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

An introduction to Quantum Field Theory. Corresponding chapters in books of Ryder.1 4. 1994.3.5 2. Zuber.2 4. 1. Other text books of Quantum Field Theory that are useful are given in refs [3-6]. Cambridge University Press. S. Veltman.4 6.3 3. Diagrammatica. II: Modern Applications.3 3.4 3.2 2.4 3.5 2.5 4.1 6. L. Peskin & Schroeder and Srednicki. 3. 2.4 4.1 3.-B.3 2.2 2. The notes contain all essential information.H.E.2 3.2 3. but are rather compact. Itzykson and J. I: Foundations. Quantum Field Theory.2 3.3 3. McGraw-Hill. Ryder.3 2. Cambridge University Press. 1995. 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). M.7 2.2 2.1 2. These notes Section 1. 1996.3 2. Addison-Wesly.3 4. 6.2 1.4 3. Vol. Cambridge University Press. 1995. Srednicki.4 2. Weinberg. 4. 1985.2. I refer to the book of Srednicki [1] and Ryder [2].1 5.3 2.3 3. Quantum Field Theory.1 2 2 3.2 3. Schroeder.8 3. Quantum Field Theory.2 36 22 . 2.3 6. 1980.3 3.1 2.3 References As most directly related books to these notes.5 Ryder Peskin & Schroeder Srednicki 1 1 1 1 3 2.2 3.3 2. Cambridge University Press. M. Vol. M. 2007.4 2.V. 3. Cambridge University Press.1 1.1. Peskin and D.5 5.1 3. 3.2 6.2 6. 5. C. The quantum theory of fields.

5 7. 4.1.2 8.4.3 20.1 9.5 6.3 11.6 9.2.4. 4. 2.3 7.3 10.4.6 3.4 4.5 4.2.4 10.10 8.1 7.4 2.8 6.7.3.4 7. 9.3 6 8 9 Srednicki 3 3 3 37 5.6 6.3 8. 6.2 9. 6.2 5.2 12.2 4.3 7.7 5.3.1 11.7 4.5 9.4 8.3 3.3 12. 9. 2. 20. 6.6 4.2 10.3 6.6 10.5 .5 4.5 Ryder 4 4 4 4 4 4 Peskin & Schroeder 2.8 5 4.5 11.2 7.3.5 9.4 9.1 10.2 8.5 2.1 20.6 3.1 8. 4. 6.6 3.1 9.1 12.5.5 7.3 9. 4.3 2. 8.4 12.2.1 5.3 8.1 20. 6.4 12.5 6 7 6 6.1.2.4 3.2 11.1.6 8.4 These notes Section 7.1 20.

3 • First two sentences of Chapter 5 have been modified.5 Corrections 2011/2012) • Extension of Exercise 3.2 . • Slight rewriting of discussion of Majorana mass term in section 6.

(1. being eigenvalues of the position operators.4) where δij is the Kronecker δ function. In special relativity a special role is played by the velocity of light c. pj ] = i δij . In the framework of classical mechanics as well as quantum mechanics the position of a particle is a well-defined concept and the position coordinates can be used as dynamical variables in the description of the particles and their interactions. In this coordinate representation the position operators r op simply acts as rop ψ(r) = r ψ(r). (1. usually given divided by 2π.6) . One can talk about states |r and the wave function ψ(r) = r||ψ . It is given by pop ψ(r) = −i ∇ψ(r). c = 299 792 458 m s−1 .054 571 68 (18) × 10−34 J s = 6. (1.Chapter 1 Introduction 1.1) In the limit that the action S is much larger than . They satisfy the well-known (canonical) operator commutation relations [ri . the action of the momentum operator in the coordinate representation is not as simple as the position operator.1 Quantum field theory In quantum field theory the theories of quantum mechanics and special relativity are united.5) One can also choose a representation in which the momenta of the particles are the dynamical variables. In quantum mechanics. S ≫ . ˜ The corresponding states are |p and the wave functions ψ(p) = p||ψ are the Fourier transforms of the coordinate space wave functions. ≡ h/2π = 1. the position can in principle be determined at any time with any accuracy. (1.2) In the limit that v ≪ c one reaches the non-relativistic domain. ˜ ψ(p) = i d3 r exp − p · r ψ(r). quantum effects do not play a role anymore and one is in the classical domain.582 119 15 (56) × 10−22 MeV s. 1 (1.3) The uncertainty principle tells us that in this representation the momenta cannot be fully determined. (1. In quantum mechanics a special role is played by Planck’s constant h. Corresponding position and momentum operators do not commute. Indeed.

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.

linear transformations that do not change the length of a four-vector are called the Lorentz transformations. (1. ˆ ˆ (1. • ds2 < 0 (spacelike intervals).Introduction 5 that can be used in the cross product of two vectors z = x × y. The lower indices are constructed in the following way.22) (1. independent of a coordinate system. The (invariant) lengths often have special names. δim δjm δkm δin δjn δkn . . we will use upper indices for the three-vectors. the points lie on the lightcone and they can be connected by a light signal. given by g00 = −g11 = −g22 = −g33 = 1 (the ˆ ˆ other components are zero).24) The quantity gµν ≡ nµ · nν is the metric tensor. (1.25) Within Minkowski space the real. 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 .3) with basis nµ (µ = ˆ 0. x2 . are .1. The Poincar´ transformations include e e Lorentz transformations and translations. One has x2 = xµ xµ = t2 − x2 .26) (1. For four-vectors in Minkowski space we will use the notation with upper indices and write x = (t. These transformations do change the components of a vector. where the coordinate t = x0 is referred to as the time component. such as eigentime τ for the position vector τ 2 ≡ x2 = t2 − x2 .µn = Λµ1ν1 .. one must distinguish tensors with upper or lower indices and one can have mixed tensors. x) = (x0 . No difference is made between upper or lower. In Minkowski space. Useful relations are ǫijk ǫℓmn ǫijk ǫimn ǫijk ǫijl = = = δiℓ δjℓ δkℓ 2 δkl . . Λµn n T ν1 . The length (squared) of a vector is obtained from the scalar ˆ product.. Unlike in Euclidean space. Vectors are denoted x = xµ nµ . The invariant distance between two points x and y in Minkowski space is determined from the length dsµ = (x − y)µ .. 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. Relations relating tensor expressions. Many other four vectors and tensors transforming like T ′ µ1 . We note that for Euclidean spaces (with a positive definite metric) vectors and tensors there is only one type of indices. x3 ) = (t. and are given by (x0 . The real. When 3-dimensional space is considered as part of Minkowski space. x2 . Because of the different signs occurring in gµν .2. x2 = x · x = xµ xν nµ · nν = xµ xν gµν .23) δjm δkn − δjn δkm . x1 . In special relativity we start with a four-dimensional real vector space E(1. • ds2 = 0 (lightlike or null intervals). however. So we could have used all upper indices in the above equations. linear transformations that leave the length of a vector invariant are called (homogeneous) Lorentz transformations. One can distinguish: • ds2 > 0 (timelike intervals).3). 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 . it is convenient to distinguish lower indices from upper indices. −x). 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.21) (1. x1 . denoted as V ′µ = Λµν V ν ..νn can be conν structed. the invariant length or distance (squared) is not positive definite. xµ = gµν xν . xi are the three space components. in which case zi = ǫijk xj yk . x3 ).

28) ν ν Note that gµ with one upper and lower index. gµ = gµρ g ρν and 0 1 2 3 is in essence a ’Kronecker delta’. . Relate this quantity to the electron Compton wavelength −e via the dimensionless λ fine structure constant α. For the contraction of a vector with the antisymmetric tensor one often uses the shorthand notation ǫABCD = ǫµνρσ Aµ Bν Cρ Dσ .∇ .27) It is easy to convince oneself of the nature of the indices in the above equation.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. Examples are the scalar products above but also relations like pµ = m dxµ /dτ for the momentum four vector. defined by ∂2 − ∇2 . because one has (1. (1.Introduction 6 called covariant. Each of these permutations leads to a minus sign. The derivative ∂µ is defined ∂µ = ∂/∂xµ and we have a four vector ∂ with components (∂0 . but more important has the same effect on lefthandside and righthandside.33) (1. explicitly written as (p0 . ∂2 . (1. starting from ǫ0123 = 1. The product of two epsilon tensors is given by g µµ ′ g νµ ′ g ρµ ′ g σµ g νν ′ g ρν ′ g σν ′ ′ ′ ǫµνρσ ǫµ ν ′ ′ ′ ρ σ′ = − g µν ′ g νν ′ g ρν ′ g σν g νρ ′ g ρρ ′ g σρ ′ ′ ′ g µρ ′ g νρ ′ g ρρ ′ g σρ g νσ ′ g ρσ ′ g σσ ′ ′ ′ g µσ ′ g νσ ′ g ρσ ′ g σσ . a∞ = 4πǫ0 2 /me e2 . ∂1 . = ∂ . It is an invariant tensor.31) ǫµνρσ ǫµν ′ ′ ρσ ′ = − (1. ∂t (1. not affected by Lorentz transformations. 0123 0123 ′ ′ ǫ µνρσ ǫµνρσ The first identity.29) 2 = ∂ µ ∂µ = ∂t2 The value of the antisymmetric tensor ǫµνρσ is determined in the same way as for ǫijk .34) (1. = −24. is timelike with invariant length (squared) p2 = p · p = pµ pµ = E 2 − p2 = m2 . p). p) = (E. . g0 = g1 = g2 = g3 = 1. (1. where m is called the mass of the system. Note that in this equation one has on left.35) ρ ǫµνρσ ǫµν σ σ ǫµνρσ ǫµνρ ′ ′ = −2 g ρρ g σσ − g ρσ g σρ = −6g σσ . (1. .32) .30) (Note that there are different conventions around and sometimes the opposite sign is used). 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. constructed via the metric tensor itself. for instance.36) Exercises Exercise 1. ∂3 ) = ∂ ∂ ∂ ∂ . ′ ′ . The length squared of ∂ is the d’Alembertian operator.1 (a) In the the Hydrogen atom (quantum system) the scale is set by the Bohr radius. ∂t ∂x ∂y ∂z ν ∂µ xν = gµ . (1.

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

1) under the substitution (in coordinate representation) E −→ Eop = i ∂ . we just want to play a bit with covariant equations and study their behavior under Lorentz transformations.3) Equations 2.1 and 2. (2.2. t).4) where M is the particle mass.5) with (k 0 )2 = k2 + M 2 . written as pµ −→ i∂µ is covariant (the same in every frame of reference). E = p2 . ∂t2 (2. p2 = pµ pµ = E 2 − p2 = M 2 . (2. ∂t 2M (2.6) The energy spectrum is not bounded from below: considering the above stationary plane wave solutions. (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. 2M (2. 8 .1 The Klein-Gordon equation In this chapter. ∂t p −→ pop = −i∇. i ∂ ∇2 ψ(r. The Schr¨dinger equation in quantum mechanics is the operator equation o corresponding to the non-relativistic expression for the energy. But the replacement 2. t) = ∂2 − ∇2 + M 2 φ(r. t) = − ψ(r. t) = 0. 2 + M 2 φ(r. Although it is straightforward to find the solutions of this equation. Substitution of operators gives the Klein-Gordon (KG) equation for a real or complex function φ. p). Thus a covariant equation can be obtained by starting with the (covariant) equation for the invariant length of the four vector (E. t) = exp(−i k 0 t + i k · r).7) k 0 = ± k2 + M 2 = ±Ek . one obtains (2. namely plane waves characterized by a wave number k.2) Acting on the wave function one finds for a free particle. φk (r.Chapter 2 Relativistic wave equations 2.3 are not covariant.

for which we need the interpretation of φ itself as an operator. we want the most general solution. It is possible to rewrite it as an integration over positive energies only but this gives two terms (use the result of exercise 2. −k) . relativistically the density is not a scalar quantity.2 Mode expansion of solutions of the KG equation Before quantizing fields. 2M 2M (2. This operator has all possible solutions in it multiplied with creation (and annihilation) operators. j) of the four-current j. Probability is not positive: in quantum mechanics one has the probability and probability current ρ j = = ψ∗ ψ ↔ i i − (ψ ∗ ∇ψ − (∇ψ ∗ )ψ) ≡ − ψ ∗ ∇ ψ. φ(x) = d4 k ˜ 2π δ(k 2 − M 2 ) e−i k·x φ(k) (2π)4 (2.9) They satisfy the continuity equation. The KG equation. The integration over k-modes clearly is covariant and restricted to the ‘mass’-shell (as required by Eq. ↔ ↔ (2.2) is j µ = i φ∗ ∂ µ φ or (ρ. −i φ∗ ∇ φ . however. 2. which is what Lorentz transformations do. having the KG equation as a space-time symmetric (classical) equation. (2π)3 2Ek (2. but rather the zero component of a four vector. j) = ↔ i φ∗ ∂0 φ.5).11) Therefore. (2.14) ˜ ˜ Introducing φ(Ek . (2. At that point the dependence on position r and time t is just a dependence on numbers/parameters on which the operator depends. rather than as a wave function.15) . k) + ei k·x φ(−Ek . 2. For this we note that an arbitrary solution for the field φ can always be written as a superposition of plane waves.13) ˜ with (in principle complex) coefficients φ(k). is a second order equation and φ and ∂φ/∂t can be fixed arbitrarily at a given time. This continuity equation can be written down o covariantly using the components (ρ. φ(x) = d3 k (2π)3 2Ek ˜ ˜ e−i k·x φ(Ek . there are solutions with negative energy. ∂µ j µ = 0. −k) ≡ b∗ (k) one has φ(x) = d3 k e−i k·x a(k) + ei k·x b∗ (k) = φ+ (x) + φ− (x). (2. k) ≡ a(k) and φ(−Ek . just as the dependence on time was in ordinary quantum mechanics. ∂ρ = −∇ · j.3). Then. there are no longer fundamental objections to mix up space and time. The appropriate current corresponding to the KG equation (see Excercise 2.10) ∂t which follows directly from the Schr¨dinger equation. As we will see later both problems are related and have to do with the existence of particles and antiparticles. it is simply a matter of being careful to find a consistent (covariant) theory.e.12) It is easy to see that this current is conserved if φ (and φ∗ ) satisfy the KG equation. And. This leads to the existence of negative densities.8) (2.Relativistic wave equations 9 i.

19) implying that for each solution φ(x) there exists a corresponding solution with the same energy. −x) ≡ (˜µ ). Another symmetry is found by complex conjugating the KG equation. 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). x) → (t.22) showing that with each solution there is a corresponding charge conjugated solution φC (x) = φ∗ (x). after restricting the energies to be positive). x ∂µ ∂ µ + M 2 φ(˜) = 0. (2. It is trivial to see that ∂µ ∂ µ + M 2 φ∗ (x) = 0. The coefficients a(k) and b∗ (k) are the amplitudes of the two independent solutions (two.Relativistic wave equations 10 In Eqs 2. x Transforming everywhere in the KG equation x → x one obtains ˜ ˜ ˜ x ∂µ ∂ µ + M 2 φ(˜) = 0.23) . one thus must have that the function φ → φ′ such that φ′ (x′ ) = φ(x) or φ′ (x) = φ(Λ−1 x). changing the sign of the spatial coordinates. for which we consider space inversion. This shows how parity transforms k-modes into −k modes. (2. 2.18) (2. (2.20) one has for the mode amplitudes aP (k) = a(−k) and bP (k) = b(−k).17) (2.e.14 and 2.6 We will explicitly discuss the example of a discrete symmetry.15 one has elimated k 0 and in both equations k · x = Ek t − k · x. (2π)3 2Ek (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) . Performing some Lorentz transformation x → x′ = Λx. It is easy to show that x φP (x) = φ(˜) x = = = or since one can define φP (x) ≡ d3 k e−i k·x aP (k) + ei k·x bP ∗ (k) . (2π)3 2Ek (2. i.3 Symmetries of the Klein-Gordon equation We arrived at the Klein-Gordon equation by constructing a covariant operator (∂µ ∂ µ + M 2 ) acting on a complex function φ. Since a · b = a · ˜ it is easy to see that ˜ b.16) The consequence of this is discussed in Exercise 2. They are referred to as mode and anti-mode amplitudes (or because of their origin positive and negative energy modes). φP (x) ≡ φ(˜) (P for parity). 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) .

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

e. express the quantities E(t) = P i (t) = d3 x (∂0 φ)∗ (∂0 φ) + ∇φ∗ · ∇φ + M 2 φ∗ φ . (c) Similarly. d3 x (∂ {0 φ)∗ (∂ i} φ). ∂ ψ(r.7 To solve the problem with positive and negative energies and get a positive definite density. {αi . and β 2 = I. B} = AB + BA. a{µ bν} ≡ aµ bν + aν bµ .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) Show that one has current conservation for the current 2 We denote the commutator as [A. We will encounter these quantities later as energy and momentum. an attempt to construct a first order differential equation for the time evolution would be to write.Relativistic wave equations 12 Exercise 2.15. Note that a{µ bν} indicates symmetrization. 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). in terms of the a(k) and b(k). From this one concludes that α and β must be matrix-valued and ψ must be a multi-component wave function. One then finds that Q is expressed as an infinite sum over number operators a† a (particles) and b† b (anti-particles). j 0 = ψ†ψ and j = ψ † αψ.2) in terms of the a(k) and b(k) using the expansion in Eq. 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 . i. t). Exercise 2. αj } = 2 δ ij I. i (a) Show that relativistic invariance. β} = 0. 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 . also using the result in (a). B] = AB − BA and the anticommutator as {A. 2. ∂t where ψ is a wave function and the nature of α and β is left open for the moment. Exercise 2. t) = (−i α · ∇ + m β) ψ(r. 1 2 (b) Express now the full charge QV (t) (exercise 2.5 (a) As an introduction to parts (b) and (c). We will see a∗ (k) → a† (k) and a(k) → aop (k) with aop and a† op op being annihilation and creation operators (as known from a harmonic oscillator) and the same for the coefficients b(k).1) for a complex scalar field current (exercise 2.

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

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

it is sufficient that the generators commute with H. The next issue is to find the appropriate symmetry operators in the Hilbert space. sometimes in more than one way (but this will be discussed later). 2 (3. For instance in a function space translations. Near the identity.e.11) Thus σx /2.e. i. one has a Lie algebra isomorphism that is linear and preserves the bilinear product. [g. immediately sees that the Lie algebras are identical. 3.10) σ ˆ · n. . .15) [g. n) π ≤ ϕ ≤ 2π. In the full parameter space. n! 2! n=0 ∞ (3.e. thus.12) One. n) −→ R(ϕ. . the above mapping corresponds to the trivial mapping of the Lie algebras.14) For a Lie group. in particular that AB → RAB = RA RB ). 2! . There exists a corresponding mapping of the groups given by µ: SU (2) −→ SO(3) ˆ ˆ A(ϕ. now with radius 2π and with all points at the surface identified (see figure). the SU (2) → SO(3) mapping is a 2 : 1 mapping where both A = ±1 are mapped into R = I. The infinitesimal generators of SU (2) are ˆ obtained by considering infinitesimal transformations. 2 2 2 (3. for fixed n. thus. with ˆ J ·n ≡ ˆ 1 ∂A(ϕ. however. (3. . The existence of symmetries means that there are operators g belonging to a symmetry group G that commute with the Hamiltonian. ˆ ˆ A(ϕ. . . SU (2) ≃ SO(3). also can be considered as a filled 3-sphere. They satisfy σi σj σk = i ǫijk . (3. φ′ (r) = φ(r + a) = φ(r) + a · ∇ φ + 1 (a · ∇)2 φ + . σy /2 and σz /2 form the basis of the Lie-algebra SU (2). H] = 0 for g ∈ G.2 Representations of symmetry groups The presence of symmetries simplifies the description of a physical system and is at the heart of physics. since any finite transformation can be constructed from the infinitesimal ones. n) i ∂ϕ = ϕ=0 (3.9) The parameter-space of SU (2).Groups and their representations 15 where we have used the (for operators new!) definition eA ≡ An 1 = 1 + A + A2 + . (3. H] = 0 for g ∈ G. n) ˆ −→ R(2π − ϕ. .13) 0≤ϕ≤π The relation A(σ · a)A−1 = σ · RA a (valid for any vector a) can be used to establish the homomorphism (Check that it satisfies the requirements of a homomorphism. Suppose we have a system described by a Hamiltonian H. i. i. n) ≈ 1 + i ϕ J · n.

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

The matrix elements of these unitary representations are known as D-functions. This works for SU (2). UE (ϕ. m′ |UE (ϕ. ǫy . . 1 ≡ ǫ1 ≡ − √ (ǫx + i ǫy ) . if χ → σχ. 0 ≡ ǫ0 ≡ ǫz . Given a representation. χ)|j. that is any closed curve in the parameter space can be contracted to a point. is not simply connected. Explicitly. the possibility thus exist that some extensions will not be well-defined representations. χ) = eim ϕ dm′ m (θ) e−imχ . however. For j = 1 another commonly used representation starts with three Cartesian basis states ǫ related to the previous basis via 1 |1.      If rotations leave H invariant. For SO(3).    (j) (3. SU (2) is the covering group of SO(3).e. χ) ≡ e−iϕJz e−iθJy e−iχJz . J =  0 0    +   0 0   0     0 0 0 0 0 √ 2 0    √   0         . however. The jump U → U ∗ implies for the generators σ/2 → −σ ∗ /2. The group SO(3). however. 2 |1. Therefore the representation and conjugate representation are equivalent in this case (see Excercise 3. the topological structure of the group is important. Because there exists a matrix ǫ such that −σ ∗ = ǫ σǫ−1 one immediately sees that appropriate linear combinations of conjugate states.19) (3.3). one can look at the conjugate representation. Of all groups of which the Lie algebras are homeomorphic. For those global representations. however. For the transformations A in SU (2) and R in SO(3) these are commonly parametrized with Euler angles. the extension from a local one must be unique. the conjugate representation is not a new one.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  . For an element in the group G the corresponding point in the parameter space can be reached in two ways. θ. e. j. If a transformation U acts on χ. or equivalently the Hilbert space can be written as a direct product space of spaces V (j) . For a decent (welldefined) global representation. to be precise the states ǫχ transform via −σ ∗ . one obtains the unitary representations Φ(g) = exp(i Φ(g)) of the full group G. θ. There exist two different types of paths. Consider the j = 1/2 representation of SU(2). This turns out to be the case for all half-integer representations of the Lie algebra. i. the conjugate transformation U ∗ acts on χ∗ . J− =  2    0        0 0 0 √ 3 0 0 0 0 √ 2 0 0 0 0       . dm′ m (θ) ≈ δm′ m − iθ(J 2 )m′ m .20) By moving through the parameter space the D-functions can be extended to global functions for all allowed angles. 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.g. (j) (j) ′ The spin matrices for that Cartesian basis {ǫx . If the group is simply connected. 1 |1. Infinitesimally (around the identity) the D-functions for SU (2) and SO(3) are the same. For SU(2). Jz =  i                0 i 0 −i 0 0 0 −i 0 0 0 0 0     . Jy =  0 0 0  . m = Dm′ m (ϕ. then ǫχ → ǫσχ = ǫσǫ−1 ǫχ = −σ∗ ǫχ. 2 From the (hermitean) representation Φ(g) of G. ǫz } are       0 0 0   0 0 i   0           Jx =  0 0 −i  . all states in the representation space have the same energy. contractable and paths that run from a point at the surface to its antipode. θ. −1 ≡ ǫ−1 ≡ √ (ǫx − i ǫy ) . the simply connected group is called the (universal) covering group.

(det Λ = +1 is called proper. 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 (3.26) (3. proof: det(ΛT GΛ) = det(G) → (det Λ)2 = 1.27) (3.3 The Lorentz group In the previous section spin has been introduced as a representation of the rotation group SU (2) without worrying much about the rest of the symmetries of the world. We considered the generators and looked for representations in finite dimensional spaces. Denoted in terms of column vectors and 4-dimensional matrices one writes for the poincar´ transe formations x′ = Λ x + a.25) (3. Thus one has the (pseudo-orthogonality) relation ΛT GΛ = G ⇔ GΛT G = Λ−1 ⇔ ΛGΛT = G.g. it is convenient to use a vector notation for the points in Minkowski space.     0 0   g00 g01 g02 g03   1 0      g     g11 g12 g13   0 −1 0 0  = . (ii) |Λ00 | ≥ 1.24) L↑ + L↓ + L↑ − L↓ − proper orthochrone proper non-orthochrone improper orthochrone improper non-orthochrone . e To derive some of the properties of the Lorentz group. consisting of the Lorentz group and translations. 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 .Groups and their representations 18 3.     G =  10 (3. From this property. explicitly x′µ = (Λ)µν xν + aµ or x′µ = Λµν xν + aµ . e. which as a matrix equation with (Λ)µν = Λµν and (G)µν = gµν gives (ΛT )ρµ (G)µν (Λ)νσ = (G)ρσ . In this section we consider the Poincar´ group. Writing x as a column vector and the metric tensor in matrix form. σ/2 in a two-dimensional (spin 1/2) case.23) The proper tensor structure of the matrix element (Λ)µν is a tensor with one upper and one lower index. it is easy to derive some properties of the matrices Λ: (i) det(Λ) = ±1. proof: (ΛT GΛ)00 = (G)00 = 1 → Λµ0 gµν Λν0 = 1 → (Λ00 )2 − i (Λi 0 )2 = 1. det Λ = −1 is called improper).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 (3.22) (Note that xT is a row vector). (3.

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. In L↑ . Using these explicit transformations. The first two sets of commutation relations exhibit the rotational behavior of J and K as vectors in E(3) under rotations. since I ∈ L↑ and + + det Λ and Λ0 change continuously along a path from the identity. 3 i 2 i=1 (Λ0 ) 3 19 (iii) Boosts along the z-direction are given by ΛB (η. J 2 . K j ] = i ǫijk K k . we have found the generators of rotations. z ) ≈ I + i ϕ J 3 . (3. Returning to the global group. Rotations around the z-axis are given by ΛR (ϕ. infinitesimally given by ΛB (η. it is easy to find the following typical examples from each of the four classes. 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  . K j ] = −i ǫijk J k . This is the origin of the Thomas precession. J j ] = i ǫijk J k . J 3 ).32)  0 −   0 −1 0      0 0 0 −1 . K 3 ). From the commutation relations one sees that the boosts (pure Lorentz transformations) do not form a group. 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).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↑ .g. infinitesimally given by ˆ ΛR (ϕ. (3.30)  0 +   0 +1 0      0 0 0 +1   0 0   −1 0    0 +1 0   0   ∈ L↓ (time inversion)   It =  (3. K 2 . since the generators K do not form a closed algebra. z ) = exp(−i ηK 3 ). βγ = sinh η. J = (J 1 . 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. i proof: use ΛT GΛ = G and ΛGΛT = G. one distinguishes rotations 0 + and boosts. z ) = exp(i ϕ J 3 ).28)                 3′    3    V 0 0 0 1 V 0 0 0 0 The parameter η runs from −∞ < η < ∞. z ) ≈ ˆ ˆ I − i η K 3 . The commutator of two boosts in different directions (e. and those of the boosts.   0 0   +1 0    0 +1 0   0   ∈ L↑ (identity)   I =  (3. [J i . K = (K 1 .Groups and their representations = i=1 (Λ0 )2 . Thus ˆ  0′       0 0 0  V 0   V   1  0 0 0 0   1′    1    V   0 cos ϕ sin ϕ 0   V   0 0 −i 0              2′  =     2    −→ J3 =   V   0 − sin ϕ cos ϕ 0   V   0 i 0 0  . [K i .

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. three of which only involve spatial coordinates (rotations) and three others involving time components (boosts). Thus the two vector operators J and K form under Lorentz transformations an antisymmetric tensor M µν . M ρσ ] = −i (g µρ M νσ − g νρ M µσ ) − i (g µσ M ρν − g νσ M ρµ ) . Also this group can be divided into four parts. 3 30 with η = ω and K = M (e. a).g. K = M ). inf (3. with ϕi = − 1 ǫijk ω jk and J i = 2 z-axis).37) The algebra of the generators of the Lorentz transformations can be obtained by an explicit calculation2 .39) (e. the Lorentz transformations form a (Lie) group (Λ1 Λ2 and Λ−1 are again Lorentz transformations. infinitesimally approximated by (I + ω.35) thus (ω)ij = −(ω)ji and (ω)0i = (ω)i0 . explicitly reading x′µ = = (Λ)µν xν + aµ = ((I)µν + (ω)µν ) xν + ǫµ Λµν xν + aµ = (g µν + ω µν ) xν + ǫµ . which are obtained by writing the infinitesimal parameters (ω)µν in terms of six antisymmetric matrices (M αβ )µν . (3.38) Explicitly. we continue using covariance arguments. We will just require Poincar´ invariance for the generators themselves. Of the four parts only L↑ forms a group.36) (3. ǫ). One can easily convince oneself that i (ω)µν = ω µν = − ωαβ (M αβ )µν . i i0 i i0 1 ijk M jk 2ǫ (3. (3. J 3 = M 12 ) and for the (infinitesimal) boosts (along (3.33) These four transformations form the Vierer group (group of Klein).34) The condition ΛT GΛ = G yields g ρµ + ω ρµ gρσ (g σν + ω σν ) = gµν =⇒ ωνµ + ωµν = 0. We now want to find a covariant form of the six generators of the Lorentz transformations. we have for the (infinitesimal) rotations (around z-axis) ΛR = I + i ϕ3 J 3 = I − i ω12 M 12 . The e 2 These commutation relations are the same as those for the ’quantummechanics’ operators i(xµ ∂ ν − xν ∂ µ ) . We therefore find (again) that there are six generators for the Lorentz group. In order to find the commutation relations including the translation generators. 3. The extension to the Poincar´ group is e ↑ straightforward. 2 (M αβ )µν = i g αµ g βν − g α g βµ ν (3.40) ΛB = I + i η 3 K 3 = I − i ω30 M 30 . Multiplying the proper orthochrone transformations with one of them gives all Lorentz transformations.g. Summarizing. This is a + normal subgroup and the factor group is the Vierer group.4 The generators of the Poincar´ group e ↑ The transformations belonging to P+ are denoted as (Λ. P+ etc. There are six generators. [M µν .

a) = Λ−1 µ P ν . P j ] = i δ ij H/c2 . (3. P j ] = [P i . as part of the set. . Of the generators of the Poincar´ group we choose first of all the generators P µ . J j ] = i ǫijk J k . that commute e among themselves. . [J i .47) (3.38 could have been obtained in the same way. They just state that M µν transforms as a tensor with two Lorentz indices. H] = i P i . We have here reinstated c. (1 + iǫγ P γ − i i ωαβ M αβ )P µ (1 − iǫδ P δ + ωρσ M ρσ ) = P µ + ω µν P ν . leads to U (Λ.46) [K i . Taking any one of the states in an irreducible representation. writing the generator P = (H/c. a)P µ U (Λ.41) 2 At this point.44) (3. the generators J± in the case of the rotation group. because one then sees that by letting c → ∞ the commutation relations of the Galilei group. .43) giving the following commutation relations by equating the coefficients of ǫµ and ωµν . For instance. the tensor M µν in terms of boosts cK i = M i0 and rotations J i = 1 ǫijk M jk . K j ] = −i ǫijk J k /c2 . [K i . [J i . (3. a)P µ U † (Λ. These form a group called the little group associated with that four vector. Explicitly. The generators of the translations are the (momentum) operators P µ . ǫ) = 1 − ωαβ M αβ + i ǫα P α .48) . H] = [J i . . a) = Λµν P ν or U † (Λ. 3. (3. P µ |p. (3. 3. 2 2 (3. one obtains 2 [P i . In order to label the states in an irreducible representation we construct a maximal set of commuting operators. To find other generators that commute with P µ we look for Lorentz transformations that leave the four vector pµ invariant. K j ] = i ǫijk K k . P ν ] [M µν .5 Representations of the Poincar´ group e where pµ is a set of four arbitrary real numbers. known from non-relativistic quantum mechanics are obtained. These define states with specified quantum numbers that are eigenvalues of these generators. . [K i . For instance. [J i . [P µ . Infinitesimally. P ) in terms of the Hamiltonian and the three-momentum operators. P ρ ] = 0. = pµ |p. other states in that representation are obtained by the action of operators outside the maximal commuting set. = −i (g µρ P ν − g νρ P µ ) . .42) (to decide on where the inverse is. H] = 0. check that U1 U2 corresponds with Λ1 Λ2 ). ν (3. the generators J 2 and J 3 in the case of the rotation group. The eigenvalues of these will be the four-momentum pµ of the state. P j ] = i ǫijk P k . the generators M αβ no longer exclusively act in Minkowski space. . Λµν pν ⇒ ωµν pν ⇒ ωµν = = = pµ + ω µν pν = pµ 0 ǫµνρσ pρ sσ .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.45) Note that the commutation relations among the generators M µν in Eq. The requirement that P µ transforms as a four-vector (for which we know the explicit behavior from the defining four-dimensional representation).

2 (3. the third case represents the vacuum. . These e operators are the Casimir operators of the algebra and can be used to define the representations of ↑ P+ .49) (3. while the others have no physical significance. .54) This will enable us to pick a suitable commuting ’spin’ operator. . 2 exp (−isµ W µ ) |p. i exp − ωµν M µν 2 i exp − ǫµνρσ M µν pρ sσ 2 (3. Given the momentum four vector pµ we choose a tetrade consisting of three orthogonal spacelike unit vectors ni (p). Wα ] = [Wµ . of which the components generate the little group of pµ or (for the third of the following relations) from explicit calculation [Mµν . . p0 > 0. Wν ] = −i (gµα Wν − gνα Wµ ) . 0. . . Pν ] = [Wµ .50) where the Pauli-Lubanski operators W µ are given by 1 Wµ = − ǫµνρσ M νρ P σ . i ǫµνρσ W ρ P σ . From the algebra of the generators one finds that P 2 = Pµ P µ and W 2 = Wµ W µ (3. Massive particles: p2 = M 2 > 0. 0.52) (3. = = i exp − ǫµνρσ M µν pρ sσ |p.56) . M ǫijk . (3.Groups and their representations 22 where s is an arbitrary vector of which the length and the component along pσ are irrelevant. satisfying gµν nµ (p)nν (p) = j i nµ (p) pµ i ǫµνρσ pµ nν nρ nσ i j k = = −δij . . (3.53) (3. .51) The following properties follow from the fact that W µ is a four-vector by construction. The elements of the little group thus are U (Λ(p)) = = and U (Λ(p))|p. .55) commute with all generators and therefore are invariants under Poincar´ transformations. 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. which can thus be chosen spacelike.

59) (3. (3. We will not do this but construct them as solutions of a wave equation to be discussed explicitly in chapter 4 Massless particles: p2 = M 2 = 0. The explicit construction of the wave function can be done using the D-functions (analogous as the rotation functions). 0). (0. 3. n2 (p) and s(p) are obtained by a Lorentz boost from the reference frame one has the property ǫµνρσ sµ nν nρ pσ > 0. In this case a set of four independent vectors is chosen starting with a reference frame in which pµ = pµ 0 in the following way: p0 = (p0 .63) where E = p2 + M 2 and ms is the z-component of the spin −s ≤ ms ≤ +s (in steps of one). . (3. show that W /M can be interpreted as the intrinsic spin. 1.57) (i. 0. M J) with components W i = (M/2)ǫijk M jk . .e.64) such that in an arbitrary frame where the four vectors p.65) .e. ms .60) and using the completeness relation nµ (p)nν (p) = − g µν − i i 2 i pµ pν M2 . 0. p. 0. S j (p)] = −i nµ (p)nν (p) j i ǫµνρσ W ρ P σ = i ǫijk S k (p). In that case the vectors ni (p) are just the space directions and W = (0. 0). the S i (p) form the generators of an SU (2) subgroup that belongs to the maximal set of commuting operators.62) 1 Thus W 2 has the eigenvalues −M 2 s(s + 1) with s = 0. Noting that S i (p) i 2 = 1 W µ (p)W ν (p) M2 niµ (p)niν (p). s. it is sufficient to choose a particular frame to investigate the coefficients in Eq. 0. The commutation relations [W i (p). . W j (p)] = i M ǫijk W k (p). Together with the four momentum states thus can be labeled as |M. 2 . (1. M2 (3. p0 > 0. W i (p) = −W · ni (p)). Having made a covariant decomposition.58) i.61) one sees that S i (p) i =− W2 . 0.57 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. i (3. M2 (3. 1. 1 2 (3.. 1. n1 (p). 0). More generally (fully covariantly) one can proceed by defining S i (p) ≡ W i (p)/M and obtain [S i (p). |p3 |).Groups and their representations 23 We can write W µ (p) = 3 W i (p) nµ (p) i i=1 (3. (0. −1) n1 (p0 ) ≡ n2 (p0 ) ≡ s(p0 ) ≡ (3. 0.

The vector W µ (p) now can be expanded W µ (p) = W 1 (p) nµ (p) + W 2 (p) nµ (p) + λ(p) pµ + W s (p) sµ (p). W 2 (3.67) (3. This state is in physical applications nondegenerate and is denoted by |0 . i W 2 (p).13. The vacuum: pµ = 0.71) which is called the helicity. We don’t know of any physical relevance.66) where W · p = 0 implies that W s (p) = −(M 2 /p · s) λ(p).69) We reproduce the algebra for the massive case with W 3 (p) = M λ(p). 3. W 2 (p)] [λ(p). |p.68) (3. (3.70) as a commuting set with zero eigenvalues for P 2 and W 2 . σ · RA a ≡ A(σ · a)A−1 ˆ (valid for any vector a). 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 . −i W 1 (p). 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. Thus we have P 2 . which can take any integer or half-integer value.72) One has P µ |0 = W µ |0 = 0 and the state is invariant under Lorentz transformations. and the eigenvalues of W 1 and W 2 are set to zero (corresponding to the limit M → 0 for the W µ Wµ eigenvalues of M 2 s(s + 1) → 0). Thus 2 λ(p) = J ·p . W µ (p) = λ(p) P µ . Note that angular momentum does not contribute to helicity as L · p = 0.1 (a) Derive from the relation that defines the mapping of SU (2) matrices A into SO(3) matrices RA in Eq. 1 2 24 (3.Groups and their representations The above does actually include the massive case. But if M = 0 one has a different algebra (Eq. Wν ]. U (Λ. a)|0 = |0 . however. thus. W 1 (p)] [λ(p). implying continuous values for the spin (actually for W i = M S i ). where A = exp(i ϕ · σ/2) = exp(iϕ σ · n/2). and gives [W 1 (p). is characterized by its momentum and the helicity. P µ .67 has a vanishing right hand side). λ . (3.73) (3. The algebra of W 1 . W 2 (p)] = = = i M 2 λ(p). Exercises Exercise 3. 3. This algebra is (for instance) isomorphic to the one generated by rotations and translations in a 2-dimensional Euclidean plane (see exercises). A massless particle. The eigenvalues of W 1 and W 2 (in that case corresponding to the translations) thus can take any continuous values. |p| (3.

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

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

This tells us how to find the representations of the group. B j ] = [Ai . K = −i . 2 2 σ σ J = .1) (4. U (ϕ) H(η) (4. Aj ] = [B i . 0) Type II : (0. The group SL(2. With this choice of parameter-spaces the plus and minus signs are actually relevant. (4. similarly as the homeomorphism between SO(3) and SU (2). becoming the defining representations of SL(2. j ′ ). Special cases are the following representations: Type I : (j. K = +i .Chapter 4 The Dirac equation 4.1 The Lorentz group and SL(2. (4. j) K = −i J (B = 0). C): Type I (denoted M ): Type II (denoted M ): 27 J= σ σ .8) ˆ ˆ ˆ with η = η n and ϕ = ϕn. C). M = exp i ϕ·σ 2 1 exp ± η · σ 2 = U (ϕ) H(η) . B j ] = i ǫijk Ak . where we restrict (for fixed n) the parameters 0 ≤ ϕ ≤ 2π and 0 ≤ η < ∞.2) This shows that the Lie algebra of the Lorentz group is identical to that of SU (2) ⊗ SU (2). 2 2 (4. 0. The matrices in SL(2. It is simply connected and forms the covering group of L↑ .9) (4. (j.10) . C) Instead of the generators J and K of the homogeneous Lorentz transformations we can use the (hermitean) combinations A = B which satisfy the commutation relations [Ai . 2 (4. 2 1 (J − i K). it also follows directly that the Lorentz group is homeomorphic with the group SL(2. They will be labeled by two angular momenta corresponding to the A and B groups. C) can be written as a product of a unitary + matrix U and a hermitean matrix H.6) (4.3) (4. respectively.4) (4. C) is the group of complex 2 × 2 matrices with determinant one. K = i J (A = 0).5) = 1 (J + i K). i ǫijk B k . They precisely correspond to the two types of representations that we have seen before.7) From the considerations above.

˜ Non-equivalence of M and M One might wonder if type I and type II representations are not equivalent. ǫ−1 = ǫ† = ǫT = −ǫ). Note that M ∗ = ǫ† M ǫ has its equivalent for the Lorentz transformations Λ(M ) and Λ(M ∗ ). Also useful is the relation H 2 (p) = σ µ pµ /M = (E + σ · p)/M . (4.14) satisfying Tr(σ µ σ ν ) = −2 gµν and Tr(σ µ σ ν ) = 2gµν = 2 δ µν (the matrices. cannot be connected by a Lorentz transformation belonging to L↑ . (4. −σ).19) .13) 2M (E + M ) (exercise 4. (4. The Lorentz transformations Λ(M ) and Λ(M ). while ±ǫη ∗ transforms as a type-I + spinor. One defines spinors ξ and η transforming similarly under unitary rotations (U † = U −1 . This shows that M ∗ ≃ M . i. U σ µ aµ U ˜ −1 H σ µ aµ H −1 = σ µ ΛB aµ H σ µ aµ H ˜ −1 (4. such that M = SM S −1 . ξ −→ Hξ. U σ µ aµ U −1 = σ µ ΛR aµ . In fact. 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). σ µ ≡ (1. thus. The following relations for + rotations and boosts are useful. H ≡ (H † )−1 = H −1 ). This is not true for the two-component spinors in SL(2.12) ξ → U ξ. ˜ = σ µ Λ B aµ . One has + −1 Λ(M ∗ ) = I2 Λ(M ) I2 .11) Considering ξ and η as spin states in the rest-frame.15) (4. H(η) = exp(−η · σ/2). there exists the matrix ǫ = iσ 2 . η → Hη. ˜ The sets of four operators defined by definitions σ µ ≡ (1. C) and. thus.1). 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. C) or L↑ . Λ(ǫ) ∈ L+ . This boost then is given by H(p) = exp η·σ 2 = cosh η η ˆ + σ · n sinh 2 2 = M +E+σ·p (4. C). ˜ (4. one can use a boost to the frame with momentum ˆ ˆ p. σ).The Dirac equation 28 Let us investigate the defining (two-dimensional) representations of SL(2.16) = σ µ Λ R aµ .18) As already discussed for SU (2). if there does not exist a unitary matrix S. namely H(η) = exp(η · σ/2). With E = M γ = M cosh(η) and p = M βγ n = M sinh(η) n. C). Eqs 4. (4. U ≡ (U † )−1 = U ) U (ϕ) = exp(i ϕ · σ/2) but differently under hermitean boosts (H † = H. (4. η → U η.e. are not covariant!) ˜ can be used to establish the homeomorphism between L↑ and SL(2.17) that can be used to relate σ µ∗ = ǫ† σ µ ǫ = ǫ−1 σ µ ǫ and hence ˜ ¯ U ∗ = ǫ† U ǫ = ǫ† U ǫ. however. ¯ ¯ H ∗ = ǫ† Hǫ = ǫ−1 Hǫ. (ǫ∗ = ǫ.18 show that ±ǫξ ∗ transforms as a type-II (η) spinor.

therefore.       = (4. ǫµνρσ will change sign. the well-known two-component spinor for a spin 1/2 particle.      (4. The behavior is the same for classical quantities. a) and ˜ a = (a0 . where a = (a0 .1. they can be connected.2 Spin 1/2 representations of the Lorentz group 1 1 Both representations (0. (axial vector). 0) (via operators J and K) one sees that under parity 2 2 1 1 (0. 4. 1 ) and ( 1 . 2 ) and ( 2 . generators.e. ) −→ ( . This provides the easiest way of seeing ˜ what is happening. 1 ) and ( 1 . The representing member of the class P− is the parity or space inversion operator Is . i. −→ −p (vector).e. 0) of SL(2.13. For a particle at rest only angular momentum is important and we can choose ξ(0. (scalar). must combine the repe resentations. the boost in Eq. The i angular momentum operators J are represented by σ i /2. It has the same effect as lowering the indices.21) In nature parity turns (often) out to be a good quantum number for elementary particle states. From the definition of the representations (0. + −   1 0    0 1  I2 =    0 0   0 0 0 0 −1 0 0 0 0 1      .22) u=     η η M (Λ) 0 where M (Λ) = ǫM ∗ (Λ)ǫ−1 with ǫ given in Eq. Examples (with between brackets the Euclidean assigments) are r t p H J λ(p) K −→ −λ(p) −→ −K −→ H −→ J −→ t (scalar). 4. (vector). but rather the operators W i (p) should be used. Within the Lorentz group. Taking M (Λ) = H(p).      0   χm   uR   H(p)      . C) are suitable for representing spin 1/2 particles. the aij elements of a tensor will not change sign. i. 0). Although these operators cannot be used to label the representations. as we have seen in section 4. 2 2 (4.g. consider the four component spinor that transforms under a Lorentz transformation as      0  ξ   ξ  . e. We also have seen that the representations (0. −a). A parity transformation changes aµ into aµ . but to L↑ .23) u(p. they cannot be connected by a unitary 2 2 transformation. and the angular momentum in the rest frame is what we are familiar with as the spin of a particle. we obtain for the two components of u which we will refer to as chiral right and chiral left components. (pseudoscalar). Thus M ∗ and M are not equivalent within SL(2. etc.20) 4.17. m) = χm .The Dirac equation 29 where The matrix I2 does not belong to L↑ . that have the correct commutation relations. but by a transformation belonging ↑ ↑ to the class P− . −→ −r (vector). For the spin 1/2 representations of the Poincar´ group including parity we. 0) are inequivalent. m) = η(0. m) =            χm uL H(p) 0 . we have seen that in the rest frame W i (p)/M → J i . C).   −→  M (Λ)   (4.

As in section 2 we can use Pµ = i∂µ as a representation for the momenta (translation operators) in function space. We will discuss another explicit realization of this algebra in the next section. γ ν } = 2 g µν .7. Applying (iγ µ ∂µ + M ) from the left gives (4.29) which is an explicit realization of the (momentum space) Dirac equation.The Dirac equation 30 with H(p) = ¯ H(p) = 2M (E + M ) E+M −σ·p 2M (E + M ) E+M +σ·p .31) To achieve also that the energy-momentum relation p2 = M 2 is satisfied for u(p).28) which is a covariant (linear) first order differential equation. ψ † −iγ 0 ∂0 +iγ i ∂i −M 1 We ← ← = 0. this can be rewritten 1 µ ν {γ . (4. The explicit realization appearing in Eq. (4. From this one obtains the Clifford algebra for the Dirac matrices. which in general is a linear equation in pµ . 4.24) (4. p (4. 2 (4. γ } ∂µ ∂ν + M 2 ψ(x) = 0. . = H −1 (p).30) γ µ γ ν ∂µ ∂ν + M 2 ψ(x) = 0.27) where γ µ are 4 × 4 matrices called the Dirac matrices1 . (4. To see this consider the equation for the hermitean † † conjugate spinor ψ † (noting that γ0 = γ0 and γi = −γi ). one must require that for ψ(x) the Klein-Gordon relation 2 + M 2 ψ(x) = 0 is valid for each component separately). We first want to investigate if the Dirac equation solves the problem with the negative densities and negative energies.      0 H 2 (p)   uR   uR        .27 is known as the Weyl representation. It is of a form that we also played with in Exercise 2.  =       (4. Since ∂µ ∂ν is symmetric. {γ µ . This leads to the Dirac equation for ψ(x) = u(p)e−i p·x in coordinate space.26)        −2    uL uL H (p) 0 or explicitly in the socalled Weyl representation  −M E+σ·p       E−σ·p −M    uR      uL      = 0.33) define for a four vector a the contraction a = aµ γµ / . The general requirements for the γ matrices are thus easily obtained. (iγ µ ∂µ − M ) ψ(x) = 0.32) suppressing on the RHS the identity matrix in Dirac space.25) It is straightforward to eliminate χm and obtain the following constraint on the components of u. (4. The first indeed is solved.   (4. (pµ γ µ − M ) u(p) ≡ (/ − M ) u(p) = 0.

but with properties such that it can be annihilated by the particle (e. Relying on the Pauli exclusion principle for spin 1/2 particles.1: The Dirac sea of negative energy states and antiparticles. The density j 0 = ψγ 0 ψ = ψ † ψ (4.g. From the equations for ψ and ψ one immediately sees that j µ = ψγ µ ψ is a conserved probability current (exercise!).34) for the adjoint spinor ψ ≡ ψ † γ0 . opposite charge). γ =ρ ⊗1= γ = iρ ⊗ σ =    0 −1 −σ k 0 2 We (4.The Dirac equation 31 E empty levels M -M filled levels E particle M -M hole Figure 4.38) The general algebra for the Dirac matrices is Two often used explicit representations are the following2 : The standard representation:     σk   1 0   0 0 3 k 2 k  . one sees that there are two positive and two negative eigenvalues. The second problem encountered before. 4. (4. the one of the negative energy states. remains. This hole forms an antiparticle with the same mass. γ ν } = 2 g µν .39) use ρi ⊗ σj with both ρ and σ being the standard 2 × 2 Pauli matrices. From the vacuum a particle can be removed. E = +M (twice) and E = −M (twice). where γ 0 p0 ψ = M ψ 0 −→ Eψ = M γ 0 ψ.27). (4. We will see how this is properly implemented (also for bosons!) when quantizing fields. This can most easily be seen in the particle rest frame. 4.36) (4. . This problem was solved by Dirac through the introduction of a negative energy sea.3 General representations of γ matrices and Dirac spinors {γ µ . (4.37) Using the explicit form of γ in the Weyl representation (see Eq. 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.  . Multiplying with γ 0 on the right and pulling it through one finds a covariant equation ψ(x) iγ µ ∂µ +M ← = 0. The Dirac sea forms the vacuum.35) is indeed positive and j 0 can serve as a probability density.

41) (4.49) i 1 with −i S ρσ = 2 σ ρσ = − 4 [γ ρ . K 3 = M 30 =    0 iσ 3 2 4 or in general one has the transformation i ψ → Lψ = exp − ωρσ S ρσ ψ.R. For example. γ 2 ].40) (4.44) and an adjoint spinor defined by ψ = ψ † γ0 . the rotation and boost generators in Weyl representation in Eqs 4.45) (γ5 )W. which in the case of the Weyl representation are just the upper and lower components.  0 (4. γ µ } = 0 and explicitly one has    0 1  1 .The Dirac equation 32 The Weyl (or chiral) representation:    0 1   .R.R. Lorentz invariance The Lorentz transformations can also be written in terms of Dirac matrices. γ 0 = ρ1 ⊗ 1 =   1 0 γµ ψ Different representations can be related to each other by unitary transformations. γ σ ]. it is easy to see that PR/L = 1 (1 ± γ5 ) 2    1 0  . = S(γµ )W. 4! (4. 2 (4. ν The latter assures Lorentz invariance of the Dirac equation (see Exercise 4. 0 σ3 2 4   1  −iσ 3 0   = − i [γ 3 .42) We note that the explicit matrix taking us from the Weyl representation to the standard representation. that project out chiral right/left states.48) are projection operators. Another matrix which is often used is γ5 defined as γ5 = i γ 0 γ 1 γ 2 γ 3 = −i γ0 γ1 γ2 γ3 = It satisfies {γ5 .10 are   1  σ3 0  i   J 3 = M 12 =   = [γ 1 . is   1  1 1    (4.46) (4. −→ Sγµ S −1 . For instance in the Weyl representation (but valid generally).R.  (γ5 )S.9 and 4. (γµ )S.47) (4.3) .  =ρ ⊗1=  0 −1 3 (4. (4. S −1 . 1 −1 2 For all representations one has † γµ = γ0 γµ γ0 . while L−1 γ µ L = Λµ γ ν . = ρ ⊗ 1 =   1 0 i ǫµνρσ µ ν ρ σ γ γ γ γ . We note that ψ → Lψ and ψ → ψ L−1 .   0 γ k = −iρ2 ⊗ σ k =  k  σ  −σ k  . −→ Sψ.43) S=√  . γ 0 ].

is a left-handed spinor.g. =    0 ǫ iσ 2 (4. ψ= (4.) and all representations connected via a real (up to an overall phase) matrix S.55) (4. ψR . ∂ with the solution ψ c (x) = ηc Cψ (x). As with parity we look for a transformation. Some properties of C are C −1 = C † and C T = −C. C is real and one has C −1 = C † = C T = −C c and [C. The latter implies that the conjugate of a right-handed spinor. −x) and via a few manipulations obtain again the ˜ Dirac equation (i/ − M ) ψ p (x) = 0. that brings ψ → ψ c . (or W. reverse the charge of the particle.51) The existence of positive and negative energy solutions implies another symmetry in the Dirac equation.53) (4.R.R. x = (t. One has ψ c = −ψ T C −1 .R.    0 −ǫ 0    0   −ǫ 0  . called charge conjugation.   ψ= (4. such that T Cγµ C −1 = −γµ . In S. ∂ (4. e.58) where ηc is an arbitrary (unobservable) phase. Note that we have (as expected) explicitly in Weyl x representation in Dirac space     P  ξ    −→ ψ p = γ 0 ψ =  η  . for instance. (C)W. but it does. T (4.6).50) Charge conjugation but with ψ p (x) ≡ ηp γ0 ψ(˜) (Exercise 4. 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. In any representation a matrix C exist. parity.57) (4.g.    −iσ 2   0 −ǫ  = .54) e. = =   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. (C)S. T (4.The Dirac equation 33 Parity There are a number of symmetries in the Dirac equation.R. in Weyl representation we find in Dirac space     T C  ξ   −ǫ η ∗     . γ5 ] = 0.56) (i/ − M ) ψ c (x) = 0. This symmetry does not change the spin 1/2 nature. It is easy to convince oneself that if ψ(x) is a solution of the Dirac equation. (i/ − M ) ψ(x) = 0. Explicitly.52)    η ξ one can apply space inversion. usually to be taken unity. ∂ (4. x) → x = (t. which is again a solution of the Dirac equation.59)  −→ ψ c = C ψ =    η ǫ ξ∗ .

62) where χs are two independent (s = ±) two-component spinors. s) + v(k. s)u(p. s)u(p. s′ ) = −2M δss′ . 2M 2M −/ + M p v(p. Note that the spinors in the negative energy modes (antiparticles) could be two different spinors. After separating positive and negative energy solutions as done in the case of the scalar field one has ψ(x) = s d3 k u(k. In an arbitrary frame one has s · p = 0 and s(p) can e.   ⇔ (/ − M ) u(p) = 0. with E = Ep = + p2 + M 2 .63) u(p.64) (4. In that frame is a spacelike unit vector sµ = (0.70) =  Ps =  ˆ 0 1∓σ·s 2 2 . s) e−i k·x b(k. s).     ∂ −iσ · ∇ −i ∂t − M (4. v (p. s) v = . s) and C v T (p.66) An arbitrary spin 1/2 field can be expanded in the independent solutions. p     −σ · p (Ep − M )     Ep + M     σ ·p ¯ Ep +M χs χs ¯ For a free massive particle.67) It is straightforward to find projection operators for the positive and negative energy states P+ P− = s p /+M u(p. ¯ (4.1). where u satisfies   −σ · p  Ep − M      u(p) = 0. s) ei k·x d∗ (k. in which γ 0 is diagonal (see discussion of negative energy states in section 4.65) (4. s)v(p. s ) = v (p. s) = u(p. the spin polarization vector in the rest frame is the starting ˆ point. s)v(p.4 Plane wave solutions Looking for positive energy solutions ∝ exp(−iEt) one finds two solutions. ¯ ′ ′ Explicit solutions in the standard representation are   χs       . p (4. s) u = .68) (4. s) . where v satisfies   σ·p  −(Ep + M )      v(p) = 0. be obtained by a Lorentz transformation.61)     σ·p −(Ep + M ) There are also two negative energy solutions. s ) = v † (p.The Dirac equation 34 4. ψ(x) = v ±s (p) ei p·x . s)u(p. s) = Ep + M      σ ·p   Ep +M χs     .60)   ψ(x) = 0. s)¯(p.. It is easy to check that   1  1±σ·s 1 ± γ5 / s ˆ 0    (4. The ¯ ¯ solutions are normalized to u(p. (2π)3 2Ek (4. the spinors satisfy C uT (p. ¯ ¯ † ′ u (p.g. 2M 2M (4. ψ(x) = u±s (p) e−i p·x . Choosing χ = −ǫχ∗ (the equivalent spin ¯ 1/2 conjugate representation). s)¯(p. The explicit Dirac equation in the standard representation reads  ∂  iσ · ∇   i ∂t − M       (4. the best representation to describe particles at rest is the standard representation. s ) = 0.   ⇔ (/ + M ) v(p) = 0. s). s′ ) = 2Ep δss′ .69) = − s In order to project out spin states. s)v(p. s) = u(p. s′ ) = 2M δss′ . s) = v(p. v(p.    (4.

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

γν ] (9) σµν γ5 = γ5 σµν = ǫµνρσ γ ρ γ σ /2. (5) γ µ γ λ γ ν γ ρ γ σ γµ = 2 γ σ γ λ γ ν γ ρ + γ ρ γ ν γ λ γ σ . s)γ µ u(k ′ .76) ul (k ′ . s′ )¯i (k ′ . (3) Tr(γ µ γ ν ) = 4 g µν or Tr(//) = 4 a · b.s′ 1 ′ (/ + m)li (γ µ )ij (/ + m)jk (γ ν )kl k k 2 1 Tr [(/ ′ + m)γ µ (/ + m)γ ν ] . It is. (7) γµ γ5 = -γ5 γµ = i ǫµνρσ γ ν γ ρ γ σ /3!.g µρ g νσ + g µσ g νρ ). s′ )) . s)γ ν u(k ′ . more convenient ¯ to use the projection operators introduced earlier. For this we first have to prove (do this) that (¯(k. k k 2 . however. Some elementary properties are: Properties of products of γ-matrices: (1) γ µ γµ = 4 (2) γ µ γ ν γµ = −2 γ ν or γ µ /γµ = −2 a. s)γ µ u(k ′ . (5) Tr(γ µ1 · · · γ µn ) = g µ1 µ2 Tr(γ µ3 · · · γ µn ) . s)¯k (k. s′ )) (¯(k. (9) Tr(γ µ1 · · · γ µ2n ) = Tr(γ µ2n · · · γ µ1 ).75) Lµν (k. Properties of traces of γ-matrices (1) Tr(γ µ1 · · · γ µn ) = 0 if n is odd. s)γ µ u(k ′ . s)(γ ν )kl u u s. s′ )(γ µ )ij uj (k. namely the calculation of the quantity 1 ∗ (¯(k. (Use matrices C to transpose the matrices in the expression under the trace and use Tr AT = Tr A. s′ )(γ µ )ij uj (k. s)¯k (k. we will give one example.5 γ gymnastics and applications An overview of properties can be found in many text books.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 ). u ¯ This leads with explicit Dirac indices to Lµν = = = = 1 2 1 2 ui (k ′ .s′ ∗ (4. s). ab (4) Tr(γ µ γ ν γ ρ γ σ ) = 4 S µνρσ . s′ )γ µ u(k. s′ ). (6) γ5 = i γ 0 γ 1 γ 2 γ 3 = −i ǫµνρσ γ µ γ ν γ ρ γ σ /4!. k ′ ) = 2 ′ s.g µ1 µ3 Tr(γ µ2 γ µ4 · · · γ µn ) + · · · . ab (4) γ µ γ ν γ ρ γ σ γµ = −2 γ σ γ ρ γ ν . (Use (γ5 )2 = 1 and pull one γ5 through. (7) Tr(γ5 γ µ γ ν ) = 0 or Tr(γ5 //) = 0. s′ )) = u(k ′ . In principle one can take a representation and just calculate the quantity u(k. u u (4. a / (3) γ µ γ ρ γ σ γµ = 4 g ρσ or γ µ //γµ = 4 a · b.The Dirac equation 36 4. etc.) (2) Tr(1) = 4. s)(γ ν )kl ul (k ′ .) In order to show the use of the above relations. s′ ) ¯ u s. (8) σµν ≡ (i/2)[γµ . (6) Tr(γ5 ) = 0. (10) γ µ γ ν = g µν − i σ µν (11) γ µ γ ν γ ρ = S µνρσ γσ + i ǫµνρσ γσ γ5 with S µνρσ = (g µν g ρσ . ab (8) Tr(γ5 γ µ γ ν γ ρ γ σ ) = −4i ǫµνρσ .

1 Prove Eq. a 4 a · b. by letting them act on the four basis spinors u(p. i S ρσ ] = g ρµ γ σ − g σµ γ ρ . 4.4 Show that in P+ ≡ s=±1/2 u(p. where x = (t. s) and v(p. 2M (E + M ) ˆ ˆ where E = M γ = M cosh(η) and p = M βγ n = M sinh(η) n.3 Imposing Lorentz invariance on the Dirac equation can be achieved by requiring L−1 γ µ L = Λµ γ ν . ν µ µ Look at the infinitesimal forms Λµ = gν + ων and L = 1 − (i/2)ωρσ S ρσ to obtain ν [γ µ . 4 2 Exercise 4.6 Apply space-inversion.2 Show that j µ = ψγ µ ψ is a conserved current if ψ en ψ are solutions of the Dirac equation Exercise 4. e.5 Show starting from the relation {γµ .77) Exercises Excercise 4. by proving that H 2 (p) = exp (η · σ) = (E + σ · p)/M = σ µ pµ /M. 0 if n is odd.g. 1 i −i S ρσ = − [γ ρ . −x) is also a solution of the Dirac equation.The Dirac equation 37 The trace is linear and can be split up in parts containing traces with up to four gamma-matrices. Show that the equality holds. H(p) = exp η·σ 2 = M +E+σ·p . s) u p /+M = 2M 2M both sides are projection operators. Exercise 4. γ σ ] = σ ρσ . of which only the traces of four and two gamma-matrices are nonzero. 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 . ˜ Exercise 4. s). Do this e. (4. to the Dirac equation and use this to show that the spinor ψ p (x) = ˜ γ0 ψ(˜). Excercise 4. x → x. γν } = 2gµν that γµ / γ µ a = Tr(//) = ab µn µ1 Tr(γ · · · γ ) = −2 /.g. x ˜ .13.

n (c) Apply the confinement condition to find the value of k for the lowest eigenmode in the spherical ˆ cavity. A condition for confining the fermion in the cavity is i / ψ = ψ (where n is the surface normal).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 ).    is a stationary solution of the (free) Dirac equation.57 MeV. The conditions for ψ and ψ imply that nµ j µ = ψ/ ψ = 0.66 MeV. you need to determine the corresponding confinement condition for ψ.7 (a) Show that (in standard representation)  j0 (kr)χm    ψ(x. i. the lepton and neutrino have spin 1/2 and are ν described by momentum-space spinors. r). mµ = 105. Do you understand why? The calculation of the life-time is completed in chapter 11. (b) Next. For such a cavity nµ = (0. Note: the outcome shows that the decay into muons is much more likely that the decay into electrons (mπ = 139. Our starting point is the transition amplitude given by M = ℓ(kℓ )¯ℓ (kν )|Hint |π(p) = GF fπ u(kℓ )/(1 − γ5 ) v(kν ). t) = N     k ˆ i E+M σ · r j1 (kr)χm    −i Et   e . .The Dirac equation 38 Exercise 4. Show that this n condition is sufficient to guarantee that nµ j µ = 0. Excercise 4. Calculate the value of k for M R = 0. 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. Note: To prove this. which is the most important factor needed to find the decay width (inverse lifetime). (d) Plot for these cases the vector densities ψ † ψ and the scalar densities ψψ as a function of r/R. The pion has spin 0. In this solution the wave number k = |k| √ = E 2 − M 2 and χm a two-component spinor while j0 and j1 are spherical Bessel functions. namely i / ψ = eiαγ5 ψ. we want to confine this solution to a spherical cavity with radius R.8 × 10−6 . ν p Note: how such parametrizations work will be explained in chapter 8. there is no current flowing through the surface of the cavity. me = 0.5 and M → ∞.511 MeV). 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. Calculate for now just |M |2 . M R = 1.e. the decay would even be forbidden. Show that the ratio is |M (π − → e− νe )|2 ¯ ≈ 54. |M (π − → µ− νµ )|2 ¯ Actually if the lepton would be massless.

These can describe spin 1 states in the rest-frame.5)  −E    B2   . These are solutions of the Klein-Gordon equation with in addition a condition. In an arbitrary frame ǫµ (k) with λ = 0. ±(1) are obtained by boosting the restframe vectors.± 5. (2 + M 2 )Vµ = 0.1) The Lorentz condition implies that k µ ǫµ (k) = 0. ǫλ ).6) where Aµ is the four-vector potential (5. λ) . just as the two-component Dirac spinors did for spin 1/2.1 Fields for spin 1 In section 3. µ (2π)3 2E µ (5. M2 (5. with ∂µ V µ = 0 (Lorentz condition). where k = k0 = (M. there are then indeed three independent possibilities.2 we have discussed the vectors ǫλ (with λ = 0 and ±1) as basis states of the spin 1 representation. The covariant generalization is to describe them with Vµ (x) = ǫµ (k) e±ik·x . In the restframe.  −B 1    0 3 With the electromagnetic field tensor (5.3) Since the vectors ǫµ (k) together with the momentum k µ form a complete set of four-vectors. one has ǫ(λ) (k) ǫ(λ)∗ (k) = −gµν + µ ν kµ kν .± (λ) d3 k ǫ(λ) (k) e−ik·x c(k. (5. 39 (5. 0). A real vector field Vµ (x) then can be expanded in modes as Vµ (x) = λ=0.Chapter 5 Vector fields and Maxwell equations 5.4) λ=0.7) .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. λ) + ǫ(λ)∗ (k) eik·x c∗ (k. A). namely ǫµ (k0 ) = (0. The vectorfield is therefore (λ) suited to describe spin 1.2) (5.

15 the d’Alembertian is replaced by 2 + M 2 . −i. the antisymmetry of F µν implies that ˜ ∂µ F µν = 0 or ∂λ Fµν + ∂µ Fνλ + ∂ν Fλµ = 0.15) (5. 2Aµ = 0. 0. Although the Lorenz condition is a constraint. . if ∂µ Aµ = 0 for large times |t|. Furthermore.11) This equation is not affected by a gauge transformation. The gauge freedom can first of all be used to impose the Lorentz condition for the electromagnetic field. Note that the incorporation of the Lorentz condition can also be done for the case of the massive vector field. in accordance with what we found in section 3. for M = 0. transverse or Coulomb gauge). (5. |k|). there is still a gauge freedom left (Aµ → Aµ + ∂µ χ with 2χ = 0). in which case one has 2Aµ = jµ . (5. Under a gauge transformation Aµ −→ Aµ + ∂µ χ.Vector fields and Maxwell equations and the (conserved) current j µ = (ρ. In that case one also has ∇·A = 0 or ki ǫi (k) = 0.14) of which the solutions give the Li´nard-Wiechert potentials. 0). only two independent vectors remain ǫ(±) µ (k) = (0.18) corresponding to the helicity states λ = ±1 of the massless photon field. ∂µ Aµ = 0. (5.13) Hence. Therefore if k µ = (|k|.10) the electric and magnetic fields are unchanged. if in Eq.9) 1 ˜ where F µν ≡ − 2 ǫµνρσ Fρσ is the tensor where E → B and B → −E. Fµν −→ Fµν + (∂µ ∂ν χ − ∂ν ∂µ χ) = Fµν . while the fact µ C ∗ that the Minkowski space is real implies that Aµ (x) = Aµ (x) = Aµ (x).e. ˜ x Note that via parity transformation one obtains another solution. 0. 40 (5. (5. shows that the electromagnetic fields correspond to the case of a massless spin 1 field.17) (radiation. it will vanish at all times. moree over. j) the Maxwell equations read ∂µ F µν = j ν . Taking the divergence of this equiation. As remarked above.5.12) (5. The Lorenz condition can actually be incorporated by changing the equations of motion to 2Aµ + (λ − 1)∂µ (∂ν Aν ) = jµ . ∓1. One possible choice is the gauge A0 = 0 (5.8) (5. It states the absence of magnetic charges and Faraday’s law.16) (5. 5. The equations of motion for the fields Aµ become 2Aµ − ∂µ (∂ν Aν ) = jµ . AP (x) = Aµ (˜). For electrodynamics one has furthermore the freedom of gauge transformations. one finds for a conserved current and λ = 0 2∂µ Aµ = 0. The equation in vacuum. it does not eliminate the freedom of gauge transformations but they are now restricted to Aµ → Aµ + ∂µ χ with 2χ = 0. i. (5.

F = e E + ev × B (5. λ L − λ λ or ∆x = (L− λ/d)δ. The situation. This causes an interference pattern as a function of ∆x. (5. This occurs in the region outside the solenoid.1: The Aharonov-Bohm experiment 5.1. To be precise. cos ϕ. . where Aµ = 0. 5. The explanation of the significance of the vector ˆ ˆ potential A for the phase of the electron wave function is found in minimal substitution ψ ∝ exp i p·r −→ exp i (p − eA) · r e A · r exp i p·r . while the electromagnetic field Aµ is not. In the presence of an electromagnetic field Aµ an additional phase difference is observed. 0). 2 2 2 BR2 BR2 y BR2 x A= . . illustrated in fig. ˆ B = 0. Nevertheless the significance of Aµ is shown in the Aharonov-Bohm experiment. may look a bit akward. A= By Bx Br ϕ= − ˆ .Vector fields and Maxwell equations 41 L ∆x d a (infinitely) long solenoid Figure 5. The phase difference between the electrons travelling two different paths is given by δ = 2π a a ∆x d = −≈ . = 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. It is however.0 . . but where E = B = 0. This phase difference is actually observed.20) e (∇ × A) · ds = B · ds = where Φ is the flux through the solenoid.3 The electromagnetic field and topology∗ The electric and magnetic fields E and B as appearing for instance in the Lorentz force on a moving charge. In this experiment an observable phase difference is measured for electrons moving through field-free space (E = B = 0 but Aµ = 0). but A = 0 (hence there is structure in the vacuum). The energy density 2 (E 2 + B 2 ) = 0 (hence a vacuum). Inside solenoid: Outside solenoid: B = B z. nothing more than a manisfes1 tation of a nontrivial vacuum (a la the Dirac sea for fermions). 0) and ϕ = (− sin ϕ. ϕ= − ˆ 2r 2r2 2r2 where r = (cos ϕ.0 .19) are gauge invariant. sin ϕ. nevertheless.

In such a space loops with different winding number (i. minimizing the space occupied by B fields. Therefore the gauge transformation must be uniquely defined in the space one is working in. i. 1 The occurrence of nontrivial possibilities. Another situation in which the topology of space can be used is in the case of superconductors. A = ∇χ with χ= BR2 ϕ 2 (5. Consider a superconductor. however.Vector fields and Maxwell equations U(1) S (space with solenoid) 42 class 0 ei φ class 1 Π1(S) = Z class 2 Figure 5. Now back to the Aharonov-Bohm experiment.21) (ϕ being the azimuthal angle). forming a simple (connected) space. This situation that A can be written as ∇χ is always true when ∇ × A = 0.21. Below the critical temperature the magnetic field is squeezed out of the superconducting material. implying that it must be single-valued1 .2: The topology of the space in the Aharonov-Bohm experiment. 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. The difference here is that we are working in a space with a ’defect’ (the infinitely long solenoid). giving no observable phase2 . nonobservable phases ϕ = 2π n. The observable phase going around in general is ∆δ = e A · dr = e ∇χ · dr = e χ|ϕ=0 . organizing itself in tiny flux tubes (Abrikosov strings). 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. If this would be be happening in empty space one would be in trouble. has been employed by Dirac in constructing magnetic monopoles in electrodynamics 2 Actually. The AharonovBohm experiment shows a realization of this possibility. Therefore space outside the solenoid doesn’t care that χ is multi-valued. the number of times they go around the defect) cannot be continuously deformed into one another. The only way for flux tubes to be formed and move around without global consequences is when each flux tube contains a flux Φ such that ∆δ = (e/ )Φ = n · 2π. ϕ=2π The function χ in Eq. is multivalued as ϕ = 0 and ϕ = 2π are the same point in space.e. the relevant charge turns out to be 2e because the electrons appear in Cooper pairs . 5. it can be obtained from the situation A = 0 by a gauge transformation.e.

∂t ∂B = 0. 5.Vector fields and Maxwell equations 43 Exercises Exercise 5.1 Show that Eqs.8 and 5.9 explicitly give the equations ∇ · E = ρ. ∂t ∇ · B = 0. ∇×B =j+ ∇×E+ ∂E . .

recall classical mechanics with the action t2 S= t1 dt L(x.6) The first term leads to the Euler-Lagrange equations. (6. ˙ 2 (6. x (t) = x(t) + δx(t).2) The principle of minimal action looks for a stationary action under variations in the coordinates and time.5) with ∆x(t) = δx(t) + x(t) δτ . The requirement δS = 0 with fixed boundaries x(t1 ) = x1 and x(t2 ) = ˙ x2 leads to t2 ′ (6. These equations can be obtained from a lagrangian using the action principle. ∂x (6.3) (6.7) The quantity ∂L/∂ x plays a special role and is known as the canonical momentum. As an example. ˙ p= ∂L . x) ˙ (6. Dirac fields (Dirac equation) and massless vector fields (Maxwell equations) and corresponding to these fields conserved currents describing the probability and probability current. x) = K − V = ˙ 1 mx2 − V (x).1 Euler-Lagrange equations In the previous chapter we have seen the equations for scalar fields (Klein-Gordon equation). t1 (6.8) 44 . thus t′ = t + δτ.1) and as an example the lagrangian 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. and the total change x′ (t′ ) = x(t) + ∆x(t). d dt ∂L ∂x ˙ = ∂L .Chapter 6 Classical lagrangian field theory 6.

˙ (6. ∂µ φ(x). dσµ .17) Furthermore the variations δφ and δ∂µ φ contribute to δS.15) d4 x′ L (φ′ . ∆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). + ∂L ∆x − H δτ ∂x ˙ t2 .9) which is done because the first term (multiplying ∆x. − d3 x . . ∂µ φ′ .14) (6. . The resulting variation of the action is δS = R µ ′ = xµ + δxµ .11) λ=0 of which p and H are the simplest examples related to space and time translations. In classical field theory one proceeds in complete analogy but using functions depending on space and time (classical fields. Variations in the action can come from the coordinates or the fields. Consider a lagrangian density L which depends on these functions. .12) Here R indicates a space-time volume bounded by (R3 .16) The change in variables x → x′ in the integration volume involves a surface variation of the form dσµ L δxµ .e. x) and an action t2 S[φ] = t1 dt L = dt d3 x L (φ(x). H is a conserved quantity. . indicated with δF [φ]/δφ should pose no problems. x). δ(∂µ φ) (6. = φ(x) + δφ(x) (6. The hamiltonian H is defined by H(p. ∂µ φ.18) 1 Taking a functional derivative. also indicated by ∂R (a more general volume in four-dimensional space-time with some boundary ∂R can also be considered). t2 ). ˙ x The second term in Eq. (R3 . ∂µ φ(x)). which in classical mechanics vanishes at the boundary) does not play a role. . t1 (6.6 can be rewritten as δS = . with ∆φ(x) = δφ(x) + (∂µ φ)δx . We will come back to it in a bit more formal way in section 9. = ∂R (R3 .Classical lagrangian field theory 45 For the lagrangian specified above this leads directly to Newton’s law p = m¨ = −∂V /∂x. . ∂µ φ(x)) = R d4 x L (φ(x). i. but which in 3 dimensions includes conserved quantities related to rotations and boosts. t1 ) and R3 .13) (6. . indicated as x′µ φ (x) or combined φ′ (x′ ) = φ(x) + ∆φ(x). x) ≡ p x − L . (6. giving1 δS = = δL δL δ(∂µ φ) + δφ + δ(∂µ φ) δφ R δL δL δφ + d4 x − ∂µ δφ δ(∂µ φ) R d4 x dσµ L δxµ ∂R dσµ ∂R δL δφ + L δxµ .2. their derivatives and possibly on the position. L (φ(x). x′ ) − d4 x L (φ. .t1 ) (6. In general any continuous transformation. R (6.10) One sees that invariance under time translations requires that H(t1 ) = H(t2 ). . ∂R Note for the specific choice of the surface for constant times t1 and t2 . 6.t2 ) d3 x . . λ=0 (6. gives rise to a conserved quantity (at least if H is invariant!) Q=p dx dλ −H dt dλ .

(6.26) The Dirac field The appropriate lagrangian from which to derive the equations of motion is L = i → i ← i ↔ ψ / ψ − M ψψ = ψ / ψ − ψ / ψ − M ψψ ∂ ∂ ∂ 2 2 2 = ψ (i/ − M ) ψ.19) δ(∂µ φ) δφ 6. leading to the Euler-Lagrange equations.28) where the second line is not symmetric but in the action only differs from the symmetric version by a surface term (partial integration).27) (6.20) (6. (6.23) (6. the second term is irrelevant.21) which differ only by surface terms. the Dirac field (spin 1/2) and the vector field (spin 1) can be found.29) . leading to (2 + M 2 )φ(x) = 0. δL δ(∂µ ψ) δL δψ one obtains immediately i / −M ∂ → i = − γ µψ 2 i → = ∂ / ψ − M ψ. (2 + M )φ (x) = 0. Using the variations in ψ (in the symmetric form). For the complex scalar field one conventionally uses L = = ∂µ φ∗ ∂ µ φ − M 2 φ∗ φ −φ∗ ∂µ ∂ µ + M 2 φ. L = 1 1 ∂µ φ∂ µ φ − M 2 φ2 2 2 1 µ = − φ ∂µ ∂ + M 2 φ.22) which can be considered as the sum of the lagrangian densities for two real scalar fields φ1 and φ2 √ with φ = (φ1 + iφ2 )/ 2. 2 ψ = 0. (6.24) (6.2 Lagrangians for spin 0. The integrand of the first term must vanish.Classical lagrangian field theory 46 With for the situation of classical fields all variations of the fields and coordinates at the surface vanishing. ∂ (6. One easily obtains (2 + M 2 )φ(x) = 0.25) (6. 1/2 and 1 fields By an appropriate choice of lagrangian density the equations of motion discussed in previous chapters for the scalar field (spin 0). 2 ∗ (6. δL δL ∂µ = . The scalar field It is straightforward to derive the equations of motion for a real scalar field φ from the lagrangian densities. 2 (6.

In that case one finds trivially L = ↔ ↔ 1 1 ψR i / ψR + ψL i / ψL − M (ψR ψL + ψL ψR ).Classical lagrangian field theory 47 and similarly from the variation with respect to ψ ψ i / +M ∂ ← = 0. L = = ↔ 1 1 c c M ψL ψL + M ∗ ψL ψL ψL i / ψL − ∂ 2 2 ↔ 1 1 M ΥR ΥL + M ∗ ΥL ΥR .    ∗  0    ⇒ Υ ≡  −ǫ η  ΥL = ψL =    η η ∗ for which η0 = ǫ η0 . We note that of which the left part coincides   . namely2 LM (Majorana) = + (6. it is in the absence c of a Dirac mass term possible to introduce a lagrangian in which only left fields ψL and ψL appear or one can work with ’real’ spinors or Majorana spinors.36) (6. ∂ ∂ 2 2 (6. The Majorana case is in fact more general. or equivalently separate the field into right. Peshkin and Schroeder or Exercise 12.30) It is often useful to link to the two-component spinors ξ and η which we started with in chapter 4.34)  (6. the mass term in the Dirac lagrangian 6. and thus (β ∗ α)∗ = α∗ β = −βα∗ . . ψL ≡ (ψL )c = C ψL =    0 (6. that the lagrangian separates into two independent parts for M = 0.39) 2 This term is written down with η being an anticommuting Grassmann number for which (αβ)∗ = β ∗ α∗ = −α∗ β ∗ .g.33) We now introduce a ’real’ (four-component) spinor satisfying Υc = Υ with ψL . (6.32) There exists another possibility to write down a mass term with only one kind of fields. 1 M η † ǫη ∗ − M ∗ η T ǫη . 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 . but note the factor 1/2 as compared to the Dirac lagrangian. which comes because we in essence use ’real’ spinors.37) This gives an expression with a mass term that is actually of the same form as the Dirac mass term. (6. 2 αβ = −βα. Using the twospinors ξ and η.7). 2 (6.31) showing e. (6.31 is given by LM (Dirac) = −M ξ † η + η † ξ .38) (6.g. The reasons for Grassmann variables will become clear in the next chapter.and lefthanded ones.35) Since the kinetic term in L separates naturally in left and right parts (or ξ and η). since a lagrangian with both Dirac and Majorana mass terms can be rewritten as the sum of two Majorana lagrangians after redefining the fields (See e.

∂µ J µ (x) = 0. δ(∂µ φ) (6.43) F (σ1 ) − F (σ2 ) = dσµ J µ (x) = ∂R R d4 x ∂µ J µ (x). (6. δ(∂µ φ) (6. which is the space-integral over the zero-component of a conserved current.Classical lagrangian field theory 48 where Fµν = ∂µ Vν − ∂ν Vµ . Q= d3 x J 0 (x.47) .3 Symmetries and conserved (Noether) currents Next. which in a consistent picture precisely generate the symmetries. δ(∂µ φ) For continuous transformations.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 . discussed in the next chapter. . Returning to δS the surface term is rewritten to δS = R d4 x . 6.45) J µ (x) = − δ(∂µ φ) dλ λ=0 dλ λ=0 Thus considering ∂R = σ2 .. (6.41) (6. ∆φ = (dφ/dλ)λ=0 δλ and δxµ = (dxµ /dλ)λ=0 δλ. Q(t1 ) = Q(t2 ). the presence of a symmetry that leaves the lagrangian invariant. implying ∂µ F µν + M 2 V ν = 0 and ∂µ V µ = 0. one can consider the variations at the initial or final surface. (6. (6. ψ(x) → ei eΛ ψ(x) or ∆ψ(x) = i eΛ ψ(x). we get (omitting the parameter δλ) dxν dφ δL + Θµν (x) . . one immediately obtains δL = −∂ µ V ν + ∂ ν V µ − λ g µν (∂ρ V ρ ) = −F µν − λ g µν (∂ρ V ρ ) δ(∂µ Vν ) leading to the equations of motion ∂µ F µν + M 2 V ν + λ ∂ ν (∂ · V ) = 2 + M 2 V ν − (1 − λ)∂ ν (∂ · V ) = 0.40) 6. (6.44) δL ∆φ + Θµν (x)δxν .42) Θµν = The variation δS thus can be expressed as δS with J µ (x) = − = δL ∂ ν φ − L g µν . In this case.18. consider U (1) transformations of the Dirac field proportional to the charge e. These do not affect the dynamics and will give rise to conserved currents. requires the presence of a conserved quantity. t).σ1 with σ = (R3 . In particular when the lagrangian is invariant under symmetries. + ∂R dσµ dσµ ∂R ≡ where d x . t).. these conserved quantities become operators. As an example for the classical field.+ R 4 δL δL ∆φ − ∂ ν φ − L g µν δxν δ(∂µ φ) δ(∂µ φ) δL ∆φ − Θµν (x)δxν . we consider the second term in Eq.

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.49) These currents are conserved as discussed already in chapter 2. (6. ↔ (6. 6. 1 (6.Classical lagrangian field theory From the lagrangian for the Dirac field.3) j µ = i e φ∗ ∂µ φ. = 0.54) (6.48) For the complex scalar field the U (1) transformations leave the lagrangian invariant and lead to the conserved current (see Exercise 6.51) (6. Translations µ Under translations (generated by Pop ) we have δxµ ∆φ(x) δφ(x) = ǫµ . . (6. H = d3 x Θ00 (x).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. The integral over the zero-component. In the next section we will see the quantity show up as the charge operator. From Noether’s theorem one sees that the current Θµν ǫν is conserved. t) is the conserved charge (Q = 0).56) δxµ = ωρσ (aρσ )µν xν 2 with (aρσ )µν = aρσ µν = g ρµ g σν − g σµ g ρν . (6. Lorentz transformations In this case the transformation of the coordinates and fields are written as δxµ ∆φi (x) = = ω µν xν . (6. we obtain (since δxµ = 0). 2 (6. = −(∂µ φ)δxµ = −ǫµ ∂µ φ(x).50) (6. ˙ Q = d3 x j 0 (x. including e the space-time translations and the Lorentz transformations.55) Note that the coordinate transformations can be written in a form similar to that for the fields (matrices in internal space). For quantized fields this will become the expressions of the hamiltonian and the momentum operators in terms of the fields.g. (ω µν antisymmetric) 1 ωρσ (−i S ρσ )ij φj (x).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 .53) These are the energy and momentum.4 Space-time symmetries One kind of symmetries that leave the lagrangian invariant are the Poincar´ transformations. e. Summarizing.

66) In section 6. In some applications (specifically coupling to gravity) it is advantageous to have an equivalent current that is symmetric. (6. Defining T µν = Θµν − ∂ρ Gρµν . γσ ] This result for the Dirac field has been discussed in chapter 3 (see also Exercise 4. is not invariant under local gauge transformations. Any lagrangian containing derivatives. • Dirac field ψ: −i Sρσ = −(1/4)[γρ .67) i. e. −i eΛ(x) ∗ i eΛ(x) (6.58) (6. ∂µ φ(x) + i e ∂µ Λ(x) e (6. however.61) A final note concernes the symmetry properties of Θµν . (6.68) (6. in which the U (1) transformation involves an angle e Λ. (6.64) (6.3). 2 (6.Classical lagrangian field theory 50 • Scalar field φ: −i Sρσ = 0.62) with Gρµν = (H ρµν + H µνρ + H νµρ )/2. where H ρµν is the quantity appearing in M ρµν . (6. ∂µ T µν (6.65) M µρσ = T µρ xσ − T µσ xρ . Gauge transformations of the second kind or local gauge transformations are transformations of the type φ(x) → ei eΛ(x) φ(x). The current deduced from Noether’s theorem is Jµ = = = = − δL ∆φi + Θµν δxν δ(∂µ φi ) δL 1 ωρσ − (−i S ρσ )ij φj + Θµρ xσ − Θµσ xρ 2 δ(∂µ φi ) 1 ωρσ {H µρσ + Θµρ xσ − Θµσ xρ } 2 1 ωρσ M µρσ . (6. The lagrangians which we have considered sofar are invariant under global gauge transformations and corresponding to this there exist a conserved Noether current.63) µν = ∂µ Θ . • Vector field Aµ : −i Sρσ = aρσ .6) T µν = T νµ .69) i eΛ(x) φ (x). one has a tensor that satisfies (exercise 6.70) .60) Therefore there are six conserved currents M µρσ labeled by ρ and σ (antisymmetric) and corresponding to it there exist conserved quantities M ρσ = d3 x M 0ρσ . φ(x) φ (x) ∂µ φ(x) ∗ → e → e → ei eΛ(x) φ(x). In general this tensor is not symmetric.5 (Abelian) gauge theories φ(x) → ei eΛ φ(x). the angle of the transformation depends on the space-time point x.59) (6.3 we have seen global transformations or gauge transformations of the first kind.g. independent of x.e. φ(x). 6.

∂µ φ) =⇒ L (φ. (6. For this purpose it is necessary to introduce a vector field Aµ . D ∂ A (6. The equation of motion for the fermion follow from δL δ(∂µ ψ) δL δψ = = i − γ µ ψ.75) The field φ is used here in a general sense standing for any possible field.71) Thus the covariant derivative has the correct transformation behavior provided Aµ → Aµ − ∂µ Λ. −e j µ Aµ = −e ρφ + e j · A.78) .Classical lagrangian field theory 51 where the last term spoils gauge invariance. 1 L (φ. A solution is the one known as minimal substitution in which the derivative ∂µ is replaced by a covariant derivative Dµ which satisfies Dµ φ(x) → ei eΛ(x) Dµ φ(x). i L = ψγ µ ∂µ ψ − (∂µ ψ)γ µ ψ − M ψψ.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 . 4 (6. 2 Minimal substitution ∂µ ψ → (∂µ + i eAµ )ψ leads to the gauge invariant lagrangian L = 1 i ↔ ψ / ψ − M ψψ − e ψγ µ ψ Aµ − Fµν F µν . Dµ φ) − Fµν F µν . Dµ φ(x) ≡ (∂µ + i eAµ (x))φ(x).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 .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). (i/ − M ) ψ = (i/ − e/ − M ) ψ = 0.73) (6. As an example consider the Dirac lagrangian. The required transformation for Dµ then demands Dµ φ(x) = = (∂µ + i eAµ (x))φ(x) ei eΛ ∂µ + i e(A′ + ∂µ Λ) φ µ → ei eΛ ∂µ φ + i e (∂µ Λ) ei eΛ φ + i eA′ ei eΛ φ µ ≡ ei eΛ (∂µ + i eAµ ) φ. (6. 2 i ∂ / ψ − M ψ − e/ ψ A 2 giving the Dirac equation in an electromagnetic field. (6.72) (6. ∂ 2 4 (6. Replacing derivatives by covariant derivatives and adding the (free) part for the massless vector fields to the original lagrangian therefore produces a gauge invariant lagrangian.

where j µ = e ψγ µ ψ. 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. giving the Maxwell equation coupling to the electromagnetic current. [iγ µ (∂µ − i e Aµ ) − M ]ψ = 0. ∂µ F µν = j ν . i. 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.Classical lagrangian field theory 52 For the photon the equations of motion follow from δL δ(∂µ Aν ) δL δAν = = −F µν . (c) Show that the interaction term is invariant under gauge transformations only if the current jµ is conserved. Exercise 6. φ → eiΛ φ. ↔ Exercise 6.2 (a) Show that the homogeneous Maxwell equations can be derived from the lagrangian density 1 L = − Fµν F µν . (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. −eψγ ν ψ.1 (a) Show that the Klein-Gordon equation for the real scalar field can be derived from the lagrangian density 1 1 L = ∂µ φ∂ µ φ − M 2 φ2 .e. .79) Exercises Exercise 6. ∂µ j µ = 0 (Note that the addition of a total derivative to the lagrangian density does not modify the equations of motion).4 Given the Dirac equation for a (negatively charged) spin-1/2 particle in an external electromagnetic T field. ∂µ F µν = j ν . and deduce what is the charge of the antiparticle as compared to a particle. give the equation which is satisfied by ψ c = Cψ .

Hints: it may be useful to realize that Gµρσ is antisymmetric in first two indices. e 1 (−i∇ + eA)2 + σ · B − eA0 ψu (r. in order to show that Θρσ − Θσρ = ∂µ H µρσ . t). A) (see also exercise 6. Comment: this relation is known as the Pauli equation. finally note that it is sufficient to show the last equation for dσµ M µρσ . 2me Exercise 6. .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 .4. 2M 2M with Enr ≡ E − M ≪ M .Classical lagrangian field theory 53 Exercise 6.6 (optional) Prove the properties of the tensor T µν in section 6.5 Show that the Dirac equation for an electron with charge −e in an external electromagnetic field Aµ = (A0 . 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. Hint ≡ −gs µe · B = −gs Qe s · B. t) = Enr ψu (r. use ∂µ T µν = 0 and ∂µ M µρσ = 0.

q) = ˙ 1 mq 2 − V (q). Actually the conserved quantities satisfy a similar algebra as the Lie algebras for continuous symmetries that they are obtained from. Further.3) For variations δO(p. 1 qx (t) = 3 d3 x φ(x.1) The Hamiltonian (also corresponding to a conserved quantity because of time translation invariance) is given by p2 + V (q).Chapter 7 Quantization of fields 7. starting from the lagrangian L(q. 6. there are conserved ˙ ˙ quantities Q (Eq. (7.1 Canonical quantization We will first recall the example of classical mechanics for one coordinate q(t). (7. in our example p = mq. namely [q. ˙ 2 (7. Q]P . Quantization of the theory is achieved by imposing canonical commutation relations between q and p. t). B]P = dB dA dA dB − . An immediate generalization for fields can be obtained by considering them as coordinates. q) = p q − L(q. however. p] = i. dλ Examples are dO = [O. q) ≈ (dO/dλ)λ=0 δλ under symmetry transformations one has dO = [O. p)) = ˙ ˙ 2m where the (canonical) momentum p = ∂L/∂ q. q(q. dq The basic Poisson bracket is actually the one between q and its canonical momentum p. The bilinear product for the conserved quantities.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. is the Poisson bracket [A.5) ∆ x ∆3 x 54 .2) H(p. (7. q) also considered in the previous chapter. p]P = 1. H]P dt and dO = [O. This is nothing else as representing the classical Poisson algebra as operators in the Hilbert space. labeled by the position. dq dp dq dp (7.11) found through Noether’s theorem. ˙ L(q. [q. p]P .

a) φi (x)U (Λ.8) (7. t).14) In fact.16) .6) ∆3 x L x (qx . i U † (Λ. (7.12) (7.Quantization of fields 55 etc. ∂ qx ˙ ∂ qx ˙ (7. t)] = 0.11) (7. albeit with ’sharp’ functions. t)] = i δ 3 (x − y). the fields φ(x) can be considered as regular operators in the Hilbert space after smearing with a test function f . δ∂0 φ) 1 1 1 = Θ00 (x) = (∂0 φ)2 + (∇φ)2 + M 2 φ2 . we have L (x) = 1 1 ∂µ φ∂ µ φ − M 2 φ2 2 2 1 1 1 2 (∂0 φ) − (∇φ)2 − M 2 φ2 = 2 2 2 δL = = ∂0 φ. discussed in the section 6. This implies specific transformation properties in the Hilbert space for the quantum fields φ. φ. formally. a) = Rj (Λ−1 ) φj (Λ−1 x − a). t). The lagrangian is given by L = = x In order to construct the hamiltonian it is necessary to find the canonical momenta. for the scalar field theory. py (t)] = i δxy or for the fields φ(x) and Π(x) the socalled equal time commutation relations [φ(x. The hamiltonian then is H= x px qx − L = ˙ = x where H (x) = δL ˙ φ(x) − L (x) = Θ00 (x). qx+∆x ) . Φ(f ) = d4 x φ(x) f (x). ∂µ φ(x)) = ˙ d3 x L (φ. for symmetry transformations. ˙ δ φ(x) (7. px (t) = ∂L ∂Lx = ∆3 x = ∆3 x Πx (t). φ(y. As an example. t)] = [Π(x.2.13) Π(x) H (x) For the quantization procedure we can formulate a number of basic axioms of quantum field theory. ˙ where Πx (t) is obtained from the continuous field Π(x) ≡ δL /δ(∂0 φ). (7. ∇φ) (7.9) Note that this indeed corresponds to the zero-zero component (Θ00 ) of the conserved energy-momentum stress tensor Θµν . 2 2 2 (7.15) with furthermore the relations [φ(x. (7. • Canonical commutation relations The quantization of the theory is achieved by imposing the canonical quantization condition [qx (t). The essential conditions to consistently quantize a theory then are the following.10) x ∆3 x [Πx qx − L x ] ˙ ∆3 x Hx (7. Π(y.7) d3 x L (φ(x). Sometimes it is useful to keep in mind that. • Poincar´ invariance e The full variation of the fields. Π(y. the discretization procedure above is an explicit example. qx . ∆φ. t).

a† ] = a† .17) (Sµν )i φj (x). P ] 2 (7.22) it is straightforward to check that the commutation relations between Q and P are equivalent with the commutation relations [a. and [N.24) It is straightforward to find the commutation relations between N and a and a† . a† ] = 1. The measurements cannot influence each other or [Φ(f ).20) where the coordinate Q and the momentum P satisfy the canonical commutation relations [Q. Q] = [P. 2 i ωµν (S µν )i . Mµν ] = i i∂µ φi (x). H = = ω ω P P Q−i √ Q+i√ 2 2 2ω 2ω 1 1 =ω N+ . (7. j 2 one has [φi (x). 1 Q = √ (a + a† ) 2ω and P = −i ω (a − a† ) 2 (7. [a. Φ(g)] = 0. i(xµ ∂ν − xν ∂µ )φ (x) + i (7.2 Creation and annihilation operators Before discussing (real and complex) scalar fields and Dirac fields we recall the analogy with the wellknown harmonic oscillator as an example of quantization using creation and annihilation operators. [φ(x). P ] = i. P ] = 0 (7. a] = −a. In simplified form the hamiltonian is given by H= 1 2 1 2 2 P + ω Q . φ(y)] = 0 if (x − y)2 < 0. • Causality Operators Φ(f ) and Φ(g) for which the test functions are space-like separated can be measured simultaneously (macroscopic causality). (7. a) = i Rj (Λ−1 ) = 1 + iǫµ P µ − 1− i ωµν M µν . 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.21) Writing Q and P in terms of creation (a† ) and annihilation (a) operators.Quantization of fields 56 or if U (Λ.19) 7. 2 2 (7. ω a† a + 2 2 ω i − [Q. sometimes referred to as second quantization.18) with S µν given in 6. a† ] = 0. (7. [N.25) . Pµ ] = [φ (x). Microscopic causality implies local commutativity. [Q.4. j (7. a] = [a† .23) The hamiltonian in this case can be expressed in the number operator N = a† a.

and (7. The quantization of the field is achieved by quantizing the coefficients in the Fourier expansion.28) (7.Quantization of fields 57 Defining states |n as eigenstates of N with eigenvalue n. † √ a and a act as raising and lowering operators.33) . 7.3 The real scalar field We have expanded the (classical) field in plane wave solutions. and we see that a state |0 must exist for which N |0 = a|0 = 0.e. i. Note that we will often write a(k) or a† (k). d3 x φ(x) i ∂0 e−i k ·x ↔ ′ ↔ (7. but one needs to realize that in that case k 0 = Ek = k2 + M 2 .e. 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.4 for the classical field) a(k) a† (k ′ ) = = d3 x ei k·x i ∂0 φ(x). N |n = n|n one finds N a† |n N a|n = = (n + 1) a† |n .32) (7. The canonical momentum becomes −i d3 k ˙ Π(x) = φ(x) = a(k) e−i k·x − a† (k) ei k·x . e. a† (k ′ )] = (2π)3 2Ek δ 3 (k − k′ ) [a(k). (2π)3 2Ek (7.26) a(k) e−i k·x + a† (k) ei k·x . which represents the ’number of particles’ with momentum k.30) [φ(x). 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 .31) The hamiltonian can be rewritten in terms of a number operator N (k) = N (k) = a† (k)a(k). [φ(x).g. φ(x) = (2π)3 2Ek where the Fourier coefficients a(k) and a† (k) are now operators. (7. which we have split into positive and negative energy pieces with (complex) coefficients a(k) and a∗ (k) multiplying them. a(k ′ )] = [a† (k).29) It is straightforward to prove that the equal time commutation relations between φ(x) and Π(x′ ) are equivalent with ’harmonic oscillator . Π(x )]x0 =x′0 = 0. the real scalar field φ(x) becomes d3 k (7. Π(x′ )]x0 =x′0 = i δ 3 (x − x′ ) ′ ′ and (7. a† (k ′ )] = 0.27) 2 (2π)3 It is easy to check that these equations can be inverted (see Exercise 2. φ(x )]x0 =x′0 = [Π(x).like’ commutation relations between a(k) and a† (k ′ ). (n − 1) a|n † i. From the normalizations one obtains a |n = √ n + 1 |n + 1 and a|n = n |n − 1 . is equivalent with [a(k).

37) (7. (2π)3 2Ek (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).39) The problem with the vacuum or zero-point energy. |k = a† (k)|0 . defining particle states |k = |k (with positive energy. writing all annihilation operators to the right of the creation operators. . which amounts to redefining H as H= d3 x : H (x) := d3 k Ek N (k).40) This procedure is known as normal ordering. (2π)3 2Ek (7. . (2π)3 2Ek (7.44) p|φ− (x)|0 = ei p·x . This term will be adressed below. k 0 = Ek = k2 + M 2 ). .36) and satisfying the completeness condition 1 = k|k ′ = (2π)3 2Ek δ 3 (k − k′ ).38) d4 k 2π δ(k 2 − M 2 ) θ(k 0 ) |k k| = (2π)4 (7. i. = normalized as (a† (k1 ))n1 (a† (k2 ))n2 √ √ .Quantization of fields 58 where the necessity to commute a(k)a† (k) (as in the case of the quantum mechanics case) leads to a zero-point energy. (2π)3 2Ek (7. For the momentum operator one has Pi = d3 x Θ0i (x) = 1 2 d3 x ∂ {0 φ ∂ i} φ = d3 k k i N (k). φ(x) φ+ (x) φ− (x) = = = φ+ (x) + φ− (x). . For the purpose of normal ordering it is convenient to decompose the field in positive and negative frequency parts.46) (7. |0 . The rest of the states are then obtained from the groundstate via the creation operator. and multiparticle states |(k1 )n1 (k2 )n2 .e. (2π)3 (7. a(k)|0 = 0.42) (7. (2π)3 2Ek d3 k a† (k) ei k·x .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).34) where V = (2π)3 δ 3 (0) is the space-volume.41) (7. . is solved by subtracting it as an (infinite) constant. assuring that the vacuum (by definition) has eigenvalue 0!. d3 k |k k|. in field theory also referred to as vacuum energy Evac = 1 V 2 d3 k Ek .35) where the vacuum contribution disappears because of rotational symmetry.45) (7. d3 k a(k) e−i k·x . n1 ! n2 ! (7. which now contains an infinite number of oscillators. The 1-particle wave function e−ip·x is obtained via 0|φ(x)|p p|φ(x)|0 = = 0|φ+ (x)|p = d3 k 0|a(k)a† (p)|0 e−i k·x = e−i p·x (2π)3 2Ek (7.

48) (7.Quantization of fields 59 In order to ensure the consistency of the theory it is necessary to check that the operators P ν and M obtained from the conserved currents Θρσ and M µρσ are indeed the generators of the Poincar´ e group.50) or as integrals over d4 k. i.52) The result for the invariant commutator function is [φ(x). (7.55) d4 k ǫ(k 0 ) 2π δ(k 2 − M 2 ) e−i k·x . x) = 0 and hence ∆(x) = 0 for x2 < 0. 2π (7.17) and (7. (iii) ∆(0. In order to calculate [φ(x).47) ′ d3 k ′ e−i k·x+i k ·y [a(k). (2π)4 d4 k − θ(k 0 ) 2π δ(k 2 − M 2 ) ei k·x (2π)4 d4 k − θ(−k 0 ) 2π δ(k 2 − M 2 ) e−i k·x = −i∆+ (−x) (2π)4 (7. φ− (y)] ≡ i∆+ (x − y). (2π)4 (7. = − (2π)3 2Ek (7. d3 k ei k·(x−y) = −i∆+ (y − x). a† (k ′ )] 3 2E ′ (2π) k [φ− (x). i∆+ (x) i∆− (x) = = = d4 k θ(k 0 ) 2π δ(k 2 − M 2 ) e−i k·x .18). ∆(x) = − ǫ(x0 ) δ(x2 ). d3 k = (2π)3 2Ek = (7. (ii) ∆(x) is a solution of the homogeneous Klein-Gordon equation. φ(y)] consider µν [φ+ (x).56) = −δ 3 (x). (2π)3 2Ek ≡ i∆− (x − y).51) (7. (7. (iv) The equal time commutation relations follow from ∂ ∆(x) ∂t (v) For M = 0.53) t=0 . φ(y)] which has the properties (i) i∆(x) = i∆+ (x) + i∆− (x) can be expressed as i∆(x) = where ǫ(k 0 ) = θ(k 0 ) − θ(−k 0 ). φ+ (y)] d3 k e−i k·(x−y) .49) (7.e.54) = i∆(x − y) = i (∆+ (x − y) + ∆− (x − y)) . The last item to be checked for the scalar field are the causality condition. that they satisfy the required commutation relations in Eqs (7.

extending it with the charge operator and the introduction of particle and antiparticle operators. Also the charge operator requires normal ordering (in order to give the vacuum eigenvalue zero). ∂0 φ† (y)]x0 =y0 = i δ 3 (x − y). we will consider it as a repetition of the quantization procedure. a† (k ′ )] = [b(k).5 The Dirac field L = ↔ i ψγ µ ∂µ ψ − M ψψ.59) (7. (2π)3 2Ek (7.60) = i φ∗ ∂µ φ.64) The commutator of φ and φ† is as for the real field given by [φ(x). ↔ (7. (2π)3 2Ek d3 k b(k) e−i k·x + a† (k) ei k·x . Q = = = i d3 x : φ† ∂0 φ − ∂0 φ† φ(x) : d3 k : a† (k)a(k) − b(k)b† (k) : (2π)3 2Ek d3 k a† (k)a(k) − b† (k)b(k) .61) (7.e. = ∂{µ φ∗ ∂ν} φ − L gµν . particles (created by a† ) and antiparticles (created by b† ) with the same momentum contribute equally to the energy.58) and satisfy the equal time commutation relation (only nonzero ones) [φ(x). The hamiltonian is as before given by the normal ordered expression H = = = d3 x : Θ00 (x) : d3 k Ek : a† (k)a(k) + b(k)b† (k) : (2π)3 2Ek d3 k Ek a† (k)a(k) + b† (k)b(k) . φ† (y)] = i∆(x − y). (2π)3 2Ek (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 .62) (7. 2 From the lagrangian density (7.66) .63) i. b† (k ′ )] = (2π)3 2Ek δ 3 (k − k′ ).57) (7. (7. which is equivalent to the relations (only nonzero ones) [a(k).65) 7.Quantization of fields 60 7.4 The complex scalar field In spite of the similarity with the case of the real field. (2π)3 2Ek (7.

jµ Θµν = = ψγµ ψ. s)vj (k.71) (7.79) Also for the field and the canonical conjugate momentum anticommutation relations are considered. s)e−i k·x + d† (k.72) ψ(x) = s In terms of the operators for the b and d quanta the hamiltonian and charge operators are (omitting mostly the spin summation in the rest of this section) H = = Q = = d3 x : ψ † (x) i∂0 ψ(x) : d3 k Ek : b† (k)b(k) − d(k)d† (k) : (2π)3 2Ek d3 x : ψ † ψ : d3 k : b† (k)b(k) + d(k)d† (k) : . (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) .Quantization of fields 61 the conserved density and energy-momentum currents are easily obtained. s)v(k. s′ )} = (2π)3 2Ek δ 3 (k − k′ ) δss′ . d† (k ′ . (2π)3 2Ek (7. ↔ i ψγµ ∂ν ψ − 2 (7. {b(k. i. s). . s) e−i k·(x−y) j s † + vi (k. † {ψi (x).74) (7.68) The canonical momentum and the hamiltonian are given by Π(x) H (x) = = δL = i ψ † (x).77) Note that achieving normal ordering.75) (7. (2π)3 2Ek d3 k b† (k. s) ei k·(x−y) x0 =y 0 . The quantized fields are written ψ(x) = s d3 k b(k.76) which seems to cause problems as the antiparticles (d-quanta) contribute negatively to the energy and the charges of particles (b-quanta) and antiparticles (d-quanta) are the same. ∂ 2 ↔ (7.78) (7. (7. s)e−i k·x . s′ )} = {d(k. s).69) (7. b† (k ′ .67) i ψ / ψ − M ψψ gµν .73) (7. s)u(k. The solution is the introduction of anticommutation relations. ˙ δ ψ(x) ↔ i Θ00 (x) = − ψγ i ∂i ψ + M ψψ 2 = iψγ 0 ∂0 ψ = iψ † ∂0 ψ.e.70) where the last line is obtained by using the Dirac equation. u v (2π)3 2Ek (7. ψj (y)}x0 =y0 = d3 k (2π)3 2Ek ui (k. s)ei k·x . s)¯(k. s)u† (k. (7. s)ei k·x + d(k. interchanging creation and annihilation operators. s)¯(k.

∂ (7. 4 2 This gives the equations of motion discussed before.80) 3 = δ 3 (x − y) δij . ˙ δ Ai (7. we would have obtained [ψi (x). as the vanishing of Π0 induces a constraint. ψ j (y)] = (i/x + M )ij i∆1 (x − y). 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). it implies the Lorentz constraint and leads to the canonical momenta Π0 Π i = = E i.82) where i∆1 = i∆+ −i∆− . ˙ δ A0 δL = F i0 = E i . (7. but we must restrict ourselves to the weaker condition B|∂µ Aµ |A = 0.85) which reflects the gauge freedom. ∂ (7.83) From the lagrangian density the canonical momenta are = = (7. one has † {ψi (x). When we would have started with commutation relations for the field ψ and the canonical momentum. 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 (7.6 The electromagnetic field 1 L = − Fµν F µν .86) (7. −λ(∂ρ Aρ ). (7. It is possible to continue in a covariant way with the lagrangian 1 λ L = − Fµν F µν − (∂µ Aµ )2 . 4 Π0 Πi δL = 0.84) (7.88) .Quantization of fields 62 Using the positive and negative energy projection operators discussed in section 4.81) where i∆ = i∆+ + i∆− is the same invariant commutator function as encountered before. For the scalar combination {ψi (x). ψ j (y)} at arbitrary times one has {ψi (x). 7.87) If one wants to impose canonical commutation relations ∂µ Aµ = 0 cannot hold as an operator identity. but has the problem of being noncovariant. 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).

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. [ℓi .93) which does exhibit problems with the time-like photon. (d) Use Poisson brackets to calculate q and p. These problems are solved by the Lorentz constraint between physical states given above.90) where Πν = F ν0 − λ g ν0 (∂ρ Aρ ) and we have furthermore [Aµ (x). λ)|A = 0. i. where Q is the conserved quantity in Eq. (c) Calculate the Poisson brackets for the coordinates x and the standard conserved quantities p and ℓ.91) (7.89) with four independent vectors ǫµ . (7. It gives 3 k µ ǫ(λ) (k)c(k. one time-like photon by itself is not allowed! This solves the problems with the normalization and the negative energies. a longitudinal photon and two transverse photons. 2. pj ]P . j = 1.1 (optional) If you haven’t seen Poisson brackets in classical mechanics you should (a) Proof that Poisson brackets (Eq. 0)c(k. λ) − c† (k.11.e. The canonical equal time commutation relations are [Aµ (x). 7.3) satisfy the conditions of a Lie algebra. Aν (y)] = [Πµ (x). ′ (7. containing a time-like photon. The quantized field is expanded as Aµ (x) = d3 k (2π)3 2Ek (λ) 3 ǫ(λ) (k) c(k. Πν (y)]x0 =y0 = i gµν δ 3 (x − y). λ)c(k. rj ]P . what are [ri . and are equivalent with [c(k. ˙ ˙ . which reproduce the Hamilton equations. In fact the commutation relations imply ˙ [Aµ (x). c† (k ′ . λ′ )] = −g λλ 2Ek (2π)3 δ 3 (k − k′ ).Quantization of fields 63 for physical states |A and |B . (b) Proof for a quantity O(p. Πν (y)] = 0. where Aµ + + is the part of the vector field containing the annihilation operators. 0. Aν (y)]x0 =y0 = i gµν δ 3 (x − y). 0) .95) Exercises Exercise 7. λ). λ)e−i k·x + c† (k. 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.e. 0. for which it is sufficient that ∂µ Aµ |A = 0.94) Choosing k µ = (|k 3 |. Q]P . (7. q) that dO/dλ = [O. pj ]P . µ λ=0 (7. λ)ei k·x . k 3 ) this reads |k 3 |c(0) (k) − k 3 c(3) (k) |A = 0 c(0) (k) ∓ c(3) (k) |A = 0. i. 6. and [ℓi . (7. µ λ=0 (7.

It is invariant.3 (a) Show that the correct implimentation of symmetries in the Hilbert space. Pµ ] = i∂µ φ(x). we can write for ∆R an integral where the k 0 integration remains along the real axis. (2 + M 2 ) ∆inhom (x) = −δ 4 (x). Pµ ] for the (real) scalar field using the expressions for the fields and momentum operator in terms of creation and annihilation operators. (e) Show that ∆R (x) = 0 if x0 < 0. Is the subtraction of ’zero point’ contributions essential in this check? µ Pµ φ(x) ei a µ Pµ = φ(x − a) Exercise 7. ∆hom (Λx) = ∆hom (x) if f is an invariant function. Give also the expression for ∆C . ∆R = 1 (2π)4 d4 k e−ikx k2 1 − M 2 + iǫk 0 (with the implicit prescription to take ǫ → 0 after integration).Quantization of fields 64 Exercise 7. (b) Check the above commutation relation [φ(x). f (Λk) = f (k) for Lorentz transformations Λ. (2 + M 2 ) ∆hom (x) = 0. (c) What are the poles in the integral under (b) in the (complex) k0 plane? (d) Depending on the paths in the k0 plane going from k0 = −∞ to k0 = +∞ one can distinguish four different Green’s functions (solutions to the above inhomogeneous Klein-Gordon equation): k0 -E E ∆ R (retarded) ∆ C (causal) ∆ ∆A (advanced) Show that by appropriate shifts of the poles into the complex plane. (Similarly one can show that for the advanced Green’s function ∆A (x) = 0 if x0 > 0). e−i a requires [φ(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.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. (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.

f (k) ∝ θ(k 0 ).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). or ǫ(k 0 ) = θ(k 0 ) − θ(−k 0 ). x) = 0 and argue that one can use Lorentz invariance to show that ∆(x) = 0 for x2 < 0. In that case one actually needs discontinuous functions f (k) such as the step function. (g) Show that the homogeneous solution ∆ in (f) satisfies ∆(0. (i) (Optional) Give similar expressions for the other Green’s functions under (d) in terms of the homogeneous solutions under (f). (h) Show that the causal Green’s function ∆C (x) = θ(x0 ) ∆+ (x) − θ(−x0 ) ∆− (x). .

Chapter 8 Discrete symmetries In this chapter we discuss the discrete symmetries. starting with the Dirac equation for ψ(x). The consequences of P. 8. T and C for classical quantities is shown in the table 1.1: The behavior of classical quantities under P. writing −1 ψ(x) −→ Pop ψ(x)Pop = ηP Aψ(˜) ≡ ψ p (x). time reversal (T) and charge conjugation (C). ˜ (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 . parity (P). r) −→ xµ ≡ xµ = (t.2) where ηP is the intrinsic parity of the field and A is a 4 × 4 matrix acting in the spinor space. x (8. We can determine A. Both ψ p and ψ satisfy the Dirac equation. (iγ µ ∂µ − M ) ψ(x) = 0. −r). T.1 Parity xµ = (t. Table 8.1) The parity operator transforms We will consider the transformation properties for a fermion field ψ(x).

In the same way as the Fermion field.5) (8. one can also consider the scalar field and vector fields.4) (8.10) i. λ)u(k.2 Charge conjugation We have already seen the particle-antiparticle symmetry with under what we will call charge conjugation the behavior T (8.13) ψ(x) −→ ψ c (x) = ηC C ψ (x). x µ (iγ ∂µ − M ) γ0 ψ(˜) = 0. λ)u(k. −λ) eik·x (2π)3 2Ek ˜ d3 k ˜ ˜ ˜ ˜ ηP b(k. m). m) = = ˜ u(k. iγ µ† ∂µ − M ψ(˜) = 0. x (8. op P (8. x It is straightforward to apply this to the explicit field operator ψ(x) using γ0 u(k. −λ) e−ik·x − d† (k. m).11) (8.Discrete symmetries 67 After parity transforming x to x the Dirac equation becomes after some manipulations ˜ ˜ x iγ µ ∂µ − M ψ(˜) = 0. −λ) e−ik·x − d† (k. choosing ηP is real (ηP = ±1) particle and antiparticle have opposite parity. λ)γ0 u(k. (2π)3 2Ek (8. λ)γ0 v(k. The result is ψ p (x) −1 = Pop ψ(x) Pop = ηP γ0 ψ(˜) x (8. λ)v(k. For the scalar field we have seen −1 φ(x) −→ Pop φ(x) Pop = ηP φ(˜). λ) eik·x .e. γ x (8.8) 3 = λ = λ = λ From this one sees immediately that −1 ˜ Pop b(k.9) (8. −λ)u(k.3) (8. 8. −λ). −λ) eik·x 3 2E (2π) ˜ k d3 k ˜ ˜ ηP b(k. λ) Pop = ηP b(k. the above operation reverses the sign of λ). λ) e−ik·x − d† (k. m) γ0 v(k. .12) and for the vector field −1 Aµ (x) −→ Pop Aµ (x) Pop = −Aµ (˜). λ) e−ik·˜ + d† (k. (i˜ µ ∂µ − M ) ψ(˜) = 0. ˜ Pop d(k. −λ)v(k. if helicity λ is used instead of the z-component of the spin m.6) (check this for the standard representation.7) = λ d k x x ηP b(k. λ) eik·˜ (2π)3 2Ek d3 k ˜ ˜ ˜ ˜ ηP b(k. λ) P −1 = −η ∗ d(k. x Therefore γ0 ψ(˜) is again a solution of the Dirac equation and we have x ψ p (x) = γ0 ψ(˜). ˜ −v(k. x The latter behavior of the vector field will be discussed further below. λ)v(k. −λ).

21) where A is a 4 × 4 matrix acting in the spinor space. λ) e−ik·x + b† (k. The same relations hold for helicity states. λ). antilinear operator is anti-unitary if A† = A−1 . λ)v(k.16) d k ηC d(k. γ x i(γ5 C)−1 γ µ γ5 C∂µ − M ψ(−˜) = 0. x (8. λ)u(k. r).19) Cop d(k.15) (where one must be aware of the choice of spinors made in the expansion. m). 8.3 Time reversal xµ = (t. λ) Cop = ηC d(k. m) ¯ = = v(k. λ) eik·x . λ) e−ik·x + b† (k. λ) eik·x ¯ ¯ (2π)3 2Ek d3 k ηC d(k. . to find ψ t (−˜) that is a solution of the Dirac equation. i. (i˜ µ∗ ∂µ − M ) ψ(−˜) = 0. x (8. Norm conservation requires Top to be anti-unitary2 .e. x µ (iγ ∂µ − M ) γ5 Cψ(−˜) = 0.14) (8. u(k. it is antilinear1 . writing −1 ψ(x) −→ Top ψ(x)Top = ηT Aψ(−˜) ≡ ψ t (−˜). ((iγ µ )∗ ∂µ − M ) ψ(x) = 0.17) 3 = λ This shows that −1 Cop b(k. The (time-reversed) Dirac equation becomes. λ) −1 Cop = ∗ ηC b(k. x iγ µT ∂µ − M ψ(−˜) = 0. λ)C v T (k. ˜ x −(iγ µ )∗ ∂µ − M ψ(−˜) = 0. λ)C uT (k. m) ¯ C v T (k.22) = Aψ|Aφ ∗ is antilinear if A(λ|φ + µ|ψ ) = λ∗ A|φ + µ∗ A|ψ . As time reversal will transform ’bra’ into ’ket’. as discussed in section 4).20) The time reversal operator transforms We will again consider the transformation properties for a fermion field ψ(x). The action on the spinors (using C = iγ 2 γ 0 = −iρ1 σ 2 in standard representation) gives C uT (k. m). x 1A 2 An −iC −1 γ µ C∂µ − M ψ(−˜) = 0. For a quantized field one has −1 ∗ −1 Top fk (x)bk Top = fk (x)Top bk Top . (8. (8. r) −→ −˜µ ≡ −xµ = (−t. One has Aφ|Aψ = φ|ψ ∗ = ψ|A† Aφ = ψ|φ . Top |φ = φt | = (|φt )∗ . λ). (2π)3 2Ek (8.Discrete symmetries 68 the latter being also a solution of the Dirac equation.18) (8. Therefore ψ c (x) = = λ −1 ¯ Cop ψ(x) Cop = ηC C ψ T (x) (8. x x (8. we start with the complex conjugated x Dirac equation for ψ.

The result is = = λ (8.4 Bi-linear combinations In quantities such as currents and lagrangians often bilinear combinations of spinor fields are encountered. −λ |¯. −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. λ)u∗ (k. λ).30) (8.24) (8. λ)iγ5 Cu(k. −p.33) (8. −λ a T a. op T (8. λ). λ)u∗ (k.26) d k x x ηT b(k. λ).23) In the standard representation iγ5 C = σ2 and it is straightforward to apply this to the explicit field operator ψ(x) using i γ5 C u(k. p. λ) eik·x + d† (k. ˜ Top d(k. there are 16 independent of these bilinear combinations. λ) T −1 = η ∗ d(k. λ) eik·x + d† (k. ˜ = 0|b(k)|A 8. λ)u∗ (k. λ) i γ5 C v(k.2: The transformation properties of physical states for particles (a) and antiparticles (¯). λ| a. ∗ ˜ v (k.28) (8. λ a |a. λ)v ∗ (k. p. a state |a. λ) ψ t (x) −1 Top ψ(x) Top = i ηT γ5 Cψ(−˜) x = = ˜ u∗ (k. λ) Top = ηT b(k. λ) e−ik·˜ (2π)3 2Ek d3 k ˜ ˜ ˜ ˜ ηT b(k. λ) e−ik·x 3 2E (2π) k ˜ d3 k ˜ ˜ ˜ ˜ ηT b(k. Therefore one has A|k = A|b† (k)|0 = A|T † T b† (k) T † T |0 ∗ = At |b† (k)|0 ∗ ˜ t = k|At . p. λ) eik·˜ + d† (k. λ).29) In table 2 the behavior of particle states under the various transformations has been summarized. λ Therefore γ5 Cψ(−˜) is again a solution of the (ordinary) Dirac equation and we can choose (phase is x convention) ψ t (x) = i γ5 C ψ(−˜). λ)v ∗ (k. p. λ)v ∗ (k.Discrete symmetries 69 Table 8. (8. −p. λ)iγ5 Cv(k. λ) eik·x + d† (k.34) . λ) e−ik·x (2π)3 2Ek (8.25) (check this for the standard representation).27) 3 = λ = λ = λ From this one obtains −1 ˜ Top b(k. Since there are 16 independent 4 × 4 matrices. λ| ¯ C |¯. λ a P |a. λ) e−ik·x 3 2E (2π) k d3 k ˜ ˜ ηT b(k.32) (8. x (8. Note that applying an anti-unitary transformation such as Top one must take for the matrix element ˜ the complex conjugate. λ |¯.31) (8.

g. s′ |V µ (x)|p. As an example consider the vector current for a point fermion. of which the expectation value between momentum eigenstates can be simply found. and T. 2M (8. In many applications the expectation values of currents are needed. F.38) (8.. ¯ ′ ′ (8. (8.37) (8.40) where the coefficients are functions of invariants constructed out of the momenta. e. s . ¯ V µ (x) =: ψ(x)γ µ ψ(x) :. Note that the lagrangian density L (x) → L (−x) under Θ. This implies p′ |V µ (x)|p = e−i (p−p )·x p′ |V µ (0)|p . F. γ ν ]. V µ (x) = ei Pop ·x V µ (0)e−i Pop ·x . for the vector current V µ (x). p) = γµ F1 (q 2 ) + iσµν q ν F2 (q 2 ).41) 2M 2M .. p)u(p) e−i (p−p )·x . As the coupling of the photon field to fermions is given by an interaction term in the ¯ lagrangian of the form : ψ(x)γ µ ψ(x) : Aµ (x) and behaves as a scalar one sees immediately that the ¯ photon field Aµ (x) behaves in the same way as the vector combination ψ(x)γ µ ψ(x). ¯ In general the expectation value between momentum states can be more complicated. C. p′ |V µ (0)|p = u(p′ )γ µ u(p). (q 2 ). γ µ . For instance for nucleons one has Γµ (p′ . γ5 γ µ or γ5 ). 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 .39) where Γµ (p′ . (q 2 ).36) (8. Applying the results from the previous sections it is straightforward to determine the behavior of the combinations under P. 8. as well as under the combined operation Θ = PCT (see Table 3). (8.5 Form factors Currents play an important role in field theory. p′ |V µ (x)|p = u(p′ )Γµ (p′ . C. σ µν . p) can be built from any combination of Dirac matrices (1. 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 ). momenta (pµ or p′µ ) or constant tensors (ǫµνρσ or gµν ).35) The x-dependence can be accounted for straightforwardly using translation invariance. called form factors.Discrete symmetries 70 Table 8. The 16 combinations of Dirac matrices appearing above are linearly independent.3: The behavior of the independent bilinear combinations of fermi fields under P. in this case only q 2 = (p′ − p)2 .

42) where q = p′ − p. or relations based on hermiticity of the operator or P . ¯ 2M (8.43) = ei q·x u(p′ )γ0 Γµ (˜′ . p ˜ • Time reversal p′ |V µ (x)|p = ei q·x u(p′ )Γµ (p′ . γ5 σµν = (1/2)ǫµνρσ γ ρ γ σ . p)γ0 .45) Exercises Exercise 8. or relations that follow from the equations of motion. p′ )u(p′ ) = u† (p′ )Γµ† (p.g.44) = p′ |Vµ (−˜)|˜ ˜ x p ∗ ˜x = ei q ·˜ [¯(˜′ )]∗ Γ∗ (˜′ . p)u(p) ¯ † † = p′ |Pop Pop V µ (x)Pop Pop |p ˜x = p′ |Vµ (˜)|˜ = ei q ·˜ u(˜′ )Γµ (˜′ . p)u∗ (˜) up p µ p ˜ = ei q·x u(p′ ) (iγ5 C)Γ∗ (˜′ . p) = (i γ5 C)Γ∗ (˜′ . p)u(˜) ˜ x p ¯ p p ˜ p (8. p)(iγ5 C)u(p). ¯ p ˜ Therefore parity invariance implies for the structure of Γµ (p′ . C and T invariance. p)(i γ5 C). p) = γ0 Γµ (˜′ . e. ¯ µ p ˜ Therefore time reversal invariance implies for the structure of Γµ (p′ . p′ )γ0 u(p).1 The matrix element of the electromagnetic current between nucleon states is written as < p′ |Jµ (x)|p >= e−i(p−p )·x u(p′ )Γµ u(p) ¯ Here Γµ iσµν q ν F2 (q 2 ) 2M pµ + pµ ′ = γµ H1 (q 2 ) − H2 (q 2 ) 2M = γµ F1 (q 2 ) + ′ . e. p)u(p) ¯ † † = p′ |Top Top V µ (x)Top Top |p ∗ (8. p′ )u(p′ )) u = ∗ † u† (p)γ0 Γµ (p. • Parity p′ |V µ (x)|p = ei q·x u(p′ )Γµ (p′ . p)γ0 u(p). p) that Γµ (p′ . p′ )γ0 = Γµ (p′ . p)u(p) ¯ = p|V µ (0)|p′ ∗ = (¯(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. µ p ˜ (8.g. p) that γ0 Γµ† (p. • Hermiticity p′ |V µ (0)|p = p′ |V µ† (0)|p = u(p′ )Γµ (p′ . p). Therefore hermiticity implies for the structure of Γµ (p′ . p) that Γµ (p′ .

p = (E. GE GM = F1 − Q2 F2 . L int = ejµ Aµ the charge is given by e GE (0) and the magnetic moment by e GM (0)/2M . 0. 0. (b) Show that such a term is also not allowed by time-reversal symmetry.2 (a) Calculate the current matrix element (using the explicit fermion spinors) in the Breit-frame in which q = (0. σµν q ν F3 (q 2 ) 2M . |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 . |q|). 0. Exercise 8. it would correspond to an electric dipole moment d = −e F3 (0)/2M (Interaction term d σ · E). 0. 4M 2 = F1 + F2 . p′ = (E. 0. Exercise 8. −|q|/2).Discrete symmetries 72 (a) Use the Dirac equation to prove the Gordon-decomposition u(p′)γ µ u(p) = ¯ 1 u(p′ )[(p′ + p)µ + iσ µν (p′ − p)ν ]u(p) ¯ 2M (b) Give the relation between Hi en Fi . (c) Show that if a term of this form would exist. 0. (b) Show that in the interaction with the electromagnetic field.3 (a) Show that because of parity conservation no term of the form γ5 can appear in the current matrix element. (c) Show that hermiticity of the current requires that F1 en F2 are real.

t . and the operators are time-dependent. t ′ QH (t′ )|q ′ . t ≡ q|q. in which the states are time-independent. t0 ) = e−i(t−t0 )H or i Two situations can be distinguished: (i) Schr¨dinger picture. t0 ) ∂t (9. (9.1 Time evolution as path integral U (t.1) The time evolution from t0 → t of a quantum mechanical system is generated by the Hamiltonian.6) Of these the Heisenberg picture is most appropriate for quantum field theory since the field operators do depend on the position and one would like to have position and time on the same footing. t0 ) = H U (t. QH (t)|q. t0 ) AH (t0 ) U (t. t ′ |q . (9. i ∂ |ψH ∂t ∂ i AH ∂t ≡ = 0.4) ∂ U (t.5) (9. in which the operators are time-independent.2) (ii) Heisenberg picture. [AH . ≡ 0. t′ ≡ q ′ |q ′ .3) (9. t0 )|ψS (t0 ) . AS (t) = AS and the states o are time dependent.Chapter 9 Path integrals and quantum mechanics 9. H]. Consider the two (time-independent) Heisenberg states: |q. AH (t) = U −1 (t. |ψS (t) = U (t. (9. t0 ). (9. i ∂ |ψS ∂t ∂ i AS ∂t = H |ψS . t′ . |ψH (t) = |ψH .7) 73 .

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

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

21) again a multidimensional integral. A useful property of functional integration is the translation invariance. y) = ∗ m. with Dα F [α] = n dαn N F (αn ) (9.n dx fm (x)fn (x) = n = n dαn 1 exp − α2 N 2 n √ 2π ≡ 1. (9.26) As an important application of translation invariance we mention the identities 1 Dα exp − α2 − ωα 2 = exp 1 2 ω .28) Dα Dα∗ exp (−α∗ α − α∗ ω − ω ∗ α) = exp (ω ∗ ω) .24) (9. In both cases N is an appropriately choosen normalization constant in such a way that the integral is finite. Consider the gaussian functional as an example.Path integrals and quantum mechanics 76 (ii) Write α in terms of a sum of orthonormal basis functions.25) (see Exercises). (9. Note that procedure (i) is an example of the more general procedure under (ii).n fm (x) Kmn fn (y) and consider F [α] = F (αn ) as a multivariable function. Using the second method and expanding in a basis set of functions one obtains for the (real) gaussian functional 1 Dα exp − α2 2 = n dαn N exp − 1 αm αn 2 m. N (9. α(x) = n αn fn and K(x. . the following integrals can be derived for a real symmetric or complex hermitean kernel K. 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) (9.23) Having defined the Gaussian integral.22) = x The last equality is obtained by defining the right measure (normalization N ) in the integration. 2 (9.27) (9. Physical answers will usually come out as the ratio of two functional integrals and are thus independent of the chosen measure. 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.

θη = −ηθ. θ∗ ] = exp (−θ∗ K θ) ≡ exp − Integration for Grassmann variables is defined as dθ 1 = 0 and dθ θ = 1. e.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. dx dy θ∗ (x) K(x. δα(x) (9. y) θ(y) . δα(x) δα(x) δ αKβ = dy K(x. products of the same pair appear. αn = δmn . −→ fn (x) δα(x) ∂αn ∂αm n Note that δ(x − y) = n ∗ fn (x) fn (y). θ2 = 0. In the expansion of the exponential (from second to third line). θ∗ ] = exp (−θ∗ θ) ≡ exp − F [θ. F [θ. (9. Examples of functional differentiation are (9. we need to consider Grassmann-valued functions involving anticommuting Grassmann variables. −→ δα(x) ∆x ∂αx ∂αx δ ∂ ∂ ∗ with .33) (9. i. n θn fn (x) dx θ∗ (x) θ(x) .29) (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. however. (9.34) (9. δα(x) or in discretized form δ 1 ∂ ∂ with .g. α(y) = δ(x − y). the Gaussian-type functionals. y) β(y) δα(x) δ 1 1 exp − α2 = −α(x) exp − α2 . αy = δxy . which vanish.38) .Path integrals and quantum mechanics 77 Functional differentiation is defined as δ . In principle the definitions of functionals is the same.35) For applications to fermion fields.30) (9. It is now easy to check that Dθ∗ Dθ exp (−θ∗ K θ) = det K. (9.31) δ δ α = dy α(y) = dy δ(x − y) = 1.e.32) (9. δα(x) 2 2 δ exp (−α β) = −β(x) exp (−α β) .

Path integrals and quantum mechanics

78

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

= δ(x − y),

(9.39)

(9.40) (9.41)

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

9.3

Time ordered products of operators and path integrals

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

q ′ , t′ |q, t =

Dq exp i
t

dτ L(q, q) , ˙

(9.44)

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

Dq exp i
t i 2

dτ Leff (q, q) , ˙

(9.45)

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

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

δ(0) ln f (q).

dqi q ′ , t′ |qn , tn . . . (9.46)

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

dτ [pq − H(p, q)] ˙

(9.47) (9.48)

Dq q(s) exp i
t

dτ L(q, q) . ˙

Path integrals and quantum mechanics

79

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

(9.49)

dτ L(q, q) . ˙

(9.50)

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

9.4

An application: time-dependent perturbation theory

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

(9.51) (9.52) (9.53)

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

= HI (t) ψI (t), = [AI , H0 ],

(9.54) (9.55)

(9.56) (9.57)

dτ HI (τ ) U (τ, t),
t

(9.58)

U (n) (t′ , t),

(9.59)

1−i (−i)

t t′ t

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

Path integrals and quantum mechanics
t′ t t′ t

80

U

(2)

(t , t) = 1 − i = (−i) . . .

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

t′

U

(n)

(t , t) = (−i)
t

dτ HI (τ ) U (n−1) (τ, t)
t′ τ1 τn−1

= (−i)n
t

dτ1
t t′ t′

dτ2 . . .
t t′

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

=

(−i)n n!

dτ1
t t

dτ2 . . .

(9.60)

The last equality is illustrated for the second term U (2) in the following. The integration
t′ τ1

dτ1
t t

dτ2 HI (τ1 ) HI (τ2 )

can also be performed by first integrating over τ2 but changing the integration limits (check!). Thus we can write the integration as the sum of the two expressions (multiplying with 1/2), = 1 2
t′ τ1

dτ1
t t

dτ2 HI (τ1 ) HI (τ2 )
t′ t′

+

1 2

dτ2
t τ2

dτ1 HI (τ1 ) HI (τ2 ),

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

dτ1
t t

dτ2 HI (τ1 ) HI (τ2 )θ(τ1 − τ2 )
t′

+

1 2

t′

dτ2
t t

dτ1 HI (τ1 ) HI (τ2 )θ(τ1 − τ2 ),

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

dτ1
t t

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

the desired result.

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

dτ1
t t′ t

dτ2 . . .
t

dτn T HI (τ1 ) . . . HI (τn ) (9.61)

≡ T exp −i

dτ HI (τ ) .
t

t = = ′ ′ V = q . T = n J t′ = = Dq exp i t dτ [L(q. Consider. ˙ J (9. L(q. a set |n of physical eigenstates. q . making the electron) and the absorption of an electron (think of a detector).1: Physical picture of vacuum to vacuum amplitude in presence of a source term This is not a surprising result. q) + J(τ )q] . q) + J(t) · q. T ′ |q ′ . ˙ The combination of exponentials in the last step is allowed because we are simply dealing with classical quantities (not operators!). i. t Q′ . t = ei Ht |q .66) .s. harmonic oscillator.65) J dq ′ dq Q′ . the creation of an electron (think of a radio-tube. φn (q. which in turn are embedded between an early time T and a future time T ′ . 9.Path integrals and quantum mechanics source experiment detector 81 T L t L+Jq t’ L T’ Figure 9. it is possible to switch on an interaction. furthermore. T . t |q. t t′ V =0 Dq D dτ HI (q) exp i t t′ t dτ [pq − H0 (p. t′ q ′ . t|Q. n (9. ˙ ˙ physically pictured as. T < t < t′ < T ′ . T with q.64) (9. The Heisenberg state |q. i. t|n n|Q. t|n = q|n e−iEn t with q|n the time-independent wave function. We can again rewrite the time-ordered products as functional integrals as derived in the previous section. t is related to the Schr¨dinger state |q by |q. Considering the source to be present between times t and t′ . q)] ˙ (9. T = n q.63) for plane waves. q)] . for example x|n x|n = φn (x) = ei kn x = e −ω 2 x2 /2 (9. one has q ′ . t q.62) p Dq D exp i 2π t dτ [pq − H(p. say. t′ |q. t |T exp −i p exp −i 2π t′ ′ ′ t′ t dτ HI (Q(τ )) |q. for the physical states o q. t|Q. T ′ |Q. Before and after these processes there is only the vacuum or ground-state |0 . for g.e. q) → L(q.e.5 The generating functional for time ordered products By introducing a source-term. t′ |q. t) φ∗ (Q) e+i En T .

(9.2: Analytic continuation of boundaries T and T ′ We can project out the groundstate by an analytic continuation in the time. = (i)n 0|T Q(t1 ) .Path integrals and quantum mechanics t 82 T t t’ T’ Figure 9.71) 9. . . eiE0 T φ∗ (Q) 0 We define the generating functional Z[J] as Z[J] = Q′ . t′ ) q ′ .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. dq dτ + J(τ )q(τ ) . T ′ |Q. T′ Z[J] = lim T ′ → −i∞ T → i∞ Dq exp i T dτ L q. t′ |q.6 Euclidean formulation The above expression (Eq. Since we have (neglecting multiplicative factors). t 0 0out |0in J . 9. t). ˜ t′ → ∞. such that t′ → −i∞ t → i∞ ↔ ↔ ˜ t → −∞. The importance of Z[J] is that the time ordered product of operators can be obtained from it. t|Q.68) (9. δJ(tn ) hence the name generating functional.70) one immediately finds δ n Z[J] δJ(t1 ) . . T = φ0 (q. in which case eiEn T → e−En ·∞ and the term eiE0 T is dominant. . T J lim −i E0 (T ′ −T ) φ (Q′ ) φ∗ (Q) 0 0 T ′ → −i∞ e T → i∞ dq ′ dq φ∗ (q ′ . Better is the use ˜ of imaginary time t = −it.70) for the generating functional is in fact ill-defined. The generating functional precisely represents the vacuum to vacuum amplitude from initial (’in’) to final(’out’) situation in the presence of a source. Q(tn )|0 . T → i ∞ and T ′ → −i ∞. .67) = = φ0 (q. thus T →i∞ lim q. J=0 (9. J (9.

when L q. dq dt = 1 2 1 2 dq dt 2 Dq exp Dq exp − ∞ d˜ L q. . for Grassmann valued functions. ensuring convergence for the functional integral ZE [J]. δJ(tn ) = (i)n J=0 1 δ n ZE [J] . δJ(tn ) J = 0 ˜ t = it (9. 1 = √ . ZE [J] = = where LE q.1 Convince yourself by choosing an appropriate (orthonormal) set of functions that 1 Dα exp − α K α 2 for real-valued functions.n fm (x) Kmn fn (y).and Grassmann-valued functions. (9. ˜ dt (9. det K Exercise 9. (9.2 Check the examples of functional derivation for real-.74) − V (q) 2 (9. δ n Z[J] 1 Z[J] δJ(t1 ) . e.Path integrals and quantum mechanics 83 In terms of the imaginary time we can write the Euclidean generating functional.75) = − LE q. As discussed in the previous section the interesting quantities are obtained from functional differentiation with respect to the sources. y) = ∗ m. i dq .76) which is a positive definite quantity. .g. i τ −∞ ∞ −∞ dq d˜ τ + J(˜)q(˜) τ τ dq d˜ τ − J(˜)q(˜) τ τ . Dα Dα∗ exp (−α∗ K α) = for complex-valued functions and Dθ∗ Dθ exp (−θ∗ K θ) = det K. In all of these cases the determinant of the (hermitean) operator K is defined as det K ≡ p Kp . where Kp are the eigenvalues of the matrix Kmn in K(x. τ dq ˜ dt ≡ −L q. . + V (q).77) where the expressions have been divided by Z[J] and ZE [J] in order to get rid of dependence on multiplicative factors. dq ˜ dt = 1 2 dq ˜ dt dq ˜ dt 2 − V (q). Exercises Exercise 9.73) d˜ LE q. complex. . det K 1 . . ˜ ˜ ZE [J] δJ(t1 ) . The differentiations in Z[J] and ZE [J] are related.72) (9.

z) = δ(x − z).78) (9. .79) (9.3 Prove the following relations 1 Dα exp − α K α − αω 2 1 = √ exp det K 1 ω K −1 ω . y) K −1 (y.80) Dα∗ Dα exp (−α∗ K α − α∗ ω − ω ∗ α) = 1 exp ω ∗ K −1 ω .Path integrals and quantum mechanics 84 Exercise 9. where dy K(x. det K Dθ∗ Dθ exp (−θ∗ K θ − θ∗ η − η ∗ θ) = det K exp η ∗ K −1 η . 2 (9.

. 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 ) .2) = ∞ −∞ d4 xE [LE (φ. ∂µ φ) − Jφ] (10. Π) (10. t = = Dφ DΠ exp i t σ′ ˙ dτ d3 x Π(x)φ(x) − H (φ. x′ . .e. .3) The Euclidean formulation is as before. . i. δJ(xn ) 85 ≡ G(n) (x1 . φ(x. t′ |φ. x. . (10.4) J=0 .1 Generating functionals for free scalar fields The generating functional for quantum fields is a generalization of the results in the previous section to a system with more degrees of freedom.Chapter 10 Feynman diagrams for scattering amplitudes 10. . . ∂µ φ) + Jφ] (10. xn ). t) is considered as a set of quantum operators φx (t) in the Heisenberg picture and t′ φ′ . . φ(xn )|0 = (−i)n δ n Z[J] δJ(x1 ) .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.

. . d4 xn G(n) (x1 . 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. J(xn ). As discussed in the previous section (exercise 9. xn ) = x2 xn . .10) J=0 In order to determine ∆F . ∆F (the socalled Feynman propagator) is the Green’s function of the Klein-Gordon equation. with Z0 [0] = 1.e.12) d4 k −i k·x ˜ e ∆(k) (2π)4 (10. = 0 2 2 2 k2 − M 2 (k ) − E 2 (k ) − k − M (10.4) one distinguishes the retarded Green’s function ∆R (x) = d4 k e−i k·x . .3) one finds 1 Z0 [J] = exp − i 2 d4 x d4 y J(x) ∆F (x − y) J(y) ≡ exp (i W0 [J]) . . (10. . (10. These Green functions appear in the expansion Z[J] = n in n! d4 x1 . . xn ) J(x1 ) . Furthermore. . (10.13) . .9) i. .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). consider the general solution of ∂µ ∂ µ + M 2 ∆(x) = −δ 4 (x).8)  1 d4 x  φ (∂µ ∂ µ + M 2 ) φ − Jφ 2 K  (10. (2π)4 (k 0 + iǫ)2 − E 2 (10.Feynman diagrams for scattering amplitudes 86 for which we introduce the picture x1 G(n) (x1 .11) Depending on the path choosen in the complex k 0 plane (see exercise 7.7) where we made the choice Z0 [0] = 1 and ∂µ ∂ µ + M 2 ∆F (x − y) = −δ 4 (x − y). which can be written as ∆(x) = ˜ with (k 2 − M 2 )∆(k) = 1 or ˜ ∆(k) = 1 1 1 = 0 2 . . 2 2 → (10. .5) Z0 [J] for the free scalar field For the (free) scalar field lagrangian L = ← 1 1 ∂µ φ∂ µ φ − M 2 φ2 .

φ+ (0)] = (2π)3 2E are also shown in exercise (7. 10.19) φ ∂E ∂E − M 2 φ + Jφ 2 1 d4 xE d4 yE J(x) (−i ∆F (x − y)) J(y) .21) ∆F (x) = = −i 0 kE =−∞ d4 kE e−i kE ·xE 2 (2π)4 kE + M 2 (10. i. in the (well-defined) Euclidean formulation starting with LE = 2 (∂E φ)2 + 2 M 2 φ2 µ 2 2 Jφ with the equation of motion (∂E ) − M φ(x) = −J(x). In order to see which is the appropriate Green’s function to be used in the generating functional Z0 [J] we will take two routes. φ(0)] = i∆+ (x) + i∆− (x) d3 k e−i k·x = [φ+ (x).18) Thus we see various solutions.1 for countours) µ (∂E )2 − M 2 (i ∆F (x − y)) = −δ 4 (x − y) 0 kE =∞ (10. t) = 0 for t > 0.Feynman diagrams for scattering amplitudes 87 satisfying ∆R (x. the causal Green’s function ∆C (x) = and finally the Green’s function d4 k e−i k·x = 4 (k 0 )2 − E 2 + iǫ (2π) ∆(x) = PV d4 k e−i k·x .16) C where C is a closed contour in the complex k 0 -plane.e. (10.15) e−i k·x d4 k . the second is to explicitly consider 0|T φ(x1 ) φ(x2 )|0 . the generating functional can be written 1 µ µ Z0 [J] = Dφ exp d4 xE (10. (2π)4 (k 0 − iǫ)2 − E 2 (10.23) k0 =i∞ k0 =−i∞ d4 k e−i k·x . µ 1 1 • Firstly. . (2π)4 k 2 − M 2 where the latter contour can be deformed to the contour for ∆C .4). ∆(x) = − d4 k e−i k·x . φ− (0)] = (2π)3 2E d3 k e−i k·x = [φ− (x). (2π)4 k 2 − M 2 (10. 4 k 2 − M 2 + iǫ (2π) (10. the advanced Green’s function ∆A (x) = satisfying ∆A (x.14) They are solutions of the inhomogeneous equation. t) = 0 for t < 0. The first possibility is to consider the well-defined Euclidean formulation.22) (10.20) = exp − 2 where or (see fig.17) d4 k e−i k·x . It is straightforward (closing contours in the appropriate half of the complex plane) to prove for instance that i ∆C (x) = θ(x0 ) i ∆+ (x) − θ(−x0 ) i ∆− (x). ∆F = ∆C . (10. (2π)4 k 2 − M 2 (10. The solution of the homogeneous equation ∂µ ∂ µ + M 2 ∆(x) = 0 can also be written as an integral in k-space. The contours for i ∆(x) i ∆+ (x) i ∆− (x) = [φ(x).

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

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. (10. Consider as an example the interaction LI (φ) = − g 4 φ 4! in the scalar field theory discussed sofar. x4 ) = G0 (x1 .32) (10. i. x4 ) = To first order in g one has Z (g ) [J] = = 1 0 0 x1 + + x2 (10. with propagators (i ∆F ) being connected to vertices (i LI ) according to the above expression for the functional Z[J].g ) (x1 . x2 . x2 ) = i ∆F (x1 − x2 ) = G(4. (10.27) When interactions are present.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 . x2 ) = G0 (x1 .g 0 ) = G0 = 1. x3 .Feynman diagrams for scattering amplitudes 89 10. ≡ −i g d4 z (10.26) 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.35) z . Z[J] = = Dφ exp i exp i d4 z LI (10.29) This expression will be the one from which Feynman rules will be derived. L (φ) = L0 (φ) + LI (φ). (10. x3 . (2) (0) (10.34) Introducing a vertex point to which four propagators are connected.e.31) G(2.28) 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 . To zeroth order in the coupling one has 0 1 Z (g ) [J] = Z0 [J] = exp − i 2 J ∆F J .30) and in zeroth order G(0.g ) (x1 . x2 .

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

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

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

a)SU −1 (Λ.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. a causality condition that can be proven to imply the absence of interactions.50) = φ(x) = √ Zφin (x) − √ Zφout (x) − d4 y d4 z ∆R (x − y) K(y.e. as it stands. φ(y)] = Z [φin (x). β. (10. in|Sφout Finally we check that as expected S does not spoil Poincar´ invariance.g. implies the strong (operator) convergence φ(x) → Zφin (x). e Proof: φin (Λx + a) = U φin (x)U −1 = U Sφout (x)S −1 U −1 = U S U −1 U φout (x) U −1 U S −1 U −1 = U SU −1 φout (Λx + a) U S −1 U −1 ⇔ U SU −1 = S. The time evolution operator (see Eq. φin (y)] = iZ ∆(x − y). √ Although the above. in|φin S → φin S = S φout (x).49) The field φ can be expressed in φin/out using retarded or advanced Green’s functions.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 ). φin (x) = d3 k ∗ a(k) fk (x) + a† (k) fk (x) . Proof: β. S is invariant under e Poincar´ transformations: U (Λ. 2 ∂µ ∂ µ + M0 φ(x) = J(x). The relation between S and Z[J] To establish this relation. i. e.62) is then T′ UT [J] T′ = T exp i T d4 x J(x)φ(x) . which in a later stage (renormalization) will become more important. The relation between S-matrix and in. The convergence therefore must be weakened to t→−∞ √ (10. a) = S. this can actually not be used as it would imply [φ(x). (2π)3 2E (10. the source term Jφ in the lagrangian density is treated in the interaction picture. (10.51) (10. z) φ(z) d4 y d4 z ∆A (x − y) K(y.54) . (10. 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) (10.53) 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). 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. z) φ(z). out|φout = β. 9.52) where Z is a constant.and out-fields is: φin (x) = S φout (x)S −1 .

63) . Thus applied to the field φ(x) = φ+ (x) + φ− (x). B] and [A. B + C]e e .Feynman diagrams for scattering amplitudes 94 and satisfies the property T ′ δUT [J] t = i UtT [J] φ(x)UT [J]. y) δ δJ(y) : F [J]. φ(x). δJ(z) (10. z) δU [J] .e. e e ] = [A.56) Since δU [J]/δJ(x) satisfies the same equations as φ(x) and we know the limits. [A. δJ(z) Taking the difference between the two expressions √ Z (φout (x) U [J] − U [J] φin (x)) =i or d4 y d4 z (∆R (x − y) − ∆A (x − y)) K(y. z) δU [J] δJ(z) δU [J] d4 y d4 z ∆A (x − y) K(y. provided [A.57) i [φin (x). Noting that 0| : eA : |0 = 1 it follows that F [J] = 0|S U [J]|0 = 0|U [J]|0 = 0out |0in ∝ Z[J]. we can. : =i d4 y ∆(x − y) f (y) : e d4 z φ(z) f (z) :.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). (10. one has [A. z) d4 y d4 z ∆(x − y)K(y. φ(y)] f (y) e e . e = i. B] (for the case that [A. (10.59) (10. t i Ut∞ [J] φ(x)U−∞ [J] (10.58) In order to find a solution to this equation note that. φ− f (10. C] are c-numbers. φ(x). Thus S U [J] =: exp 1 √ Z d4 x d4 y φin (x) K(x. : e d4 z φ(z) f (z) d4 x φ− (x) f (x) B C B C (10.61) e φ+ f where the normal ordered expression : e φ f : is used. eB ] = [A. just as we did for φ(x). B]eB . 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. express it in terms of advanced and retarded Green’s functions. z) .62) where F [J] is some (arbitrary) functional. δJ(x) ∞ or for U [J] ≡ U−∞ [J] ′ (10. B] is a c-number). δJ(z) (10. with the use of the Baker-Campbell-Hausdorff formula e−B A e+B = A + [A. S U [J]] = √ Z δ S U [J]. which is equal to the expression e in which the creation operators are placed left of annihilation operators.60) e d4 y φ+ (y) f (y) d4 z φ− (z) f (z) d4 z φ+ (z) f (z) d4 y [φ(x).

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

Veltman. = −i g . Corresponding to external particles wave functions are introduced k k 1 1 In order to calculate the connected amplitude −i Mf i appearing in the S-matrix element these ingredients are combined using Eqs 10.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. ’t Hooft and M.66 and 10. or calculating full Green’s functions external lines are treated as propagators. At these vertices each line can be assigned a momentum. to be precise iLI vertices in momentum space are introduced. for example up to order g 2 the scattering of two neutral spin 0 particles (real scalar field) is described by + + + (Rule 2) The contribution of each diagram is obtained by multiplying the contributions from propagators. vertices and external particle wave functions in that diagram. which cancels the factor 1/2 in the quadratic piece. = −i g (multiplied with 4! corresponding to the allowed number of permutations of identical particles). For the interaction terms in the lagrangian. but overall momentum conservation at a vertex is understood. 3! 4! The vertices are: = −i f Consider first the lowest order self-energy diagram. Diagrammar) in the case of the interaction terms g f (10. which is summarized in the following rules: (Rule 1) Start with external legs (incoming particles/outgoing particles) and draw all possible topologically different connected diagrams. 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. Note that in calculating amputated Green’s functions external lines are neglected. (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).69) LI (φ) = − φ3 − φ4 . the factor 2 corresponds to the two-points Green’s function having two identical ends).27.27.

External line 1 can be attached in six. 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. (10. 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 . Divide by the permutational factors of the vertices. which have been included in the definition of vertices (here 3! for each vertex). there is no combinatorial factor like in the scalar case. .e. The generating functional in the interacting case can be written as Z[J. after that line 2 in three ways. Finally divide by the number of permutation of the points that have identical vertices (here 2!). Then there are two ways to connect the remaining lines such that the desired diagram results. ∗ (z) i δJ(z) i δJ d4 x d4 y J ∗ (x) i∆F (x − y) J(y) .70) = exp i exp − In Feynman diagrams the propagator is still given by i∆F (k).Feynman diagrams for scattering amplitudes 97 Draw two points corresponding to the two vertices and draw in each of these points the lines coming out of the vertices: Now count in how many ways the lines can be connected with the same topological result. S 3! 3! 2! 2 As a second example consider the diagram 1 2 There are three vertices. i. the result is 1 6×4×6×3×2 1 = = . S 3! 3! 4! 2! 2 Complex scalar fields The case of complex scalar fields can be considered as two independent fields. Dividing by vertex factors and permutations of identical vertices. but it connects a source with its complex conjugate and therefore is oriented. denoted k = i k 2 − M 2 + iǫ Note that in this case the propagator does not have identical ends. The total result is 6×3×2 1 1 = = .

. which is the solution of (i/ − M )SF (x) = −δ 4 (x) ∂ (i. namely (Rule 5) Feynman diagrams which only differ by exchanging identical fermions in initial or final state have a relative minus sign. (Rule 6) Each closed fermion loop gets a sign −1. in e− e− → e− e− scattering (Møller scattering) the lowest order contribution is (see next section for e-e-γ vertex). 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. minus the inverse of the operator appearing in the quadratic piece) and is given by iSF (x − y) = 0|T ψ(x)ψ(y)|0 = = (i/x + M ) i∆F (x − y) ∂ d4 p −i p·x p /+M = e . − g2 2 dz dz ′ δ2 δ2 δη(z) δη(z) δη(z ′ ) δη(z ′ ) and the quadratic term in Z0 [η.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. (10. but the anticommutating properties of Grassmann variables imply some additional minus sign in Feynman diagrams. η). − 1 2 dx dy dx′ dy ′ η(x)S(x − y)η(y) η(x′ )S(x′ − y ′ )η(y ′ ).g. 4 2 − M 2 + iǫ (2π) p and the (oriented) propagator in Feynman diagrams involving fermions is i p j (10.72) (10. η] = Dψ Dψ exp i d4 z LI exp − d4 x ψ (i/ − M ) ψ + LI (ψ) + ηψ + ψη ∂ 1 δ 1 δ . which arises from the quadratic term in exp i d4 z LI ..71) = exp i where iSF is the Feynman propagator for fermions. e..− i δη(z) i δη(z) d4 x d4 y η(x) iSF (x − y) η(y) .Feynman diagrams for scattering amplitudes 98 Dirac fields For fermions the generating functional is given by Z[η. η]. The two-points Green’s function 0|T φ(x) φ(y)|0 contains a fermionic loop contribution. = + ..

1 k 2 − M 2 + iǫ λ (k 2 − M 2 + iǫ)(k 2 − λ M 2 + iǫ) leads to . 4 2 In addition to the fermion propagator and fermion wave functions discussed in the previous section we have the photon propagator1 µ k ν = −i k2 gµν 1 − 1− + iǫ λ (k 2 kµ kν . writing down the propagators till the beginning of the ¯ line (incoming fermion u or outgoing antifermion v) has been reached. 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 . Vector fields and Quantum Electrodynamics As the most important example of vector fields consider the lagrangian density for quantum electrodynamics (QED). It is necessary to start at the end of a fermion line (with the above arrow convention an outgoing fermion wave function u or an incoming antifermion ¯ wave function v ) and keep on following that line.Feynman diagrams for scattering amplitudes 99 The result is which contains an extra minus sign as compared to a bosonic loop. Particular choices of λ are λ = 1 (Feynman propagator or Feynman gauge) and λ = ∞ (Landau gauge). 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. 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. + 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.74) L = − Fµν F µν − (∂ · A)2 + ψ(i/ − M )ψ − e ψγ µ ψ Aµ . The wave functions for fermions are given by ∝ g 2 S(z − z ′ ) S(z ′ − z) = −g 2 iS(z − z ′ ) iS(z ′ − z). λ 1 ∂ (10. 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 .

s′ ′ ′ = 2 kµ kν + 2 kν kµ + q 2 gµν 1 Tr [(/ ′ + m)γµ (/ + m)γν ] k k 2 ′ ′ = 2 kµ kν + kν kµ − gµν (k · k ′ − m2 ) (10. s′ )γµ u(k. The diagram and momenta and the commonly used invariants (Mandelstam variables) for a 2 → 2 scattering process are 1 k 3 k’ s t u = (k + p)2 = m2 + M 2 + 2 k · p = (k ′ + p′ )2 = m2 + M 2 + 2 k ′ · p′ = (k − k ′ )2 = q 2 = 2 m2 − 2 k · k ′ = (p − p′ )2 = 2 M 2 − 2 p · p′ p 2 p’ 4 s = (k ′ − p)2 = m2 + M 2 − 2 k ′ · p + t+u= m2 = 2 m2 + 2 M 2 i = (k − p′ )2 = m2 + M 2 − 2 k · p′ The scattering amplitude is given by −iM = u(k ′ .s4 e4 (m) µν (M) L L . s1 ) u e2 u(p′ .79) . s4 )(−ie)γ ν u(p.76) 1 . s′ ) ¯ u s.78) = 2 α2 2 s + u2 + 4t(M 2 + m2 ) − 2(M 2 + m2 )2 . where u ¯ L(m) µν = = 1 2 u(k ′ . s3 )(−ie)γ µ u(k. q 4 µν [using that (¯(k ′ )γµ u(k))∗ = u(k)γµ u(k ′ )]. s3 )γ µ u(k.75) Note that the q µ q ν term in the photon propagator are irrelevant because the photon couples to a conserved current. To lowest order (∝ α = e2 /4π) only one diagram contributes. 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 . t2 (10.s2 ) which can be written as (see also chapter 4) |M |2 = = 1 4s |¯(k ′ . s1 ) ¯ −igµν u(p′ . s2 )|2 ¯ q2 (10.5 Some examples eµ scattering The first example is the electromagnetic scattering of an electron and a muon. s2 ). s)¯(k.s2 . s4 )γµ u(p. s)γν u(k ′ .s3 . ¯ q2 (10.s4 ) and averaged over spins in the initial state (1/2 × 1/2 × s1 .Feynman diagrams for scattering amplitudes 100 10.77) Combining Lµν and Lµν (M) one obtains L(m) Lµν (M) = 2 s2 + u2 + 4t(M 2 + m2 ) − 2(M 2 + m2 )2 µν and M 4π 2 (m) (10.

Basically the same diagram is calculated and the result is the same after the interchange of s ↔ t. s3 )γ µ v(p′ .Feynman diagrams for scattering amplitudes 101 The second example is the annihilation of an electron pair and creation of a muon pair.1 (a) Give in diagrammatic notation the full Green functions G(4) (x1 . 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. s1 )γν v(k ′ . for instance Møller scattering (e− e− → e− e− ) and Bhabha scattering (e− e+ → e− e+ ) are related using crossing symmetry.s2 3 . (c) Do the same for the connected Green function Gc . The diagram. s2 )γµ u(k. s2 (10. s2 ) ¯ ¯ 1 2s u(p. s4 )γ ν u(p. s3 ) ¯ ¯ 1 . (b) Use the definition of the source-connected Green function Gsc and show that indeed sourceconnected diagrams survive. 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. . Exercises Exercise 10. To lowest order (∝ α = e2 /4π) only one diagram contributes. This is known as crossing symmetry. s1 ) u(k.s4 = 1 1 e Tr(/ − m)γµ (/ ′ + m)γν k k Tr(/ + M )γ µ (/′ − M )γ ν p p s2 2 2 1 e4 1 ′ ′ = 4 2 kµ kν + kν kµ − gµν s pµ p′ν + pν p′µ − g µν s s 2 2 M 4π 2 4 and = 2 α2 2 t + u2 + 4s(M 2 + m2 ) − 2(M 2 + m2 )2 . s4 ) u e2 v (k ′ .s2 .2.80) Note the similarity in the amplitudes for eµ scattering and e− e+ → µ− µ+ . Similarly. the masses. .s4 = 1 2s × v (k ′ . s1 )|2 ¯ s 1 . s4 ) v (p′ . s3 )γ µ v(p′ . (4) (4) . . s2 )γµ u(k.s3 .

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

.1) = (2π)3 2E (2π)4 (proven in Chapter 2). It is determined by the energy-momentum four vector p = (E. (11. and the total momentum four-vector P = pa + pb . i. (2π)3 2Ei i=1 (11. p) which satisfies p2 = E 2 −p2 = m2 . a a b b the quantity s = P 2 = (pa + pb )2 . pn ) = (2π)4 δ 4 (P − pi ) dR(p1 . . In that case cm pa = (Ea . s is for obvious reasons known as the center of mass (CM) energy. pb we start with the four vectors pa = (Ea . . A physical state has positive energy. pn ). .Chapter 11 Scattering theory 11.e. . the 1particle phase space is d3 p d4 p θ(p0 ) (2π)δ(p2 − m2 ). In the CM frame the threshold value for s obviously is n 2 sthreshold = i=1 mi . .3) which is useful because the total 4-momentum of the final state usually is fixed by overall momentum conservation. 103 . . + pn )2 . pn ) = and the reduced phase space element by dR(s.4) For two particle states |pa . p1 . It is a useful quantity. . i n d3 pi . . . + n. . for instance for determining the threshold energy for the production of a final state 1 + 2 + . s = (p1 + . This is generalized to the multi-particle phase space dR(p1 . Here s is the invariant mass of the n-particle system.5) √ is referred to as the invariant mass squared.2) (11. (11.7) pb = cm (Eb . . . For two particles. pa ) and pb = (Eb . (11. . The first is the CM system. The phase space is determined by the weight factors assigned to each state in the summation or integration over states.1 kinematics in scattering processes Phase space The 1-particle state is denoted |p . To be specific let us consider two frequently used frames. pb ) satisfying p2 = m2 and p2 = m2 . (11. . −q).6) (11. Its square root. q). .

4π s 4π 8π s 4π (11. 2mb (11. trf Ea (threshold) = 1 2 mb ( i mi )2 − m2 − m2 a b . 0). m2 . 1 2 d(E1 + E2 ) √ |q| dΩ(ˆ) q δ( s − E1 − E2 ) 2 (2π) 4(E1 + E2 ) √ q |q| dΩ(ˆ) λ12 dΩ(ˆ) q √ = . = 2 s λ(s. m2 .9) (11. 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. which in the specific case a b also can be expressed as λ(s.14) One can.10) The function λ(s. m2 . Prove that trf Ea = |ptrf | = a s − m2 − m2 a b . m2 . mb and s). a (11. 2mb λ(s. p1 . m2 ). In that case trf pa = (Ea .8) (11.16) . p1 . m2 . 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. (11. m2 ) a b . 2 s s − m2 + m2 a b √ . p2 ) = = where λ12 denotes λ(s. Prove that |q| = cm Ea cm Eb (s − m2 − m2 )2 − 4 m2 m2 a a b b = 4s s + m2 − m2 a b √ = . ptrf ).15) Explicit calculation of the reduced two-body phase space element gives dR(s.Scattering theory 104 It is straightforward to prove that the unknowns in the particular system can be expressed in the invariants (ma . m2 ) a b . for instance. m2 ) = 4(pa · pb )2 − 4p2 p2 . Also in this case one can express the energy and momentum in the invariants. m2 ) is a function symmetric in its three arguments.13) (11.11) (11. 4s (11. 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.12) pb = (mb .

An often used set of invariants are the Mandelstam variables. Writing E = mT cosh y.24) Using the relation between t and cos θcm it is straightforward to express dΩcm in dt. The minimum and maximum values for t correspond to θcm CM = m2 + m2 − 2 Ea Ec ± 2 qq ′ a c = m2 a + m2 c (s + m2 − m2 )(s + m2 − m2 ) a c b d − ± 2s √ λab λcd . where the first is referred to as elastic channel and the set of all other channels as the inelastic channels. however. Consider the 1-particle inclusive case. the rapidity y is defined T T x y y= 1 ln 2 E + pz E − pz = ln E + pz mT = tanh−1 pz .Scattering theory 105 Kinematics of 2 → 2 scattering processes π− + p The simplest scattering process is 2 particles in and 2 particles out. there is usually a preferred direction. Consider the process a + b → c + d. pT = (px .23) which are not independent as s + t + u = m2 + m2 + m2 + m2 . = (11. py ). 11.25) dR(s. tmax min λcd /4s. The variable s is always larger than a c b d the minimal value (ma + mb )2 .22) (11. The initial state. . At high energies.. H1 + H2 → h + X.20) → π0 + n → π+ + π− + n → .19) (11.18) (11. 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 . E . usually many particles are produced. Of course there are not only 2-particle channels. for instance the momenta of incoming (colliding) hadrons and it is useful to use for the produced particle rapidity as a variable. a c with q = λab /4s and q ′ = being 0 or 180 degrees.17) (11. pd ) = 4π s 4π 8π s 4π dt dt √ √ . Examples appear in → π− + p (11. s = (pa + pb )2 = (pc + pd )2 t = (pa − pc ) = (pb − pd ) u = (pa − pd )2 = (pb − pc )2 2 2 (11. A specific reaction channel starts contributing at the threshold value (Eq.. pc . while the final state often arises from a series of 2-particle processes combined with the decay of an intermediate particle (resonance). in which one particle is detected. The various possibilities are referred to as different reaction channels. pz = mT sinh y.27) with m2 = m2 − p2 = m2 + p2 + p2 . usually is a 2-particle state. dt = 2 qq ′ d cos θcm and obtain for the two-body phase space element √ q ′ dΩcm λcd dΩcm √ = (11. Instead of the scattering angle.4).26) = 16π q s 8π λab Kinematics of inclusive hard scattering processes In high energy (hard) scattering processes.21) (11. In inclusive measurements no particles are detected in the final state. in exclusive measurements all particles are detected. (11. 2s (11.

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

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

T V. Microscopically one has Ndecay = Na · Γ T or Γ= N 1 . the number of scattered or decayed particles is given by 2 N = (2π)4 δ 4 (Pi − Pf ) iMf i dR(p1 . From the solution N (t) = N (0) e−Γ t one knows that the decay time τ = 1/Γ. . pn ). N 1 .35) This quantity is not covariant. dt (11. as expected. σ= √ 2 λab T · V (11. the amount of decaying particles is proportional to the number of particles with proportionality factor the em decay width Γ.Scattering theory 108 Although this at first sight does not look covariant. T · V ρa (11. For this we employ in essence Fermi’s Golden rule stating that when the S-matrix element is written as Sf i = δf i − (2π)4 δ 4 (Pi − Pf ) iMf i (11. (11. .T d4 x = V · T (2π)4 δ 4 (Pi − Pf ). N and T · V are covariant.33) Decay of particles For the decay of particle a one has macroscopically dN = −Γ N. .e. (2π)4 δ 4 (Pi − Pf ) 2 = (2π)4 δ 4 (Pi − Pf ) = (2π)4 δ 4 (Pi − Pf ) d4 x ei(Pi −Pf )·x V.39) One of the δ functions can be rewritten as T · V (remember the normalization of plane waves).38) (in which we can calculate −iMf i using Feynman diagrams).36) (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. 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. . .34) i. For the (proper) decay width one thus has 1 N Γ0 = . 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 . (11.

δf n + i(2π)4 δ 4 (Pf − Pn ) (M † )f n or −i Mf i − (M † )f i = − n δni − i(2π)4 δ 4 (Pi − Pn ) Mni = δf i . . 2 λab and for instance for two particles dσ = q ′ M (s. .Scattering theory 109 (Using normalized wave packets these somewhat ill-defined manipulations can be made more rigorous). pn ).42) 2−body decay The differential cross section (final state not integrated over) is given by 1 dσ = √ |Mf i |2 dR(s. . θ)|2 . θ) in quantum mechanics. . for which one has dσ/dΩ = |f (E. pn ) |M |2 q 32π 2 m2 dΩ |M |2 . θ).e. θcm ) √ q 8π s 2 2 (11.46) Partial wave expansion Often it is useful to make a partial wave expansion for the amplitude M (s. . for which the √ amplitudes squared have been calculated in the previous chapter.47) M (s. implies for the scattering matrix M .43) dΩcm = π M (s. . i M† M √ − √ 8π s 8π s = fi n dΩn ∗ Mnf qn Mni √ √ . 16π 2 s (11. q f ). q n ). note the sign and cos θ = qi · qf ).44) This can be used to get the full expression for dσ/dt for eµ and e+ e− scattering. The result is N = |Mf i |2 dR(s. 11. see 11.41) (11. . pn ). ˆ ˆ Inserted in the unitarity condition for M . The amplitude −M /8π s is the one to be compared with the quantum mechanical scattering amplitude f (E. respectively.4 Unitarity condition (S † )f n Sni = δf i The unitarity of the S-matrix. (11.25) −i Mf i − (M † )f i = − ∗ dΩ(ˆn )Mnf (q f . t) λab 4π dt. √ (2ℓ + 1)Mℓ (s) Pℓ (cos θ). i.40) T ·V Combining this with the expressions for the width or the cross section one obtains for the decay width Γ = 1 2m = dR(m2 . . . p1 . The sign difference comes from the (conventional) sign in relation between S and quantummechanical and relativistic scattering amplitude. (M † )f n (2π)4 δ 4 (Pi − Pn ) Mni . . p1 . 8π s 2π 8π s . θ) or M (q i . (11. (11. q n ) q n qn √ Mni (q i . . (11. . (11. p1 .45) Since the amplitudes also depend on all momenta the full result for two-particle intermediate states is (in CM. θ) = −8π s ℓ (in analogy with the expansion for f (E.

(11.e.50) where Sℓ (s) satisfies |Sℓ (s)| = 1 and δℓ (s) is the phase shift.49) or Im Mℓ = q |Mℓ |2 . η = 1) the result σel = 4π q2 (2ℓ + 1) sin2 δℓ .Scattering theory 110 we obtain LHS = −i while for the RHS use is made of ℓ † (2ℓ + 1) Pℓ (ˆi · qf ) (Mℓ )f i − (Mℓ )f i . In general a given channel has |Sℓ (s)| ≤ 1.51) and for the case that this is the only channel (purely elastic scattering. † −i (Mℓ )f i − (Mℓ )f i = 2 † (Mℓ )f n qn (Mℓ )ni . q ˆ Pℓ (ˆ · q ′ ) = q ˆ (ℓ) m 4π Y (ℓ) (ˆ) Ym (ˆ′ ) q (ℓ)∗ q 2ℓ + 1 m and the orthogonality of the Ym functions to prove that RHS = 2 n ℓ † (2ℓ + 1)Pℓ (ˆi · qf ) (Mℓ )f n qn (Mℓ )ni . which allows writing Mℓ (s) = Sℓ (s) − 1 e2i δℓ (s) − 1 = .52) From the imaginary part of M (s.53) The difference is the inelastic cross section. 2i q 2i q (11. q ˆ i. (11. 0). the total cross section can be determined. Show that σT = = 4π q 2π q2 (2ℓ + 1) Im Mℓ (s) ℓ ℓ (2ℓ + 1) (1 − ηℓ cos 2δℓ ). parametrized as Sℓ (s) = ηℓ (s) exp(2i δℓ (s)). σinel = π q2 2 (2ℓ + 1) (1 − ηℓ ). θcm ) √ q 8π s 2 . in combination with the partial wave expansion for the amplitudes M and the orthogonality of the Legendre polynomials immediately gives for the elastic channel.54) ℓ . Using σ= dΩ q ′ M (s.48) If only one channel is present this simplifies to ∗ ∗ −i (Mℓ − Mℓ ) = 2 qMℓ Mℓ . (11. n (11. ℓ (11. (11. σel = 4π ℓ (2ℓ + 1)|Mℓ (s)|2 (2ℓ + 1) ℓ = 4π q2 ηℓ e 2i δℓ − 1 2i 2 .

however.e. We can also invert the situation and use unitarity to find the imaginary part of an amplitude at some order from a lower order calculation. but still real. s − M 2 + iM Γ (11. (q Mℓ )ij (s) = −M Γi Γj . pn )MRn (2π)4 δ 4 (k − Pn ) MnR = 2M n Γn = 2M Γ. . 0). 0)|2 = e−Γt . The quantity Γ is precisely the width for unstable particles. . For an unstable particle one expects that U (t. in which case. The prescription for the pole structure. This is achieved with a propagator i∆R (k) = i k2 − M2 + iM Γ (11. η = 0.55) (note that we have disregarded spin). i. σel = σinel . which is given by a sum of the partial widths into the different channels.56) (again disregarding spin). For instance the tree-level calculation of Feynman diagrams at order e2 was real.5 Unstable particles i k 2 − M 2 + iǫ For a stable particle the propagator is i∆(k) = (11. This is what we will do to discuss decay widths. such that |U (t. 11. This is (somewhat sloppy!) seen by considering the (amputated) 1PI two-point vertex Γ(2) = −i ∆ as the amplitude −iM for scattering a particle into itself (forward!) through the decay channels as intermediate states. i. ∝ U (t. (11. one has poles at k 0 = ±(Ek − iǫ) where Ek = + k2 + M 2 guarantees the correct behavior.Scattering theory 111 Note that the total cross section is maximal in the case of full absorption. We note that unitarity is generally broken in a finite order calculation. . For the amplitude in a scattering process going through a resonance. 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 . 0) ∝ e−i(E−iΓ/2)t .58) .e. specifically one has for t > 0 that the Fourier transform is dk e 0 −ik0 t ∆(k) ∝ e−ik t dk (k 0 − Ek + iǫ)(k 0 + Ek − iǫ) 0 0 (t>0) ∝ e−iEk t . it is straightforward to write down the partial wave amplitude. The unitarity condition then states 2 Im ∆−1 (k) = R n † dR(p1 .57) This shows that Γ is the width of the resonance. To check unitarity we need the amplitude at order e4 . . The product of amplitudes is of order e4 . the time-evolution operator.

where furthermore σinel = 0. This characteristic shape of a resonance is called the Breit-Wigner shape. Exercises Exercise 11.e.49 GeV.26 keV. Three neutrinos explain the resonance shape. At resonance the cross section σT (π + p) is about 210 mb.59) reaching the unitarity limit for s = M 2 . Three famous resonances are: • The ∆-resonance seen in pion-nucleon scattering. u) + f (u. All these decays can be seen and leave an ’invisible’ width of 498 MeV. The cross section at resonance is about 30 nb.1 Show that the cross section for electron-electron scattering (exercise 10. one can reconstruct the resonance shape for different numbers of neutrino species. a fast change in the phase shift for a narrow resonance. dt s(s − 4m2 ) with f (t. Limiting ourselves to a resonance in one channel. 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). i. Its mass is M = 3096. it is furthermore easy to prove that the cross section is given by σel = M 2 Γ2 4π (2ℓ + 1) . which is attributed to neutrinos. The phase shift in the resonating channel near the resonance is given by tan δℓ (s) = MΓ . u) + g(t. M2 − s (11.4) can be written as 4πα2 dσ = {f (t. • The Z 0 resonance in e+ e− scattering with M = 91.88 MeV. Note that because of the presence of a background the phase shift at resonance may actually be shifted.Scattering theory 112 (Prove this using the unitarity condition for partial waves). Knowing that each neutrino contributes about 160 MeV (see next chapter).2 GeV. 2 q (s − M 2 )2 + M 2 Γ2 (11. the full width is Γ = 88 keV. the partial width into e+ e− is Γee = 5. • The J/ψ resonance in e+ e− scattering. This is a narrow resonance discovered in 1974. u) = g(t. From this one sees that a resonance has the same shape in all channels but different strength. Γ = 2. Essentially this resonance can decay into quark-antiquark pairs or into pairs of charged leptons. The half-width of the resonance is M Γ. t) + g(u. t)} .60) showing that the phase shift at resonance rises through δ = π/2 with a ’velocity’ ∂δ/∂s = 1/M Γ. its width Γ = 120 MeV. The cross section σT (π − p) also shows a resonance with the same width with a value of about 70 mb. u) = 1 1 2 (s + u2 ) + 4m2 (t − m2 ) t2 2 2 1 1 ( s − m2 )( s − 3m2 ) tu 2 2 . Its mass is M = 1232 MeV.

4. we have here already used unitarity to find a finite imaginary part that is proportional to the total decay width.999 877 Γ and Γ(π − → e− νe ) = 0. Exercise 11.000 123 Γ.8. Show that unitarity can be used one step further to fix also the real part of the residue of the pole in the amplitude.Scattering theory 113 Exercise 11. Calculate now fπ . writing N (s) M1 (s) = 2 + iM Γ . Compare the result with the factor trf 1/ρa ρb va that we used in the derivation of the basic expression for cross sections. ρcm ρcm a b 1 cm . cm |va − vb | in the center of mass system for two colliding particles a and b.3 Complete the calculation of the decay width and lifetime of the pion by including the appropriate phase space factors and using the calculation of the squared matrix element in Exc. As discussed for unstable particles.2).6 × 10−8 s and the relative decay widths Γ(π − → µ− νµ ) = 0. Compare the results with the experimental π − lifetime τ = 2.4 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. Thus. Exercise 11. a quantity which we will ¯ ¯ encounter in the next chapter (section 12. s − M∆ ∆ ∆ 2 calculate Re N (M∆ ) .

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

However. requiring that the ’connection’ Aµ (x) is such that Dµ → S(x)D µ S −1 (x).16) which in the presence of the ’connection’ Aµ differs from the total change between x ψ(x) and x+dx ψ(x+ dx).14) (12.19) This gives rise to a (path dependent) phase in each point.e. Dν ] ψ(x) 1 − ig dσ µν Gµν ψ(x). should be independent of such transformations. if two different paths to the same point give different phases the effects can be observed. We note that local gauge invariance requires that we can modify all local systems with a (local) unitary transformation S(x). x+dx ψ(x + dx) = 1 + dxµ Dµ x+dx ψ(x).The standard model the basis for x G and x+dx 117 G. (12.15) ψ(x). (12. ds s dxµ ds′ P exp ig A (s′ ) ψ(0). but expresses them with respect to different bases. = . (12. the real change thus is given by ψ(x) = (1 − dy ρ Dρ )(1 − dxσ Dσ )(1 + dy ν Dν )(1 + dxµ Dµ )ψ(x) = 1 + dxµ dy ν [Dµ . 12.16.17) The three equations given so far immediately give Dµ = ∂µ − ig Aµ (x). ds which is solved by dψ(s) ds = dxµ . which connects two identical vectors. In principle such a phase in a given point is not observable. i. 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). A ’constant’ vector that only rotates because of the arbitrary definitions of local frames satisfies Dµ ψ(x) = 0. The relation in Eq. (12. x+dx ψ(x) = = 1 + ig dxµ Aa (x)Ta µ 1 + ig dx Aµ (x) µ x x ψ(x) (12. -dx -dy µ µ dyµ µ dx For a vector. x x+dx ψ(x + dx) = (1 + dxµ ∂µ ) ψ(x).e. i. If there is a ’true’ difference in the vector ψ(x) and ψ(x + dx) it is denoted with the covariant derivative connecting the vectors expressed with respect to the same basis. What is this physical effect by which the ’connection’ Aµ can be observed? For this consider the phase around a closed loop. ∂µ − ig Aµ (x) ψ(x) = 0 dxµ ∂µ − ig Aµ (x(s)) ψ(x) = 0.18) or considering a path xµ (s) from a fixed origin (0) to point x.

The standard model 118 where Gµν = (i/g)[Dµ . In geometric language the effect on parallel transport of a vector depends on the ’curvature’ Gµν . 2 2 4 −V (φ) The potential V (φ) is shown in fig. If it is zero one has dxµ Aµ (x) = 0.20) with a lagrangian density of the form L = 1 1 1 ∂µ φ · ∂ µ φ − m2 φ · φ − λ(φ · φ)2 .  φ= 2      φ3 (12. . external magnetic fields. Nevertheless. Consider as an example a ferromagnet with an interaction hamiltonian of the form H=− Jij Si · Sj . 12. and equivalently Aµ can be considered as a pure gauge effect.    φ1    φ .2 Spontaneous symmetry breaking In this section we consider the situation that the groundstate of a physical system is degenerate. a theory for a scalar degree of freedom (a scalar field) having three (real) components. 12. Nature will choose one. As there can be one and only one groundstate. Classically the (time-independent) ground state is found for V(φ ) F |φ| Figure 12. which means that by an appropriate transformation S(x) it can be gauged away (see the example of Aharonov-Bohm effect).2: The symmetry-breaking ’potential’ in the lagrangian for the case that m2 < 0. the groundstate itself appears degenerate. If the temperature is high enough the spins are oriented randomly and the (macroscopic) ground state is spherically symmetric. we can disregard those ’perturbations’ and look at the ideal situation. in reality a not perfectly symmetric situation. usually being (slightly) prejudiced by impurities. This characterizes spontaneous symmetry breaking. i. e. 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). this means that there is more than one possibility for the groundstate. Dν ]. In this case there are many possible groundstates (determined by a fixed direction in space). Only if this quantity is nonzero a physically observable effect of Aµ exists. 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.2.e.g. i>j which is rotationally invariant.

It is only invariant under rotations around the 3-axis. The situation now is the following. In this mode the lagrangian and the vacuum are both invariant under a set of symmetry transformations generated by Qa . . They are known as the Weyl mode or the Goldstone mode: Weyl mode. while the lagrangian including the nonzero groundstate expectation value chosen by nature. has less symmetry. The original lagrangian contained an SO(3) invariance under (length conserving) rotations among the three fields. It is appropriate to redefine the field as    ϕ1      φ =  ϕ2  .The standard model 119 a constant field (∇φ = 0) and the condition ∂V ∂φ ϕc =0 −→ ϕc · ϕc = 0 or ϕc · ϕc = − m2 ≡ F 2. . In the case of degeneracy. . A quantum field theory has only one nondegenerate groundstate |0 . etc. In this case the spectrum is described . for the vacuum Qa |0 = 0.23)     F +η the latter only forming a minimum for m2 < 0. (12.21) Stability of the action requires the classical groundstate ϕc to have a well-defined value (which can be nonzero). The η massless particles are called Goldstone bosons.25) 2 2 2 Therefore there are 2 massless scalar particles.) terms constitute interaction terms.   ϕc = 0|φ |0 =  (12. The field along the third axis plays a special role because of the choice of the vacuum expectation value. ϕ2 and η.24) 1 2 2 4 1 1 1 = (∂µ ϕ1 )2 + (∂µ ϕ2 )2 + (∂µ η)2 + m2 η 2 + . Any constant field with ’length’ F is a possible groundstate. 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.e. Writing the field φ as a sum of a classical and a quantum field. It is sufficient to do this to second order in the fields as the higher (cubic. In order to see the consequences for the particle spectrum of the theory we construct the lagrangian in terms of the fields ϕ1 . so one has 0|φquantum |0 = 0 and hence 0|φ|0 = ϕc .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. 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 . while the quadratic terms must correspond with non-negative masses. The classical groundstate appears degenerate. say    0    0 . therefore a choice must be made. In this situation one speaks of spontaneous symmetry breaking. i. The presence of a nonzero value for the classical groundstate value of the field will have an effect when the field is quantized. λ such that 0|ϕ1 |0 = 0|ϕ2 |0 = 0|η|0 = 0. (12. φ = ϕc + φquantum where for the (operator-valued) coefficients in the quantum field one wants 0|c† = c|0 = 0. (12.

32) (12. called flavors. it is easy to see that one gets ∂µ V µ = i ψ [M. if the generators Qa generate a symmetry. This lagrangian density then is invariant under unitary (vector) transformations in the flavor space.28) where ψ is extended to a vector in flavor space and M is a diagonal matrix in flavor space.  . i. a and a′ are degenerate states. In this mode the lagrangian is invariant but Qa |0 = 0 for a number of generators.      ψu   mu           md     ψ =  ψd  . suggesting the stronger relation ∂ µ Jµ (x) = fπ m2 a π a (x). ψ −→ ei α·T ψ. ∂ ∂ (12. strange. T = τ /2.The standard model 120 by degenerate representations of the symmetry group. a massless Goldstone boson for each ’broken’ generator. a 0|∂ µ Jµ (x)|π b (k) = fπ k 2 ei k·x δab = fπ m2 a ei k·x δab . T ] ψ. This means that they are operators that create states from the vacuum.30) which in the case that we include only two flavors form an SU (2)V symmetry (isospin symmetry) generated by the Pauli matrices.31) . down.29) (Note that each of the entries in the vector for ψ in the above equation is still a 4-component Dirac spinor). a ∂ µ Jµ = 0. (12. V µ = ψγ µ T ψ. H] = 0. Using translation invariance and as kµ is the only four vector on which this matrix element could depend one may write a 0|Jµ (x)|π b (k) = fπ kµ ei k·x δab (fπ = 0) (12. (12.) one can write L = f ψ f (i/ − Mf )ψf .    . and |a and |a′ belong to the same multiplet (there is a Qa such that |a′ = Qa |a ) then H|a = Ea |a implies that H|a′ = Ea |a′ . M =      . (12. π Example: chiral symmetry An example of spontaneous symmetry breaking is chiral symmetry breaking in QCD. there must be a nonzero expectation value 0|Jµ (x)|π (k) . Using the Dirac equation. 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. Note that for the fields π a (x) one would have a the relation 0|π a (x)|π a (k) = ei k·x . .    . = ψ(i/ − M )ψ. The conserved currents corresponding to this symmetry transformation are found directly using Noether’s theorem (see chapter 6).e. [Qa . Goldstone mode. the lagrangian for the quarks consists of the free Dirac lagrangian for each of the types of quarks. Taking the derivative.. etc. and one must have mπa = 0.e. Neglecting at this point the local color symmetry. Known examples are rotational symmetry and the fact that the spectrum shows multiplets labeled by angular momentum ℓ (with members labeled by m). As the a generators for a symmetry are precisely the zero-components of a conserved current Jµ (x) integrated a a over space. i. i. denoted |π a (k) . irrespective of the fact if they annihilate the vacuum. π (12.e.26) for all the states labeled by a corresponding to ’broken’ generators. The generators Qa (in that case the rotation operators Lz .27) If the transformations in the lagrangian give rise to a symmetry the Noether currents are conserved. A bit more formal. Including a sum over the different flavors (up.

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

4 2 where Since the explicit (adjoint. Initially there are 3 massless gauge fields (each. 2 massive vector fields or spin 1 bosons (2 × 3) and one scalar field (1). Dµ φ = Gµν = ∂µ φ + g Wµ × φ. again 9 degrees of freedom. One may wonder about the degrees of freedom. .The standard model 122 12. like a photon.    0      ϕc = 0|φ |0 =  0  . . Considering the same example with rotational symmetry (SO(3)) as for spontaneous symmetry breaking of a scalar field (Higgs field) with three components. After symmetry breaking the same number (as expected) comes out. thus 9 independent degrees of freedom. .     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 + . having two independent spin components) and three scalar fields (one degree of freedom each). is made.40) a Dµ φ = ∂µ φ − ig Wµ La φ.42) 2 from which one reads off that the particle content of the theory consists of one massless gauge boson 3 1 2 (Wµ ).e. (12. as in this case there are no massless Goldstone bosons. The difference comes when we reparametrize the field φ.   The symmetry is broken in the same way as before and the same choice for the vacuum. Therefore we can always rotate the field φ into the 3-direction in order to simplify the calculation. (12. two massive bosons (Wµ and Wµ with MW = gF ) and a massive scalar particle (η with m2 = η 2 −2 m . made into a gauge theory. in this case three-dimensional) representation reads (La )ij = −i ǫaij . . while the number of massive gauge bosons coincides with the number of ’broken’ generators.39) (12. one sees that the fields Wµ and Gµν also can be represented as three-component fields.37) (12.3 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. i.     0  0         0 = . 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). (12.    0 φ= (12. We have the possibility to perform local gauge transformations. but one has 1 massless gauge field (2). 1 1 L = − Gµν · Gµν + Dµ φ · Dµ φ − V (φ).41)         φ3 F +η Explicitly one then has  2 2  gF Wµ + g Wµ η   1 1  −gF Wµ − g Wµ η Dµ φ = ∂µ φ + g Wµ × φ =   ∂µ η     . The latter is a spin 0 particle (real scalar field) called a Higgs particle.38) ∂µ Wν − ∂ν Wµ + g Wµ × Wν .

2 2 = ∂µ R + i g ′ Bµ R.44) which has an SU (2)W symmetry under transformations eiα·TW . muon µ− or tau τ − ).48) (12.45) (12. In the standard model one starts with a very simple basic lagrangian for (massless) fermions which exhibits more symmetry than observed in nature.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. explicitly Dµ L Dµ R i i g Wµ · τ L + g ′ Bµ L. its corresponding neutrino (νe . The lagrangian for the leptons consists of three families each containing an elementary fermion (electron e− . the quarks (up. (12. νµ and ντ ) and their antiparticles.and righthanded particles. down. i. ψR/L = 2 (1 ± γ5 )ψ decouple. electromagnetic and weak forces.The standard model 123 12. (12. The procedure.51) . R −→ e−i β R.   0   1 +1/2  L   νe  3   − = with TW = and TW = L=    −1/2 eL L 2 R = R − = eR 3 with TW = 0 and TW = 0. explicitly L −→ e−i β/2 L. muons. Thus L (f ) = i eR / eR + i eL / eL + i νeL / νeL + (µ. neutrinos. left. the leptons (elektrons. 2 (12.49) 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.). etc.e. = ∂µ L − (12. U(1)Y U(1)Y (12. τ ). Thus the lagrangian density is L (f ) ∂ ∂ = i L/ L + i R/ R.) and the gauge bosons responsible for the strong.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. ∂ ∂ ∂ (12. etc. implies certain nontrivial relations between masses and mixing angles that can be tested experimentally and sofar are in excellent agreement with experiment.50) (12. This is written as 3 Q = TW + YW .47) and YW is considered as an operator that generates a U (1)Y symmetry. For the neutrino only a lefthanded particle (and righthanded antiparticle) exist. explicitly L R SU(2)W SU(2)W −→ ei α·τ /2 L. under which the lefthanded and righthanded particles with YW (L) = −1 and YW (R) = −2 transform with eiβYW /2 . By introducing gauge fields and breaking the symmetry a more complex lagrangian is obtained. while the lefthanded particles form a doublet. 1 As they are massless.43) and One introduces a (weak) SU (2)W symmetry under which eR forms a singlet. however. R. that gives a good description of the physical world.

this leads to the lagrangian L (h1)     . which assures with the choice of YW for the Higgs field that Q generates the remaining U (1) symmetry.53) φ=   √ (θ4 − iθ3 )  φ 2 with TW = 1/2 and YW = 1 is introduced. with the following lagrangian density consisting of a symmetry breaking piece and a coupling to the fermions. leading to the parametrization   1  0    φ(x) = √  (12. . . −V (φ) L and (h2) = −Ge (LφR + Rφ L).54) = (Dµ φ)† (Dµ φ) −m2 φ† φ − λ (φ† φ)2 . . 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. (12. the U (1)Q symmetry (generated by the charge operator). † i i g Wµ · τ − g ′ Bµ )φ.The standard model 124 3 3 where Wµ is a triplet of gauge bosons with TW = 1.59) . a complex Higgs field   +   1  φ   √2 (θ2 + iθ1 )    0 = 1   (12. + 8 g2 v2 1 + − (Wµ )2 + (Wµ )2 (∂µ h)2 + m2 h2 + 2 8 (g 2 + g ′2 ) v 2 (Zµ )2 + .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 + .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.   h)  (12. . Putting this in leads to L (f ) =L (f 1) +L (f 2) . (12. + 8 (12. L where L (h1) (h) =L (h1) +L (h2) .52) L L (f 1) (f 2) = = i Rγ µ (∂µ + ig ′ Bµ )R + i Lγ µ (∂µ + i ′ i g Bµ − g Wµ · τ )L 2 2 1 1 − (∂µ Wν − ∂ν Wµ + g Wµ × Wν )2 − (∂µ Bν − ∂ν Bµ )2 . Using local gauge invariance θi for i = 1. . (12. 4 4 In order to break the symmetry to the symmetry of the physical world. 2 and 3 may be eliminated (the necessary SU (2)W rotation is precisely e−iθ(x)·τ ). For the classical field the choice θ4 = v is made.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.

(12. 2θ 2θ 4 cos W cos W 0.60) (12.63) (12. 2 Zµ (12. (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 + . σ (νµ γ ρ µL ) 2 2 (eL γ νe ) 2 k + MW (12. 2 + g ′2 g 3 g ′ Wµ + g Bµ (12.68) g2 ρ 5 2 (νµ γρ (1 − γ5 )µ) (eγ (1 − γ )νe ) 8 MW (µ)† (jL )ρ (e) (jL )ρ GF (µ)† (e) ≡ i √ (jL )ρ (jL )ρ . g ′ /g = tan θW .62) g 2 + g ′2 3 ≡ sin θW Wµ + cos θW Bµ .67) The coupling of electrons or muons to their respective neutrinos. . . we can read off e = g sin θW = g ′ cos θW . and correspond to three massive particle fields (W ± and Z 0 ) and one massless field (photon γ) with 2 MW 2 MZ 2 Mγ = = = g2 v2 .61) (12.69) .65) The weak mixing angle is related to the ratio of coupling constants.66) From the coupling to the photon. 2 3 g Wµ − g ′ Bµ 3 ≡ cos θW Wµ − sin θW Bµ . 4 2 MW g2 v2 = . 2 (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.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 .

The coupling to the Higgs particle is weak as the value for v calculated e. one has the relation g2 e2 1 GF √ = .05 and CA = −1/2 we calculate Γ(e+ e− ) ≈ 78. v (12.166 4 × 10 0. The masses of the fermions and the coupling to the Higgs particle are contained in L (h2) .e. 3 = TW . one obtains L (h2) Ge v Ge = − √ (eL eR + eR eL ) − √ (eL eR + eR eL ) h 2 2 me = −me ee − eeh. Γinvisible = 480 MeV one sees that three families of (light) neutrinos are allowed. leptons with CV = −1/2 + 2 sin2 θW ≈ −0.72) (12.70) = 2 = 2 sin2 θ 8 MW 2 v2 8 MW 2 W In this way the parameters g. muon or tau. Finally we want to say something about the weak properties of the quarks. 2 The value of α deviates from the known value α ≈ 1/137 because of higher order contributions.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 . 80.76) with CV CA 3 = TW − 2 sin2 θW Q. i. as appear for instance in the decay of the neutron or the decay of the Λ (quark content uds). 1. i.e.79) Γ(Z 0 → f f ) = 48π cos2 θW For the electron.19 GeV. from the MW mass is about 250 GeV.78) From this coupling it is straightforward to calculate the partial width for Z 0 into a fermion-antifermion pair. (12. (12.40 GeV. −5 (12. g2 MZ 2 2 (CV + CA ). .80) First.The standard model 126 the good old four-point interaction introduced by Fermi to explain the weak interactions.77) (12.71) GeV −2 = = = = . giving rise to a running coupling constant after renormalization of the field theory.231 2. Comparing this with the total width into (invisible!) channels. With the choosen vacuum expectation value for the Higgs field. 91. (12. Γe ≈ Γµ ≈ Γτ ≈ 83 MeV). such as2 α = e2 /4π GF sin2 θW MW MZ ≈ 1/127. 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. me /v is extremely small. For each neutrino species (with CV = 1/2 and CA = 1/2 one expects Γ(νν) ≈ 155 MeV.g. the mass of the electron comes from the spontaneous symmetry breaking but is not predicted (it is in the coupling Ge ).74) (12. (12.5 MeV (exp.73) (12. 2 2 2 (12. g ′ and v determine a number of experimentally measurable quantities.

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

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. † Λu Λ† = Vu G2 Vu . ms and mb ). Note that we need the conjugate Higgs field to get a U (1)Y singlet in the case of the charge +2/3 quarks. Thus one has † † † =⇒ −DL Vd Md Wd DR − DR Wd Md Vd† DL − UL Vu Mu Wu UR − UR Wu Mu Vu UL (12.g.The standard model 128 using the socalled Wolfenstein      + O(λ4 )   with λ ≈ 0. Working with the mass eigenstates one simply sees that the weak current coupling to the W ± mass † mass becomes U L γ µ DL = U L γ µ Vu Vd DL . the weak mass eigenstates are L (h2.34. e. A ≈ 0. The Λu and Λd are complex matrices in the 3×3 family and Λd Λ† = Vd G2 Vd† . i. The Higgs field is still limited to one complex doublet.84) with Gu and Gd being real and positive and Wu and Wd being different unitary matrices.85) √ √ with Mu = Gu v/ 2 (diagonal matrix containing mu .227.5 Family mixing in the Higgs sector and neutrino masses The quark sector Allowing for the most general (Dirac) mass generating term in the lagrangian one starts with L (h2.   φ =    −φ− 0 2 For the (squared) complex matrices we can find positive eigenvalues. (12. u u where we include now the three lefthanded quark doublets in QL . . d d (12. 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 three righthanded quarks with charge +2/3 in UR and the three righthanded quarks with charges −1/3 in DR . mc and mt ) and Md = Gd v/ 2 (diagonal matrix containing md .22 and η ≈ 0. respectively).e. UR = uR cR tR .83) where Vu and Vd are unitary matrices. The imaginary part i η gives rise to CP violation in decays of K and B-mesons (containing s and b quarks. each of these containing   the three families. for which we need the appropriate weak isospin doublet  0∗    1    φ c  = √  v +h .q) = −QL φΛd DR − DR Λ† φ† QL − QL φc Λu UR − UR Λ† φc† QL u d (12.q) ′ weak † mass mass DL = DL = Vu Vd DL = VCKM DL .86) the unitary CKM-matrix introduced above in an ad hoc way. (12.82 and ρ ≈ 0. allowing us to write † Λu = Vu Gu Wu † and Λd = Vd Gd Wd .82) space.

e. and obtain L (h2. As before.ℓ) † † † =⇒ −EL Ve Me We ER − ER We Me Ve† EL − NL Vn Mn Wn NR − NR Wn Mn Vn NL . (12. decoupling from all known interactions.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    . 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 . The lepton sector (massive Dirac neutrinos) In principle a massive Dirac neutrino could be accounted for by a lagrangian of the type L (h2.  =  − 1/6 2/3 0           1/2 1/2 0 0 1 0 1/3 1/2 − 1/6 (12. mµ and mτ .87) is a weak doublet containing the three families of neutrinos (NL ) and charged leptons (EL ) and ER is † a three-family weak singlet.88) √ mass with Me = Ge v/ 2 the diagonal mass matrix with masses me . EL = Ve† EL .ℓ) = −LφΛe ER − ER Λ† φ† L. while θ23 then is determined by atmospheric neutrino oscillations requiring |∆m2 | ≈ 2. (12. e   N L= L  EL     (12. † † Again we continue as before now with matrices Λe = Ve Ge We and Λn = Vn Gn Wn . For the (massless) neutrino fields we just can redefine fields into NL = Ve† NL . EL = Ve EL . it is particularly useful because θ12 is essentially determined by solar neutrino oscillations requiring ∆m2 ≈ 8 × 10−5 12 eV2 (convention m2 > m1 ). we find massless neutrinos. there is no family mixing for massless neutrinos.ℓ) =⇒ −Me EL Ve We ER − ER We Ve† EL . i.92) .5×10−3 eV2 . In this case. 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.91) † with UPMNS = Vn Ve known as the Pontecorvo-Maki-Nakagawa-Sakata mixing matrix.    (12. NL = Vn NL and NR = Wn NR µ mass γ µ V † V N mass . UPMNS a parametrization that in principle also could have been used for quarks. 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  . The W -current then becomes mass mass E L γ µ NL = E L γ µ NL . since the weak current is the only place where they show up. one can write Λe = Ve Ge We and we find † L (h2. (12.ℓ) = −LφΛe ER − ER Λ† φ† L − Lφc Λn NR − NR Λ† φc† L e n (12.90) mass † mass mass † mass † We note that there are mass fields ER = We ER . The mass fields ER † mass † mass = We ER .89) with three righthanded neutrinos added to the previous 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 massless and massive Majorana neutrinos. Υ′ and Υ′ (satisfying Υc = Υ and as discussed in chapter 6.96) VPMNS = UPMNS K with K =  0 eiα2 /2 0  . For these light Majorana neutrinos one has. 1 L Υ ′ = nR + nc . however. This is called the see-saw mechanism (outlined below).95) 2 In order to have more than a completely decoupled sector.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.ν =− containing three (CP-violating) phases (α1 . −    MD MR nR 2 (12. We use here the primes starting with the weak eigenstates. is to add in Eq. hence the mixing matrix becomes   iα /2 0 0   e 1     (12. Υc ≡ (ΥL )c = ΥR 1 2 L and Υc = ΥL ).The standard model 130 The lepton sector (massive Majorana fields) An even simpler option than sterile righthanded Dirac neutrinos. as above. one must for the neutrinos as well as charged leptons. α2 and δ). The mass matrix can be written as   ML |MD | eiφ     M = (12.ν = − MR NR NR + MR NR NR . 2 R ψ = nR + nL . The see-saw mechanism Consider (for one family N = n) the most general Lorentz invariant mass term for two independent Majorana spinors. Thus 2 Υ ′ = nc + nL . (12.       0 0 1 L mass. couple the right.94) M L NL NL + M L NL NL .99) . a unitary matrix relating them to the weak eigenstates. For the charged leptons there cannot exist a Majorana mass term as this would break the U(1) electromagnetic symmetry.98) (12. 1 c ∗ c L mass. For the leptons.88 a Majorana mass term for the (lefthanded) neutrino mass eigenstates. Absorption of phases in the states is not possible for Majorana neutrinos. the left.and righthanded species then just form a Dirac fermion. with for the righthanded sector a mass term MR . For the neutrino sector. Actually. coupled by a Dirac mass term. (12. If the Majorana mass MR ≫ MD one actually obtains in a natural way one Majorana neutrino with a very small mass.100)  |MD | eiφ MR 1 1 ∗ ∗ M L nc nL + M L nL nc − M R nR nc + M R nc nR L R L R 2 2 1 1 ∗ ∗ − M D nc nc + M D nL nR − M D nR nL + M D nc nc L R R L 2 2     1  nc nR   M L M D   nL   L   c  + h. one being massive. 12. 2 although this option is not attractive as it violates the electroweak symmetry. are equivalent to two decoupled Majorana neutrinos (see below). Thus (disregarding family structure) one has two Majorana neutrinos.c. it is easy to see R that this incorporates the Dirac case by considering the lefthanded part of Υ′ and the righthanded 1 part of Υ′ as a Dirac spinor ψ.and lefthanded species through Dirac mass terms coming from the coupling to the Higgs sector as in the previous section. 1 ∗ c c (12. The way to circumvent this is to introduce as in the previous section righthanded neutrinos.

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

Γ(Z 0 → νe νe ).) are proportional to the mass squared for bosons or mass for fermions of the particles. Note that you can find the answer without explicit construction of all interaction terms in the lagrangian. The mass of the Z 0 is MZ = 91 GeV. to be precise one has for bosons (b) or fermions (f). ZZh. e+ e− h. (b) Calculate the width to electron-positron pair. Exercise 12. 3 TW . and the width to a pair of neutrino’s. ghbb = b v v where v is the vacuum expectation value of the Higgs field.4 Calculate the lifetime τ = 1/Γ for the top quark (t) assuming that the dominant decay mode is t → b + W +.6 Check that the use of the Wolfenstein parametrization in the CKM matrix indeed gives a unitary matrix. the weak mixing angle is given by ¯ sin2 θW = 0. Neglect the masses of fermions and use the fact that the sum over polarizations is 3 pµ pν . etc. Γ(Z 0 → e+ e− ). Exercise 12.The standard model 132 with CV CA = = 3 TW − 2Q sin2 θW . In the standard model this coupling is described by the vertex −i g µ √ (γ − γ µ γ5 ) . 2 2 The masses are mt ≈ 175 GeV. hhh. mb ≈ 5 GeV and MW ≈ 80 GeV. 0 ¯ Γ(Z 0 → f f ) = MZ g2 48π cos2 θW f f CV 2 + CA 2 .7 In this exercise two limits are investigated for the two-Majorana case. at least up to a high (which?) order in λ.231.5 Show that the couplings to the Higgs (for W + W − h. . Write down the matrix element squared (averaged over initial spins and summed over final spins) for the decay of the Z 0 . Mf M2 and ghf f = . Exercise 12. Exercise 12. ǫ(λ) (p)ǫ(λ)∗ (p) = −gµν + µ ν M2 λ=1 ¯ to calculate the width Γ(Z → f f ).

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

Sign up to vote on this title
UsefulNot useful