QUANTUM FIELD THEORY

P.J. Mulders Department of Theoretical Physics, Department of Physics and Astronomy, Faculty of Sciences, VU University, 1081 HV Amsterdam, the Netherlands E-mail: mulders@few.vu.nl

November 2011 (version 6.04)

Contents
1 Introduction 1.1 Quantum field theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.2 Units . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.3 Conventions for vectors and tensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Relativistic wave equations 2.1 The Klein-Gordon equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2 Mode expansion of solutions of the KG equation . . . . . . . . . . . . . . . . . . . . . 2.3 Symmetries of the Klein-Gordon equation . . . . . . . . . . . . . . . . . . . . . . . . . 3 Groups and their representations 3.1 The rotation group and SU (2) . . . . 3.2 Representations of symmetry groups . 3.3 The Lorentz group . . . . . . . . . . . 3.4 The generators of the Poincar´ group . e 3.5 Representations of the Poincar´ group e 4 The 4.1 4.2 4.3 4.4 4.5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1 2 4 8 8 9 10 13 13 15 18 20 21 27 27 29 31 34 36 39 39 39 41 44 44 46 48 49 50 54 54 56 57 60

Dirac equation The Lorentz group and SL(2, C) . . . . . . . . . . . . . Spin 1/2 representations of the Lorentz group . . . . . . General representations of γ matrices and Dirac spinors Plane wave solutions . . . . . . . . . . . . . . . . . . . . γ gymnastics and applications . . . . . . . . . . . . . . .

5 Vector fields and Maxwell equations 5.1 Fields for spin 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2 The electromagnetic field . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.3 The electromagnetic field and topology∗ . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Classical lagrangian field theory 6.1 Euler-Lagrange equations . . . . . . . . . . . 6.2 Lagrangians for spin 0, 1/2 and 1 fields . . . 6.3 Symmetries and conserved (Noether) currents 6.4 Space-time symmetries . . . . . . . . . . . . . 6.5 (Abelian) gauge theories . . . . . . . . . . . . 7 Quantization of fields 7.1 Canonical quantization . . . . . . . . 7.2 Creation and annihilation operators 7.3 The real scalar field . . . . . . . . . 7.4 The complex scalar field . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1

2 7.5 7.6 The Dirac field . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The electromagnetic field . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60 62 66 66 67 68 69 70 73 73 75 78 79 81 82

8 Discrete symmetries 8.1 Parity . . . . . . . . . 8.2 Charge conjugation . . 8.3 Time reversal . . . . . 8.4 Bi-linear combinations 8.5 Form factors . . . . .

9 Path integrals and quantum mechanics 9.1 Time evolution as path integral . . . . . . . . . . . . . 9.2 Functional integrals . . . . . . . . . . . . . . . . . . . 9.3 Time ordered products of operators and path integrals 9.4 An application: time-dependent perturbation theory . 9.5 The generating functional for time ordered products . 9.6 Euclidean formulation . . . . . . . . . . . . . . . . . . 10 Feynman diagrams for scattering amplitudes 10.1 Generating functionals for free scalar fields . . . . 10.2 Generating functionals for interacting scalar fields 10.3 Interactions and the S-matrix . . . . . . . . . . . . 10.4 Feynman rules . . . . . . . . . . . . . . . . . . . . 10.5 Some examples . . . . . . . . . . . . . . . . . . . . 11 Scattering theory 11.1 kinematics in scattering processes 11.2 Crossing symmetry . . . . . . . . 11.3 Cross sections and lifetimes . . . 11.4 Unitarity condition . . . . . . . . 11.5 Unstable particles . . . . . . . . 12 The 12.1 12.2 12.3 12.4 12.5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

85 . 85 . 89 . 92 . 95 . 100 . . . . . . . . . . 103 103 106 107 109 111 114 114 118 122 123 128

standard model Non-abelian gauge theories . . . . . . . . . . . . . . . Spontaneous symmetry breaking . . . . . . . . . . . . The Higgs mechanism . . . . . . . . . . . . . . . . . . The standard model SU (2)W ⊗ U (1)Y . . . . . . . . . Family mixing in the Higgs sector and neutrino masses

3.5 2. Cambridge University Press. Ryder. 4.4 2. 2. Cambridge University Press.2 3. I: Foundations. 1.3 2.3 2. 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). 3.3 3. M. Cambridge University Press. An introduction to Quantum Field Theory.2 6.4 3. Corresponding chapters in books of Ryder. Itzykson and J.2 2. M. 2.2 2. but are rather compact. 1994. Zuber.2.1 5. 1995. The notes contain all essential information.4 3. Cambridge University Press.4 6.1 6.3 3.3 References As most directly related books to these notes.5 5.3 2.1 4. Vol. 1996.H.5 Ryder Peskin & Schroeder Srednicki 1 1 1 1 3 2. 3. McGraw-Hill.8 3.2 1. These notes Section 1. 1995. Veltman. 2007.2 36 22 . Addison-Wesly. Schroeder.1 3.4 4.2 3.3 2. Diagrammatica. 5. 1980. Weinberg.1 2 2 3. Other text books of Quantum Field Theory that are useful are given in refs [3-6].3 2. Cambridge University Press.2 3. Peskin and D. L. S.2 3. Quantum Field Theory. Peskin & Schroeder and Srednicki.4 2.5 2.1 2. I refer to the book of Srednicki [1] and Ryder [2].2 3.2 2.-B. II: Modern Applications. Vol. Quantum Field Theory.E. Srednicki.V. Quantum Field Theory.5 4. 1985.3 4.2 6.3 3.1 2.3 6. C.1 1.2 4.4 3. M. The quantum theory of fields.3 3.7 2.1. 3. 6.3 3.1 3.

1.2 5.3 3.3 12.3 11.6 4.1 9.1 20.3.3.5 6 7 6 6.1 12.8 6.5 11.2 9.6 3. 4.5 6.4 10. 6. 4.4 7.5 9.3 8.2 4.10 8.4 12.7.1 7.4 9. 20.3 6.3 9.2.2.6 8.2. 6.6 9.1 8.7 5.3 7. 9.2 8. 6.6 3.5 7. 6.3 6 8 9 Srednicki 3 3 3 37 5.2 10.4 8.4 3.5 7.3 8.1 20.3.1 11.1 20.2 8.3 7.5 Ryder 4 4 4 4 4 4 Peskin & Schroeder 2.3 2.5. 4.4 4.1 9. 6.2.7 4.6 10.4.6 3.2 12.4 12.1 5.5 4.6 6.5 2.4. 4.8 5 4.4 2.2 11.3 10.3 20.5 9. 9.2 7. 2. 8.5 4.1. 2.1 10.4 These notes Section 7.1.5 .4.

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

In quantum mechanics a special role is played by Planck’s constant h. ˜ ψ(p) = i d3 r exp − p · r ψ(r). usually given divided by 2π. S ≫ .1 Quantum field theory In quantum field theory the theories of quantum mechanics and special relativity are united. (1. being eigenvalues of the position operators. (1.Chapter 1 Introduction 1.5) One can also choose a representation in which the momenta of the particles are the dynamical variables. It is given by pop ψ(r) = −i ∇ψ(r). In quantum mechanics. (1.054 571 68 (18) × 10−34 J s = 6. In this coordinate representation the position operators r op simply acts as rop ψ(r) = r ψ(r). 1 (1. Indeed.582 119 15 (56) × 10−22 MeV s. ˜ The corresponding states are |p and the wave functions ψ(p) = p||ψ are the Fourier transforms of the coordinate space wave functions.6) .3) The uncertainty principle tells us that in this representation the momenta cannot be fully determined. c = 299 792 458 m s−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. the position can in principle be determined at any time with any accuracy.2) In the limit that v ≪ c one reaches the non-relativistic domain. the action of the momentum operator in the coordinate representation is not as simple as the position operator. pj ] = i δij .1) In the limit that the action S is much larger than . (1. ≡ h/2π = 1. They satisfy the well-known (canonical) operator commutation relations [ri . In special relativity a special role is played by the velocity of light c. (1. Corresponding position and momentum operators do not commute. One can talk about states |r and the wave function ψ(r) = r||ψ .4) where δij is the Kronecker δ function. quantum effects do not play a role anymore and one is in the classical domain.

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.

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

0123 0123 ′ ′ ǫ µνρσ ǫµνρσ The first identity.33) (1.27) It is easy to convince oneself of the nature of the indices in the above equation. . constructed via the metric tensor itself. ∂3 ) = ∂ ∂ ∂ ∂ . but more important has the same effect on lefthandside and righthandside.35) ρ ǫµνρσ ǫµν σ σ ǫµνρσ ǫµνρ ′ ′ = −2 g ρρ g σσ − g ρσ g σρ = −6g σσ .Introduction 6 called covariant. = −24. (1.32) . explicitly written as (p0 . defined by ∂2 − ∇2 .31) ǫµνρσ ǫµν ′ ′ ρσ ′ = − (1. g0 = g1 = g2 = g3 = 1. (1. a∞ = 4πǫ0 2 /me e2 . 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 .30) (Note that there are different conventions around and sometimes the opposite sign is used).28) ν ν Note that gµ with one upper and lower index. ′ ′ . .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. starting from ǫ0123 = 1. = ∂ . not affected by Lorentz transformations.34) (1.∇ . The derivative ∂µ is defined ∂µ = ∂/∂xµ and we have a four vector ∂ with components (∂0 . (1. Note that in this equation one has on left. The length squared of ∂ is the d’Alembertian operator.29) 2 = ∂ µ ∂µ = ∂t2 The value of the antisymmetric tensor ǫµνρσ is determined in the same way as for ǫijk . (1. Each of these permutations leads to a minus sign. ∂1 . is timelike with invariant length (squared) p2 = p · p = pµ pµ = E 2 − p2 = m2 . 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. because one has (1. for instance. ∂2 . gµ = gµρ g ρν and 0 1 2 3 is in essence a ’Kronecker delta’. where m is called the mass of the system. ∂t ∂x ∂y ∂z ν ∂µ xν = gµ .1 (a) In the the Hydrogen atom (quantum system) the scale is set by the Bohr radius. . It is an invariant tensor. (1. Relate this quantity to the electron Compton wavelength −e via the dimensionless λ fine structure constant α. p). For the contraction of a vector with the antisymmetric tensor one often uses the shorthand notation ǫABCD = ǫµνρσ Aµ Bν Cρ Dσ . p) = (E. ∂t (1.36) Exercises Exercise 1. Examples are the scalar products above but also relations like pµ = m dxµ /dτ for the momentum four vector.

me . e. 0) ˆ n2 = (0. a1 . n2 . 2. 0. aT ]) are defined through √ a± ≡ (a0 ± a3 )/ 2. Also construct and calculate the Planck length Lpl . 0.(A · B) C using the properties of the tensor ǫijk given in section 1. 1. a+ . Note that ˆ ˆ ˆ ˆ a · nα = aα and derive from that the components nα of the four vector nα . ]. This demonstrates how a careful use of units can save a lot of work. α and me c2 = 0. n1 . a2 ] or [a− . a+ .4 Lightcone coordinates for a four vector a (which we will denote with square brackets as [a− .13 to give its actual value in kg. a1 . a3 ) are the expansion coefficients using the basis vectors n0 . Exercise 1. (b) The coordinates (a0 . 0. 3} the index α can only be equal to one of the indices in ǫµνρσ . 1. a1 . which is the Compton wavelength for the Planck mass.3. .3 Prove the following relation ǫµνρσ g αβ = ǫανρσ g µβ + ǫµαρσ g νβ + ǫµνασ g ρβ + ǫµνρα g σβ . a+ . 0) ˆ n3 = (0. 0. Also for the coordinates [a− .Introduction 7 (b) Relate the classical radius of the electron (a relativistic concept). 1) ˆ . (c) Calculate the Compton wavelength of the electron and the quantities under (a) and (b) using the value of c. 0. by a simple few-line reasoning [For instance: If {µ. Compare it with the proton mass and use Eq. Use that to give the ˆ ˆµ ˆ components of the basis vectors n+ and n− .2 Prove the identity A × (B × C) = (A · C) B . 1. (a) Express the scalar product a · b in lightcone coordinates and deduce from this the values of g++ . (d) Use the value of the gravitational constant GN / c5 = 6. σ} is a permutation of {0. ǫ0 . [Note: what is the MKS unit for V/T?] Exercise 1. One does not need to know . (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). n+ . ν. How are a+ and a− related to a+ and a− . n2 . Exercise 1.71 × 10−39 GeV−2 to construct a mass Mpl (Planck mass).511 MeV. 1. n3 . a2 ] we can find basis vectors n− . but only appropriate combinations. c. n1 . ˆ ˆ . 0) ˆ n0 = (1. g+− and g−+ . . ρ. ˆ ˆ ˆ ˆ These are n1 = (0. a2 . . re = e2 /4πǫ0 me c2 to the Compton wavelength. 0. g−− .

t) = − ψ(r. ∂t p −→ pop = −i∇. But the replacement 2. p2 = pµ pµ = E 2 − p2 = M 2 . The Schr¨dinger equation in quantum mechanics is the operator equation o corresponding to the non-relativistic expression for the energy. 2 + M 2 φ(r. written as pµ −→ i∂µ is covariant (the same in every frame of reference).Chapter 2 Relativistic wave equations 2. the interpretation of this equation as a single-particle equation in which φ is a complex wave function poses problems because the energy spectrum is not bounded from below and the probability is not positive definite. (2.2.3) Equations 2. t) = 0. p).2) Acting on the wave function one finds for a free particle.4) where M is the particle mass. 2M (2. we just want to play a bit with covariant equations and study their behavior under Lorentz transformations. (2. Substitution of operators gives the Klein-Gordon (KG) equation for a real or complex function φ.6) The energy spectrum is not bounded from below: considering the above stationary plane wave solutions. ∂t 2M (2. Although it is straightforward to find the solutions of this equation. t) = ∂2 − ∇2 + M 2 φ(r. t) = exp(−i k 0 t + i k · r).7) k 0 = ± k2 + M 2 = ±Ek . t). one obtains (2.3 are not covariant.5) with (k 0 )2 = k2 + M 2 . E = p2 . namely plane waves characterized by a wave number k. Thus a covariant equation can be obtained by starting with the (covariant) equation for the invariant length of the four vector (E. φk (r. ∂t2 (2.1 The Klein-Gordon equation In this chapter. i ∂ ∇2 ψ(r. (2. 8 .1) under the substitution (in coordinate representation) E −→ Eop = i ∂ .1 and 2.

It is possible to rewrite it as an integration over positive energies only but this gives two terms (use the result of exercise 2. 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 operator has all possible solutions in it multiplied with creation (and annihilation) operators.13) ˜ with (in principle complex) coefficients φ(k). φ(x) = d4 k ˜ 2π δ(k 2 − M 2 ) e−i k·x φ(k) (2π)4 (2.Relativistic wave equations 9 i. As we will see later both problems are related and have to do with the existence of particles and antiparticles. And.3). −k) ≡ b∗ (k) one has φ(x) = d3 k e−i k·x a(k) + ei k·x b∗ (k) = φ+ (x) + φ− (x).8) (2. it is simply a matter of being careful to find a consistent (covariant) theory.2) is j µ = i φ∗ ∂ µ φ or (ρ.5). 2.12) It is easy to see that this current is conserved if φ (and φ∗ ) satisfy the KG equation. which is what Lorentz transformations do. For this we note that an arbitrary solution for the field φ can always be written as a superposition of plane waves.9) They satisfy the continuity equation.11) Therefore.e. Then. k) + ei k·x φ(−Ek . −i φ∗ ∇ φ . At that point the dependence on position r and time t is just a dependence on numbers/parameters on which the operator depends. (2. relativistically the density is not a scalar quantity. ∂ρ = −∇ · j. there are solutions with negative energy. 2. we want the most general solution.14) ˜ ˜ Introducing φ(Ek . (2. ↔ ↔ (2. The appropriate current corresponding to the KG equation (see Excercise 2. This continuity equation can be written down o covariantly using the components (ρ. just as the dependence on time was in ordinary quantum mechanics. having the KG equation as a space-time symmetric (classical) equation. (2π)3 2Ek (2. but rather the zero component of a four vector. is a second order equation and φ and ∂φ/∂t can be fixed arbitrarily at a given time. j) of the four-current j. j) = ↔ i φ∗ ∂0 φ. (2. however. ∂µ j µ = 0.10) ∂t which follows directly from the Schr¨dinger equation. for which we need the interpretation of φ itself as an operator. −k) . φ(x) = d3 k (2π)3 2Ek ˜ ˜ e−i k·x φ(Ek . rather than as a wave function. The KG equation. The integration over k-modes clearly is covariant and restricted to the ‘mass’-shell (as required by Eq.15) . This leads to the existence of negative densities. k) ≡ a(k) and φ(−Ek . 2M 2M (2.2 Mode expansion of solutions of the KG equation Before quantizing fields. there are no longer fundamental objections to mix up space and time.

after restricting the energies to be positive). The coefficients a(k) and b∗ (k) are the amplitudes of the two independent solutions (two.14 and 2. Another symmetry is found by complex conjugating the KG equation. (2π)3 2Ek (2. It is easy to show that x φP (x) = φ(˜) x = = = or since one can define φP (x) ≡ d3 k e−i k·x aP (k) + ei k·x bP ∗ (k) . (2.19) implying that for each solution φ(x) there exists a corresponding solution with the same energy. Since a · b = a · ˜ it is easy to see that ˜ b. They are referred to as mode and anti-mode amplitudes (or because of their origin positive and negative energy modes).3 Symmetries of the Klein-Gordon equation We arrived at the Klein-Gordon equation by constructing a covariant operator (∂µ ∂ µ + M 2 ) acting on a complex function φ. (2. (2π)3 2Ek (2. which implies (xµ ) = (t.21) d3 k x x e−i k·˜ a(k) + ei k·˜ b∗ (k) (2π)3 2Ek d3 k ˜ ˜ e−i k·x a(k) + ei k·x b∗ (k) (2π)3 2Ek d3 k e−i k·x a(−k) + ei k·x b∗ (−k) . 2. This shows how parity transforms k-modes into −k modes.17) (2. Performing some Lorentz transformation x → x′ = Λx. x Transforming everywhere in the KG equation x → x one obtains ˜ ˜ ˜ x ∂µ ∂ µ + M 2 φ(˜) = 0.22) showing that with each solution there is a corresponding charge conjugated solution φC (x) = φ∗ (x).e.15 one has elimated k 0 and in both equations k · x = Ek t − k · x. φP (x) ≡ φ(˜) (P for parity). changing the sign of the spatial coordinates. i. for which we consider space inversion. It is trivial to see that ∂µ ∂ µ + M 2 φ∗ (x) = 0. (2π)3 2Ek (2.23) . In terms of modes one has φC (x) = φ∗ (x) = ≡ d3 k e−i k·x b(k) + ei k·x a∗ (k) (2π)3 2Ek d3 k e−i k·x aC (k) + ei k·x bC∗ (k) . −x) ≡ (˜µ ). 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).16) The consequence of this is discussed in Exercise 2. x ∂µ ∂ µ + M 2 φ(˜) = 0.18) (2.Relativistic wave equations 10 In Eqs 2. one thus must have that the function φ → φ′ such that φ′ (x′ ) = φ(x) or φ′ (x) = φ(Λ−1 x).20) one has for the mode amplitudes aP (k) = a(−k) and bP (k) = b(−k). x) → (t.6 We will explicitly discuss the example of a discrete symmetry. (2.

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

in terms of the a(k) and b(k). .e. i (a) Show that relativistic invariance. d3 x (∂ {0 φ)∗ (∂ i} φ). ∂ ψ(r. We will encounter these quantities later as energy and momentum. j 0 = ψ†ψ and j = ψ † αψ. {αi . (b) Show that one has current conservation for the current 2 We denote the commutator as [A. Note that a{µ bν} indicates symmetrization. β} = 0. an attempt to construct a first order differential equation for the time evolution would be to write. From this one concludes that α and β must be matrix-valued and ψ must be a multi-component wave function.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). B} = AB + BA.7 To solve the problem with positive and negative energies and get a positive definite density. 2. t). B] = AB − BA and the anticommutator as {A. express the quantities E(t) = P i (t) = d3 x (∂0 φ)∗ (∂0 φ) + ∇φ∗ · ∇φ + M 2 φ∗ φ . 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).15. One then finds that Q is expressed as an infinite sum over number operators a† a (particles) and b† b (anti-particles).1) for a complex scalar field current (exercise 2.2) in terms of the a(k) and b(k) using the expansion in Eq. αj } = 2 δ ij I. 1 2 (b) Express now the full charge QV (t) (exercise 2. Exercise 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). 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 . making sure that each component of ψ satisfies the KleinGordon equation (in terms of differential operators ∂µ ∂ µ +m2 = 0) requires the anticommutation relation2 {αi . ∂t where ψ is a wave function and the nature of α and β is left open for the moment. t) = (−i α · ∇ + m β) ψ(r. a{µ bν} ≡ aµ bν + aν bµ . also using the result in (a). (c) Similarly. Exercise 2.5 (a) As an introduction to parts (b) and (c). and β 2 = I.Relativistic wave equations 12 Exercise 2. i.

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

thus is as the surface i=0 of a sphere in 4 Euclidean dimensions. π ) π -sphere (antipodes identified) 2 π -sphere (surface identified) Figure 3. These generators form a threedimensional Lie algebra SO(3). 2 (3. They satisfy the commutation relations [Li .6) Summarizing. n) = 1 cos ˆ 1 Note ϕ ϕ ˆ + i (σ · n) sin 2 2 = exp i ϕ ˆ σ·n .1: the parameter spaces of SO(3) (left) and SU (2) (right). Next. ˆ therefore. ] that satisfies • ∀ x. [y. and thus (σ · n)2 = 1. . [x. In the same way we can consider rotations around the x. is a 3-dimensional Lie-group.7) = a0         0 −1 i 0 1 0 0 1 • [x. [z. π) (n.5) Rotations in general do not commute. namely that there exists a bilinear product [. z ) = lim N →∞ that are generated by  i    0 . It is straightforward to check that any (finite) rotation can be obtained from a combination of infinitesimal rotations. ˆ R(ϕ. y ∈ A ⇒ [x. Locally this is a 3-dimensional Euclidean space and SU (2).2 π) (-n.    Lx =  Ly =  0 0       0 i 0 −i 0 ϕ ˆ .and y-axes     0 0 0   0 0    0 0 −i  . Lj ] = i ǫijk Lk . z]] + [y. (3.4) R . x]). x]] + [z. It is straightforward to check that the conditions require    a0 + i a3 +i a1 + a2    = a0 1 + i a · σ A =   +i a1 − a2 a0 − i a3          1 0          + i a1  0 1  + i a2  0 −i  + i a3  1 0  (3. (3. y]] = 0 (Jacobi identity). These matrices can be defined as acting on 2-component spinors (χ → Aχ) or equivalently as acting on 2 × 2 matrices (B → ABA† ). Writing a0 = cos(ϕ/2) and a = n sin(ϕ/2) we have1 A = A(ϕ. we turn to the group SU (2) of special (det A = 1) unitary (A† = A−1 ) 2 × 2 matrices. • [x. One way of viewing the parameter space. y] = −[y. x] = 0 (thus [x. which reflects itself in the noncommutation of the generators. for rotations around z for instance.z N N or (Lk )ij = −i ǫijk .   0 (3. with real a’s and 3 (ai )2 = 1.8) ˆ that (σ · a)(σ · b) = a · b + i σ · (a × b). y] ∈ A. the rotations in SO(3) can be generated from infinitesimal rotations that can be expressed in terms of a basis of three generators Lx . With matrix commutation this algebra satisfies the requirements for a Lie algebra. Ly and Lz .Groups and their representations 14 (n.

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

m = j(j + 1)|j. it is sufficient to consider the representations Φ of the Lie-algebra G. z )φ(r. . for the function space. pj ] = i δij . j − 1. These are mappings of G into a finite dimensional vector space. . Indeed. J+ = 0 J− = 0 . 3. ϕ + α) = U (α. −j. J− =  2 √           0 0 0 0 2 0 . . The other generators Jx and Jy (or J± ≡ Jx ± iJy ) then transform between the states in V (j) . the Poisson bracket operation satisfies all properties of a Lie algebra. . for rotations the operator Jz and the (quadratic Casimir) operator J 2 . Casimir operators commute with all the generators and the eigenvalue of J 2 can be used to label the representation (j). m . To get explicit representations. Similarly one has the time translation operator φ(t + τ ) = U (τ )φ(t). It may seem trivial. for j = 1/2:  0  1/2 Jz =   0 −1/2 0 0 −1         . m = j. ˆ (3. the commutation relations. The finite dimensional vector space just represents new degrees of freedom. φ(r. In order to find local representations Φ of a Lie-group G. θ. z ) = exp (+iα ℓz ) . . indeed. one looks among the generators for a maximal commuting set of operators. In many cases one can simply construct a suitable Hilbert space or part of it by considering the representations Φ of a group G. ℓj ] = i ǫijk ℓk that are identical to those in Eq.6. 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). θ. From the algebra one derives J 2 |j.     2  . J− =  0      1 0 0  0  . The latter is sufficient to get the commutation relations of the (quantum) rotation operators. m ± 1 with 2j + 1 being integer and m = j. i. m = j(j + 1) − m(m ± 1)|j. . m = m|j. which preserve the Lie-algebra structure.17) (3. e. p) and B(x. J+ =  0 1  . with H = i ∂/∂t and the rotation operator. with pop = −i∇.Groups and their representations 16 are generated by the derivative acting on the functions. The above set of operators H. ˆ U (α. more precise φ(r + a) = U (a)φ(r) where U (a) = exp +ia pop . U (τ ) = exp (−iτ H) . Jz |j. Note that the same requirements would apply to classical mechanics. but it is at the heart of being able to construct a suitable Hilbert space. m with eigenvalues of Jz . −j one has for j = 0:       Jz = 0 . j − 1.18) (3. Somehow the nontrivial structure of rotations should show up and it. pop and ℓop work in Hilbert space and their commutators indeed follow the required commutation relations as symmetry operators.  0 for j = 1:   1 0   Jz =  0 0   0 0     √  0 0  2 √  0   √  0   0       .16) with ℓz = −i∂/∂ϕ = −i(x ∂/∂y − y ∂/∂x) or ℓop = r op × pop . p) (we will come back to this also in Chapter 7).g. Within the (2j + 1)-dimensional representation space V (j) one can label the eigenstates |j. . while J± |j. Explicit representations using the basis states |j. preserving the group structure. m with m-values running from the heighest to the lowest. In particular the quantum operators ℓop obey in quantum mechanics commutation relations [ℓi .e. J =  0 0  +  0 0 . m . . ϕ). does in the nontrivial behavior of Poisson brackets of quantities A(x. just starting off with the (basic) canonical commutation relations [ri . These are mappings from G into the same finite dimensional vector space (its dimension is the dimension of the representation).

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

22) (Note that xT is a row vector). σ/2 in a two-dimensional (spin 1/2) case. In this section we consider the Poincar´ group. it is easy to derive some properties of the matrices Λ: (i) det(Λ) = ±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.24) L↑ + L↓ + L↑ − L↓ − proper orthochrone proper non-orthochrone improper orthochrone improper non-orthochrone . From this property. which as a matrix equation with (Λ)µν = Λµν and (G)µν = gµν gives (ΛT )ρµ (G)µν (Λ)νσ = (G)ρσ . e. Writing x as a column vector and the metric tensor in matrix form. (ii) |Λ00 | ≥ 1. Denoted in terms of column vectors and 4-dimensional matrices one writes for the poincar´ transe formations x′ = Λ x + a. proof: (ΛT GΛ)00 = (G)00 = 1 → Λµ0 gµν Λν0 = 1 → (Λ00 )2 − i (Λi 0 )2 = 1. 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. (3.25) (3.21)  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. (det Λ = +1 is called proper.Groups and their representations 18 3.23) The proper tensor structure of the matrix element (Λ)µν is a tensor with one upper and one lower index. consisting of the Lorentz group and translations. explicitly x′µ = (Λ)µν xν + aµ or x′µ = Λµν xν + aµ .     G =  10 (3. it is convenient to use a vector notation for the points in Minkowski space. e To derive some of the properties of the Lorentz group.27) (3. det Λ = −1 is called improper).26) (3. Thus one has the (pseudo-orthogonality) relation ΛT GΛ = G ⇔ GΛT G = Λ−1 ⇔ ΛGΛT = G.g. proof: det(ΛT GΛ) = det(G) → (det Λ)2 = 1.     0 0   g00 g01 g02 g03   1 0      g     g11 g12 g13   0 −1 0 0  = . We considered the generators and looked for representations in finite dimensional spaces. 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 .

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

we have for the (infinitesimal) rotations (around z-axis) ΛR = I + i ϕ3 J 3 = I − i ω12 M 12 . Thus the two vector operators J and K form under Lorentz transformations an antisymmetric tensor M µν .36) (3. Summarizing.40) ΛB = I + i η 3 K 3 = I − i ω30 M 30 . three of which only involve spatial coordinates (rotations) and three others involving time components (boosts). a). We now want to find a covariant form of the six generators of the Lorentz transformations. Also this group can be divided into four parts. [M µν .39) (e. we continue using covariance arguments. The e 2 These commutation relations are the same as those for the ’quantummechanics’ operators i(xµ ∂ ν − xν ∂ µ ) . M ρσ ] = −i (g µρ M νσ − g νρ M µσ ) − i (g µσ M ρν − g νσ M ρµ ) .4 The generators of the Poincar´ group e ↑ The transformations belonging to P+ are denoted as (Λ.37) The algebra of the generators of the Lorentz transformations can be obtained by an explicit calculation2 .35) thus (ω)ij = −(ω)ji and (ω)0i = (ω)i0 .38) Explicitly. In order to find the commutation relations including the translation generators. (3. explicitly reading x′µ = = (Λ)µν xν + aµ = ((I)µν + (ω)µν ) xν + ǫµ Λµν xν + aµ = (g µν + ω µν ) xν + ǫµ . i i0 i i0 1 ijk M jk 2ǫ (3. We will just require Poincar´ invariance for the generators themselves. The extension to the Poincar´ group is e ↑ straightforward.g. Multiplying the proper orthochrone transformations with one of them gives all Lorentz transformations. P+ etc. with ϕi = − 1 ǫijk ω jk and J i = 2 z-axis). inf (3. infinitesimally approximated by (I + ω. (3. which are obtained by writing the infinitesimal parameters (ω)µν in terms of six antisymmetric matrices (M αβ )µν .33) These four transformations form the Vierer group (group of Klein). 3 30 with η = ω and K = M (e. K = M ).Groups and their representations  0 0  −1 0   0 −1 0  0  =   0  0 −1 0   0 0 0 −1       ∈ L↓   +    20 Is It (space-time inversion) (3. We therefore find (again) that there are six generators for the Lorentz group. There are six generators. the Lorentz transformations form a (Lie) group (Λ1 Λ2 and Λ−1 are again Lorentz transformations.g. This is a + normal subgroup and the factor group is the Vierer group. ǫ).34) The condition ΛT GΛ = G yields g ρµ + ω ρµ gρσ (g σν + ω σν ) = gµν =⇒ ωνµ + ωµν = 0. 2 (M αβ )µν = i g αµ g βν − g α g βµ ν (3. Of the four parts only L↑ forms a group. 3. J 3 = M 12 ) and for the (infinitesimal) boosts (along (3. One can easily convince oneself that i (ω)µν = ω µν = − ωαβ (M αβ )µν .

P j ] = i δ ij H/c2 . We have here reinstated c. Taking any one of the states in an irreducible representation.45) Note that the commutation relations among the generators M µν in Eq. as part of the set.41) 2 At this point. H] = [J i . J j ] = i ǫijk J k . P ) in terms of the Hamiltonian and the three-momentum operators. (1 + iǫγ P γ − i i ωαβ M αβ )P µ (1 − iǫδ P δ + ωρσ M ρσ ) = P µ + ω µν P ν . . H] = 0.44) (3. For instance. In order to label the states in an irreducible representation we construct a maximal set of commuting operators. = pµ |p.46) [K i . . a)P µ U (Λ. that commute e among themselves. . The generators of the translations are the (momentum) operators P µ . K j ] = −i ǫijk J k /c2 . other states in that representation are obtained by the action of operators outside the maximal commuting set. These define states with specified quantum numbers that are eigenvalues of these generators. [J i . [J i .48) . Explicitly. check that U1 U2 corresponds with Λ1 Λ2 ). Of the generators of the Poincar´ group we choose first of all the generators P µ . the generators J± in the case of the rotation group. H] = i P i . P ν ] [M µν . Λµν pν ⇒ ωµν pν ⇒ ωµν = = = pµ + ω µν pν = pµ 0 ǫµνρσ pρ sσ . [J i . because one then sees that by letting c → ∞ the commutation relations of the Galilei group. 3.42) (to decide on where the inverse is. one obtains 2 [P i . They just state that M µν transforms as a tensor with two Lorentz indices. To find other generators that commute with P µ we look for Lorentz transformations that leave the four vector pµ invariant. ν (3. a) = Λ−1 µ P ν . These form a group called the little group associated with that four vector. The requirement that P µ transforms as a four-vector (for which we know the explicit behavior from the defining four-dimensional representation). (3. Infinitesimally. 2 2 (3. The eigenvalues of these will be the four-momentum pµ of the state. P ρ ] = 0. = −i (g µρ P ν − g νρ P µ ) . P j ] = i ǫijk P k . the tensor M µν in terms of boosts cK i = M i0 and rotations J i = 1 ǫijk M jk . known from non-relativistic quantum mechanics are obtained. K j ] = i ǫijk K k . [K i . leads to U (Λ.5 Representations of the Poincar´ group e where pµ is a set of four arbitrary real numbers. the generators M αβ no longer exclusively act in Minkowski space. a)P µ U † (Λ. P j ] = [P i . a) = Λµν P ν or U † (Λ. .43) giving the following commutation relations by equating the coefficients of ǫµ and ωµν . (3. (3. P µ |p. the generators J 2 and J 3 in the case of the rotation group. [P µ .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. (3. ǫ) = 1 − ωαβ M αβ + i ǫα P α .38 could have been obtained in the same way. . . .47) (3. 3. writing the generator P = (H/c. [K i . For instance.

Wα ] = [Wµ . while the others have no physical significance. M ǫijk . which can thus be chosen spacelike. satisfying gµν nµ (p)nν (p) = j i nµ (p) pµ i ǫµνρσ pµ nν nρ nσ i j k = = −δij .55) commute with all generators and therefore are invariants under Poincar´ transformations.56) . p0 > 0. Pν ] = [Wµ . = = i exp − ǫµνρσ M µν pρ sσ |p.49) (3. Massive particles: p2 = M 2 > 0. Given the momentum four vector pµ we choose a tetrade consisting of three orthogonal spacelike unit vectors ni (p). . . .50) where the Pauli-Lubanski operators W µ are given by 1 Wµ = − ǫµνρσ M νρ P σ . From the algebra of the generators one finds that P 2 = Pµ P µ and W 2 = Wµ W µ (3. . .Groups and their representations 22 where s is an arbitrary vector of which the length and the component along pσ are irrelevant. i exp − ωµν M µν 2 i exp − ǫµνρσ M µν pρ sσ 2 (3. (3. . for which we distinguish the following cases p 2 = m2 > 0 p2 = 0 pµ ≡ 0 p 2 = m2 > 0 p2 = 0 p2 < 0 p0 > 0 p0 > 0 p0 < 0 p0 < 0 Only the first two cases correspond to physical states. the third case represents the vacuum. The elements of the little group thus are U (Λ(p)) = = and U (Λ(p))|p. of which the components generate the little group of pµ or (for the third of the following relations) from explicit calculation [Mµν . 0. 2 exp (−isµ W µ ) |p.54) This will enable us to pick a suitable commuting ’spin’ operator.52) (3. 0.53) (3.51) The following properties follow from the fact that W µ is a four-vector by construction. (3. . 2 (3. . These e operators are the Casimir operators of the algebra and can be used to define the representations of ↑ P+ . Wν ] = −i (gµα Wν − gνα Wµ ) . i ǫµνρσ W ρ P σ . .

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. (3. 0. . s. Together with the four momentum states thus can be labeled as |M. −1) n1 (p0 ) ≡ n2 (p0 ) ≡ s(p0 ) ≡ (3. (3. the S i (p) form the generators of an SU (2) subgroup that belongs to the maximal set of commuting operators. W j (p)] = i M ǫijk W k (p). ms . . 1. Noting that S i (p) i 2 = 1 W µ (p)W ν (p) M2 niµ (p)niν (p).63) where E = p2 + M 2 and ms is the z-component of the spin −s ≤ ms ≤ +s (in steps of one).60) and using the completeness relation nµ (p)nν (p) = − g µν − i i 2 i pµ pν M2 . 0. (1. M2 (3. p0 > 0. S j (p)] = −i nµ (p)nν (p) j i ǫµνρσ W ρ P σ = i ǫijk S k (p).57) (i. M J) with components W i = (M/2)ǫijk M jk . The explicit construction of the wave function can be done using the D-functions (analogous as the rotation functions). 0).62) 1 Thus W 2 has the eigenvalues −M 2 s(s + 1) with s = 0.e. 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 . |p3 |). 0). (0. 1. 0.58) i.e.65) . 0. 0. i (3. W i (p) = −W · ni (p)). Having made a covariant decomposition. 1 2 (3. 2 .59) (3. p. 0). We will not do this but construct them as solutions of a wave equation to be discussed explicitly in chapter 4 Massless particles: p2 = M 2 = 0. More generally (fully covariantly) one can proceed by defining S i (p) ≡ W i (p)/M and obtain [S i (p). 1.Groups and their representations 23 We can write W µ (p) = 3 W i (p) nµ (p) i i=1 (3. (0. 3. show that W /M can be interpreted as the intrinsic spin.64) such that in an arbitrary frame where the four vectors p. 0. M2 (3. n1 (p). In that case the vectors ni (p) are just the space directions and W = (0. The commutation relations [W i (p). .61) one sees that S i (p) i =− W2 . it is sufficient to choose a particular frame to investigate the coefficients in Eq.. n2 (p) and s(p) are obtained by a Lorentz boost from the reference frame one has the property ǫµνρσ sµ nν nρ pσ > 0.

The 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. implying continuous values for the spin (actually for W i = M S i ). We don’t know of any physical relevance. (3. where A = exp(i ϕ · σ/2) = exp(iϕ σ · n/2). Wν ]. 3.66) where W · p = 0 implies that W s (p) = −(M 2 /p · s) λ(p). Thus we have P 2 . |p. P µ . 1 2 24 (3. Thus 2 λ(p) = J ·p . i W 2 (p). thus. Note that angular momentum does not contribute to helicity as L · p = 0. W 2 (3. The eigenvalues of W 1 and W 2 (in that case corresponding to the translations) thus can take any continuous values. U (Λ. 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). −i W 1 (p).Groups and their representations The above does actually include the massive case. 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). W µ (p) = λ(p) P µ .70) as a commuting set with zero eigenvalues for P 2 and W 2 . Exercises Exercise 3. 3.73) (3.67 has a vanishing right hand side).1 (a) Derive from the relation that defines the mapping of SU (2) matrices A into SO(3) matrices RA in Eq. W 2 (p)] = = = i M 2 λ(p). A massless particle.68) (3. however. which can take any integer or half-integer value. is characterized by its momentum and the helicity. This state is in physical applications nondegenerate and is denoted by |0 .69) We reproduce the algebra for the massive case with W 3 (p) = M λ(p). W 1 (p)] [λ(p). The vacuum: pµ = 0. the action of RA b · RA a = = ˆ ˆ ˆ b · a cos ϕ + n · (b × a) sin ϕ + 2(b · n)(a · n) sin2 (ϕ/2) ˆ ˆ b · a + b · (ˆ × n) sin ϕ − 2 (b · a − (b · n)(a · n)) sin2 (ϕ/2) a . W 2 and λ is derived from the general commutation relations for [Wµ . λ .72) One has P µ |0 = W µ |0 = 0 and the state is invariant under Lorentz transformations. and gives [W 1 (p). This algebra is (for instance) isomorphic to the one generated by rotations and translations in a 2-dimensional Euclidean plane (see exercises). But if M = 0 one has a different algebra (Eq. σ · RA a ≡ A(σ · a)A−1 ˆ (valid for any vector a). W 2 (p)] [λ(p). (3. |p| (3. The algebra of W 1 . a)|0 = |0 .71) which is called the helicity.67) (3.13.

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

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

K = +i . i ǫijk B k . (4. 2 (4.1 The Lorentz group and SL(2. becoming the defining representations of SL(2. (j. They will be labeled by two angular momenta corresponding to the A and B groups. B j ] = i ǫijk Ak . 2 2 (4.5) = 1 (J + i K). C) can be written as a product of a unitary + matrix U and a hermitean matrix H. it also follows directly that the Lorentz group is homeomorphic with the group SL(2. 0) Type II : (0. 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. B j ] = [Ai . respectively. where we restrict (for fixed n) the parameters 0 ≤ ϕ ≤ 2π and 0 ≤ η < ∞. The group SL(2.8) ˆ ˆ ˆ with η = η n and ϕ = ϕn.Chapter 4 The Dirac equation 4.9) (4. j) K = −i J (B = 0). It is simply connected and forms the covering group of L↑ . K = −i . similarly as the homeomorphism between SO(3) and SU (2). Special cases are the following representations: Type I : (j. K = i J (A = 0). C) is the group of complex 2 × 2 matrices with determinant one.2) This shows that the Lie algebra of the Lorentz group is identical to that of SU (2) ⊗ SU (2). C): Type I (denoted M ): Type II (denoted M ): 27 J= σ σ .7) From the considerations above. The matrices in SL(2.3) (4. This tells us how to find the representations of the group. C).4) (4. Aj ] = [B i . With this choice of parameter-spaces the plus and minus signs are actually relevant. 0. j ′ ).6) (4. U (ϕ) H(η) (4.10) . They precisely correspond to the two types of representations that we have seen before. M = exp i ϕ·σ 2 1 exp ± η · σ 2 = U (ϕ) H(η) .1) (4. 2 1 (J − i K). 2 2 σ σ J = .

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

22) u=     η η M (Λ) 0 where M (Λ) = ǫM ∗ (Λ)ǫ−1 with ǫ given in Eq. the well-known two-component spinor for a spin 1/2 particle. The representing member of the class P− is the parity or space inversion operator Is . e. 0) of SL(2. −→ −r (vector). (pseudoscalar). a) and ˜ a = (a0 . the boost in Eq. (vector). The i angular momentum operators J are represented by σ i /2. We also have seen that the representations (0. the aij elements of a tensor will not change sign. For the spin 1/2 representations of the Poincar´ group including parity we.      0   χm   uR   H(p)      .21) In nature parity turns (often) out to be a good quantum number for elementary particle states. but rather the operators W i (p) should be used.17.23) u(p.20) 4. consider the four component spinor that transforms under a Lorentz transformation as      0  ξ   ξ  . ǫµνρσ will change sign. i. as we have seen in section 4. Examples (with between brackets the Euclidean assigments) are r t p H J λ(p) K −→ −λ(p) −→ −K −→ H −→ J −→ t (scalar). 0) are inequivalent. (axial vector). A parity transformation changes aµ into aµ . we obtain for the two components of u which we will refer to as chiral right and chiral left components.2 Spin 1/2 representations of the Lorentz group 1 1 Both representations (0. C). 2 ) and ( 2 .       = (4. The behavior is the same for classical quantities. we have seen that in the rest frame W i (p)/M → J i . 0) (via operators J and K) one sees that under parity 2 2 1 1 (0. etc. that have the correct commutation relations. From the definition of the representations (0. but by a transformation belonging ↑ ↑ to the class P− . m) =            χm uL H(p) 0 . For a particle at rest only angular momentum is important and we can choose ξ(0.e. m) = χm . they cannot be connected by a unitary 2 2 transformation. 4.   −→  M (Λ)   (4. where a = (a0 .13. i. ) −→ ( . This provides the easiest way of seeing ˜ what is happening. + −   1 0    0 1  I2 =    0 0   0 0 0 0 −1 0 0 0 0 1      . Within the Lorentz group. and the angular momentum in the rest frame is what we are familiar with as the spin of a particle.g. −→ −p (vector). −a).      (4. but to L↑ . they can be connected. 4.1. C) are suitable for representing spin 1/2 particles. 2 2 (4. (scalar). Thus M ∗ and M are not equivalent within SL(2. therefore. must combine the repe resentations. 0). m) = η(0. It has the same effect as lowering the indices.The Dirac equation 29 where The matrix I2 does not belong to L↑ . generators. Taking M (Λ) = H(p). 1 ) and ( 1 . Although these operators cannot be used to label the representations. 1 ) and ( 1 .e.

The general requirements for the γ matrices are thus easily obtained. (pµ γ µ − M ) u(p) ≡ (/ − M ) u(p) = 0. = H −1 (p). (4. To see this consider the equation for the hermitean † † conjugate spinor ψ † (noting that γ0 = γ0 and γi = −γi ). .29) which is an explicit realization of the (momentum space) Dirac equation. γ } ∂µ ∂ν + M 2 ψ(x) = 0.27) where γ µ are 4 × 4 matrices called the Dirac matrices1 .7.26)        −2    uL uL H (p) 0 or explicitly in the socalled Weyl representation  −M E+σ·p       E−σ·p −M    uR      uL      = 0. which in general is a linear equation in pµ .32) suppressing on the RHS the identity matrix in Dirac space. From this one obtains the Clifford algebra for the Dirac matrices. (iγ µ ∂µ − M ) ψ(x) = 0. Since ∂µ ∂ν is symmetric. We first want to investigate if the Dirac equation solves the problem with the negative densities and negative energies. The first indeed is solved. (4.31) To achieve also that the energy-momentum relation p2 = M 2 is satisfied for u(p). ψ † −iγ 0 ∂0 +iγ i ∂i −M 1 We ← ← = 0. 2 (4. (4. The explicit realization appearing in Eq.The Dirac equation 30 with H(p) = ¯ H(p) = 2M (E + M ) E+M −σ·p 2M (E + M ) E+M +σ·p . this can be rewritten 1 µ ν {γ . 4. It is of a form that we also played with in Exercise 2. γ ν } = 2 g µν . We will discuss another explicit realization of this algebra in the next section.25) It is straightforward to eliminate χm and obtain the following constraint on the components of u. {γ µ .33) define for a four vector a the contraction a = aµ γµ / .      0 H 2 (p)   uR   uR        . p (4.30) γ µ γ ν ∂µ ∂ν + M 2 ψ(x) = 0. Applying (iγ µ ∂µ + M ) from the left gives (4. one must require that for ψ(x) the Klein-Gordon relation 2 + M 2 ψ(x) = 0 is valid for each component separately).27 is known as the Weyl representation.28) which is a covariant (linear) first order differential equation. (4. This leads to the Dirac equation for ψ(x) = u(p)e−i p·x in coordinate space.   (4.24) (4. As in section 2 we can use Pµ = i∂µ as a representation for the momenta (translation operators) in function space.  =       (4.

1: The Dirac sea of negative energy states and antiparticles.37) Using the explicit form of γ in the Weyl representation (see Eq.g.  . The second problem encountered before.34) for the adjoint spinor ψ ≡ ψ † γ0 .3 General representations of γ matrices and Dirac spinors {γ µ .38) The general algebra for the Dirac matrices is Two often used explicit representations are the following2 : The standard representation:     σk   1 0   0 0 3 k 2 k  . 4.The Dirac equation 31 E empty levels M -M filled levels E particle M -M hole Figure 4. remains. This problem was solved by Dirac through the introduction of a negative energy sea. From the vacuum a particle can be removed. From the equations for ψ and ψ one immediately sees that j µ = ψγ µ ψ is a conserved probability current (exercise!). E = +M (twice) and E = −M (twice). We will see how this is properly implemented (also for bosons!) when quantizing fields.35) is indeed positive and j 0 can serve as a probability density. opposite charge). γ =ρ ⊗1= γ = iρ ⊗ σ =    0 −1 −σ k 0 2 We (4. The Dirac sea forms the vacuum. Relying on the Pauli exclusion principle for spin 1/2 particles. γ ν } = 2 g µν . (4. Dirac supposed that the negative energy states were already completely filled and the exclusion principle prevents any more particles to enter the sea of negative energy states. where γ 0 p0 ψ = M ψ 0 −→ Eψ = M γ 0 ψ. one sees that there are two positive and two negative eigenvalues. Multiplying with γ 0 on the right and pulling it through one finds a covariant equation ψ(x) iγ µ ∂µ +M ← = 0. . This hole forms an antiparticle with the same mass. 4. The density j 0 = ψγ 0 ψ = ψ † ψ (4. but with properties such that it can be annihilated by the particle (e. the one of the negative energy states. This can most easily be seen in the particle rest frame.39) use ρi ⊗ σj with both ρ and σ being the standard 2 × 2 Pauli matrices. (4. (4.27).36) (4.

the rotation and boost generators in Weyl representation in Eqs 4.  (γ5 )S. For example.49) i 1 with −i S ρσ = 2 σ ρσ = − 4 [γ ρ .43) S=√  .10 are   1  σ3 0  i   J 3 = M 12 =   = [γ 1 .40) (4. −→ Sγµ S −1 . 2 (4.R.48) are projection operators.R. For instance in the Weyl representation (but valid generally).41) (4. while L−1 γ µ L = Λµ γ ν .42) We note that the explicit matrix taking us from the Weyl representation to the standard representation.R.  0 (4.44) and an adjoint spinor defined by ψ = ψ † γ0 . γ 0 = ρ1 ⊗ 1 =   1 0 γµ ψ Different representations can be related to each other by unitary transformations. γ 2 ]. 0 σ3 2 4   1  −iσ 3 0   = − i [γ 3 .The Dirac equation 32 The Weyl (or chiral) representation:    0 1   . γ µ } = 0 and explicitly one has    0 1  1 . 4! (4.9 and 4.45) (γ5 )W. K 3 = M 30 =    0 iσ 3 2 4 or in general one has the transformation i ψ → Lψ = exp − ωρσ S ρσ ψ. (4. −→ Sψ. is   1  1 1    (4. that project out chiral right/left states. Lorentz invariance The Lorentz transformations can also be written in terms of Dirac matrices. S −1 . γ 0 ]. (γµ )S. ν The latter assures Lorentz invariance of the Dirac equation (see Exercise 4.R. = ρ ⊗ 1 =   1 0 i ǫµνρσ µ ν ρ σ γ γ γ γ .  =ρ ⊗1=  0 −1 3 (4. We note that ψ → Lψ and ψ → ψ L−1 . 1 −1 2 For all representations one has † γµ = γ0 γµ γ0 .3) .   0 γ k = −iρ2 ⊗ σ k =  k  σ  −σ k  .46) (4. which in the case of the Weyl representation are just the upper and lower components. γ σ ]. 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 . = S(γµ )W.47) (4. it is easy to see that PR/L = 1 (1 ± γ5 ) 2    1 0  .

(C)W. ∂ (4.g.51) The existence of positive and negative energy solutions implies another symmetry in the Dirac equation.The Dirac equation 33 Parity There are a number of symmetries in the Dirac equation.R. parity. ∂ with the solution ψ c (x) = ηc Cψ (x).53) (4. (i/ − M ) ψ(x) = 0.R. reverse the charge of the particle. The latter implies that the conjugate of a right-handed spinor.R. ψR . x = (t. x) → x = (t. γ5 ] = 0. e. for instance.    −iσ 2   0 −ǫ  = .50) Charge conjugation but with ψ p (x) ≡ ηp γ0 ψ(˜) (Exercise 4. that brings ψ → ψ c . In any representation a matrix C exist. but it does. Explicitly. in Weyl representation we find in Dirac space     T C  ξ   −ǫ η ∗     . It is easy to convince oneself that if ψ(x) is a solution of the Dirac equation. =    0 ǫ iσ 2 (4. Note that we have (as expected) explicitly in Weyl x representation in Dirac space     P  ξ    −→ ψ p = γ 0 ψ =  η  .g. ∂ (4. (C)S. usually to be taken unity. (or W. 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. One has ψ c = −ψ T C −1 . This symmetry does not change the spin 1/2 nature.    0 −ǫ 0    0   −ǫ 0  . C is real and one has C −1 = C † = C T = −C c and [C. T (4. called charge conjugation.58) where ηc is an arbitrary (unobservable) phase. such that T Cγµ C −1 = −γµ .R.57) (4. T (4.55) (4.56) (i/ − M ) ψ c (x) = 0. As with parity we look for a transformation. = =   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. −x) and via a few manipulations obtain again the ˜ Dirac equation (i/ − M ) ψ p (x) = 0.54) e. Some properties of C are C −1 = C † and C T = −C. In S. ψ= (4.59)  −→ ψ c = C ψ =    η ǫ ξ∗ .   ψ= (4.) and all representations connected via a real (up to an overall phase) matrix S.52)    η ξ one can apply space inversion. is a left-handed spinor.6).

s). s)u(p. the spinors satisfy C uT (p. s)¯(p. s) + v(k. s) u = . with E = Ep = + p2 + M 2 . 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. It is easy to check that   1  1±σ·s 1 ± γ5 / s ˆ 0    (4. In that frame is a spacelike unit vector sµ = (0. s) v = .1).    (4. in which γ 0 is diagonal (see discussion of negative energy states in section 4. ψ(x) = u±s (p) e−i p·x . ψ(x) = v ±s (p) ei p·x .65) (4. where u satisfies   −σ · p  Ep − M      u(p) = 0. s) = u(p. ¯ (4. be obtained by a Lorentz transformation. s)v(p.69) = − s In order to project out spin states. s) ei k·x d∗ (k. 2M 2M −/ + M p v(p.. p (4.62) where χs are two independent (s = ±) two-component spinors. where v satisfies   σ·p  −(Ep + M )      v(p) = 0. s) = u(p. s) e−i k·x b(k. s ) = 0. s ) = v (p.g. s′ ) = 2M δss′ .     ∂ −iσ · ∇ −i ∂t − M (4. Note that the spinors in the negative energy modes (antiparticles) could be two different spinors.The Dirac equation 34 4. v (p.   ⇔ (/ − M ) u(p) = 0.60)   ψ(x) = 0.64) (4.66) An arbitrary spin 1/2 field can be expanded in the independent solutions. s)u(p. (2π)3 2Ek (4.4 Plane wave solutions Looking for positive energy solutions ∝ exp(−iEt) one finds two solutions.61)     σ·p −(Ep + M ) There are also two negative energy solutions. v(p. s). the spin polarization vector in the rest frame is the starting ˆ point. s) = v(p. s)¯(p. Choosing χ = −ǫχ∗ (the equivalent spin ¯ 1/2 conjugate representation). s′ ) = 2Ep δss′ .   ⇔ (/ + M ) v(p) = 0. The explicit Dirac equation in the standard representation reads  ∂  iσ · ∇   i ∂t − M       (4. s) and C v T (p. s′ ) = −2M δss′ . the best representation to describe particles at rest is the standard representation.63) u(p. s)v(p. s) . The ¯ ¯ solutions are normalized to u(p. p     −σ · p (Ep − M )     Ep + M     σ ·p ¯ Ep +M χs χs ¯ For a free massive particle. s) = Ep + M      σ ·p   Ep +M χs     . ¯ ′ ′ Explicit solutions in the standard representation are   χs       .68) (4. ¯ ¯ † ′ u (p. 2M 2M (4.70) =  Ps =  ˆ 0 1∓σ·s 2 2 . In an arbitrary frame one has s · p = 0 and s(p) can e.67) It is straightforward to find projection operators for the positive and negative energy states P+ P− = s p /+M u(p. s)u(p. s)v(p. s ) = v † (p.

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

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

s). e. Do this e. 4 2 Exercise 4. x ˜ . −x) is also a solution of the Dirac equation. γ σ ] = σ ρσ . 2M (E + M ) ˆ ˆ where E = M γ = M cosh(η) and p = M βγ n = M sinh(η) n. of which only the traces of four and two gamma-matrices are nonzero.6 Apply space-inversion. to the Dirac equation and use this to show that the spinor ψ p (x) = ˜ γ0 ψ(˜).77) Exercises Excercise 4. Show that the equality holds. by proving that H 2 (p) = exp (η · σ) = (E + σ · p)/M = σ µ pµ /M. 1 i −i S ρσ = − [γ ρ . s) u p /+M = 2M 2M both sides are projection operators.13. H(p) = exp η·σ 2 = M +E+σ·p .g. a 4 a · b.1 Prove Eq. x → x. 4.4 Show that in P+ ≡ s=±1/2 u(p. where x = (t.3 Imposing Lorentz invariance on the Dirac equation can be achieved by requiring L−1 γ µ L = Λµ γ ν .5 Show starting from the relation {γµ . by letting them act on the four basis spinors u(p. Exercise 4. γν } = 2gµν that γµ / γ µ a = Tr(//) = ab µn µ1 Tr(γ · · · γ ) = −2 /.g. i S ρσ ] = g ρµ γ σ − g σµ γ ρ . s)¯(p. 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. Excercise 4. 0 if n is odd. ν µ µ Look at the infinitesimal forms Λµ = gν + ων and L = 1 − (i/2)ωρσ S ρσ to obtain ν [γ µ . ˜ Exercise 4. (4.2 Show that j µ = ψγ µ ψ is a conserved current if ψ en ψ are solutions of the Dirac equation Exercise 4. s) and v(p.

e.57 MeV. i. (d) Plot for these cases the vector densities ψ † ψ and the scalar densities ψψ as a function of r/R. Note: use first the appropriate equation (which?) that is obeyed by each individual component of the Dirac equation to show that the upper two components are allowed solutions. |M (π − → µ− νµ )|2 ¯ Actually if the lepton would be massless. M R = 1. n (c) Apply the confinement condition to find the value of k for the lowest eigenmode in the spherical ˆ cavity. 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. we want to confine this solution to a spherical cavity with radius R.7 (a) Show that (in standard representation)  j0 (kr)χm    ψ(x. mµ = 105. the decay would even be forbidden. Calculate for now just |M |2 .    is a stationary solution of the (free) Dirac equation. (b) Next. Show that the ratio is |M (π − → e− νe )|2 ¯ ≈ 54.8 A negatively charged pion (π − ) decays in a lepton ℓ (ℓ can be an electron or a muon) and a corresponding anti-neutrino (¯ℓ ). Our starting point is the transition amplitude given by M = ℓ(kℓ )¯ℓ (kν )|Hint |π(p) = GF fπ u(kℓ )/(1 − γ5 ) v(kν ). ν p Note: how such parametrizations work will be explained in chapter 8. which is the most important factor needed to find the decay width (inverse lifetime). t) = N     k ˆ i E+M σ · r j1 (kr)χm    −i Et   e .8 × 10−6 . you need to determine the corresponding confinement condition for ψ. there is no current flowing through the surface of the cavity. Note: To prove this. Excercise 4. me = 0. Show that this n condition is sufficient to guarantee that nµ j µ = 0. For such a cavity nµ = (0. r). the lepton and neutrino have spin 1/2 and are ν described by momentum-space spinors. The pion has spin 0. Do you understand why? The calculation of the life-time is completed in chapter 11.511 MeV). We will not pursue this chiral freedom at this point and take α = 0. . n It turns out that there is actually a whole family of conditions that provide confinement. namely i / ψ = eiαγ5 ψ.5 and M → ∞.The Dirac equation 38 Exercise 4. The conditions for ψ and ψ imply that nµ j µ = ψ/ ψ = 0. A condition for confining the fermion in the cavity is i / ψ = ψ (where n is the surface normal).66 MeV. Note: the outcome shows that the decay into muons is much more likely that the decay into electrons (mπ = 139. Calculate the value of k for M R = 0.

3) Since the vectors ǫµ (k) together with the momentum k µ form a complete set of four-vectors. just as the two-component Dirac spinors did for spin 1/2. namely ǫµ (k0 ) = (0. µ (2π)3 2E µ (5.1 Fields for spin 1 In section 3.7) .1) The Lorentz condition implies that k µ ǫµ (k) = 0. (2 + M 2 )Vµ = 0. where k = k0 = (M. In an arbitrary frame ǫµ (k) with λ = 0. λ) .5)  −E    B2   . These are solutions of the Klein-Gordon equation with in addition a condition.2) (5.± (λ) d3 k ǫ(λ) (k) e−ik·x c(k.2 we have discussed the vectors ǫλ (with λ = 0 and ±1) as basis states of the spin 1 representation. one has ǫ(λ) (k) ǫ(λ)∗ (k) = −gµν + µ ν kµ kν . A real vector field Vµ (x) then can be expanded in modes as Vµ (x) = λ=0. These can describe spin 1 states in the rest-frame. 0). with ∂µ V µ = 0 (Lorentz condition). (5. The vectorfield is therefore (λ) suited to describe spin 1.± 5.  −B 1    0 3 With the electromagnetic field tensor (5. In the restframe. A). there are then indeed three independent possibilities. 39 (5.6) where Aµ is the four-vector potential (5. ±(1) are obtained by boosting the restframe vectors. λ) + ǫ(λ)∗ (k) eik·x c∗ (k.Chapter 5 Vector fields and Maxwell equations 5.4) λ=0. The covariant generalization is to describe them with Vµ (x) = ǫµ (k) e±ik·x .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. M2 (5. ǫλ ).

only two independent vectors remain ǫ(±) µ (k) = (0.15 the d’Alembertian is replaced by 2 + M 2 . if in Eq. ∂µ Aµ = 0. while the fact µ C ∗ that the Minkowski space is real implies that Aµ (x) = Aµ (x) = Aµ (x). i.15) (5. The equations of motion for the fields Aµ become 2Aµ − ∂µ (∂ν Aν ) = jµ . . |k|). The Lorenz condition can actually be incorporated by changing the equations of motion to 2Aµ + (λ − 1)∂µ (∂ν Aν ) = jµ . (5. Therefore if k µ = (|k|. it does not eliminate the freedom of gauge transformations but they are now restricted to Aµ → Aµ + ∂µ χ with 2χ = 0. Under a gauge transformation Aµ −→ Aµ + ∂µ χ.5. ∓1. (5. there is still a gauge freedom left (Aµ → Aµ + ∂µ χ with 2χ = 0). transverse or Coulomb gauge). 40 (5. (5. (5. shows that the electromagnetic fields correspond to the case of a massless spin 1 field. 5. in accordance with what we found in section 3. Fµν −→ Fµν + (∂µ ∂ν χ − ∂ν ∂µ χ) = Fµν . Furthermore.9) 1 ˜ where F µν ≡ − 2 ǫµνρσ Fρσ is the tensor where E → B and B → −E. The gauge freedom can first of all be used to impose the Lorentz condition for the electromagnetic field.e. 0. (5. Although the Lorenz condition is a constraint. it will vanish at all times. In that case one also has ∇·A = 0 or ki ǫi (k) = 0. ˜ x Note that via parity transformation one obtains another solution. It states the absence of magnetic charges and Faraday’s law. AP (x) = Aµ (˜). 0. For electrodynamics one has furthermore the freedom of gauge transformations. moree over.Vector fields and Maxwell equations and the (conserved) current j µ = (ρ.13) Hence. 2Aµ = 0. the antisymmetry of F µν implies that ˜ ∂µ F µν = 0 or ∂λ Fµν + ∂µ Fνλ + ∂ν Fλµ = 0.16) (5. Note that the incorporation of the Lorentz condition can also be done for the case of the massive vector field. The equation in vacuum. As remarked above.12) (5. j) the Maxwell equations read ∂µ F µν = j ν .11) This equation is not affected by a gauge transformation.8) (5. One possible choice is the gauge A0 = 0 (5.14) of which the solutions give the Li´nard-Wiechert potentials. for M = 0.10) the electric and magnetic fields are unchanged. −i.17) (radiation. Taking the divergence of this equiation. if ∂µ Aµ = 0 for large times |t|. in which case one has 2Aµ = jµ . 0).18) corresponding to the helicity states λ = ±1 of the massless photon field. one finds for a conserved current and λ = 0 2∂µ Aµ = 0.

but A = 0 (hence there is structure in the vacuum). F = e E + ev × B (5.1: The Aharonov-Bohm experiment 5.20) e (∇ × A) · ds = B · ds = where Φ is the flux through the solenoid. nevertheless. ˆ B = 0. illustrated in fig. A= By Bx Br ϕ= − ˆ . cos ϕ. . To be precise. In the presence of an electromagnetic field Aµ an additional phase difference is observed. 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 . where Aµ = 0. This occurs in the region outside the solenoid. . sin ϕ.19) are gauge invariant. = 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.0 .Vector fields and Maxwell equations 41 L ∆x d a (infinitely) long solenoid Figure 5. The situation. It is however. Nevertheless the significance of Aµ is shown in the Aharonov-Bohm experiment. This phase difference is actually observed. The energy density 2 (E 2 + B 2 ) = 0 (hence a vacuum). The phase difference between the electrons travelling two different paths is given by δ = 2π a a ∆x d = −≈ .0 . 2 2 2 BR2 BR2 y BR2 x A= . 0). . This causes an interference pattern as a function of ∆x.1. In this experiment an observable phase difference is measured for electrons moving through field-free space (E = B = 0 but Aµ = 0). may look a bit akward. λ L − λ λ or ∆x = (L− λ/d)δ. but where E = B = 0. 5. (5. Inside solenoid: Outside solenoid: B = B z. ϕ= − ˆ 2r 2r2 2r2 where r = (cos ϕ.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. while the electromagnetic field Aµ is not. 0) and ϕ = (− sin ϕ. nothing more than a manisfes1 tation of a nontrivial vacuum (a la the Dirac sea for fermions).

minimizing the space occupied by B fields. illustrating also that the group structure of the mapping of U (1) into this space is that of the group of integers Indeed when one considers the A-field in the example of the Aharonov-Bohm effect outside the solenoid. Now back to the Aharonov-Bohm experiment.21) (ϕ being the azimuthal angle).21. Below the critical temperature the magnetic field is squeezed out of the superconducting material. Therefore the gauge transformation must be uniquely defined in the space one is working in.Vector fields and Maxwell equations U(1) S (space with solenoid) 42 class 0 ei φ class 1 Π1(S) = Z class 2 Figure 5. nonobservable phases ϕ = 2π n. The difference here is that we are working in a space with a ’defect’ (the infinitely long solenoid). i. the number of times they go around the defect) cannot be continuously deformed into one another. By going around an arbitrary loop a different number of times the electron wave function would acquire different phases or it would acquire an arbitrary phase in a point by contracting a loop continuously into that point. 5. it can be obtained from the situation A = 0 by a gauge transformation. A = ∇χ with χ= BR2 ϕ 2 (5. however. Consider a superconductor. 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. Therefore space outside the solenoid doesn’t care that χ is multi-valued. If this would be be happening in empty space one would be in trouble. the relevant charge turns out to be 2e because the electrons appear in Cooper pairs . This situation that A can be written as ∇χ is always true when ∇ × A = 0. organizing itself in tiny flux tubes (Abrikosov strings). implying that it must be single-valued1 . giving no observable phase2 .2: The topology of the space in the Aharonov-Bohm experiment. is multivalued as ϕ = 0 and ϕ = 2π are the same point in space. 1 The occurrence of nontrivial possibilities. In such a space loops with different winding number (i. has been employed by Dirac in constructing magnetic monopoles in electrodynamics 2 Actually. forming a simple (connected) space. The AharonovBohm experiment shows a realization of this possibility.e.e. The observable phase going around in general is ∆δ = e A · dr = e ∇χ · dr = e χ|ϕ=0 . ϕ=2π The function χ in Eq. Another situation in which the topology of space can be used is in the case of superconductors.

8 and 5. . 5. ∂t ∇ · B = 0. ∂t ∂B = 0. ∇×B =j+ ∇×E+ ∂E .1 Show that Eqs.9 explicitly give the equations ∇ · E = ρ.Vector fields and Maxwell equations 43 Exercises Exercise 5.

1) and as an example the lagrangian L(x. ˙ 2 (6. and the total change x′ (t′ ) = x(t) + ∆x(t). t1 (6. The requirement δS = 0 with fixed boundaries x(t1 ) = x1 and x(t2 ) = ˙ x2 leads to t2 ′ (6. As an example. These equations can be obtained from a lagrangian using the action principle.1 Euler-Lagrange equations In the previous chapter we have seen the equations for scalar fields (Klein-Gordon equation). Dirac fields (Dirac equation) and massless vector fields (Maxwell equations) and corresponding to these fields conserved currents describing the probability and probability current. x) ˙ (6.8) 44 . x (t) = x(t) + δx(t). ∂x ˙ (6.6) The first term leads to the Euler-Lagrange equations. x) = K − V = ˙ 1 mx2 − V (x). recall classical mechanics with the action t2 S= t1 dt L(x.4) δS = t1 t2 dt dt t1 ∂L ∂L δx + δ x + L δτ ˙ ∂x ∂x ˙ d ∂L − ∂x dt ∂L ∂x ˙ δx + t2 t1 = ∂L δx + L δτ ∂x ˙ t2 . ∂x (6.5) with ∆x(t) = δx(t) + x(t) δτ .Chapter 6 Classical lagrangian field theory 6.3) (6. thus t′ = t + δτ. d dt ∂L ∂x ˙ = ∂L . ˙ p= ∂L .7) The quantity ∂L/∂ x plays a special role and is known as the canonical momentum.2) The principle of minimal action looks for a stationary action under variations in the coordinates and time. (6.

which in classical mechanics vanishes at the boundary) does not play a role. − d3 x . + ∂L ∆x − H δτ ∂x ˙ t2 . giving1 δS = = δL δL δ(∂µ φ) + δφ + δ(∂µ φ) δφ R δL δL δφ + d4 x − ∂µ δφ δ(∂µ φ) R d4 x dσµ L δxµ ∂R dσµ ∂R δL δφ + L δxµ . λ=0 (6. t1 (6. gives rise to a conserved quantity (at least if H is invariant!) Q=p dx dλ −H dt dλ . ∂µ φ(x)) = R d4 x L (φ(x). (6. The hamiltonian H is defined by H(p. .11) λ=0 of which p and H are the simplest examples related to space and time translations. but which in 3 dimensions includes conserved quantities related to rotations and boosts. x). = φ(x) + δφ(x) (6.t1 ) (6. . 6. ∂µ φ′ . ∂µ φ(x)).e. Variations in the action can come from the coordinates or the fields. . ∆x = (dx/dλ)λ=0 δλ and δt = (dt/dλ)λ=0 δλ. think for instance of a temperature or density distribution or of an electromagnetic field). . i. In classical field theory one proceeds in complete analogy but using functions depending on space and time (classical fields.t2 ) d3 x . (R3 .18) 1 Taking a functional derivative. L (φ(x). ˙ (6.13) (6. .2.6 can be rewritten as δS = .10) One sees that invariance under time translations requires that H(t1 ) = H(t2 ). In general any continuous transformation. R (6. The resulting variation of the action is δS = R µ ′ = xµ + δxµ . H is a conserved quantity.17) Furthermore the variations δφ and δ∂µ φ contribute to δS. . . = ∂R (R3 . We will come back to it in a bit more formal way in section 9.Classical lagrangian field theory 45 For the lagrangian specified above this leads directly to Newton’s law p = m¨ = −∂V /∂x. indicated with δF [φ]/δφ should pose no problems. their derivatives and possibly on the position. also indicated by ∂R (a more general volume in four-dimensional space-time with some boundary ∂R can also be considered). δ(∂µ φ) (6. .9) which is done because the first term (multiplying ∆x. ˙ x The second term in Eq. with ∆φ(x) = δφ(x) + (∂µ φ)δx . Consider a lagrangian density L which depends on these functions.15) d4 x′ L (φ′ . ∂R Note for the specific choice of the surface for constant times t1 and t2 . ∂µ φ.14) (6.16) The change in variables x → x′ in the integration volume involves a surface variation of the form dσµ L δxµ . x) and an action t2 S[φ] = t1 dt L = dt d3 x L (φ(x). . dσµ . indicated as x′µ φ (x) or combined φ′ (x′ ) = φ(x) + ∆φ(x). x′ ) − d4 x L (φ. x) ≡ p x − L . t2 ). t1 ) and R3 . ∂µ φ(x).12) Here R indicates a space-time volume bounded by (R3 . .

25) (6.19) δ(∂µ φ) δφ 6. (6.Classical lagrangian field theory 46 With for the situation of classical fields all variations of the fields and coordinates at the surface vanishing. δL δL ∂µ = .21) which differ only by surface terms. (6.22) which can be considered as the sum of the lagrangian densities for two real scalar fields φ1 and φ2 √ with φ = (φ1 + iφ2 )/ 2. ∂ (6. δL δ(∂µ ψ) δL δψ one obtains immediately i / −M ∂ → i = − γ µψ 2 i → = ∂ / ψ − M ψ. (6. the Dirac field (spin 1/2) and the vector field (spin 1) can be found.29) .2 Lagrangians for spin 0.23) (6. 2 ψ = 0.26) The Dirac field The appropriate lagrangian from which to derive the equations of motion is L = i → i ← i ↔ ψ / ψ − M ψψ = ψ / ψ − ψ / ψ − M ψψ ∂ ∂ ∂ 2 2 2 = ψ (i/ − M ) ψ. The integrand of the first term must vanish. The scalar field It is straightforward to derive the equations of motion for a real scalar field φ from the lagrangian densities. 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). 2 ∗ (6. leading to the Euler-Lagrange equations. One easily obtains (2 + M 2 )φ(x) = 0. For the complex scalar field one conventionally uses L = = ∂µ φ∗ ∂ µ φ − M 2 φ∗ φ −φ∗ ∂µ ∂ µ + M 2 φ. (2 + M )φ (x) = 0.20) (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). the second term is irrelevant.24) (6. leading to (2 + M 2 )φ(x) = 0. 2 (6. Using the variations in ψ (in the symmetric form). L = 1 1 ∂µ φ∂ µ φ − M 2 φ2 2 2 1 µ = − φ ∂µ ∂ + M 2 φ.27) (6.

31 is given by LM (Dirac) = −M ξ † η + η † ξ .g. The Majorana case is in fact more general.38) (6.33) We now introduce a ’real’ (four-component) spinor satisfying Υc = Υ with ψL . 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 . Υi / Υ − ∂ 4 2   T  −ǫ η ∗  c  = ΥR . 2 αβ = −βα. . but note the factor 1/2 as compared to the Dirac lagrangian. In that case one finds trivially L = ↔ ↔ 1 1 ψR i / ψR + ψL i / ψL − M (ψR ψL + ψL ψR ). 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.30) It is often useful to link to the two-component spinors ξ and η which we started with in chapter 4. namely2 LM (Majorana) = + (6. (6. the mass term in the Dirac lagrangian 6. that the lagrangian separates into two independent parts for M = 0. L = = ↔ 1 1 c c M ψL ψL + M ∗ ψL ψL ψL i / ψL − ∂ 2 2 ↔ 1 1 M ΥR ΥL + M ∗ ΥL ΥR .34)  (6. or equivalently separate the field into right.and lefthanded ones.7). Peshkin and Schroeder or Exercise 12. We note that of which the left part coincides   . Using the twospinors ξ and η. and thus (β ∗ α)∗ = α∗ β = −βα∗ .32) There exists another possibility to write down a mass term with only one kind of fields. which comes because we in essence use ’real’ spinors. ψL ≡ (ψL )c = C ψL =    0 (6. (6.g.31) showing e. 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.    ∗  0    ⇒ Υ ≡  −ǫ η  ΥL = ψL =    η η ∗ for which η0 = ǫ η0 .35) Since the kinetic term in L separates naturally in left and right parts (or ξ and η). ∂ ∂ 2 2 (6. (6.Classical lagrangian field theory 47 and similarly from the variation with respect to ψ ψ i / +M ∂ ← = 0.37) This gives an expression with a mass term that is actually of the same form as the Dirac mass term.36) (6.39) 2 This term is written down with η being an anticommuting Grassmann number for which (αβ)∗ = β ∗ α∗ = −α∗ β ∗ . The reasons for Grassmann variables will become clear in the next chapter. 2 (6. 1 M η † ǫη ∗ − M ∗ η T ǫη .

∂µ J µ (x) = 0. ∆φ = (dφ/dλ)λ=0 δλ and δxµ = (dxµ /dλ)λ=0 δλ. ψ(x) → ei eΛ ψ(x) or ∆ψ(x) = i eΛ ψ(x). In this case. Q= d3 x J 0 (x. which in a consistent picture precisely generate the symmetries.47) ..40) 6. (6. the presence of a symmetry that leaves the lagrangian invariant. Returning to δS the surface term is rewritten to δS = R d4 x . which is the space-integral over the zero-component of a conserved current.42) Θµν = The variation δS thus can be expressed as δS with J µ (x) = − = δL ∂ ν φ − L g µν .3 Symmetries and conserved (Noether) currents Next. these conserved quantities become operators. t).45) J µ (x) = − δ(∂µ φ) dλ λ=0 dλ λ=0 Thus considering ∂R = σ2 . (6.σ1 with σ = (R3 . δ(∂µ φ) For continuous transformations.. requires the presence of a conserved quantity. t). .46) For quantum fields (φ operator!) things become more subtle as it it is not possible to specify ˙ for instance φ and φ on the surface σ1 .18. + ∂R dσµ dσµ ∂R ≡ where d x . δ(∂µ φ) (6. . implying ∂µ F µν + M 2 V ν = 0 and ∂µ V µ = 0. These do not affect the dynamics and will give rise to conserved currents. (6.41) (6. 6. In particular when the lagrangian is invariant under symmetries. 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. δ(∂µ φ) (6. we get (omitting the parameter δλ) dxν dφ δL + Θµν (x) . (6. we consider the second term in Eq. As an example for the classical field.Classical lagrangian field theory 48 where Fµν = ∂µ Vν − ∂ν Vµ .44) δL ∆φ + Θµν (x)δxν . (6.+ R 4 δL δL ∆φ − ∂ ν φ − L g µν δxν δ(∂µ φ) δ(∂µ φ) δL ∆φ − Θµν (x)δxν . one can consider the variations at the initial or final surface. consider U (1) transformations of the Dirac field proportional to the charge e. discussed in the next chapter. Q(t1 ) = Q(t2 ).43) F (σ1 ) − F (σ2 ) = dσµ J µ (x) = ∂R R d4 x ∂µ J µ (x).

we obtain (since δxµ = 0).g.52) The behavior of the field under translations (δφ) is governed by the translational behavior of the argument in such a way that ∆φ = 0.3) j µ = i e φ∗ ∂µ φ.51) (6.50) (6. Translations µ Under translations (generated by Pop ) we have δxµ ∆φ(x) δφ(x) = ǫµ .54) (6. Lorentz transformations In this case the transformation of the coordinates and fields are written as δxµ ∆φi (x) = = ω µν xν . . (ω µν antisymmetric) 1 ωρσ (−i S ρσ )ij φj (x). = 0. 6.48) 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. The integral over the zero-component.4 Space-time symmetries One kind of symmetries that leave the lagrangian invariant are the Poincar´ transformations. e. (6. = −(∂µ φ)δxµ = −ǫµ ∂µ φ(x).Classical lagrangian field theory From the lagrangian for the Dirac field. From Noether’s theorem one sees that the current Θµν ǫν is conserved. Summarizing. (6.56) δxµ = ωρσ (aρσ )µν xν 2 with (aρσ )µν = aρσ µν = g ρµ g σν − g σµ g ρν . In the next section we will see the quantity show up as the charge operator.53) These are the energy and momentum. t) is the conserved charge (Q = 0). (6.49) These currents are conserved as discussed already in chapter 2.57) For the fields the behavior under Lorentz transformations has been the subject of previous chapters. Therefore there are four conserved currents Θµν (ν labeling the currents!) and four conserved quantities Pν = dσµ Θµν = d3 x Θ0ν . (6. H = d3 x Θ00 (x). ↔ (6. 1 (6.55) Note that the coordinate transformations can be written in a form similar to that for the fields (matrices in internal space). 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. 2 (6. including e the space-time translations and the Lorentz transformations. For quantized fields this will become the expressions of the hamiltonian and the momentum operators in terms of the fields.

however. one has a tensor that satisfies (exercise 6.65) M µρσ = T µρ xσ − T µσ xρ . φ(x) φ (x) ∂µ φ(x) ∗ → e → e → ei eΛ(x) φ(x). where H ρµν is the quantity appearing in M ρµν .67) i.63) µν = ∂µ Θ . −i eΛ(x) ∗ i eΛ(x) (6.3 we have seen global transformations or gauge transformations of the first kind. In some applications (specifically coupling to gravity) it is advantageous to have an equivalent current that is symmetric. 2 (6. (6.g.59) (6.Classical lagrangian field theory 50 • Scalar field φ: −i Sρσ = 0.69) i eΛ(x) φ (x).64) (6. (6. independent of x. ∂µ φ(x) + i e ∂µ Λ(x) e (6. (6. 6. γσ ] This result for the Dirac field has been discussed in chapter 3 (see also Exercise 4. • Dirac field ψ: −i Sρσ = −(1/4)[γρ . ∂µ T µν (6. φ(x).68) (6. Gauge transformations of the second kind or local gauge transformations are transformations of the type φ(x) → ei eΛ(x) φ(x). Any lagrangian containing derivatives. e. is not invariant under local gauge transformations.5 (Abelian) gauge theories φ(x) → ei eΛ φ(x). the angle of the transformation depends on the space-time point x.66) In section 6.3).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 .58) (6.62) with Gρµν = (H ρµν + H µνρ + H νµρ )/2. • Vector field Aµ : −i Sρσ = aρσ .61) A final note concernes the symmetry properties of Θµν .70) . (6. (6. In general this tensor is not symmetric. 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 . Defining T µν = Θµν − ∂ρ Gρµν . The lagrangians which we have considered sofar are invariant under global gauge transformations and corresponding to this there exist a conserved Noether current.e.6) T µν = T νµ . in which the U (1) transformation involves an angle e Λ.

For this purpose it is necessary to introduce a vector field Aµ .Classical lagrangian field theory 51 where the last term spoils gauge invariance. As an example consider the Dirac lagrangian. 4 (6. D ∂ A (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.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 . Dµ φ) − Fµν F µν .76) We note first of all that the coupling of the Dirac field (electron) to the vector field (photon) can be written in the familiar interaction form Lint = −e ψγ µ ψ Aµ = −e j µ Aµ . 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. ∂ 2 4 (6. 1 L (φ. 2 i ∂ / ψ − M ψ − e/ ψ A 2 giving the Dirac equation in an electromagnetic field. (6.72) (6. (6. (i/ − M ) ψ = (i/ − e/ − M ) ψ = 0. The equation of motion for the fermion follow from δL δ(∂µ ψ) δL δψ = = i − γ µ ψ. (6. i L = ψγ µ ∂µ ψ − (∂µ ψ)γ µ ψ − M ψψ. Dµ φ(x) ≡ (∂µ + i eAµ (x))φ(x).71) Thus the covariant derivative has the correct transformation behavior provided Aµ → Aµ − ∂µ Λ. ∂µ φ) =⇒ L (φ.78) . 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).73) (6.75) The field φ is used here in a general sense standing for any possible field. 2 Minimal substitution ∂µ ψ → (∂µ + i eAµ )ψ leads to the gauge invariant lagrangian L = 1 i ↔ ψ / ψ − M ψψ − e ψγ µ ψ Aµ − Fµν F µν .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). −e j µ Aµ = −e ρφ + e j · A.

i. where j µ = e ψγ µ ψ. giving the Maxwell equation coupling to the electromagnetic current. ↔ Exercise 6. 4 (b) What is the form of the interaction term involving a current jµ and the field Aµ that will give the inhomogeneous Maxwell equations. Exercise 6.3 Show that the current jµ = i φ∗ ∂µ φ for a complex scalar field is connected to a U (1) transformation on the fields. ∂µ j µ = 0 (Note that the addition of a total derivative to the lagrangian density does not modify the equations of motion).1 (a) Show that the Klein-Gordon equation for the real scalar field can be derived from the lagrangian density 1 1 L = ∂µ φ∂ µ φ − M 2 φ2 . ∂µ F µν = j ν . φ → eiΛ φ. 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.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. (c) Show that the interaction term is invariant under gauge transformations only if the current jµ is conserved. give the equation which is satisfied by ψ c = Cψ . (6. ∂µ F µν = j ν .e. Exercise 6.79) Exercises Exercise 6. [iγ µ (∂µ − i e Aµ ) − M ]ψ = 0. .4 Given the Dirac equation for a (negatively charged) spin-1/2 particle in an external electromagnetic T field. −eψγ ν ψ.Classical lagrangian field theory 52 For the photon the equations of motion follow from δL δ(∂µ Aν ) δL δAν = = −F µν .

This relation is known as the Pauli equation. Comment: this relation is known as the Pauli equation. 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. 2me Exercise 6. e 1 (−i∇ + eA)2 + σ · B − eA0 ψu (r.6 (optional) Prove the properties of the tensor T µν in section 6.4. Hint ≡ −gs µe · B = −gs Qe s · B. t) = Enr ψu (r.5 Show that the Dirac equation for an electron with charge −e in an external electromagnetic field Aµ = (A0 . use ∂µ T µν = 0 and ∂µ M µρσ = 0. A) (see also exercise 6. . 2M 2M with Enr ≡ E − M ≪ M .4) for a stationary solution in the non-relativistic (|Enr | ≪ M ) and weak-field (|A0 | ≪ M ) limits yields in the standard representation for the ’upper (two) components’ ψu . in order to show that Θρσ − Θσρ = ∂µ H µρσ . finally note that it is sufficient to show the last equation for dσµ M µρσ .Classical lagrangian field theory 53 Exercise 6. Hints: it may be useful to realize that Gµρσ is antisymmetric in first two indices. t).

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

the discretization procedure above is an explicit example.2. The hamiltonian then is H= x px qx − L = ˙ = x where H (x) = δL ˙ φ(x) − L (x) = Θ00 (x). Φ(f ) = d4 x φ(x) f (x). (7. px (t) = ∂L ∂Lx = ∆3 x = ∆3 x Πx (t).15) with furthermore the relations [φ(x.12) (7.6) ∆3 x L x (qx . qx+∆x ) . (7.13) 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.9) Note that this indeed corresponds to the zero-zero component (Θ00 ) of the conserved energy-momentum stress tensor Θµν . The lagrangian is given by L = = x In order to construct the hamiltonian it is necessary to find the canonical momenta. t)] = i δ 3 (x − y). t)] = [Π(x. a) φi (x)U (Λ. 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. Π(y. ∂µ φ(x)) = ˙ d3 x L (φ. py (t)] = i δxy or for the fields φ(x) and Π(x) the socalled equal time commutation relations [φ(x. ∆φ. 2 2 2 (7. • Poincar´ invariance e The full variation of the fields. ˙ where Πx (t) is obtained from the continuous field Π(x) ≡ δL /δ(∂0 φ).8) (7. qx . t). φ. for the scalar field theory. t). t)] = 0. discussed in the section 6. for symmetry transformations. formally. t). (7. a) = Rj (Λ−1 ) φj (Λ−1 x − a). i U † (Λ. Sometimes it is useful to keep in mind that. ∂ qx ˙ ∂ qx ˙ (7.16) . φ(y. the fields φ(x) can be considered as regular operators in the Hilbert space after smearing with a test function f . ˙ δ φ(x) (7.11) (7. albeit with ’sharp’ functions. ∇φ) (7. Π(y. As an example.10) x ∆3 x [Πx qx − L x ] ˙ ∆3 x Hx (7. The essential conditions to consistently quantize a theory then are the following. δ∂0 φ) 1 1 1 = Θ00 (x) = (∂0 φ)2 + (∇φ)2 + M 2 φ2 .7) d3 x L (φ(x).14) In fact. • Canonical commutation relations The quantization of the theory is achieved by imposing the canonical quantization condition [qx (t). This implies specific transformation properties in the Hilbert space for the quantum fields φ.

1 Q = √ (a + a† ) 2ω and P = −i ω (a − a† ) 2 (7. In simplified form the hamiltonian is given by H= 1 2 1 2 2 P + ω Q .19) 7. sometimes referred to as second quantization.Quantization of fields 56 or if U (Λ.22) it is straightforward to check that the commutation relations between Q and P are equivalent with the commutation relations [a. Mµν ] = i i∂µ φi (x).23) The hamiltonian in this case can be expressed in the number operator N = a† a. [Q. (7. a] = [a† . i(xµ ∂ν − xν ∂µ )φ (x) + i (7. [a. (7. P ] = i. • Causality Operators Φ(f ) and Φ(g) for which the test functions are space-like separated can be measured simultaneously (macroscopic causality). and [N.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. ω a† a + 2 2 ω i − [Q. Q] = [P.21) Writing Q and P in terms of creation (a† ) and annihilation (a) operators. 2 i ωµν (S µν )i . a] = −a.18) with S µν given in 6. [φ(x).20) where the coordinate Q and the momentum P satisfy the canonical commutation relations [Q.4. a† ] = 0. (7. a† ] = a† . H = = ω ω P P Q−i √ Q+i√ 2 2 2ω 2ω 1 1 =ω N+ . This must be valid for the – now operators – generators P µ and M µν that in the previous chapter have been found in terms of the fields through Noether’s theorem. P ] = 0 (7.25) . a) = i Rj (Λ−1 ) = 1 + iǫµ P µ − 1− i ωµν M µν .24) It is straightforward to find the commutation relations between N and a and a† . The measurements cannot influence each other or [Φ(f ). Microscopic causality implies local commutativity. 2 2 (7. [N. φ(y)] = 0 if (x − y)2 < 0. j 2 one has [φi (x). Pµ ] = [φ (x). a† ] = 1. Φ(g)] = 0. j (7.17) (Sµν )i φj (x). P ] 2 (7.

28) (7.like’ commutation relations between a(k) and a† (k ′ ). a† (k ′ )] = 0.e.30) [φ(x). Note that we will often write a(k) or a† (k). 7.3 The real scalar field We have expanded the (classical) field in plane wave solutions. but one needs to realize that in that case k 0 = Ek = k2 + M 2 .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. (2π)3 2Ek (7.Quantization of fields 57 Defining states |n as eigenstates of N with eigenvalue n. Π(x )]x0 =x′0 = 0. which we have split into positive and negative energy pieces with (complex) coefficients a(k) and a∗ (k) multiplying them. N |n = n|n one finds N a† |n N a|n = = (n + 1) a† |n . 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. The quantization of the field is achieved by quantizing the coefficients in the Fourier expansion. and we see that a state |0 must exist for which N |0 = a|0 = 0. a(k ′ )] = [a† (k).33) . i. a† (k ′ )] = (2π)3 2Ek δ 3 (k − k′ ) [a(k). e.g. which represents the ’number of particles’ with momentum k. d3 x φ(x) i ∂0 e−i k ·x ↔ ′ ↔ (7.4 for the classical field) a(k) a† (k ′ ) = = d3 x ei k·x i ∂0 φ(x).e. The canonical momentum becomes −i d3 k ˙ Π(x) = φ(x) = a(k) e−i k·x − a† (k) ei k·x .27) 2 (2π)3 It is easy to check that these equations can be inverted (see Exercise 2.32) (7. φ(x) = (2π)3 2Ek where the Fourier coefficients a(k) and a† (k) are now operators. (7.26) a(k) e−i k·x + a† (k) ei k·x . [φ(x). the real scalar field φ(x) becomes d3 k (7. and (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 . φ(x )]x0 =x′0 = [Π(x). From the normalizations one obtains a |n = √ n + 1 |n + 1 and a|n = n |n − 1 .31) The hamiltonian can be rewritten in terms of a number operator N (k) = N (k) = a† (k)a(k). † √ a and a act as raising and lowering operators. is equivalent with [a(k). Π(x′ )]x0 =x′0 = i δ 3 (x − x′ ) ′ ′ and (7.

This term will be adressed below. (2π)3 2Ek d3 k a† (k) ei k·x . = normalized as (a† (k1 ))n1 (a† (k2 ))n2 √ √ . 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 ). The rest of the states are then obtained from the groundstate via the creation operator.40) This procedure is known as normal ordering. a(k)|0 = 0.42) (7. For the purpose of normal ordering it is convenient to decompose the field in positive and negative frequency parts. |0 .38) d4 k 2π δ(k 2 − M 2 ) θ(k 0 ) |k k| = (2π)4 (7. which now contains an infinite number of oscillators. 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).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).45) (7. . Just as in the case of the harmonic oscillator it is essential (axiom) that there exists a ground state |0 that is annihilated by a(k).44) p|φ− (x)|0 = ei p·x . (2π)3 2Ek (7. (2π)3 2Ek (7. d3 k |k k|.39) The problem with the vacuum or zero-point energy. . n1 ! n2 ! (7. (2π)3 2Ek (7. (2π)3 (7. The 1-particle wave function e−ip·x is obtained via 0|φ(x)|p p|φ(x)|0 = = 0|φ+ (x)|p = d3 k 0|a(k)a† (p)|0 e−i k·x = e−i p·x (2π)3 2Ek (7. writing all annihilation operators to the right of the creation operators. (2π)3 2Ek (7.37) (7. and multiparticle states |(k1 )n1 (k2 )n2 .36) and satisfying the completeness condition 1 = k|k ′ = (2π)3 2Ek δ 3 (k − k′ ).Quantization of fields 58 where the necessity to commute a(k)a† (k) (as in the case of the quantum mechanics case) leads to a zero-point energy.46) (7. i. φ(x) φ+ (x) φ− (x) = = = φ+ (x) + φ− (x). . . d3 k a(k) e−i k·x .41) (7.35) where the vacuum contribution disappears because of rotational symmetry. assuring that the vacuum (by definition) has eigenvalue 0!. defining particle states |k = |k (with positive energy. |k = a† (k)|0 .e. .34) where V = (2π)3 δ 3 (0) is the space-volume. is solved by subtracting it as an (infinite) constant.

(2π)3 2Ek ≡ i∆− (x − y). φ+ (y)] d3 k e−i k·(x−y) .51) (7. d3 k ei k·(x−y) = −i∆+ (y − x).53) t=0 .e.55) d4 k ǫ(k 0 ) 2π δ(k 2 − M 2 ) e−i k·x . φ(y)] which has the properties (i) i∆(x) = i∆+ (x) + i∆− (x) can be expressed as i∆(x) = where ǫ(k 0 ) = θ(k 0 ) − θ(−k 0 ).54) = i∆(x − y) = i (∆+ (x − y) + ∆− (x − y)) .48) (7. (7. (7.47) ′ d3 k ′ e−i k·x+i k ·y [a(k). = − (2π)3 2Ek (7. (iv) The equal time commutation relations follow from ∂ ∆(x) ∂t (v) For M = 0.49) (7. 2π (7.17) and (7. (2π)4 d4 k − θ(k 0 ) 2π δ(k 2 − M 2 ) ei k·x (2π)4 d4 k − θ(−k 0 ) 2π δ(k 2 − M 2 ) e−i k·x = −i∆+ (−x) (2π)4 (7. (ii) ∆(x) is a solution of the homogeneous Klein-Gordon equation. i∆+ (x) i∆− (x) = = = d4 k θ(k 0 ) 2π δ(k 2 − M 2 ) e−i k·x . In order to calculate [φ(x). d3 k = (2π)3 2Ek = (7. φ− (y)] ≡ i∆+ (x − y). (2π)4 (7.18).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.52) The result for the invariant commutator function is [φ(x). that they satisfy the required commutation relations in Eqs (7. The last item to be checked for the scalar field are the causality condition. ∆(x) = − ǫ(x0 ) δ(x2 ). a† (k ′ )] 3 2E ′ (2π) k [φ− (x).56) = −δ 3 (x).50) or as integrals over d4 k. (iii) ∆(0. i. φ(y)] consider µν [φ+ (x). x) = 0 and hence ∆(x) = 0 for x2 < 0.

extending it with the charge operator and the introduction of particle and antiparticle operators. (2π)3 2Ek (7.57) (7.60) = i φ∗ ∂µ φ. a† (k ′ )] = [b(k). which is equivalent to the relations (only nonzero ones) [a(k).62) (7. Q = = = i d3 x : φ† ∂0 φ − ∂0 φ† φ(x) : d3 k : a† (k)a(k) − b(k)b† (k) : (2π)3 2Ek d3 k a† (k)a(k) − b† (k)b(k) .5 The Dirac field L = ↔ i ψγ µ ∂µ ψ − M ψψ. (2π)3 2Ek d3 k b(k) e−i k·x + a† (k) ei k·x . Also the charge operator requires normal ordering (in order to give the vacuum eigenvalue zero).66) . 2 From the lagrangian density (7.58) and satisfy the equal time commutation relation (only nonzero ones) [φ(x). ↔ (7.e. (7.Quantization of fields 60 7. we will consider it as a repetition of the quantization procedure. = ∂{µ φ∗ ∂ν} φ − L gµν .61) (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 . φ† (y)] = i∆(x − y).59) (7. particles (created by a† ) and antiparticles (created by b† ) with the same momentum contribute equally to the energy.63) i. (2π)3 2Ek (7. The hamiltonian is as before given by the normal ordered expression H = = = d3 x : Θ00 (x) : d3 k Ek : a† (k)a(k) + b(k)b† (k) : (2π)3 2Ek d3 k Ek a† (k)a(k) + b† (k)b(k) .4 The complex scalar field In spite of the similarity with the case of the real field. (2π)3 2Ek (7.65) 7. ∂0 φ† (y)]x0 =y0 = i δ 3 (x − y).64) The commutator of φ and φ† is as for the real field given by [φ(x). b† (k ′ )] = (2π)3 2Ek δ 3 (k − k′ ).

s)u† (k.e. The solution is the introduction of anticommutation relations. b† (k ′ . s)e−i k·x .78) (7. s)e−i k·x + d† (k.70) where the last line is obtained by using the Dirac equation. u v (2π)3 2Ek (7. . then leads to additional minus signs and H Q = = d3 k Ek b† (k)b(k) + d† (k)d(k) (2π)3 2Ek d3 k b† (k)b(k) − d† (k)d(k) . ↔ i ψγµ ∂ν ψ − 2 (7.77) Note that achieving normal ordering. (2π)3 2Ek d3 k b† (k.79) Also for the field and the canonical conjugate momentum anticommutation relations are considered. s) e−i k·(x−y) j s † + vi (k. (7. s). † {ψi (x).67) i ψ / ψ − M ψψ gµν .69) (7. The quantized fields are written ψ(x) = s d3 k b(k. ˙ δ ψ(x) ↔ i Θ00 (x) = − ψγ i ∂i ψ + M ψψ 2 = iψγ 0 ∂0 ψ = iψ † ∂0 ψ. ∂ 2 ↔ (7. s)ei k·x + d(k. s′ )} = {d(k. s). s)¯(k. s)v(k. s) ei k·(x−y) x0 =y 0 . s′ )} = (2π)3 2Ek δ 3 (k − k′ ) δss′ .75) (7. (7.Quantization of fields 61 the conserved density and energy-momentum currents are easily obtained. ψj (y)}x0 =y0 = d3 k (2π)3 2Ek ui (k.74) (7. i. s)vj (k.68) The canonical momentum and the hamiltonian are given by Π(x) H (x) = = δL = i ψ † (x). {b(k. (2π)3 2Ek (7. interchanging creation and annihilation operators. s)¯(k.71) (7.76) which seems to cause problems as the antiparticles (d-quanta) contribute negatively to the energy and the charges of particles (b-quanta) and antiparticles (d-quanta) are the same. s)u(k. d† (k ′ .73) (7. jµ Θµν = = ψγµ ψ.72) ψ(x) = s In terms of the operators for the b and d quanta the hamiltonian and charge operators are (omitting mostly the spin summation in the rest of this section) H = = Q = = d3 x : ψ † (x) i∂0 ψ(x) : d3 k Ek : b† (k)b(k) − d(k)d† (k) : (2π)3 2Ek d3 x : ψ † ψ : d3 k : b† (k)b(k) + d(k)d† (k) : . (2π)3 2Ek (7. s)ei k·x .

84) (7. one has † {ψi (x). When we would have started with commutation relations for the field ψ and the canonical momentum. 4 Π0 Πi δL = 0.82) where i∆1 = i∆+ −i∆− . ψ j (y)} at arbitrary times one has {ψi (x). but we must restrict ourselves to the weaker condition B|∂µ Aµ |A = 0. 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).Quantization of fields 62 Using the positive and negative energy projection operators discussed in section 4. (7.87) If one wants to impose canonical commutation relations ∂µ Aµ = 0 cannot hold as an operator identity. ˙ δ A0 δL = F i0 = E i . ∂ (7. (7. ˙ δ Ai (7.83) From the lagrangian density the canonical momenta are = = (7.80) 3 = δ 3 (x − y) δij . −λ(∂ρ Aρ ). as the vanishing of Π0 induces a constraint.86) (7. ψ j (y)] = (i/x + M )ij i∆1 (x − y).88) . we would have obtained [ψi (x). it implies the Lorentz constraint and leads to the canonical momenta Π0 Π i = = E i. but has the problem of being noncovariant. 4 2 This gives the equations of motion discussed before. 7. For the scalar combination {ψi (x).85) which reflects the gauge freedom. ∂ (7.81) where i∆ = i∆+ + i∆− is the same invariant commutator function as encountered before. 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. It is possible to continue in a covariant way with the lagrangian 1 λ L = − Fµν F µν − (∂µ Aµ )2 .6 The electromagnetic field 1 L = − Fµν F µν . ψ j (y)} = d3 k (/ + M )ij e−i k·(x−y) + (/ − M )ij ei k·(x−y) k k (2π)3 2Ek d3 k e−i k·(x−y) + −ei k·(x−y) = (i/x + M )ij ∂ (2π)3 2Ek = (i/x + M )ij i∆(x − y).

what are [ri . Aν (y)]x0 =y0 = i gµν δ 3 (x − y). c† (k ′ . i. rj ]P . and are equivalent with [c(k. It gives 3 k µ ǫ(λ) (k)c(k. The canonical equal time commutation relations are [Aµ (x). ′ (7. which reproduce the Hamilton equations. 7. λ)ei k·x . λ) − c† (k. Πν (y)] = 0.93) which does exhibit problems with the time-like photon.11. 0) .e. (b) Proof for a quantity O(p. [ℓi . where Aµ + + is the part of the vector field containing the annihilation operators. pj ]P . where Q is the conserved quantity in Eq. Aν (y)] = [Πµ (x). λ)|A = 0. and [ℓi . λ). (c) Calculate the Poisson brackets for the coordinates x and the standard conserved quantities p and ℓ. λ)e−i k·x + c† (k. j = 1.94) Choosing k µ = (|k 3 |.95) Exercises Exercise 7. Πν (y)]x0 =y0 = i gµν δ 3 (x − y). In fact the commutation relations imply ˙ [Aµ (x). (7. λ′ )] = −g λλ 2Ek (2π)3 δ 3 (k − k′ ). a longitudinal photon and two transverse photons. pj ]P . for which it is sufficient that ∂µ Aµ |A = 0. (7.89) with four independent vectors ǫµ . i. 0)c(k. containing a time-like photon. µ λ=0 (7.e.Quantization of fields 63 for physical states |A and |B . The quantized field is expanded as Aµ (x) = d3 k (2π)3 2Ek (λ) 3 ǫ(λ) (k) c(k. µ λ=0 (7. k 3 ) this reads |k 3 |c(0) (k) − k 3 c(3) (k) |A = 0 c(0) (k) ∓ c(3) (k) |A = 0. (d) Use Poisson brackets to calculate q and p. q) that dO/dλ = [O.91) (7. The hamiltonian in terms of the creation and annihilation operators is (after normal ordering) given by H= d3 k Ek (2π)3 2Ek 3 λ=1 c† (k. 0.3) satisfy the conditions of a Lie algebra.1 (optional) If you haven’t seen Poisson brackets in classical mechanics you should (a) Proof that Poisson brackets (Eq. 2. Q]P . λ)c(k. (7. 6. 0. These problems are solved by the Lorentz constraint between physical states given above. ˙ ˙ . one time-like photon by itself is not allowed! This solves the problems with the normalization and the negative energies.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.90) where Πν = F ν0 − λ g ν0 (∂ρ Aρ ) and we have furthermore [Aµ (x).

(b) Check the above commutation relation [φ(x). ∆hom (Λx) = ∆hom (x) if f is an invariant function.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. f (Λk) = f (k) for Lorentz transformations Λ. (Similarly one can show that for the advanced Green’s function ∆A (x) = 0 if x0 > 0). Pµ ] for the (real) scalar field using the expressions for the fields and momentum operator in terms of creation and annihilation operators. 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. Is the subtraction of ’zero point’ contributions essential in this check? µ Pµ φ(x) ei a µ Pµ = φ(x − a) Exercise 7. Pµ ] = i∂µ φ(x). (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. . we can write for ∆R an integral where the k 0 integration remains along the real axis. (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. (e) Show that ∆R (x) = 0 if x0 < 0.Quantization of fields 64 Exercise 7. e−i a requires [φ(x). It is invariant. (2 + M 2 ) ∆inhom (x) = −δ 4 (x).4 (Green’s functions) (a) Show that ∆hom (x) = 1 (2π)4 d4 k e−ik·x δ(k 2 − M 2 )f (k) is a solution of the homogeneous Klein-Gordon equation. ∆R = 1 (2π)4 d4 k e−ikx k2 1 − M 2 + iǫk 0 (with the implicit prescription to take ǫ → 0 after integration).

or ǫ(k 0 ) = θ(k 0 ) − θ(−k 0 ). (h) Show that the causal Green’s function ∆C (x) = θ(x0 ) ∆+ (x) − θ(−x0 ) ∆− (x). .Quantization of fields 65 (f) By performing the k0 integration over any closed path C (encircling one or both of the ples) show that 1 1 d4 k e−ikx 2 ∆(x) = − 4 (2π) C k − M2 are homogeneous solutions for the contours: k0 −E ∆_ ∆+ E ∆ One way of showing this is to write these contours as integrals of the form under (a). (i) (Optional) Give similar expressions for the other Green’s functions under (d) in terms of the homogeneous solutions under (f). (g) Show that the homogeneous solution ∆ in (f) satisfies ∆(0. In that case one actually needs discontinuous functions f (k) such as the step function. 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). (iγ µ ∂µ − M ) ψ(x) = 0. We can determine A. −r). T and C for classical quantities is shown in the table 1.1) The parity operator transforms We will consider the transformation properties for a fermion field ψ(x). time reversal (T) and charge conjugation (C). T. 8. 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 .1 Parity xµ = (t.Chapter 8 Discrete symmetries In this chapter we discuss the discrete symmetries. x (8.1: The behavior of classical quantities under P. starting with the Dirac equation for ψ(x).2) where ηP is the intrinsic parity of the field and A is a 4 × 4 matrix acting in the spinor space. ˜ (8. Table 8. r) −→ xµ ≡ xµ = (t. Both ψ p and ψ satisfy the Dirac equation. parity (P). The consequences of P.

−λ) eik·x 3 2E (2π) ˜ k d3 k ˜ ˜ ηP b(k. λ) e−ik·x − d† (k. −λ) e−ik·x − d† (k. λ)γ0 u(k.6) (check this for the standard representation. −λ). the above operation reverses the sign of λ).5) (8.11) (8. ˜ −v(k. m).7) = λ d k x x ηP b(k.e. (i˜ µ ∂µ − M ) ψ(˜) = 0. m) γ0 v(k. op P (8.4) (8. m). x It is straightforward to apply this to the explicit field operator ψ(x) using γ0 u(k. . λ) P −1 = −η ∗ d(k. γ x (8.13) ψ(x) −→ ψ c (x) = ηC C ψ (x). x µ (iγ ∂µ − M ) γ0 ψ(˜) = 0. iγ µ† ∂µ − M ψ(˜) = 0. λ)v(k. 8. The result is ψ p (x) −1 = Pop ψ(x) Pop = ηP γ0 ψ(˜) x (8. In the same way as the Fermion field.10) i. −λ). λ) eik·˜ (2π)3 2Ek d3 k ˜ ˜ ˜ ˜ ηP b(k. λ) Pop = ηP b(k.8) 3 = λ = λ = λ From this one sees immediately that −1 ˜ Pop b(k. choosing ηP is real (ηP = ±1) particle and antiparticle have opposite parity. −λ)v(k.Discrete symmetries 67 After parity transforming x to x the Dirac equation becomes after some manipulations ˜ ˜ x iγ µ ∂µ − M ψ(˜) = 0. λ)γ0 v(k. λ)v(k. λ)u(k. λ) e−ik·˜ + d† (k. ˜ Pop d(k. −λ)u(k. (2π)3 2Ek (8. λ)u(k. one can also consider the scalar field and vector fields. For the scalar field we have seen −1 φ(x) −→ Pop φ(x) Pop = ηP φ(˜).2 Charge conjugation We have already seen the particle-antiparticle symmetry with under what we will call charge conjugation the behavior T (8.9) (8. λ) eik·x . x (8.12) and for the vector field −1 Aµ (x) −→ Pop Aµ (x) Pop = −Aµ (˜). if helicity λ is used instead of the z-component of the spin m. x The latter behavior of the vector field will be discussed further below. −λ) e−ik·x − d† (k. x Therefore γ0 ψ(˜) is again a solution of the Dirac equation and we have x ψ p (x) = γ0 ψ(˜). −λ) eik·x (2π)3 2Ek ˜ d3 k ˜ ˜ ˜ ˜ ηP b(k. m) = = ˜ u(k.3) (8.

22) = Aψ|Aφ ∗ is antilinear if A(λ|φ + µ|ψ ) = λ∗ A|φ + µ∗ A|ψ . One has Aφ|Aψ = φ|ψ ∗ = ψ|A† Aφ = ψ|φ . x x (8. λ). m). ((iγ µ )∗ ∂µ − M ) ψ(x) = 0.21) where A is a 4 × 4 matrix acting in the spinor space.3 Time reversal xµ = (t. For a quantized field one has −1 ∗ −1 Top fk (x)bk Top = fk (x)Top bk Top . (8. it is antilinear1 .15) (where one must be aware of the choice of spinors made in the expansion. x µ (iγ ∂µ − M ) γ5 Cψ(−˜) = 0. m). λ)C uT (k. to find ψ t (−˜) that is a solution of the Dirac equation. ˜ x −(iγ µ )∗ ∂µ − M ψ(−˜) = 0. x (8. λ) e−ik·x + b† (k. λ) e−ik·x + b† (k. as discussed in section 4).19) Cop d(k.20) The time reversal operator transforms We will again consider the transformation properties for a fermion field ψ(x). m) ¯ C v T (k. Norm conservation requires Top to be anti-unitary2 . λ) Cop = ηC d(k. λ)C v T (k. The same relations hold for helicity states. antilinear operator is anti-unitary if A† = A−1 .Discrete symmetries 68 the latter being also a solution of the Dirac equation.e. . Therefore ψ c (x) = = λ −1 ¯ Cop ψ(x) Cop = ηC C ψ T (x) (8. 8. we start with the complex conjugated x Dirac equation for ψ. γ x i(γ5 C)−1 γ µ γ5 C∂µ − M ψ(−˜) = 0. (i˜ µ∗ ∂µ − M ) ψ(−˜) = 0. λ). (8. r) −→ −˜µ ≡ −xµ = (−t. Top |φ = φt | = (|φt )∗ .17) 3 = λ This shows that −1 Cop b(k. r). λ) −1 Cop = ∗ ηC b(k. m) ¯ = = v(k. λ) eik·x ¯ ¯ (2π)3 2Ek d3 k ηC d(k.16) d k ηC d(k. λ)v(k. writing −1 ψ(x) −→ Top ψ(x)Top = ηT Aψ(−˜) ≡ ψ t (−˜). λ) eik·x . u(k. (2π)3 2Ek (8. The (time-reversed) Dirac equation becomes. x (8. x 1A 2 An −iC −1 γ µ C∂µ − M ψ(−˜) = 0.14) (8. As time reversal will transform ’bra’ into ’ket’. x iγ µT ∂µ − M ψ(−˜) = 0. λ)u(k.18) (8. i. The action on the spinors (using C = iγ 2 γ 0 = −iρ1 σ 2 in standard representation) gives C uT (k.

λ a P |a. λ) e−ik·x 3 2E (2π) k d3 k ˜ ˜ ηT b(k. λ). λ)v ∗ (k. λ)u∗ (k. λ)v ∗ (k. p. λ) ψ t (x) −1 Top ψ(x) Top = i ηT γ5 Cψ(−˜) x = = ˜ u∗ (k. p.27) 3 = λ = λ = λ From this one obtains −1 ˜ Top b(k.23) In the standard representation iγ5 C = σ2 and it is straightforward to apply this to the explicit field operator ψ(x) using i γ5 C u(k.30) (8. λ) e−ik·x 3 2E (2π) k ˜ d3 k ˜ ˜ ˜ ˜ ηT b(k.4 Bi-linear combinations In quantities such as currents and lagrangians often bilinear combinations of spinor fields are encountered. Since there are 16 independent 4 × 4 matrices.26) d k x x ηT b(k.24) (8. Note that applying an anti-unitary transformation such as Top one must take for the matrix element ˜ the complex conjugate. −λ |¯. λ) Top = ηT b(k. The result is = = λ (8. λ) i γ5 C v(k. λ)iγ5 Cv(k. (8.25) (check this for the standard representation). x (8. λ) eik·x + d† (k. −λ a T a. ˜ = 0|b(k)|A 8.28) (8. λ).34) . ˜ Top d(k. ∗ ˜ v (k. λ)v ∗ (k. p. λ) eik·x + d† (k. λ) e−ik·x (2π)3 2Ek (8.31) (8. λ| ¯ C |¯. λ). p.29) In table 2 the behavior of particle states under the various transformations has been summarized.2: The transformation properties of physical states for particles (a) and antiparticles (¯). λ)u∗ (k. a state |a.Discrete symmetries 69 Table 8. −p. They are the following S(x) V µ (x) T µν (x) Aµ (x) P (x) ¯ = ψ(x)ψ(x) (scalar) ¯ = ψ(x)γ µ ψ(x) (vector) µν ¯ = ψ(x)σ ψ(x) (tensor) µ ¯ = ψ(x)γ5 γ ψ(x) (axial vector) ¯ = iψ(x)γ5 ψ(x) (pseudoscalar) (8. −p. there are 16 independent of these bilinear combinations. λ |¯. op T (8. λ) T −1 = η ∗ d(k. λ)iγ5 Cu(k. −p. λ)u∗ (k. λ Therefore γ5 Cψ(−˜) is again a solution of the (ordinary) Dirac equation and we can choose (phase is x convention) ψ t (x) = i γ5 C ψ(−˜). −p. λ) eik·x + d† (k. λ) e−ik·˜ (2π)3 2Ek d3 k ˜ ˜ ˜ ˜ ηT b(k. λ). λ a |a.32) (8.33) (8. λ) eik·˜ + d† (k. Therefore one has A|k = A|b† (k)|0 = A|T † T b† (k) T † T |0 ∗ = At |b† (k)|0 ∗ ˜ t = k|At . λ| a.

p′ |V µ (0)|p = u(p′ )γ µ u(p). p′ |V µ (x)|p = u(p′ )Γµ (p′ ... (q 2 ). C. Note that the lagrangian density L (x) → L (−x) under Θ. (8. σ µν . for the vector current V µ (x). 2M (8.38) (8.39) where Γµ (p′ .41) 2M 2M . ¯ ′ ′ (8. s′ |V µ (x)|p. (q 2 ). in this case only q 2 = (p′ − p)2 .Discrete symmetries 70 Table 8. As an example consider the vector current for a point fermion. This implies p′ |V µ (x)|p = e−i (p−p )·x p′ |V µ (0)|p . In many applications the expectation values of currents are needed. γ ν ]. and T. e. momenta (pµ or p′µ ) or constant tensors (ǫµνρσ or gµν ). p′ . γ µ .40) where the coefficients are functions of invariants constructed out of the momenta. 8. F.5 Form factors Currents play an important role in field theory. of which the expectation value between momentum eigenstates can be simply found. Applying the results from the previous sections it is straightforward to determine the behavior of the combinations under 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 ). (8. For instance for nucleons one has Γµ (p′ . 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 . p) = γµ F1 (q 2 ) + iσµν q ν F2 (q 2 ). as well as under the combined operation Θ = PCT (see Table 3).3: The behavior of the independent bilinear combinations of fermi fields under P. C. V µ (x) = ei Pop ·x V µ (0)e−i Pop ·x . ¯ In general the expectation value between momentum states can be more complicated. called form factors. ¯ V µ (x) =: ψ(x)γ µ ψ(x) :.g.37) (8. p) can be built from any combination of Dirac matrices (1. s . γ5 γ µ or γ5 ). p)u(p) e−i (p−p )·x . F. The 16 combinations of Dirac matrices appearing above are linearly independent.35) The x-dependence can be accounted for straightforwardly using translation invariance.36) (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).

p′ )γ0 = Γµ (p′ . p)(i γ5 C). Therefore hermiticity implies for the structure of Γµ (p′ .g. ¯ 2M (8. p) that γ0 Γµ† (p. p)γ0 . ¯ p ˜ Therefore parity invariance implies for the structure of Γµ (p′ . e. p)u(p) ¯ † † = p′ |Top Top V µ (x)Top Top |p ∗ (8. ¯ µ p ˜ Therefore time reversal invariance implies for the structure of Γµ (p′ . γ5 σµν = (1/2)ǫµνρσ γ ρ γ σ . p). or relations based on hermiticity of the operator or P .43) = ei q·x u(p′ )γ0 Γµ (˜′ . C and T invariance. p)u(˜) ˜ x p ¯ p p ˜ p (8. or relations that follow from the equations of motion. • Hermiticity p′ |V µ (0)|p = p′ |V µ† (0)|p = u(p′ )Γµ (p′ . p ˜ • Time reversal p′ |V µ (x)|p = ei q·x u(p′ )Γµ (p′ . the Gordon decomposition u(p′)γ µ u(p) = ¯ 1 u(p′ ) [(p′ + p)µ + iσ µν qν ] u(p).Discrete symmetries 71 but that are eliminated because of relations between Dirac matrices. e. p) that Γµ (p′ . µ p ˜ (8. p)u(p) ¯ = p|V µ (0)|p′ ∗ = (¯(p)Γµ (p. p′ )γ0 u(p).45) Exercises Exercise 8. p)γ0 u(p). p) that Γµ (p′ .1 The matrix element of the electromagnetic current between nucleon states is written as < p′ |Jµ (x)|p >= e−i(p−p )·x u(p′ )Γµ u(p) ¯ Here Γµ iσµν q ν F2 (q 2 ) 2M pµ + pµ ′ = γµ H1 (q 2 ) − H2 (q 2 ) 2M = γµ F1 (q 2 ) + ′ . p)u∗ (˜) up p µ p ˜ = ei q·x u(p′ ) (iγ5 C)Γ∗ (˜′ . p) = (i γ5 C)Γ∗ (˜′ .44) = p′ |Vµ (−˜)|˜ ˜ x p ∗ ˜x = ei q ·˜ [¯(˜′ )]∗ Γ∗ (˜′ . p) = γ0 Γµ (˜′ . p′ )u(p′ )) u = ∗ † u† (p)γ0 Γµ (p.g. p)u(p) ¯ † † = p′ |Pop Pop V µ (x)Pop Pop |p ˜x = p′ |Vµ (˜)|˜ = ei q ·˜ u(˜′ )Γµ (˜′ .42) where q = p′ − p. • Parity p′ |V µ (x)|p = ei q·x u(p′ )Γµ (p′ . p)(iγ5 C)u(p). p′ )u(p′ ) = u† (p′ )Γµ† (p.

0. it would correspond to an electric dipole moment d = −e F3 (0)/2M (Interaction term d σ · E). Exercise 8. 0. −|q|/2). (c) Show that hermiticity of the current requires that F1 en F2 are real. (b) Show that such a term is also not allowed by time-reversal symmetry. Exercise 8. (c) Show that if a term of this form would exist. 0. p = (E. 0.2 (a) Calculate the current matrix element (using the explicit fermion spinors) in the Breit-frame in which q = (0.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 . |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 . p′ = (E. 0. |q|). GE GM = F1 − Q2 F2 .3 (a) Show that because of parity conservation no term of the form γ5 can appear in the current matrix element. 4M 2 = F1 + F2 . 0. σµν q ν F3 (q 2 ) 2M . (b) Show that in the interaction with the electromagnetic field. L int = ejµ Aµ the charge is given by e GE (0) and the magnetic moment by e GM (0)/2M .

(9. AH (t) = U −1 (t. (9. t0 ) ∂t (9. in which the states are time-independent. and the operators are time-dependent. (9. t0 )|ψS (t0 ) .Chapter 9 Path integrals and quantum mechanics 9. t0 ).3) (9. QH (t)|q.4) ∂ U (t. H]. |ψS (t) = U (t. t′ ≡ q ′ |q ′ .1 Time evolution as path integral U (t. AS (t) = AS and the states o are time dependent. t ′ QH (t′ )|q ′ . i ∂ |ψH ∂t ∂ i AH ∂t ≡ = 0. in which the operators are time-independent. ≡ 0. t ′ |q .7) 73 .2) (ii) Heisenberg picture. t0 ) = e−i(t−t0 )H or i Two situations can be distinguished: (i) Schr¨dinger picture.1) The time evolution from t0 → t of a quantum mechanical system is generated by the Hamiltonian. |ψH (t) = |ψH . t . t0 ) = H U (t. [AH . t′ . i ∂ |ψS ∂t ∂ i AS ∂t = H |ψS .6) Of these the Heisenberg picture is most appropriate for quantum field theory since the field operators do depend on the position and one would like to have position and time on the same footing. t ≡ q|q.5) (9. Consider the two (time-independent) Heisenberg states: |q. t0 ) AH (t0 ) U (t. (9.

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

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

22) = x The last equality is obtained by defining the right measure (normalization N ) in the integration. N (9.26) As an important application of translation invariance we mention the identities 1 Dα exp − α2 − ωα 2 = exp 1 2 ω . 1 = √ det K 1 Dα Dα∗ exp (−α∗ K α) = det K 1 Dα exp − α K α 2 (α real-valued) (α complex-valued) (9. α(x) = n αn fn and K(x. In both cases N is an appropriately choosen normalization constant in such a way that the integral is finite. (9.21) again a multidimensional integral. Consider the gaussian functional as an example. 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.n fm (x) Kmn fn (y) and consider F [α] = F (αn ) as a multivariable function.27) (9. y) = ∗ m. 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. with Dα F [α] = n dαn N F (αn ) (9.28) Dα Dα∗ exp (−α∗ α − α∗ ω − ω ∗ α) = exp (ω ∗ ω) .n dx fm (x)fn (x) = n = n dαn 1 exp − α2 N 2 n √ 2π ≡ 1.23) Having defined the Gaussian integral. Physical answers will usually come out as the ratio of two functional integrals and are thus independent of the chosen measure. the following integrals can be derived for a real symmetric or complex hermitean kernel K. . (9.24) (9. Note that procedure (i) is an example of the more general procedure under (ii).Path integrals and quantum mechanics 76 (ii) Write α in terms of a sum of orthonormal basis functions.25) (see Exercises). 2 (9. A useful property of functional integration is the translation invariance.

31) δ δ α = dy α(y) = dy δ(x − y) = 1. −→ fn (x) δα(x) ∂αn ∂αm n Note that δ(x − y) = n ∗ fn (x) fn (y).g. δα(x) δα(x) δ αKβ = dy K(x. i. we need to consider Grassmann-valued functions involving anticommuting Grassmann variables. αn = δmn . y) θ(y) .35) For applications to fermion fields. which vanish. δα(x) (9.33) (9. Examples of functional differentiation are (9. In the expansion of the exponential (from second to third line). n θn fn (x) dx θ∗ (x) θ(x) .34) (9.Path integrals and quantum mechanics 77 Functional differentiation is defined as δ .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. θ∗ ] = exp (−θ∗ θ) ≡ exp − F [θ. e.30) (9.36) 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 ) = 1.32) (9. −→ δα(x) ∆x ∂αx ∂αx δ ∂ ∂ ∗ with . y) β(y) δα(x) δ 1 1 exp − α2 = −α(x) exp − α2 . (9. products of the same pair appear. δα(x) 2 2 δ exp (−α β) = −β(x) exp (−α β) . F [θ.e. α(y) = δ(x − y). the Gaussian-type functionals. θ∗ ] = exp (−θ∗ K θ) ≡ exp − Integration for Grassmann variables is defined as dθ 1 = 0 and dθ θ = 1. (9. αy = δxy .38) . dx dy θ∗ (x) K(x. It is now easy to check that Dθ∗ Dθ exp (−θ∗ K θ) = det K. δα(x) or in discretized form δ 1 ∂ ∂ with . (9. In principle the definitions of functionals is the same.29) (9. θη = −ηθ. θ2 = 0. however.

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

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

T ′ |Q. t). thus T →i∞ lim q. such that t′ → −i∞ t → i∞ ↔ ↔ ˜ t → −∞. t|Q. (9.68) (9. Since we have (neglecting multiplicative factors). J (9. Better is the use ˜ of imaginary time t = −it.2: Analytic continuation of boundaries T and T ′ We can project out the groundstate by an analytic continuation in the time. T′ Z[J] = lim T ′ → −i∞ T → i∞ Dq exp i T dτ L q.70) one immediately finds δ n Z[J] δJ(t1 ) . T = φ0 (q.67) = = φ0 (q. .Path integrals and quantum mechanics t 82 T t t’ T’ Figure 9. . in which case eiEn T → e−En ·∞ and the term eiE0 T is dominant. t′ ) q ′ . t 0 0out |0in J . T → i ∞ and T ′ → −i ∞. ˜ t′ → ∞. δJ(tn ) hence the name generating functional. = (i)n 0|T Q(t1 ) .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) (9. . T J lim −i E0 (T ′ −T ) φ (Q′ ) φ∗ (Q) 0 0 T ′ → −i∞ e T → i∞ dq ′ dq φ∗ (q ′ . 9.6 Euclidean formulation The above expression (Eq. The generating functional precisely represents the vacuum to vacuum amplitude from initial (’in’) to final(’out’) situation in the presence of a source. eiE0 T φ∗ (Q) 0 We define the generating functional Z[J] as Z[J] = Q′ . The importance of Z[J] is that the time ordered product of operators can be obtained from it.71) 9. t′ |q. .70) for the generating functional is in fact ill-defined. J=0 (9. dq dτ + J(τ )q(τ ) . Q(tn )|0 . .

det K 1 . ˜ ˜ ZE [J] δJ(t1 ) . δ n Z[J] 1 Z[J] δJ(t1 ) . δJ(tn ) = (i)n J=0 1 δ n ZE [J] . (9. ensuring convergence for the functional integral ZE [J]. (9.g. . i τ −∞ ∞ −∞ dq d˜ τ + J(˜)q(˜) τ τ dq d˜ τ − J(˜)q(˜) τ τ . + V (q).74) − V (q) 2 (9. ZE [J] = = where LE q. dq ˜ dt = 1 2 dq ˜ dt dq ˜ dt 2 − V (q). det K Exercise 9.1 Convince yourself by choosing an appropriate (orthonormal) set of functions that 1 Dα exp − α K α 2 for real-valued functions. when L q.2 Check the examples of functional derivation for real-.75) = − LE q. where Kp are the eigenvalues of the matrix Kmn in K(x. complex. i dq . As discussed in the previous section the interesting quantities are obtained from functional differentiation with respect to the sources. y) = ∗ m. e. 1 = √ . for Grassmann valued functions. . Dα Dα∗ exp (−α∗ K α) = for complex-valued functions and Dθ∗ Dθ exp (−θ∗ K θ) = det K. .and Grassmann-valued functions. . The differentiations in Z[J] and ZE [J] are related.76) which is a positive definite quantity.72) (9.Path integrals and quantum mechanics 83 In terms of the imaginary time we can write the Euclidean generating functional.73) d˜ LE q. In all of these cases the determinant of the (hermitean) operator K is defined as det K ≡ p Kp . τ dq ˜ dt ≡ −L q. ˜ dt (9. Exercises Exercise 9. .n fm (x) Kmn fn (y). δJ(tn ) J = 0 ˜ t = it (9. dq dt = 1 2 1 2 dq dt 2 Dq exp Dq exp − ∞ d˜ L q.77) where the expressions have been divided by Z[J] and ZE [J] in order to get rid of dependence on multiplicative factors.

y) K −1 (y. det K Dθ∗ Dθ exp (−θ∗ K θ − θ∗ η − η ∗ θ) = det K exp η ∗ K −1 η .78) (9. 2 (9.79) (9.80) Dα∗ Dα exp (−α∗ K α − α∗ ω − ω ∗ α) = 1 exp ω ∗ K −1 ω . . where dy K(x.Path integrals and quantum mechanics 84 Exercise 9.3 Prove the following relations 1 Dα exp − α K α − αω 2 1 = √ exp det K 1 ω K −1 ω . z) = δ(x − z).

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. x′ .1 Generating functionals for free scalar fields The generating functional for quantum fields is a generalization of the results in the previous section to a system with more degrees of freedom.2) = ∞ −∞ d4 xE [LE (φ. xn ). . .4) J=0 . . t′ |φ. t = = Dφ DΠ exp i t σ′ ˙ dτ d3 x Π(x)φ(x) − H (φ. t) is considered as a set of quantum operators φx (t) in the Heisenberg picture and t′ φ′ . ∂µ φ) + Jφ] (10. Π) (10.Chapter 10 Feynman diagrams for scattering amplitudes 10. (10. . i. φ(xn )|0 = (−i)n δ n Z[J] δJ(x1 ) . x.e. . δJ(xn ) 85 ≡ G(n) (x1 . 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 ) . . . . ∂µ φ) − Jφ] (10.3) The Euclidean formulation is as before.

Feynman diagrams for scattering amplitudes 86 for which we introduce the picture x1 G(n) (x1 .12) d4 k −i k·x ˜ e ∆(k) (2π)4 (10.4) one distinguishes the retarded Green’s function ∆R (x) = d4 k e−i k·x . These Green functions appear in the expansion Z[J] = n in n! d4 x1 . . xn ) = x2 xn .8)  1 d4 x  φ (∂µ ∂ µ + M 2 ) φ − Jφ 2 K  (10. . Furthermore. ∆F (the socalled Feynman propagator) is the Green’s function of the Klein-Gordon equation. J(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. .9) i. .11) Depending on the path choosen in the complex k 0 plane (see exercise 7. (2π)4 (k 0 + iǫ)2 − E 2 (10. which can be written as ∆(x) = ˜ with (k 2 − M 2 )∆(k) = 1 or ˜ ∆(k) = 1 1 1 = 0 2 . .6) the generating functional is (using ∂µ = − ∂µ ) given by   Z0 [J] =  Dφ exp −i where Kxy = (∂µ ∂ µ + M 2 ) δ 4 (x − y). = 0 2 2 2 k2 − M 2 (k ) − E 2 (k ) − k − M (10. 2 2 → (10.7) where we made the choice Z0 [0] = 1 and ∂µ ∂ µ + M 2 ∆F (x − y) = −δ 4 (x − y). . d4 xn G(n) (x1 . . consider the general solution of ∂µ ∂ µ + M 2 ∆(x) = −δ 4 (x). (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]) . . with Z0 [0] = 1. (10.10) J=0 In order to determine ∆F . xn ) J(x1 ) .e. (10. . .5) Z0 [J] for the free scalar field For the (free) scalar field lagrangian L = ← 1 1 ∂µ φ∂ µ φ − M 2 φ2 .13) . As discussed in the previous section (exercise 9. .

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. µ 1 1 • Firstly. (2π)4 k 2 − M 2 (10.e. φ+ (0)] = (2π)3 2E are also shown in exercise (7. the second is to explicitly consider 0|T φ(x1 ) φ(x2 )|0 . . 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 .1 for countours) µ (∂E )2 − M 2 (i ∆F (x − y)) = −δ 4 (x − y) 0 kE =∞ (10. t) = 0 for t > 0. (2π)4 k 2 − M 2 (10.21) ∆F (x) = = −i 0 kE =−∞ d4 kE e−i kE ·xE 2 (2π)4 kE + M 2 (10. (10. ∆F = ∆C .Feynman diagrams for scattering amplitudes 87 satisfying ∆R (x. in the (well-defined) Euclidean formulation starting with LE = 2 (∂E φ)2 + 2 M 2 φ2 µ 2 2 Jφ with the equation of motion (∂E ) − M φ(x) = −J(x). φ(0)] = i∆+ (x) + i∆− (x) d3 k e−i k·x = [φ+ (x).23) k0 =i∞ k0 =−i∞ d4 k e−i k·x .16) C where C is a closed contour in the complex k 0 -plane.14) They are solutions of the inhomogeneous equation. The contours for i ∆(x) i ∆+ (x) i ∆− (x) = [φ(x). (2π)4 k 2 − M 2 where the latter contour can be deformed to the contour for ∆C .20) = exp − 2 where or (see fig. the advanced Green’s function ∆A (x) = satisfying ∆A (x.18) Thus we see various solutions. ∆(x) = − d4 k e−i k·x . the generating functional can be written 1 µ µ Z0 [J] = Dφ exp d4 xE (10.4). The first possibility is to consider the well-defined Euclidean formulation.15) e−i k·x d4 k . (10. 10. (2π)4 (k 0 − iǫ)2 − E 2 (10. The solution of the homogeneous equation ∂µ ∂ µ + M 2 ∆(x) = 0 can also be written as an integral in k-space. 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). 4 k 2 − M 2 + iǫ (2π) (10.19) φ ∂E ∂E − M 2 φ + Jφ 2 1 d4 xE d4 yE J(x) (−i ∆F (x − y)) J(y) .22) (10. i. t) = 0 for t < 0.17) d4 k e−i k·x . φ− (0)] = (2π)3 2E d3 k e−i k·x = [φ− (x).

25) x1 x2 x3 = x4 + + .3) i ∆F (x − y) = 0|T φ(x) φ(y)|0 = θ(x0 − y 0 ) 0|φ(x) φ(y)|0 + θ(y 0 − x0 ) 0|φ(y) φ(x)|0 = θ(x0 − y 0 ) 0|φ+ (x) φ− (y)|0 + θ(y 0 − x0 ) 0|φ+ (y) φ− (x)|0 = i ∆C (x − y).1: Contour in the k 0 plane for the Feynman propagator in Euclidean and Minkowski space • As the second possibility. or diagrammatically (4) (10. the same result as above. x4 ) = 0|T φ(x1 ) φ(x2 ) φ(x3 ) φ(x4 )|0 1 δ4 exp − i J ∆F J = (−i)4 δJ(x1 ) . . we have G0 (x1 . because this is the only quantity entering Z0 [J].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. 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. x2 . δJ(x4 ) 2 J=0 = − [∆F (x1 − x2 ) ∆F (x3 − x4 ) + ∆F (x1 − x3 ) ∆F (x2 − x4 ) + ∆F (x1 − x4 ) ∆F (x2 − x3 )] . φ− (y)]|0 − θ(y 0 − x0 ) 0|[φ− (x). . x3 . φ+ (y)]|0 = θ(x0 − y 0 ) i ∆+ (x − y) − θ(y 0 − x0 ) i ∆− (x − y) (10. (4) (2) It should be clear that G0 can be expressed in terms of G0 = i ∆F . = θ(x0 − y 0 ) 0|[φ+ (x). Neglecting multiplicative factors or equivalently assuming that Z0 [0] = 1. We find (see also section 7.24) Knowing the explicit form of Z0 [J] it is straightforward to calculate the 4-points Green’s function.

with propagators (i ∆F ) being connected to vertices (i LI ) according to the above expression for the functional Z[J].29) This expression will be the one from which Feynman rules will be derived.26) Dφ exp i = exp i d4 z LI 1 δ i δJ(z) d4 x d4 y J(x) ∆F (x − y) J(y) (10. the generating functional can be written.30) and in zeroth order G(0.g 0 ) = G0 = 1. L (φ) = L0 (φ) + LI (φ).g ) (x1 .34) Introducing a vertex point to which four propagators are connected.e. x2 . i. x4 ) = G0 (x1 .Feynman diagrams for scattering amplitudes 89 10.2 Generating functionals for interacting scalar fields d4 x [L0 (φ) + LI (φ) + Jφ] 1 δ i δJ(z) d4 x [L0 (φ) + Jφ] (10.32) (10.g ) (x1 . (2) (0) (10. To zeroth order in the coupling one has 0 1 Z (g ) [J] = Z0 [J] = exp − i 2 J ∆F J .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 . (10. Consider as an example the interaction LI (φ) = − g 4 φ 4! in the scalar field theory discussed sofar. x3 . ≡ −i g d4 z (10. x4 ) = To first order in g one has Z (g ) [J] = = 1 0 0 x1 + + x2 (10.31) G(2. Z[J] = = Dφ exp i exp i d4 z LI (10. x2 ) = i ∆F (x1 − x2 ) = G(4. (10. x2 ) = G0 (x1 . x2 .27) When interactions are present.35) z .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 . x3 . (10.

.g 1 ) = 1 8 z 1 8 . d4 xn G(n) (x1 . . J(xn ) c (10. . . . δJ(xn ) . Z[J] = n in n! d4 x1 . . . In the free case we have seen that the exponent of Z[J] contained all essential information. . J(xn ). xn ) J(x1 ) . x2 ) = c (vacuum bubble) (tadpole) (connected propagator) . which also was the only Green’s function for which the diagram was connected. . .Feynman diagrams for scattering amplitudes 90 one has J Z (g1 ) [J] = 1 8 1 − 4 J J 1 + 24 J Z0 [J].e.38) The result for the 4-points Green’s function is left as an exercise. G(n) (x1 . xn ) = x2 c xn Specific examples are G(0) = c G(1) (x1 ) = c G(2) (x1 .37) G(2.g ) (x1 .41) in terms of socalled connected Green’s functions.40) in−1 n! d4 x1 . Diagrammatically this exponent only contained the two-point function. xn ) J(x1 ) . . This remains also true for the interacting case. . . 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]. . . xn ) = (−i)n−1 c which are denoted δ n W [J] δJ(x1 ) . (10. (10. .42) x1 G(n) (x1 . . . d4 xn G(n) (x1 . . i. x2 ) = 1 z x1 x2 + 1 2 x1 z x2 . (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]. . .36) J J from which one obtains G(0. . did not contain parts that could be written as products of simpler Green’s functions. (10. . (10. . .39) We have calculated the function up to first order in the coupling constant for φ4 -theroy. J=0 (10. .

Feynman diagrams for scattering amplitudes 91 To see that Z[J] = exp(i W [J]) contains all non-connected diagrams can be performed inductively. 2! 3! +i d4 x1 + i2 2! d4 x1 d4 x2 . (1) (1) (iii) Gsc = Gc . 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). δJ(x) J=0 ... + +i d4 x1        1 2! + 1 3! +i d4 x1 + 1 2! i2 2! d4 x1 d4 x2              + . (iv) If δZ[J] = i 0|φ(x)|0 = 0... These can be divided out.. (ii) Note that Z[0] = exp(iW [0]) appears as a multiplicative factor. in the expansion of which all vacuum blobs are contained... If i W [J] = then exp(i W [J]) = 1 + i W [J] + = 1 + 1 2! +i d4 x1        i2 + d4 x1 d4 x2 2!         i2 4 4 d x1 d x2 +   2!     +                                        + (i W [J])2 (i W [J])3 + + . In the expansion of Z[J]/Z[0] one (by definition) has the socalled source (n) connected Green’s functions Gsc . but is easily illustrated by writing down the first few terms. 1 2! Z[0] = 1 + + + .

out .g. considered explicitly. in = |α. in|S = β. in = β. in|SS † |α. we have W0 [J] = − 1 2 d4 x d4 y J(x) ∆F (x − y) J(y). thus β. Proof: α. this will be translated to the action on fields. For the free scalar theory. out = |p′ . (2) (10. in = S|α. in ⇔ β. in into final state particle states (out-states) |β. out = p|p′ . the vacuum expectation value of the field φ(x) is zero. out| ⇔ |α. Next. .e. (10. (10. out|α. translation invariance). p. Sβα ≡ β. . out|p′ .47) . x3 . in = p. in = |p1 . (3) S is unitary (it conserves the scalar product from initial to final state). out (suppressing except momenta all other 1 m quantum numbers). in|S = 0. out| and S † |α. Proof: 0. implying the absence of tadpoles. . in = p. φout and the interpolating field φ(x). .46) The properties of the S-matrix are (1) The vacuum is invariant or |S00 | = 1.g ) (x1 . in|S|p′ . e. (2) The one-particle state is invariant (conservation of energy and momentum. ∂µ ∂ µ + M 2 φin (x) = 0. Also for free field (Heisenberg) operators a distinction is made between φin and φout . . in|S = α. one finds for the connected 4-point Green’s function at order g. t = −∞ φin (x) t = +∞ φout (x) φ(x) where φin and φout transform under the Poincar´ group as scalar fields and satisfy the homogeneous e Klein-Gordon equation with the physical mass M . pn . x4 10. out .Feynman diagrams for scattering amplitudes 92 i. . In line with the consideration of the generating functional representing the vacuum to vacuum amplitude we consider fields φin . x2 . out|α.3 Interactions and the S-matrix The S-matrix The S-matrix transforms initial state free particle states (in-states) |α. x1 G(4. 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 ). out| = eiϕ0 0. then (n) (n) Gsc = Gc for n ≤ 3.44) J=0 For the interacting theory. (10. in| (choose ϕ0 = 0).45) x2 as the only surviving diagram (see exercises). in = p. x4 ) = c 1 x3 z (10. in|p′ . out = δαβ ↔ SS † = 1. out|p′ .

51) (10. (10. z) φ(z) d4 y d4 z ∆A (x − y) K(y. out|φout = β. β.and out-fields is: φin (x) = S φout (x)S −1 .49) The field φ can be expressed in φin/out using retarded or advanced Green’s functions. The relation between S and Z[J] To establish this relation. φin (y)] = iZ ∆(x − y).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. which in a later stage (renormalization) will become more important. 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. 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. (2π)3 2E (10. (10. this can actually not be used as it would imply [φ(x). in|φin S → φin S = S φout (x).54) . 2 ∂µ ∂ µ + M0 φ(x) = J(x). (10.53) 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). 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. in|Sφout Finally we check that as expected S does not spoil Poincar´ invariance. a) = S.g.62) is then T′ UT [J] T′ = T exp i T d4 x J(x)φ(x) . i. φin (x) = d3 k ∗ a(k) fk (x) + a† (k) fk (x) . φ(y)] = Z [φin (x). implies the strong (operator) convergence φ(x) → Zφin (x). a causality condition that can be proven to imply the absence of interactions. The convergence therefore must be weakened to t→−∞ √ (10. as it stands. S is invariant under e Poincar´ transformations: U (Λ. a)SU −1 (Λ.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 ). the source term Jφ in the lagrangian density is treated in the interaction picture. √ Although the above. Proof: β. The time evolution operator (see Eq. z) φ(z). 9. The relation between S-matrix and in.50) = φ(x) = √ Zφin (x) − √ Zφout (x) − d4 y d4 z ∆R (x − y) K(y. e.e.

one has [A. z) . eB ] = [A. B] and [A.59) (10. C] are c-numbers. 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.57) i [φin (x). with the use of the Baker-Campbell-Hausdorff formula e−B A e+B = A + [A.61) e φ+ f where the normal ordered expression : e φ f : is used. Thus S U [J] =: exp 1 √ Z d4 x d4 y φin (x) K(x.60) e d4 y φ+ (y) f (y) d4 z φ− (z) f (z) d4 z φ+ (z) f (z) d4 y [φ(x). Thus applied to the field φ(x) = φ+ (x) + φ− (x). δJ(z) (10.56) Since δU [J]/δJ(x) satisfies the same equations as φ(x) and we know the limits. : =i d4 y ∆(x − y) f (y) : e d4 z φ(z) f (z) :. φ(y)] f (y) e e .63) . e = i. φ− f (10. y) δ δJ(y) : F [J].Feynman diagrams for scattering amplitudes 94 and satisfies the property T ′ δUT [J] t = i UtT [J] φ(x)UT [J]. express it in terms of advanced and retarded Green’s functions. [A. (10.58) In order to find a solution to this equation note that. δJ(x) ∞ or for U [J] ≡ U−∞ [J] ′ (10. z) δU [J] . we can. 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). z) d4 y d4 z ∆(x − y)K(y. which is equal to the expression e in which the creation operators are placed left of annihilation operators. δJ(z) (10.62) where F [J] is some (arbitrary) functional. just as we did for φ(x). B] (for the case that [A. e e ] = [A.e. S U [J]] = √ Z δ S U [J].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). B] is a c-number). t i Ut∞ [J] φ(x)U−∞ [J] (10. φ(x). B]eB . provided [A. : e d4 z φ(z) f (z) d4 x φ− (x) f (x) B C B C (10. z) δU [J] δJ(z) δU [J] d4 y d4 z ∆A (x − y) K(y. Noting that 0| : eA : |0 = 1 it follows that F [J] = 0|S U [J]|0 = 0|U [J]|0 = 0out |0in ∝ Z[J]. B + C]e e . (10.

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

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

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

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

+ 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.Feynman diagrams for scattering amplitudes 99 The result is which contains an extra minus sign as compared to a bosonic loop. 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. 4 2 In addition to the fermion propagator and fermion wave functions discussed in the previous section we have the photon propagator1 µ k ν = −i k2 gµν 1 − 1− + iǫ λ (k 2 kµ kν . Particular choices of λ are λ = 1 (Feynman propagator or Feynman gauge) and λ = ∞ (Landau gauge). Vector fields and Quantum Electrodynamics As the most important example of vector fields consider the lagrangian density for quantum electrodynamics (QED). λ (∂µ Aµ )2 2 1 For a massive vector boson inversion of the quadratic term including the Lorentz constraint ∝ the propagator kµ kν 1 −gµν + 1− i Dµν (k) = i . 1 k 2 − M 2 + iǫ λ (k 2 − M 2 + iǫ)(k 2 − λ M 2 + iǫ) leads to . 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. 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 . 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 . λ 1 ∂ (10.

75) Note that the q µ q ν term in the photon propagator are irrelevant because the photon couples to a conserved current.79) . where u ¯ L(m) µν = = 1 2 u(k ′ . The diagram and momenta and the commonly used invariants (Mandelstam variables) for a 2 → 2 scattering process are 1 k 3 k’ s t u = (k + p)2 = m2 + M 2 + 2 k · p = (k ′ + p′ )2 = m2 + M 2 + 2 k ′ · p′ = (k − k ′ )2 = q 2 = 2 m2 − 2 k · k ′ = (p − p′ )2 = 2 M 2 − 2 p · p′ p 2 p’ 4 s = (k ′ − p)2 = m2 + M 2 − 2 k ′ · p + t+u= m2 = 2 m2 + 2 M 2 i = (k − p′ )2 = m2 + M 2 − 2 k · p′ The scattering amplitude is given by −iM = u(k ′ . s1 ) u e2 u(p′ . s1 ) ¯ −igµν 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. s2 )|2 ¯ q2 (10.s2 . s)γν u(k ′ . s′ ) ¯ u s. 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 . s4 )(−ie)γ ν u(p. To lowest order (∝ α = e2 /4π) only one diagram contributes. t2 (10. s′ )γµ u(k.5 Some examples eµ scattering The first example is the electromagnetic scattering of an electron and a muon.77) Combining Lµν and Lµν (M) one obtains L(m) Lµν (M) = 2 s2 + u2 + 4t(M 2 + m2 ) − 2(M 2 + m2 )2 µν and M 4π 2 (m) (10. s3 )γ µ u(k. s2 ).78) = 2 α2 2 s + u2 + 4t(M 2 + m2 ) − 2(M 2 + m2 )2 . ¯ q2 (10. q 4 µν [using that (¯(k ′ )γµ u(k))∗ = u(k)γµ u(k ′ )].s4 ) and averaged over spins in the initial state (1/2 × 1/2 × s1 . s)¯(k.76) 1 .s2 ) which can be written as (see also chapter 4) |M |2 = = 1 4s |¯(k ′ .s3 . s4 )γµ u(p.Feynman diagrams for scattering amplitudes 100 10. s3 )(−ie)γ µ u(k.s4 e4 (m) µν (M) L L .

2. s2 )γµ u(k.s4 = 1 1 e Tr(/ − m)γµ (/ ′ + m)γν k k Tr(/ + M )γ µ (/′ − M )γ ν p p s2 2 2 1 e4 1 ′ ′ = 4 2 kµ kν + kν kµ − gµν s pµ p′ν + pν p′µ − g µν s s 2 2 M 4π 2 4 and = 2 α2 2 t + u2 + 4s(M 2 + m2 ) − 2(M 2 + m2 )2 . This is known as crossing symmetry. (b) Use the definition of the source-connected Green function Gsc and show that indeed sourceconnected diagrams survive. s4 )γ ν u(p. The diagram.s2 .Feynman diagrams for scattering amplitudes 101 The second example is the annihilation of an electron pair and creation of a muon pair. s2 ) ¯ ¯ 1 2s u(p. s2 )γµ u(k. .s4 = 1 2s × v (k ′ . .80) Note the similarity in the amplitudes for eµ scattering and e− e+ → µ− µ+ . Basically the same diagram is calculated and the result is the same after the interchange of s ↔ t. the masses. To lowest order (∝ α = e2 /4π) only one diagram contributes. 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. s1 )γν v(k ′ . s3 )γ µ v(p′ . s4 ) u e2 v (k ′ . (4) (4) . s1 )|2 ¯ s 1 . s3 ) ¯ ¯ 1 . s2 (10. Similarly. . s3 )γ µ v(p′ . s4 ) v (p′ . Exercises Exercise 10.1 (a) Give in diagrammatic notation the full Green functions G(4) (x1 . s1 ) u(k.s2 3 .s3 . 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. for instance Møller scattering (e− e− → e− e− ) and Bhabha scattering (e− e+ → e− e+ ) are related using crossing symmetry. (c) Do the same for the connected Green function Gc .

4 (a) Write down the Feynman diagrams contributing to electron-electron scattering. . Note that the function G(n) (p1 .38. xn ). + pn ) G(n) (p1 . pn ). |M |2 = T11 + T22 − T12 − T21 . xn ) ≡ (2π)4 δ 4 (p1 + .Feynman diagrams for scattering amplitudes 102 Exercise 10. that the two-point Green’s function in momentum space has an incoming and an outgoing leg (as for the x-space Green’s function). . . Exercise 10. . Express the contributions in invariants s. . e(p1 ) + e(p2 ) → e(p′ ) + e(p′ ) in lowest order in α. . . 1 2 (b) Calculate the quadratic pieces and interference terms. xn + a) = G(n) (x1 . (b) Show (by using x1 as shift-varible) that this implies that n i=1 d4 xi ei pi ·xi G(n) (x1 . . xn ) ∝ (2π)4 δ 4 (p1 + . . .3 reproduce for the diagrams precisely the combinatorial factors that appear in Eqs 10. . Exercise 10. . their sum is zero.37 and 10. Pictorially this implies. for instance. but the incoming and outgoing momenta are identical. . . . . It is of the form −i M = A1 − A2 . . . . which means overall momentum conservation in Green functions in momentum space. .3 Show that the combinatorial factors found using the rules given in section 10. + pn ). in the amplitude (Tij = A∗ Aj ). hence we can write n i=1 d4 xi ei pi ·xi G(n) (x1 . or if we in general would consider all momenta as incoming. . . . . . t and u. .2 (a) Show that the translation properties of the fields and the vacuum imply G(n) (x1 + a. . . pn ) only has n − 1 independent variables in it. Show that i the amplitude is symmetric under the interchange of t ↔ u. .

This is generalized to the multi-particle phase space dR(p1 . a a b b the quantity s = P 2 = (pa + pb )2 . . −q). . Here s is the invariant mass of the n-particle system. the 1particle phase space is d3 p d4 p θ(p0 ) (2π)δ(p2 − m2 ).6) (11. . . . p) which satisfies p2 = E 2 −p2 = m2 . (11. pn ) = (2π)4 δ 4 (P − pi ) dR(p1 . . . For two particles. pn ). and the total momentum four-vector P = pa + pb .2) (11. (11. . . pa ) and pb = (Eb . To be specific let us consider two frequently used frames. pn ) = and the reduced phase space element by dR(s. p1 .7) pb = cm (Eb . It is a useful quantity. for instance for determining the threshold energy for the production of a final state 1 + 2 + .4) For two particle states |pa . In the CM frame the threshold value for s obviously is n 2 sthreshold = i=1 mi . + pn )2 .5) √ is referred to as the invariant mass squared. . pb we start with the four vectors pa = (Ea . . A physical state has positive energy. + n.3) which is useful because the total 4-momentum of the final state usually is fixed by overall momentum conservation. pb ) satisfying p2 = m2 and p2 = m2 . . . i. . q). The phase space is determined by the weight factors assigned to each state in the summation or integration over states. Its square root.1) = (2π)3 2E (2π)4 (proven in Chapter 2).Chapter 11 Scattering theory 11. . It is determined by the energy-momentum four vector p = (E. .e. In that case cm pa = (Ea . (2π)3 2Ei i=1 (11. The first is the CM system. s is for obvious reasons known as the center of mass (CM) energy. (11. s = (p1 + . (11.1 kinematics in scattering processes Phase space The 1-particle state is denoted |p . 103 . i n d3 pi .

p1 .8) (11. a (11. m2 ) = 4(pa · pb )2 − 4p2 p2 . 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. use the first relation and the abovementioned threshold value for s to calculate the threshold for a specific n-particle final state in the target rest frame. 2 s s − m2 + m2 a b √ . = 2 s λ(s. 0). m2 . p1 . ptrf ). m2 ) a b . 2mb (11. (11. m2 ). mb and s).15) Explicit calculation of the reduced two-body phase space element gives dR(s. 2mb λ(s. p2 ) = = where λ12 denotes λ(s. 4π s 4π 8π s 4π (11. Prove that trf Ea = |ptrf | = a s − m2 − m2 a b .16) . m2 .14) One can. m2 ) is a function symmetric in its three arguments.13) (11. 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 √ = .10) The function λ(s.12) pb = (mb . 4s (11.11) (11. m2 . Also in this case one can express the energy and momentum in the invariants. 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.9) (11. Prove that |q| = cm Ea cm Eb (s − m2 − m2 )2 − 4 m2 m2 a a b b = 4s s + m2 − m2 a b √ = . m2 ) a b . In that case trf pa = (Ea . which in the specific case a b also can be expressed as λ(s. trf Ea (threshold) = 1 2 mb ( i mi )2 − m2 − m2 a b .Scattering theory 104 It is straightforward to prove that the unknowns in the particular system can be expressed in the invariants (ma . m2 . for instance. m2 .

E .22) (11. 11.18) (11. py ). The initial state. s = (pa + pb )2 = (pc + pd )2 t = (pa − pc ) = (pb − pd ) u = (pa − pd )2 = (pb − pc )2 2 2 (11. while the final state often arises from a series of 2-particle processes combined with the decay of an intermediate particle (resonance). pc . Consider the process a + b → c + d. pT = (px .21) (11. Of course there are not only 2-particle channels. where the first is referred to as elastic channel and the set of all other channels as the inelastic channels. Instead of the scattering angle. Writing E = mT cosh y. however. tmax min λcd /4s. Consider the 1-particle inclusive case. 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 . for instance the momenta of incoming (colliding) hadrons and it is useful to use for the produced particle rapidity as a variable. there is usually a preferred direction. The various possibilities are referred to as different reaction channels. pz = mT sinh y. An often used set of invariants are the Mandelstam variables. usually many particles are produced.25) dR(s.4).24) Using the relation between t and cos θcm it is straightforward to express dΩcm in dt. H1 + H2 → h + X. Examples appear in → π− + p (11. a c with q = λab /4s and q ′ = being 0 or 180 degrees.17) (11. . The variable s is always larger than a c b d the minimal value (ma + mb )2 .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. = (11.27) with m2 = m2 − p2 = m2 + p2 + p2 .19) (11. (11. dt = 2 qq ′ d cos θcm and obtain for the two-body phase space element √ q ′ dΩcm λcd dΩcm √ = (11. usually is a 2-particle state... The minimum and maximum values for t correspond to θcm CM = m2 + m2 − 2 Ea Ec ± 2 qq ′ a c = m2 a + m2 c (s + m2 − m2 )(s + m2 − m2 ) a c b d − ± 2s √ λab λcd . pd ) = 4π s 4π 8π s 4π dt dt √ √ . in which one particle is detected. A specific reaction channel starts contributing at the threshold value (Eq. in exclusive measurements all particles are detected. the rapidity y is defined T T x y y= 1 ln 2 E + pz E − pz = ln E + pz mT = tanh−1 pz . At high energies. 2s (11.20) → π0 + n → π+ + π− + n → .23) which are not independent as s + t + u = m2 + m2 + m2 + m2 .26) = 16π q s 8π λab Kinematics of inclusive hard scattering processes In high energy (hard) scattering processes.

tt = s. tu = s. ut = u) = Mab→cd (s. u) with s = (p1 + p2 )2 = (p3 + p4 )2 . b ut = (pa − pd )2 = u. t. The one-particle phase space in terms of these variables becomes d3 p dy d|pT |2 dφ dy d2 pT |p| dη d|pT |2 dφ = = = . For the t-channel process (a¯ → ¯ one has c bd) st = (pa + pc )2 = t. ¯ while for the u-channel process (ad → c¯ one has b) su = (pa + pd )2 = u. which means that rapidity distributions dN/dy maintain their shape. b Analiticity of the field theoretically calculated result implies Ma¯→¯ (st = t. u = (p1 − p4 )2 = (p2 − p3 )2 . η = − ln(tan(θ/2)) = tanh−1 (cos θ).31) .2 Crossing symmetry In the previous chapter. uu = s) = Mab→cd (s. u). ¯ tu = (pa − pc )2 = t. (11. t = (p1 − p3 )2 = (p2 − p4 )2 . t-channel and u-channel processes respectively. (11. ¯ b (11. t. u). t. c bd Mad→c¯(su = u. Given a two-to-two scattering process ab → cd one can relate the processes b p p a 2 1 d p 4 _ c −p p 1 3 d p 4 _ d −p p a 1 4 _ b −p 2 p 3 c ab cd a _ ac −p 2 _ _ b bd _ ad p 3 _ c cb They are referred to as s-channel.30) (11.28) which (only involving angles) is easier to determine. 11. 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. For large energies and not too small scattering angles (θ ≫ 1/γ) y is approximately equal to the pseudo-rapidity η. These variables are precisely the Mandelstam variables for the s-channel process (ab → cd). uu = (pa − p¯)2 = s. ¯ tt = (pa − p¯)2 = s.Scattering theory 106 It is convenient because under a boost (along z) with velocity β it changes as y → y − tanh−1 β.29) 3 2E 3 2 (2π) 2 (2π) 16π 2π E 16π 2 2π with at high energies the factor |p|/E ≃ 1. This is known as crossing symmetry. With the momenta defined as in the figures above one has for all these processes the same amplitude M (s.

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

(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.Scattering theory 108 Although this at first sight does not look covariant. the amount of decaying particles is proportional to the number of particles with proportionality factor the em decay width Γ. . (11.35) This quantity is not covariant. 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. . For the (proper) decay width one thus has 1 N Γ0 = . .34) i. the number of scattered or decayed particles is given by 2 N = (2π)4 δ 4 (Pi − Pf ) iMf i dR(p1 . . T · V ρa (11.39) One of the δ functions can be rewritten as T · V (remember the normalization of plane waves). (11. Microscopically one has Ndecay = Na · Γ T or Γ= N 1 . 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 . it is.33) Decay of particles For the decay of particle a one has macroscopically dN = −Γ N. From the solution N (t) = N (0) e−Γ t one knows that the decay time τ = 1/Γ.T V. pn ).38) (in which we can calculate −iMf i using Feynman diagrams). 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. as expected. N 1 . dt (11.36) (11. N and T · V are covariant.T d4 x = V · T (2π)4 δ 4 (Pi − Pf ).e. 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 . .

for which one has dσ/dΩ = |f (E. q n ) q n qn √ Mni (q i .46) Partial wave expansion Often it is useful to make a partial wave expansion for the amplitude M (s.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 ). t) λab 4π dt. pn ) |M |2 q 32π 2 m2 dΩ |M |2 . respectively. √ (2ℓ + 1)Mℓ (s) Pℓ (cos θ).40) T ·V Combining this with the expressions for the width or the cross section one obtains for the decay width Γ = 1 2m = dR(m2 . (11. . .25) −i Mf i − (M † )f i = − ∗ dΩ(ˆn )Mnf (q f .45) Since the amplitudes also depend on all momenta the full result for two-particle intermediate states is (in CM. 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 . θcm ) √ q 8π s 2 2 (11. .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.e. . implies for the scattering matrix M . 11. p1 . The amplitude −M /8π s is the one to be compared with the quantum mechanical scattering amplitude f (E. . θ) or M (q i . pn ). (11.Scattering theory 109 (Using normalized wave packets these somewhat ill-defined manipulations can be made more rigorous). θ)|2 . 8π s 2π 8π s . see 11.4 Unitarity condition (S † )f n Sni = δf i The unitarity of the S-matrix. . p1 .41) (11. θ) = −8π s ℓ (in analogy with the expansion for f (E. .47) M (s. (11. . 16π 2 s (11. . ˆ ˆ Inserted in the unitarity condition for M . . The result is N = |Mf i |2 dR(s. . pn ). (M † )f n (2π)4 δ 4 (Pi − Pn ) Mni . (11. θ). p1 . i. note the sign and cos θ = qi · qf ). (11. 2 λab and for instance for two particles dσ = q ′ M (s.43) dΩcm = π M (s. . i M† M √ − √ 8π s 8π s = fi n dΩn ∗ Mnf qn Mni √ √ . θ) in quantum mechanics. for which the √ amplitudes squared have been calculated in the previous chapter. q f ).

e.54) ℓ . Using σ= dΩ q ′ M (s. θcm ) √ q 8π s 2 . parametrized as Sℓ (s) = ηℓ (s) exp(2i δℓ (s)). ℓ (11. q ˆ i. 2i q 2i q (11. the total cross section can be determined.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 . (11. which allows writing Mℓ (s) = Sℓ (s) − 1 e2i δℓ (s) − 1 = . In general a given channel has |Sℓ (s)| ≤ 1. σinel = π q2 2 (2ℓ + 1) (1 − ηℓ ).53) The difference is the inelastic cross section. † −i (Mℓ )f i − (Mℓ )f i = 2 † (Mℓ )f n qn (Mℓ )ni . η = 1) the result σel = 4π q2 (2ℓ + 1) sin2 δℓ . 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 .48) If only one channel is present this simplifies to ∗ ∗ −i (Mℓ − Mℓ ) = 2 qMℓ Mℓ . 0). (11.49) or Im Mℓ = q |Mℓ |2 . (11.52) From the imaginary part of M (s. σel = 4π ℓ (2ℓ + 1)|Mℓ (s)|2 (2ℓ + 1) ℓ = 4π q2 ηℓ e 2i δℓ − 1 2i 2 . (11.50) where Sℓ (s) satisfies |Sℓ (s)| = 1 and δℓ (s) is the phase shift. Show that σT = = 4π q 2π q2 (2ℓ + 1) Im Mℓ (s) ℓ ℓ (2ℓ + 1) (1 − ηℓ cos 2δℓ ). n (11. in combination with the partial wave expansion for the amplitudes M and the orthogonality of the Legendre polynomials immediately gives for the elastic channel.51) and for the case that this is the only channel (purely elastic scattering.

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

1 Show that the cross section for electron-electron scattering (exercise 10. a fast change in the phase shift for a narrow resonance. u) = g(t. 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). Its mass is M = 1232 MeV. This is a narrow resonance discovered in 1974. its width Γ = 120 MeV. 2 q (s − M 2 )2 + M 2 Γ2 (11. where furthermore σinel = 0.49 GeV. Its mass is M = 3096. Limiting ourselves to a resonance in one channel. Note that because of the presence of a background the phase shift at resonance may actually be shifted.59) reaching the unitarity limit for s = M 2 . Three famous resonances are: • The ∆-resonance seen in pion-nucleon scattering. the partial width into e+ e− is Γee = 5.60) showing that the phase shift at resonance rises through δ = π/2 with a ’velocity’ ∂δ/∂s = 1/M Γ. The cross section σT (π − p) also shows a resonance with the same width with a value of about 70 mb. Knowing that each neutrino contributes about 160 MeV (see next chapter). • The Z 0 resonance in e+ e− scattering with M = 91. The phase shift in the resonating channel near the resonance is given by tan δℓ (s) = MΓ . u) = 1 1 2 (s + u2 ) + 4m2 (t − m2 ) t2 2 2 1 1 ( s − m2 )( s − 3m2 ) tu 2 2 .4) can be written as 4πα2 dσ = {f (t. t)} . the full width is Γ = 88 keV.e. t) + g(u. Essentially this resonance can decay into quark-antiquark pairs or into pairs of charged leptons. i. From this one sees that a resonance has the same shape in all channels but different strength. • The J/ψ resonance in e+ e− scattering.88 MeV. M2 − s (11. Γ = 2. All these decays can be seen and leave an ’invisible’ width of 498 MeV. which is attributed to neutrinos.Scattering theory 112 (Prove this using the unitarity condition for partial waves). This characteristic shape of a resonance is called the Breit-Wigner shape.26 keV. The cross section at resonance is about 30 nb. Three neutrinos explain the resonance shape. dt s(s − 4m2 ) with f (t. u) + f (u. one can reconstruct the resonance shape for different numbers of neutrino species. u) + g(t. Exercises Exercise 11.2 GeV. it is furthermore easy to prove that the cross section is given by σel = M 2 Γ2 4π (2ℓ + 1) . At resonance the cross section σT (π + p) is about 210 mb. The half-width of the resonance is M Γ.

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

Chapter 12

The standard model
12.1 Non-abelian gauge theories

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

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

La

φ(x) = (1 + i θa La ) φ(x)

inf

(12.2)

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

(12.3)

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

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

The standard model

115

which implies W ′ = U (θ) W µ U −1 (θ) − µ or infinitesimal
′a a c Wµ = Wµ − cabc θb Wµ +

i (∂µ U (θ)) U −1 (θ), g

(12.7)

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

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

(12.8)

(12.9) (12.10)

The gauge-invariant lagrangian density is now constructed as

Gµν → U (θ) Gµν U −1 (θ).

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

(12.11)

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

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

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

The standard model
xψ x+dx

116

(x)

ψ(x)

x+dx ψ(x+dx)

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

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

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

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

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

    λ7 =   

i

    −i   

  1  1  λ8 = √   3

−2

the 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 .17) The three equations given so far immediately give Dµ = ∂µ − ig Aµ (x). (12. ds s dxµ ds′ P exp ig A (s′ ) ψ(0). 12.14) (12. (12. What is this physical effect by which the ’connection’ Aµ can be observed? For this consider the phase around a closed loop. In principle such a phase in a given point is not observable. which connects two identical vectors. The relation in Eq.The standard model the basis for x G and x+dx 117 G. i. i. if two different paths to the same point give different phases the effects can be observed. but expresses them with respect to different bases. should be independent of such transformations.e. requiring that the ’connection’ Aµ (x) is such that Dµ → S(x)D µ S −1 (x). -dx -dy µ µ dyµ µ dx For a vector.16) which in the presence of the ’connection’ Aµ differs from the total change between x ψ(x) and x+dx ψ(x+ dx).15) ψ(x).e. x+dx ψ(x) = = 1 + ig dxµ Aa (x)Ta µ 1 + ig dx Aµ (x) µ x x ψ(x) (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. Dν ] ψ(x) 1 − ig dσ µν Gµν ψ(x). x+dx ψ(x + dx) = 1 + dxµ Dµ x+dx 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 x+dx ψ(x + dx) = (1 + dxµ ∂µ ) ψ(x). (12. = . A ’constant’ vector that only rotates because of the arbitrary definitions of local frames satisfies Dµ ψ(x) = 0. ds which is solved by dψ(s) ds = dxµ .18) or considering a path xµ (s) from a fixed origin (0) to point x. However.16. (12.19) This gives rise to a (path dependent) phase in each point. 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).

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

while the quadratic terms must correspond with non-negative masses. (12.23)     F +η the latter only forming a minimum for m2 < 0.) terms constitute interaction terms. They are known as the Weyl mode or the Goldstone mode: Weyl mode. The field along the third axis plays a special role because of the choice of the vacuum expectation value. for the vacuum Qa |0 = 0. It is only invariant under rotations around the 3-axis. so one has 0|φquantum |0 = 0 and hence 0|φ|0 = ϕc .25) 2 2 2 Therefore there are 2 massless scalar particles. ϕ2 and η.24) 1 2 2 4 1 1 1 = (∂µ ϕ1 )2 + (∂µ ϕ2 )2 + (∂µ η)2 + m2 η 2 + . The classical groundstate appears degenerate. In the case of degeneracy. It is appropriate to redefine the field as    ϕ1      φ =  ϕ2  .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. The η massless particles are called Goldstone bosons. Writing the field φ as a sum of a classical and a quantum field. The presence of a nonzero value for the classical groundstate value of the field will have an effect when the field is quantized. In this case the spectrum is described . say    0    0 . It is sufficient to do this to second order in the fields as the higher (cubic.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. Any constant field with ’length’ F is a possible groundstate. The situation now is the following. therefore a choice must be made. . while the lagrangian including the nonzero groundstate expectation value chosen by nature.   ϕc = 0|φ |0 =  (12. (12.21) Stability of the action requires the classical groundstate ϕc to have a well-defined value (which can be nonzero). The original lagrangian contained an SO(3) invariance under (length conserving) rotations among the three fields. A quantum field theory has only one nondegenerate groundstate |0 . The result is 1 1 1 1 (∂µ ϕ1 )2 + (∂µ ϕ2 )2 + (∂µ η)2 − m2 (ϕ2 + ϕ2 ) L = 1 2 2 2 2 2 1 2 1 − m (F + η)2 − λ (ϕ2 + ϕ2 + F 2 + η 2 + 2 F η)2 (12. etc. In this mode the lagrangian and the vacuum are both invariant under a set of symmetry transformations generated by Qa . (12. λ such that 0|ϕ1 |0 = 0|ϕ2 |0 = 0|η|0 = 0. i. . .e. In order to see the consequences for the particle spectrum of the theory we construct the lagrangian in terms of the fields ϕ1 . φ = ϕc + φquantum where for the (operator-valued) coefficients in the quantum field one wants 0|c† = c|0 = 0. In this situation one speaks of spontaneous symmetry breaking. corresponding to the number of broken generators (in this case rotations around 1 and 2 axis) and 1 massive scalar particle with mass m2 = −2m2 . has less symmetry.

called flavors.e. a ∂ µ Jµ = 0.31) . π (12. 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′ .    . T ] ψ. down. i. etc. π Example: chiral symmetry An example of spontaneous symmetry breaking is chiral symmetry breaking in QCD. Goldstone mode.27) If the transformations in the lagrangian give rise to a symmetry the Noether currents are conserved. denoted |π a (k) . V µ = ψγ µ T ψ. suggesting the stronger relation ∂ µ Jµ (x) = fπ m2 a π a (x). = ψ(i/ − M )ψ. Note that for the fields π a (x) one would have a the relation 0|π a (x)|π a (k) = ei k·x . strange. . Taking the derivative.) one can write L = f ψ f (i/ − Mf )ψf . i. Using translation invariance and as kµ is the only four vector on which this matrix element could depend one may write a 0|Jµ (x)|π b (k) = fπ kµ ei k·x δab (fπ = 0) (12. The conserved currents corresponding to this symmetry transformation are found directly using Noether’s theorem (see chapter 6).      ψu   mu           md     ψ =  ψd  . a and a′ are degenerate states. As the a generators for a symmetry are precisely the zero-components of a conserved current Jµ (x) integrated a a over space. M =      . H] = 0. i. the lagrangian for the quarks consists of the free Dirac lagrangian for each of the types of quarks.e.28) where ψ is extended to a vector in flavor space and M is a diagonal matrix in flavor space. 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).30) which in the case that we include only two flavors form an SU (2)V symmetry (isospin symmetry) generated by the Pauli matrices. a 0|∂ µ Jµ (x)|π b (k) = fπ k 2 ei k·x δab = fπ m2 a ei k·x δab . [Qa . and one must have mπa = 0.26) for all the states labeled by a corresponding to ’broken’ generators. (12. In this mode the lagrangian is invariant but Qa |0 = 0 for a number of generators. A bit more formal.  .32) (12.    . This means that they are operators that create states from the vacuum. This lagrangian density then is invariant under unitary (vector) transformations in the flavor space. T = τ /2. a massless Goldstone boson for each ’broken’ generator. irrespective of the fact if they annihilate the vacuum. 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.e.The standard model 120 by degenerate representations of the symmetry group. (12. it is easy to see that one gets ∂µ V µ = i ψ [M. (12. there must be a nonzero expectation value 0|Jµ (x)|π (k) . ∂ ∂ (12. Known examples are rotational symmetry and the fact that the spectrum shows multiplets labeled by angular momentum ℓ (with members labeled by m).. The generators Qa (in that case the rotation operators Lz . Neglecting at this point the local color symmetry. ψ −→ ei α·T ψ. Including a sum over the different flavors (up. if the generators Qa generate a symmetry.

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

4 2 where Since the explicit (adjoint. made into a gauge theory. (12.37) (12. 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). 1 1 L = − Gµν · Gµν + Dµ φ · Dµ φ − V (φ).    0      ϕc = 0|φ |0 =  0  . is made.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     .e. . Dµ φ = Gµν = ∂µ φ + g Wµ × φ. in this case three-dimensional) representation reads (La )ij = −i ǫaij .39) (12. The difference comes when we reparametrize the field φ. (12. like a photon. We have the possibility to perform local gauge transformations. (12. one sees that the fields Wµ and Gµν also can be represented as three-component fields.The standard model 122 12.     F which gives for the lagrangian density up to quadratic terms L 1 1 λ 1 = − Gµν · Gµν + Dµ φ · Dµ φ − m2 φ · φ − (φ · φ)2 4 2 2 4 1 2 2 1 1 2 = − (∂µ Wν − ∂ν Wµ ) · (∂ µ W ν − ∂ ν W µ ) − g F Wµ W µ 1 + Wµ W µ 2 4 2 1 + (∂µ η)2 + m2 η 2 + . 2 massive vector fields or spin 1 bosons (2 × 3) and one scalar field (1). After symmetry breaking the same number (as expected) comes out.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. Considering the same example with rotational symmetry (SO(3)) as for spontaneous symmetry breaking of a scalar field (Higgs field) with three components. Initially there are 3 massless gauge fields (each.    0 φ= (12. two massive bosons (Wµ and Wµ with MW = gF ) and a massive scalar particle (η with m2 = η 2 −2 m . as in this case there are no massless Goldstone bosons. having two independent spin components) and three scalar fields (one degree of freedom each).40) a Dµ φ = ∂µ φ − ig Wµ La φ. i. . Therefore we can always rotate the field φ into the 3-direction in order to simplify the calculation. but one has 1 massless gauge field (2). One may wonder about the degrees of freedom.     0  0         0 = . . thus 9 independent degrees of freedom.38) ∂µ Wν − ∂ν Wµ + g Wµ × Wν . . while the number of massive gauge bosons coincides with the number of ’broken’ generators.   The symmetry is broken in the same way as before and the same choice for the vacuum. again 9 degrees of freedom.42) 2 from which one reads off that the particle content of the theory consists of one massless gauge boson 3 1 2 (Wµ ). The latter is a spin 0 particle (real scalar field) called a Higgs particle.

while the lefthanded particles form a doublet. electromagnetic and weak forces. ∂ ∂ ∂ (12. ψR/L = 2 (1 ± γ5 )ψ decouple. (12.50) (12. the leptons (elektrons. muons. left. explicitly L R SU(2)W SU(2)W −→ ei α·τ /2 L.   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.and righthanded particles.45) (12. etc.43) and One introduces a (weak) SU (2)W symmetry under which eR forms a singlet.). explicitly Dµ L Dµ R i i g Wµ · τ L + g ′ Bµ L. the quarks (up.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. Thus the lagrangian density is L (f ) ∂ ∂ = i L/ L + i R/ R.44) which has an SU (2)W symmetry under transformations eiα·TW .) and the gauge bosons responsible for the strong. (12. i. The lagrangian for the leptons consists of three families each containing an elementary fermion (electron e− . R −→ e−i β R. 1 As they are massless. νµ and ντ ) and their antiparticles. For the neutrino only a lefthanded particle (and righthanded antiparticle) exist.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. U(1)Y U(1)Y (12. Thus L (f ) = i eR / eR + i eL / eL + i νeL / νeL + (µ.48) (12.47) and YW is considered as an operator that generates a U (1)Y symmetry. The procedure. that gives a good description of the physical world. explicitly L −→ e−i β/2 L. This is written as 3 Q = TW + YW . implies certain nontrivial relations between masses and mixing angles that can be tested experimentally and sofar are in excellent agreement with experiment.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. down. neutrinos. In the standard model one starts with a very simple basic lagrangian for (massless) fermions which exhibits more symmetry than observed in nature. however. under which the lefthanded and righthanded particles with YW (L) = −1 and YW (R) = −2 transform with eiβYW /2 . 2 2 = ∂µ R + i g ′ Bµ R. 2 (12. its corresponding neutrino (νe . By introducing gauge fields and breaking the symmetry a more complex lagrangian is obtained. = ∂µ L − (12.51) . etc.e.The standard model 123 12. τ ). R. muon µ− or tau τ − ).

(12. with the following lagrangian density consisting of a symmetry breaking piece and a coupling to the fermions.53) φ=   √ (θ4 − iθ3 )  φ 2 with TW = 1/2 and YW = 1 is introduced.59) . a complex Higgs field   +   1  φ   √2 (θ2 + iθ1 )    0 = 1   (12. Using local gauge invariance θi for i = 1. Putting this in leads to L (f ) =L (f 1) +L (f 2) . leading to the parametrization   1  0    φ(x) = √  (12. this leads to the lagrangian L (h1)     . .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 .The standard model 124 3 3 where Wµ is a triplet of gauge bosons with TW = 1. (12. L where L (h1) (h) =L (h1) +L (h2) . . † i i g Wµ · τ − g ′ Bµ )φ. −V (φ) L and (h2) = −Ge (LφR + Rφ L). . + 8 g2 v2 1 + − (Wµ )2 + (Wµ )2 (∂µ h)2 + m2 h2 + 2 8 (g 2 + g ′2 ) v 2 (Zµ )2 + . the U (1)Q symmetry (generated by the charge operator).58) (12.57) = 1 g2 v2 1 2 (Wµ )2 + (Wµ )2 (∂µ h)2 + m2 h2 + 2 8 v2 2 3 gWµ − g ′ Bµ + . which assures with the choice of YW for the Higgs field that Q generates the remaining U (1) symmetry. .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. 4 4 In order to break the symmetry to the symmetry of the physical world. For the classical field the choice θ4 = v is made. 2 and 3 may be eliminated (the necessary SU (2)W rotation is precisely e−iθ(x)·τ ). 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 = 0.56)  v + h(x) 2 Dµ φ = (∂µ − and     Dµ φ =     = 1 √ 2 − ig 2 ∂µ h + 1 2 Wµ −iWµ √ (v + h) 2 3 gWµ −g′ Bµ i √ (v + 2 2 Up to cubic terms. + 8 (12. .54) = (Dµ φ)† (Dµ φ) −m2 φ† φ − λ (φ† φ)2 . (12.   h)  (12.

63) (12.62) g 2 + g ′2 3 ≡ sin θW Wµ + cos θW Bµ . g ′ /g = tan θW . 2 (12. for instance in the amplitude for the decay of the muon νµ µ− W − e− νµ = µ− e− νe νe is given by −i M = − ≈ i g2 −i gρσ + . .61) (12. σ (νµ γ ρ µL ) 2 2 (eL γ νe ) 2 k + MW (12. The coupling of the fermions to the physical gauge bosons are contained in L (f 1) giving L (f 1) = i eγ µ ∂µ e + i νe γ µ ∂µ νe − g sin θW eγ µ e Aµ g 1 1 + sin2 θW eR γ µ eR − cos 2θW eL γ µ eL + νe γ µ νe cos θW 2 2 g µ + µ − + √ νe γ eL Wµ + eL γ νe Wµ .64) (12.67) The coupling of electrons or muons to their respective neutrinos.60) (12. (12.68) g2 ρ 5 2 (νµ γρ (1 − γ5 )µ) (eγ (1 − γ )νe ) 8 MW (µ)† (jL )ρ (e) (jL )ρ GF (µ)† (e) ≡ i √ (jL )ρ (jL )ρ .66) From the coupling to the photon.69) . 2 3 g Wµ − g ′ Bµ 3 ≡ cos θW Wµ − sin θW Bµ .65) The weak mixing angle is related to the ratio of coupling constants. 4 2 MW g2 v2 = . . 2 Zµ (12. we can read off e = g sin θW = g ′ cos θW . (12. 2θ 2θ 4 cos W cos W 0. 2 + g ′2 g 3 g ′ Wµ + g Bµ (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 .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 .

Γe ≈ Γµ ≈ Γτ ≈ 83 MeV). Comparing this with the total width into (invisible!) channels.40 GeV. 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. 2 The value of α deviates from the known value α ≈ 1/137 because of higher order contributions. one obtains L (h2) Ge v Ge = − √ (eL eR + eR eL ) − √ (eL eR + eR eL ) h 2 2 me = −me ee − eeh.73) (12. leptons with CV = −1/2 + 2 sin2 θW ≈ −0. −5 (12. 1. The coupling to the Higgs particle is weak as the value for v calculated e.05 and CA = −1/2 we calculate Γ(e+ e− ) ≈ 78. i.77) (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 . (12. 2 2 2 (12.166 4 × 10 0.74) (12.76) with CV CA 3 = TW − 2 sin2 θW Q. such as2 α = e2 /4π GF sin2 θW MW MZ ≈ 1/127.80) First. Γinvisible = 480 MeV one sees that three families of (light) neutrinos are allowed. as appear for instance in the decay of the neutron or the decay of the Λ (quark content uds).g. one has the relation g2 e2 1 GF √ = .78) From this coupling it is straightforward to calculate the partial width for Z 0 into a fermion-antifermion pair.19 GeV. (12. Finally we want to say something about the weak properties of the quarks. giving rise to a running coupling constant after renormalization of the field theory.70) = 2 = 2 sin2 θ 8 MW 2 v2 8 MW 2 W In this way the parameters g. v (12. the mass of the electron comes from the spontaneous symmetry breaking but is not predicted (it is in the coupling Ge ). i. from the MW mass is about 250 GeV.5 MeV (exp.79) Γ(Z 0 → f f ) = 48π cos2 θW For the electron.231 2. muon or tau.The standard model 126 the good old four-point interaction introduced by Fermi to explain the weak interactions. With the choosen vacuum expectation value for the Higgs field. 3 = TW . g2 MZ 2 2 (CV + CA ).71) GeV −2 = = = = . For each neutrino species (with CV = 1/2 and CA = 1/2 one expects Γ(νν) ≈ 155 MeV. me /v is extremely small.e. The masses of the fermions and the coupling to the Higgs particle are contained in L (h2) .e. (12. 91.72) (12. 80. . g ′ and v determine a number of experimentally measurable quantities. (12.

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

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 . ms and mb ). UR = uR cR tR . d d (12.34. i. The imaginary part i η gives rise to CP violation in decays of K and B-mesons (containing s and b quarks. respectively). (12. (12.e.The standard model 128 using the socalled Wolfenstein      + O(λ4 )   with λ ≈ 0. the weak mass eigenstates are L (h2. 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 . The Λu and Λd are complex matrices in the 3×3 family and Λd Λ† = Vd G2 Vd† . The Higgs field is still limited to one complex doublet. e.q) ′ weak † mass mass DL = DL = Vu Vd DL = VCKM DL .q) = −QL φΛd DR − DR Λ† φ† QL − QL φc Λu UR − UR Λ† φc† QL u d (12. .   φ =    −φ− 0 2 For the (squared) complex matrices we can find positive eigenvalues. for which we need the appropriate weak isospin doublet  0∗    1    φ c  = √  v +h .86) the unitary CKM-matrix introduced above in an ad hoc way. Thus one has † † † =⇒ −DL Vd Md Wd DR − DR Wd Md Vd† DL − UL Vu Mu Wu UR − UR Wu Mu Vu UL (12.227.82) space. allowing us to write † Λu = Vu Gu Wu † and Λd = Vd Gd Wd . u u where we include now the three lefthanded quark doublets in QL .5 Family mixing in the Higgs sector and neutrino masses The quark sector Allowing for the most general (Dirac) mass generating term in the lagrangian one starts with L (h2. 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. the three righthanded quarks with charge +2/3 in UR and the three righthanded quarks with charges −1/3 in DR .g. Note that we need the conjugate Higgs field to get a U (1)Y singlet in the case of the charge +2/3 quarks. mc and mt ) and Md = Gd v/ 2 (diagonal matrix containing md . each of these containing   the three families.84) with Gu and Gd being real and positive and Wu and Wd being different unitary matrices. A ≈ 0.85) √ √ with Mu = Gu v/ 2 (diagonal matrix containing mu . † Λu Λ† = Vu G2 Vu .82 and ρ ≈ 0.83) where Vu and Vd are unitary matrices.22 and η ≈ 0.

i.92) . For neutrino’s this matrix is parametrized in terms of three angles θij with cij = cos θij and sij = sin θij and one angle δ. † † Again we continue as before now with matrices Λe = Ve Ge We and Λn = Vn Gn Wn . UPMNS a parametrization that in principle also could have been used for quarks.    (12. EL = Ve† EL . e   N L= L  EL     (12. The mixing is intriguingly close to the Harrison-Perkins-Scott tri-bimaximal 23 mixing matrix      0 0 2/3 1/3 0 2/3 1/3 0     1                 1/2 − 1/2   − 1/3 UHPS =  0 1/3 − 1/2  .ℓ) =⇒ −Me EL Ve We ER − ER We Ve† EL . The lepton sector (massive Dirac neutrinos) In principle a massive Dirac neutrino could be accounted for by a lagrangian of the type L (h2.88) √ mass with Me = Ge v/ 2 the diagonal mass matrix with masses me . and obtain L (h2.89) with three righthanded neutrinos added to the previous case. decoupling from all known interactions.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 . since the weak current is the only place where they show up.5×10−3 eV2 . it is particularly useful because θ12 is essentially determined by solar neutrino oscillations requiring ∆m2 ≈ 8 × 10−5 12 eV2 (convention m2 > m1 ). In this case.The standard model 129 The lepton sector (massless neutrinos) For a lepton sector with a lagrangian density of the form L in which (h2. The mass fields ER † mass † mass = We ER . For the (massless) neutrino fields we just can redefine fields into NL = Ve† NL .90) mass † mass mass † mass † We note that there are mass fields ER = We ER . (12. As before.  =  − 1/6 2/3 0           1/2 1/2 0 0 1 0 1/3 1/2 − 1/6 (12.ℓ) = −LφΛe ER − ER Λ† φ† L − Lφc Λn NR − NR Λ† φc† L e n (12. one can write Λe = Ve Ge We and we find † L (h2. mµ and mτ .e. there is no family mixing for massless neutrinos.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    . we find massless neutrinos.ℓ) = −LφΛe ER − ER Λ† φ† L.87) is a weak doublet containing the three families of neutrinos (NL ) and charged leptons (EL ) and ER is † a three-family weak singlet. (12. while θ23 then is determined by atmospheric neutrino oscillations requiring |∆m2 | ≈ 2. (12. NL = Vn NL and NR = Wn NR µ mass γ µ V † V N mass . The W -current then becomes mass mass E L γ µ NL = E L γ µ NL . EL = Ve EL .ℓ) † † † =⇒ −EL Ve Me We ER − ER We Me Ve† EL − NL Vn Mn Wn NR − NR Wn Mn Vn NL .

couple the right.98) (12. 1 L Υ ′ = nR + nc . The way to circumvent this is to introduce as in the previous section righthanded neutrinos. 12. the left.The standard model 130 The lepton sector (massive Majorana fields) An even simpler option than sterile righthanded Dirac neutrinos. 2 R ψ = nR + nL . with for the righthanded sector a mass term MR . the massless and massive Majorana neutrinos.99) . Υc ≡ (ΥL )c = ΥR 1 2 L and Υc = ΥL ). −    MD MR nR 2 (12.100)  |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. If the Majorana mass MR ≫ MD one actually obtains in a natural way one Majorana neutrino with a very small mass. The mass matrix can be written as   ML |MD | eiφ     M = (12. are equivalent to two decoupled Majorana neutrinos (see below). however. For the charged leptons there cannot exist a Majorana mass term as this would break the U(1) electromagnetic symmetry.and lefthanded species through Dirac mass terms coming from the coupling to the Higgs sector as in the previous section. as above. We use here the primes starting with the weak eigenstates. a unitary matrix relating them to the weak eigenstates. coupled by a Dirac mass term. one being massive. 1 ∗ c c (12. Actually. Thus (disregarding family structure) one has two Majorana neutrinos.ν =− containing three (CP-violating) phases (α1 . Thus 2 Υ ′ = nc + nL . Absorption of phases in the states is not possible for Majorana neutrinos. 1 c ∗ c L mass.ν = − MR NR NR + MR NR NR . The see-saw mechanism Consider (for one family N = n) the most general Lorentz invariant mass term for two independent Majorana spinors.96) VPMNS = UPMNS K with K =  0 eiα2 /2 0  . For the leptons. hence the mixing matrix becomes   iα /2 0 0   e 1     (12. (12.95) 2 In order to have more than a completely decoupled sector. Υ′ and Υ′ (satisfying Υc = Υ and as discussed in chapter 6. it is easy to see R that this incorporates the Dirac case by considering the lefthanded part of Υ′ and the righthanded 1 part of Υ′ as a Dirac spinor ψ.       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. 2 although this option is not attractive as it violates the electroweak symmetry.c.88 a Majorana mass term for the (lefthanded) neutrino mass eigenstates. For these light Majorana neutrinos one has. is to add in Eq. (12. one must for the neutrinos as well as charged leptons.94) M L NL NL + M L NL NL .and righthanded species then just form a Dirac fermion. For the neutrino sector. This is called the see-saw mechanism (outlined below). α2 and δ).

The standard model 131 taking ML and MR real and non-negative.  L      Υ2L nR Υ1 and Υ2 . H] = 0. Exercise 12.104) ↔ 1 1 Υi i / Υi − Mi Υi Υi ∂ 4 2 (12. (12. related via    c   Υ1R     = U  nL  . This is a mixing problem with a symmetric (complex) mass matrix leading to two (real) mass eigenstates. T ] 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 . For the situation ML = 0 and MR ≫ MD (taking MD real) one 2 finds M1 ≈ MD /MR and M2 ≈ MR . 2 with real masses Mi . 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′ . which implies U ∗ M † U † = U ∗ M ∗ U † = M0 and a ’normal’ diagonalization of the (hermitean) matrix M M † . i ψ {M. [Qa . (12.1 Consider the case of the Weyl mode for symmetries. T } γ5 ψ.102) = 1 2 2 ML + MR + 2|MD |2 2 ± 2 2 2 2 (ML − MR )2 + 4|MD |2 (ML + MR + 2ML MR cos(2φ)) .103) for each of which one finds the lagrangians L = and we are left with two decoupled Majorana fields      Υ1L    = U ∗  nc  . 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).e.      Υ2R nR (12. V µ = ψγ µ T ψ and Aµ = ψγ µ γ5 T ψ ∂µ V µ ∂µ A µ = = i ψ [M. Prove that if the generators Qa generate a symmetry. a and a′ are degenerate states.  (12. i.2 Derive for the vector and axial vector currents. i.105) for i = 1. Exercises Exercise 12. Exercise 12. 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 . 2 U (M M † ) U † = M0 .e.101) Thus one obtains from  2 ML + |MD |2  † MM =   |MD | ML e+iφ + MR e−iφ the eigenvalues 2 M1/2 |MD | ML e−iφ + MR e+iφ 2 MR + |MD |2   .

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

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

Sign up to vote on this title
UsefulNot useful

Master Your Semester with Scribd & The New York Times

Special offer: Get 4 months of Scribd and The New York Times for just $1.87 per week!

Master Your Semester with a Special Offer from Scribd & The New York Times