. Cambridge University Press.2 3.-B. 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). 2.4 2.3 3.3 2.2 2.3 3. Ryder.1 2.3 2.2 2. 5. Cambridge University Press.2 3.8 3.4 6.3 2. Veltman. Addison-Wesly.4 3. 2007.4 2.1 2.3 References As most directly related books to these notes. II: Modern Applications.2 3. 1. Other text books of Quantum Field Theory that are useful are given in refs [3-6].1 3. Quantum Field Theory.

4.1 20.1.5 4.1.6 3.6 6.2 9. 6.7.4 These notes Section 7.4 12.2 7.3 2. 4.2 10.1 20.6 3.8 6. 4.5 9.4 9.10 8.3 20.3 9.2 8.2 8.4 4.1 8.3 12.4 12.3.2 12. 9. 8.2 11.4 2.5 7.4 7.2. 9.2 5.2.2. 6.3 7.5.4.4 10.5 11.1 20.4.6 9.5 2.5 6.5 4.3 8.6 8.3 11.7 4.1 9. 20.1 10.2.1 12.1.6 3.4.3.8 5 4.3 6.3 8.3 10.1 5. 2.6 4.3 3. 6. 6. 6.5 9.2 4.3 6 8 9 Srednicki 3 3 3 37 5. 2.3 7.6 10.7 5.5 6 7 6 6.4 3.1 9.5 7.5 Ryder 4 4 4 4 4 4 Peskin & Schroeder 2.1 7.3. 4.4 8.1 11.5 .

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.

Chapter 1 Introduction 1.582 119 15 (56) × 10−22 MeV s.6) . ˜ ψ(p) = i d3 r exp − p · r ψ(r). S ≫ . (1. In this coordinate representation the position operators r op simply acts as rop ψ(r) = r ψ(r). It is given by pop ψ(r) = −i ∇ψ(r). 1 (1. the action of the momentum operator in the coordinate representation is not as simple as the position operator. pj ] = i δij . They satisfy the well-known (canonical) operator commutation relations [ri . ≡ h/2π = 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. In quantum mechanics.1 Quantum ﬁeld theory In quantum ﬁeld theory the theories of quantum mechanics and special relativity are united. One can talk about states |r and the wave function ψ(r) = r||ψ .2) In the limit that v ≪ c one reaches the non-relativistic domain. c = 299 792 458 m s−1 . Corresponding position and momentum operators do not commute. (1.4) where δij is the Kronecker δ function. the position can in principle be determined at any time with any accuracy.3) The uncertainty principle tells us that in this representation the momenta cannot be fully determined.1) In the limit that the action S is much larger than . In special relativity a special role is played by the velocity of light c. (1. usually given divided by 2π.5) One can also choose a representation in which the momenta of the particles are the dynamical variables.054 571 68 (18) × 10−34 J s = 6. ˜ The corresponding states are |p and the wave functions ψ(p) = p||ψ are the Fourier transforms of the coordinate space wave functions. In quantum mechanics a special role is played by Planck’s constant h. Indeed. (1. quantum eﬀects do not play a role anymore and one is in the classical domain. being eigenvalues of the position operators.

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.

. where the coordinate t = x0 is referred to as the time component. These transformations do change the components of a vector. in which case zi = ǫijk xj yk .21) (1. xi are the three space components. such as eigentime τ for the position vector τ 2 ≡ x2 = t2 − x2 . we will use upper indices for the three-vectors. 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 . . When 3-dimensional space is considered as part of Minkowski space.Introduction 5 that can be used in the cross product of two vectors z = x × y. For four-vectors in Minkowski space we will use the notation with upper indices and write x = (t. x1 . x1 . the invariant length or distance (squared) is not positive deﬁnite. Relations relating tensor expressions.23) δjm δkn − δjn δkm .νn can be conν structed.24) The quantity gµν ≡ nµ · nν is the metric tensor. xµ = gµν xν . x2 . x3 ). No diﬀerence is made between upper or lower. 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. Many other four vectors and tensors transforming like T ′ µ1 . in this case an inertial system exists in which the two points are at the same time and ds2 just represents minus the spatial distance squred ds2 = −dx2 ..22) (1. The lower indices are constructed in the following way. Λµn n T ν1 . the points lie on the lightcone and they can be connected by a light signal. The real. Vectors are denoted x = xµ nµ . The length (squared) of a vector is obtained from the scalar ˆ product. it is convenient to distinguish lower indices from upper indices. x2 = x · x = xµ xν nµ · nν = xµ xν gµν ..3). Because of the diﬀerent signs occurring in gµν . One has x2 = xµ xµ = t2 − x2 . (1. The (invariant) lengths often have special names. and are given by (x0 . One can distinguish: • ds2 > 0 (timelike intervals). −x).25) Within Minkowski space the real. So we could have used all upper indices in the above equations. (1.1. independent of a coordinate system. In Minkowski space. one must distinguish tensors with upper or lower indices and one can have mixed tensors. Unlike in Euclidean space. x2 . however. denoted as V ′µ = Λµν V ν . linear transformations that do not change the length of a four-vector are called the Lorentz transformations. x) = (x0 .26) (1. given by g00 = −g11 = −g22 = −g33 = 1 (the ˆ ˆ other components are zero).2. • ds2 = 0 (lightlike or null intervals). are . linear transformations that leave the length of a vector invariant are called (homogeneous) Lorentz transformations. δim δjm δkm δin δjn δkn . The Poincar´ transformations include e e Lorentz transformations and translations. ˆ ˆ (1. The invariant distance between two points x and y in Minkowski space is determined from the length dsµ = (x − y)µ . We note that for Euclidean spaces (with a positive deﬁnite metric) vectors and tensors there is only one type of indices.µn = Λµ1ν1 . In special relativity we start with a four-dimensional real vector space E(1.. Useful relations are ǫijk ǫℓmn ǫijk ǫimn ǫijk ǫijl = = = δiℓ δjℓ δkℓ 2 δkl . x3 ) = (t.. • ds2 < 0 (spacelike intervals). The scalar product of two diﬀerent vectors x and y is denoted x · y = xµ y ν gµν = xµ yµ = xµ y µ = x0 y 0 − x · y.3) with basis nµ (µ = ˆ 0.

35) ρ ǫµνρσ ǫµν σ σ ǫµνρσ ǫµνρ ′ ′ = −2 g ρρ g σσ − g ρσ g σρ = −6g σσ . starting from ǫ0123 = 1. Examples are the scalar products above but also relations like pµ = m dxµ /dτ for the momentum four vector. p) = (E. Note that in this equation one has on left. but more important has the same eﬀect on lefthandside and righthandside.32) .30) (Note that there are diﬀerent conventions around and sometimes the opposite sign is used). where m is called the mass of the system. g0 = g1 = g2 = g3 = 1.29) 2 = ∂ µ ∂µ = ∂t2 The value of the antisymmetric tensor ǫµνρσ is determined in the same way as for ǫijk . deﬁned by ∂2 − ∇2 . . constructed via the metric tensor itself.34) (1.28) ν ν Note that gµ with one upper and lower index. ′ ′ .36) Exercises Exercise 1. The length squared of ∂ is the d’Alembertian operator. not aﬀected by Lorentz transformations. because one has (1. Each of these permutations leads to a minus sign.Introduction 6 called covariant. ∂3 ) = ∂ ∂ ∂ ∂ . 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. ∂t (1. (1. . a∞ = 4πǫ0 2 /me e2 . The derivative ∂µ is deﬁned ∂µ = ∂/∂xµ and we have a four vector ∂ with components (∂0 . 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 . ∂2 .33) (1. = −24. is timelike with invariant length (squared) p2 = p · p = pµ pµ = E 2 − p2 = m2 . Relate this quantity to the electron Compton wavelength −e via the dimensionless λ ﬁne structure constant α.31) ǫµνρσ ǫµν ′ ′ ρσ ′ = − (1. = ∂ . (1. gµ = gµρ g ρν and 0 1 2 3 is in essence a ’Kronecker delta’. (1. p).27) It is easy to convince oneself of the nature of the indices in the above equation. for instance. For the contraction of a vector with the antisymmetric tensor one often uses the shorthand notation ǫABCD = ǫµνρσ Aµ Bν Cρ Dσ . (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. 0123 0123 ′ ′ ǫ µνρσ ǫµνρσ The ﬁrst identity. ∂1 . . It is an invariant tensor.1 (a) In the the Hydrogen atom (quantum system) the scale is set by the Bohr radius. ∂t ∂x ∂y ∂z ν ∂µ xν = gµ .

This demonstrates how a careful use of units can save a lot of work. a1 . n2 . ]. Exercise 1. a1 . Use that to give the ˆ ˆµ ˆ components of the basis vectors n+ and n− . a2 . 0) ˆ n0 = (1. (c) Calculate the Compton wavelength of the electron and the quantities under (a) and (b) using the value of c. a2 ] or [a− . a3 ) are the expansion coeﬃcients using the basis vectors n0 . which is the Compton wavelength for the Planck mass. Note that ˆ ˆ ˆ ˆ a · nα = aα and derive from that the components nα of the four vector nα .511 MeV. 1. e. g−− . 1. n+ . re = e2 /4πǫ0 me c2 to the Compton wavelength. 1.Introduction 7 (b) Relate the classical radius of the electron (a relativistic concept).3. 2. n1 . (d) Use the value of the gravitational constant GN / c5 = 6. (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). [Note: what is the MKS unit for V/T?] Exercise 1. a+ . (b) The coordinates (a0 . ν. 0. ˆ ˆ ˆ ˆ These are n1 = (0. ˆ ˆ . . n3 .3 Prove the following relation ǫµνρσ g αβ = ǫανρσ g µβ + ǫµαρσ g νβ + ǫµνασ g ρβ + ǫµνρα g σβ . but only appropriate combinations. 0) ˆ n3 = (0. Also for the coordinates [a− . 0.71 × 10−39 GeV−2 to construct a mass Mpl (Planck mass). Exercise 1. a+ .13 to give its actual value in kg. g+− and g−+ . 1. ρ. Also construct and calculate the Planck length Lpl . One does not need to know . 0. a2 ] we can ﬁnd basis vectors n− . me . ǫ0 . 0. n2 . n1 . 0) ˆ n2 = (0. aT ]) are deﬁned through √ a± ≡ (a0 ± a3 )/ 2. 3} the index α can only be equal to one of the indices in ǫµνρσ .(A · B) C using the properties of the tensor ǫijk given in section 1. 1) ˆ . (a) Express the scalar product a · b in lightcone coordinates and deduce from this the values of g++ . How are a+ and a− related to a+ and a− . Compare it with the proton mass and use Eq. a+ . σ} is a permutation of {0. c. by a simple few-line reasoning [For instance: If {µ. 0. a1 . . α and me c2 = 0.2 Prove the identity A × (B × C) = (A · C) B . .4 Lightcone coordinates for a four vector a (which we will denote with square brackets as [a− . 0.

3 are not covariant.2) Acting on the wave function one ﬁnds for a free particle. t) = exp(−i k 0 t + i k · r).1 The Klein-Gordon equation In this chapter. t) = − ψ(r. one obtains (2. 2M (2. written as pµ −→ i∂µ is covariant (the same in every frame of reference). p).4) where M is the particle mass.3) Equations 2.1) under the substitution (in coordinate representation) E −→ Eop = i ∂ . p2 = pµ pµ = E 2 − p2 = M 2 .5) with (k 0 )2 = k2 + M 2 . i ∂ ∇2 ψ(r. But the replacement 2. The Schr¨dinger equation in quantum mechanics is the operator equation o corresponding to the non-relativistic expression for the energy. Although it is straightforward to ﬁnd the solutions of this equation. Substitution of operators gives the Klein-Gordon (KG) equation for a real or complex function φ. (2. ∂t p −→ pop = −i∇. φk (r. t) = ∂2 − ∇2 + M 2 φ(r. t). (2.6) The energy spectrum is not bounded from below: considering the above stationary plane wave solutions. 8 .2.7) k 0 = ± k2 + M 2 = ±Ek . 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. we just want to play a bit with covariant equations and study their behavior under Lorentz transformations. ∂t2 (2. namely plane waves characterized by a wave number k. ∂t 2M (2. E = p2 .1 and 2. 2 + M 2 φ(r. t) = 0.Chapter 2 Relativistic wave equations 2. Thus a covariant equation can be obtained by starting with the (covariant) equation for the invariant length of the four vector (E. (2.

(2. k) + ei k·x φ(−Ek . 2.15) . it is simply a matter of being careful to ﬁnd a consistent (covariant) theory.9) They satisfy the continuity equation. φ(x) = d3 k (2π)3 2Ek ˜ ˜ e−i k·x φ(Ek . ∂ρ = −∇ · j. there are no longer fundamental objections to mix up space and time.5).e.13) ˜ with (in principle complex) coeﬃcients φ(k). (2. This leads to the existence of negative densities.8) (2. k) ≡ a(k) and φ(−Ek .10) ∂t which follows directly from the Schr¨dinger equation. This operator has all possible solutions in it multiplied with creation (and annihilation) operators. For this we note that an arbitrary solution for the ﬁeld φ can always be written as a superposition of plane waves. It is possible to rewrite it as an integration over positive energies only but this gives two terms (use the result of exercise 2. −k) ≡ b∗ (k) one has φ(x) = d3 k e−i k·x a(k) + ei k·x b∗ (k) = φ+ (x) + φ− (x). for which we need the interpretation of φ itself as an operator. φ(x) = d4 k ˜ 2π δ(k 2 − M 2 ) e−i k·x φ(k) (2π)4 (2. 2. having the KG equation as a space-time symmetric (classical) equation. The integration over k-modes clearly is covariant and restricted to the ‘mass’-shell (as required by Eq. −i φ∗ ∇ φ .2) is j µ = i φ∗ ∂ µ φ or (ρ.3). ∂µ j µ = 0. ↔ ↔ (2. Probability is not positive: in quantum mechanics one has the probability and probability current ρ j = = ψ∗ ψ ↔ i i − (ψ ∗ ∇ψ − (∇ψ ∗ )ψ) ≡ − ψ ∗ ∇ ψ. This continuity equation can be written down o covariantly using the components (ρ. j) of the four-current j. −k) . there are solutions with negative energy. At that point the dependence on position r and time t is just a dependence on numbers/parameters on which the operator depends. 2M 2M (2. The appropriate current corresponding to the KG equation (see Excercise 2.2 Mode expansion of solutions of the KG equation Before quantizing ﬁelds. just as the dependence on time was in ordinary quantum mechanics. is a second order equation and φ and ∂φ/∂t can be ﬁxed arbitrarily at a given time. relativistically the density is not a scalar quantity. which is what Lorentz transformations do.Relativistic wave equations 9 i.11) Therefore. (2π)3 2Ek (2. j) = ↔ i φ∗ ∂0 φ.14) ˜ ˜ Introducing φ(Ek .12) It is easy to see that this current is conserved if φ (and φ∗ ) satisfy the KG equation. however. we want the most general solution. (2. Then. but rather the zero component of a four vector. And. 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. The KG equation.

19) implying that for each solution φ(x) there exists a corresponding solution with the same energy. (2.22) showing that with each solution there is a corresponding charge conjugated solution φC (x) = φ∗ (x). It is trivial to see that ∂µ ∂ µ + M 2 φ∗ (x) = 0. Another symmetry is found by complex conjugating the KG equation.20) one has for the mode amplitudes aP (k) = a(−k) and bP (k) = b(−k). This shows how parity transforms k-modes into −k modes. after restricting the energies to be positive). The choice of a and b∗ allows an easier distinction between the cases that φ is real (a = b) or complex (a and b are independent amplitudes). for which we consider space inversion. changing the sign of the spatial coordinates. 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) . 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. Since a · b = a · ˜ it is easy to see that ˜ b. Performing some Lorentz transformation x → x′ = Λx.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) . 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) .e.16) The consequence of this is discussed in Exercise 2. −x) ≡ (˜µ ). The coeﬃcients a(k) and b∗ (k) are the amplitudes of the two independent solutions (two.Relativistic wave equations 10 In Eqs 2.15 one has elimated k 0 and in both equations k · x = Ek t − k · x.23) . i. (2. which implies (xµ ) = (t. x) → (t. (2π)3 2Ek (2. one thus must have that the function φ → φ′ such that φ′ (x′ ) = φ(x) or φ′ (x) = φ(Λ−1 x).17) (2. (2. φP (x) ≡ φ(˜) (P for parity).14 and 2. (2π)3 2Ek (2. x ∂µ ∂ µ + M 2 φ(˜) = 0. x Transforming everywhere in the KG equation x → x one obtains ˜ ˜ ˜ x ∂µ ∂ µ + M 2 φ(˜) = 0. (2π)3 2Ek (2.18) (2.6 We will explicitly discuss the example of a discrete symmetry. They are referred to as mode and anti-mode amplitudes (or because of their origin positive and negative energy modes).

t) exp(−i k · x) and its time ˜ derivative i∂0 φ(k. ˙ and thus for any normalized solution the full ‘charge’. This shows how charge conjugation transforms ’particle’ modes into ’antiparticle’ modes and vice versa. k).e. For the real ﬁeld one has aC (k) = a(k).4 ˜ Write down the 3-dimensional Fourier transform φ(k. t) = ei Ek t (i∂0 + Ek ) φ(k. ˜ b(k) = ei Ek t i ∂0 φ∗ (−k. d3 k F (Ek . Exercise 2. S V d3 x j 0 (x). t) for the mode expansion in Eq. is conserved. 1 Use ↔ ↔ the following property of delta functions δ (f (x)) = zeros xn 1 δ(x − xn ). QV ≡ satisﬁes ˙ QV = − ds · j. for the mode amplitudes aC (k) = b(k) and bC (k) = a(k).1 Show that for a conserved current (∂µ j µ = 0) the charge in a ﬁnite volume. (2π)3 2Ek where Ek = Exercise 2. 2. letting V → ∞. Exercises Exercise 2.15. |f ′ (xn )| .3 Show that1 d4 k 2π δ(k 2 − M 2 ) θ(k 0 ) F (k 0 . ↔ ↔ Exercise 2. t).Relativistic wave equations 11 i. k) = (2π)4 k2 + M 2 .2 Show that if φ and φ∗ are solutions of the KG equation. t) ≡ d3 x φ(x. Q = 0. t). Use these to show that ˜ ˜ a(k) = ei Ek t i ∂0 φ(k. that j µ = i φ∗ ∂ µ φ is a conserved current Note: A ∂µ B ≡ A∂µ B − (∂µ A)B). Note that a(k) and b(k) are independent of t.

t) = (−i α · ∇ + m β) ψ(r. αj } = 2 δ ij I. B} = AB + BA. and β 2 = I. 1 2 (b) Express now the full charge QV (t) (exercise 2. Exercise 2.6 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). 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). We will encounter these quantities later as energy and momentum.5 (a) As an introduction to parts (b) and (c).Relativistic wave equations 12 Exercise 2. i (a) Show that relativistic invariance. Exercise 2. making sure that each component of ψ satisﬁes the KleinGordon equation (in terms of diﬀerential operators ∂µ ∂ µ +m2 = 0) requires the anticommutation relation2 {αi . Note that a{µ bν} indicates symmetrization. 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. (b) Show that one has current conservation for the current 2 We denote the commutator as [A. 2. We will see a∗ (k) → a† (k) and a(k) → aop (k) with aop and a† op op being annihilation and creation operators (as known from a harmonic oscillator) and the same for the coeﬃcients b(k).1) for a complex scalar ﬁeld current (exercise 2. From this one concludes that α and β must be matrix-valued and ψ must be a multi-component wave function. d3 x (∂ {0 φ)∗ (∂ i} φ). a{µ bν} ≡ aµ bν + aν bµ . ∂ ψ(r. express the quantities E(t) = P i (t) = d3 x (∂0 φ)∗ (∂0 φ) + ∇φ∗ · ∇φ + M 2 φ∗ φ . also using the result in (a).15. {αi . 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 . . in terms of the a(k) and b(k).2) in terms of the a(k) and b(k) using the expansion in Eq. i. j 0 = ψ†ψ and j = ψ † αψ.e. β} = 0.7 To solve the problem with positive and negative energies and get a positive deﬁnite density. ∂t where ψ is a wave function and the nature of α and β is left open for the moment. t). One then ﬁnds that Q is expressed as an inﬁnite sum over number operators a† a (particles) and b† b (anti-particles). (c) Similarly.

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

Next. These generators form a threedimensional Lie algebra SO(3). These matrices can be deﬁned as acting on 2-component spinors (χ → Aχ) or equivalently as acting on 2 × 2 matrices (B → ABA† ). n) = 1 cos ˆ 1 Note ϕ ϕ ˆ + i (σ · n) sin 2 2 = exp i ϕ ˆ σ·n .5) Rotations in general do not commute. Lj ] = i ǫijk Lk . x]] + [z. z ) = lim N →∞ that are generated by i 0 . . which reﬂects itself in the noncommutation of the generators. π ) π -sphere (antipodes identified) 2 π -sphere (surface identified) Figure 3. 2 (3. y]] = 0 (Jacobi identity). Writing a0 = cos(ϕ/2) and a = n sin(ϕ/2) we have1 A = A(ϕ. y ∈ A ⇒ [x. It is straightforward to check that any (ﬁnite) rotation can be obtained from a combination of inﬁnitesimal rotations. the rotations in SO(3) can be generated from inﬁnitesimal rotations that can be expressed in terms of a basis of three generators Lx . (3. It is straightforward to check that the conditions require a0 + i a3 +i a1 + a2 = a0 1 + i a · σ A = +i a1 − a2 a0 − i a3 1 0 + i a1 0 1 + i a2 0 −i + i a3 1 0 (3. x]). π) (n. ˆ therefore. x] = 0 (thus [x.Groups and their representations 14 (n. Ly and Lz . and thus (σ · n)2 = 1. In the same way we can consider rotations around the x. y] ∈ A. z]] + [y. Lx = Ly = 0 0 0 i 0 −i 0 ϕ ˆ .and y-axes 0 0 0 0 0 0 0 −i . y] = −[y. namely that there exists a bilinear product [. for rotations around z for instance. Locally this is a 3-dimensional Euclidean space and SU (2). They satisfy the commutation relations [Li . thus is as the surface i=0 of a sphere in 4 Euclidean dimensions. (3.7) = a0 0 −1 i 0 1 0 0 1 • [x. One way of viewing the parameter space. [y. with real a’s and 3 (ai )2 = 1. is a 3-dimensional Lie-group.1: the parameter spaces of SO(3) (left) and SU (2) (right). ] that satisﬁes • ∀ x. [z. 0 (3.8) ˆ that (σ · a)(σ · b) = a · b + i σ · (a × b). With matrix commutation this algebra satisﬁes the requirements for a Lie algebra.z N N or (Lk )ij = −i ǫijk . • [x. we turn to the group SU (2) of special (det A = 1) unitary (A† = A−1 ) 2 × 2 matrices.6) Summarizing.4) R . ˆ R(ϕ.2 π) (-n. [x.

n) ≈ 1 + i ϕ J · n. with ˆ J ·n ≡ ˆ 1 ∂A(ϕ.2 Representations of symmetry groups The presence of symmetries simpliﬁes the description of a physical system and is at the heart of physics. 2 2 2 (3.15) [g. For instance in a function space translations. sometimes in more than one way (but this will be discussed later). since any ﬁnite transformation can be constructed from the inﬁnitesimal ones. There exists a corresponding mapping of the groups given by µ: SU (2) −→ SO(3) ˆ ˆ A(ϕ.e. immediately sees that the Lie algebras are identical. 2 (3.10) σ ˆ · n. i. i. .12) One. the SU (2) → SO(3) mapping is a 2 : 1 mapping where both A = ±1 are mapped into R = I. also can be considered as a ﬁlled 3-sphere.11) Thus σx /2.14) For a Lie group.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. n) ˆ −→ R(2π − ϕ. n) i ∂ϕ = ϕ=0 (3. however. . In the full parameter space. The inﬁnitesimal generators of SU (2) are ˆ obtained by considering inﬁnitesimal transformations. . The next issue is to ﬁnd the appropriate symmetry operators in the Hilbert space. . . Near the identity. Suppose we have a system described by a Hamiltonian H. φ′ (r) = φ(r + a) = φ(r) + a · ∇ φ + 1 (a · ∇)2 φ + .e. now with radius 2π and with all points at the surface identiﬁed (see ﬁgure). for ﬁxed n. (3. n) π ≤ ϕ ≤ 2π. it is suﬃcient that the generators commute with H. thus. (3.9) The parameter-space of SU (2). . the above mapping corresponds to the trivial mapping of the Lie algebras. H] = 0 for g ∈ G. thus. (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.e. SU (2) ≃ SO(3). n! 2! n=0 ∞ (3. 2! . n) −→ R(ϕ. The existence of symmetries means that there are operators g belonging to a symmetry group G that commute with the Hamiltonian. H] = 0 for g ∈ G.Groups and their representations 15 where we have used the (for operators new!) deﬁnition eA ≡ An 1 = 1 + A + A2 + . They satisfy σi σj σk = i ǫijk . i. 3. ˆ ˆ A(ϕ. in particular that AB → RAB = RA RB ).

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

J = 0 0 + 0 0 0 0 0 0 0 0 √ 2 0 √ 0 . χ) ≡ e−iϕJz e−iθJy e−iχJz . θ.19) (3. This turns out to be the case for all half-integer representations of the Lie algebra. For SU(2). the extension from a local one must be unique. SU (2) is the covering group of SO(3). m′ |UE (ϕ. or equivalently the Hilbert space can be written as a direct product space of spaces V (j) . 2 From the (hermitean) representation Φ(g) of G.e. the conjugate representation is not a new one. ǫz } are 0 0 0 0 0 i 0 Jx = 0 0 −i .g. Jy = 0 0 0 . 1 ≡ ǫ1 ≡ − √ (ǫx + i ǫy ) . Jz = i 0 i 0 −i 0 0 0 −i 0 0 0 0 0 . the conjugate transformation U ∗ acts on χ∗ . . χ) = eim ϕ dm′ m (θ) e−imχ . the topological structure of the group is important. however. If rotations leave H invariant. is not simply connected. all states in the representation space have the same energy. then ǫχ → ǫσχ = ǫσǫ−1 ǫχ = −σ∗ ǫχ. m = Dm′ m (ϕ. For a decent (welldeﬁned) global representation. however. i. one can look at the conjugate representation. Therefore the representation and conjugate representation are equivalent in this case (see Excercise 3.20) By moving through the parameter space the D-functions can be extended to global functions for all allowed angles. For an element in the group G the corresponding point in the parameter space can be reached in two ways. If the group is simply connected. Because there exists a matrix ǫ such that −σ ∗ = ǫ σǫ−1 one immediately sees that appropriate linear combinations of conjugate states. For SO(3). χ)|j. the possibility thus exist that some extensions will not be well-deﬁned representations. θ. For those global representations. 1 |1.3). UE (ϕ. This works for SU (2). dm′ m (θ) ≈ δm′ m − iθ(J 2 )m′ m . however. one obtains the unitary representations Φ(g) = exp(i Φ(g)) of the full group G. Given a representation. (j) (3. Inﬁnitesimally (around the identity) the D-functions for SU (2) and SO(3) are the same. For j = 1 another commonly used representation starts with three Cartesian basis states ǫ related to the previous basis via 1 |1. the simply connected group is called the (universal) covering group. (j) (j) ′ The spin matrices for that Cartesian basis {ǫx .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 . 2 |1. θ. If a transformation U acts on χ. −1 ≡ ǫ−1 ≡ √ (ǫx − i ǫy ) . if χ → σχ. e. any point in the parameter space can be reached in a unique way and any local (inﬁnitesimal) representation can be extended to a global one. For the transformations A in SU (2) and R in SO(3) these are commonly parametrized with Euler angles. 0 ≡ ǫ0 ≡ ǫz . The group SO(3). The jump U → U ∗ implies for the generators σ/2 → −σ ∗ /2. however. contractable and paths that run from a point at the surface to its antipode. ǫy . J− = 2 0 0 0 0 √ 3 0 0 0 0 √ 2 0 0 0 0 . Explicitly. Consider the j = 1/2 representation of SU(2). that is any closed curve in the parameter space can be contracted to a point. j. The matrix elements of these unitary representations are known as D-functions. Of all groups of which the Lie algebras are homeomorphic. There exist two diﬀerent types of paths. to be precise the states ǫχ transform via −σ ∗ .

Groups and their representations 18 3. it is easy to derive some properties of the matrices Λ: (i) det(Λ) = ±1.25) (3. proof: (ΛT GΛ)00 = (G)00 = 1 → Λµ0 gµν Λν0 = 1 → (Λ00 )2 − i (Λi 0 )2 = 1.21) 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. Writing x as a column vector and the metric tensor in matrix form. consisting of the Lorentz group and translations.26) (3.g.23) The proper tensor structure of the matrix element (Λ)µν is a tensor with one upper and one lower index. σ/2 in a two-dimensional (spin 1/2) case. e. proof: det(ΛT GΛ) = det(G) → (det Λ)2 = 1.3 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. explicitly x′µ = (Λ)µν xν + aµ or x′µ = Λµν xν + aµ . 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 (3.22) (Note that xT is a row vector). Denoted in terms of column vectors and 4-dimensional matrices one writes for the poincar´ transe formations x′ = Λ x + a. We considered the generators and looked for representations in ﬁnite dimensional spaces. From this property. e To derive some of the properties of the Lorentz group. (ii) |Λ00 | ≥ 1. In this section we consider the Poincar´ group. (3. which as a matrix equation with (Λ)µν = Λµν and (G)µν = gµν gives (ΛT )ρµ (G)µν (Λ)νσ = (G)ρσ .24) L↑ + L↓ + L↑ − L↓ − proper orthochrone proper non-orthochrone improper orthochrone improper non-orthochrone . G = 10 (3.27) (3. (det Λ = +1 is called proper. det Λ = −1 is called improper). it is convenient to use a vector notation for the points in Minkowski space. 0 0 g00 g01 g02 g03 1 0 g g11 g12 g13 0 −1 0 0 = . Invariance of the length of a vector requires for the Lorentz transformations x′2 = gµν x′µ x′ν = gµν Λµρ xρ Λνσ xσ = x2 = gρσ xρ xσ or Λµρ gµν Λνσ = gρσ . Thus one has the (pseudo-orthogonality) relation ΛT GΛ = G ⇔ GΛT G = Λ−1 ⇔ ΛGΛT = G.

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

We therefore ﬁnd (again) that there are six generators for the Lorentz group. In order to ﬁnd the commutation relations including the translation generators. Of the four parts only L↑ forms a group. three of which only involve spatial coordinates (rotations) and three others involving time components (boosts). we continue using covariance arguments. K = M ). (3. P+ etc.33) These four transformations form the Vierer group (group of Klein).36) (3.35) thus (ω)ij = −(ω)ji and (ω)0i = (ω)i0 .38) Explicitly.34) The condition ΛT GΛ = G yields g ρµ + ω ρµ gρσ (g σν + ω σν ) = gµν =⇒ ωνµ + ωµν = 0. J 3 = M 12 ) and for the (inﬁnitesimal) boosts (along (3. with ϕi = − 1 ǫijk ω jk and J i = 2 z-axis).4 The generators of the Poincar´ group e ↑ The transformations belonging to P+ are denoted as (Λ. This is a + normal subgroup and the factor group is the Vierer group.37) The algebra of the generators of the Lorentz transformations can be obtained by an explicit calculation2 . 2 (M αβ )µν = i g αµ g βν − g α g βµ ν (3. One can easily convince oneself that i (ω)µν = ω µν = − ωαβ (M αβ )µν . Summarizing.g. Multiplying the proper orthochrone transformations with one of them gives all Lorentz transformations. 3. M ρσ ] = −i (g µρ M νσ − g νρ M µσ ) − i (g µσ M ρν − g νσ M ρµ ) . explicitly reading x′µ = = (Λ)µν xν + aµ = ((I)µν + (ω)µν ) xν + ǫµ Λµν xν + aµ = (g µν + ω µν ) xν + ǫµ . ǫ). we have for the (inﬁnitesimal) rotations (around z-axis) ΛR = I + i ϕ3 J 3 = I − i ω12 M 12 . We will just require Poincar´ invariance for the generators themselves. Also this group can be divided into four parts. Thus the two vector operators J and K form under Lorentz transformations an antisymmetric tensor M µν .39) (e. i i0 i i0 1 ijk M jk 2ǫ (3. inf (3.g. We now want to ﬁnd a covariant form of the six generators of the Lorentz transformations.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. a).40) ΛB = I + i η 3 K 3 = I − i ω30 M 30 . The e 2 These commutation relations are the same as those for the ’quantummechanics’ operators i(xµ ∂ ν − xν ∂ µ ) . [M µν . inﬁnitesimally approximated by (I + ω. the Lorentz transformations form a (Lie) group (Λ1 Λ2 and Λ−1 are again Lorentz transformations. 3 30 with η = ω and K = M (e. which are obtained by writing the inﬁnitesimal parameters (ω)µν in terms of six antisymmetric matrices (M αβ )µν . There are six generators. The extension to the Poincar´ group is e ↑ straightforward. (3.

a) = Λ−1 µ P ν . P j ] = i δ ij H/c2 . . J j ] = i ǫijk J k . . (3.42) (to decide on where the inverse is. P ρ ] = 0.45) Note that the commutation relations among the generators M µν in Eq. (3. P j ] = i ǫijk P k . . as part of the set. K j ] = i ǫijk K k . leads to U (Λ. the tensor M µν in terms of boosts cK i = M i0 and rotations J i = 1 ǫijk M jk . writing the generator P = (H/c. other states in that representation are obtained by the action of operators outside the maximal commuting set. because one then sees that by letting c → ∞ the commutation relations of the Galilei group. H] = i P i . known from non-relativistic quantum mechanics are obtained. [K i . To ﬁnd other generators that commute with P µ we look for Lorentz transformations that leave the four vector pµ invariant.38 could have been obtained in the same way. the generators M αβ no longer exclusively act in Minkowski space. (3. The generators of the translations are the (momentum) operators P µ . (3. H] = 0. P ) in terms of the Hamiltonian and the three-momentum operators.Groups and their representations 21 inﬁnitesimal form of any representation of the Poincar´ group (thus including the translations) is of e the form i U (I + ω. the generators J 2 and J 3 in the case of the rotation group. 2 2 (3.5 Representations of the Poincar´ group e where pµ is a set of four arbitrary real numbers. check that U1 U2 corresponds with Λ1 Λ2 ). . [J i . These form a group called the little group associated with that four vector. P µ |p. a)P µ U † (Λ.47) (3. [J i . [P µ . For instance. (1 + iǫγ P γ − i i ωαβ M αβ )P µ (1 − iǫδ P δ + ωρσ M ρσ ) = P µ + ω µν P ν . Λµν pν ⇒ ωµν pν ⇒ ωµν = = = pµ + ω µν pν = pµ 0 ǫµνρσ pρ sσ . In order to label the states in an irreducible representation we construct a maximal set of commuting operators. the generators J± in the case of the rotation group. K j ] = −i ǫijk J k /c2 . Of the generators of the Poincar´ group we choose ﬁrst of all the generators P µ . They just state that M µν transforms as a tensor with two Lorentz indices. [K i . = pµ |p. [J i . .43) giving the following commutation relations by equating the coeﬃcients of ǫµ and ωµν . We have here reinstated c. For instance. a)P µ U (Λ. one obtains 2 [P i .46) [K i . 3. Taking any one of the states in an irreducible representation. Explicitly. a) = Λµν P ν or U † (Λ. The requirement that P µ transforms as a four-vector (for which we know the explicit behavior from the deﬁning four-dimensional representation).48) . = −i (g µρ P ν − g νρ P µ ) . ǫ) = 1 − ωαβ M αβ + i ǫα P α .44) (3. P j ] = [P i . These deﬁne states with speciﬁed quantum numbers that are eigenvalues of these generators. H] = [J i . ν (3.41) 2 At this point. The eigenvalues of these will be the four-momentum pµ of the state. . that commute e among themselves. Inﬁnitesimally. P ν ] [M µν . . 3.

51) The following properties follow from the fact that W µ is a four-vector by construction. satisfying gµν nµ (p)nν (p) = j i nµ (p) pµ i ǫµνρσ pµ nν nρ nσ i j k = = −δij .52) (3. p0 > 0. . These e operators are the Casimir operators of the algebra and can be used to deﬁne the representations of ↑ P+ . From the algebra of the generators one ﬁnds that P 2 = Pµ P µ and W 2 = Wµ W µ (3. 0. . for which we distinguish the following cases p 2 = m2 > 0 p2 = 0 pµ ≡ 0 p 2 = m2 > 0 p2 = 0 p2 < 0 p0 > 0 p0 > 0 p0 < 0 p0 < 0 Only the ﬁrst two cases correspond to physical states. 2 exp (−isµ W µ ) |p. . .56) .54) This will enable us to pick a suitable commuting ’spin’ operator. . Wν ] = −i (gµα Wν − gνα Wµ ) .Groups and their representations 22 where s is an arbitrary vector of which the length and the component along pσ are irrelevant. Wα ] = [Wµ . The elements of the little group thus are U (Λ(p)) = = and U (Λ(p))|p. . i ǫµνρσ W ρ P σ .50) where the Pauli-Lubanski operators W µ are given by 1 Wµ = − ǫµνρσ M νρ P σ . 2 (3.49) (3.53) (3. Pν ] = [Wµ . 0. Massive particles: p2 = M 2 > 0. (3. . = = i exp − ǫµνρσ M µν pρ sσ |p. Given the momentum four vector pµ we choose a tetrade consisting of three orthogonal spacelike unit vectors ni (p). . i exp − ωµν M µν 2 i exp − ǫµνρσ M µν pρ sσ 2 (3.55) commute with all generators and therefore are invariants under Poincar´ transformations. while the others have no physical signiﬁcance. the third case represents the vacuum. M ǫijk . (3. of which the components generate the little group of pµ or (for the third of the following relations) from explicit calculation [Mµν . which can thus be chosen spacelike. .

0). (0. The commutation relations [W i (p).59) (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. . In that case the vectors ni (p) are just the space directions and W = (0. 0. 0. W i (p) = −W · ni (p)). the S i (p) form the generators of an SU (2) subgroup that belongs to the maximal set of commuting operators.e. 0. 1.60) and using the completeness relation nµ (p)nν (p) = − g µν − i i 2 i pµ pν M2 .e. Noting that S i (p) i 2 = 1 W µ (p)W ν (p) M2 niµ (p)niν (p). (1. ms . show that W /M can be interpreted as the intrinsic spin.62) 1 Thus W 2 has the eigenvalues −M 2 s(s + 1) with s = 0. (0. . . n1 (p). Having made a covariant decomposition. M2 (3. W j (p)] = i M ǫijk W k (p). 0). |p3 |). 3. i (3. In this case a set of four independent vectors is chosen starting with a reference frame in which pµ = pµ 0 in the following way: p0 = (p0 . S j (p)] = −i nµ (p)nν (p) j i ǫµνρσ W ρ P σ = i ǫijk S k (p). p0 > 0. 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. 0.. M2 (3. it is suﬃcient to choose a particular frame to investigate the coeﬃcients in Eq. 1. 0. s.63) where E = p2 + M 2 and ms is the z-component of the spin −s ≤ ms ≤ +s (in steps of one). Together with the four momentum states thus can be labeled as |M. 1.64) such that in an arbitrary frame where the four vectors p. M J) with components W i = (M/2)ǫijk M jk . (3.65) . 2 . More generally (fully covariantly) one can proceed by deﬁning S i (p) ≡ W i (p)/M and obtain [S i (p). The explicit construction of the wave function can be done using the D-functions (analogous as the rotation functions). 0.61) one sees that S i (p) i =− W2 . p.58) i. −1) n1 (p0 ) ≡ n2 (p0 ) ≡ s(p0 ) ≡ (3. 1 2 (3.57) (i. 0). (3.Groups and their representations 23 We can write W µ (p) = 3 W i (p) nµ (p) i i=1 (3.57 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.

1 2 24 (3.66) where W · p = 0 implies that W s (p) = −(M 2 /p · s) λ(p).71) which is called the helicity. This state is in physical applications nondegenerate and is denoted by |0 . W 2 (3. (3. The vacuum: pµ = 0. σ · RA a ≡ A(σ · a)A−1 ˆ (valid for any vector a). 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.67) (3. A massless particle. 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 . implying continuous values for the spin (actually for W i = M S i ).72) One has P µ |0 = W µ |0 = 0 and the state is invariant under Lorentz transformations. P µ . Thus 2 λ(p) = J ·p . The algebra of W 1 . λ . i W 2 (p).70) as a commuting set with zero eigenvalues for P 2 and W 2 .1 (a) Derive from the relation that deﬁnes the mapping of SU (2) matrices A into SO(3) matrices RA in Eq. is characterized by its momentum and the helicity. however. Exercises Exercise 3. |p. and the eigenvalues of W 1 and W 2 are set to zero (corresponding to the limit M → 0 for the W µ Wµ eigenvalues of M 2 s(s + 1) → 0). W 2 (p)] = = = i M 2 λ(p).13. W 2 (p)] [λ(p). W 1 (p)] [λ(p). The eigenvalues of W 1 and W 2 (in that case corresponding to the translations) thus can take any continuous values. which can take any integer or half-integer value. and gives [W 1 (p). 3.69) We reproduce the algebra for the massive case with W 3 (p) = M λ(p). 3. W µ (p) = λ(p) P µ . |p| (3.67 has a vanishing right hand side).68) (3. W 2 and λ is derived from the general commutation relations for [Wµ . (3. But if M = 0 one has a diﬀerent algebra (Eq.Groups and their representations The above does actually include the massive case.73) (3. where A = exp(i ϕ · σ/2) = exp(iϕ σ · n/2). Wν ]. Thus we have P 2 . Note that angular momentum does not contribute to helicity as L · p = 0. −i W 1 (p). U (Λ. thus. a)|0 = |0 . This algebra is (for instance) isomorphic to the one generated by rotations and translations in a 2-dimensional Euclidean plane (see exercises). The vector W µ (p) now can be expanded W µ (p) = W 1 (p) nµ (p) + W 2 (p) nµ (p) + λ(p) pµ + W s (p) sµ (p). We don’t know of any physical relevance.

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

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

9) (4. C): Type I (denoted M ): Type II (denoted M ): 27 J= σ σ . 2 1 (J − i K). M = exp i ϕ·σ 2 1 exp ± η · σ 2 = U (ϕ) H(η) . K = +i . 2 2 (4. B j ] = i ǫijk Ak .3) (4. 2 (4. The group SL(2. The matrices in SL(2. j) K = −i J (B = 0). 2 2 σ σ J = .1 The Lorentz group and SL(2. B j ] = [Ai . i ǫijk B k . They will be labeled by two angular momenta corresponding to the A and B groups. respectively. U (ϕ) H(η) (4. 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. similarly as the homeomorphism between SO(3) and SU (2).10) .4) (4. (j.6) (4. Aj ] = [B i .Chapter 4 The Dirac equation 4. It is simply connected and forms the covering group of L↑ .5) = 1 (J + i K). Special cases are the following representations: Type I : (j. With this choice of parameter-spaces the plus and minus signs are actually relevant. becoming the deﬁning representations of SL(2. 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 . (4. 0. C). K = −i . 0) Type II : (0.2) This shows that the Lie algebra of the Lorentz group is identical to that of SU (2) ⊗ SU (2). it also follows directly that the Lorentz group is homeomorphic with the group SL(2. (4.7) From the considerations above. where we restrict (for ﬁxed n) the parameters 0 ≤ ϕ ≤ 2π and 0 ≤ η < ∞. K = i J (A = 0).8) ˆ ˆ ˆ with η = η n and ϕ = ϕn.1) (4. C) can be written as a product of a unitary + matrix U and a hermitean matrix H. This tells us how to ﬁnd the representations of the group. j ′ ).

are not covariant!) ˜ can be used to establish the homeomorphism between L↑ and SL(2. if there does not exist a unitary matrix S. With E = M γ = M cosh(η) and p = M βγ n = M sinh(η) n. ξ −→ Hξ.18 show that ±ǫξ ∗ transforms as a type-II (η) spinor. The following relations for + rotations and boosts are useful. one can use a boost to the frame with momentum ˆ ˆ p. such that M = SM S −1 . however. Also useful is the relation H 2 (p) = σ µ pµ /M = (E + σ · p)/M . 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(2.17) 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. Λ(ǫ) ∈ L+ . i.e. 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).16) = σ µ Λ R aµ . namely H(η) = exp(η · σ/2). One has + −1 Λ(M ∗ ) = I2 Λ(M ) I2 . This is not true for the two-component spinors in SL(2. ˜ Non-equivalence of M and M One might wonder if type I and type II representations are not equivalent. C) or L↑ . thus. U σ µ aµ U ˜ −1 H σ µ aµ H −1 = σ µ ΛB aµ H σ µ aµ H ˜ −1 (4.14) satisfying Tr(σ µ σ ν ) = −2 gµν and Tr(σ µ σ ν ) = 2gµν = 2 δ µν (the matrices. ˜ The sets of four operators deﬁned by deﬁnitions σ µ ≡ (1. H(η) = exp(−η · σ/2).19) . ¯ ¯ H ∗ = ǫ† Hǫ = ǫ−1 Hǫ.15) (4. (ǫ∗ = ǫ. C) and. C). 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. C). (4. σ µ ≡ (1. η → U η. while ±ǫη ∗ transforms as a type-I + spinor. H ≡ (H † )−1 = H −1 ). σ). U ≡ (U † )−1 = U ) U (ϕ) = exp(i ϕ · σ/2) but diﬀerently under hermitean boosts (H † = H. One deﬁnes spinors ξ and η transforming similarly under unitary rotations (U † = U −1 . Note that M ∗ = ǫ† M ǫ has its equivalent for the Lorentz transformations Λ(M ) and Λ(M ∗ ). −σ). ˜ (4. (4. ˜ = σ µ Λ B aµ . cannot be connected by a Lorentz transformation belonging to L↑ . U σ µ aµ U −1 = σ µ ΛR aµ .18) As already discussed for SU (2). (4.The Dirac equation 28 Let us investigate the deﬁning (two-dimensional) representations of SL(2. ǫ−1 = ǫ† = ǫT = −ǫ).12) ξ → U ξ. The Lorentz transformations Λ(M ) and Λ(M ). thus.1). η → Hη.13) 2M (E + M ) (exercise 4. This shows that M ∗ ≃ M . (4. Eqs 4.11) Considering ξ and η as spin states in the rest-frame. there exists the matrix ǫ = iσ 2 . In fact. (4.

(4. a) and ˜ a = (a0 . we have seen that in the rest frame W i (p)/M → J i . For a particle at rest only angular momentum is important and we can choose ξ(0. The representing member of the class P− is the parity or space inversion operator Is . 1 ) and ( 1 . but by a transformation belonging ↑ ↑ to the class P− . but to L↑ .e. ) −→ ( . Although these operators cannot be used to label the representations. as we have seen in section 4. C) are suitable for representing spin 1/2 particles. m) = η(0. (scalar). 0). Within the Lorentz group. i.20) 4. must combine the repe resentations. We also have seen that the representations (0. generators. + − 1 0 0 1 I2 = 0 0 0 0 0 0 −1 0 0 0 0 1 . 0) (via operators J and K) one sees that under parity 2 2 1 1 (0. the aij elements of a tensor will not change sign. For the spin 1/2 representations of the Poincar´ group including parity we. Taking M (Λ) = H(p). where a = (a0 . This provides the easiest way of seeing ˜ what is happening. −→ −p (vector). m) = χm . the well-known two-component spinor for a spin 1/2 particle.g. e. 2 2 (4. that have the correct commutation relations. 2 ) and ( 2 .e. the boost in Eq.22) u= η η M (Λ) 0 where M (Λ) = ǫM ∗ (Λ)ǫ−1 with ǫ given in Eq. It has the same eﬀect as lowering the indices.1. 1 ) and ( 1 . we obtain for the two components of u which we will refer to as chiral right and chiral left components. i. From the deﬁnition of the representations (0. 4. Examples (with between brackets the Euclidean assigments) are r t p H J λ(p) K −→ −λ(p) −→ −K −→ H −→ J −→ t (scalar). they can be connected. −a).2 Spin 1/2 representations of the Lorentz group 1 1 Both representations (0. A parity transformation changes aµ into aµ . Thus M ∗ and M are not equivalent within SL(2. C). and the angular momentum in the rest frame is what we are familiar with as the spin of a particle. 0) of SL(2. −→ −r (vector). but rather the operators W i (p) should be used. they cannot be connected by a unitary 2 2 transformation. m) = χm uL H(p) 0 .23) u(p. consider the four component spinor that transforms under a Lorentz transformation as 0 ξ ξ .13. = (4. 0 χm uR H(p) . (vector). therefore. (pseudoscalar). The behavior is the same for classical quantities. 0) are inequivalent.21) In nature parity turns (often) out to be a good quantum number for elementary particle states. ǫµνρσ will change sign. The i angular momentum operators J are represented by σ i /2. etc.The Dirac equation 29 where The matrix I2 does not belong to L↑ . 4.17. −→ M (Λ) (4. (axial vector).

(4. The explicit realization appearing in Eq.32) suppressing on the RHS the identity matrix in Dirac space.7. To see this consider the equation for the hermitean † † conjugate spinor ψ † (noting that γ0 = γ0 and γi = −γi ). γ } ∂µ ∂ν + M 2 ψ(x) = 0. one must require that for ψ(x) the Klein-Gordon relation 2 + M 2 ψ(x) = 0 is valid for each component separately). As in section 2 we can use Pµ = i∂µ as a representation for the momenta (translation operators) in function space. ψ † −iγ 0 ∂0 +iγ i ∂i −M 1 We ← ← = 0. We ﬁrst want to investigate if the Dirac equation solves the problem with the negative densities and negative energies. It is of a form that we also played with in Exercise 2. γ ν } = 2 g µν . (4. {γ µ . 2 (4. 0 H 2 (p) uR uR . 4. (4.28) which is a covariant (linear) ﬁrst order diﬀerential equation. The general requirements for the γ matrices are thus easily obtained. which in general is a linear equation in pµ . (pµ γ µ − M ) u(p) ≡ (/ − M ) u(p) = 0.25) It is straightforward to eliminate χm and obtain the following constraint on the components of u.27 is known as the Weyl representation. = H −1 (p). From this one obtains the Cliﬀord algebra for the Dirac matrices.The Dirac equation 30 with H(p) = ¯ H(p) = 2M (E + M ) E+M −σ·p 2M (E + M ) E+M +σ·p . This leads to the Dirac equation for ψ(x) = u(p)e−i p·x in coordinate space. We will discuss another explicit realization of this algebra in the next section. Since ∂µ ∂ν is symmetric.29) which is an explicit realization of the (momentum space) Dirac equation. (4. . (4.33) deﬁne for a four vector a the contraction a = aµ γµ / .31) To achieve also that the energy-momentum relation p2 = M 2 is satisﬁed for u(p). p (4. (iγ µ ∂µ − M ) ψ(x) = 0.24) (4. The ﬁrst indeed is solved.26) −2 uL uL H (p) 0 or explicitly in the socalled Weyl representation −M E+σ·p E−σ·p −M uR uL = 0.27) where γ µ are 4 × 4 matrices called the Dirac matrices1 . Applying (iγ µ ∂µ + M ) from the left gives (4. this can be rewritten 1 µ ν {γ . = (4.30) γ µ γ ν ∂µ ∂ν + M 2 ψ(x) = 0.

Multiplying with γ 0 on the right and pulling it through one ﬁnds a covariant equation ψ(x) iγ µ ∂µ +M ← = 0.27). Relying on the Pauli exclusion principle for spin 1/2 particles. The Dirac sea forms the vacuum. . 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.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 . the one of the negative energy states.35) is indeed positive and j 0 can serve as a probability density. From the vacuum a particle can be removed. The density j 0 = ψγ 0 ψ = ψ † ψ (4. From the equations for ψ and ψ one immediately sees that j µ = ψγ µ ψ is a conserved probability current (exercise!). (4. γ ν } = 2 g µν . remains. . Dirac supposed that the negative energy states were already completely ﬁlled and the exclusion principle prevents any more particles to enter the sea of negative energy states. (4.37) Using the explicit form of γ in the Weyl representation (see Eq.36) (4.1: The Dirac sea of negative energy states and antiparticles. (4. This problem was solved by Dirac through the introduction of a negative energy sea. where γ 0 p0 ψ = M ψ 0 −→ Eψ = M γ 0 ψ. opposite charge). We will see how this is properly implemented (also for bosons!) when quantizing ﬁelds.g.3 General representations of γ matrices and Dirac spinors {γ µ . but with properties such that it can be annihilated by the particle (e. γ =ρ ⊗1= γ = iρ ⊗ σ = 0 −1 −σ k 0 2 We (4. 4.39) use ρi ⊗ σj with both ρ and σ being the standard 2 × 2 Pauli matrices. The second problem encountered before. This can most easily be seen in the particle rest frame. one sees that there are two positive and two negative eigenvalues.34) for the adjoint spinor ψ ≡ ψ † γ0 . E = +M (twice) and E = −M (twice).

while L−1 γ µ L = Λµ γ ν . Lorentz invariance The Lorentz transformations can also be written in terms of Dirac matrices. γ µ } = 0 and explicitly one has 0 1 1 . For example. −→ Sψ. Another matrix which is often used is γ5 deﬁned as γ5 = i γ 0 γ 1 γ 2 γ 3 = −i γ0 γ1 γ2 γ3 = It satisﬁes {γ5 .10 are 1 σ3 0 i J 3 = M 12 = = [γ 1 . (γ5 )S. =ρ ⊗1= 0 −1 3 (4.9 and 4.40) (4. γ 0 = ρ1 ⊗ 1 = 1 0 γµ ψ Diﬀerent representations can be related to each other by unitary transformations. 0 (4.47) (4. which in the case of the Weyl representation are just the upper and lower components. 0 γ k = −iρ2 ⊗ σ k = k σ −σ k . γ 0 ]. the rotation and boost generators in Weyl representation in Eqs 4. γ 2 ].R.49) i 1 with −i S ρσ = 2 σ ρσ = − 4 [γ ρ . 4! (4. γ σ ]. (γµ )S. 2 (4.The Dirac equation 32 The Weyl (or chiral) representation: 0 1 .R.R. that project out chiral right/left states. For instance in the Weyl representation (but valid generally). = ρ ⊗ 1 = 1 0 i ǫµνρσ µ ν ρ σ γ γ γ γ . 1 −1 2 For all representations one has † γµ = γ0 γµ γ0 . (4. ν The latter assures Lorentz invariance of the Dirac equation (see Exercise 4. −→ Sγµ S −1 .3) . K 3 = M 30 = 0 iσ 3 2 4 or in general one has the transformation i ψ → Lψ = exp − ωρσ S ρσ ψ.46) (4. S −1 .41) (4. = S(γµ )W. is 1 1 1 (4.44) and an adjoint spinor deﬁned by ψ = ψ † γ0 .48) are projection operators.R. We note that ψ → Lψ and ψ → ψ L−1 .43) S=√ . 0 σ3 2 4 1 −iσ 3 0 = − i [γ 3 .45) (γ5 )W. it is easy to see that PR/L = 1 (1 ± γ5 ) 2 1 0 .42) We note that the explicit matrix taking us from the Weyl representation to the standard representation.

ψ= (4. that brings ψ → ψ c . is a left-handed spinor.52) η ξ one can apply space inversion. x = (t.R.g. In S.58) where ηc is an arbitrary (unobservable) phase. One has ψ c = −ψ T C −1 . T (4. = = 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. (i/ − M ) ψ(x) = 0. (C)W. reverse the charge of the particle.R.53) (4. in Weyl representation we ﬁnd in Dirac space T C ξ −ǫ η ∗ .The Dirac equation 33 Parity There are a number of symmetries in the Dirac equation. ψR .59) −→ ψ c = C ψ = η ǫ ξ∗ .g. T (4.50) Charge conjugation but with ψ p (x) ≡ ηp γ0 ψ(˜) (Exercise 4. This symmetry does not change the spin 1/2 nature. parity.55) (4. for instance. x) → x = (t. but it does.57) (4.R. Note that we have (as expected) explicitly in Weyl x representation in Dirac space P ξ −→ ψ p = γ 0 ψ = η . C is real and one has C −1 = C † = C T = −C c and [C. ∂ (4. As with parity we look for a transformation. −x) and via a few manipulations obtain again the ˜ Dirac equation (i/ − M ) ψ p (x) = 0. (C)S. Explicitly.56) (i/ − M ) ψ c (x) = 0. 0 −ǫ 0 0 −ǫ 0 .54) e. It is easy to convince oneself that if ψ(x) is a solution of the Dirac equation. In any representation a matrix C exist. ψ= (4. such that T Cγµ C −1 = −γµ . = 0 ǫ iσ 2 (4. called charge conjugation. ∂ (4. −iσ 2 0 −ǫ = . ∂ with the solution ψ c (x) = ηc Cψ (x). γ5 ] = 0. Some properties of C are C −1 = C † and C T = −C. 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. which is again a solution of the Dirac equation.) and all representations connected via a real (up to an overall phase) matrix S.R. usually to be taken unity. (or W. The latter implies that the conjugate of a right-handed spinor. e.6).51) The existence of positive and negative energy solutions implies another symmetry in the Dirac equation.

g.. 2M 2M −/ + M p v(p. ψ(x) = u±s (p) e−i p·x .65) (4. s)u(p. The ¯ ¯ solutions are normalized to u(p.69) = − s In order to project out spin states. s) . s ) = 0. be obtained by a Lorentz transformation. s′ ) = 2Ep δss′ .68) (4. the spinors satisfy C uT (p. in which γ 0 is diagonal (see discussion of negative energy states in section 4.61) σ·p −(Ep + M ) There are also two negative energy solutions. ¯ (4. v (p.62) where χs are two independent (s = ±) two-component spinors. s) = v(p. s)v(p. ¯ ¯ † ′ u (p. v(p. where u satisﬁes −σ · p Ep − M u(p) = 0. 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. In that frame is a spacelike unit vector sµ = (0. s).70) = Ps = ˆ 0 1∓σ·s 2 2 . s)u(p. the best representation to describe particles at rest is the standard representation. s) v = . s) + v(k. s) = u(p. ⇔ (/ − M ) u(p) = 0. The explicit Dirac equation in the standard representation reads ∂ iσ · ∇ i ∂t − M (4. the spin polarization vector in the rest frame is the starting ˆ point.63) u(p. s) e−i k·x b(k. 2M 2M (4. s) u = .67) It is straightforward to ﬁnd projection operators for the positive and negative energy states P+ P− = s p /+M u(p. s)v(p.1).The Dirac equation 34 4.66) An arbitrary spin 1/2 ﬁeld can be expanded in the independent solutions. s′ ) = 2M δss′ . s ) = v (p. In an arbitrary frame one has s · p = 0 and s(p) can e. ∂ −iσ · ∇ −i ∂t − M (4.64) (4. s) = Ep + M σ ·p Ep +M χs . Note that the spinors in the negative energy modes (antiparticles) could be two diﬀerent spinors. s). p (4. with E = Ep = + p2 + M 2 . (2π)3 2Ek (4. (4. p −σ · p (Ep − M ) Ep + M σ ·p ¯ Ep +M χs χs ¯ For a free massive particle. s′ ) = −2M δss′ .4 Plane wave solutions Looking for positive energy solutions ∝ exp(−iEt) one ﬁnds two solutions. ¯ ′ ′ Explicit solutions in the standard representation are χs . s) and C v T (p. Choosing χ = −ǫχ∗ (the equivalent spin ¯ 1/2 conjugate representation). It is easy to check that 1 1±σ·s 1 ± γ5 / s ˆ 0 (4. s) ei k·x d∗ (k. where v satisﬁes σ·p −(Ep + M ) v(p) = 0. s)v(p. s)u(p. s ) = v † (p. s)¯(p. s)¯(p. ⇔ (/ + M ) v(p) = 0. s) = u(p. ψ(x) = v ±s (p) ei p·x .60) ψ(x) = 0.

λ = −1/2) = u(p.72) ˆ Helicity states (p along z ) in Standard representation are: √ √ 0 E+M E +M . helicity states. = √ =⇒ 2E E+M + E−M ( E + M + E − M )2 which vanishes in the ultra-relativistic limit E ≫ M or in the massless case. 1 u(p. p3 E +M E+M 1 2 3 p + ip −p 1 2 p − ip 3 1 −p v(p. For instance with the z-axis as spin quantization axis. λ = + 1 ) = √ + E +M − E −M . λ) = u(p. λ = −1/2) = √ − E+M √ 0 0 E+M (4. λ = − 1 ) = √ √ v(p. −p3 −p1 − i p2 1 v(p.73) Also for helicity states C uT (p. −1/2) = √ E + M −(E + M ) 0 .48 replace the spin projection operators which are not deﬁned (by lack of a rest frame). u(p. The chirality projection operators in Eq. The ratio of the components is (directly or inversely) proportional to √ √ E+M − E−M 2M M ≪E M √ √ √ . λ) = v(p. v(p.71) p − i p2 . 2 2 0 √ 2 √ 2 − E+M − E−M 0 . γ5 /ψ = 0 but also p p p /γ5 ψ = −γ5 /ψ = 0. (4. λ = +1/2) = v(p. etc. . while the negative helicity (λ = −1/2) is in essence lefthanded. +1/2) = √ u(p. Explicit examples of spinors are useful to illustrate spin eigenstates. This means that in the solution space for massless fermions the chirality states. +1/2) = √ 0 E +M E+M . In principle massless fermions can be described by twocomponent spinors. λ) and C uT (p. λ = +1/2) = E−M √ 0 − E −M 0 √ √ 0 − E−M − E −M 0 . one has in standard representation: 0 E+M E+M 1 1 0 . We note that for high energy ¯ ¯ the positive helicity (λ = +1/2) is in essence righthanded. √ 0 u(p. (4. u(p. By writing the helicity states in Weyl representation it is easy to project out righthanded (upper components) and lefthanded (lower components). v(p. −1/2) = √ (4. λ = + 2 ) = √ E +M − E −M 2 0√ 2 2 √ 0 E +M + E −M √ √ 0√ √ − E +M − E −M 1 1 E+M − E−M 0 √ . λ = − 1 ) = √ E + M − E − M . Therefore. chirality.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 ). One ﬁnds for the helicity states in Weyl representation: √ √ 0√ E +M + E −M √ 1 √ 1 0√ 1 . λ). 4. p ψR/L ≡ PR/L ψ are independent solutions. Note that for solutions of the massless Dirac equation /ψ = 0.74) .

s)γ µ u(k ′ . s′ )¯i (k ′ . a / (3) γ µ γ ρ γ σ γµ = 4 g ρσ or γ µ //γµ = 4 a · b. s)(γ ν )kl u u s. s). s)¯k (k. u u (4. γν ] (9) σµν γ5 = γ5 σµν = ǫµνρσ γ ρ γ σ /2. ab (4) Tr(γ µ γ ν γ ρ γ σ ) = 4 S µνρσ . s′ )(γ µ )ij uj (k. (9) Tr(γ µ1 · · · γ µ2n ) = Tr(γ µ2n · · · γ µ1 ). s′ )γ µ u(k. (5) Tr(γ µ1 · · · γ µn ) = g µ1 µ2 Tr(γ µ3 · · · γ µn ) . (5) γ µ γ λ γ ν γ ρ γ σ γµ = 2 γ σ γ λ γ ν γ ρ + γ ρ γ ν γ λ γ σ .s′ ∗ (4.76) ul (k ′ . (3) Tr(γ µ γ ν ) = 4 g µν or Tr(//) = 4 a · b. (Use (γ5 )2 = 1 and pull one γ5 through. namely the calculation of the quantity 1 ∗ (¯(k. (7) Tr(γ5 γ µ γ ν ) = 0 or Tr(γ5 //) = 0.g µρ g νσ + g µσ g νρ ). s′ ) ¯ u s. k ′ ) = 2 ′ s. s)¯k (k. s′ )) = u(k ′ . (8) σµν ≡ (i/2)[γµ . For this we ﬁrst have to prove (do this) that (¯(k. Properties of traces of γ-matrices (1) Tr(γ µ1 · · · γ µn ) = 0 if n is odd.) In order to show the use of the above relations. 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′ )) (¯(k. s′ )(γ µ )ij uj (k. (6) Tr(γ5 ) = 0. s)(γ ν )kl ul (k ′ . more convenient ¯ to use the projection operators introduced earlier.) (2) Tr(1) = 4. It is. u ¯ This leads with explicit Dirac indices to Lµν = = = = 1 2 1 2 ui (k ′ . however.s′ 1 ′ (/ + m)li (γ µ )ij (/ + m)jk (γ ν )kl k k 2 1 Tr [(/ ′ + m)γ µ (/ + m)γ ν ] . ab (8) Tr(γ5 γ µ γ ν γ ρ γ σ ) = −4i ǫµνρσ . In principle one can take a representation and just calculate the quantity u(k. (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 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 ).5 γ gymnastics and applications An overview of properties can be found in many text books. (7) γµ γ5 = -γ5 γµ = i ǫµνρσ γ ν γ ρ γ σ /3!. (6) γ5 = i γ 0 γ 1 γ 2 γ 3 = −i ǫµνρσ γ µ γ ν γ ρ γ σ /4!. s)γ µ u(k ′ . s′ )) . we will give one example.75) Lµν (k. k k 2 . etc.The Dirac equation 36 4. ab (4) γ µ γ ν γ ρ γ σ γµ = −2 γ σ γ ρ γ ν . s)γ ν u(k ′ .g µ1 µ3 Tr(γ µ2 γ µ4 · · · γ µn ) + · · · . (Use matrices C to transpose the matrices in the expression under the trace and use Tr AT = Tr A. s)γ µ u(k ′ . s′ ).

e.g.3 Imposing Lorentz invariance on the Dirac equation can be achieved by requiring L−1 γ µ L = Λµ γ ν . ˜ Exercise 4. s). by proving that H 2 (p) = exp (η · σ) = (E + σ · p)/M = σ µ pµ /M. Do this e. x → x.13. 4 2 Exercise 4.4 Show that in P+ ≡ s=±1/2 u(p. of which only the traces of four and two gamma-matrices are nonzero. where x = (t. to the Dirac equation and use this to show that the spinor ψ p (x) = ˜ γ0 ψ(˜). by letting them act on the four basis spinors u(p. γ σ ] = σ ρσ . Show that the equality holds.g. Excercise 4. 2M (E + M ) ˆ ˆ where E = M γ = M cosh(η) and p = M βγ n = M sinh(η) n. Exercise 4. 0 if n is odd. γν } = 2gµν that γµ / γ µ a = Tr(//) = ab µn µ1 Tr(γ · · · γ ) = −2 /. 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 .The Dirac equation 37 The trace is linear and can be split up in parts containing traces with up to four gamma-matrices. a 4 a · b.2 Show that j µ = ψγ µ ψ is a conserved current if ψ en ψ are solutions of the Dirac equation Exercise 4. 4. −x) is also a solution of the Dirac equation. 1 i −i S ρσ = − [γ ρ . ν µ µ Look at the inﬁnitesimal forms Λµ = gν + ων and L = 1 − (i/2)ωρσ S ρσ to obtain ν [γ µ .1 Prove Eq.5 Show starting from the relation {γµ . s) and v(p.6 Apply space-inversion. H(p) = exp η·σ 2 = M +E+σ·p . s)¯(p. x ˜ .77) Exercises Excercise 4. (4. i S ρσ ] = g ρµ γ σ − g σµ γ ρ . s) u p /+M = 2M 2M both sides are projection operators.

Excercise 4. Note: the outcome shows that the decay into muons is much more likely that the decay into electrons (mπ = 139. M R = 1. which is the most important factor needed to ﬁnd the decay width (inverse lifetime). Show that the ratio is |M (π − → e− νe )|2 ¯ ≈ 54.7 (a) Show that (in standard representation) j0 (kr)χm ψ(x.The Dirac equation 38 Exercise 4. (b) Next. namely i / ψ = eiαγ5 ψ. is a stationary solution of the (free) Dirac equation. . (d) Plot for these cases the vector densities ψ † ψ and the scalar densities ψψ as a function of r/R. Do you understand why? The calculation of the life-time is completed in chapter 11. |M (π − → µ− νµ )|2 ¯ Actually if the lepton would be massless. Calculate for now just |M |2 . there is no current ﬂowing through the surface of the cavity. the decay would even be forbidden. 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. We will not pursue this chiral freedom at this point and take α = 0. you need to determine the corresponding conﬁnement condition for ψ.8 × 10−6 . The pion has spin 0.8 A negatively charged pion (π − ) decays in a lepton ℓ (ℓ can be an electron or a muon) and a corresponding anti-neutrino (¯ℓ ). n (c) Apply the conﬁnement condition to ﬁnd the value of k for the lowest eigenmode in the spherical ˆ cavity. i. The conditions for ψ and ψ imply that nµ j µ = ψ/ ψ = 0. 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. A condition for conﬁning the fermion in the cavity is i / ψ = ψ (where n is the surface normal).511 MeV).5 and M → ∞.66 MeV. Calculate the value of k for M R = 0. Show that this n condition is suﬃcient to guarantee that nµ j µ = 0. we want to conﬁne this solution to a spherical cavity with radius R. 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 ). t) = N k ˆ i E+M σ · r j1 (kr)χm −i Et e .e. Note: To prove this. me = 0. r). the lepton and neutrino have spin 1/2 and are ν described by momentum-space spinors. n It turns out that there is actually a whole family of conditions that provide conﬁnement. mµ = 105. For such a cavity nµ = (0.57 MeV. ν p Note: how such parametrizations work will be explained in chapter 8.

These can describe spin 1 states in the rest-frame. A). In an arbitrary frame ǫµ (k) with λ = 0. one has ǫ(λ) (k) ǫ(λ)∗ (k) = −gµν + µ ν kµ kν .Chapter 5 Vector ﬁelds and Maxwell equations 5.4) λ=0. namely ǫµ (k0 ) = (0.2 we have discussed the vectors ǫλ (with λ = 0 and ±1) as basis states of the spin 1 representation. 0). The covariant generalization is to describe them with Vµ (x) = ǫµ (k) e±ik·x . ǫλ ).± 5.1 Fields for spin 1 In section 3. 39 (5.2 The electromagnetic ﬁeld F µν = = ∂ µ Aν − ∂ ν Aµ 1 −E 2 0 −E 1 E 0 −B 3 2 E 3 B 0 3 E −B 2 B 1 Aµ = (φ. (2 + M 2 )Vµ = 0.2) (5. (5.7) .1) The Lorentz condition implies that k µ ǫµ (k) = 0. ±(1) are obtained by boosting the restframe vectors. where k = k0 = (M.± (λ) d3 k ǫ(λ) (k) e−ik·x c(k.5) −E B2 . with ∂µ V µ = 0 (Lorentz condition). −B 1 0 3 With the electromagnetic ﬁeld tensor (5. λ) + ǫ(λ)∗ (k) eik·x c∗ (k.6) where Aµ is the four-vector potential (5. there are then indeed three independent possibilities.3) Since the vectors ǫµ (k) together with the momentum k µ form a complete set of four-vectors. µ (2π)3 2E µ (5. M2 (5. In the restframe. These are solutions of the Klein-Gordon equation with in addition a condition. The vectorﬁeld is therefore (λ) suited to describe spin 1. just as the two-component Dirac spinors did for spin 1/2. A real vector ﬁeld Vµ (x) then can be expanded in modes as Vµ (x) = λ=0. λ) .

0). Therefore if k µ = (|k|. while the fact µ C ∗ that the Minkowski space is real implies that Aµ (x) = Aµ (x) = Aµ (x). It states the absence of magnetic charges and Faraday’s law. if ∂µ Aµ = 0 for large times |t|. One possible choice is the gauge A0 = 0 (5.16) (5. it will vanish at all times.15 the d’Alembertian is replaced by 2 + M 2 . The equations of motion for the ﬁelds Aµ become 2Aµ − ∂µ (∂ν Aν ) = jµ . In that case one also has ∇·A = 0 or ki ǫi (k) = 0. 0. Furthermore. 0. −i.18) corresponding to the helicity states λ = ±1 of the massless photon ﬁeld. (5. As remarked above. shows that the electromagnetic ﬁelds correspond to the case of a massless spin 1 ﬁeld.9) 1 ˜ where F µν ≡ − 2 ǫµνρσ Fρσ is the tensor where E → B and B → −E. (5. only two independent vectors remain ǫ(±) µ (k) = (0.Vector ﬁelds and Maxwell equations and the (conserved) current j µ = (ρ.17) (radiation.13) Hence. Although the Lorenz condition is a constraint. AP (x) = Aµ (˜). it does not eliminate the freedom of gauge transformations but they are now restricted to Aµ → Aµ + ∂µ χ with 2χ = 0. The equation in vacuum.e. ˜ x Note that via parity transformation one obtains another solution. 40 (5.14) of which the solutions give the Li´nard-Wiechert potentials. (5. moree over.10) the electric and magnetic ﬁelds are unchanged. Taking the divergence of this equiation. j) the Maxwell equations read ∂µ F µν = j ν . 5. if in Eq. one ﬁnds for a conserved current and λ = 0 2∂µ Aµ = 0. The gauge freedom can ﬁrst of all be used to impose the Lorentz condition for the electromagnetic ﬁeld.11) This equation is not aﬀected by a gauge transformation. .5. transverse or Coulomb gauge). (5. For electrodynamics one has furthermore the freedom of gauge transformations. in accordance with what we found in section 3. ∓1. (5. Fµν −→ Fµν + (∂µ ∂ν χ − ∂ν ∂µ χ) = Fµν .12) (5. ∂µ Aµ = 0. Note that the incorporation of the Lorentz condition can also be done for the case of the massive vector ﬁeld. The Lorenz condition can actually be incorporated by changing the equations of motion to 2Aµ + (λ − 1)∂µ (∂ν Aν ) = jµ . the antisymmetry of F µν implies that ˜ ∂µ F µν = 0 or ∂λ Fµν + ∂µ Fνλ + ∂ν Fλµ = 0. i. in which case one has 2Aµ = jµ . there is still a gauge freedom left (Aµ → Aµ + ∂µ χ with 2χ = 0). |k|).15) (5.8) (5. 2Aµ = 0. Under a gauge transformation Aµ −→ Aµ + ∂µ χ. for M = 0.

0 . This phase diﬀerence is actually observed. It is however. nothing more than a manisfes1 tation of a nontrivial vacuum (a la the Dirac sea for fermions). This causes an interference pattern as a function of ∆x. Nevertheless the signiﬁcance of Aµ is shown in the Aharonov-Bohm experiment. may look a bit akward. 0). nevertheless. .20) e (∇ × A) · ds = B · ds = where Φ is the ﬂux through the solenoid. To be precise. but where E = B = 0. Inside solenoid: Outside solenoid: B = B z. illustrated in ﬁg. In this experiment an observable phase diﬀerence is measured for electrons moving through ﬁeld-free space (E = B = 0 but Aµ = 0). ˆ B = 0. λ L − λ λ or ∆x = (L− λ/d)δ. 5. F = e E + ev × B (5.Vector ﬁelds and Maxwell equations 41 L ∆x d a (infinitely) long solenoid Figure 5. sin ϕ. . A= By Bx Br ϕ= − ˆ . 2 2 2 BR2 BR2 y BR2 x A= . ϕ= − ˆ 2r 2r2 2r2 where r = (cos ϕ.1. . but A = 0 (hence there is structure in the vacuum). 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 . while the electromagnetic ﬁeld Aµ is not.0 . The situation. The phase diﬀerence between the electrons travelling two diﬀerent paths is given by δ = 2π a a ∆x d = −≈ . cos ϕ. This occurs in the region outside the solenoid.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. The energy density 2 (E 2 + B 2 ) = 0 (hence a vacuum). = exp −i The phase diﬀerence between the two paths of the electron then are given by ∆δ = − = e 1 A · dr + e 2 A · dr = e e A · dr e Φ.1: The Aharonov-Bohm experiment 5. 0) and ϕ = (− sin ϕ. (5.19) are gauge invariant. In the presence of an electromagnetic ﬁeld Aµ an additional phase diﬀerence is observed. where Aµ = 0.

illustrating also that the group structure of the mapping of U (1) into this space is that of the group of integers Indeed when one considers the A-ﬁeld in the example of the Aharonov-Bohm eﬀect outside the solenoid. i. 5. is multivalued as ϕ = 0 and ϕ = 2π are the same point in space. giving no observable phase2 . the number of times they go around the defect) cannot be continuously deformed into one another.e. however. If this would be be happening in empty space one would be in trouble. minimizing the space occupied by B ﬁelds.2: The topology of the space in the Aharonov-Bohm experiment. the relevant charge turns out to be 2e because the electrons appear in Cooper pairs . forming a simple (connected) space. it can be obtained from the situation A = 0 by a gauge transformation.21) (ϕ being the azimuthal angle). Therefore the gauge transformation must be uniquely deﬁned in the space one is working in. Now back to the Aharonov-Bohm experiment. The diﬀerence here is that we are working in a space with a ’defect’ (the inﬁnitely long solenoid). organizing itself in tiny ﬂux tubes (Abrikosov strings).e. has been employed by Dirac in constructing magnetic monopoles in electrodynamics 2 Actually. The AharonovBohm experiment shows a realization of this possibility. The observable phase going around in general is ∆δ = e A · dr = e ∇χ · dr = e χ|ϕ=0 . This situation that A can be written as ∇χ is always true when ∇ × A = 0. implying that it must be single-valued1 . In such a space loops with diﬀerent winding number (i. Another situation in which the topology of space can be used is in the case of superconductors.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. 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. ϕ=2π The function χ in Eq. Therefore space outside the solenoid doesn’t care that χ is multi-valued. 1 The occurrence of nontrivial possibilities. Below the critical temperature the magnetic ﬁeld is squeezed out of the superconducting material. 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. nonobservable phases ϕ = 2π n. Consider a superconductor. A = ∇χ with χ= BR2 ϕ 2 (5.21.

1 Show that Eqs.Vector ﬁelds and Maxwell equations 43 Exercises Exercise 5.8 and 5. ∂t ∇ · B = 0. ∇×B =j+ ∇×E+ ∂E . . 5.9 explicitly give the equations ∇ · E = ρ. ∂t ∂B = 0.

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

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

24) (6. The scalar ﬁeld It is straightforward to derive the equations of motion for a real scalar ﬁeld φ from the lagrangian densities. (6.25) (6.23) (6. leading to the Euler-Lagrange equations. One easily obtains (2 + M 2 )φ(x) = 0. L = 1 1 ∂µ φ∂ µ φ − M 2 φ2 2 2 1 µ = − φ ∂µ ∂ + M 2 φ. the second term is irrelevant.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. the Dirac ﬁeld (spin 1/2) and the vector ﬁeld (spin 1) can be found. 2 ψ = 0. 2 ∗ (6.20) (6.27) (6.21) which diﬀer only by surface terms.22) which can be considered as the sum of the lagrangian densities for two real scalar ﬁelds φ1 and φ2 √ with φ = (φ1 + iφ2 )/ 2.2 Lagrangians for spin 0. For the complex scalar ﬁeld one conventionally uses L = = ∂µ φ∗ ∂ µ φ − M 2 φ∗ φ −φ∗ ∂µ ∂ µ + M 2 φ. (6. δL δ(∂µ ψ) δL δψ one obtains immediately i / −M ∂ → i = − γ µψ 2 i → = ∂ / ψ − M ψ.19) δ(∂µ φ) δφ 6.28) where the second line is not symmetric but in the action only diﬀers from the symmetric version by a surface term (partial integration). 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). The integrand of the ﬁrst term must vanish. ∂ (6. δL δL ∂µ = . (6. leading to (2 + M 2 )φ(x) = 0.29) . (2 + M )φ (x) = 0. Using the variations in ψ (in the symmetric form). 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.

31 is given by LM (Dirac) = −M ξ † η + η † ξ . 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.g.39) 2 This term is written down with η being an anticommuting Grassmann number for which (αβ)∗ = β ∗ α∗ = −α∗ β ∗ .and lefthanded ones. We note that of which the left part coincides . (6. 2 αβ = −βα. namely2 LM (Majorana) = + (6. ψL ≡ (ψL )c = C ψL = 0 (6. The reasons for Grassmann variables will become clear in the next chapter. (6.35) Since the kinetic term in L separates naturally in left and right parts (or ξ and η).36) (6.33) We now introduce a ’real’ (four-component) spinor satisfying Υc = Υ with ψL . 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. but note the factor 1/2 as compared to the Dirac lagrangian. 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∗ . 2 (6. 1 M η † ǫη ∗ − M ∗ η T ǫη . In that case one ﬁnds trivially L = ↔ ↔ 1 1 ψR i / ψR + ψL i / ψL − M (ψR ψL + ψL ψR ).g.38) (6. or equivalently separate the ﬁeld into right.7). Υi / Υ − ∂ 4 2 T −ǫ η ∗ c = ΥR .37) This gives an expression with a mass term that is actually of the same form as the Dirac mass term. Peshkin and Schroeder or Exercise 12. Using the twospinors ξ and η.Classical lagrangian ﬁeld theory 47 and similarly from the variation with respect to ψ ψ i / +M ∂ ← = 0. which comes because we in essence use ’real’ spinors.31) showing e. ∗ 0 ⇒ Υ ≡ −ǫ η ΥL = ψL = η η ∗ for which η0 = ǫ η0 . ∂ ∂ 2 2 (6. the mass term in the Dirac lagrangian 6. L = = ↔ 1 1 c c M ψL ψL + M ∗ ψL ψL ψL i / ψL − ∂ 2 2 ↔ 1 1 M ΥR ΥL + M ∗ ΥL ΥR .32) There exists another possibility to write down a mass term with only one kind of ﬁelds. that the lagrangian separates into two independent parts for M = 0. The Majorana case is in fact more general. . (6.30) It is often useful to link to the two-component spinors ξ and η which we started with in chapter 4.34) (6.

3 Symmetries and conserved (Noether) currents Next.+ R 4 δL δL ∆φ − ∂ ν φ − L g µν δxν δ(∂µ φ) δ(∂µ φ) δL ∆φ − Θµν (x)δxν . These do not aﬀect the dynamics and will give rise to conserved currents. Returning to δS the surface term is rewritten to δS = R d4 x .40) 6. In particular when the lagrangian is invariant under symmetries. In this case.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 .42) Θµν = The variation δS thus can be expressed as δS with J µ (x) = − = δL ∂ ν φ − L g µν . As an example for the classical ﬁeld. discussed in the next chapter. one can consider the variations at the initial or ﬁnal surface. which in a consistent picture precisely generate the symmetries.44) δL ∆φ + Θµν (x)δxν . these conserved quantities become operators. which is the space-integral over the zero-component of a conserved current. (6. + ∂R dσµ dσµ ∂R ≡ where d x . ∆φ = (dφ/dλ)λ=0 δλ and δxµ = (dxµ /dλ)λ=0 δλ.43) F (σ1 ) − F (σ2 ) = dσµ J µ (x) = ∂R R d4 x ∂µ J µ (x). 6. δ(∂µ φ) (6. the presence of a symmetry that leaves the lagrangian invariant.Classical lagrangian ﬁeld theory 48 where Fµν = ∂µ Vν − ∂ν Vµ .47) .45) J µ (x) = − δ(∂µ φ) dλ λ=0 dλ λ=0 Thus considering ∂R = σ2 . t). . we consider the second term in Eq. .σ1 with σ = (R3 . (6. (6. one immediately obtains δL = −∂ µ V ν + ∂ ν V µ − λ g µν (∂ρ V ρ ) = −F µν − λ g µν (∂ρ V ρ ) δ(∂µ Vν ) leading to the equations of motion ∂µ F µν + M 2 V ν + λ ∂ ν (∂ · V ) = 2 + M 2 V ν − (1 − λ)∂ ν (∂ · V ) = 0. we get (omitting the parameter δλ) dxν dφ δL + Θµν (x) .. Q(t1 ) = Q(t2 ). consider U (1) transformations of the Dirac ﬁeld proportional to the charge e. ∂µ J µ (x) = 0. (6.18. δ(∂µ φ) For continuous transformations.. requires the presence of a conserved quantity. ψ(x) → ei eΛ ψ(x) or ∆ψ(x) = i eΛ ψ(x).41) (6. Q= d3 x J 0 (x. implying ∂µ F µν + M 2 V ν = 0 and ∂µ V µ = 0. t). (6. δ(∂µ φ) (6.

From Noether’s theorem one sees that the current Θµν ǫν is conserved. (6. 2 (6.Classical lagrangian ﬁeld theory From the lagrangian for the Dirac ﬁeld.57) For the ﬁelds the behavior under Lorentz transformations has been the subject of previous chapters.3) j µ = i e φ∗ ∂µ φ.g.48) For the complex scalar ﬁeld the U (1) transformations leave the lagrangian invariant and lead to the conserved current (see Exercise 6. ˙ Q = d3 x j 0 (x. (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. we obtain (since δxµ = 0). = −(∂µ φ)δxµ = −ǫµ ∂µ φ(x). = 0. Summarizing. (6.53) These are the energy and momentum. 1 (6. Therefore there are four conserved currents Θµν (ν labeling the currents!) and four conserved quantities Pν = dσµ Θµν = d3 x Θ0ν . Lorentz transformations In this case the transformation of the coordinates and ﬁelds are written as δxµ ∆φi (x) = = ω µν xν . t) is the conserved charge (Q = 0).54) (6.4 Space-time symmetries One kind of symmetries that leave the lagrangian invariant are the Poincar´ transformations. including e the space-time translations and the Lorentz transformations. ↔ (6. e.49) These currents are conserved as discussed already in chapter 2.56) δxµ = ωρσ (aρσ )µν xν 2 with (aρσ )µν = aρσ µν = g ρµ g σν − g σµ g ρν . Translations µ Under translations (generated by Pop ) we have δxµ ∆φ(x) δφ(x) = ǫµ .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. .55) Note that the coordinate transformations can be written in a form similar to that for the ﬁelds (matrices in internal space). In the next section we will see the quantity show up as the charge operator.51) (6. H = d3 x Θ00 (x). (6. For quantized ﬁelds this will become the expressions of the hamiltonian and the momentum operators in terms of the ﬁelds. 6. The integral over the zero-component. (ω µν antisymmetric) 1 ωρσ (−i S ρσ )ij φj (x).50) (6.

Any lagrangian containing derivatives. φ(x).58) (6. The lagrangians which we have considered sofar are invariant under global gauge transformations and corresponding to this there exist a conserved Noether current.3). Deﬁning T µν = Θµν − ∂ρ Gρµν . ∂µ φ(x) + i e ∂µ Λ(x) e (6.67) i. −i eΛ(x) ∗ i eΛ(x) (6. ∂µ T µν (6. where H ρµν is the quantity appearing in M ρµν .63) µν = ∂µ Θ . the angle of the transformation depends on the space-time point x. The current deduced from Noether’s theorem is Jµ = = = = − δL ∆φi + Θµν δxν δ(∂µ φi ) δL 1 ωρσ − (−i S ρσ )ij φj + Θµρ xσ − Θµσ xρ 2 δ(∂µ φi ) 1 ωρσ {H µρσ + Θµρ xσ − Θµσ xρ } 2 1 ωρσ M µρσ .g.62) with Gρµν = (H ρµν + H µνρ + H νµρ )/2.68) (6. is not invariant under local gauge transformations.59) (6.60) Therefore there are six conserved currents M µρσ labeled by ρ and σ (antisymmetric) and corresponding to it there exist conserved quantities M ρσ = d3 x M 0ρσ .3 we have seen global transformations or gauge transformations of the ﬁrst kind. • Vector ﬁeld Aµ : −i Sρσ = aρσ . one has a tensor that satisﬁes (exercise 6. In some applications (speciﬁcally coupling to gravity) it is advantageous to have an equivalent current that is symmetric. however. (6.65) M µρσ = T µρ xσ − T µσ xρ .69) i eΛ(x) φ (x).64) (6.70) . 2 (6.6) T µν = T νµ . (6. 6. • Dirac ﬁeld ψ: −i Sρσ = −(1/4)[γρ . e. in which the U (1) transformation involves an angle e Λ.66) In section 6. In general this tensor is not symmetric.e. independent of x. γσ ] This result for the Dirac ﬁeld has been discussed in chapter 3 (see also Exercise 4. (6. (6.5 (Abelian) gauge theories φ(x) → ei eΛ φ(x).Classical lagrangian ﬁeld theory 50 • Scalar ﬁeld φ: −i Sρσ = 0.61) A ﬁnal note concernes the symmetry properties of Θµν . Gauge transformations of the second kind or local gauge transformations are transformations of the type φ(x) → ei eΛ(x) φ(x). (6. φ(x) φ (x) ∂µ φ(x) ∗ → e → e → ei eΛ(x) φ(x).

∂µ φ) =⇒ L (φ. 2 i ∂ / ψ − M ψ − e/ ψ A 2 giving the Dirac equation in an electromagnetic ﬁeld. (6.75) The ﬁeld φ is used here in a general sense standing for any possible ﬁeld. ∂ 2 4 (6. Replacing derivatives by covariant derivatives and adding the (free) part for the massless vector ﬁelds to the original lagrangian therefore produces a gauge invariant lagrangian. (i/ − M ) ψ = (i/ − e/ − M ) ψ = 0.72) (6. 1 L (φ. The required transformation for Dµ then demands Dµ φ(x) = = (∂µ + i eAµ (x))φ(x) ei eΛ ∂µ + i e(A′ + ∂µ Λ) φ µ → ei eΛ ∂µ φ + i e (∂µ Λ) ei eΛ φ + i eA′ ei eΛ φ µ ≡ ei eΛ (∂µ + i eAµ ) φ. As an example consider the Dirac lagrangian.73) (6. A solution is the one known as minimal substitution in which the derivative ∂µ is replaced by a covariant derivative Dµ which satisﬁes Dµ φ(x) → ei eΛ(x) Dµ φ(x). D ∂ A (6. i L = ψγ µ ∂µ ψ − (∂µ ψ)γ µ ψ − M ψψ.77) involving the interaction of the charge (ρ = j 0 ) and three-current density (j) with the electric potential (φ = A0 ) and the vector potential (A). (6. 4 (6. The equation of motion for the fermion follow from δL δ(∂µ ψ) δL δψ = = i − γ µ ψ. Dµ φ(x) ≡ (∂µ + i eAµ (x))φ(x). 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.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 .Classical lagrangian ﬁeld theory 51 where the last term spoils gauge invariance.78) . (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 . Dµ φ) − Fµν F µν . −e j µ Aµ = −e ρφ + e j · A.74) the behavior which we have encountered before as a gauge freedom for massless vector ﬁelds with the (free) lagrangian density L = −(1/4)Fµν F µν .

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

2me Exercise 6. It shows that for a spin 1/2 particle the g-factor in the coupling of spin to a magnetic ﬁeld takes the value gs = 2. 2M 2M with Enr ≡ E − M ≪ M .6 (optional) Prove the properties of the tensor T µν in section 6.4. Hint ≡ −gs µe · B = −gs Qe s · B. ﬁnally note that it is suﬃcient to show the last equation for dσµ M µρσ . in order to show that Θρσ − Θσρ = ∂µ H µρσ . t) = Enr ψu (r.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 . 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. e 1 (−i∇ + eA)2 + σ · B − eA0 ψu (r.5 Show that the Dirac equation for an electron with charge −e in an external electromagnetic ﬁeld Aµ = (A0 . Comment: this relation is known as the Pauli equation. t). A) (see also exercise 6.Classical lagrangian ﬁeld theory 53 Exercise 6. . This relation is known as the Pauli equation.

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. Quantization of the theory is achieved by imposing canonical commutation relations between q and p.2) H(p.5) ∆ x ∆3 x 54 . q) = ˙ 1 mq 2 − V (q). (7. dq The basic Poisson bracket is actually the one between q and its canonical momentum p. p]P .Chapter 7 Quantization of ﬁelds 7. Further. p]P = 1. q) also considered in the previous chapter. 1 qx (t) = 3 d3 x φ(x. dλ Examples are dO = [O. q(q. ˙ L(q. ˙ 2 (7.3) For variations δO(p. starting from the lagrangian L(q. (7. q) ≈ (dO/dλ)λ=0 δλ under symmetry transformations one has dO = [O. (7. 6. B]P = dB dA dA dB − .1) The Hamiltonian (also corresponding to a conserved quantity because of time translation invariance) is given by p2 + V (q). Q]P . dq dp dq dp (7. t). p)) = ˙ ˙ 2m where the (canonical) momentum p = ∂L/∂ q. An immediate generalization for ﬁelds can be obtained by considering them as coordinates. This is nothing else as representing the classical Poisson algebra as operators in the Hilbert space.1 Canonical quantization We will ﬁrst recall the example of classical mechanics for one coordinate q(t).11) found through Noether’s theorem. however. q) = p q − L(q. H]P dt and dO = [O. in our example p = mq. is the Poisson bracket [A. namely [q. Actually the conserved quantities satisfy a similar algebra as the Lie algebras for continuous symmetries that they are obtained from. The bilinear product for the conserved quantities. there are conserved ˙ ˙ quantities Q (Eq. [q. p] = i. labeled by the position.

The essential conditions to consistently quantize a theory then are the following.8) (7. As an example. The lagrangian is given by L = = x In order to construct the hamiltonian it is necessary to ﬁnd the canonical momenta. a) φi (x)U (Λ. The hamiltonian then is H= x px qx − L = ˙ = x where H (x) = δL ˙ φ(x) − L (x) = Θ00 (x). ∆φ.16) . py (t)] = i δxy or for the ﬁelds φ(x) and Π(x) the socalled equal time commutation relations [φ(x. ∂ qx ˙ ∂ qx ˙ (7. the discretization procedure above is an explicit example. This implies speciﬁc transformation properties in the Hilbert space for the quantum ﬁelds φ. albeit with ’sharp’ functions. Sometimes it is useful to keep in mind that. (7.10) x ∆3 x [Πx qx − L x ] ˙ ∆3 x Hx (7. t)] = 0.9) Note that this indeed corresponds to the zero-zero component (Θ00 ) of the conserved energy-momentum stress tensor Θµν . • Poincar´ invariance e The full variation of the ﬁelds. t)] = i δ 3 (x − y). the ﬁelds φ(x) can be considered as regular operators in the Hilbert space after smearing with a test function f .15) with furthermore the relations [φ(x.7) d3 x L (φ(x). Π(y. for the scalar ﬁeld theory.13) Π(x) H (x) For the quantization procedure we can formulate a number of basic axioms of quantum ﬁeld theory.Quantization of ﬁelds 55 etc. for symmetry transformations. formally.12) (7. ˙ δ φ(x) (7. t)] = [Π(x.6) ∆3 x L x (qx .11) (7. we have L (x) = 1 1 ∂µ φ∂ µ φ − M 2 φ2 2 2 1 1 1 2 (∂0 φ) − (∇φ)2 − M 2 φ2 = 2 2 2 δL = = ∂0 φ. a) = Rj (Λ−1 ) φj (Λ−1 x − a). t). 2 2 2 (7. px (t) = ∂L ∂Lx = ∆3 x = ∆3 x Πx (t). Π(y. ∂µ φ(x)) = ˙ d3 x L (φ. Φ(f ) = d4 x φ(x) f (x). qx+∆x ) . qx . ˙ where Πx (t) is obtained from the continuous ﬁeld Π(x) ≡ δL /δ(∂0 φ). φ(y.14) In fact. t). t). ∇φ) (7. δ∂0 φ) 1 1 1 = Θ00 (x) = (∂0 φ)2 + (∇φ)2 + M 2 φ2 . (7. discussed in the section 6. φ.2. • Canonical commutation relations The quantization of the theory is achieved by imposing the canonical quantization condition [qx (t). i U † (Λ. (7.

17) (Sµν )i φj (x). Pµ ] = [φ (x). sometimes referred to as second quantization. 1 Q = √ (a + a† ) 2ω and P = −i ω (a − a† ) 2 (7.19) 7.22) it is straightforward to check that the commutation relations between Q and P are equivalent with the commutation relations [a. [φ(x). a† ] = 1. i(xµ ∂ν − xν ∂µ )φ (x) + i (7.Quantization of ﬁelds 56 or if U (Λ. j (7. Mµν ] = i i∂µ φi (x). 2 i ωµν (S µν )i .4.23) The hamiltonian in this case can be expressed in the number operator N = a† a. (7. Microscopic causality implies local commutativity. and [N. Q] = [P. [a. [Q. Φ(g)] = 0. The measurements cannot inﬂuence each other or [Φ(f ). ω a† a + 2 2 ω i − [Q. a) = i Rj (Λ−1 ) = 1 + iǫµ P µ − 1− i ωµν M µν . (7. P ] 2 (7. P ] = 0 (7. In simpliﬁed form the hamiltonian is given by H= 1 2 1 2 2 P + ω Q . [N.18) with S µν given in 6. This must be valid for the – now operators – generators P µ and M µν that in the previous chapter have been found in terms of the ﬁelds through Noether’s theorem. a] = [a† . a† ] = 0.25) . • Causality Operators Φ(f ) and Φ(g) for which the test functions are space-like separated can be measured simultaneously (macroscopic causality).24) It is straightforward to ﬁnd the commutation relations between N and a and a† .21) Writing Q and P in terms of creation (a† ) and annihilation (a) operators. P ] = i. a] = −a. (7. 2 2 (7.2 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. j 2 one has [φi (x). H = = ω ω P P Q−i √ Q+i√ 2 2 2ω 2ω 1 1 =ω N+ . φ(y)] = 0 if (x − y)2 < 0. a† ] = a† .20) where the coordinate Q and the momentum P satisfy the canonical commutation relations [Q.

g. (7.e.28) (7. From the normalizations one obtains a |n = √ n + 1 |n + 1 and a|n = n |n − 1 . The canonical momentum becomes −i d3 k ˙ Π(x) = φ(x) = a(k) e−i k·x − a† (k) ei k·x . (2π)3 2Ek (7. a† (k ′ )] = 0.e. a(k ′ )] = [a† (k). which we have split into positive and negative energy pieces with (complex) coeﬃcients a(k) and a∗ (k) multiplying them. The quantization of the ﬁeld is achieved by quantizing the coeﬃcients in the Fourier expansion.32) (7. H = = = d3 x 1 1 1 (∂0 φ)2 + (∇φ)2 + M 2 φ2 2 2 2 d3 k Ek † a (k)a(k) + a(k)a† (k) (2π)3 2Ek 2 d3 k Ek N (k) + Evac . the real scalar ﬁeld φ(x) becomes d3 k (7.30) [φ(x).33) . N |n = n|n one ﬁnds N a† |n N a|n = = (n + 1) a† |n . which represents the ’number of particles’ with momentum k. φ(x) = (2π)3 2Ek where the Fourier coeﬃcients a(k) and a† (k) are now operators. and we see that a state |0 must exist for which N |0 = a|0 = 0. 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.29) It is straightforward to prove that the equal time commutation relations between φ(x) and Π(x′ ) are equivalent with ’harmonic oscillator . (n − 1) a|n † i. [φ(x).4 for the classical ﬁeld) a(k) a† (k ′ ) = = d3 x ei k·x i ∂0 φ(x). is equivalent with [a(k).Quantization of ﬁelds 57 Deﬁning states |n as eigenstates of N with eigenvalue n. and (7.31) The hamiltonian can be rewritten in terms of a number operator N (k) = N (k) = a† (k)a(k).27) 2 (2π)3 It is easy to check that these equations can be inverted (see Exercise 2. Π(x′ )]x0 =x′0 = i δ 3 (x − x′ ) ′ ′ and (7. d3 x φ(x) i ∂0 e−i k ·x ↔ ′ ↔ (7. a† (k ′ )] = (2π)3 2Ek δ 3 (k − k′ ) [a(k). Π(x )]x0 =x′0 = 0. 7. φ(x )]x0 =x′0 = [Π(x). e. Note that we will often write a(k) or a† (k). † √ a and a act as raising and lowering operators.26) a(k) e−i k·x + a† (k) ei k·x . but one needs to realize that in that case k 0 = Ek = k2 + M 2 .3 The real scalar ﬁeld We have expanded the (classical) ﬁeld in plane wave solutions. i.like’ commutation relations between a(k) and a† (k ′ ).

40) This procedure is known as normal ordering. is solved by subtracting it as an (inﬁnite) constant.44) p|φ− (x)|0 = ei p·x . in ﬁeld theory also referred to as vacuum energy Evac = 1 V 2 d3 k Ek . which amounts to redeﬁning H as H= d3 x : H (x) := d3 k Ek N (k). (2π)3 2Ek (7. and multiparticle states |(k1 )n1 (k2 )n2 . . . This term will be adressed below. . assuring that the vacuum (by deﬁnition) has eigenvalue 0!. 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). d3 k |k k|.46) (7.e.34) where V = (2π)3 δ 3 (0) is the space-volume. 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. φ(x) φ+ (x) φ− (x) = = = φ+ (x) + φ− (x).45) (7. d3 k a(k) e−i k·x . 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). k 0 = Ek = k2 + M 2 ). |0 . The rest of the states are then obtained from the groundstate via the creation operator. which now contains an inﬁnite number of oscillators. For the purpose of normal ordering it is convenient to decompose the ﬁeld in positive and negative frequency parts. (2π)3 2Ek (7.37) (7. (2π)3 2Ek (7. .38) d4 k 2π δ(k 2 − M 2 ) θ(k 0 ) |k k| = (2π)4 (7. deﬁning particle states |k = |k (with positive energy.35) where the vacuum contribution disappears because of rotational symmetry.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. |k = a† (k)|0 . = normalized as (a† (k1 ))n1 (a† (k2 ))n2 √ √ .39) The problem with the vacuum or zero-point energy. n1 ! n2 ! (7. a(k)|0 = 0.43) 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). i. (2π)3 (7.42) (7. (2π)3 2Ek d3 k a† (k) ei k·x . (2π)3 2Ek (7. .41) (7. writing all annihilation operators to the right of the creation operators.36) and satisfying the completeness condition 1 = k|k ′ = (2π)3 2Ek δ 3 (k − k′ ).

54) = i∆(x − y) = i (∆+ (x − y) + ∆− (x − y)) . (ii) ∆(x) is a solution of the homogeneous Klein-Gordon equation.51) (7.49) (7.53) t=0 . ∆(x) = − ǫ(x0 ) δ(x2 ).17) and (7.e. i∆+ (x) i∆− (x) = = = d4 k θ(k 0 ) 2π δ(k 2 − M 2 ) e−i k·x . (iv) The equal time commutation relations follow from ∂ ∆(x) ∂t (v) For M = 0. φ(y)] which has the properties (i) i∆(x) = i∆+ (x) + i∆− (x) can be expressed as i∆(x) = where ǫ(k 0 ) = θ(k 0 ) − θ(−k 0 ). (2π)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. (2π)3 2Ek ≡ i∆− (x − y). a† (k ′ )] 3 2E ′ (2π) k [φ− (x). (2π)4 (7. φ− (y)] ≡ i∆+ (x − y). = − (2π)3 2Ek (7.18). The last item to be checked for the scalar ﬁeld are the causality condition. (7.56) = −δ 3 (x).Quantization of ﬁelds 59 In order to ensure the consistency of the theory it is necessary to check that the operators P ν and M obtained from the conserved currents Θρσ and M µρσ are indeed the generators of the Poincar´ e group. In order to calculate [φ(x).48) (7. (iii) ∆(0.50) or as integrals over d4 k. d3 k ei k·(x−y) = −i∆+ (y − x). x) = 0 and hence ∆(x) = 0 for x2 < 0. φ(y)] consider µν [φ+ (x).55) d4 k ǫ(k 0 ) 2π δ(k 2 − M 2 ) e−i k·x . i. (7. d3 k = (2π)3 2Ek = (7. 2π (7.47) ′ d3 k ′ e−i k·x+i k ·y [a(k). that they satisfy the required commutation relations in Eqs (7.52) The result for the invariant commutator function is [φ(x). φ+ (y)] d3 k e−i k·(x−y) .

(2π)3 2Ek d3 k b(k) e−i k·x + a† (k) ei k·x . φ† (y)] = i∆(x − y). extending it with the charge operator and the introduction of particle and antiparticle operators. b† (k ′ )] = (2π)3 2Ek δ 3 (k − k′ ).61) (7.57) (7. (2π)3 2Ek (7. = ∂{µ φ∗ ∂ν} φ − L gµν .59) (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 . a† (k ′ )] = [b(k). particles (created by a† ) and antiparticles (created by b† ) with the same momentum contribute equally to the energy. ↔ (7.4 The complex scalar ﬁeld In spite of the similarity with the case of the real ﬁeld.60) = i φ∗ ∂µ φ. which is equivalent to the relations (only nonzero ones) [a(k). Also the charge operator requires normal ordering (in order to give the vacuum eigenvalue zero). (2π)3 2Ek (7.e. 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) .Quantization of ﬁelds 60 7.62) (7.58) and satisfy the equal time commutation relation (only nonzero ones) [φ(x).5 The Dirac ﬁeld L = ↔ i ψγ µ ∂µ ψ − M ψψ.64) The commutator of φ and φ† is as for the real ﬁeld given by [φ(x). we will consider it as a repetition of the quantization procedure. (2π)3 2Ek (7. 2 From the lagrangian density (7.63) i. (7.66) .65) 7. ∂0 φ† (y)]x0 =y0 = i δ 3 (x − y). 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) .

s)¯(k. ↔ i ψγµ ∂ν ψ − 2 (7. {b(k. s) ei k·(x−y) x0 =y 0 . d† (k ′ . s) e−i k·(x−y) j s † + vi (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. ψj (y)}x0 =y0 = d3 k (2π)3 2Ek ui (k. u v (2π)3 2Ek (7. s)ei k·x . s)u† (k. s)u(k. jµ Θµν = = ψγµ ψ.67) i ψ / ψ − M ψψ gµν . s)e−i k·x .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) : . i. 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)¯(k.75) (7. s)ei k·x + d(k. s′ )} = {d(k.70) where the last line is obtained by using the Dirac equation.74) (7. (2π)3 2Ek (7. s)vj (k. b† (k ′ . s′ )} = (2π)3 2Ek δ 3 (k − k′ ) δss′ .71) (7. (7. ∂ 2 ↔ (7. (2π)3 2Ek d3 k b† (k.73) (7. The solution is the introduction of anticommutation relations.e. s)e−i k·x + d† (k. s). (2π)3 2Ek (7. (7.79) Also for the ﬁeld and the canonical conjugate momentum anticommutation relations are considered. † {ψi (x).77) Note that achieving normal ordering.78) (7. ˙ δ ψ(x) ↔ i Θ00 (x) = − ψγ i ∂i ψ + M ψψ 2 = iψγ 0 ∂0 ψ = iψ † ∂0 ψ. s). interchanging creation and annihilation operators.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. .Quantization of ﬁelds 61 the conserved density and energy-momentum currents are easily obtained. s)v(k.69) (7.

we would have obtained [ψi (x).86) (7.6 The electromagnetic ﬁeld 1 L = − Fµν F µν .82) where i∆1 = i∆+ −i∆− . −λ(∂ρ Aρ ). 7.83) From the lagrangian density the canonical momenta are = = (7. one has † {ψi (x). it implies the Lorentz constraint and leads to the canonical momenta Π0 Π i = = E i. ˙ δ Ai (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. as the vanishing of Π0 induces a constraint.88) . For the scalar combination {ψi (x). 4 2 This gives the equations of motion discussed before.84) (7.Quantization of ﬁelds 62 Using the positive and negative energy projection operators discussed in section 4. (7. When we would have started with commutation relations for the ﬁeld ψ and the canonical momentum.87) If one wants to impose canonical commutation relations ∂µ Aµ = 0 cannot hold as an operator identity.81) where i∆ = i∆+ + i∆− is the same invariant commutator function as encountered before. but we must restrict ourselves to the weaker condition B|∂µ Aµ |A = 0. ψ j (y)] = (i/x + M )ij i∆1 (x − y). 4 Π0 Πi δL = 0.85) which reﬂects the gauge freedom.80) 3 = δ 3 (x − y) δij . It is possible to continue in a covariant way with the lagrangian 1 λ L = − Fµν F µν − (∂µ Aµ )2 . but has the problem of being noncovariant. ∂ (7. ∂ (7. ψ j (y)} at arbitrary times one has {ψi (x). ˙ δ A0 δL = F i0 = E i . which however has wrong causality properties! Therefore the relation between spin and statistics is required to get micro-causality (which is also known as the spin statistics theorem). (7. ψ j (y)} = 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).

where Aµ + + is the part of the vector ﬁeld containing the annihilation operators. ′ (7. c† (k ′ . 6. 0. Πν (y)] = 0. Πν (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. It gives 3 k µ ǫ(λ) (k)c(k. ˙ ˙ .e. rj ]P . The quantized ﬁeld is expanded as Aµ (x) = d3 k (2π)3 2Ek (λ) 3 ǫ(λ) (k) c(k. (7. which reproduce the Hamilton equations. λ).Quantization of ﬁelds 63 for physical states |A and |B . and [ℓi . The canonical equal time commutation relations are [Aµ (x).1 (optional) If you haven’t seen Poisson brackets in classical mechanics you should (a) Proof that Poisson brackets (Eq.90) where Πν = F ν0 − λ g ν0 (∂ρ Aρ ) and we have furthermore [Aµ (x). λ)ei k·x .e. In fact the commutation relations imply ˙ [Aµ (x). containing a time-like photon. and are equivalent with [c(k. 0)c(k. µ λ=0 (7. where Q is the conserved quantity in Eq.3) satisfy the conditions of a Lie algebra. pj ]P . i. 0) . one time-like photon by itself is not allowed! This solves the problems with the normalization and the negative energies. [ℓi . λ) − c† (k. Aν (y)]x0 =y0 = i gµν δ 3 (x − y). λ)e−i k·x + c† (k. 0. i. µ λ=0 (7. j = 1.94) Choosing k µ = (|k 3 |. (c) Calculate the Poisson brackets for the coordinates x and the standard conserved quantities p and ℓ.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.95) Exercises Exercise 7. (7. λ)|A = 0. pj ]P . λ)c(k. (7. (b) Proof for a quantity O(p. q) that dO/dλ = [O. for which it is suﬃcient that ∂µ Aµ |A = 0. These problems are solved by the Lorentz constraint between physical states given above. 2.91) (7. 7. what are [ri .11. a longitudinal photon and two transverse photons.93) which does exhibit problems with the time-like photon. (d) Use Poisson brackets to calculate q and p. Q]P .89) with four independent vectors ǫµ . 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′ )] = −g λλ 2Ek (2π)3 δ 3 (k − k′ ). Aν (y)] = [Πµ (x).

∆hom (Λx) = ∆hom (x) if f is an invariant function. (2 + M 2 ) ∆inhom (x) = −δ 4 (x). Give also the expression for ∆C . (2 + M 2 ) ∆hom (x) = 0.3 (a) Show that the correct implimentation of symmetries in the Hilbert space.Quantization of ﬁelds 64 Exercise 7. Is the subtraction of ’zero point’ contributions essential in this check? µ Pµ φ(x) ei a µ Pµ = φ(x − a) Exercise 7. (e) Show that ∆R (x) = 0 if x0 < 0. . e−i a requires [φ(x). It is invariant. (b) Check the above commutation relation [φ(x). (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. (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. Pµ ] for the (real) scalar ﬁeld using the expressions for the ﬁelds and momentum operator in terms of creation and annihilation operators.2 Prove that the equal time commutation relations between quantum ﬁelds φ(x) and Π(x′ ) are equivalent with the commutation relations between a(k) and a† (k ′ ) Exercise 7. Pµ ] = i∂µ φ(x). f (Λk) = f (k) for Lorentz transformations Λ.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. (Similarly one can show that for the advanced Green’s function ∆A (x) = 0 if x0 > 0). ∆R = 1 (2π)4 d4 k e−ikx k2 1 − M 2 + iǫk 0 (with the implicit prescription to take ǫ → 0 after integration). we can write for ∆R an integral where the k 0 integration remains along the real axis.

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

writing −1 ψ(x) −→ Pop ψ(x)Pop = ηP Aψ(˜) ≡ ψ p (x).Chapter 8 Discrete symmetries In this chapter we discuss the discrete symmetries.1: The behavior of classical quantities under P. 8. ˜ (8. starting with the Dirac equation for ψ(x). Both ψ p and ψ satisfy the Dirac equation. We can determine A. and C quantity t r xµ E p pµ L s ˆ λ = s·p P t -r xµ ≡ xµ ˜ E -p pµ ˜ L s −λ T -t r −˜µ x E -p pµ ˜ -L -s λ C t r xµ E p pµ L s λ 66 .2) where ηP is the intrinsic parity of the ﬁeld and A is a 4 × 4 matrix acting in the spinor space. r) −→ xµ ≡ xµ = (t. T and C for classical quantities is shown in the table 1. T. (iγ µ ∂µ − M ) ψ(x) = 0. parity (P).1) The parity operator transforms We will consider the transformation properties for a fermion ﬁeld ψ(x). The consequences of P. −r). Table 8.1 Parity xµ = (t. time reversal (T) and charge conjugation (C). x (8.

m) = = ˜ u(k. λ) Pop = ηP b(k. x (8. λ) e−ik·˜ + d† (k.6) (check this for the standard representation. ˜ −v(k. λ) eik·x . ˜ Pop d(k. λ)v(k. x µ (iγ ∂µ − M ) γ0 ψ(˜) = 0. (2π)3 2Ek (8. −λ) e−ik·x − d† (k. m) γ0 v(k. m). λ)γ0 v(k. In the same way as the Fermion ﬁeld.4) (8. (i˜ µ ∂µ − M ) ψ(˜) = 0. λ) e−ik·x − d† (k. iγ µ† ∂µ − M ψ(˜) = 0.Discrete symmetries 67 After parity transforming x to x the Dirac equation becomes after some manipulations ˜ ˜ x iγ µ ∂µ − M ψ(˜) = 0. x It is straightforward to apply this to the explicit ﬁeld operator ψ(x) using γ0 u(k. m).2 Charge conjugation We have already seen the particle-antiparticle symmetry with under what we will call charge conjugation the behavior T (8. −λ). −λ). λ) P −1 = −η ∗ d(k.12) and for the vector ﬁeld −1 Aµ (x) −→ Pop Aµ (x) Pop = −Aµ (˜). λ) eik·˜ (2π)3 2Ek d3 k ˜ ˜ ˜ ˜ ηP b(k. one can also consider the scalar ﬁeld and vector ﬁelds.3) (8. λ)u(k. −λ)v(k. The result is ψ p (x) −1 = Pop ψ(x) Pop = ηP γ0 ψ(˜) x (8.5) (8. x Therefore γ0 ψ(˜) is again a solution of the Dirac equation and we have x ψ p (x) = γ0 ψ(˜). λ)γ0 u(k. the above operation reverses the sign of λ). if helicity λ is used instead of the z-component of the spin m.10) i.13) ψ(x) −→ ψ c (x) = ηC C ψ (x). λ)v(k. For the scalar ﬁeld we have seen −1 φ(x) −→ Pop φ(x) Pop = ηP φ(˜). λ)u(k.9) (8. γ x (8.11) (8. . −λ)u(k. 8.8) 3 = λ = λ = λ From this one sees immediately that −1 ˜ Pop b(k. −λ) eik·x 3 2E (2π) ˜ k d3 k ˜ ˜ ηP b(k. x The latter behavior of the vector ﬁeld will be discussed further below. −λ) eik·x (2π)3 2Ek ˜ d3 k ˜ ˜ ˜ ˜ ηP b(k. choosing ηP is real (ηP = ±1) particle and antiparticle have opposite parity.e.7) = λ d k x x ηP b(k. op P (8. −λ) e−ik·x − d† (k.

17) 3 = λ This shows that −1 Cop b(k. Therefore ψ c (x) = = λ −1 ¯ Cop ψ(x) Cop = ηC C ψ T (x) (8. u(k. 8.18) (8.16) d k ηC d(k. For a quantized ﬁeld one has −1 ∗ −1 Top fk (x)bk Top = fk (x)Top bk Top . The (time-reversed) Dirac equation becomes. (i˜ µ∗ ∂µ − M ) ψ(−˜) = 0. λ) eik·x ¯ ¯ (2π)3 2Ek d3 k ηC d(k. to ﬁnd ψ t (−˜) that is a solution of the Dirac equation. λ).14) (8. As time reversal will transform ’bra’ into ’ket’. i. x (8. λ)v(k. Norm conservation requires Top to be anti-unitary2 . writing −1 ψ(x) −→ Top ψ(x)Top = ηT Aψ(−˜) ≡ ψ t (−˜). we start with the complex conjugated x Dirac equation for ψ. m) ¯ = = v(k. One has Aφ|Aψ = φ|ψ ∗ = ψ|A† Aφ = ψ|φ .3 Time reversal xµ = (t.20) The time reversal operator transforms We will again consider the transformation properties for a fermion ﬁeld ψ(x). m). m).22) = Aψ|Aφ ∗ is antilinear if A(λ|φ + µ|ψ ) = λ∗ A|φ + µ∗ A|ψ . λ)C uT (k. ((iγ µ )∗ ∂µ − M ) ψ(x) = 0. r) −→ −˜µ ≡ −xµ = (−t. x 1A 2 An −iC −1 γ µ C∂µ − M ψ(−˜) = 0.15) (where one must be aware of the choice of spinors made in the expansion.e. . λ) e−ik·x + b† (k.19) Cop d(k. λ)C v T (k. Top |φ = φt | = (|φt )∗ .21) where A is a 4 × 4 matrix acting in the spinor space. γ x i(γ5 C)−1 γ µ γ5 C∂µ − M ψ(−˜) = 0. m) ¯ C v T (k. The same relations hold for helicity states. λ)u(k. (8. λ) Cop = ηC d(k.Discrete symmetries 68 the latter being also a solution of the Dirac equation. (8. as discussed in section 4). x µ (iγ ∂µ − M ) γ5 Cψ(−˜) = 0. antilinear operator is anti-unitary if A† = A−1 . λ) e−ik·x + b† (k. λ) −1 Cop = ∗ ηC b(k. The action on the spinors (using C = iγ 2 γ 0 = −iρ1 σ 2 in standard representation) gives C uT (k. it is antilinear1 . x x (8. x iγ µT ∂µ − M ψ(−˜) = 0. x (8. λ). r). λ) eik·x . (2π)3 2Ek (8. ˜ x −(iγ µ )∗ ∂µ − M ψ(−˜) = 0.

λ) i γ5 C v(k. −p. op T (8. λ) e−ik·x (2π)3 2Ek (8. ˜ = 0|b(k)|A 8.23) In the standard representation iγ5 C = σ2 and it is straightforward to apply this to the explicit ﬁeld operator ψ(x) using i γ5 C u(k.4 Bi-linear combinations In quantities such as currents and lagrangians often bilinear combinations of spinor ﬁelds are encountered. λ). there are 16 independent of these bilinear combinations. Therefore one has A|k = A|b† (k)|0 = A|T † T b† (k) T † T |0 ∗ = At |b† (k)|0 ∗ ˜ t = k|At . λ). λ a P |a. λ).34) .29) In table 2 the behavior of particle states under the various transformations has been summarized. λ| a.31) (8. (8. λ) e−ik·x 3 2E (2π) k d3 k ˜ ˜ ηT b(k. λ)v ∗ (k. λ) eik·x + d† (k. λ)v ∗ (k. λ) T −1 = η ∗ d(k. The result is = = λ (8. λ)u∗ (k. λ) eik·˜ + d† (k. λ)u∗ (k. λ)iγ5 Cu(k. −p. λ) eik·x + d† (k.28) (8. λ a |a. a state |a. ˜ Top d(k.33) (8. They are the following S(x) V µ (x) T µν (x) Aµ (x) P (x) ¯ = ψ(x)ψ(x) (scalar) ¯ = ψ(x)γ µ ψ(x) (vector) µν ¯ = ψ(x)σ ψ(x) (tensor) µ ¯ = ψ(x)γ5 γ ψ(x) (axial vector) ¯ = iψ(x)γ5 ψ(x) (pseudoscalar) (8. ∗ ˜ v (k.27) 3 = λ = λ = λ From this one obtains −1 ˜ Top b(k. λ) ψ t (x) −1 Top ψ(x) Top = i ηT γ5 Cψ(−˜) x = = ˜ u∗ (k. −λ |¯. −p.25) (check this for the standard representation).26) d k x x ηT b(k. λ Therefore γ5 Cψ(−˜) is again a solution of the (ordinary) Dirac equation and we can choose (phase is x convention) ψ t (x) = i γ5 C ψ(−˜). λ)v ∗ (k. λ |¯. x (8.2: The transformation properties of physical states for particles (a) and antiparticles (¯).24) (8. Since there are 16 independent 4 × 4 matrices.Discrete symmetries 69 Table 8. λ) eik·x + d† (k. λ). λ) e−ik·˜ (2π)3 2Ek d3 k ˜ ˜ ˜ ˜ ηT b(k. Note that applying an anti-unitary transformation such as Top one must take for the matrix element ˜ the complex conjugate.30) (8. p. p. λ)iγ5 Cv(k. p. λ)u∗ (k. λ) e−ik·x 3 2E (2π) k ˜ d3 k ˜ ˜ ˜ ˜ ηT b(k. −p. λ) Top = ηT b(k.32) (8. λ| ¯ C |¯. −λ a T a. p.

For instance for nucleons one has Γµ (p′ . p′ |V µ (x)|p = u(p′ )Γµ (p′ . p) = γµ F1 (q 2 ) + iσµν q ν F2 (q 2 ). 8. F. γ µ . ¯ ′ ′ (8.41) 2M 2M . γ ν ]. of which the expectation value between momentum eigenstates can be simply found.35) The x-dependence can be accounted for straightforwardly using translation invariance. (8. 2M (8. 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). momenta (pµ or p′µ ) or constant tensors (ǫµνρσ or gµν ). In many applications the expectation values of currents are needed. s′ |V µ (x)|p.36) (8. e.40) where the coeﬃcients are functions of invariants constructed out of the momenta. γ5 γ µ or γ5 ). for the vector current V µ (x). p)u(p) e−i (p−p )·x . s . Applying the results from the previous sections it is straightforward to determine the behavior of the combinations under P.3: The behavior of the independent bilinear combinations of fermi ﬁelds under P. p) can be built from any combination of Dirac matrices (1.37) (8. (8. p′ . (q 2 ). p′ |V µ (0)|p = u(p′ )γ µ u(p). F.Discrete symmetries 70 Table 8.. σ µν . in this case only q 2 = (p′ − p)2 .39) where Γµ (p′ . There are several other terms that also have the correct tensorial behavior such as (pµ + p′µ ) γ5 σ µν qν q µ F3 (q 2 ). C. As an example consider the vector current for a point fermion. (q 2 ). and T P S(˜) x Vµ (˜) x Tµν (˜) x −Aµ (˜) x −P (˜) x C S(x) −V µ (x) −T µν (x) Aµ (x) P (x) T S(−˜) x Vµ (−˜) x −Tµν (−˜) x Aµ (−˜) x −P (−˜) x Θ = PCT S(−x) −V µ (−x) T µν (−x) −Aµ (−x) P (−x) S(x) V µ (x) T µν (x) Aµ (x) P (x) The matrix σ µν ≡ (i/2)[γ µ . ¯ V µ (x) =: ψ(x)γ µ ψ(x) :. The 16 combinations of Dirac matrices appearing above are linearly independent. Note that the lagrangian density L (x) → L (−x) under Θ. called form factors. ¯ In general the expectation value between momentum states can be more complicated. This implies p′ |V µ (x)|p = e−i (p−p )·x p′ |V µ (0)|p .38) (8. and T. C.g.5 Form factors Currents play an important role in ﬁeld theory. V µ (x) = ei Pop ·x V µ (0)e−i Pop ·x . as well as under the combined operation Θ = PCT (see Table 3)..

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

−|q|/2). 0. 0. p′ = (E. σµν q ν F3 (q 2 ) 2M . Exercise 8. it would correspond to an electric dipole moment d = −e F3 (0)/2M (Interaction term d σ · E). (c) Show that if a term of this form would exist. (b) Show that such a term is also not allowed by time-reversal symmetry. |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 . GE GM = F1 − Q2 F2 . (b) Show that in the interaction with the electromagnetic ﬁeld. |q|).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 . p = (E.2 (a) Calculate the current matrix element (using the explicit fermion spinors) in the Breit-frame in which q = (0. 0. 0.3 (a) Show that because of parity conservation no term of the form γ5 can appear in the current matrix element. 0. Exercise 8. 0. L int = ejµ Aµ the charge is given by e GE (0) and the magnetic moment by e GM (0)/2M . 4M 2 = F1 + F2 . (c) Show that hermiticity of the current requires that F1 en F2 are real.

t ′ |q . and the operators are time-dependent. t ′ QH (t′ )|q ′ .1 Time evolution as path integral U (t. H]. |ψH (t) = |ψH . t0 ) = e−i(t−t0 )H or i Two situations can be distinguished: (i) Schr¨dinger picture. t′ ≡ q ′ |q ′ . AH (t) = U −1 (t. t0 )|ψS (t0 ) . QH (t)|q. |ψS (t) = U (t.4) ∂ U (t. t0 ) ∂t (9. t0 ) = H U (t. t . t′ . ≡ 0. (9.1) The time evolution from t0 → t of a quantum mechanical system is generated by the Hamiltonian. in which the operators are time-independent. (9. i ∂ |ψS ∂t ∂ i AS ∂t = H |ψS . AS (t) = AS and the states o are time dependent. t0 ) AH (t0 ) U (t.3) (9. i ∂ |ψH ∂t ∂ i AH ∂t ≡ = 0. in which the states are time-independent.2) (ii) Heisenberg picture. t0 ). [AH . (9. Consider the two (time-independent) Heisenberg states: |q.5) (9. (9. t ≡ q|q.7) 73 .Chapter 9 Path integrals and quantum mechanics 9.6) 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.

2π (9. eA eB = eC with C = A+B+ (9.16) this. use the Campbell-Baker-Hausdorﬀ formula.10) dqn−1 q ′ |e−i H∆τ |qn−1 qn−1 | . 2 12 12 (9. These can be written in coordinate or momentum representation as Q = P = dq |q q q| dq |q −i ∂ ∂q q| = dp |p p p|. Q) expressed in terms of the operators P and Q. which is a hamiltonian function in which the operators are replaced by real-numbered variables.Path integrals and quantum mechanics 74 and Schr¨dinger states o |q |q ′ QS |q ≡ q|q . t and QH (t) = QS and study the evolution of the system by calculating the quantum mechanical overlap amplitude q ′ .11) (9. QS |q ′ ≡ q ′ |q ′ . t′ |q. tj+1 |qj . 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 .. . ′ (9. H(P. ˙ 2π 1 1 1 [A.9) Dividing the interval from t ≡ t0 to t′ ≡ tn into n pieces of length ∆τ and using completeness (at each time ti ) one writes q ′ . 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 . Combining the two terms gives qj+1 . . Then qj+1 . [A.12) where the transformation between coordinate and momentum space involves q|p = ei p. The hamiltonian is an operator H = H(P.qj ) e 2π dpj exp (i ∆τ [pj qj − H(pj . The purpose of this is to calculate the evolution for an inﬁnitesimal time interval. t = = q ′ |e−i n H∆τ |q = q ′ | e−i H∆τ dq1 . (9. t′ |q. . qj ). with the correction1 being of order (∆τ )2 . B]] + [[A. B] + [A.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.. t = q ′ |e−i H(t −t) |q . qj )]) . tj = = = 1 For dpj i pj (qj+1 −qj ) −i H(pj . |q1 q1 |e−i H∆τ |q . tj+1 |qj .qj )∆τ e e 2π dpj i pj (qj+1 −qj )−i ∆τ H(pj . .q .13) At least for a simple hamiltonian such as consisting of a kinetic energy term and a local potential. B] + . B]. 2π (9.8) Choose t as the starting point with |q = |q. n |q (9.14) .

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

n dx fm (x)fn (x) = n = n dαn 1 exp − α2 N 2 n √ 2π ≡ 1.n fm (x) Kmn fn (y) and consider F [α] = F (αn ) as a multivariable function. .27) (9.23) Having deﬁned the Gaussian integral. (9. N (9. y) = ∗ m. Physical answers will usually come out as the ratio of two functional integrals and are thus independent of the chosen measure.22) = x The last equality is obtained by deﬁning the right measure (normalization N ) in the integration.Path integrals and quantum mechanics 76 (ii) Write α in terms of a sum of orthonormal basis functions. Using the second method and expanding in a basis set of functions one obtains for the (real) gaussian functional 1 Dα exp − α2 2 = n dαn N exp − 1 αm αn 2 m. A useful property of functional integration is the translation invariance.26) As an important application of translation invariance we mention the identities 1 Dα exp − α2 − ωα 2 = exp 1 2 ω . Dα F [α] = Dα F [α + β]. 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 ≡ 1.28) Dα Dα∗ exp (−α∗ α − α∗ ω − ω ∗ α) = exp (ω ∗ ω) . Note that procedure (i) is an example of the more general procedure under (ii).25) (see Exercises). 2 (9. the following integrals can be derived for a real symmetric or complex hermitean kernel K. In both cases N is an appropriately choosen normalization constant in such a way that the integral is ﬁnite.24) (9. (9.21) again a multidimensional integral. Consider the gaussian functional as an example. with Dα F [α] = n dαn N F (αn ) (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. α(x) = n αn fn and K(x.

α(y) = δ(x − y).38) . (9. δα(x) or in discretized form δ 1 ∂ ∂ with . which vanish.e. It is now easy to check that Dθ∗ Dθ exp (−θ∗ K θ) = det K.33) (9. however. αn = δmn . products of the same pair appear. −→ δα(x) ∆x ∂αx ∂αx δ ∂ ∂ ∗ with . αy = δxy . we need to consider Grassmann-valued functions involving anticommuting Grassmann variables.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. θ2 = 0. δα(x) δα(x) δ αKβ = dy K(x. δα(x) (9.34) (9. n θn fn (x) dx θ∗ (x) θ(x) .31) δ δ α = dy α(y) = dy δ(x − y) = 1. y) β(y) δα(x) δ 1 1 exp − α2 = −α(x) exp − α2 . In the expansion of the exponential (from second to third line).29) (9. θ∗ ] = exp (−θ∗ θ) ≡ exp − F [θ.32) (9. −→ fn (x) δα(x) ∂αn ∂αm n Note that δ(x − y) = n ∗ fn (x) fn (y). e.30) (9. y) θ(y) . F [θ. (9. Examples of functional diﬀerentiation are (9. the Gaussian-type functionals. In principle the deﬁnitions of functionals is the same.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. θη = −ηθ. θ∗ ] = exp (−θ∗ K θ) ≡ exp − Integration for Grassmann variables is deﬁned as dθ 1 = 0 and dθ θ = 1. dx dy θ∗ (x) K(x.g. (9.35) For applications to fermion ﬁelds.Path integrals and quantum mechanics 77 Functional diﬀerentiation is deﬁned as δ . δα(x) 2 2 δ exp (−α β) = −β(x) exp (−α β) . i.

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

harmonic oscillator.1: Physical picture of vacuum to vacuum amplitude in presence of a source term This is not a surprising result. t|n = q|n e−iEn t with q|n the time-independent wave function. q . which in turn are embedded between an early time T and a future time T ′ . for g. The Heisenberg state |q. t t′ V =0 Dq D dτ HI (q) exp i t t′ t dτ [pq − H0 (p. t) φ∗ (Q) e+i En T . ˙ J (9. the creation of an electron (think of a radio-tube. it is possible to switch on an interaction.66) .65) J dq ′ dq Q′ .e. t Q′ . Considering the source to be present between times t and t′ . one has q ′ . making the electron) and the absorption of an electron (think of a detector). t|n n|Q. We can again rewrite the time-ordered products as functional integrals as derived in the previous section. t = ei Ht |q . q)] .62) p Dq D exp i 2π t dτ [pq − H(p.s. furthermore.5 The generating functional for time ordered products By introducing a source-term. q)] ˙ (9. for example x|n x|n = φn (x) = ei kn x = e −ω 2 x2 /2 (9. φn (q. say. L(q. T with q.64) (9. 9. T . i. q) + J(τ )q] .e. T = n J t′ = = Dq exp i t dτ [L(q. t′ |q. T ′ |q ′ . Before and after these processes there is only the vacuum or ground-state |0 . t′ |q. t |q. t|Q. T = n q. q) → L(q. a set |n of physical eigenstates. T < t < t′ < T ′ . T ′ |Q. for the physical states o q. t = = ′ ′ V = q . t|Q. n (9.63) for plane waves. t |T exp −i p exp −i 2π t′ ′ ′ t′ t dτ HI (Q(τ )) |q. q) + J(t) · q. ˙ The combination of exponentials in the last step is allowed because we are simply dealing with classical quantities (not operators!). t′ q ′ . ˙ ˙ physically pictured as.Path integrals and quantum mechanics source experiment detector 81 T L t L+Jq t’ L T’ Figure 9. t is related to the Schr¨dinger state |q by |q. t q. i. Consider.

68) (9.70) one immediately ﬁnds δ n Z[J] δJ(t1 ) . The importance of Z[J] is that the time ordered product of operators can be obtained from it. in which case eiEn T → e−En ·∞ and the term eiE0 T is dominant.71) 9.Path integrals and quantum mechanics t 82 T t t’ T’ Figure 9. such that t′ → −i∞ t → i∞ ↔ ↔ ˜ t → −∞. t|Q. T′ Z[J] = lim T ′ → −i∞ T → i∞ Dq exp i T dτ L q. Since we have (neglecting multiplicative factors). (9.70) for the generating functional is in fact ill-deﬁned. t′ ) q ′ .2: Analytic continuation of boundaries T and T ′ We can project out the groundstate by an analytic continuation in the time. ˜ t′ → ∞. eiE0 T φ∗ (Q) 0 We deﬁne the generating functional Z[J] as Z[J] = Q′ . . dq dτ + J(τ )q(τ ) . 9. T J lim −i E0 (T ′ −T ) φ (Q′ ) φ∗ (Q) 0 0 T ′ → −i∞ e T → i∞ dq ′ dq φ∗ (q ′ . J (9.6 Euclidean formulation The above expression (Eq. t′ |q. δJ(tn ) hence the name generating functional. = (i)n 0|T Q(t1 ) . T → i ∞ and T ′ → −i ∞. T ′ |Q. . J=0 (9. t) (9.69) The factor that has been divided out in the ﬁrst line of this equation is a numerical factor depending on the boundaries of the space-time volume (T and T ′ ). t 0 0out |0in J . Q(tn )|0 . . T = φ0 (q.67) = = φ0 (q. thus T →i∞ lim q. t). Better is the use ˜ of imaginary time t = −it. . . The generating functional precisely represents the vacuum to vacuum amplitude from initial (’in’) to ﬁnal(’out’) situation in the presence of a source.

1 Convince yourself by choosing an appropriate (orthonormal) set of functions that 1 Dα exp − α K α 2 for real-valued functions.73) d˜ LE q. for Grassmann valued functions. ZE [J] = = where LE q. . . As discussed in the previous section the interesting quantities are obtained from functional diﬀerentiation with respect to the sources. dq dt = 1 2 1 2 dq dt 2 Dq exp Dq exp − ∞ d˜ L q. ˜ ˜ ZE [J] δJ(t1 ) . . (9. e. i dq . .and Grassmann-valued functions. δ n Z[J] 1 Z[J] δJ(t1 ) .75) = − LE q. τ dq ˜ dt ≡ −L q. when L q. (9. where Kp are the eigenvalues of the matrix Kmn in K(x. Exercises Exercise 9.74) − V (q) 2 (9. δJ(tn ) J = 0 ˜ t = it (9.2 Check the examples of functional derivation for real-. dq ˜ dt = 1 2 dq ˜ dt dq ˜ dt 2 − V (q). 1 = √ . δJ(tn ) = (i)n J=0 1 δ n ZE [J] . ensuring convergence for the functional integral ZE [J]. i τ −∞ ∞ −∞ dq d˜ τ + J(˜)q(˜) τ τ dq d˜ τ − J(˜)q(˜) τ τ . + V (q).g. y) = ∗ m. complex. The diﬀerentiations in Z[J] and ZE [J] are related.77) where the expressions have been divided by Z[J] and ZE [J] in order to get rid of dependence on multiplicative factors. .72) (9.76) which is a positive deﬁnite quantity.Path integrals and quantum mechanics 83 In terms of the imaginary time we can write the Euclidean generating functional. ˜ dt (9. In all of these cases the determinant of the (hermitean) operator K is deﬁned as det K ≡ p Kp . det K 1 . Dα Dα∗ exp (−α∗ K α) = for complex-valued functions and Dθ∗ Dθ exp (−θ∗ K θ) = det K.n fm (x) Kmn fn (y). det K Exercise 9.

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

xn ). . (10. δJ(xn ) 85 ≡ G(n) (x1 . . Π) (10. ∂µ φ) + Jφ] (10.1) Dφ exp i σ d4 x L (x) and T′ Z[J] = lim T → i∞ T ′ → −i∞ Dφ exp − Dφ exp i T d4 x [L (φ. φ(x. ∂µ φ) − Jφ] (10. implying at the level of four vectors in coordinate and momentum space (when k · x ≡ kE · xE ) for instance x4 = i x0 = i t x · x = (x0 )2 − d4 xE = i d4 x L = 1 2 3 i 2 i=1 (x ) k 4 = −i k 0 =− 4 µ 2 µ=1 (x ) k · k = (k 0 )2 − d4 kE = −i d4 k 3 i 2 i=1 (k ) =− 4 µ 2 µ=1 (k ) ∂µ φ∂ µ φ − 1 2 M 2 φ2 LE = 1 2 µ (∂E φ)2 + 1 2 M 2 φ2 Furthermore we ﬁnd the n-point Green functions 0|T φ(x1 ) .e. x′ .Chapter 10 Feynman diagrams for scattering amplitudes 10. . .3) The Euclidean formulation is as before. . t = = Dφ DΠ exp i t σ′ ˙ dτ d3 x Π(x)φ(x) − H (φ. . . i. .4) J=0 .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. t) is considered as a set of quantum operators φx (t) in the Heisenberg picture and t′ φ′ .2) = ∞ −∞ d4 xE [LE (φ. x. φ(xn )|0 = (−i)n δ n Z[J] δJ(x1 ) . t′ |φ.

These Green functions appear in the expansion Z[J] = n in n! d4 x1 .4) one distinguishes the retarded Green’s function ∆R (x) = d4 k e−i k·x . (2π)4 (k 0 + iǫ)2 − E 2 (10.9) i. ∆F (the socalled Feynman propagator) is the Green’s function of the Klein-Gordon equation.8) 1 d4 x φ (∂µ ∂ µ + M 2 ) φ − Jφ 2 K (10.12) d4 k −i k·x ˜ e ∆(k) (2π)4 (10. .13) .10) J=0 In order to determine ∆F . with Z0 [0] = 1. . . = 0 2 2 2 k2 − M 2 (k ) − E 2 (k ) − k − M (10. xn ) = x2 xn . . 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. Furthermore. .6) the generating functional is (using ∂µ = − ∂µ ) given by Z0 [J] = Dφ exp −i where Kxy = (∂µ ∂ µ + M 2 ) δ 4 (x − y).5) Z0 [J] for the free scalar ﬁeld For the (free) scalar ﬁeld lagrangian L = ← 1 1 ∂µ φ∂ µ φ − M 2 φ2 . . . consider the general solution of ∂µ ∂ µ + M 2 ∆(x) = −δ 4 (x). As discussed in the previous section (exercise 9. 2 2 → (10. . J(xn ). . . d4 xn G(n) (x1 .e.11) Depending on the path choosen in the complex k 0 plane (see exercise 7.7) where we made the choice Z0 [0] = 1 and ∂µ ∂ µ + M 2 ∆F (x − y) = −δ 4 (x − y). . which can be written as ∆(x) = ˜ with (k 2 − M 2 )∆(k) = 1 or ˜ ∆(k) = 1 1 1 = 0 2 . (10. . (10. xn ) J(x1 ) . (10.3) one ﬁnds 1 Z0 [J] = exp − i 2 d4 x d4 y J(x) ∆F (x − y) J(y) ≡ exp (i W0 [J]) .Feynman diagrams for scattering amplitudes 86 for which we introduce the picture x1 G(n) (x1 .

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 .18) Thus we see various solutions. 10.14) They are solutions of the inhomogeneous equation.19) φ ∂E ∂E − M 2 φ + Jφ 2 1 d4 xE d4 yE J(x) (−i ∆F (x − y)) J(y) . (2π)4 k 2 − M 2 (10.4).20) = exp − 2 where or (see ﬁg. the generating functional can be written 1 µ µ Z0 [J] = Dφ exp d4 xE (10. µ 1 1 • Firstly. The ﬁrst possibility is to consider the well-deﬁned Euclidean formulation. the advanced Green’s function ∆A (x) = satisfying ∆A (x. (10.21) ∆F (x) = = −i 0 kE =−∞ d4 kE e−i kE ·xE 2 (2π)4 kE + M 2 (10. (2π)4 k 2 − M 2 where the latter contour can be deformed to the contour for ∆C .1 for countours) µ (∂E )2 − M 2 (i ∆F (x − y)) = −δ 4 (x − y) 0 kE =∞ (10. in the (well-deﬁned) Euclidean formulation starting with LE = 2 (∂E φ)2 + 2 M 2 φ2 µ 2 2 Jφ with the equation of motion (∂E ) − M φ(x) = −J(x). The solution of the homogeneous equation ∂µ ∂ µ + M 2 ∆(x) = 0 can also be written as an integral in k-space. 4 k 2 − M 2 + iǫ (2π) (10.17) d4 k e−i k·x . i. the second is to explicitly consider 0|T φ(x1 ) φ(x2 )|0 .23) k0 =i∞ k0 =−i∞ d4 k e−i k·x . .e.15) e−i k·x d4 k . 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 are also shown in exercise (7.16) C where C is a closed contour in the complex k 0 -plane.22) (10. The contours for i ∆(x) i ∆+ (x) i ∆− (x) = [φ(x). It is straightforward (closing contours in the appropriate half of the complex plane) to prove for instance that i ∆C (x) = θ(x0 ) i ∆+ (x) − θ(−x0 ) i ∆− (x).Feynman diagrams for scattering amplitudes 87 satisfying ∆R (x. φ− (0)] = (2π)3 2E d3 k e−i k·x = [φ− (x). t) = 0 for t < 0. (2π)4 (k 0 − iǫ)2 − E 2 (10. ∆F = ∆C . ∆(x) = − d4 k e−i k·x . (2π)4 k 2 − M 2 (10. φ(0)] = i∆+ (x) + i∆− (x) d3 k e−i k·x = [φ+ (x). (10. t) = 0 for t > 0.

1: Contour in the k 0 plane for the Feynman propagator in Euclidean and Minkowski space • As the second possibility. or diagrammatically (4) (10. because this is the only quantity entering Z0 [J]. We ﬁnd (see also section 7.3) 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). δJ(x4 ) 2 J=0 = − [∆F (x1 − x2 ) ∆F (x3 − x4 ) + ∆F (x1 − x3 ) ∆F (x2 − x4 ) + ∆F (x1 − x4 ) ∆F (x2 − x3 )] . Neglecting multiplicative factors or equivalently assuming that Z0 [0] = 1. x2 . φ+ (y)]|0 = θ(x0 − y 0 ) i ∆+ (x − y) − θ(y 0 − x0 ) i ∆− (x − y) (10. the same result as above. .24) Knowing the explicit form of Z0 [J] it is straightforward to calculate the 4-points Green’s function. (4) (2) It should be clear that G0 can be expressed in terms of G0 = i ∆F . 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. = θ(x0 − y 0 ) 0|[φ+ (x).25) x1 x2 x3 = x4 + + . φ− (y)]|0 − θ(y 0 − x0 ) 0|[φ− (x). we have G0 (x1 . . x4 ) = 0|T φ(x1 ) φ(x2 ) φ(x3 ) φ(x4 )|0 1 δ4 exp − i J ∆F J = (−i)4 δJ(x1 ) .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 .

35) z . x2 ) = G0 (x1 .27) When interactions are present.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.29) This expression will be the one from which Feynman rules will be derived.30) and in zeroth order G(0. (10. L (φ) = L0 (φ) + LI (φ).31) G(2.Feynman diagrams for scattering amplitudes 89 10. x3 . x3 . To zeroth order in the coupling one has 0 1 Z (g ) [J] = Z0 [J] = exp − i 2 J ∆F J .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. with propagators (i ∆F ) being connected to vertices (i LI ) according to the above expression for the functional Z[J]. (10. Z[J] = = Dφ exp i exp i d4 z LI (10. x4 ) = To ﬁrst order in g one has Z (g ) [J] = = 1 0 0 x1 + + x2 (10.28) 1 exp − i 2 = exp 1 2 d4 x d4 y exp i δ δ i ∆F (x − y) δφ(x) δφ(y) d4 z [LI (φ) + J(z)φ(z)] φ=0 .33) (4) i −i d4 z LI g 4! d4 z 1 δ i δJ(z) 1 exp − i 2 J ∆F J 2 −3 ∆2 (0) + 6i ∆F (0) F + d4 x ∆F (z − x) J(x) 4 d4 x ∆F (z − x) J(x) 1 exp − i 2 J ∆F J .g ) (x1 . x2 . (2) (0) (10. i.32) (10. (10. x2 .g ) (x1 .e. ≡ −i g d4 z (10. Consider as an example the interaction LI (φ) = − g 4 φ 4! in the scalar ﬁeld theory discussed sofar. x2 ) = i ∆F (x1 − x2 ) = G(4.34) Introducing a vertex point to which four propagators are connected. x4 ) = G0 (x1 .g 0 ) = G0 = 1.

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

(iv) If δZ[J] = i 0|φ(x)|0 = 0... 1 2! Z[0] = 1 + + + . but is easily illustrated by writing down the ﬁrst few terms... (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. These can be divided out. In the expansion of Z[J]/Z[0] one (by deﬁnition) has the socalled source (n) connected Green’s functions Gsc .. If i W [J] = then exp(i W [J]) = 1 + i W [J] + = 1 + 1 2! +i d4 x1 i2 + d4 x1 d4 x2 2! i2 4 4 d x1 d x2 + 2! + + (i W [J])2 (i W [J])3 + + . Notes: (n) (i) In G(n) the connected Green functions Gc appear with particular combinatorial factors (accounted for in the deﬁnition of Feynman rules to be discussed later). in the expansion of which all vacuum blobs are contained. δJ(x) J=0 .. + +i d4 x1 1 2! + 1 3! +i d4 x1 + 1 2! i2 2! d4 x1 d4 x2 + . 2! 3! +i d4 x1 + i2 2! d4 x1 d4 x2 . (ii) Note that Z[0] = exp(iW [0]) appears as a multiplicative factor.

. in = p. in = |p1 . 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 . Next.g.45) x2 as the only surviving diagram (see exercises). Proof: α. in = |α. (2) (10. e. in|SS † |α. translation invariance). in into ﬁnal state particle states (out-states) |β. p. x2 . In line with the consideration of the generating functional representing the vacuum to vacuum amplitude we consider ﬁelds φin . (10. out (suppressing except momenta all other 1 m quantum numbers). out| and S † |α. . pn . (3) S is unitary (it conserves the scalar product from initial to ﬁnal state). in = S|α.3 Interactions and the S-matrix The S-matrix The S-matrix transforms initial state free particle states (in-states) |α.47) . (10. φout and the interpolating ﬁeld φ(x). For the free scalar theory. in = p. p′ . in| (choose ϕ0 = 0). . in|S|p′ . out|α. (10. the vacuum expectation value of the ﬁeld φ(x) is zero.Feynman diagrams for scattering amplitudes 92 i. one ﬁnds for the connected 4-point Green’s function at order g. x3 . in|p′ . considered explicitly. out = |p′ .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 ). x4 10. Sβα ≡ β. out|p′ . implying the absence of tadpoles. out = p|p′ . out . out| = eiϕ0 0. . out|p′ . in = β. Also for free ﬁeld (Heisenberg) operators a distinction is made between φin and φout . in = p. we have W0 [J] = − 1 2 d4 x d4 y J(x) ∆F (x − y) J(y). x1 G(4. in|S = 0. out . this will be translated to the action on ﬁelds. out|α. in ⇔ β. ∂µ ∂ µ + M 2 φin (x) = 0. thus β. out = δαβ ↔ SS † = 1. Proof: 0.44) J=0 For the interacting theory. then (n) (n) Gsc = Gc for n ≤ 3. (2) The one-particle state is invariant (conservation of energy and momentum. in|S = α. x4 ) = c 1 x3 z (10. . out| ⇔ |α.46) The properties of the S-matrix are (1) The vacuum is invariant or |S00 | = 1. .e. in|S = β.g ) (x1 .

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

S U [J]] = √ Z δ S U [J].63) . B]eB . B] is a c-number).57) i [φin (x). Thus S U [J] =: exp 1 √ Z d4 x d4 y φin (x) K(x. Thus applied to the ﬁeld φ(x) = φ+ (x) + φ− (x).56) Since δU [J]/δJ(x) satisﬁes the same equations as φ(x) and we know the limits. 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.62) where F [J] is some (arbitrary) functional. provided [A. which is equal to the expression e in which the creation operators are placed left of annihilation operators. δJ(x) ∞ or for U [J] ≡ U−∞ [J] ′ (10. : =i d4 y ∆(x − y) f (y) : e d4 z φ(z) f (z) :. B] and [A. φ(x). φ− f (10. B] (for the case that [A.59) (10. we can. z) δU [J] δJ(z) δU [J] d4 y d4 z ∆A (x − y) K(y.Feynman diagrams for scattering amplitudes 94 and satisﬁes the property T ′ δUT [J] t = i UtT [J] φ(x)UT [J]. e e ] = [A. δJ(z) (10.60) e d4 y φ+ (y) f (y) d4 z φ− (z) f (z) d4 z φ+ (z) f (z) d4 y [φ(x). [A.e. t i Ut∞ [J] φ(x)U−∞ [J] (10. φ(y)] f (y) e e . Noting that 0| : eA : |0 = 1 it follows that F [J] = 0|S U [J]|0 = 0|U [J]|0 = 0out |0in ∝ Z[J]. (10. just as we did for φ(x). one has [A.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).58) In order to ﬁnd a solution to this equation note that. z) δU [J] . δU [J] δJ(x) = = √ i Z U [J] φin (x) − √ i Z φout (x) U [J] − d4 y d4 z ∆R (x − y) K(y. φ(x). eB ] = [A. (10. with the use of the Baker-Campbell-Hausdorﬀ formula e−B A e+B = A + [A. express it in terms of advanced and retarded Green’s functions. y) δ δJ(y) : F [J]. δJ(z) (10. z) d4 y d4 z ∆(x − y)K(y. B + C]e e . C] are c-numbers. z) . : e d4 z φ(z) f (z) d4 x φ− (x) f (x) B C B C (10.61) e φ+ f where the normal ordered expression : e φ f : is used. e = i.

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

If problems arise go back to the deﬁning expression for the generating function in 10.Feynman diagrams for scattering amplitudes 96 (in fact the inverse of the operator found in the quadratic term is for real scalar ﬁelds also multiplied by a factor 2. 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. At these vertices each line can be assigned a momentum. Note that in calculating amputated Green’s functions external lines are neglected.69) LI (φ) = − φ3 − φ4 . Veltman. Diagrammar) in the case of the interaction terms g f (10. or calculating full Green’s functions external lines are treated as propagators. For the symmetry factor consider the examples (given in G. Corresponding to external particles wave functions are introduced k k 1 1 In order to calculate the connected amplitude −i Mf i appearing in the S-matrix element these ingredients are combined using Eqs 10.27. but overall momentum conservation at a vertex is understood. = −i g . 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. ’t Hooft and M. (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). the factor 2 corresponds to the two-points Green’s function having two identical ends). = −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. to be precise iLI vertices in momentum space are introduced. which cancels the factor 1/2 in the quadratic piece. For the interaction terms in the lagrangian.66 and 10. 3! 4! The vertices are: = −i f Consider ﬁrst the lowest order self-energy diagram.

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

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

λ 1 ∂ (10. 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 . writing down the propagators till the beginning of the ¯ line (incoming fermion u or outgoing antifermion v) has been reached. The wave functions for fermions are given by ∝ g 2 S(z − z ′ ) S(z ′ − z) = −g 2 iS(z − z ′ ) iS(z ′ − z). It is necessary to start at the end of a fermion line (with the above arrow convention an outgoing fermion wave function u or an incoming antifermion ¯ wave function v ) and keep on following that line. λ (∂µ Aµ )2 2 1 For a massive vector boson inversion of the quadratic term including the Lorentz constraint ∝ the propagator kµ kν 1 −gµν + 1− i Dµν (k) = i . i p p i ui (p) ui (p) ¯ vi (p) ¯ incoming fermion outgoing fermion incoming antifermion outgoing antifermion i p p i vi (p) In writing down the expressions for Feynman diagrams one has to be aware that the wave functions and the propagators have more than one component. 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.74) L = − Fµν F µν − (∂ · A)2 + ψ(i/ − M )ψ − e ψγ µ ψ Aµ . 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 .Feynman diagrams for scattering amplitudes 99 The result is which contains an extra minus sign as compared to a bosonic loop. Vector ﬁelds and Quantum Electrodynamics As the most important example of vector ﬁelds consider the lagrangian density for quantum electrodynamics (QED). 1 k 2 − M 2 + iǫ λ (k 2 − M 2 + iǫ)(k 2 − λ M 2 + iǫ) leads to . Particular choices of λ are λ = 1 (Feynman propagator or Feynman gauge) and λ = ∞ (Landau gauge). + 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.

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) (10.75) Note that the q µ q ν term in the photon propagator are irrelevant because the photon couples to a conserved current.s2 . 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 . q 4 µν [using that (¯(k ′ )γµ u(k))∗ = u(k)γµ u(k ′ )]. s2 )|2 ¯ q2 (10. s3 )γ µ u(k.s2 ) which can be written as (see also chapter 4) |M |2 = = 1 4s |¯(k ′ .s3 .76) 1 . t2 (10. s4 )(−ie)γ ν u(p.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′ )γµ u(k. s2 ). s3 )(−ie)γ µ u(k. where u ¯ L(m) µν = = 1 2 u(k ′ .79) .78) = 2 α2 2 s + u2 + 4t(M 2 + m2 ) − 2(M 2 + m2 )2 .Feynman diagrams for scattering amplitudes 100 10.s4 ) and averaged over spins in the initial state (1/2 × 1/2 × s1 . s1 ) ¯ −igµν u(p′ . To lowest order (∝ α = e2 /4π) only one diagram contributes. s1 ) u e2 u(p′ . The diagram and momenta and the commonly used invariants (Mandelstam variables) for a 2 → 2 scattering process are 1 k 3 k’ s t u = (k + p)2 = m2 + M 2 + 2 k · p = (k ′ + p′ )2 = m2 + M 2 + 2 k ′ · p′ = (k − k ′ )2 = q 2 = 2 m2 − 2 k · k ′ = (p − p′ )2 = 2 M 2 − 2 p · p′ p 2 p’ 4 s = (k ′ − p)2 = m2 + M 2 − 2 k ′ · p + t+u= m2 = 2 m2 + 2 M 2 i = (k − p′ )2 = m2 + M 2 − 2 k · p′ The scattering amplitude is given by −iM = u(k ′ . s′ ) ¯ u s. s)γν u(k ′ . ¯ q2 (10. s4 )γµ u(p.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. s)¯(k.

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

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

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

m2 .14) One can. m2 . Prove that |q| = cm Ea cm Eb (s − m2 − m2 )2 − 4 m2 m2 a a b b = 4s s + m2 − m2 a b √ = . = 2 s λ(s. (11. In that case trf pa = (Ea . ptrf ). p2 ) = = where λ12 denotes λ(s.8) (11. 4π s 4π 8π s 4π (11. 2mb (11.10) The function λ(s. m2 ). 2mb λ(s. 4s (11.11) (11.16) . 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.12) pb = (mb . 0). 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. which in the speciﬁc case a b also can be expressed as λ(s. Prove that trf Ea = |ptrf | = a s − m2 − m2 a b . m2 .15) Explicit calculation of the reduced two-body phase space element gives dR(s. 2 s s − m2 + m2 a b √ .9) (11. m2 ) a b . m2 ) is a function symmetric in its three arguments. 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. Also in this case one can express the energy and momentum in the invariants. m2 ) = 4(pa · pb )2 − 4p2 p2 . p1 . for instance. m2 .Scattering theory 104 It is straightforward to prove that the unknowns in the particular system can be expressed in the invariants (ma . mb and s).13) (11. m2 . p1 . trf Ea (threshold) = 1 2 mb ( i mi )2 − m2 − m2 a b . m2 ) a b . 1 2 d(E1 + E2 ) √ |q| dΩ(ˆ) q δ( s − E1 − E2 ) 2 (2π) 4(E1 + E2 ) √ q |q| dΩ(ˆ) λ12 dΩ(ˆ) q √ = . a (11.

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

These variables are precisely the Mandelstam variables for the s-channel process (ab → cd).28) which (only involving angles) is easier to determine. ¯ b (11. ¯ tt = (pa − p¯)2 = s. 11.31) . This is known as crossing symmetry. b Analiticity of the ﬁeld theoretically calculated result implies Ma¯→¯ (st = t. (11. For large energies and not too small scattering angles (θ ≫ 1/γ) y is approximately equal to the pseudo-rapidity η. t-channel and u-channel processes respectively. b ut = (pa − pd )2 = u.29) 3 2E 3 2 (2π) 2 (2π) 16π 2π E 16π 2 2π with at high energies the factor |p|/E ≃ 1. tt = s. uu = s) = Mab→cd (s. With the momenta deﬁned as in the ﬁgures above one has for all these processes the same amplitude M (s. t. u). u = (p1 − p4 )2 = (p2 − p3 )2 . ¯ while for the u-channel process (ad → c¯ one has b) su = (pa + pd )2 = u. η = − ln(tan(θ/2)) = tanh−1 (cos θ). ¯ tu = (pa − pc )2 = t. u) with s = (p1 + p2 )2 = (p3 + p4 )2 .Scattering theory 106 It is convenient because under a boost (along z) with velocity β it changes as y → y − tanh−1 β. u).30) (11. t. Given a two-to-two scattering process ab → cd one can relate the processes b p p a 2 1 d p 4 _ c −p p 1 3 d p 4 _ d −p p a 1 4 _ b −p 2 p 3 c ab cd a _ ac −p 2 _ _ b bd _ ad p 3 _ c cb They are referred to as s-channel. tu = s. we have seen that the amplitudes for the processes e− µ− → e− µ− and the process e− e+ → µ+ µ− are simply related by an interchange of the variables s and t. ut = u) = Mab→cd (s. t = (p1 − p3 )2 = (p2 − p4 )2 . 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. uu = (pa − p¯)2 = s.2 Crossing symmetry In the previous chapter. (11. t. For the t-channel process (a¯ → ¯ one has c bd) st = (pa + pc )2 = t. which means that rapidity distributions dN/dy maintain their shape.

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

dt (11.e. . (2π)4 δ 4 (Pi − Pf ) 2 = (2π)4 δ 4 (Pi − Pf ) = (2π)4 δ 4 (Pi − Pf ) d4 x ei(Pi −Pf )·x V. . σ= √ 2 λab T · V (11.37) 2 ma T · V Fermi’s Golden Rule In both the scattering cross section and the decay constant the quantity N/T V appears. N and T · V are covariant. pn ).39) One of the δ functions can be rewritten as T · V (remember the normalization of plane waves). The decay time for moving particles τ is related to the decay time in the rest frame of that particle (the proper decay time τ0 ) by τ = γ τ0 . it is. 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 ).Scattering theory 108 Although this at ﬁrst sight does not look covariant. From the solution N (t) = N (0) e−Γ t one knows that the decay time τ = 1/Γ. 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. the number of scattered or decayed particles is given by 2 N = (2π)4 δ 4 (Pi − Pf ) iMf i dR(p1 .33) Decay of particles For the decay of particle a one has macroscopically dN = −Γ N. T · V ρa (11. N 1 . .T V. the amount of decaying particles is proportional to the number of particles with proportionality factor the em decay width Γ. (11. Microscopically one has Ndecay = Na · Γ T or Γ= N 1 . as expected.36) (11. For the (proper) decay width one thus has 1 N Γ0 = . (11. . .34) i.35) This quantity is not covariant.38) (in which we can calculate −iMf i using Feynman diagrams).

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

θcm ) √ q 8π s 2 .e.49) or Im Mℓ = q |Mℓ |2 . η = 1) the result σel = 4π q2 (2ℓ + 1) sin2 δℓ . σinel = π q2 2 (2ℓ + 1) (1 − ηℓ ). † −i (Mℓ )f i − (Mℓ )f i = 2 † (Mℓ )f n qn (Mℓ )ni . which allows writing Mℓ (s) = Sℓ (s) − 1 e2i δℓ (s) − 1 = . (11. 0). q ˆ i. (11.Scattering theory 110 we obtain LHS = −i while for the RHS use is made of ℓ † (2ℓ + 1) Pℓ (ˆi · qf ) (Mℓ )f i − (Mℓ )f i . n (11.53) The diﬀerence is the inelastic cross section. q ˆ Pℓ (ˆ · q ′ ) = q ˆ (ℓ) m 4π Y (ℓ) (ˆ) Ym (ˆ′ ) q (ℓ)∗ q 2ℓ + 1 m and the orthogonality of the Ym functions to prove that RHS = 2 n ℓ † (2ℓ + 1)Pℓ (ˆi · qf ) (Mℓ )f n qn (Mℓ )ni . (11.48) If only one channel is present this simpliﬁes to ∗ ∗ −i (Mℓ − Mℓ ) = 2 qMℓ Mℓ . Show that σT = = 4π q 2π q2 (2ℓ + 1) Im Mℓ (s) ℓ ℓ (2ℓ + 1) (1 − ηℓ cos 2δℓ ).51) and for the case that this is the only channel (purely elastic scattering. Using σ= dΩ q ′ M (s. parametrized as Sℓ (s) = ηℓ (s) exp(2i δℓ (s)). In general a given channel has |Sℓ (s)| ≤ 1. σel = 4π ℓ (2ℓ + 1)|Mℓ (s)|2 (2ℓ + 1) ℓ = 4π q2 ηℓ e 2i δℓ − 1 2i 2 . in combination with the partial wave expansion for the amplitudes M and the orthogonality of the Legendre polynomials immediately gives for the elastic channel.52) From the imaginary part of M (s.54) ℓ . (11. 2i q 2i q (11.50) where Sℓ (s) satisﬁes |Sℓ (s)| = 1 and δℓ (s) is the phase shift. the total cross section can be determined. ℓ (11.

57) This shows that Γ is the width of the resonance. i. The prescription for the pole structure. one has poles at k 0 = ±(Ek − iǫ) where Ek = + k2 + M 2 guarantees the correct behavior. 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 . (q Mℓ )ij (s) = −M Γi Γj . 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 .55) (note that we have disregarded spin).56) (again disregarding spin). For an unstable particle one expects that U (t. 0) ∝ e−i(E−iΓ/2)t . however. it is straightforward to write down the partial wave amplitude. . σel = σinel . in which case.Scattering theory 111 Note that the total cross section is maximal in the case of full absorption. For the amplitude in a scattering process going through a resonance. (11. such that |U (t. 0)|2 = e−Γt . which is given by a sum of the partial widths into the diﬀerent channels. The quantity Γ is precisely the width for unstable particles. The unitarity condition then states 2 Im ∆−1 (k) = R n † dR(p1 . .5 Unstable particles i k 2 − M 2 + iǫ For a stable particle the propagator is i∆(k) = (11. but still real. The product of amplitudes is of order e4 . This is achieved with a propagator i∆R (k) = i k2 − M2 + iM Γ (11. 0). 11.58) . This is what we will do to discuss decay widths. . i. 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.e. For instance the tree-level calculation of Feynman diagrams at order e2 was real. ∝ U (t. . To check unitarity we need the amplitude at order e4 . η = 0. pn )MRn (2π)4 δ 4 (k − Pn ) MnR = 2M n Γn = 2M Γ.e. s − M 2 + iM Γ (11. the time-evolution operator. We note that unitarity is generally broken in a ﬁnite order calculation. This is (somewhat sloppy!) seen by considering the (amputated) 1PI two-point vertex Γ(2) = −i ∆ as the amplitude −iM for scattering a particle into itself (forward!) through the decay channels as intermediate states.

From this one sees that a resonance has the same shape in all channels but diﬀerent strength. Three neutrinos explain the resonance shape. u) + g(t. 2 q (s − M 2 )2 + M 2 Γ2 (11. The half-width of the resonance is M Γ. At resonance the cross section σT (π + p) is about 210 mb. Limiting ourselves to a resonance in one channel. its width Γ = 120 MeV. the partial width into e+ e− is Γee = 5.59) reaching the unitarity limit for s = M 2 . u) = g(t. • The J/ψ resonance in e+ e− scattering. u) = 1 1 2 (s + u2 ) + 4m2 (t − m2 ) t2 2 2 1 1 ( s − m2 )( s − 3m2 ) tu 2 2 . Three famous resonances are: • The ∆-resonance seen in pion-nucleon scattering. Note that because of the presence of a background the phase shift at resonance may actually be shifted.2 GeV. The cross section σT (π − p) also shows a resonance with the same width with a value of about 70 mb.1 Show that the cross section for electron-electron scattering (exercise 10. t)} . the full width is Γ = 88 keV. This characteristic shape of a resonance is called the Breit-Wigner shape.Scattering theory 112 (Prove this using the unitarity condition for partial waves). where furthermore σinel = 0. Γ = 2. • The Z 0 resonance in e+ e− scattering with M = 91.60) showing that the phase shift at resonance rises through δ = π/2 with a ’velocity’ ∂δ/∂s = 1/M Γ. it is furthermore easy to prove that the cross section is given by σel = M 2 Γ2 4π (2ℓ + 1) .49 GeV. i. dt s(s − 4m2 ) with f (t. which is attributed to neutrinos.88 MeV. t) + g(u. a fast change in the phase shift for a narrow resonance. one can reconstruct the resonance shape for diﬀerent numbers of neutrino species.4) can be written as 4πα2 dσ = {f (t. Exercises Exercise 11. Knowing that each neutrino contributes about 160 MeV (see next chapter). The phase shift in the resonating channel near the resonance is given by tan δℓ (s) = MΓ . Its mass is M = 3096. M2 − s (11. The cross section at resonance is about 30 nb.e. u) + f (u. Its mass is M = 1232 MeV. All these decays can be seen and leave an ’invisible’ width of 498 MeV.26 keV. This is a narrow resonance discovered in 1974. Essentially this resonance can decay into quark-antiquark pairs or into pairs of charged leptons. 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).

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.999 877 Γ and Γ(π − → e− νe ) = 0. we have here already used unitarity to ﬁnd a ﬁnite imaginary part that is proportional to the total decay width. Calculate now fπ .8.000 123 Γ.2). Exercise 11.Scattering theory 113 Exercise 11. 4. Compare the result with the factor trf 1/ρa ρb va that we used in the derivation of the basic expression for cross sections. writing N (s) M1 (s) = 2 + iM Γ . cm |va − vb | in the center of mass system for two colliding particles a and b.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∆ .2 Calculate the ﬂux factor. As discussed for unstable particles. s − M∆ ∆ ∆ 2 calculate Re N (M∆ ) . Exercise 11. Thus. Show that unitarity can be used one step further to ﬁx also the real part of the residue of the pole in the amplitude.6 × 10−8 s and the relative decay widths Γ(π − → µ− νµ ) = 0. a quantity which we will ¯ ¯ encounter in the next chapter (section 12. Compare the results with the experimental π − lifetime τ = 2. ρcm ρcm a b 1 cm .

Chapter 12

**The standard model
**

12.1 Non-abelian gauge theories

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

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

a

La

φ(x) = (1 + i θa La ) φ(x)

inf

(12.2)

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

inf

(12.3)

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

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

The standard model

115

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

′a a c Wµ = Wµ − cabc θb Wµ +

i (∂µ U (θ)) U −1 (θ), g

(12.7)

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

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

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

(12.8)

(12.9) (12.10)

The gauge-invariant lagrangian density is now constructed as

Gµν → U (θ) Gµν U −1 (θ).

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

(12.11)

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

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

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

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

**The standard model
**

xψ x+dx

116

(x)

ψ(x)

x+dx ψ(x+dx)

**x-space internal space
**

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

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

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

**A geometric picture of gauge theories
**

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

1 The

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

λ7 =

i

−i

1 1 λ8 = √ 3

−2

(12. We note that local gauge invariance requires that we can modify all local systems with a (local) unitary transformation S(x). ∂µ − ig Aµ (x) ψ(x) = 0 dxµ ∂µ − ig Aµ (x(s)) ψ(x) = 0. i. should be independent of such transformations. (12.19) This gives rise to a (path dependent) phase in each point. Dν ] ψ(x) 1 − ig dσ µν Gµν ψ(x). If there is a ’true’ diﬀerence in the vector ψ(x) and ψ(x + dx) it is denoted with the covariant derivative connecting the vectors expressed with respect to the same basis. x+dx ψ(x) = = 1 + ig dxµ Aa (x)Ta µ 1 + ig dx Aµ (x) µ x x ψ(x) (12. However.e. In principle such a phase in a given point is not observable. (12. (12. requiring that the ’connection’ Aµ (x) is such that Dµ → S(x)D µ S −1 (x). ds′ µ 0 ig Aµ (s) ψ(s) dxµ Aµ (x) ψ(s) = which is the path-ordered integral denoted ψ(x) = P eig P ψ(0). if two diﬀerent paths to the same point give diﬀerent phases the eﬀects can be observed. ds s dxµ ds′ P exp ig A (s′ ) ψ(0).16) which in the presence of the ’connection’ Aµ diﬀers from the total change between x ψ(x) and x+dx ψ(x+ dx). x x+dx ψ(x + dx) = (1 + dxµ ∂µ ) ψ(x).e.18) or considering a path xµ (s) from a ﬁxed origin (0) to point x.17) The three equations given so far immediately give Dµ = ∂µ − ig Aµ (x). A ’constant’ vector that only rotates because of the arbitrary deﬁnitions of local frames satisﬁes Dµ ψ(x) = 0.The standard model the basis for x G and x+dx 117 G. i. The relation in Eq. which connects two identical vectors.15) ψ(x).16. but expresses them with respect to diﬀerent bases.14) (12. 12. ds which is solved by dψ(s) ds = dxµ . -dx -dy µ µ dyµ µ dx For a vector. x+dx ψ(x + dx) = 1 + dxµ Dµ x+dx ψ(x). What is this physical eﬀect by which the ’connection’ Aµ can be observed? For this consider the phase around a closed loop. the real change thus is given by ψ(x) = (1 − dy ρ Dρ )(1 − dxσ Dσ )(1 + dy ν Dν )(1 + dxµ Dµ )ψ(x) = 1 + dxµ dy ν [Dµ . = .

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

(12.) terms constitute interaction terms.23) F +η the latter only forming a minimum for m2 < 0.21) Stability of the action requires the classical groundstate ϕc to have a well-deﬁned value (which can be nonzero). etc. say 0 0 .e. 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. . φ = ϕc + φquantum where for the (operator-valued) coeﬃcients in the quantum ﬁeld one wants 0|c† = c|0 = 0. . therefore a choice must be made. The situation now is the following. In the case of degeneracy. 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 standard model 119 a constant ﬁeld (∇φ = 0) and the condition ∂V ∂φ ϕc =0 −→ ϕc · ϕc = 0 or ϕc · ϕc = − m2 ≡ F 2. They are known as the Weyl mode or the Goldstone mode: Weyl mode. The classical groundstate appears degenerate. . Writing the ﬁeld φ as a sum of a classical and a quantum ﬁeld. The ﬁeld along the third axis plays a special role because of the choice of the vacuum expectation value.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.24) 1 2 2 4 1 1 1 = (∂µ ϕ1 )2 + (∂µ ϕ2 )2 + (∂µ η)2 + m2 η 2 + . The η massless particles are called Goldstone bosons. A quantum ﬁeld theory has only one nondegenerate groundstate |0 . In this situation one speaks of spontaneous symmetry breaking. It is suﬃcient to do this to second order in the ﬁelds as the higher (cubic. ϕc = 0|φ |0 = (12. It is only invariant under rotations around the 3-axis. (12. λ such that 0|ϕ1 |0 = 0|ϕ2 |0 = 0|η|0 = 0. has less symmetry. The presence of a nonzero value for the classical groundstate value of the ﬁeld will have an eﬀect when the ﬁeld is quantized. In order to see the consequences for the particle spectrum of the theory we construct the lagrangian in terms of the ﬁelds ϕ1 . while the quadratic terms must correspond with non-negative masses. Any constant ﬁeld with ’length’ F is a possible groundstate. The original lagrangian contained an SO(3) invariance under (length conserving) rotations among the three ﬁelds. so one has 0|φquantum |0 = 0 and hence 0|φ|0 = ϕc . for the vacuum Qa |0 = 0. In this mode the lagrangian and the vacuum are both invariant under a set of symmetry transformations generated by Qa . ϕ2 and η. It is appropriate to redeﬁne the ﬁeld as ϕ1 φ = ϕ2 . In this case the spectrum is described . (12. while the lagrangian including the nonzero groundstate expectation value chosen by nature.25) 2 2 2 Therefore there are 2 massless scalar particles. i.

Goldstone mode. Using the Dirac equation.29) (Note that each of the entries in the vector for ψ in the above equation is still a 4-component Dirac spinor).31) . Taking the derivative. suggesting the stronger relation ∂ µ Jµ (x) = fπ m2 a π a (x). .27) If the transformations in the lagrangian give rise to a symmetry the Noether currents are conserved.26) for all the states labeled by a corresponding to ’broken’ generators. called ﬂavors. M = .e. a and a′ are degenerate states. (12. and one must have mπa = 0. A bit more formal. ψ −→ ei α·T ψ. .) one can write L = f ψ f (i/ − Mf )ψf . the lagrangian for the quarks consists of the free Dirac lagrangian for each of the types of quarks. etc. [Qa . As the a generators for a symmetry are precisely the zero-components of a conserved current Jµ (x) integrated a a over space. H] = 0. down. a 0|∂ µ Jµ (x)|π b (k) = fπ k 2 ei k·x δab = fπ m2 a ei k·x δab . π (12. (12.e. there must be a nonzero expectation value 0|Jµ (x)|π (k) . This lagrangian density then is invariant under unitary (vector) transformations in the ﬂavor space. i. .. Lx and Ly or instead of the latter two L+ and L− ) are used to label the multiplet members or transform them into one another. i. and |a and |a′ belong to the same multiplet (there is a Qa such that |a′ = Qa |a ) then H|a = Ea |a implies that H|a′ = Ea |a′ . The conserved currents corresponding to this symmetry transformation are found directly using Noether’s theorem (see chapter 6). T = τ /2. π Example: chiral symmetry An example of spontaneous symmetry breaking is chiral symmetry breaking in QCD. Known examples are rotational symmetry and the fact that the spectrum shows multiplets labeled by angular momentum ℓ (with members labeled by m). if the generators Qa generate a symmetry. 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) (12.28) where ψ is extended to a vector in ﬂavor space and M is a diagonal matrix in ﬂavor space. T ] ψ.e. a massless Goldstone boson for each ’broken’ generator.30) which in the case that we include only two ﬂavors form an SU (2)V symmetry (isospin symmetry) generated by the Pauli matrices. ∂ ∂ (12. The generators Qa (in that case the rotation operators Lz . a ∂ µ Jµ = 0. denoted |π a (k) . Including a sum over the diﬀerent ﬂavors (up. Note that for the ﬁelds π a (x) one would have a the relation 0|π a (x)|π a (k) = ei k·x . ψu mu md ψ = ψd . In this mode the lagrangian is invariant but Qa |0 = 0 for a number of generators. it is easy to see that one gets ∂µ V µ = i ψ [M. (12. = ψ(i/ − M )ψ. strange. This means that they are operators that create states from the vacuum. irrespective of the fact if they annihilate the vacuum.32) (12.The standard model 120 by degenerate representations of the symmetry group. V µ = ψγ µ T ψ. i. . Neglecting at this point the local color symmetry.

Note that these transformations also work on the spinor indices. (leaving out the ﬂavor structure) the same as ψγ5 ψ. socalled axial vector transformations. is what happens in the real world.e. fπ = 0 remains. This symmetry is realized in the Weyl mode with the spectrum of QCD showing an almost perfect isospin symmetry. The Goldstone bosons of QCD are the pions for which fπ = 93 MeV and which have a mass of mπ ≈ 138 MeV/c2 . There exists another set of symmetry transformations. etc. i. 12. . I. in terms of which the Dirac lagrangian 2 density looks like L = i ψL / ψL + i ψR / ψR − ψR M ψL − ψL M ψR . proton and neutron. From the number of broken generators it is clear that one expects three massless Goldstone bosons. Therefore the world of up and down quarks describing pions. 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.27. (12. M = m · 1. generated by TR/L = 1 (1 ± γ5 )T . with almost degenerate masses (Mp = 938. nucleons and atomic nuclei has not only an isospin or vector symmetry SU (2)V but also an axial vector symmetry SU (2)A .35 is nonzero. or.3 MeV/c2 and Mn = 939. ψ −→ ei α·T γ5 ψ. if in ﬂavor space M is proportional to the unit matrix.6 MeV/c2 ). which is equivalent 2 to the V-A symmetry. much smaller than any of the other mesons or baryons. Still the basic fact that the generators acting on the vacuum give a nonzero result (i. ψR/L = 1 (1 ± γ5 )ψ. parity would not play a role in the world. T γ5 = τ γ5 /2. This situation. i. nature has choosen the path SU (2)L ⊗ SU (2)R =⇒ SU (2)V . We know that mesons and baryons (such as the nucleons) have a well-deﬁned parity that is conserved. ∂ ∂ (12. T ] = 0.e. Aµ = ψγ µ T γ5 ψ. etc. which is approximately true for the up and down quarks. 12.35) In this case ∂µ Aµ = 0 will be true if the quarks have zero mass. T } γ5 ψ.The standard model 121 Furthermore ∂µ V µ = 0 ⇐⇒ [M. it is easy to see that one gets ∂µ Aµ = i ψ {M. but also a triplet (isospin 1) of pions. including in Dirac space a γ5 matrix.e. The chiral ﬁelds ψR and ψL are transformed into each other under parity. but the fact that the symmetry is not perfect and the right hand side of Eq. however. parity minus). The currents corresponding to this symmetry transformation are again found using Noether’s theorem. is by nature not realized in the Weyl mode. e. (12.e. chiral symmetry is not perfect. both are very small.33) which in the case of two ﬂavors form SU (2)A transformations generated by the Pauli matrices.e. SU (2)V (isospin) symmetry is good if the up and down quark masses are identical. Therefore realization in the Weyl mode would require that all particles come double with positive and negative parity. as the quantity ∂µ Aµ (x). From group theory (Schur’s theorem) one knows that the latter can only be true. How can we see this. a doublet (isospin 1/2) of nucleons. (12. This combined symmetry is what one calls chiral symmetry. where the quark masses are not completely zero. That the massless theory has this symmetry can also be seen by writing it down for the socalled lefthanded and righthanded fermions. i. the lagrangian density is invariant under left (L) and right (R) rotations independently. while the groundstate is only invariant under isospin rotations (R = L). gives also rise to a nonzero mass for the Goldstone bosons according to Eq.g.36) If the mass is zero the lagrangian is split into two disjunct parts for L and R showing that there is a direct product SU (2)L ⊗ SU (2)R symmetry. In the real world. for which the ﬁeld (according to the discussion above) has the same behavior under parity. behaves as a pseudoscalar particle (spin zero. This symmetry.34) Using the Dirac equation. stated equivalently.

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. while the number of massive gauge bosons coincides with the number of ’broken’ generators. 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 + . one sees that the ﬁelds Wµ and Gµν also can be represented as three-component ﬁelds.37) (12. The diﬀerence comes when we reparametrize the ﬁeld 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 .The standard model 122 12.38) ∂µ Wν − ∂ν Wµ + g Wµ × Wν . 1 1 L = − Gµν · Gµν + Dµ φ · Dµ φ − V (φ). is made. like a photon. thus 9 independent degrees of freedom. in this case three-dimensional) representation reads (La )ij = −i ǫaij . . (12.42) 2 from which one reads oﬀ that the particle content of the theory consists of one massless gauge boson 3 1 2 (Wµ ). Dµ φ = Gµν = ∂µ φ + g Wµ × φ.40) a Dµ φ = ∂µ φ − ig Wµ La φ. 0 0 0 = . 0 φ= (12.39) (12. but one has 1 massless gauge ﬁeld (2). . We have the possibility to perform local gauge transformations. . The latter is a spin 0 particle (real scalar ﬁeld) called a Higgs particle. 4 2 where Since the explicit (adjoint. two massive bosons (Wµ and Wµ with MW = gF ) and a massive scalar particle (η with m2 = η 2 −2 m . After symmetry breaking the same number (as expected) comes out. 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). made into a gauge theory. 0 ϕc = 0|φ |0 = 0 . 2 massive vector ﬁelds or spin 1 bosons (2 × 3) and one scalar ﬁeld (1). . having two independent spin components) and three scalar ﬁelds (one degree of freedom each). Considering the same example with rotational symmetry (SO(3)) as for spontaneous symmetry breaking of a scalar ﬁeld (Higgs ﬁeld) with three components. i. The symmetry is broken in the same way as before and the same choice for the vacuum. again 9 degrees of freedom. Therefore we can always rotate the ﬁeld φ into the 3-direction in order to simplify the calculation. (12. Initially there are 3 massless gauge ﬁelds (each. (12.e. as in this case there are no massless Goldstone bosons. One may wonder about the degrees of freedom.

U(1)Y U(1)Y (12. 1 As they are massless.51) .43) and One introduces a (weak) SU (2)W symmetry under which eR forms a singlet. 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. neutrinos.e. however.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. its corresponding neutrino (νe . 2 2 = ∂µ R + i g ′ Bµ R. muon µ− or tau τ − ).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. under which the lefthanded and righthanded particles with YW (L) = −1 and YW (R) = −2 transform with eiβYW /2 . ψR/L = 2 (1 ± γ5 )ψ decouple. implies certain nontrivial relations between masses and mixing angles that can be tested experimentally and sofar are in excellent agreement with experiment. In the standard model one starts with a very simple basic lagrangian for (massless) fermions which exhibits more symmetry than observed in nature. By introducing gauge ﬁelds and breaking the symmetry a more complex lagrangian is obtained. R. = ∂µ L − (12. Thus the lagrangian density is L (f ) ∂ ∂ = i L/ L + i R/ R. that gives a good description of the physical world.The standard model 123 12. explicitly L −→ e−i β/2 L.47) and YW is considered as an operator that generates a U (1)Y symmetry. explicitly Dµ L Dµ R i i g Wµ · τ L + g ′ Bµ L. the quarks (up. For the neutrino only a lefthanded particle (and righthanded antiparticle) exist. the leptons (elektrons. Thus L (f ) = i eR / eR + i eL / eL + i νeL / νeL + (µ. The lagrangian for the leptons consists of three families each containing an elementary fermion (electron e− .44) which has an SU (2)W symmetry under transformations eiα·TW . muons. R −→ e−i β R. left. etc. τ ). νµ and ντ ) and their antiparticles.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. The procedure. etc.). 2 (12. electromagnetic and weak forces. while the lefthanded particles form a doublet.50) (12. explicitly L R SU(2)W SU(2)W −→ ei α·τ /2 L. (12. i. down. (12.45) (12.and righthanded particles. This is written as 3 Q = TW + YW . ∂ ∂ ∂ (12.) and the gauge bosons responsible for the strong.48) (12.

(12. L where L (h1) (h) =L (h1) +L (h2) .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.53) φ= √ (θ4 − iθ3 ) φ 2 with TW = 1/2 and YW = 1 is introduced. 2 and 3 may be eliminated (the necessary SU (2)W rotation is precisely e−iθ(x)·τ ).57) = 1 g2 v2 1 2 (Wµ )2 + (Wµ )2 (∂µ h)2 + m2 h2 + 2 8 v2 2 3 gWµ − g ′ Bµ + . −V (φ) L and (h2) = −Ge (LφR + Rφ L). (12. which assures with the choice of YW for the Higgs ﬁeld that Q generates the remaining U (1) symmetry.58) (12. . † i i g Wµ · τ − g ′ Bµ )φ. Using local gauge invariance θi for i = 1. the U (1)Q symmetry (generated by the charge operator). h) (12. . . with the following lagrangian density consisting of a symmetry breaking piece and a coupling to the fermions. . .55) 2 2 The Higgs potential V (φ) is choosen such that it gives rise to spontaneous symmetry breaking with ϕ† ϕ = −m2 /2λ ≡ v 2 /2. this leads to the lagrangian L (h1) .54) = (Dµ φ)† (Dµ φ) −m2 φ† φ − λ (φ† φ)2 . a complex Higgs ﬁeld + 1 φ √2 (θ2 + iθ1 ) 0 = 1 (12. (12. TW = ±1 or 0 and YW = 0 (thus Q = TW ) and 3 Bµ is a singlet under SU (2)W (TW = TW = 0) and also has YW = 0. + 8 (12. Putting this in leads to L (f ) =L (f 1) +L (f 2) .59) .56) v + h(x) 2 Dµ φ = (∂µ − and Dµ φ = = 1 √ 2 − ig 2 ∂µ h + 1 2 Wµ −iWµ √ (v + h) 2 3 gWµ −g′ Bµ i √ (v + 2 2 Up to cubic terms. + 8 g2 v2 1 + − (Wµ )2 + (Wµ )2 (∂µ h)2 + m2 h2 + 2 8 (g 2 + g ′2 ) v 2 (Zµ )2 + .The standard model 124 3 3 where Wµ is a triplet of gauge bosons with TW = 1. leading to the parametrization 1 0 φ(x) = √ (12. For the classical ﬁeld the choice θ4 = v is made.

2θ 2θ 4 cos W cos W 0.68) 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 .66) From the coupling to the photon. (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 . 2 (12. 4 2 MW g2 v2 = .65) The weak mixing angle is related to the ratio of coupling constants.69) .60) (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 .63) (12. (12. .The standard model 125 where the quadratically appearing gauge ﬁelds that are furthermore eigenstates of the charge operator are ± Wµ = = = Zµ Aµ 1 1 2 √ Wµ ± i Wµ .67) The coupling of electrons or muons to their respective neutrinos. we can read oﬀ e = g sin θW = g ′ cos θW . 2 Zµ (12.61) (12. 2 + g ′2 g 3 g ′ Wµ + g Bµ (12. σ (νµ γ ρ µL ) 2 2 (eL γ νe ) 2 k + MW (12. g ′ /g = tan θW .64) (12.62) g 2 + g ′2 3 ≡ sin θW Wµ + cos θW Bµ . 2 3 g Wµ − g ′ Bµ 3 ≡ cos θW Wµ − sin θW Bµ . 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 + . .

(12. such as2 α = e2 /4π GF sin2 θW MW MZ ≈ 1/127. 91. Γe ≈ Γµ ≈ Γτ ≈ 83 MeV).05 and CA = −1/2 we calculate Γ(e+ e− ) ≈ 78.73) (12. 3 = TW . Finally we want to say something about the weak properties of the quarks.74) (12. from the MW mass is about 250 GeV. leptons with CV = −1/2 + 2 sin2 θW ≈ −0.e. The masses of the fermions and the coupling to the Higgs particle are contained in L (h2) . g ′ and v determine a number of experimentally measurable quantities. 2 2 2 (12. one obtains L (h2) Ge v Ge = − √ (eL eR + eR eL ) − √ (eL eR + eR eL ) h 2 2 me = −me ee − eeh.The standard model 126 the good old four-point interaction introduced by Fermi to explain the weak interactions.79) Γ(Z 0 → f f ) = 48π cos2 θW For the electron. 2 The value of α deviates from the known value α ≈ 1/137 because of higher order contributions. muon or tau. v (12. me /v is extremely small. For each neutrino species (with CV = 1/2 and CA = 1/2 one expects Γ(νν) ≈ 155 MeV. Comparing this with the total width into (invisible!) channels. g2 MZ 2 2 (CV + CA ). (12. With the choosen vacuum expectation value for the Higgs ﬁeld. −5 (12.19 GeV. giving rise to a running coupling constant after renormalization of the ﬁeld theory.g.78) From this coupling it is straightforward to calculate the partial width for Z 0 into a fermion-antifermion pair. The coupling to the Higgs particle is weak as the value for v calculated e. the mass of the electron comes from the spontaneous symmetry breaking but is not predicted (it is in the coupling Ge ). i.71) GeV −2 = = = = .166 4 × 10 0.70) = 2 = 2 sin2 θ 8 MW 2 v2 8 MW 2 W In this way the parameters g. (12.40 GeV. . i. as appear for instance in the decay of the neutron or the decay of the Λ (quark content uds). Actually including corrections corresponding to higher order diagrams the agreement for the decay width into electrons can be calculated much more accurately and the number of allowed (light) neutrinos turns to be even closer to three. 1.72) (12.75) The coupling of the Z 0 to fermions is given by (g/ cos θW )γ µ multiplied with 3 TW 1 1 1 (1 − γ5 ) − sin2 θW Q ≡ CV − CA γ5 .5 MeV (exp.77) (12. one has the relation g2 e2 1 GF √ = .231 2. 80.e.80) First. Γinvisible = 480 MeV one sees that three families of (light) neutrinos are allowed.76) with CV CA 3 = TW − 2 sin2 θW Q. (12.

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

The imaginary part i η gives rise to CP violation in decays of K and B-mesons (containing s and b quarks. † Λu Λ† = Vu G2 Vu . Working with the mass eigenstates one simply sees that the weak current coupling to the W ± mass † mass becomes U L γ µ DL = U L γ µ Vu Vd DL . The Higgs ﬁeld is still limited to one complex doublet. mc and mt ) and Md = Gd v/ 2 (diagonal matrix containing md .The standard model 128 using the socalled Wolfenstein + O(λ4 ) with λ ≈ 0. Note that we need the conjugate Higgs ﬁeld to get a U (1)Y singlet in the case of the charge +2/3 quarks. UR = uR cR tR .g. d d (12.86) the unitary CKM-matrix introduced above in an ad hoc way. u u where we include now the three lefthanded quark doublets in QL .85) √ √ with Mu = Gu v/ 2 (diagonal matrix containing mu . for which we need the appropriate weak isospin doublet 0∗ 1 φ c = √ v +h . e. 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 . ms and mb ). φ = −φ− 0 2 For the (squared) complex matrices we can ﬁnd positive eigenvalues.83) where Vu and Vd are unitary matrices.82) space. (12.84) with Gu and Gd being real and positive and Wu and Wd being diﬀerent unitary matrices. the three righthanded quarks with charge +2/3 in UR and the three righthanded quarks with charges −1/3 in DR . Thus one has † † † =⇒ −DL Vd Md Wd DR − DR Wd Md Vd† DL − UL Vu Mu Wu UR − UR Wu Mu Vu UL (12.82 and ρ ≈ 0. each of these containing the three families.227. allowing us to write † Λu = Vu Gu Wu † and Λd = Vd Gd Wd . 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 12.22 and η ≈ 0. A ≈ 0. the weak mass eigenstates are L (h2.q) = −QL φΛd DR − DR Λ† φ† QL − QL φc Λu UR − UR Λ† φc† QL u d (12. i. respectively). . The Λu and Λd are complex matrices in the 3×3 family and Λd Λ† = Vd G2 Vd† . (12.e.34.q) ′ weak † mass mass DL = DL = Vu Vd DL = VCKM DL .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.

For the (massless) neutrino ﬁelds we just can redeﬁne ﬁelds into NL = Ve† NL .e.91) † with UPMNS = Vn Ve known as the Pontecorvo-Maki-Nakagawa-Sakata mixing matrix. 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 . EL = Ve EL .89) with three righthanded neutrinos added to the previous case. we ﬁnd massless neutrinos. since the weak current is the only place where they show up. 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. decoupling from all known interactions. i. † † 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. and obtain L (h2. EL = Ve† EL .87) is a weak doublet containing the three families of neutrinos (NL ) and charged leptons (EL ) and ER is † a three-family weak singlet.5×10−3 eV2 . As before. e N L= L EL (12. The mass ﬁelds ER † mass † mass = We ER . = − 1/6 2/3 0 1/2 1/2 0 0 1 0 1/3 1/2 − 1/6 (12. while θ23 then is determined by atmospheric neutrino oscillations requiring |∆m2 | ≈ 2. it is particularly useful because θ12 is essentially determined by solar neutrino oscillations requiring ∆m2 ≈ 8 × 10−5 12 eV2 (convention m2 > m1 ).90) mass † mass mass † mass † We note that there are mass ﬁelds ER = We ER . (12. UPMNS a parametrization that in principle also could have been used for quarks. one can write Λe = Ve Ge We and we ﬁnd † L (h2. The lepton sector (massive Dirac neutrinos) In principle a massive Dirac neutrino could be accounted for by a lagrangian of the type L (h2. NL = Vn NL and NR = Wn NR µ mass γ µ V † V N mass . there is no family mixing for massless neutrinos.ℓ) = −LφΛe ER − ER Λ† φ† L.88) √ mass with Me = Ge v/ 2 the diagonal mass matrix with masses me .ℓ) † † † =⇒ −EL Ve Me We ER − ER We Me Ve† EL − NL Vn Mn Wn NR − NR Wn Mn Vn NL .The standard model 129 The lepton sector (massless neutrinos) For a lepton sector with a lagrangian density of the form L in which (h2.93) 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 .92) . (12. (12.ℓ) =⇒ −Me EL Ve We ER − ER We Ve† EL . (12. mµ and mτ . The W -current then becomes mass mass E L γ µ NL = E L γ µ NL . 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 . In this case.

0 0 1 L mass.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. We use here the primes starting with the weak eigenstates. − MD MR nR 2 (12. 2 although this option is not attractive as it violates the electroweak symmetry.99) . Absorption of phases in the states is not possible for Majorana neutrinos. 1 c ∗ c L mass.94) M L NL NL + M L NL NL .88 a Majorana mass term for the (lefthanded) neutrino mass eigenstates.The standard model 130 The lepton sector (massive Majorana ﬁelds) An even simpler option than sterile righthanded Dirac neutrinos. are equivalent to two decoupled Majorana neutrinos (see below). The way to circumvent this is to introduce as in the previous section righthanded neutrinos. couple the right. Thus (disregarding family structure) one has two Majorana neutrinos.and righthanded species then just form a Dirac fermion. Υc ≡ (ΥL )c = ΥR 1 2 L and Υc = ΥL ). one being massive. (12. α2 and δ). This is called the see-saw mechanism (outlined below).96) VPMNS = UPMNS K with K = 0 eiα2 /2 0 . If the Majorana mass MR ≫ MD one actually obtains in a natural way one Majorana neutrino with a very small mass. a unitary matrix relating them to the weak eigenstates. (12. For the leptons. the massless and massive Majorana neutrinos. the left. hence the mixing matrix becomes iα /2 0 0 e 1 (12. with for the righthanded sector a mass term MR . 12. 2 R ψ = nR + nL . For the neutrino sector. coupled by a Dirac mass term. The see-saw mechanism Consider (for one family N = n) the most general Lorentz invariant mass term for two independent Majorana spinors. For the charged leptons there cannot exist a Majorana mass term as this would break the U(1) electromagnetic symmetry. 1 ∗ c c (12. Υ′ and Υ′ (satisfying Υc = Υ and as discussed in chapter 6.95) 2 In order to have more than a completely decoupled sector.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. 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.c. one must for the neutrinos as well as charged leptons. For these light Majorana neutrinos one has. 1 L Υ ′ = nR + nc . The mass matrix can be written as ML |MD | eiφ M = (12.ν =− containing three (CP-violating) phases (α1 . as above. Thus 2 Υ ′ = nc + nL .ν = − MR NR NR + MR NR NR .and lefthanded species through Dirac mass terms coming from the coupling to the Higgs sector as in the previous section. Actually. is to add in Eq. however.98) (12.

For the situation ML = 0 and MR ≫ MD (taking MD real) one 2 ﬁnds M1 ≈ MD /MR and M2 ≈ MR .e.104) ↔ 1 1 Υi i / Υi − Mi Υi Υi ∂ 4 2 (12. Exercise 12.2 Derive for the vector and axial vector currents. V µ = ψγ µ T ψ and Aµ = ψγ µ γ5 T ψ ∂µ V µ ∂µ A µ = = i ψ [M. 2 U (M M † ) U † = M0 . Exercise 12. Υ2R nR (12. a and a′ are degenerate states. (12.101) 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 . 2 with real masses Mi .e. which implies U ∗ M † U † = U ∗ M ∗ U † = M0 and a ’normal’ diagonalization of the (hermitean) matrix M M † . related via c Υ1R = U nL . T ] ψ. (12. Prove that if the generators Qa generate a symmetry. . i. H] = 0. i.103) for each of which one ﬁnds the lagrangians L = and we are left with two decoupled Majorana ﬁelds Υ1L = U ∗ nc . 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 . T } γ5 ψ.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 . 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). [Qa . i ψ {M. (12.105) for i = 1. and |a and |a′ belong to the same multiplet (there is a Qa such that |a′ = Qa |a ) then H|a = Ea |a implies that H|a′ = Ea |a′ . This is a mixing problem with a symmetric (complex) mass matrix leading to two (real) mass eigenstates. Exercises Exercise 12. L Υ2L nR Υ1 and Υ2 .The standard model 131 taking ML and MR real and non-negative.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)) .1 Consider the case of the Weyl mode for symmetries.

7 In this exercise two limits are investigated for the two-Majorana case.231. etc. hhh. mb ≈ 5 GeV and MW ≈ 80 GeV. Exercise 12. 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 +. and the width to a pair of neutrino’s. Neglect the masses of fermions and use the fact that the sum over polarizations is 3 pµ pν . 0 ¯ Γ(Z 0 → f f ) = MZ g2 48π cos2 θW f f CV 2 + CA 2 . 3 TW . Mf M2 and ghf f = .The standard model 132 with CV CA = = 3 TW − 2Q sin2 θW . Γ(Z 0 → νe νe ). e+ e− h. Exercise 12. ZZh. Write down the matrix element squared (averaged over initial spins and summed over ﬁnal spins) for the decay of the Z 0 .6 Check that the use of the Wolfenstein parametrization in the CKM matrix indeed gives a unitary matrix. ǫ(λ) (p)ǫ(λ)∗ (p) = −gµν + µ ν M2 λ=1 ¯ to calculate the width Γ(Z → f f ). Exercise 12. the weak mixing angle is given by ¯ sin2 θW = 0.5 Show that the couplings to the Higgs (for W + W − h. In the standard model this coupling is described by the vertex −i g µ √ (γ − γ µ γ5 ) . Γ(Z 0 → e+ e− ). Note that you can ﬁnd the answer without explicit construction of all interaction terms in the lagrangian. at least up to a high (which?) order in λ.) are proportional to the mass squared for bosons or mass for fermions of the particles. (b) Calculate the width to electron-positron pair. The mass of the Z 0 is MZ = 91 GeV. Exercise 12. to be precise one has for bosons (b) or fermions (f). . ghbb = b v v where v is the vacuum expectation value of the Higgs ﬁeld.

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

relativistic quantum mechanics

relativistic quantum mechanics

- Book 7
- Quantum Field Theory 2013
- M. Kozlov, E. Levin, V. Khachatryan and J. Miller-The BFKL Pomeron Calculus in zero transverse dimensions
- 6.17.docx
- vak
- Chapter 14- Iso, Geno and Hyper-Grand-Unifications and Iso,Geno, Hyper-Cosmologies:14.1- Iso, Geno and Hyper-Grand-Unification
- FunkyQuantumConcepts.pdf
- Chapter 10- Grand-Unification and Cosmology
- Monopole condensation in two-ﬂavour Adjoint QCD
- 9806024v1[1]
- Quantum Mechanics - Concepts and Applications - T. Biswas
- RevModPhys.66.129
- Notes 2003
- Paper on the Aokiphase 2008
- 2Q_Fradkin
- 2nd_lecture1
- UCB 221A - Postulates of Quantum Mechanics
- PhysRevLett.55.2850
- Proc_vietnam_2007
- Hydrogen Atom 2D
- Probability Density in the Complex Plane
- Infrared Behavior of Scalar Condensates in Effective Holographic Theories
- A Report on How Friction Factor is Obtained
- V.A.Berezin, A.M.Boyarsky and A.Yu.Neronov- Towards the mass spectrum of quantum black holes and wormholes
- lotka ecuaciones 2
- LC_lock
- Thermodynamic Bubble Ansatz
- Gauge Theories Notes

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