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

Schroeder. I refer to the book of Srednicki [1] and Ryder [2].2 2.-B.2 6. Weinberg. Peskin and D. 1996.3 3. Vol. The quantum theory of fields. The notes contain all essential information.2.2 36 22 . M. 1.5 2.2 6.E.2 2.1 4.3 2.4 2. Quantum Field Theory. Peskin & Schroeder and Srednicki. Cambridge University Press. Quantum Field Theory. Cambridge University Press. Ryder. Addison-Wesly.H. 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). Diagrammatica. 2007. I: Foundations.4 3. 1994. L.3 2.2 3.1 6. 6.4 6.5 2.1.4 2.4 3.2 4. Quantum Field Theory. Veltman. Cambridge University Press. 3. Cambridge University Press.3 3. 2.1 2 2 3.5 Ryder Peskin & Schroeder Srednicki 1 1 1 1 3 2. S.2 3.3 3.7 2. Vol.1 3. II: Modern Applications. Cambridge University Press. 3. 5.3 2.3 6. C.5 4. Srednicki.2 3. Itzykson and J.V.3 2. M.3 3. 1995. 4.4 4. 1980. Corresponding chapters in books of Ryder.2 3.2 3. 2. Zuber.3 3. 3.2 1.1 3. McGraw-Hill.3 2.3 References As most directly related books to these notes.1 2. These notes Section 1. 1985.3 4. but are rather compact. Other text books of Quantum Field Theory that are useful are given in refs [3-6].5 5.2 2. 1995.4 3.3.1 2.1 5.8 3. An introduction to Quantum Field Theory. M.1 1.

6 8.6 3. 8. 20. 2.1 7.5 .4.2 4.5 9.2 8.3 12.2 5.3 2.1 11.5 4.6 3. 6.2 9.1.3 8.2 10. 6.2 8.3 8.6 3.4 3.3 10.6 10.1 8.3.7.4.1 9. 4.5 4.1 20. 9.5 7.1 10.5 11.3 11.7 5.1.5 6 7 6 6.1 12. 9.2 12.10 8.3 7.1 20.4. 4.2.3.4 12.4 8.4 7. 6.5 7.2.3 7.3.6 6.4 9.6 9.1 20.3 6.5 6. 6. 6.1 9.3 6 8 9 Srednicki 3 3 3 37 5.1 5.6 4.4 These notes Section 7. 4.5 2.4 12.5 9.4 2.4 4.5 Ryder 4 4 4 4 4 4 Peskin & Schroeder 2. 2.3 9.4 10.2.2 7.8 6.8 5 4.5.7 4. 4.2.1.3 3.2 11.3 20.

• Slight rewriting of discussion of Majorana mass term in section 6.5 Corrections 2011/2012) • Extension of Exercise 3.2 .3 • First two sentences of Chapter 5 have been modified.

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

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.

ˆ ˆ (1. (1.µn = Λµ1ν1 . x) = (x0 . 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.1. The lower indices are constructed in the following way. x2 . Many other four vectors and tensors transforming like T ′ µ1 .. in this case an inertial system exists in which the two points are at the same space point and in that frame ds2 just represents the time difference ds2 = dt2 . x3 ).21) (1. In Minkowski space. where the coordinate t = x0 is referred to as the time component.. The transformations that leave invariant the distance ds2 = dt2 − (dx2 + dy 2 + dz 2 ) between two points are called inhomogeneous Lorentz transformations or Poincar´ transformations. x1 . One can distinguish: • ds2 > 0 (timelike intervals). independent of a coordinate system. The Poincar´ transformations include e e Lorentz transformations and translations. Unlike in Euclidean space. such as eigentime τ for the position vector τ 2 ≡ x2 = t2 − x2 . In special relativity we start with a four-dimensional real vector space E(1. x2 . xi are the three space components. given by g00 = −g11 = −g22 = −g33 = 1 (the ˆ ˆ other components are zero). in which case zi = ǫijk xj yk . Because of the different signs occurring in gµν .. No difference is made between upper or lower. δim δjm δkm δin δjn δkn .3) with basis nµ (µ = ˆ 0. x1 .. Vectors are denoted x = xµ nµ . the invariant length or distance (squared) is not positive definite. xµ = gµν xν .26) (1. one must distinguish tensors with upper or lower indices and one can have mixed tensors.2.23) δjm δkn − δjn δkm .25) Within Minkowski space the real. are . Λµn n T ν1 . 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 . . linear transformations that do not change the length of a four-vector are called the Lorentz transformations.24) The quantity gµν ≡ nµ · nν is the metric tensor. the points lie on the lightcone and they can be connected by a light signal.νn can be conν structed. linear transformations that leave the length of a vector invariant are called (homogeneous) Lorentz transformations. The length (squared) of a vector is obtained from the scalar ˆ product. So we could have used all upper indices in the above equations. and are given by (x0 . it is convenient to distinguish lower indices from upper indices. x2 = x · x = xµ xν nµ · nν = xµ xν gµν . • ds2 = 0 (lightlike or null intervals). x3 ) = (t. When 3-dimensional space is considered as part of Minkowski space.22) (1. Useful relations are ǫijk ǫℓmn ǫijk ǫimn ǫijk ǫijl = = = δiℓ δjℓ δkℓ 2 δkl .Introduction 5 that can be used in the cross product of two vectors z = x × y. We note that for Euclidean spaces (with a positive definite metric) vectors and tensors there is only one type of indices. The real. • ds2 < 0 (spacelike intervals). One has x2 = xµ xµ = t2 − x2 . The invariant distance between two points x and y in Minkowski space is determined from the length dsµ = (x − y)µ . These transformations do change the components of a vector. Relations relating tensor expressions. . For four-vectors in Minkowski space we will use the notation with upper indices and write x = (t. denoted as V ′µ = Λµν V ν . The (invariant) lengths often have special names. (1.3). we will use upper indices for the three-vectors. however. −x).

34) (1. (1. is timelike with invariant length (squared) p2 = p · p = pµ pµ = E 2 − p2 = m2 . defined by ∂2 − ∇2 . p) = (E.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. .Introduction 6 called covariant.1 (a) In the the Hydrogen atom (quantum system) the scale is set by the Bohr radius. The derivative ∂µ is defined ∂µ = ∂/∂xµ and we have a four vector ∂ with components (∂0 . Relate this quantity to the electron Compton wavelength −e via the dimensionless λ fine structure constant α. but more important has the same effect on lefthandside and righthandside. = −24. Each of these permutations leads to a minus sign. for instance. 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 . (1.36) Exercises Exercise 1. . (1. 0123 0123 ′ ′ ǫ µνρσ ǫµνρσ The first identity.27) It is easy to convince oneself of the nature of the indices in the above equation. ∂t ∂x ∂y ∂z ν ∂µ xν = gµ . because one has (1. It is an invariant tensor. p). ′ ′ . ∂1 . (1. .31) ǫµνρσ ǫµν ′ ′ ρσ ′ = − (1. Examples are the scalar products above but also relations like pµ = m dxµ /dτ for the momentum four vector. explicitly written as (p0 . not affected by Lorentz transformations.32) .29) 2 = ∂ µ ∂µ = ∂t2 The value of the antisymmetric tensor ǫµνρσ is determined in the same way as for ǫijk .35) ρ ǫµνρσ ǫµν σ σ ǫµνρσ ǫµνρ ′ ′ = −2 g ρρ g σσ − g ρσ g σρ = −6g σσ . (1. constructed via the metric tensor itself. where m is called the mass of the system. a∞ = 4πǫ0 2 /me e2 . The length squared of ∂ is the d’Alembertian operator.28) ν ν Note that gµ with one upper and lower index.33) (1. ∂2 .and righthandside a four vector because τ is a scalar quantity! The equation with t = x0 instead of τ simply would not make sense! The momentum four vector. = ∂ . starting from ǫ0123 = 1. ∂3 ) = ∂ ∂ ∂ ∂ . ∂t (1.∇ . g0 = g1 = g2 = g3 = 1. gµ = gµρ g ρν and 0 1 2 3 is in essence a ’Kronecker delta’. For the contraction of a vector with the antisymmetric tensor one often uses the shorthand notation ǫABCD = ǫµνρσ Aµ Bν Cρ Dσ . Note that in this equation one has on left.

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

2) Acting on the wave function one finds for a free particle. p2 = pµ pµ = E 2 − p2 = M 2 . i ∂ ∇2 ψ(r.4) where M is the particle mass.1 and 2. one obtains (2. t) = exp(−i k 0 t + i k · r). Although it is straightforward to find the solutions of this equation. But the replacement 2. (2. E = p2 . 8 . t) = ∂2 − ∇2 + M 2 φ(r. Substitution of operators gives the Klein-Gordon (KG) equation for a real or complex function φ. p).3 are not covariant. we just want to play a bit with covariant equations and study their behavior under Lorentz transformations.3) Equations 2.1) under the substitution (in coordinate representation) E −→ Eop = i ∂ . (2. (2.7) k 0 = ± k2 + M 2 = ±Ek .1 The Klein-Gordon equation In this chapter. t) = − ψ(r.6) The energy spectrum is not bounded from below: considering the above stationary plane wave solutions. written as pµ −→ i∂µ is covariant (the same in every frame of reference). t) = 0. 2M (2. t). 2 + M 2 φ(r.5) with (k 0 )2 = k2 + M 2 . ∂t2 (2.2. namely plane waves characterized by a wave number k. Thus a covariant equation can be obtained by starting with the (covariant) equation for the invariant length of the four vector (E. ∂t 2M (2. The Schr¨dinger equation in quantum mechanics is the operator equation o corresponding to the non-relativistic expression for the energy. ∂t p −→ pop = −i∇. φk (r.Chapter 2 Relativistic wave equations 2. the interpretation of this equation as a single-particle equation in which φ is a complex wave function poses problems because the energy spectrum is not bounded from below and the probability is not positive definite.

but rather the zero component of a four vector.8) (2.Relativistic wave equations 9 i. The KG equation. rather than as a wave function. The integration over k-modes clearly is covariant and restricted to the ‘mass’-shell (as required by Eq. Then. 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. however. (2. we want the most general solution. which is what Lorentz transformations do. This continuity equation can be written down o covariantly using the components (ρ. 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. there are no longer fundamental objections to mix up space and time.2) is j µ = i φ∗ ∂ µ φ or (ρ.10) ∂t which follows directly from the Schr¨dinger equation. j) = ↔ i φ∗ ∂0 φ. φ(x) = d3 k (2π)3 2Ek ˜ ˜ e−i k·x φ(Ek . −k) . ∂µ j µ = 0.2 Mode expansion of solutions of the KG equation Before quantizing fields. for which we need the interpretation of φ itself as an operator. At that point the dependence on position r and time t is just a dependence on numbers/parameters on which the operator depends. 2. relativistically the density is not a scalar quantity. (2π)3 2Ek (2.12) It is easy to see that this current is conserved if φ (and φ∗ ) satisfy the KG equation. (2. As we will see later both problems are related and have to do with the existence of particles and antiparticles. k) ≡ a(k) and φ(−Ek . it is simply a matter of being careful to find a consistent (covariant) theory. k) + ei k·x φ(−Ek . ∂ρ = −∇ · j. −i φ∗ ∇ φ . This leads to the existence of negative densities. It is possible to rewrite it as an integration over positive energies only but this gives two terms (use the result of exercise 2. 2M 2M (2. The appropriate current corresponding to the KG equation (see Excercise 2. −k) ≡ b∗ (k) one has φ(x) = d3 k e−i k·x a(k) + ei k·x b∗ (k) = φ+ (x) + φ− (x). For this we note that an arbitrary solution for the field φ can always be written as a superposition of plane waves. is a second order equation and φ and ∂φ/∂t can be fixed arbitrarily at a given time. 2.11) Therefore. And.13) ˜ with (in principle complex) coefficients φ(k). (2. there are solutions with negative energy. j) of the four-current j.e. This operator has all possible solutions in it multiplied with creation (and annihilation) operators.3).5).14) ˜ ˜ Introducing φ(Ek . having the KG equation as a space-time symmetric (classical) equation. ↔ ↔ (2. just as the dependence on time was in ordinary quantum mechanics.15) .

i. (2.e. after restricting the energies to be positive). 2.20) one has for the mode amplitudes aP (k) = a(−k) and bP (k) = b(−k). It is trivial to see that ∂µ ∂ µ + M 2 φ∗ (x) = 0. which implies (xµ ) = (t.23) . (2π)3 2Ek (2.18) (2. It is easy to show that x φP (x) = φ(˜) x = = = or since one can define φP (x) ≡ d3 k e−i k·x aP (k) + ei k·x bP ∗ (k) .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. φP (x) ≡ φ(˜) (P for parity). for which we consider space inversion. (2.19) implying that for each solution φ(x) there exists a corresponding solution with the same energy. one thus must have that the function φ → φ′ such that φ′ (x′ ) = φ(x) or φ′ (x) = φ(Λ−1 x).6 We will explicitly discuss the example of a discrete symmetry. x ∂µ ∂ µ + M 2 φ(˜) = 0. (2. (2π)3 2Ek (2. changing the sign of the spatial coordinates.14 and 2.Relativistic wave equations 10 In Eqs 2. The choice of a and b∗ allows an easier distinction between the cases that φ is real (a = b) or complex (a and b are independent amplitudes). The coefficients a(k) and b∗ (k) are the amplitudes of the two independent solutions (two. They are referred to as mode and anti-mode amplitudes (or because of their origin positive and negative energy modes). Another symmetry is found by complex conjugating the KG equation.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) .17) (2. (2π)3 2Ek (2.16) The consequence of this is discussed in Exercise 2. Performing some Lorentz transformation x → x′ = Λx. −x) ≡ (˜µ ).15 one has elimated k 0 and in both equations k · x = Ek t − k · x. In terms of modes one has φC (x) = φ∗ (x) = ≡ d3 k e−i k·x b(k) + ei k·x a∗ (k) (2π)3 2Ek d3 k e−i k·x aC (k) + ei k·x bC∗ (k) . x Transforming everywhere in the KG equation x → x one obtains ˜ ˜ ˜ x ∂µ ∂ µ + M 2 φ(˜) = 0. Since a · b = a · ˜ it is easy to see that ˜ b. This shows how parity transforms k-modes into −k modes. x) → (t.22) showing that with each solution there is a corresponding charge conjugated solution φC (x) = φ∗ (x).

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

Relativistic wave equations 12 Exercise 2.7 To solve the problem with positive and negative energies and get a positive definite density. B] = AB − BA and the anticommutator as {A. Exercise 2. We will see a∗ (k) → a† (k) and a(k) → aop (k) with aop and a† op op being annihilation and creation operators (as known from a harmonic oscillator) and the same for the coefficients b(k).2) in terms of the a(k) and b(k) using the expansion in Eq. and β 2 = I. One then finds that Q is expressed as an infinite sum over number operators a† a (particles) and b† b (anti-particles). B} = AB + BA. 2. 1 2 (b) Express now the full charge QV (t) (exercise 2. an attempt to construct a first order differential equation for the time evolution would be to write. 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). (b) Show that one has current conservation for the current 2 We denote the commutator as [A. a{µ bν} ≡ aµ bν + aν bµ . making sure that each component of ψ satisfies the KleinGordon equation (in terms of differential operators ∂µ ∂ µ +m2 = 0) requires the anticommutation relation2 {αi . d3 x (∂ {0 φ)∗ (∂ i} φ).5 (a) As an introduction to parts (b) and (c). i.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). also using the result in (a). t). 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 . express the quantities E(t) = P i (t) = d3 x (∂0 φ)∗ (∂0 φ) + ∇φ∗ · ∇φ + M 2 φ∗ φ . Exercise 2. ∂ ψ(r. β} = 0. (c) Similarly. {αi . .e. 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. ∂t where ψ is a wave function and the nature of α and β is left open for the moment.15. We will encounter these quantities later as energy and momentum. j 0 = ψ†ψ and j = ψ † αψ. αj } = 2 δ ij I.1) for a complex scalar field current (exercise 2. t) = (−i α · ∇ + m β) ψ(r. Note that a{µ bν} indicates symmetrization. i (a) Show that relativistic invariance.

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

7) = a0         0 −1 i 0 1 0 0 1 • [x. . It is straightforward to check that any (finite) rotation can be obtained from a combination of infinitesimal rotations.z N N or (Lk )ij = −i ǫijk . [x. z ) = lim N →∞ that are generated by  i    0 .Groups and their representations 14 (n. which reflects itself in the noncommutation of the generators. In the same way we can consider rotations around the x. ] that satisfies • ∀ x. ˆ R(ϕ. These matrices can be defined as acting on 2-component spinors (χ → Aχ) or equivalently as acting on 2 × 2 matrices (B → ABA† ). for rotations around z for instance.1: the parameter spaces of SO(3) (left) and SU (2) (right). Lj ] = i ǫijk Lk . x]).6) Summarizing. x] = 0 (thus [x. x]] + [z. ˆ therefore. we turn to the group SU (2) of special (det A = 1) unitary (A† = A−1 ) 2 × 2 matrices. and thus (σ · n)2 = 1. namely that there exists a bilinear product [. It is straightforward to check that the conditions require    a0 + i a3 +i a1 + a2    = a0 1 + i a · σ A =   +i a1 − a2 a0 − i a3          1 0          + i a1  0 1  + i a2  0 −i  + i a3  1 0  (3. (3.5) Rotations in general do not commute. y] = −[y.8) ˆ that (σ · a)(σ · b) = a · b + i σ · (a × b). π) (n. • [x. y]] = 0 (Jacobi identity). y ∈ A ⇒ [x.2 π) (-n. is a 3-dimensional Lie-group. They satisfy the commutation relations [Li . Locally this is a 3-dimensional Euclidean space and SU (2). Writing a0 = cos(ϕ/2) and a = n sin(ϕ/2) we have1 A = A(ϕ. y] ∈ A. [z. z]] + [y. One way of viewing the parameter space. Next.and y-axes     0 0 0   0 0    0 0 −i  .   0 (3. These generators form a threedimensional Lie algebra SO(3). (3. Ly and Lz .4) R . n) = 1 cos ˆ 1 Note ϕ ϕ ˆ + i (σ · n) sin 2 2 = exp i ϕ ˆ σ·n . π ) π -sphere (antipodes identified) 2 π -sphere (surface identified) Figure 3. thus is as the surface i=0 of a sphere in 4 Euclidean dimensions. the rotations in SO(3) can be generated from infinitesimal rotations that can be expressed in terms of a basis of three generators Lx . with real a’s and 3 (ai )2 = 1. [y.    Lx =  Ly =  0 0       0 i 0 −i 0 ϕ ˆ . With matrix commutation this algebra satisfies the requirements for a Lie algebra. 2 (3.

n) i ∂ϕ = ϕ=0 (3. n) ≈ 1 + i ϕ J · n.11) Thus σx /2. i. (3. Near the identity. with ˆ J ·n ≡ ˆ 1 ∂A(ϕ. 2 2 2 (3. The existence of symmetries means that there are operators g belonging to a symmetry group G that commute with the Hamiltonian.2 Representations of symmetry groups The presence of symmetries simplifies the description of a physical system and is at the heart of physics. For instance in a function space translations. The next issue is to find the appropriate symmetry operators in the Hilbert space. (3.15) [g. Suppose we have a system described by a Hamiltonian H.14) For a Lie group. now with radius 2π and with all points at the surface identified (see figure). 3.e. . immediately sees that the Lie algebras are identical. however. thus. ˆ ˆ A(ϕ. since any finite transformation can be constructed from the infinitesimal ones. 2 (3. (3.10) σ ˆ · n.Groups and their representations 15 where we have used the (for operators new!) definition eA ≡ An 1 = 1 + A + A2 + . φ′ (r) = φ(r + a) = φ(r) + a · ∇ φ + 1 (a · ∇)2 φ + . They satisfy σi σj σk = i ǫijk . . There exists a corresponding mapping of the groups given by µ: SU (2) −→ SO(3) ˆ ˆ A(ϕ. .12) One. . n! 2! n=0 ∞ (3.13) 0≤ϕ≤π The relation A(σ · a)A−1 = σ · RA a (valid for any vector a) can be used to establish the homomorphism (Check that it satisfies the requirements of a homomorphism. [g. the above mapping corresponds to the trivial mapping of the Lie algebras. In the full parameter space. H] = 0 for g ∈ G. . i. n) π ≤ ϕ ≤ 2π. for fixed n.e. one has a Lie algebra isomorphism that is linear and preserves the bilinear product. i. in particular that AB → RAB = RA RB ).e. . n) ˆ −→ R(2π − ϕ. also can be considered as a filled 3-sphere. . σy /2 and σz /2 form the basis of the Lie-algebra SU (2). sometimes in more than one way (but this will be discussed later). it is sufficient that the generators commute with H.9) The parameter-space of SU (2). the SU (2) → SO(3) mapping is a 2 : 1 mapping where both A = ±1 are mapped into R = I. thus. H] = 0 for g ∈ G. The infinitesimal generators of SU (2) are ˆ obtained by considering infinitesimal transformations. SU (2) ≃ SO(3). 2! . n) −→ R(ϕ.

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

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

Thus one has the (pseudo-orthogonality) relation ΛT GΛ = G ⇔ GΛT G = Λ−1 ⇔ ΛGΛT = G. (ii) |Λ00 | ≥ 1. which as a matrix equation with (Λ)µν = Λµν and (G)µν = gµν gives (ΛT )ρµ (G)µν (Λ)νσ = (G)ρσ . proof: (ΛT GΛ)00 = (G)00 = 1 → Λµ0 gµν Λν0 = 1 → (Λ00 )2 − i (Λi 0 )2 = 1.     G =  10 (3. it is convenient to use a vector notation for the points in Minkowski space.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 (3.27) (3. proof: det(ΛT GΛ) = det(G) → (det Λ)2 = 1.g. explicitly x′µ = (Λ)µν xν + aµ or x′µ = Λµν xν + aµ . (3. det Λ = −1 is called improper). In this section we consider the Poincar´ group. (det Λ = +1 is called proper.23) The proper tensor structure of the matrix element (Λ)µν is a tensor with one upper and one lower index.Groups and their representations 18 3. e.3 The Lorentz group In the previous section spin has been introduced as a representation of the rotation group SU (2) without worrying much about the rest of the symmetries of the world.24) L↑ + L↓ + L↑ − L↓ − proper orthochrone proper non-orthochrone improper orthochrone improper non-orthochrone . From this property. consisting of the Lorentz group and translations.26) (3.     0 0   g00 g01 g02 g03   1 0      g     g11 g12 g13   0 −1 0 0  = . σ/2 in a two-dimensional (spin 1/2) case. Writing x as a column vector and the metric tensor in matrix form. it is easy to derive some properties of the matrices Λ: (i) det(Λ) = ±1. We considered the generators and looked for representations in finite dimensional spaces. 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. 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 .22) (Note that xT is a row vector). Denoted in terms of column vectors and 4-dimensional matrices one writes for the poincar´ transe formations x′ = Λ x + a. e To derive some of the properties of the Lorentz group.25) (3.

From the commutation relations one sees that the boosts (pure Lorentz transformations) do not form a group. 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  .32)  0 −   0 −1 0      0 0 0 −1 . J j ] = i ǫijk J k . 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. Returning to the global group. [K i . J 3 ).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↑ . which satisfy the commutation relations (check!) [J i . K = (K 1 . In L↑ . it is easy to find the following typical examples from each of the four classes. z ) ≈ I + i ϕ J 3 . [J i . This is the origin of the Thomas precession. (3. one distinguishes rotations 0 + and boosts. i proof: use ΛT GΛ = G and ΛGΛT = G. we have found the generators of rotations. Rotations around the z-axis are given by ΛR (ϕ.g. Note that the velocity β = v = v/c and the Lorentz contraction factor γ = (1 − β 2 )−1/2 corresponding to the boost are related to η as γ = cosh η. (3.31)  0 −   0 +1 0      0 0 0 +1   0 0   +1 0    0 −1 0   0   ∈ L↑ (space inversion)   Is =  (3. 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  . since I ∈ L↑ and + + det Λ and Λ0 change continuously along a path from the identity. K 2 . J 2 .30)  0 +   0 +1 0      0 0 0 +1   0 0   −1 0    0 +1 0   0   ∈ L↓ (time inversion)   It =  (3. K j ] = i ǫijk K k . since the generators K do not form a closed algebra. z ) = exp(i ϕ J 3 ). infinitesimally given by ˆ ΛR (ϕ. and those of the boosts. 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). The commutator of two boosts in different directions (e.   0 0   +1 0    0 +1 0   0   ∈ L↑ (identity)   I =  (3. 3 i 2 i=1 (Λ0 ) 3 19 (iii) Boosts along the z-direction are given by ΛB (η.28)                 3′    3    V 0 0 0 1 V 0 0 0 0 The parameter η runs from −∞ < η < ∞.Groups and their representations = i=1 (Λ0 )2 . J = (J 1 . βγ = sinh η. z ) ≈ ˆ ˆ I − i η K 3 . K 3 ). infinitesimally given by ΛB (η. z ) = exp(−i ηK 3 ). K j ] = −i ǫijk J k .

M ρσ ] = −i (g µρ M νσ − g νρ M µσ ) − i (g µσ M ρν − g νσ M ρµ ) .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) (3.4 The generators of the Poincar´ group e ↑ The transformations belonging to P+ are denoted as (Λ. This is a + normal subgroup and the factor group is the Vierer group.37) The algebra of the generators of the Lorentz transformations can be obtained by an explicit calculation2 . three of which only involve spatial coordinates (rotations) and three others involving time components (boosts). K = M ). we have for the (infinitesimal) rotations (around z-axis) ΛR = I + i ϕ3 J 3 = I − i ω12 M 12 . The extension to the Poincar´ group is e ↑ straightforward. 2 (M αβ )µν = i g αµ g βν − g α g βµ ν (3. There are six generators.38) Explicitly. we continue using covariance arguments. Of the four parts only L↑ forms a group. i i0 i i0 1 ijk M jk 2ǫ (3. with ϕi = − 1 ǫijk ω jk and J i = 2 z-axis).36) (3. The e 2 These commutation relations are the same as those for the ’quantummechanics’ operators i(xµ ∂ ν − xν ∂ µ ) . We now want to find a covariant form of the six generators of the Lorentz transformations.39) (e.40) ΛB = I + i η 3 K 3 = I − i ω30 M 30 . the Lorentz transformations form a (Lie) group (Λ1 Λ2 and Λ−1 are again Lorentz transformations. J 3 = M 12 ) and for the (infinitesimal) boosts (along (3. inf (3. 3 30 with η = ω and K = M (e. ǫ). infinitesimally approximated by (I + ω. (3. P+ etc. In order to find the commutation relations including the translation generators. We will just require Poincar´ invariance for the generators themselves. (3.34) The condition ΛT GΛ = G yields g ρµ + ω ρµ gρσ (g σν + ω σν ) = gµν =⇒ ωνµ + ωµν = 0. Thus the two vector operators J and K form under Lorentz transformations an antisymmetric tensor M µν .g. which are obtained by writing the infinitesimal parameters (ω)µν in terms of six antisymmetric matrices (M αβ )µν .33) These four transformations form the Vierer group (group of Klein).g. [M µν . 3. We therefore find (again) that there are six generators for the Lorentz group. One can easily convince oneself that i (ω)µν = ω µν = − ωαβ (M αβ )µν . Also this group can be divided into four parts. explicitly reading x′µ = = (Λ)µν xν + aµ = ((I)µν + (ω)µν ) xν + ǫµ Λµν xν + aµ = (g µν + ω µν ) xν + ǫµ . Summarizing. a). Multiplying the proper orthochrone transformations with one of them gives all Lorentz transformations.35) thus (ω)ij = −(ω)ji and (ω)0i = (ω)i0 .

(1 + iǫγ P γ − i i ωαβ M αβ )P µ (1 − iǫδ P δ + ωρσ M ρσ ) = P µ + ω µν P ν . P µ |p. H] = i P i . They just state that M µν transforms as a tensor with two Lorentz indices. one obtains 2 [P i . the tensor M µν in terms of boosts cK i = M i0 and rotations J i = 1 ǫijk M jk . = −i (g µρ P ν − g νρ P µ ) . (3. a) = Λ−1 µ P ν . writing the generator P = (H/c. the generators J 2 and J 3 in the case of the rotation group. These define states with specified quantum numbers that are eigenvalues of these generators. [J i . Explicitly. For instance. P ρ ] = 0. These form a group called the little group associated with that four vector. K j ] = −i ǫijk J k /c2 . P j ] = i ǫijk P k .41) 2 At this point. H] = 0. We have here reinstated c. To find other generators that commute with P µ we look for Lorentz transformations that leave the four vector pµ invariant. . (3. ǫ) = 1 − ωαβ M αβ + i ǫα P α . check that U1 U2 corresponds with Λ1 Λ2 ). a)P µ U † (Λ.45) Note that the commutation relations among the generators M µν in Eq. (3. (3. Λµν pν ⇒ ωµν pν ⇒ ωµν = = = pµ + ω µν pν = pµ 0 ǫµνρσ pρ sσ . . the generators J± in the case of the rotation group. Taking any one of the states in an irreducible representation. . = pµ |p.43) giving the following commutation relations by equating the coefficients of ǫµ and ωµν . as part of the set.46) [K i . The generators of the translations are the (momentum) operators P µ . leads to U (Λ. a) = Λµν P ν or U † (Λ. The requirement that P µ transforms as a four-vector (for which we know the explicit behavior from the defining four-dimensional representation). Infinitesimally. ν (3. The eigenvalues of these will be the four-momentum pµ of the state.38 could have been obtained in the same way. a)P µ U (Λ.47) (3. P ) in terms of the Hamiltonian and the three-momentum operators. .Groups and their representations 21 infinitesimal form of any representation of the Poincar´ group (thus including the translations) is of e the form i U (I + ω. P ν ] [M µν . [P µ . In order to label the states in an irreducible representation we construct a maximal set of commuting operators. P j ] = [P i .48) . P j ] = i δ ij H/c2 . [K i . . [J i .42) (to decide on where the inverse is. [J i . known from non-relativistic quantum mechanics are obtained. Of the generators of the Poincar´ group we choose first of all the generators P µ . J j ] = i ǫijk J k . 3. [K i . K j ] = i ǫijk K k .44) (3. . 2 2 (3. 3. the generators M αβ no longer exclusively act in Minkowski space. other states in that representation are obtained by the action of operators outside the maximal commuting set. that commute e among themselves. . H] = [J i .5 Representations of the Poincar´ group e where pµ is a set of four arbitrary real numbers. because one then sees that by letting c → ∞ the commutation relations of the Galilei group. For instance.

. M ǫijk .51) The following properties follow from the fact that W µ is a four-vector by construction. From the algebra of the generators one finds that P 2 = Pµ P µ and W 2 = Wµ W µ (3. . (3. = = i exp − ǫµνρσ M µν pρ sσ |p. of which the components generate the little group of pµ or (for the third of the following relations) from explicit calculation [Mµν . i exp − ωµν M µν 2 i exp − ǫµνρσ M µν pρ sσ 2 (3. .50) where the Pauli-Lubanski operators W µ are given by 1 Wµ = − ǫµνρσ M νρ P σ .Groups and their representations 22 where s is an arbitrary vector of which the length and the component along pσ are irrelevant. i ǫµνρσ W ρ P σ . Wν ] = −i (gµα Wν − gνα Wµ ) . . Wα ] = [Wµ . satisfying gµν nµ (p)nν (p) = j i nµ (p) pµ i ǫµνρσ pµ nν nρ nσ i j k = = −δij . . Given the momentum four vector pµ we choose a tetrade consisting of three orthogonal spacelike unit vectors ni (p). (3. Massive particles: p2 = M 2 > 0.54) This will enable us to pick a suitable commuting ’spin’ operator.55) commute with all generators and therefore are invariants under Poincar´ transformations. 2 (3. Pν ] = [Wµ .56) . which can thus be chosen spacelike. 0. while the others have no physical significance. .49) (3.52) (3. 0.53) (3. for which we distinguish the following cases p 2 = m2 > 0 p2 = 0 pµ ≡ 0 p 2 = m2 > 0 p2 = 0 p2 < 0 p0 > 0 p0 > 0 p0 < 0 p0 < 0 Only the first two cases correspond to physical states. . These e operators are the Casimir operators of the algebra and can be used to define the representations of ↑ P+ . The elements of the little group thus are U (Λ(p)) = = and U (Λ(p))|p. . p0 > 0. 2 exp (−isµ W µ ) |p. . the third case represents the vacuum.

Having made a covariant decomposition.61) one sees that S i (p) i =− W2 .58) i. 0. The commutation relations [W i (p). n1 (p). 0.57) (i. .. 1. The explicit construction of the wave function can be done using the D-functions (analogous as the rotation functions). 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 . 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. p0 > 0. 0).62) 1 Thus W 2 has the eigenvalues −M 2 s(s + 1) with s = 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. Noting that S i (p) i 2 = 1 W µ (p)W ν (p) M2 niµ (p)niν (p). In that case the vectors ni (p) are just the space directions and W = (0. 0).60) and using the completeness relation nµ (p)nν (p) = − g µν − i i 2 i pµ pν M2 .59) (3. W i (p) = −W · ni (p)). 0. S j (p)] = −i nµ (p)nν (p) j i ǫµνρσ W ρ P σ = i ǫijk S k (p). n2 (p) and s(p) are obtained by a Lorentz boost from the reference frame one has the property ǫµνρσ sµ nν nρ pσ > 0. 0. −1) n1 (p0 ) ≡ n2 (p0 ) ≡ s(p0 ) ≡ (3. (3. s. 3. M J) with components W i = (M/2)ǫijk M jk . 2 . p. Together with the four momentum states thus can be labeled as |M. More generally (fully covariantly) one can proceed by defining S i (p) ≡ W i (p)/M and obtain [S i (p). i (3. 1 2 (3. M2 (3. (1. 1.65) . (0. 0.63) where E = p2 + M 2 and ms is the z-component of the spin −s ≤ ms ≤ +s (in steps of one). (3.e. 1. it is sufficient to choose a particular frame to investigate the coefficients in Eq.e. the S i (p) form the generators of an SU (2) subgroup that belongs to the maximal set of commuting operators. M2 (3. show that W /M can be interpreted as the intrinsic spin. |p3 |). 0). 0. ms .64) such that in an arbitrary frame where the four vectors p.Groups and their representations 23 We can write W µ (p) = 3 W i (p) nµ (p) i i=1 (3. W j (p)] = i M ǫijk W k (p). (0. .

i W 2 (p). W µ (p) = λ(p) P µ . W 2 and λ is derived from the general commutation relations for [Wµ .69) We reproduce the algebra for the massive case with W 3 (p) = M λ(p). |p. where A = exp(i ϕ · σ/2) = exp(iϕ σ · n/2). thus. We don’t know of any physical relevance. The algebra of W 1 . This algebra is (for instance) isomorphic to the one generated by rotations and translations in a 2-dimensional Euclidean plane (see exercises). Thus 2 λ(p) = J ·p .71) which is called the helicity. (3. is characterized by its momentum and the helicity.68) (3. A massless particle.72) One has P µ |0 = W µ |0 = 0 and the state is invariant under Lorentz transformations.67) (3. The vacuum: pµ = 0.73) (3.70) as a commuting set with zero eigenvalues for P 2 and W 2 .13. a)|0 = |0 . The vector W µ (p) now can be expanded W µ (p) = W 1 (p) nµ (p) + W 2 (p) nµ (p) + λ(p) pµ + W s (p) sµ (p).67 has a vanishing right hand side). But if M = 0 one has a different algebra (Eq. Thus we have P 2 . Wν ]. W 1 (p)] [λ(p).Groups and their representations The above does actually include the massive case.1 (a) Derive from the relation that defines the mapping of SU (2) matrices A into SO(3) matrices RA in Eq. λ . (3. 1 2 24 (3. implying continuous values for the spin (actually for W i = M S i ). W 2 (p)] [λ(p). −i W 1 (p). Note that angular momentum does not contribute to helicity as L · p = 0. 3. This state is in physical applications nondegenerate and is denoted by |0 . |p| (3. σ · RA a ≡ A(σ · a)A−1 ˆ (valid for any vector a). and the eigenvalues of W 1 and W 2 are set to zero (corresponding to the limit M → 0 for the W µ Wµ eigenvalues of M 2 s(s + 1) → 0). 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.66) where W · p = 0 implies that W s (p) = −(M 2 /p · s) λ(p). W 2 (3. however. The eigenvalues of W 1 and W 2 (in that case corresponding to the translations) thus can take any continuous values. W 2 (p)] = = = i M 2 λ(p). the action of RA b · RA a = = ˆ ˆ ˆ b · a cos ϕ + n · (b × a) sin ϕ + 2(b · n)(a · n) sin2 (ϕ/2) ˆ ˆ b · a + b · (ˆ × n) sin ϕ − 2 (b · a − (b · n)(a · n)) sin2 (ϕ/2) a . and gives [W 1 (p). Exercises Exercise 3. 3. which can take any integer or half-integer value. U (Λ. P µ .

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

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

2 2 σ σ J = . The matrices in SL(2. (4. 2 2 (4. Special cases are the following representations: Type I : (j. It is simply connected and forms the covering group of L↑ . M = exp i ϕ·σ 2 1 exp ± η · σ 2 = U (ϕ) H(η) . 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 .6) (4. They will be labeled by two angular momenta corresponding to the A and B groups.1 The Lorentz group and SL(2. 0) Type II : (0. 2 1 (J − i K).1) (4. This tells us how to find the representations of the group.7) From the considerations above. B j ] = [Ai . U (ϕ) H(η) (4.5) = 1 (J + i K). C) is the group of complex 2 × 2 matrices with determinant one. (j.3) (4. The group SL(2.9) (4. where we restrict (for fixed n) the parameters 0 ≤ ϕ ≤ 2π and 0 ≤ η < ∞. K = i J (A = 0). C) can be written as a product of a unitary + matrix U and a hermitean matrix H. becoming the defining representations of SL(2. 2 (4.2) This shows that the Lie algebra of the Lorentz group is identical to that of SU (2) ⊗ SU (2). B j ] = i ǫijk Ak . C): Type I (denoted M ): Type II (denoted M ): 27 J= σ σ . 0. respectively.8) ˆ ˆ ˆ with η = η n and ϕ = ϕn. j) K = −i J (B = 0). They precisely correspond to the two types of representations that we have seen before. With this choice of parameter-spaces the plus and minus signs are actually relevant. j ′ ). K = +i .Chapter 4 The Dirac equation 4. similarly as the homeomorphism between SO(3) and SU (2).10) . it also follows directly that the Lorentz group is homeomorphic with the group SL(2. i ǫijk B k . K = −i . Aj ] = [B i . C).4) (4. (4.

(ǫ∗ = ǫ. σ µ ≡ (1. one can use a boost to the frame with momentum ˆ ˆ p. 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). are not covariant!) ˜ can be used to establish the homeomorphism between L↑ and SL(2. This is not true for the two-component spinors in SL(2. Eqs 4. ǫ−1 = ǫ† = ǫT = −ǫ). η → Hη. cannot be connected by a Lorentz transformation belonging to L↑ . C) and. One defines spinors ξ and η transforming similarly under unitary rotations (U † = U −1 . Also useful is the relation H 2 (p) = σ µ pµ /M = (E + σ · p)/M . namely H(η) = exp(η · σ/2). ˜ = σ µ Λ B aµ .12) ξ → U ξ. U σ µ aµ U ˜ −1 H σ µ aµ H −1 = σ µ ΛB aµ H σ µ aµ H ˜ −1 (4. U ≡ (U † )−1 = U ) U (ϕ) = exp(i ϕ · σ/2) but differently under hermitean boosts (H † = H. ˜ The sets of four operators defined by definitions σ µ ≡ (1. Λ(ǫ) ∈ L+ .11) Considering ξ and η as spin states in the rest-frame.The Dirac equation 28 Let us investigate the defining (two-dimensional) representations of SL(2. σ).15) (4. 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. ¯ ¯ H ∗ = ǫ† Hǫ = ǫ−1 Hǫ. there exists the matrix ǫ = iσ 2 .e. if there does not exist a unitary matrix S.13) 2M (E + M ) (exercise 4. (4.16) = σ µ Λ R aµ . (4. (4. C) or L↑ .18 show that ±ǫξ ∗ transforms as a type-II (η) spinor.1). (4. The Lorentz transformations Λ(M ) and Λ(M ). while ±ǫη ∗ transforms as a type-I + spinor.18) As already discussed for SU (2). η → U η. This shows that M ∗ ≃ M . ˜ Non-equivalence of M and M One might wonder if type I and type II representations are not equivalent. (4. In fact. such that M = SM S −1 . H ≡ (H † )−1 = H −1 ). C). 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. The following relations for + rotations and boosts are useful. thus. however. Note that M ∗ = ǫ† M ǫ has its equivalent for the Lorentz transformations Λ(M ) and Λ(M ∗ ). C). H(η) = exp(−η · σ/2). −σ).19) . One has + −1 Λ(M ∗ ) = I2 Λ(M ) I2 .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. thus. ˜ (4. With E = M γ = M cosh(η) and p = M βγ n = M sinh(η) n. U σ µ aµ U −1 = σ µ ΛR aµ . ξ −→ Hξ. i.14) satisfying Tr(σ µ σ ν ) = −2 gµν and Tr(σ µ σ ν ) = 2gµν = 2 δ µν (the matrices.

1 ) and ( 1 . This provides the easiest way of seeing ˜ what is happening. i. e. −→ −r (vector). C) are suitable for representing spin 1/2 particles. C). (scalar). but to L↑ . the aij elements of a tensor will not change sign. Within the Lorentz group. (axial vector).e.g. m) =            χm uL H(p) 0 . m) = χm . but by a transformation belonging ↑ ↑ to the class P− .23) u(p. 2 2 (4. must combine the repe resentations.22) u=     η η M (Λ) 0 where M (Λ) = ǫM ∗ (Λ)ǫ−1 with ǫ given in Eq.2 Spin 1/2 representations of the Lorentz group 1 1 Both representations (0.      0   χm   uR   H(p)      .20) 4. The representing member of the class P− is the parity or space inversion operator Is . (pseudoscalar). We also have seen that the representations (0. + −   1 0    0 1  I2 =    0 0   0 0 0 0 −1 0 0 0 0 1      . (vector). It has the same effect as lowering the indices.      (4. −→ −p (vector).       = (4.   −→  M (Λ)   (4.21) In nature parity turns (often) out to be a good quantum number for elementary particle states. and the angular momentum in the rest frame is what we are familiar with as the spin of a particle. the well-known two-component spinor for a spin 1/2 particle. they can be connected. The i angular momentum operators J are represented by σ i /2. m) = η(0. ǫµνρσ will change sign. but rather the operators W i (p) should be used. Although these operators cannot be used to label the representations. 1 ) and ( 1 . Taking M (Λ) = H(p). −a).The Dirac equation 29 where The matrix I2 does not belong to L↑ . Thus M ∗ and M are not equivalent within SL(2. i. From the definition of the representations (0. For the spin 1/2 representations of the Poincar´ group including parity we. etc.17. 0) are inequivalent. we obtain for the two components of u which we will refer to as chiral right and chiral left components. 2 ) and ( 2 . the boost in Eq. For a particle at rest only angular momentum is important and we can choose ξ(0.1. The behavior is the same for classical quantities. 0) of SL(2. that have the correct commutation relations.13. a) and ˜ a = (a0 . they cannot be connected by a unitary 2 2 transformation. 0) (via operators J and K) one sees that under parity 2 2 1 1 (0. we have seen that in the rest frame W i (p)/M → J i . A parity transformation changes aµ into aµ .e. where a = (a0 . Examples (with between brackets the Euclidean assigments) are r t p H J λ(p) K −→ −λ(p) −→ −K −→ H −→ J −→ t (scalar). consider the four component spinor that transforms under a Lorentz transformation as      0  ξ   ξ  . therefore. 0). 4. ) −→ ( . 4. as we have seen in section 4. generators.

 =       (4.33) define for a four vector a the contraction a = aµ γµ / . (4. which in general is a linear equation in pµ . this can be rewritten 1 µ ν {γ . one must require that for ψ(x) the Klein-Gordon relation 2 + M 2 ψ(x) = 0 is valid for each component separately).The Dirac equation 30 with H(p) = ¯ H(p) = 2M (E + M ) E+M −σ·p 2M (E + M ) E+M +σ·p . 2 (4. γ ν } = 2 g µν .30) γ µ γ ν ∂µ ∂ν + M 2 ψ(x) = 0. (4. As in section 2 we can use Pµ = i∂µ as a representation for the momenta (translation operators) in function space. p (4.24) (4. (iγ µ ∂µ − M ) 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 ). γ } ∂µ ∂ν + M 2 ψ(x) = 0. We first want to investigate if the Dirac equation solves the problem with the negative densities and negative energies. .31) To achieve also that the energy-momentum relation p2 = M 2 is satisfied for u(p). The first indeed is solved.25) It is straightforward to eliminate χm and obtain the following constraint on the components of u. It is of a form that we also played with in Exercise 2.   (4.27 is known as the Weyl representation. Applying (iγ µ ∂µ + M ) from the left gives (4. We will discuss another explicit realization of this algebra in the next section. (4. From this one obtains the Clifford algebra for the Dirac matrices.26)        −2    uL uL H (p) 0 or explicitly in the socalled Weyl representation  −M E+σ·p       E−σ·p −M    uR      uL      = 0. (4. The explicit realization appearing in Eq. This leads to the Dirac equation for ψ(x) = u(p)e−i p·x in coordinate space. (pµ γ µ − M ) u(p) ≡ (/ − M ) u(p) = 0.28) which is a covariant (linear) first order differential equation.29) which is an explicit realization of the (momentum space) Dirac equation.32) suppressing on the RHS the identity matrix in Dirac space. The general requirements for the γ matrices are thus easily obtained. Since ∂µ ∂ν is symmetric.      0 H 2 (p)   uR   uR        . = H −1 (p). ψ † −iγ 0 ∂0 +iγ i ∂i −M 1 We ← ← = 0.7. {γ µ .

39) use ρi ⊗ σj with both ρ and σ being the standard 2 × 2 Pauli matrices.37) Using the explicit form of γ in the Weyl representation (see Eq.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  .g. but with properties such that it can be annihilated by the particle (e.35) is indeed positive and j 0 can serve as a probability density. Relying on the Pauli exclusion principle for spin 1/2 particles. 4. From the equations for ψ and ψ one immediately sees that j µ = ψγ µ ψ is a conserved probability current (exercise!). 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. From the vacuum a particle can be removed. opposite charge). The Dirac sea forms the vacuum. γ ν } = 2 g µν . where γ 0 p0 ψ = M ψ 0 −→ Eψ = M γ 0 ψ. remains. γ =ρ ⊗1= γ = iρ ⊗ σ =    0 −1 −σ k 0 2 We (4. 4.36) (4.  .27). Multiplying with γ 0 on the right and pulling it through one finds a covariant equation ψ(x) iγ µ ∂µ +M ← = 0. This hole forms an antiparticle with the same mass. We will see how this is properly implemented (also for bosons!) when quantizing fields.34) for the adjoint spinor ψ ≡ ψ † γ0 . (4. E = +M (twice) and E = −M (twice). This problem was solved by Dirac through the introduction of a negative energy sea. (4. This can most easily be seen in the particle rest frame.The Dirac equation 31 E empty levels M -M filled levels E particle M -M hole Figure 4. the one of the negative energy states. one sees that there are two positive and two negative eigenvalues. . (4.1: The Dirac sea of negative energy states and antiparticles.3 General representations of γ matrices and Dirac spinors {γ µ . The second problem encountered before. The density j 0 = ψγ 0 ψ = ψ † ψ (4.

the rotation and boost generators in Weyl representation in Eqs 4.43) S=√  . −→ Sγµ S −1 .R. = ρ ⊗ 1 =   1 0 i ǫµνρσ µ ν ρ σ γ γ γ γ .  0 (4.49) i 1 with −i S ρσ = 2 σ ρσ = − 4 [γ ρ .45) (γ5 )W.R. it is easy to see that PR/L = 1 (1 ± γ5 ) 2    1 0  . 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 . γ 2 ]. ν The latter assures Lorentz invariance of the Dirac equation (see Exercise 4.47) (4. S −1 . For example. = S(γµ )W.40) (4.3) . which in the case of the Weyl representation are just the upper and lower components.   0 γ k = −iρ2 ⊗ σ k =  k  σ  −σ k  .The Dirac equation 32 The Weyl (or chiral) representation:    0 1   . (γµ )S.42) We note that the explicit matrix taking us from the Weyl representation to the standard representation.46) (4. 2 (4. γ µ } = 0 and explicitly one has    0 1  1 . that project out chiral right/left states.9 and 4. γ σ ]. while L−1 γ µ L = Λµ γ ν .44) and an adjoint spinor defined by ψ = ψ † γ0 . 1 −1 2 For all representations one has † γµ = γ0 γµ γ0 .48) are projection operators. For instance in the Weyl representation (but valid generally). Lorentz invariance The Lorentz transformations can also be written in terms of Dirac matrices.  =ρ ⊗1=  0 −1 3 (4. (4. K 3 = M 30 =    0 iσ 3 2 4 or in general one has the transformation i ψ → Lψ = exp − ωρσ S ρσ ψ.41) (4. −→ Sψ.R. 4! (4. γ 0 = ρ1 ⊗ 1 =   1 0 γµ ψ Different representations can be related to each other by unitary transformations. 0 σ3 2 4   1  −iσ 3 0   = − i [γ 3 . is   1  1 1    (4. γ 0 ].  (γ5 )S.10 are   1  σ3 0  i   J 3 = M 12 =   = [γ 1 .R. We note that ψ → Lψ and ψ → ψ L−1 .

which is again a solution of the Dirac equation. ∂ with the solution ψ c (x) = ηc Cψ (x). One has ψ c = −ψ T C −1 .51) The existence of positive and negative energy solutions implies another symmetry in the Dirac equation.g.The Dirac equation 33 Parity There are a number of symmetries in the Dirac equation. (i/ − M ) ψ(x) = 0. γ5 ] = 0. e.R. called charge conjugation.R. The latter implies that the conjugate of a right-handed spinor. (C)W.50) Charge conjugation but with ψ p (x) ≡ ηp γ0 ψ(˜) (Exercise 4.55) (4.54) e.    0 −ǫ 0    0   −ǫ 0  .57) (4. such that T Cγµ C −1 = −γµ .59)  −→ ψ c = C ψ =    η ǫ ξ∗ . ∂ (4. parity.    −iσ 2   0 −ǫ  = . −x) and via a few manipulations obtain again the ˜ Dirac equation (i/ − M ) ψ p (x) = 0. reverse the charge of the particle. Explicitly.58) where ηc is an arbitrary (unobservable) phase.56) (i/ − M ) ψ c (x) = 0. This symmetry does not change the spin 1/2 nature. T (4. In S.   ψ= (4. in Weyl representation we find in Dirac space     T C  ξ   −ǫ η ∗     . x = (t. (C)S. ∂ (4. (or W. It is easy to convince oneself that if ψ(x) is a solution of the Dirac equation. As with parity we look for a transformation. ψR . = =   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. usually to be taken unity. for instance. Some properties of C are C −1 = C † and C T = −C.6).R. 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. x) → x = (t.) and all representations connected via a real (up to an overall phase) matrix S. =    0 ǫ iσ 2 (4. T (4. In any representation a matrix C exist. but it does. that brings ψ → ψ c . Note that we have (as expected) explicitly in Weyl x representation in Dirac space     P  ξ    −→ ψ p = γ 0 ψ =  η  .g.53) (4. ψ= (4. C is real and one has C −1 = C † = C T = −C c and [C. is a left-handed spinor.52)    η ξ one can apply space inversion.R.

v(p. s) u = . s) + v(k. s) and C v T (p. in which γ 0 is diagonal (see discussion of negative energy states in section 4. Choosing χ = −ǫχ∗ (the equivalent spin ¯ 1/2 conjugate representation). the spinors satisfy C uT (p.   ⇔ (/ + M ) v(p) = 0. the spin polarization vector in the rest frame is the starting ˆ point.60)   ψ(x) = 0. be obtained by a Lorentz transformation. where u satisfies   −σ · p  Ep − M      u(p) = 0. s′ ) = 2M δss′ . After separating positive and negative energy solutions as done in the case of the scalar field one has ψ(x) = s d3 k u(k. s).63) u(p. The explicit Dirac equation in the standard representation reads  ∂  iσ · ∇   i ∂t − M       (4. ¯ ′ ′ Explicit solutions in the standard representation are   χs       . where v satisfies   σ·p  −(Ep + M )      v(p) = 0.70) =  Ps =  ˆ 0 1∓σ·s 2 2 .1). In an arbitrary frame one has s · p = 0 and s(p) can e.The Dirac equation 34 4. ψ(x) = u±s (p) e−i p·x .     ∂ −iσ · ∇ −i ∂t − M (4.61)     σ·p −(Ep + M ) There are also two negative energy solutions. s)v(p.66) An arbitrary spin 1/2 field can be expanded in the independent solutions. s)v(p. s ) = v † (p. It is easy to check that   1  1±σ·s 1 ± γ5 / s ˆ 0    (4. p     −σ · p (Ep − M )     Ep + M     σ ·p ¯ Ep +M χs χs ¯ For a free massive particle. s)u(p. s ) = 0. s) v = . p (4. ¯ ¯ † ′ u (p. s) = v(p. the best representation to describe particles at rest is the standard representation. In that frame is a spacelike unit vector sµ = (0. with E = Ep = + p2 + M 2 . (2π)3 2Ek (4. s′ ) = 2Ep δss′ . s)v(p. s)u(p. s)u(p. s) .g.64) (4. The ¯ ¯ solutions are normalized to u(p. Note that the spinors in the negative energy modes (antiparticles) could be two different spinors.    (4. s). s)¯(p.67) It is straightforward to find projection operators for the positive and negative energy states P+ P− = s p /+M u(p. 2M 2M (4. ψ(x) = v ±s (p) ei p·x . v (p. s′ ) = −2M δss′ .68) (4. s ) = v (p.65) (4.4 Plane wave solutions Looking for positive energy solutions ∝ exp(−iEt) one finds two solutions. 2M 2M −/ + M p v(p.. s) = Ep + M      σ ·p   Ep +M χs     . s) = u(p.69) = − s In order to project out spin states. s) e−i k·x b(k.   ⇔ (/ − M ) u(p) = 0. s)¯(p.62) where χs are two independent (s = ±) two-component spinors. s) = u(p. ¯ (4. s) ei k·x d∗ (k.

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

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

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 . by letting them act on the four basis spinors u(p. to the Dirac equation and use this to show that the spinor ψ p (x) = ˜ γ0 ψ(˜). x → x.77) Exercises Excercise 4.6 Apply space-inversion. γ σ ] = σ ρσ . 1 i −i S ρσ = − [γ ρ .g. s) u p /+M = 2M 2M both sides are projection operators. s)¯(p. where x = (t. Exercise 4. ν µ µ Look at the infinitesimal forms Λµ = gν + ων and L = 1 − (i/2)ωρσ S ρσ to obtain ν [γ µ .The Dirac equation 37 The trace is linear and can be split up in parts containing traces with up to four gamma-matrices.2 Show that j µ = ψγ µ ψ is a conserved current if ψ en ψ are solutions of the Dirac equation Exercise 4.3 Imposing Lorentz invariance on the Dirac equation can be achieved by requiring L−1 γ µ L = Λµ γ ν . γν } = 2gµν that γµ / γ µ a = Tr(//) = ab µn µ1 Tr(γ · · · γ ) = −2 /.g. s) and v(p. of which only the traces of four and two gamma-matrices are nonzero. e.4 Show that in P+ ≡ s=±1/2 u(p. −x) is also a solution of the Dirac equation. s). (4.5 Show starting from the relation {γµ . by proving that H 2 (p) = exp (η · σ) = (E + σ · p)/M = σ µ pµ /M.1 Prove Eq. Excercise 4. a 4 a · b. Show that the equality holds. i S ρσ ] = g ρµ γ σ − g σµ γ ρ . 4. 0 if n is odd. 2M (E + M ) ˆ ˆ where E = M γ = M cosh(η) and p = M βγ n = M sinh(η) n. 4 2 Exercise 4. x ˜ . Do this e. H(p) = exp η·σ 2 = M +E+σ·p . ˜ Exercise 4.13.

Do you understand why? The calculation of the life-time is completed in chapter 11.66 MeV. . A condition for confining the fermion in the cavity is i / ψ = ψ (where n is the surface normal). |M (π − → µ− νµ )|2 ¯ Actually if the lepton would be massless. namely i / ψ = eiαγ5 ψ. Note: the outcome shows that the decay into muons is much more likely that the decay into electrons (mπ = 139. we want to confine this solution to a spherical cavity with radius R. Show that this n condition is sufficient to guarantee that nµ j µ = 0. Calculate for now just |M |2 . i.511 MeV). Calculate the value of k for M R = 0. t) = N     k ˆ i E+M σ · r j1 (kr)χm    −i Et   e .7 (a) Show that (in standard representation)  j0 (kr)χm    ψ(x. The conditions for ψ and ψ imply that nµ j µ = ψ/ ψ = 0. (d) Plot for these cases the vector densities ψ † ψ and the scalar densities ψψ as a function of r/R.The Dirac equation 38 Exercise 4. For such a cavity nµ = (0. We will not pursue this chiral freedom at this point and take α = 0.e.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 ). r). Note: To prove this. Show that the ratio is |M (π − → e− νe )|2 ¯ ≈ 54. 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 ). M R = 1.5 and M → ∞. me = 0. you need to determine the corresponding confinement condition for ψ.    is a stationary solution of the (free) Dirac equation. the decay would even be forbidden. The pion has spin 0. which is the most important factor needed to find the decay width (inverse lifetime).57 MeV. Excercise 4. there is no current flowing through the surface of the cavity. ν p Note: how such parametrizations work will be explained in chapter 8. n (c) Apply the confinement condition to find the value of k for the lowest eigenmode in the spherical ˆ cavity. 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. 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.8 × 10−6 . mµ = 105. (b) Next. the lepton and neutrino have spin 1/2 and are ν described by momentum-space spinors. n It turns out that there is actually a whole family of conditions that provide confinement.

1) The Lorentz condition implies that k µ ǫµ (k) = 0.6) where Aµ is the four-vector potential (5.4) λ=0. 0). there are then indeed three independent possibilities.7) . M2 (5. These can describe spin 1 states in the rest-frame. namely ǫµ (k0 ) = (0.± 5. λ) . one has ǫ(λ) (k) ǫ(λ)∗ (k) = −gµν + µ ν kµ kν . with ∂µ V µ = 0 (Lorentz condition).2 The electromagnetic field F µν = = ∂ µ Aν − ∂ ν Aµ  1 −E 2  0 −E  1  E  0 −B 3   2  E 3  B 0   3 E −B 2 B 1 Aµ = (φ.5)  −E    B2   .2) (5.Chapter 5 Vector fields and Maxwell equations 5. (2 + M 2 )Vµ = 0.± (λ) d3 k ǫ(λ) (k) e−ik·x c(k.1 Fields for spin 1 In section 3. In the restframe.  −B 1    0 3 With the electromagnetic field tensor (5. These are solutions of the Klein-Gordon equation with in addition a condition. λ) + ǫ(λ)∗ (k) eik·x c∗ (k.2 we have discussed the vectors ǫλ (with λ = 0 and ±1) as basis states of the spin 1 representation. 39 (5. The covariant generalization is to describe them with Vµ (x) = ǫµ (k) e±ik·x . A). A real vector field Vµ (x) then can be expanded in modes as Vµ (x) = λ=0. ǫλ ). The vectorfield is therefore (λ) suited to describe spin 1. (5.3) Since the vectors ǫµ (k) together with the momentum k µ form a complete set of four-vectors. just as the two-component Dirac spinors did for spin 1/2. µ (2π)3 2E µ (5. ±(1) are obtained by boosting the restframe vectors. where k = k0 = (M. In an arbitrary frame ǫµ (k) with λ = 0.

5. it does not eliminate the freedom of gauge transformations but they are now restricted to Aµ → Aµ + ∂µ χ with 2χ = 0.17) (radiation. the antisymmetry of F µν implies that ˜ ∂µ F µν = 0 or ∂λ Fµν + ∂µ Fνλ + ∂ν Fλµ = 0. transverse or Coulomb gauge). −i. one finds for a conserved current and λ = 0 2∂µ Aµ = 0. only two independent vectors remain ǫ(±) µ (k) = (0. 0).18) corresponding to the helicity states λ = ±1 of the massless photon field.12) (5. One possible choice is the gauge A0 = 0 (5.8) (5. AP (x) = Aµ (˜). 40 (5. . for M = 0. Therefore if k µ = (|k|. ∓1.9) 1 ˜ where F µν ≡ − 2 ǫµνρσ Fρσ is the tensor where E → B and B → −E. i.5. Furthermore. (5. Fµν −→ Fµν + (∂µ ∂ν χ − ∂ν ∂µ χ) = Fµν . in accordance with what we found in section 3.14) of which the solutions give the Li´nard-Wiechert potentials. 0. (5.13) Hence.e. (5. 0. in which case one has 2Aµ = jµ .11) This equation is not affected by a gauge transformation. if in Eq. Note that the incorporation of the Lorentz condition can also be done for the case of the massive vector field. if ∂µ Aµ = 0 for large times |t|. shows that the electromagnetic fields correspond to the case of a massless spin 1 field. moree over. As remarked above.Vector fields and Maxwell equations and the (conserved) current j µ = (ρ.10) the electric and magnetic fields are unchanged. Under a gauge transformation Aµ −→ Aµ + ∂µ χ. For electrodynamics one has furthermore the freedom of gauge transformations.15 the d’Alembertian is replaced by 2 + M 2 . In that case one also has ∇·A = 0 or ki ǫi (k) = 0. it will vanish at all times.16) (5. ˜ x Note that via parity transformation one obtains another solution. ∂µ Aµ = 0. there is still a gauge freedom left (Aµ → Aµ + ∂µ χ with 2χ = 0). The equations of motion for the fields Aµ become 2Aµ − ∂µ (∂ν Aν ) = jµ . j) the Maxwell equations read ∂µ F µν = j ν . The Lorenz condition can actually be incorporated by changing the equations of motion to 2Aµ + (λ − 1)∂µ (∂ν Aν ) = jµ . |k|). 2Aµ = 0. The equation in vacuum.15) (5. It states the absence of magnetic charges and Faraday’s law. (5. while the fact µ C ∗ that the Minkowski space is real implies that Aµ (x) = Aµ (x) = Aµ (x). Taking the divergence of this equiation. The gauge freedom can first of all be used to impose the Lorentz condition for the electromagnetic field. Although the Lorenz condition is a constraint. (5.

illustrated in fig. A= By Bx Br ϕ= − ˆ . F = e E + ev × B (5.3 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.19) are gauge invariant. sin ϕ. In the presence of an electromagnetic field Aµ an additional phase difference is observed. Nevertheless the significance of Aµ is shown in the Aharonov-Bohm experiment. It is however. while the electromagnetic field Aµ is not. 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 . .Vector fields and Maxwell equations 41 L ∆x d a (infinitely) long solenoid Figure 5. This causes an interference pattern as a function of ∆x. but A = 0 (hence there is structure in the vacuum). . nevertheless. This phase difference is actually observed. . ˆ B = 0. but where E = B = 0. may look a bit akward. 2 2 2 BR2 BR2 y BR2 x A= . λ L − λ λ or ∆x = (L− λ/d)δ.20) e (∇ × A) · ds = B · ds = where Φ is the flux through the solenoid. Inside solenoid: Outside solenoid: B = B z. This occurs in the region outside the solenoid. The situation. The phase difference between the electrons travelling two different paths is given by δ = 2π a a ∆x d = −≈ . cos ϕ.0 . 5.1. = exp −i The phase difference between the two paths of the electron then are given by ∆δ = − = e 1 A · dr + e 2 A · dr = e e A · dr e Φ. 0) and ϕ = (− sin ϕ. To be precise. ϕ= − ˆ 2r 2r2 2r2 where r = (cos ϕ.0 . The energy density 2 (E 2 + B 2 ) = 0 (hence a vacuum). 0).1: The Aharonov-Bohm experiment 5. In this experiment an observable phase difference is measured for electrons moving through field-free space (E = B = 0 but Aµ = 0). nothing more than a manisfes1 tation of a nontrivial vacuum (a la the Dirac sea for fermions). (5. where Aµ = 0.

Therefore the gauge transformation must be uniquely defined in the space one is working in.21. 5.2: The topology of the space in the Aharonov-Bohm experiment. nonobservable phases ϕ = 2π n. Therefore space outside the solenoid doesn’t care that χ is multi-valued. The difference here is that we are working in a space with a ’defect’ (the infinitely long solenoid). is multivalued as ϕ = 0 and ϕ = 2π are the same point in space. Consider a superconductor.Vector fields and Maxwell equations U(1) S (space with solenoid) 42 class 0 ei φ class 1 Π1(S) = Z class 2 Figure 5. the number of times they go around the defect) cannot be continuously deformed into one another.e. has been employed by Dirac in constructing magnetic monopoles in electrodynamics 2 Actually. The observable phase going around in general is ∆δ = e A · dr = e ∇χ · dr = e χ|ϕ=0 .21) (ϕ being the azimuthal angle). minimizing the space occupied by B fields. it can be obtained from the situation A = 0 by a gauge transformation. 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. Another situation in which the topology of space can be used is in the case of superconductors. however. In such a space loops with different winding number (i. implying that it must be single-valued1 . The AharonovBohm experiment shows a realization of this possibility. A = ∇χ with χ= BR2 ϕ 2 (5.e. 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. 1 The occurrence of nontrivial possibilities. 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. This situation that A can be written as ∇χ is always true when ∇ × A = 0. forming a simple (connected) space. giving no observable phase2 . Below the critical temperature the magnetic field is squeezed out of the superconducting material. Now back to the Aharonov-Bohm experiment. If this would be be happening in empty space one would be in trouble. i. ϕ=2π The function χ in Eq. the relevant charge turns out to be 2e because the electrons appear in Cooper pairs . organizing itself in tiny flux tubes (Abrikosov strings).

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

recall classical mechanics with the action t2 S= t1 dt L(x. and the total change x′ (t′ ) = x(t) + ∆x(t). x) ˙ (6. ˙ p= ∂L .7) The quantity ∂L/∂ x plays a special role and is known as the canonical momentum.6) The first term leads to the Euler-Lagrange equations.5) with ∆x(t) = δx(t) + x(t) δτ .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 .1 Euler-Lagrange equations In the previous chapter we have seen the equations for scalar fields (Klein-Gordon equation). thus t′ = t + δτ. ∂x (6. ∂x ˙ (6. x) = K − V = ˙ 1 mx2 − V (x). These equations can be obtained from a lagrangian using the action principle.8) 44 . As an example. Dirac fields (Dirac equation) and massless vector fields (Maxwell equations) and corresponding to these fields conserved currents describing the probability and probability current.3) (6. The requirement δS = 0 with fixed boundaries x(t1 ) = x1 and x(t2 ) = ˙ x2 leads to t2 ′ (6. ˙ 2 (6. d dt ∂L ∂x ˙ = ∂L .Chapter 6 Classical lagrangian field theory 6. (6. t1 (6.1) and as an example the lagrangian L(x.2) The principle of minimal action looks for a stationary action under variations in the coordinates and time. x (t) = x(t) + δx(t).

t2 ).t2 ) d3 x . δ(∂µ φ) (6. gives rise to a conserved quantity (at least if H is invariant!) Q=p dx dλ −H dt dλ . = φ(x) + δφ(x) (6. ∆x = (dx/dλ)λ=0 δλ and δt = (dt/dλ)λ=0 δλ.9) which is done because the first term (multiplying ∆x. . ∂µ φ. R (6. . .14) (6. which in classical mechanics vanishes at the boundary) does not play a role. In general any continuous transformation. x′ ) − d4 x L (φ.18) 1 Taking a functional derivative.e. ∂µ φ(x)). . dσµ . 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 . (6. x). We will come back to it in a bit more formal way in section 9. think for instance of a temperature or density distribution or of an electromagnetic field). . ∂µ φ′ . (R3 . x) and an action t2 S[φ] = t1 dt L = dt d3 x L (φ(x). The resulting variation of the action is δS = R µ ′ = xµ + δxµ . x) ≡ p x − L . also indicated by ∂R (a more general volume in four-dimensional space-time with some boundary ∂R can also be considered).11) λ=0 of which p and H are the simplest examples related to space and time translations.10) One sees that invariance under time translations requires that H(t1 ) = H(t2 ). L (φ(x).Classical lagrangian field theory 45 For the lagrangian specified above this leads directly to Newton’s law p = m¨ = −∂V /∂x.2.t1 ) (6. indicated as x′µ φ (x) or combined φ′ (x′ ) = φ(x) + ∆φ(x). t1 (6. + ∂L ∆x − H δτ ∂x ˙ t2 .6 can be rewritten as δS = . t1 ) and R3 . ∂µ φ(x). . . 6. . their derivatives and possibly on the position.15) d4 x′ L (φ′ . ∂µ φ(x)) = R d4 x L (φ(x).13) (6. The hamiltonian H is defined by H(p. − d3 x . Variations in the action can come from the coordinates or the fields.17) Furthermore the variations δφ and δ∂µ φ contribute to δS. with ∆φ(x) = δφ(x) + (∂µ φ)δx .12) Here R indicates a space-time volume bounded by (R3 . λ=0 (6. ˙ x The second term in Eq. ˙ (6. In classical field theory one proceeds in complete analogy but using functions depending on space and time (classical fields. . indicated with δF [φ]/δφ should pose no problems.16) The change in variables x → x′ in the integration volume involves a surface variation of the form dσµ L δxµ . Consider a lagrangian density L which depends on these functions. = ∂R (R3 . i. . H is a conserved quantity. ∂R Note for the specific choice of the surface for constant times t1 and t2 . but which in 3 dimensions includes conserved quantities related to rotations and boosts.

20) (6. δL δ(∂µ ψ) δL δψ one obtains immediately i / −M ∂ → i = − γ µψ 2 i → = ∂ / ψ − M ψ. ∂ (6.21) which differ only by surface terms. 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).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. (6. The integrand of the first term must vanish. (6. One easily obtains (2 + M 2 )φ(x) = 0. 2 ∗ (6. The scalar field It is straightforward to derive the equations of motion for a real scalar field φ from the lagrangian densities.24) (6.Classical lagrangian field theory 46 With for the situation of classical fields all variations of the fields and coordinates at the surface vanishing.19) δ(∂µ φ) δφ 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.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. L = 1 1 ∂µ φ∂ µ φ − M 2 φ2 2 2 1 µ = − φ ∂µ ∂ + M 2 φ. For the complex scalar field one conventionally uses L = = ∂µ φ∗ ∂ µ φ − M 2 φ∗ φ −φ∗ ∂µ ∂ µ + M 2 φ. (2 + M )φ (x) = 0. the second term is irrelevant. the Dirac field (spin 1/2) and the vector field (spin 1) can be found.29) .23) (6. leading to the Euler-Lagrange equations. (6. δL δL ∂µ = .25) (6. Using the variations in ψ (in the symmetric form).2 Lagrangians for spin 0. leading to (2 + M 2 )φ(x) = 0. 2 (6.

34)  (6. but note the factor 1/2 as compared to the Dirac lagrangian. (6. The reasons for Grassmann variables will become clear in the next chapter. or equivalently separate the field into right. Υi / Υ − ∂ 4 2   T  −ǫ η ∗  c  = ΥR . 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. . ψL ≡ (ψL )c = C ψL =    0 (6.36) (6.39) 2 This term is written down with η being an anticommuting Grassmann number for which (αβ)∗ = β ∗ α∗ = −α∗ β ∗ .7).g. 1 M η † ǫη ∗ − M ∗ η T ǫη . In that case one finds trivially L = ↔ ↔ 1 1 ψR i / ψR + ψL i / ψL − M (ψR ψL + ψL ψR ).g.38) (6.Classical lagrangian field theory 47 and similarly from the variation with respect to ψ ψ i / +M ∂ ← = 0. Using the twospinors ξ and η.33) We now introduce a ’real’ (four-component) spinor satisfying Υc = Υ with ψL . which comes because we in essence use ’real’ spinors. 2 αβ = −βα. (6. Peshkin and Schroeder or Exercise 12.32) There exists another possibility to write down a mass term with only one kind of fields. 2 (6. We note that of which the left part coincides   .and lefthanded ones. 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 .    ∗  0    ⇒ Υ ≡  −ǫ η  ΥL = ψL =    η η ∗ for which η0 = ǫ η0 . the mass term in the Dirac lagrangian 6. 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.35) Since the kinetic term in L separates naturally in left and right parts (or ξ and η).30) It is often useful to link to the two-component spinors ξ and η which we started with in chapter 4. namely2 LM (Majorana) = + (6.31) showing e. that the lagrangian separates into two independent parts for M = 0.37) This gives an expression with a mass term that is actually of the same form as the Dirac mass term. and thus (β ∗ α)∗ = α∗ β = −βα∗ .31 is given by LM (Dirac) = −M ξ † η + η † ξ . ∂ ∂ 2 2 (6. (6. The Majorana case is in fact more general. L = = ↔ 1 1 c c M ψL ψL + M ∗ ψL ψL ψL i / ψL − ∂ 2 2 ↔ 1 1 M ΥR ΥL + M ∗ ΥL ΥR .

. (6. ∆φ = (dφ/dλ)λ=0 δλ and δxµ = (dxµ /dλ)λ=0 δλ.18. which in a consistent picture precisely generate the symmetries.42) Θµν = The variation δS thus can be expressed as δS with J µ (x) = − = δL ∂ ν φ − L g µν . consider U (1) transformations of the Dirac field proportional to the charge e..σ1 with σ = (R3 . δ(∂µ φ) (6. discussed in the next chapter. In this case. . (6. As an example for the classical field. these conserved quantities become operators. one immediately obtains δL = −∂ µ V ν + ∂ ν V µ − λ g µν (∂ρ V ρ ) = −F µν − λ g µν (∂ρ V ρ ) δ(∂µ Vν ) leading to the equations of motion ∂µ F µν + M 2 V ν + λ ∂ ν (∂ · V ) = 2 + M 2 V ν − (1 − λ)∂ ν (∂ · V ) = 0. we get (omitting the parameter δλ) dxν dφ δL + Θµν (x) . In particular when the lagrangian is invariant under symmetries.3 Symmetries and conserved (Noether) currents Next.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 . (6. These do not affect the dynamics and will give rise to conserved currents. Q(t1 ) = Q(t2 ). t). the presence of a symmetry that leaves the lagrangian invariant.45) J µ (x) = − δ(∂µ φ) dλ λ=0 dλ λ=0 Thus considering ∂R = σ2 . + ∂R dσµ dσµ ∂R ≡ where d x . ψ(x) → ei eΛ ψ(x) or ∆ψ(x) = i eΛ ψ(x). implying ∂µ F µν + M 2 V ν = 0 and ∂µ V µ = 0.40) 6. we consider the second term in Eq. Q= d3 x J 0 (x. δ(∂µ φ) For continuous transformations. ∂µ J µ (x) = 0.Classical lagrangian field theory 48 where Fµν = ∂µ Vν − ∂ν Vµ .44) δL ∆φ + Θµν (x)δxν . Returning to δS the surface term is rewritten to δS = R d4 x . requires the presence of a conserved quantity.43) F (σ1 ) − F (σ2 ) = dσµ J µ (x) = ∂R R d4 x ∂µ J µ (x)..41) (6. δ(∂µ φ) (6. 6.+ R 4 δL δL ∆φ − ∂ ν φ − L g µν δxν δ(∂µ φ) δ(∂µ φ) δL ∆φ − Θµν (x)δxν . one can consider the variations at the initial or final surface. t). which is the space-integral over the zero-component of a conserved current. (6. (6.47) .

In the next section we will see the quantity show up as the charge operator. (6. = 0. (ω µν antisymmetric) 1 ωρσ (−i S ρσ )ij φj (x). (6. 6.51) (6. 2 (6.50) (6.48) For the complex scalar field the U (1) transformations leave the lagrangian invariant and lead to the conserved current (see Exercise 6. . t) is the conserved charge (Q = 0). H = d3 x Θ00 (x).57) For the fields the behavior under Lorentz transformations has been the subject of previous chapters. From Noether’s theorem one sees that the current Θµν ǫν is conserved. For quantized fields this will become the expressions of the hamiltonian and the momentum operators in terms of the fields. we obtain (since δxµ = 0). (6. including e the space-time translations and the Lorentz transformations. The integral over the zero-component. 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. 1 (6. e.54) (6. (6.g.4 Space-time symmetries One kind of symmetries that leave the lagrangian invariant are the Poincar´ transformations. ↔ (6.Classical lagrangian field theory From the lagrangian for the Dirac field. = −(∂µ φ)δxµ = −ǫµ ∂µ φ(x). Translations µ Under translations (generated by Pop ) we have δxµ ∆φ(x) δφ(x) = ǫµ .3) j µ = i e φ∗ ∂µ φ. Summarizing.53) These are the energy and momentum.55) Note that the coordinate transformations can be written in a form similar to that for the fields (matrices in internal space). ˙ Q = d3 x j 0 (x. Lorentz transformations In this case the transformation of the coordinates and fields are written as δxµ ∆φi (x) = = ω µν xν .56) δxµ = ωρσ (aρσ )µν xν 2 with (aρσ )µν = aρσ µν = g ρµ g σν − g σµ g ρν .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. Therefore there are four conserved currents Θµν (ν labeling the currents!) and four conserved quantities Pν = dσµ Θµν = d3 x Θ0ν .49) These currents are conserved as discussed already in chapter 2.

70) . • Dirac field ψ: −i Sρσ = −(1/4)[γρ . In general this tensor is not symmetric. In some applications (specifically coupling to gravity) it is advantageous to have an equivalent current that is symmetric.66) In section 6.62) with Gρµν = (H ρµν + H µνρ + H νµρ )/2. 2 (6. Any lagrangian containing derivatives.68) (6.69) i eΛ(x) φ (x). (6.5 (Abelian) gauge theories φ(x) → ei eΛ φ(x). φ(x) φ (x) ∂µ φ(x) ∗ → e → e → ei eΛ(x) φ(x).63) µν = ∂µ Θ . is not invariant under local gauge transformations.58) (6.65) M µρσ = T µρ xσ − T µσ xρ . The lagrangians which we have considered sofar are invariant under global gauge transformations and corresponding to this there exist a conserved Noether current.67) i.3). (6. ∂µ T µν (6. • Vector field Aµ : −i Sρσ = aρσ .60) Therefore there are six conserved currents M µρσ labeled by ρ and σ (antisymmetric) and corresponding to it there exist conserved quantities M ρσ = d3 x M 0ρσ .3 we have seen global transformations or gauge transformations of the first kind.64) (6.6) T µν = T νµ . where H ρµν is the quantity appearing in M ρµν . (6. in which the U (1) transformation involves an angle e Λ. 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 . however. the angle of the transformation depends on the space-time point x. one has a tensor that satisfies (exercise 6.Classical lagrangian field theory 50 • Scalar field φ: −i Sρσ = 0. ∂µ φ(x) + i e ∂µ Λ(x) e (6. Gauge transformations of the second kind or local gauge transformations are transformations of the type φ(x) → ei eΛ(x) φ(x). independent of x. −i eΛ(x) ∗ i eΛ(x) (6.61) A final note concernes the symmetry properties of Θµν .e. φ(x). (6. e.g.59) (6. Defining T µν = Θµν − ∂ρ Gρµν . γσ ] This result for the Dirac field has been discussed in chapter 3 (see also Exercise 4. 6. (6.

72) (6. (i/ − M ) ψ = (i/ − e/ − M ) ψ = 0. (6. 2 i ∂ / ψ − M ψ − e/ ψ A 2 giving the Dirac equation in an electromagnetic field.73) (6. D ∂ A (6.78) .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). Dµ φ) − Fµν F µν .76) We note first of all that the coupling of the Dirac field (electron) to the vector field (photon) can be written in the familiar interaction form Lint = −e ψγ µ ψ Aµ = −e j µ Aµ . −e j µ Aµ = −e ρφ + e j · A. i L = ψγ µ ∂µ ψ − (∂µ ψ)γ µ ψ − M ψψ. (6.Classical lagrangian field theory 51 where the last term spoils gauge invariance. 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. (6. For this purpose it is necessary to introduce a vector field Aµ . 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 . The required transformation for Dµ then demands Dµ φ(x) = = (∂µ + i eAµ (x))φ(x) ei eΛ ∂µ + i e(A′ + ∂µ Λ) φ µ → ei eΛ ∂µ φ + i e (∂µ Λ) ei eΛ φ + i eA′ ei eΛ φ µ ≡ ei eΛ (∂µ + i eAµ ) φ. ∂ 2 4 (6. 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.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 . ∂µ φ) =⇒ L (φ. As an example consider the Dirac lagrangian. 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).71) Thus the covariant derivative has the correct transformation behavior provided Aµ → Aµ − ∂µ Λ. 4 (6. 1 L (φ. Dµ φ(x) ≡ (∂µ + i eAµ (x))φ(x).

. −eψγ ν ψ. ∂µ F µν = j ν . ∂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. where j µ = e ψγ µ ψ. Exercise 6. giving the Maxwell equation coupling to the electromagnetic current. 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.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. ↔ 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. ∂µ F µν = j ν . give the equation which is satisfied by ψ c = Cψ . φ → eiΛ φ. (c) Show that the interaction term is invariant under gauge transformations only if the current jµ is conserved. and deduce what is the charge of the antiparticle as compared to a particle. i. Exercise 6.2 (a) Show that the homogeneous Maxwell equations can be derived from the lagrangian density 1 L = − Fµν F µν .79) Exercises Exercise 6. (6.e.Classical lagrangian field theory 52 For the photon the equations of motion follow from δL δ(∂µ Aν ) δL δAν = = −F µν .1 (a) Show that the Klein-Gordon equation for the real scalar field can be derived from the lagrangian density 1 1 L = ∂µ φ∂ µ φ − M 2 φ2 . [iγ µ (∂µ − i e Aµ ) − M ]ψ = 0.

t). 2me Exercise 6.5 Show that the Dirac equation for an electron with charge −e in an external electromagnetic field Aµ = (A0 . finally note that it is sufficient to show the last equation for dσµ M µρσ . 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 = 2.4. 2M 2M with Enr ≡ E − M ≪ M . A) (see also exercise 6.6 (optional) Prove the properties of the tensor T µν in section 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 . e 1 (−i∇ + eA)2 + σ · B − eA0 ψu (r. in order to show that Θρσ − Θσρ = ∂µ H µρσ . Hints: it may be useful to realize that Gµρσ is antisymmetric in first two indices. t) = Enr ψu (r. This relation is known as the Pauli equation.Classical lagrangian field theory 53 Exercise 6. . Comment: this relation is known as the Pauli equation. use ∂µ T µν = 0 and ∂µ M µρσ = 0. Hint ≡ −gs µe · B = −gs Qe s · B.

dq The basic Poisson bracket is actually the one between q and its canonical momentum p. ˙ L(q. 6.Chapter 7 Quantization of fields 7. An immediate generalization for fields can be obtained by considering them as coordinates. Quantization of the theory is achieved by imposing canonical commutation relations between q and p.5) ∆ x ∆3 x 54 . dλ Examples are dO = [O.3) For variations δO(p. starting from the lagrangian L(q. is the Poisson bracket [A. (7. p] = i. in our example p = mq. ˙ 2 (7.4) with a possible realization as operators in the Hilbert space of (coordinate space) wave functions through qop ψ(q) = qψ(q) and pop ψ(q) = −idψ/dq. p)) = ˙ ˙ 2m where the (canonical) momentum p = ∂L/∂ q. This is nothing else as representing the classical Poisson algebra as operators in the Hilbert space.2) H(p. however.11) found through Noether’s theorem. dq dp dq dp (7. namely [q. [q. q) also considered in the previous chapter. H]P dt and dO = [O. t). Further. The bilinear product for the conserved quantities. (7. Actually the conserved quantities satisfy a similar algebra as the Lie algebras for continuous symmetries that they are obtained from. q) = ˙ 1 mq 2 − V (q). q) ≈ (dO/dλ)λ=0 δλ under symmetry transformations one has dO = [O.1) The Hamiltonian (also corresponding to a conserved quantity because of time translation invariance) is given by p2 + V (q). q) = p q − L(q. there are conserved ˙ ˙ quantities Q (Eq.1 Canonical quantization We will first recall the example of classical mechanics for one coordinate q(t). 1 qx (t) = 3 d3 x φ(x. (7. p]P = 1. q(q. p]P . labeled by the position. B]P = dB dA dA dB − . Q]P .

The lagrangian is given by L = = x In order to construct the hamiltonian it is necessary to find the canonical momenta. 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. i U † (Λ. Φ(f ) = d4 x φ(x) f (x).6) ∆3 x L x (qx .8) (7. t)] = 0. t). • Canonical commutation relations The quantization of the theory is achieved by imposing the canonical quantization condition [qx (t). Sometimes it is useful to keep in mind that. As an example.14) In fact. δ∂0 φ) 1 1 1 = Θ00 (x) = (∂0 φ)2 + (∇φ)2 + M 2 φ2 . The hamiltonian then is H= x px qx − L = ˙ = x where H (x) = δL ˙ φ(x) − L (x) = Θ00 (x). qx . for symmetry transformations. for the scalar field theory.2.7) d3 x L (φ(x). discussed in the section 6. • Poincar´ invariance e The full variation of the fields. the fields φ(x) can be considered as regular operators in the Hilbert space after smearing with a test function f .9) Note that this indeed corresponds to the zero-zero component (Θ00 ) of the conserved energy-momentum stress tensor Θµν . (7. Π(y. (7. t)] = [Π(x. φ. ˙ δ φ(x) (7. t).13) Π(x) H (x) For the quantization procedure we can formulate a number of basic axioms of quantum field theory. formally. qx+∆x ) .16) . ∂ qx ˙ ∂ qx ˙ (7. the discretization procedure above is an explicit example. ∇φ) (7.Quantization of fields 55 etc. py (t)] = i δxy or for the fields φ(x) and Π(x) the socalled equal time commutation relations [φ(x. ˙ where Πx (t) is obtained from the continuous field Π(x) ≡ δL /δ(∂0 φ).12) (7.15) with furthermore the relations [φ(x. (7.11) (7. t). 2 2 2 (7.10) x ∆3 x [Πx qx − L x ] ˙ ∆3 x Hx (7. The essential conditions to consistently quantize a theory then are the following. ∂µ φ(x)) = ˙ d3 x L (φ. ∆φ. px (t) = ∂L ∂Lx = ∆3 x = ∆3 x Πx (t). This implies specific transformation properties in the Hilbert space for the quantum fields φ. albeit with ’sharp’ functions. Π(y. a) = Rj (Λ−1 ) φj (Λ−1 x − a). φ(y. t)] = i δ 3 (x − y). a) φi (x)U (Λ.

The measurements cannot influence each other or [Φ(f ). j 2 one has [φi (x). a† ] = 1. a† ] = 0. Φ(g)] = 0.18) with S µν given in 6. P ] = i. 2 i ωµν (S µν )i .24) It is straightforward to find the commutation relations between N and a and a† .20) where the coordinate Q and the momentum P satisfy the canonical commutation relations [Q. [Q. In simplified form the hamiltonian is given by H= 1 2 1 2 2 P + ω Q .4. P ] = 0 (7. Mµν ] = i i∂µ φi (x).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.22) it is straightforward to check that the commutation relations between Q and P are equivalent with the commutation relations [a. • Causality Operators Φ(f ) and Φ(g) for which the test functions are space-like separated can be measured simultaneously (macroscopic causality). Pµ ] = [φ (x). (7.25) . φ(y)] = 0 if (x − y)2 < 0.17) (Sµν )i φj (x). 2 2 (7. i(xµ ∂ν − xν ∂µ )φ (x) + i (7. j (7. (7. P ] 2 (7. Q] = [P. Microscopic causality implies local commutativity. and [N. H = = ω ω P P Q−i √ Q+i√ 2 2 2ω 2ω 1 1 =ω N+ . sometimes referred to as second quantization. ω a† a + 2 2 ω i − [Q. 1 Q = √ (a + a† ) 2ω and P = −i ω (a − a† ) 2 (7.23) The hamiltonian in this case can be expressed in the number operator N = a† a.21) Writing Q and P in terms of creation (a† ) and annihilation (a) operators. [N.19) 7. (7. a] = −a.Quantization of fields 56 or if U (Λ. a] = [a† . [φ(x). 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. [a. a† ] = a† .

The quantization of the field is achieved by quantizing the coefficients in the Fourier expansion. [φ(x).Quantization of fields 57 Defining states |n as eigenstates of N with eigenvalue n. but one needs to realize that in that case k 0 = Ek = k2 + M 2 . d3 x φ(x) i ∂0 e−i k ·x ↔ ′ ↔ (7.32) (7.30) [φ(x). a† (k ′ )] = (2π)3 2Ek δ 3 (k − k′ ) [a(k). e. (7.29) It is straightforward to prove that the equal time commutation relations between φ(x) and Π(x′ ) are equivalent with ’harmonic oscillator . φ(x) = (2π)3 2Ek where the Fourier coefficients a(k) and a† (k) are now operators.e.g. which we have split into positive and negative energy pieces with (complex) coefficients a(k) and a∗ (k) multiplying them.3 The real scalar field We have expanded the (classical) field in plane wave solutions.4 for the classical field) a(k) a† (k ′ ) = = d3 x ei k·x i ∂0 φ(x). and we see that a state |0 must exist for which N |0 = a|0 = 0.e. H = = = d3 x 1 1 1 (∂0 φ)2 + (∇φ)2 + M 2 φ2 2 2 2 d3 k Ek † a (k)a(k) + a(k)a† (k) (2π)3 2Ek 2 d3 k Ek N (k) + Evac . i. the real scalar field φ(x) becomes d3 k (7. 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. From the normalizations one obtains a |n = √ n + 1 |n + 1 and a|n = n |n − 1 . φ(x )]x0 =x′0 = [Π(x). N |n = n|n one finds N a† |n N a|n = = (n + 1) a† |n . † √ a and a act as raising and lowering operators. The canonical momentum becomes −i d3 k ˙ Π(x) = φ(x) = a(k) e−i k·x − a† (k) ei k·x . Π(x )]x0 =x′0 = 0. a† (k ′ )] = 0. 7.27) 2 (2π)3 It is easy to check that these equations can be inverted (see Exercise 2. is equivalent with [a(k).like’ commutation relations between a(k) and a† (k ′ ).26) a(k) e−i k·x + a† (k) ei k·x . (2π)3 2Ek (7.31) The hamiltonian can be rewritten in terms of a number operator N (k) = N (k) = a† (k)a(k). Note that we will often write a(k) or a† (k). a(k ′ )] = [a† (k). (n − 1) a|n † i. and (7. Π(x′ )]x0 =x′0 = i δ 3 (x − x′ ) ′ ′ and (7.33) . which represents the ’number of particles’ with momentum k.28) (7.

and multiparticle states |(k1 )n1 (k2 )n2 .46) (7.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. For the purpose of normal ordering it is convenient to decompose the field in positive and negative frequency parts.44) p|φ− (x)|0 = ei p·x . . (2π)3 2Ek (7.45) (7. in field theory also referred to as vacuum energy Evac = 1 V 2 d3 k Ek . Just as in the case of the harmonic oscillator it is essential (axiom) that there exists a ground state |0 that is annihilated by a(k). |k = a† (k)|0 . d3 k |k k|. 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. (2π)3 2Ek (7. writing all annihilation operators to the right of the creation operators. i. . The rest of the states are then obtained from the groundstate via the creation operator.42) (7. This term will be adressed below.43) The normal ordered product can be expressed as : φ(x)φ(y) := φ+ (x)φ+ (y) + φ− (x)φ+ (y) + φ− (y)φ+ (x) + φ− (x)φ− (y).40) This procedure is known as normal ordering.35) where the vacuum contribution disappears because of rotational symmetry. a(k)|0 = 0. defining particle states |k = |k (with positive energy.38) d4 k 2π δ(k 2 − M 2 ) θ(k 0 ) |k k| = (2π)4 (7. φ(x) φ+ (x) φ− (x) = = = φ+ (x) + φ− (x). (2π)3 (7. which amounts to redefining H as H= d3 x : H (x) := d3 k Ek N (k).e.37) (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). .34) where V = (2π)3 δ 3 (0) is the space-volume. (2π)3 2Ek d3 k a† (k) ei k·x . n1 ! n2 ! (7. (2π)3 2Ek (7.41) (7. d3 k a(k) e−i k·x .39) The problem with the vacuum or zero-point energy.36) and satisfying the completeness condition 1 = k|k ′ = (2π)3 2Ek δ 3 (k − k′ ). is solved by subtracting it as an (infinite) constant. = normalized as (a† (k1 ))n1 (a† (k2 ))n2 √ √ . . (2π)3 2Ek (7. assuring that the vacuum (by definition) has eigenvalue 0!. k 0 = Ek = k2 + M 2 ). |0 .

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. φ(y)] consider µν [φ+ (x).48) (7. that they satisfy the required commutation relations in Eqs (7. φ+ (y)] d3 k e−i k·(x−y) . d3 k ei k·(x−y) = −i∆+ (y − x).50) or as integrals over d4 k. The last item to be checked for the scalar field are the causality condition. φ− (y)] ≡ i∆+ (x − y). i. ∆(x) = − ǫ(x0 ) δ(x2 ). (7. (iii) ∆(0.18).49) (7. (2π)3 2Ek ≡ i∆− (x − y). (7.53) t=0 .17) and (7. = − (2π)3 2Ek (7. 2π (7.47) ′ d3 k ′ e−i k·x+i k ·y [a(k). 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 ). (iv) The equal time commutation relations follow from ∂ ∆(x) ∂t (v) For M = 0. (2π)4 (7. d3 k = (2π)3 2Ek = (7. (ii) ∆(x) is a solution of the homogeneous Klein-Gordon equation.52) The result for the invariant commutator function is [φ(x). In order to calculate [φ(x). a† (k ′ )] 3 2E ′ (2π) k [φ− (x). (2π)4 d4 k − θ(k 0 ) 2π δ(k 2 − M 2 ) ei k·x (2π)4 d4 k − θ(−k 0 ) 2π δ(k 2 − M 2 ) e−i k·x = −i∆+ (−x) (2π)4 (7.56) = −δ 3 (x).51) (7. x) = 0 and hence ∆(x) = 0 for x2 < 0.e.54) = i∆(x − y) = i (∆+ (x − y) + ∆− (x − y)) .55) d4 k ǫ(k 0 ) 2π δ(k 2 − M 2 ) e−i k·x . i∆+ (x) i∆− (x) = = = d4 k θ(k 0 ) 2π δ(k 2 − M 2 ) e−i k·x .

e.64) The commutator of φ and φ† is as for the real field given by [φ(x). we will consider it as a repetition of the quantization procedure.66) .4 The complex scalar field In spite of the similarity with the case of the real field. (2π)3 2Ek (7. extending it with the charge operator and the introduction of particle and antiparticle operators. (2π)3 2Ek (7. which is equivalent to the relations (only nonzero ones) [a(k). (2π)3 2Ek (7. (7.58) and satisfy the equal time commutation relation (only nonzero ones) [φ(x).63) i. 2 From the lagrangian density (7. φ† (y)] = i∆(x − y). Also the charge operator requires normal ordering (in order to give the vacuum eigenvalue zero).62) (7.59) (7. = ∂{µ φ∗ ∂ν} φ − L gµν .60) = i φ∗ ∂µ φ. 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 .65) 7.61) (7.5 The Dirac field L = ↔ i ψγ µ ∂µ ψ − M ψψ.57) (7. Q = = = i d3 x : φ† ∂0 φ − ∂0 φ† φ(x) : d3 k : a† (k)a(k) − b(k)b† (k) : (2π)3 2Ek d3 k a† (k)a(k) − b† (k)b(k) . particles (created by a† ) and antiparticles (created by b† ) with the same momentum contribute equally to the energy. (2π)3 2Ek d3 k b(k) e−i k·x + a† (k) ei k·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) . a† (k ′ )] = [b(k). b† (k ′ )] = (2π)3 2Ek δ 3 (k − k′ ). ∂0 φ† (y)]x0 =y0 = i δ 3 (x − y).Quantization of fields 60 7. ↔ (7.

s)ei k·x . s).69) (7. u v (2π)3 2Ek (7.72) ψ(x) = s In terms of the operators for the b and d quanta the hamiltonian and charge operators are (omitting mostly the spin summation in the rest of this section) H = = Q = = d3 x : ψ † (x) i∂0 ψ(x) : d3 k Ek : b† (k)b(k) − d(k)d† (k) : (2π)3 2Ek d3 x : ψ † ψ : d3 k : b† (k)b(k) + d(k)d† (k) : . s)u(k. s) e−i k·(x−y) j s † + vi (k. The quantized fields are written ψ(x) = s d3 k b(k. s) ei k·(x−y) x0 =y 0 . i. {b(k. s′ )} = (2π)3 2Ek δ 3 (k − k′ ) δss′ .75) (7. s)vj (k.79) Also for the field and the canonical conjugate momentum anticommutation relations are considered. ↔ i ψγµ ∂ν ψ − 2 (7. (2π)3 2Ek (7. ∂ 2 ↔ (7.78) (7. s)e−i k·x . s)e−i k·x + d† (k. ψj (y)}x0 =y0 = d3 k (2π)3 2Ek ui (k. interchanging creation and annihilation operators. (7.68) The canonical momentum and the hamiltonian are given by Π(x) H (x) = = δL = i ψ † (x).73) (7. d† (k ′ . † {ψi (x). s)ei k·x + d(k. b† (k ′ . s). s′ )} = {d(k.76) which seems to cause problems as the antiparticles (d-quanta) contribute negatively to the energy and the charges of particles (b-quanta) and antiparticles (d-quanta) are the same. The solution is the introduction of anticommutation relations. (2π)3 2Ek (7. s)v(k. s)u† (k. (7.77) Note that achieving normal ordering.74) (7. (2π)3 2Ek d3 k b† (k. . 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) .67) i ψ / ψ − M ψψ gµν . s)¯(k. ˙ δ ψ(x) ↔ i Θ00 (x) = − ψγ i ∂i ψ + M ψψ 2 = iψγ 0 ∂0 ψ = iψ † ∂0 ψ.Quantization of fields 61 the conserved density and energy-momentum currents are easily obtained. s)¯(k. jµ Θµν = = ψγµ ψ.e.70) where the last line is obtained by using the Dirac equation.71) (7.

one has † {ψi (x). we would have obtained [ψi (x).85) which reflects the gauge freedom. For the scalar combination {ψi (x). ψ j (y)} at arbitrary times one has {ψi (x). but has the problem of being noncovariant.88) . ∂ (7.87) If one wants to impose canonical commutation relations ∂µ Aµ = 0 cannot hold as an operator identity.6 The electromagnetic field 1 L = − Fµν F µν .86) (7. ψj (y)}x0 =y0 = d3 k (/ + M )il (γ0 )lj e−i k·(x−y) k (2π)3 2Ek + (/ − M )il (γ0 )lj ei k·(x−y) k x0 =y 0 = d k 2Ek (γ0 γ0 )ij ei k·(x−y ) (2π)3 2Ek (7.82) where i∆1 = i∆+ −i∆− . which however has wrong causality properties! Therefore the relation between spin and statistics is required to get micro-causality (which is also known as the spin statistics theorem). When we would have started with commutation relations for the field ψ and the canonical momentum. ψ j (y)] = (i/x + M )ij i∆1 (x − y). (7.81) where i∆ = i∆+ + i∆− is the same invariant commutator function as encountered before. 4 2 This gives the equations of motion discussed before. ˙ δ Ai (7. (7. ∂ (7. ψ j (y)} = d3 k (/ + M )ij e−i k·(x−y) + (/ − M )ij ei k·(x−y) k k (2π)3 2Ek d3 k e−i k·(x−y) + −ei k·(x−y) = (i/x + M )ij ∂ (2π)3 2Ek = (i/x + M )ij i∆(x − y). −λ(∂ρ Aρ ).Quantization of fields 62 Using the positive and negative energy projection operators discussed in section 4. as the vanishing of Π0 induces a constraint. it implies the Lorentz constraint and leads to the canonical momenta Π0 Π i = = E i. 7. 4 Π0 Πi δL = 0.83) From the lagrangian density the canonical momenta are = = (7. ˙ δ A0 δL = F i0 = E i .84) (7.80) 3 = δ 3 (x − y) δij . It is possible to continue in a covariant way with the lagrangian 1 λ L = − Fµν F µν − (∂µ Aµ )2 . but we must restrict ourselves to the weaker condition B|∂µ Aµ |A = 0.

i. 0)c(k.1 (optional) If you haven’t seen Poisson brackets in classical mechanics you should (a) Proof that Poisson brackets (Eq. (7. (7.90) where Πν = F ν0 − λ g ν0 (∂ρ Aρ ) and we have furthermore [Aµ (x).94) Choosing k µ = (|k 3 |. ′ (7. for which it is sufficient that ∂µ Aµ |A = 0. Aν (y)]x0 =y0 = i gµν δ 3 (x − y). one time-like photon by itself is not allowed! This solves the problems with the normalization and the negative energies. λ)ei k·x . Πν (y)] = 0.e. (7. pj ]P . (c) Calculate the Poisson brackets for the coordinates x and the standard conserved quantities p and ℓ. It gives 3 k µ ǫ(λ) (k)c(k. i.92) Note that for the transverse states there are no problems with the normalization and the statistics of the states as −g ij = δ ij for i. λ). j = 1.95) Exercises Exercise 7.91) (7. λ′ )] = −g λλ 2Ek (2π)3 δ 3 (k − k′ ). λ)|A = 0. containing a time-like photon.3) satisfy the conditions of a Lie algebra. (b) Proof for a quantity O(p. 0. and are equivalent with [c(k. µ λ=0 (7. Q]P . ˙ ˙ . rj ]P .Quantization of fields 63 for physical states |A and |B .11. 2. These problems are solved by the Lorentz constraint between physical states given above.e. The quantized field is expanded as Aµ (x) = d3 k (2π)3 2Ek (λ) 3 ǫ(λ) (k) c(k. 0) . 6. µ λ=0 (7. λ)e−i k·x + c† (k. q) that dO/dλ = [O. (d) Use Poisson brackets to calculate q and p. Aν (y)] = [Πµ (x). what are [ri . 7. λ) − c† (k. pj ]P . where Aµ + + is the part of the vector field containing the annihilation operators. c† (k ′ .93) which does exhibit problems with the time-like photon. Πν (y)]x0 =y0 = i gµν δ 3 (x − y). In fact the commutation relations imply ˙ [Aµ (x). and [ℓi . 0. λ)c(k.89) with four independent vectors ǫµ . where Q is the conserved quantity in Eq. which reproduce the Hamilton equations. 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. The canonical equal time commutation relations are [Aµ (x). a longitudinal photon and two transverse photons. [ℓi . k 3 ) this reads |k 3 |c(0) (k) − k 3 c(3) (k) |A = 0 c(0) (k) ∓ c(3) (k) |A = 0.

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

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

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

x µ (iγ ∂µ − M ) γ0 ψ(˜) = 0. op P (8. (2π)3 2Ek (8. x It is straightforward to apply this to the explicit field operator ψ(x) using γ0 u(k. λ) Pop = ηP b(k.5) (8. m). λ)v(k.3) (8. x (8. λ)u(k.e. (i˜ µ ∂µ − M ) ψ(˜) = 0. λ) P −1 = −η ∗ d(k. choosing ηP is real (ηP = ±1) particle and antiparticle have opposite parity. λ)v(k. ˜ −v(k.4) (8. −λ) eik·x (2π)3 2Ek ˜ d3 k ˜ ˜ ˜ ˜ ηP b(k. λ) eik·x .11) (8. the above operation reverses the sign of λ). m) = = ˜ u(k. −λ) eik·x 3 2E (2π) ˜ k d3 k ˜ ˜ ηP b(k. m). −λ)v(k. −λ). The result is ψ p (x) −1 = Pop ψ(x) Pop = ηP γ0 ψ(˜) x (8. if helicity λ is used instead of the z-component of the spin m.6) (check this for the standard representation. λ) eik·˜ (2π)3 2Ek d3 k ˜ ˜ ˜ ˜ ηP b(k. γ x (8. x Therefore γ0 ψ(˜) is again a solution of the Dirac equation and we have x ψ p (x) = γ0 ψ(˜). For the scalar field we have seen −1 φ(x) −→ Pop φ(x) Pop = ηP φ(˜).9) (8. −λ) e−ik·x − d† (k. iγ µ† ∂µ − M ψ(˜) = 0. x The latter behavior of the vector field will be discussed further below. one can also consider the scalar field and vector fields. ˜ Pop d(k. 8. −λ) e−ik·x − d† (k. .13) ψ(x) −→ ψ c (x) = ηC C ψ (x). λ) e−ik·x − d† (k.8) 3 = λ = λ = λ From this one sees immediately that −1 ˜ Pop b(k. In the same way as the Fermion field.2 Charge conjugation We have already seen the particle-antiparticle symmetry with under what we will call charge conjugation the behavior T (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)γ0 v(k. λ)u(k. m) γ0 v(k. λ)γ0 u(k. −λ)u(k. λ) e−ik·˜ + d† (k.10) i.12) and for the vector field −1 Aµ (x) −→ Pop Aµ (x) Pop = −Aµ (˜). −λ).7) = λ d k x x ηP b(k.

One has Aφ|Aψ = φ|ψ ∗ = ψ|A† Aφ = ψ|φ . antilinear operator is anti-unitary if A† = A−1 .18) (8. 8. Therefore ψ c (x) = = λ −1 ¯ Cop ψ(x) Cop = ηC C ψ T (x) (8.14) (8. γ x i(γ5 C)−1 γ µ γ5 C∂µ − M ψ(−˜) = 0.Discrete symmetries 68 the latter being also a solution of the Dirac equation. m) ¯ C v T (k. x µ (iγ ∂µ − M ) γ5 Cψ(−˜) = 0. Top |φ = φt | = (|φt )∗ . λ)u(k.21) where A is a 4 × 4 matrix acting in the spinor space. (i˜ µ∗ ∂µ − M ) ψ(−˜) = 0. x (8. r) −→ −˜µ ≡ −xµ = (−t. m) ¯ = = v(k. writing −1 ψ(x) −→ Top ψ(x)Top = ηT Aψ(−˜) ≡ ψ t (−˜). λ) −1 Cop = ∗ ηC b(k.3 Time reversal xµ = (t. (2π)3 2Ek (8. (8. λ) eik·x ¯ ¯ (2π)3 2Ek d3 k ηC d(k.20) The time reversal operator transforms We will again consider the transformation properties for a fermion field ψ(x). x (8. ((iγ µ )∗ ∂µ − M ) ψ(x) = 0. The (time-reversed) Dirac equation becomes. As time reversal will transform ’bra’ into ’ket’. u(k. λ) e−ik·x + b† (k. x 1A 2 An −iC −1 γ µ C∂µ − M ψ(−˜) = 0. Norm conservation requires Top to be anti-unitary2 .19) Cop d(k. i. it is antilinear1 .15) (where one must be aware of the choice of spinors made in the expansion. m). as discussed in section 4). λ). λ)C uT (k. λ) Cop = ηC d(k.16) d k ηC d(k.e. For a quantized field one has −1 ∗ −1 Top fk (x)bk Top = fk (x)Top bk Top . m). r). The action on the spinors (using C = iγ 2 γ 0 = −iρ1 σ 2 in standard representation) gives C uT (k. The same relations hold for helicity states. λ)v(k. . λ) eik·x . λ) e−ik·x + b† (k. x iγ µT ∂µ − M ψ(−˜) = 0. (8. to find ψ t (−˜) that is a solution of the Dirac equation.17) 3 = λ This shows that −1 Cop b(k. ˜ x −(iγ µ )∗ ∂µ − M ψ(−˜) = 0. x x (8. λ). λ)C v T (k.22) = Aψ|Aφ ∗ is antilinear if A(λ|φ + µ|ψ ) = λ∗ A|φ + µ∗ A|ψ . we start with the complex conjugated x Dirac equation for ψ.

λ)u∗ (k.27) 3 = λ = λ = λ From this one obtains −1 ˜ Top b(k. ˜ = 0|b(k)|A 8. Note that applying an anti-unitary transformation such as Top one must take for the matrix element ˜ the complex conjugate. −λ a T a. Since there are 16 independent 4 × 4 matrices. λ) e−ik·x 3 2E (2π) k d3 k ˜ ˜ ηT b(k.26) d k x x ηT b(k. −p. λ |¯. λ)iγ5 Cv(k. λ a P |a.24) (8. p. λ)iγ5 Cu(k.30) (8. λ) e−ik·x 3 2E (2π) k ˜ d3 k ˜ ˜ ˜ ˜ ηT b(k. p.31) (8. a state |a. The result is = = λ (8. p. λ) eik·x + d† (k. ˜ Top d(k.Discrete symmetries 69 Table 8. −p. λ).28) (8. λ)v ∗ (k. λ)v ∗ (k. −p. λ) eik·x + d† (k.2: The transformation properties of physical states for particles (a) and antiparticles (¯). λ| ¯ C |¯. x (8. Therefore one has A|k = A|b† (k)|0 = A|T † T b† (k) T † T |0 ∗ = At |b† (k)|0 ∗ ˜ t = k|At . λ) eik·x + d† (k. λ a |a. λ) ψ t (x) −1 Top ψ(x) Top = i ηT γ5 Cψ(−˜) x = = ˜ u∗ (k. λ Therefore γ5 Cψ(−˜) is again a solution of the (ordinary) Dirac equation and we can choose (phase is x convention) ψ t (x) = i γ5 C ψ(−˜). λ). ∗ ˜ v (k. p. λ)u∗ (k.34) . 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. λ) Top = ηT b(k. λ)v ∗ (k. λ). λ) e−ik·x (2π)3 2Ek (8. λ| a. op T (8.23) In the standard representation iγ5 C = σ2 and it is straightforward to apply this to the explicit field operator ψ(x) using i γ5 C u(k. (8.33) (8. λ)u∗ (k.4 Bi-linear combinations In quantities such as currents and lagrangians often bilinear combinations of spinor fields are encountered. λ) i γ5 C v(k. −p. −λ |¯.32) (8.29) In table 2 the behavior of particle states under the various transformations has been summarized. λ). there are 16 independent of these bilinear combinations. λ) T −1 = η ∗ d(k.25) (check this for the standard representation). λ) eik·˜ + d† (k. λ) e−ik·˜ (2π)3 2Ek d3 k ˜ ˜ ˜ ˜ ηT b(k.

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 ). for the vector current V µ (x). s . of which the expectation value between momentum eigenstates can be simply found. C.36) (8.5 Form factors Currents play an important role in field theory. as well as under the combined operation Θ = PCT (see Table 3)..37) (8. Applying the results from the previous sections it is straightforward to determine the behavior of the combinations under P. p) can be built from any combination of Dirac matrices (1.38) (8. p′ |V µ (x)|p = u(p′ )Γµ (p′ . momenta (pµ or p′µ ) or constant tensors (ǫµνρσ or gµν ). F. As an example consider the vector current for a point fermion. in this case only q 2 = (p′ − p)2 .g. σ µν . C. The 16 combinations of Dirac matrices appearing above are linearly independent. γ ν ].35) The x-dependence can be accounted for straightforwardly using translation invariance. In many applications the expectation values of currents are needed. and T. This implies p′ |V µ (x)|p = e−i (p−p )·x p′ |V µ (0)|p . ¯ ′ ′ (8.41) 2M 2M .39) where Γµ (p′ . γ µ . 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 . s′ |V µ (x)|p. As the coupling of the photon field to fermions is given by an interaction term in the ¯ lagrangian of the form : ψ(x)γ µ ψ(x) : Aµ (x) and behaves as a scalar one sees immediately that the ¯ photon field Aµ (x) behaves in the same way as the vector combination ψ(x)γ µ ψ(x). p′ |V µ (0)|p = u(p′ )γ µ u(p). p)u(p) e−i (p−p )·x . F. (q 2 ). p) = γµ F1 (q 2 ) + iσµν q ν F2 (q 2 ). 8. γ5 γ µ or γ5 ).3: The behavior of the independent bilinear combinations of fermi fields under P. 2M (8. Note that the lagrangian density L (x) → L (−x) under Θ.40) where the coefficients are functions of invariants constructed out of the momenta. called form factors. (8. p′ . (q 2 ). ¯ In general the expectation value between momentum states can be more complicated.Discrete symmetries 70 Table 8. ¯ V µ (x) =: ψ(x)γ µ ψ(x) :. For instance for nucleons one has Γµ (p′ . e. (8.. V µ (x) = ei Pop ·x V µ (0)e−i Pop ·x .

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

0.2 (a) Calculate the current matrix element (using the explicit fermion spinors) in the Breit-frame in which q = (0. 0. 0. |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 . (c) Show that if a term of this form would exist. GE GM = F1 − Q2 F2 .Discrete symmetries 72 (a) Use the Dirac equation to prove the Gordon-decomposition u(p′)γ µ u(p) = ¯ 1 u(p′ )[(p′ + p)µ + iσ µν (p′ − p)ν ]u(p) ¯ 2M (b) Give the relation between Hi en Fi . 0. 4M 2 = F1 + F2 . (b) Show that such a term is also not allowed by time-reversal symmetry. p = (E. σµν q ν F3 (q 2 ) 2M . L int = ejµ Aµ the charge is given by e GE (0) and the magnetic moment by e GM (0)/2M . 0. 0. Exercise 8. (b) Show that in the interaction with the electromagnetic field. −|q|/2). |q|). Exercise 8. it would correspond to an electric dipole moment d = −e F3 (0)/2M (Interaction term d σ · E). p′ = (E.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.

in which the states are time-independent.7) 73 . t ′ |q . and the operators are time-dependent. Consider the two (time-independent) Heisenberg states: |q. AH (t) = U −1 (t.4) ∂ U (t.2) (ii) Heisenberg picture. AS (t) = AS and the states o are time dependent. i ∂ |ψH ∂t ∂ i AH ∂t ≡ = 0. t0 ) = e−i(t−t0 )H or i Two situations can be distinguished: (i) Schr¨dinger picture. in which the operators are time-independent. (9. (9.1 Time evolution as path integral U (t. t0 ) ∂t (9. t0 ) AH (t0 ) U (t. (9. |ψH (t) = |ψH . i ∂ |ψS ∂t ∂ i AS ∂t = H |ψS .6) Of these the Heisenberg picture is most appropriate for quantum field theory since the field operators do depend on the position and one would like to have position and time on the same footing. t ′ QH (t′ )|q ′ . t0 ). QH (t)|q. t0 ) = H U (t.5) (9. t′ ≡ q ′ |q ′ . |ψS (t) = U (t. t0 )|ψS (t0 ) . ≡ 0. [AH . t . (9. H].1) The time evolution from t0 → t of a quantum mechanical system is generated by the Hamiltonian. t′ . t ≡ q|q.3) (9.Chapter 9 Path integrals and quantum mechanics 9.

t and QH (t) = QS and study the evolution of the system by calculating the quantum mechanical overlap amplitude q ′ .Path integrals and quantum mechanics 74 and Schr¨dinger states o |q |q ′ QS |q ≡ q|q . with the correction1 being of order (∆τ )2 . [A.qj ) e 2π dpj exp (i ∆τ [pj qj − H(pj . (9. 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 .q . Combining the two terms gives qj+1 . t′ |q. Then qj+1 . t′ |q. n |q (9. which is a hamiltonian function in which the operators are replaced by real-numbered variables. t = q ′ |e−i H(t −t) |q .11) (9. use the Campbell-Baker-Hausdorff formula. tj = = = 1 For dpj i pj (qj+1 −qj ) −i H(pj . qj ).16) this. 2π (9. tj+1 |qj . ˙ 2π 1 1 1 [A. B] + [A. 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|. . . B] + .14) . The hamiltonian is an operator H = H(P. QS |q ′ ≡ q ′ |q ′ .. . . tj+1 |qj .15) By letting the kinetic and potential parts act to left and right respectively one can express the expectation values in terms of integrals containing H(pj . 2π (9..13) At least for a simple hamiltonian such as consisting of a kinetic energy term and a local potential.qj )∆τ e e 2π dpj i pj (qj+1 −qj )−i ∆τ H(pj . qj )]) . The purpose of this is to calculate the evolution for an infinitesimal time interval.10) dqn−1 q ′ |e−i H∆τ |qn−1 qn−1 | . B].8) Choose t as the starting point with |q = |q. 2 12 12 (9. B]] + [[A. Q) expressed in terms of the operators P and Q. ′ (9. (9. |q1 q1 |e−i H∆τ |q .12) where the transformation between coordinate and momentum space involves q|p = ei p. H(P.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 n H∆τ |q = q ′ | e−i H∆τ dq1 . eA eB = eC with C = A+B+ (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′ |q1 . . . y) in the above examples can be (hermitean) operators acting on the functions. α∗ ] = exp (−α∗ K α) ≡ exp − The kernels K(x. while Dα F [α] = x dαx N F (αx ) (9. y) β(y). . q)] ˙ (9. . ˙ where Dq and Dp indicate functional integrals. .20) becomes a multidimensional integral. Before proceeding we also give the straightforward extension to more than one degree of freedom. . . including differential operators. 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. 9. ′ ′ q1 . α(x) → αx and F [α] = F (αx ) changes into a multivariable function. (9.2 Functional integrals In this section I want to give a fairly heuristic discussion of functional integrals. dx dy α∗ (x) K(x. dx dy α(x) K(x.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.18) t ′ = n=1 Dqn D pn 2π exp i t dτ n pn qn − H(p1 . .e. q)] .Path integrals and quantum mechanics 75 or for the full interval q ′ . . pN . qN . i. F [α] ∈ R. qN ) ˙ . What one is after is the meaning of Dα F [α]. . t′ |q. t = τ  dq(τ ) dp(τ ) exp i 2π p exp i 2π t′ t t′ t ∆τ [pq − H(p. ≡ exp − 1 2 dx α2 (x) . i. q1 . etc. qN . Examples are F1 [α] = α ≡ F2 [α] = α2 ≡ dx α(x). dx α2 (x). .17)  = Dq D dτ [pq − H(p. . t N (9. . Two working approaches for functional integration are the following: (i) the heuristic division of the space on which α acts into cells. y) α(y) . β] = α K β ≡ 1 F4 [α] = exp − α2 2 F5 [α. F3 [α.e.

Consider the gaussian functional as an example.n dx fm (x)fn (x) = n = n dαn 1 exp − α2 N 2 n √ 2π ≡ 1.24) (9. y) = ∗ m. . the following integrals can be derived for a real symmetric or complex hermitean kernel K.25) (see Exercises). Physical answers will usually come out as the ratio of two functional integrals and are thus independent of the chosen measure. (9.n fm (x) Kmn fn (y) and consider F [α] = F (αn ) as a multivariable function. with Dα F [α] = n dαn N F (αn ) (9. Note that procedure (i) is an example of the more general procedure under (ii). 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) (9.27) (9.21) again a multidimensional integral. 2 (9.23) Having defined the Gaussian integral. 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. A useful property of functional integration is the translation invariance. α(x) = n αn fn and K(x.22) = x The last equality is obtained by defining the right measure (normalization N ) in the integration. Dα F [α] = Dα F [α + β].26) As an important application of translation invariance we mention the identities 1 Dα exp − α2 − ωα 2 = exp 1 2 ω . In both cases N is an appropriately choosen normalization constant in such a way that the integral is finite.28) Dα Dα∗ exp (−α∗ α − α∗ ω − ω ∗ α) = exp (ω ∗ ω) . 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. (9.Path integrals and quantum mechanics 76 (ii) Write α in terms of a sum of orthonormal basis functions.

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.g. (9. products of the same pair appear.29) (9. F [θ. n θn fn (x) dx θ∗ (x) θ(x) . In principle the definitions of functionals is the same.30) (9. however. i.32) (9.33) (9. α(y) = δ(x − y).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. It is now easy to check that Dθ∗ Dθ exp (−θ∗ K θ) = det K. δα(x) or in discretized form δ 1 ∂ ∂ with . Examples of functional differentiation are (9. (9. δα(x) 2 2 δ exp (−α β) = −β(x) exp (−α β) .e.31) δ δ α = dy α(y) = dy δ(x − y) = 1. the Gaussian-type functionals.34) (9. y) β(y) δα(x) δ 1 1 exp − α2 = −α(x) exp − α2 . δα(x) (9.Path integrals and quantum mechanics 77 Functional differentiation is defined as δ . −→ fn (x) δα(x) ∂αn ∂αm n Note that δ(x − y) = n ∗ fn (x) fn (y). θη = −ηθ. δα(x) δα(x) δ αKβ = dy K(x. θ2 = 0. which vanish.38) . θ∗ ] = exp (−θ∗ K θ) ≡ exp − Integration for Grassmann variables is defined as dθ 1 = 0 and dθ θ = 1. −→ δα(x) ∆x ∂αx ∂αx δ ∂ ∂ ∗ with . αy = δxy . αn = δmn . y) θ(y) . (9.35) For applications to fermion fields. e. θ∗ ] = exp (−θ∗ θ) ≡ exp − F [θ. we need to consider Grassmann-valued functions involving anticommuting Grassmann variables. dx dy θ∗ (x) K(x. In the expansion of the exponential (from second to third line).

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 with q. t Q′ . for example x|n x|n = φn (x) = ei kn x = e −ω 2 x2 /2 (9. n (9. furthermore. t|n = q|n e−iEn t with q|n the time-independent wave function. q)] . Before and after these processes there is only the vacuum or ground-state |0 . t|Q. the creation of an electron (think of a radio-tube. Consider. t|Q. i. t |q. ˙ The combination of exponentials in the last step is allowed because we are simply dealing with classical quantities (not operators!).64) (9.65) J dq ′ dq Q′ . for g. a set |n of physical eigenstates. T = n q. one has q ′ . t t′ V =0 Dq D dτ HI (q) exp i t t′ t dτ [pq − H0 (p. t|n n|Q. t q.66) . T ′ |q ′ . ˙ ˙ physically pictured as. q) + J(τ )q] . it is possible to switch on an interaction. ˙ J (9. t′ |q.63) for plane waves.62) p Dq D exp i 2π t dτ [pq − H(p. i. t′ |q. 9.1: Physical picture of vacuum to vacuum amplitude in presence of a source term This is not a surprising result. The Heisenberg state |q. t = = ′ ′ V = q . Considering the source to be present between times t and t′ . 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. T = n J t′ = = Dq exp i t dτ [L(q. t′ q ′ . T < t < t′ < T ′ . q) + J(t) · q. say. t |T exp −i p exp −i 2π t′ ′ ′ t′ t dτ HI (Q(τ )) |q.5 The generating functional for time ordered products By introducing a source-term. φn (q.e. for the physical states o q. T . t = ei Ht |q . q .Path integrals and quantum mechanics source experiment detector 81 T L t L+Jq t’ L T’ Figure 9. which in turn are embedded between an early time T and a future time T ′ . T ′ |Q. L(q. q) → L(q.e. making the electron) and the absorption of an electron (think of a detector).s. harmonic oscillator. t) φ∗ (Q) e+i En T . q)] ˙ (9.

67) = = φ0 (q. t 0 0out |0in J . t|Q. in which case eiEn T → e−En ·∞ and the term eiE0 T is dominant. . . T ′ |Q. 9.71) 9. T → i ∞ and T ′ → −i ∞. The importance of Z[J] is that the time ordered product of operators can be obtained from it. t). . J=0 (9. t′ ) q ′ .Path integrals and quantum mechanics t 82 T t t’ T’ Figure 9. thus T →i∞ lim q. = (i)n 0|T Q(t1 ) . such that t′ → −i∞ t → i∞ ↔ ↔ ˜ t → −∞. . t′ |q.68) (9. eiE0 T φ∗ (Q) 0 We define the generating functional Z[J] as Z[J] = Q′ . ˜ t′ → ∞. dq dτ + J(τ )q(τ ) . . T J lim −i E0 (T ′ −T ) φ (Q′ ) φ∗ (Q) 0 0 T ′ → −i∞ e T → i∞ dq ′ dq φ∗ (q ′ .2: Analytic continuation of boundaries T and T ′ We can project out the groundstate by an analytic continuation in the time. Since we have (neglecting multiplicative factors). J (9.70) for the generating functional is in fact ill-defined.6 Euclidean formulation The above expression (Eq.70) one immediately finds δ n Z[J] δJ(t1 ) . The generating functional precisely represents the vacuum to vacuum amplitude from initial (’in’) to final(’out’) situation in the presence of a source. δJ(tn ) hence the name generating functional. T = φ0 (q. (9. Q(tn )|0 . T′ Z[J] = lim T ′ → −i∞ T → i∞ Dq exp i T dτ L q. t) (9.69) The factor that has been divided out in the first line of this equation is a numerical factor depending on the boundaries of the space-time volume (T and T ′ ). Better is the use ˜ of imaginary time t = −it.

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

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

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

which can be written as ∆(x) = ˜ with (k 2 − M 2 )∆(k) = 1 or ˜ ∆(k) = 1 1 1 = 0 2 .11) Depending on the path choosen in the complex k 0 plane (see exercise 7. . 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. . consider the general solution of ∂µ ∂ µ + M 2 ∆(x) = −δ 4 (x).9) i. with Z0 [0] = 1. . As discussed in the previous section (exercise 9.6) the generating functional is (using ∂µ = − ∂µ ) given by   Z0 [J] =  Dφ exp −i where Kxy = (∂µ ∂ µ + M 2 ) δ 4 (x − y). These Green functions appear in the expansion Z[J] = n in n! d4 x1 . .12) d4 k −i k·x ˜ e ∆(k) (2π)4 (10. (10.3) one finds 1 Z0 [J] = exp − i 2 d4 x d4 y J(x) ∆F (x − y) J(y) ≡ exp (i W0 [J]) .Feynman diagrams for scattering amplitudes 86 for which we introduce the picture x1 G(n) (x1 . (2π)4 (k 0 + iǫ)2 − E 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 . . . (10.8)  1 d4 x  φ (∂µ ∂ µ + M 2 ) φ − Jφ 2 K  (10. 2 2 → (10.13) . . Furthermore. ∆F (the socalled Feynman propagator) is the Green’s function of the Klein-Gordon equation. = 0 2 2 2 k2 − M 2 (k ) − E 2 (k ) − k − M (10. .e.10) J=0 In order to determine ∆F . (10.7) where we made the choice Z0 [0] = 1 and ∂µ ∂ µ + M 2 ∆F (x − y) = −δ 4 (x − y). . . xn ) = x2 xn . .4) one distinguishes the retarded Green’s function ∆R (x) = d4 k e−i k·x . J(xn ). xn ) J(x1 ) . d4 xn G(n) (x1 .

φ+ (0)] = (2π)3 2E are also shown in exercise (7.19) φ ∂E ∂E − M 2 φ + Jφ 2 1 d4 xE d4 yE J(x) (−i ∆F (x − y)) J(y) .17) d4 k e−i k·x . 4 k 2 − M 2 + iǫ (2π) (10. φ(0)] = i∆+ (x) + i∆− (x) d3 k e−i k·x = [φ+ (x). in the (well-defined) Euclidean formulation starting with LE = 2 (∂E φ)2 + 2 M 2 φ2 µ 2 2 Jφ with the equation of motion (∂E ) − M φ(x) = −J(x). The contours for i ∆(x) i ∆+ (x) i ∆− (x) = [φ(x). ∆(x) = − d4 k e−i k·x . the advanced Green’s function ∆A (x) = satisfying ∆A (x.23) k0 =i∞ k0 =−i∞ d4 k e−i k·x . t) = 0 for t > 0.18) Thus we see various solutions.4). (10.21) ∆F (x) = = −i 0 kE =−∞ d4 kE e−i kE ·xE 2 (2π)4 kE + M 2 (10. (2π)4 k 2 − M 2 (10. (2π)4 k 2 − M 2 (10. the generating functional can be written 1 µ µ Z0 [J] = Dφ exp d4 xE (10.1 for countours) µ (∂E )2 − M 2 (i ∆F (x − y)) = −δ 4 (x − y) 0 kE =∞ (10.22) (10. (2π)4 k 2 − M 2 where the latter contour can be deformed to the contour for ∆C . . 10.Feynman diagrams for scattering amplitudes 87 satisfying ∆R (x.20) = exp − 2 where or (see fig. φ− (0)] = (2π)3 2E d3 k e−i k·x = [φ− (x). t) = 0 for t < 0. ∆F = ∆C .16) C where C is a closed contour in the complex k 0 -plane. i. 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 . It is straightforward (closing contours in the appropriate half of the complex plane) to prove for instance that i ∆C (x) = θ(x0 ) i ∆+ (x) − θ(−x0 ) i ∆− (x). (2π)4 (k 0 − iǫ)2 − E 2 (10.15) e−i k·x d4 k . The first possibility is to consider the well-defined Euclidean formulation. (10. The solution of the homogeneous equation ∂µ ∂ µ + M 2 ∆(x) = 0 can also be written as an integral in k-space. In order to see which is the appropriate Green’s function to be used in the generating functional Z0 [J] we will take two routes. µ 1 1 • Firstly.e.14) They are solutions of the inhomogeneous equation. the second is to explicitly consider 0|T φ(x1 ) φ(x2 )|0 .

(4) (2) It should be clear that G0 can be expressed in terms of G0 = i ∆F . δJ(x4 ) 2 J=0 = − [∆F (x1 − x2 ) ∆F (x3 − x4 ) + ∆F (x1 − x3 ) ∆F (x2 − x4 ) + ∆F (x1 − x4 ) ∆F (x2 − x3 )] . . because this is the only quantity entering Z0 [J]. 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. Neglecting multiplicative factors or equivalently assuming that Z0 [0] = 1. x3 .24) Knowing the explicit form of Z0 [J] it is straightforward to calculate the 4-points Green’s function. the same result as above. or diagrammatically (4) (10. We find (see also section 7.Feynman diagrams for scattering amplitudes 0 1 1 0 0 0 1 1 1 0 1 0 0 0 1 0 1 0 11 00 k 1 0 1 0 k 11 00 E 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 -E 1 0 1 0 1 0 1111111111111111111111111111111111111 0000000000000000000000000000000000000 1111111 1 0000000 0 11111111111111111 00000000000000000 11111111111111 00000000000000 11 00 11 00 1 0 1 0 1 0 11111111 1 00000000 0 1 0 1 0 E 1 0 1 0 1 0 1 0 11 00 1 0 1 0 1 0 1 0 1 0 1 0 0 1 1 0 1 0 1 0 1 0 1 0 88 Figure 10.25) x1 x2 x3 = x4 + + . we have G0 (x1 . = θ(x0 − y 0 ) 0|[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 ) . . x2 .3) i ∆F (x − y) = 0|T φ(x) φ(y)|0 = θ(x0 − y 0 ) 0|φ(x) φ(y)|0 + θ(y 0 − x0 ) 0|φ(y) φ(x)|0 = θ(x0 − y 0 ) 0|φ+ (x) φ− (y)|0 + θ(y 0 − x0 ) 0|φ+ (y) φ− (x)|0 = i ∆C (x − y).1: Contour in the k 0 plane for the Feynman propagator in Euclidean and Minkowski space • As the second possibility. φ− (y)]|0 − θ(y 0 − x0 ) 0|[φ− (x).

29) This expression will be the one from which Feynman rules will be derived.35) z . x3 . (2) (0) (10.33) (4) i −i d4 z LI g 4! d4 z 1 δ i δJ(z) 1 exp − i 2 J ∆F J 2 −3 ∆2 (0) + 6i ∆F (0) F + d4 x ∆F (z − x) J(x) 4 d4 x ∆F (z − x) J(x) 1 exp − i 2 J ∆F J .g ) (x1 . L (φ) = L0 (φ) + LI (φ). Z[J] = = Dφ exp i exp i d4 z LI (10.32) (10. x4 ) = G0 (x1 . To zeroth order in the coupling one has 0 1 Z (g ) [J] = Z0 [J] = exp − i 2 J ∆F J . ≡ −i g d4 z (10. x3 .30) and in zeroth order G(0. Consider as an example the interaction LI (φ) = − g 4 φ 4! in the scalar field theory discussed sofar.e.g 0 ) = G0 = 1. x2 .2 Generating functionals for interacting scalar fields d4 x [L0 (φ) + LI (φ) + Jφ] 1 δ i δJ(z) d4 x [L0 (φ) + Jφ] (10.28) 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 . the generating functional can be written.Feynman diagrams for scattering amplitudes 89 10.34) Introducing a vertex point to which four propagators are connected. x4 ) = To first order in g one has Z (g ) [J] = = 1 0 0 x1 + + x2 (10.31) G(2.26) 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. x2 ) = i ∆F (x1 − x2 ) = G(4. (10. x2 ) = G0 (x1 . (10. (10. i.27) When interactions are present. with propagators (i ∆F ) being connected to vertices (i LI ) according to the above expression for the functional Z[J]. x2 .g ) (x1 .

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

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

e. p′ .44) J=0 For the interacting theory. p. pn . the vacuum expectation value of the field φ(x) is zero. in|S = α. out = |p′ . Next.47) .e. thus β. one finds for the connected 4-point Green’s function at order g. Proof: α. (2) (10. in|p′ . in = p.3 Interactions and the S-matrix The S-matrix The S-matrix transforms initial state free particle states (in-states) |α. out .43) implying as expected as the only nonzero connected Green’s function G(2) (x1 − x2 ) = (−i) c δ 2 W [J] δJ(x1 ) δJ(x2 ) = G0 (x1 − x2 ) = i ∆F (x1 − x2 ). Proof: 0. (3) S is unitary (it conserves the scalar product from initial to final state). out| and S † |α. . in ⇔ β. (10. out = p|p′ . translation invariance). out| ⇔ |α. in|S = β. out|α. . out| = eiϕ0 0. out = δαβ ↔ SS † = 1. in = |p1 . out|p′ .46) The properties of the S-matrix are (1) The vacuum is invariant or |S00 | = 1. out . out (suppressing except momenta all other 1 m quantum numbers). out|p′ .g. Also for free field (Heisenberg) operators a distinction is made between φin and φout . in = p. φout and the interpolating field φ(x). (2) The one-particle state is invariant (conservation of energy and momentum. . out|α. . in|S|p′ . x4 ) = c 1 x3 z (10. In line with the consideration of the generating functional representing the vacuum to vacuum amplitude we consider fields φin .45) x2 as the only surviving diagram (see exercises).Feynman diagrams for scattering amplitudes 92 i. we have W0 [J] = − 1 2 d4 x d4 y J(x) ∆F (x − y) J(y). in| (choose ϕ0 = 0). in = S|α. x2 . . in into final state particle states (out-states) |β. in = β. then (n) (n) Gsc = Gc for n ≤ 3. this will be translated to the action on fields.g ) (x1 . . implying the absence of tadpoles. x4 10. For the free scalar theory. Sβα ≡ β. in|S = 0. (10. considered explicitly. 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 . in|SS † |α. x1 G(4. x3 . in = |α. ∂µ ∂ µ + M 2 φin (x) = 0. (10. in = p.

The time evolution operator (see Eq. (10. √ Although the above. 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.62) is then T′ UT [J] T′ = T exp i T d4 x J(x)φ(x) .52) where Z is a constant.51) (10. e. φin (x) = d3 k ∗ a(k) fk (x) + a† (k) fk (x) . S is invariant under e Poincar´ transformations: U (Λ. out|φout = β.49) The field φ can be expressed in φin/out using retarded or advanced Green’s functions. β. implies the strong (operator) convergence φ(x) → Zφin (x). in|φin S → φin S = S φout (x).54) . z) φ(z) d4 y d4 z ∆A (x − y) K(y. a) = S.50) = φ(x) = √ Zφin (x) − √ Zφout (x) − d4 y d4 z ∆R (x − y) K(y. (2π)3 2E (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) (10.48) 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. φin (y)] = iZ ∆(x − y). The relation between S and Z[J] To establish this relation. The relation between S-matrix and in. Proof: β. i. the source term Jφ in the lagrangian density is treated in the interaction picture.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 ).and out-fields is: φin (x) = S φout (x)S −1 . which in a later stage (renormalization) will become more important.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). this can actually not be used as it would imply [φ(x). 2 ∂µ ∂ µ + M0 φ(x) = J(x). φ(y)] = Z [φin (x). a causality condition that can be proven to imply the absence of interactions. a)SU −1 (Λ.g. 9. The convergence therefore must be weakened to t→−∞ √ (10. (10. (10. as it stands.e. z) φ(z). in|Sφout Finally we check that as expected S does not spoil Poincar´ invariance. 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.

: e d4 z φ(z) f (z) d4 x φ− (x) f (x) B C B C (10. φ− f (10. y) δ δJ(y) : F [J]. : =i d4 y ∆(x − y) f (y) : e d4 z φ(z) f (z) :. one has [A. S U [J]] = √ Z δ S U [J]. δJ(x) ∞ or for U [J] ≡ U−∞ [J] ′ (10. B] (for the case that [A. we can. B]eB .63) . Thus applied to the field φ(x) = φ+ (x) + φ− (x). with the use of the Baker-Campbell-Hausdorff formula e−B A e+B = A + [A. φ(y)] f (y) e e . δJ(z) (10. [A. t i Ut∞ [J] φ(x)U−∞ [J] (10. (10. φ(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].e.60) e d4 y φ+ (y) f (y) d4 z φ− (z) f (z) d4 z φ+ (z) f (z) d4 y [φ(x).62) where F [J] is some (arbitrary) functional. z) d4 y d4 z ∆(x − y)K(y. (10. φ(x). eB ] = [A.56) Since δU [J]/δJ(x) satisfies the same equations as φ(x) and we know the limits. B + C]e e . C] are c-numbers. z) δU [J] . provided [A. Thus S U [J] =: exp 1 √ Z d4 x d4 y φin (x) K(x. B] and [A. z) δU [J] δJ(z) δU [J] d4 y d4 z ∆A (x − y) K(y.61) e φ+ f where the normal ordered expression : e φ f : is used. e e ] = [A.Feynman diagrams for scattering amplitudes 94 and satisfies the property T ′ δUT [J] t = i UtT [J] φ(x)UT [J].59) (10. δJ(z) (10. just as we did for φ(x).57) i [φin (x). e = i. express it in terms of advanced and retarded Green’s functions. 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. which is equal to the expression e in which the creation operators are placed left of annihilation operators. 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. B] is a c-number).58) In order to find a solution to this equation note that. 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).

←− − → × G(n) (x′ . J=0 (10. (λ) v(p).65) we get the Lehman-Symanzik-Zimmerman (LSZ) reduction formula. . . which is obtained considering only connected diagrams. i. .e. d4 x . . |S|p. y). S =: exp 1 √ Z d4 x d4 y φin (x) K(x.δ. . . which is determined by the inverse of the operator found in the quadratic term in the lagrangian i.2). .. . .. . .64) Therefore. iKx′ = (2x′ + M 2 ) precisely annihilating an external propagator i∆F in the Green’s function.66) where Gc is the connected Green’s function.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. Sf i = δf i − i (2π)4 δ 4 (Pi − Pf ) Mf i . . pn ) =  G (p1 . i. Note that the Green’s function connecting the external point xi with the bubble is annihilated by K(x. pn ) = i=1 d4 xi ei pi ·xi G(n) (x1 . . ∆(pj ) j=1 (10. using that φin (x) = d3 k ∗ a(k) fk (x) + a† (k) fk (x) .e. ¯ ¯ and the amputated or one-particle irreducible (1PI) Green’s functions   n −i  (n) Γ(n) (p1 . Next we introduce the Fourier transform (after extracting a momentum conserving delta function coming from translation invariance.. . Usually we are interested in the part of the S-matrix describing the scattering.Feynman diagrams for scattering amplitudes 95 while for J = 0 one has U [0] = 1. . . . 1 for scalar case. . c (n) (10. .67) where ∆(p) is the Fourier transform of the (full) propagator ∆F (x).) iKx . . p′ . . It is straightforward to check that the S-matrix element now precisely is given by the amputated Green’s function multiplied with the momentum space wave functions of the particles in initial and final state (by which we refer to the quantities multiplying the plane wave e±i p·x in the field expansion. fp (x) . (10.68) 10. .e. . . . see exercise 10. . n (2π)4 δ 4 (p1 + . (2π)3 2E n (10.. . . . y) δ δJ(y) : Z[J] Z[0] . k = k2 i − M 2 + iǫ . x. Explicitly. u(p). . 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). . xn ). = . . . . fp′ (x′ ) . . . iKx′ . . u(p) and v (p) for fermions and ǫµ (p) for vector fields). . . + pn ) G(n) (p1 .+ 1 √ Z ∗ d4 x′ . . . . pn ). They start with the propagator (i∆F (k)) in momentum space. . .

’t Hooft and M. Veltman. At these vertices each line can be assigned a momentum. = −i g (multiplied with 4! corresponding to the allowed number of permutations of identical particles). Note that in calculating amputated Green’s functions external lines are neglected.69) LI (φ) = − φ3 − φ4 . 3! 4! The vertices are: = −i f Consider first the lowest order self-energy diagram. For the symmetry factor consider the examples (given in G. but overall momentum conservation at a vertex is understood. 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. vertices and external particle wave functions in that diagram. the factor 2 corresponds to the two-points Green’s function having two identical ends).27. to be precise iLI vertices in momentum space are introduced. (Rule 3) Carry out the integration over all internal momenta (keeping track of momentum conservation at all vertices!) (Rule 4) Add a symmetry factor 1/S corresponding to permutation of internal lines and vertices (keeping external lines fixed). for example up to order g 2 the scattering of two neutral spin 0 particles (real scalar field) is described by + + + (Rule 2) The contribution of each diagram is obtained by multiplying the contributions from propagators. = −i g . Diagrammar) in the case of the interaction terms g f (10. or calculating full Green’s functions external lines are treated as propagators. If problems arise go back to the defining expression for the generating function in 10.66 and 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. For the interaction terms in the lagrangian.27. Corresponding to external particles wave functions are introduced k k 1 1 In order to calculate the connected amplitude −i Mf i appearing in the S-matrix element these ingredients are combined using Eqs 10. which cancels the factor 1/2 in the quadratic piece.

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

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

The wave functions for fermions are given by ∝ g 2 S(z − z ′ ) S(z ′ − z) = −g 2 iS(z − z ′ ) iS(z ′ − z).Feynman diagrams for scattering amplitudes 99 The result is which contains an extra minus sign as compared to a bosonic loop.74) L = − Fµν F µν − (∂ · A)2 + ψ(i/ − M )ψ − e ψγ µ ψ Aµ . 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 . λ 1 ∂ (10. 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 . + 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. Particular choices of λ are λ = 1 (Feynman propagator or Feynman gauge) and λ = ∞ (Landau gauge). It is necessary to start at the end of a fermion line (with the above arrow convention an outgoing fermion wave function u or an incoming antifermion ¯ wave function v ) and keep on following that line. 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. 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 . Vector fields and Quantum Electrodynamics As the most important example of vector fields consider the lagrangian density for quantum electrodynamics (QED). As an example some scattering processes in quantum electrodynamics will be discussed in the next section after the introduction of the Feynman rules for vector fields. 1 k 2 − M 2 + iǫ λ (k 2 − M 2 + iǫ)(k 2 − λ M 2 + iǫ) leads to .

s′ )γµ u(k. q 4 µν [using that (¯(k ′ )γµ u(k))∗ = u(k)γµ u(k ′ )]. s4 )γµ u(p.75) Note that the q µ q ν term in the photon propagator are irrelevant because the photon couples to a conserved current.5 Some examples eµ scattering The first example is the electromagnetic scattering of an electron and a muon. s4 )(−ie)γ ν u(p. To lowest order (∝ α = e2 /4π) only one diagram contributes. s3 )γ µ u(k. s′ ) ¯ u s.76) 1 .78) = 2 α2 2 s + u2 + 4t(M 2 + m2 ) − 2(M 2 + m2 )2 . s1 ) ¯ −igµν u(p′ . s2 ).s2 ) which can be written as (see also chapter 4) |M |2 = = 1 4s |¯(k ′ . s)¯(k.s4 ) and averaged over spins in the initial state (1/2 × 1/2 × s1 .s4 e4 (m) µν (M) L L . s2 )|2 ¯ q2 (10.Feynman diagrams for scattering amplitudes 100 10. ¯ q2 (10. s1 ) u e2 u(p′ . s3 )(−ie)γ µ u(k. 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 (10.79) .s2 .s′ ′ ′ = 2 kµ kν + 2 kν kµ + q 2 gµν 1 Tr [(/ ′ + m)γµ (/ + m)γν ] k k 2 ′ ′ = 2 kµ kν + kν kµ − gµν (k · k ′ − m2 ) (10.s3 . 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 ′ . where u ¯ L(m) µν = = 1 2 u(k ′ .77) Combining Lµν and Lµν (M) one obtains L(m) Lµν (M) = 2 s2 + u2 + 4t(M 2 + m2 ) − 2(M 2 + m2 )2 µν and M 4π 2 (m) (10. s)γν u(k ′ .

s3 . s4 )γ ν u(p. s3 )γ µ v(p′ . the masses. s1 )γν v(k ′ . s4 ) v (p′ . s2 )γµ u(k. s2 (10. This is known as crossing symmetry.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 . 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 10.1 (a) Give in diagrammatic notation the full Green functions G(4) (x1 . .s2 3 . s3 ) ¯ ¯ 1 .Feynman diagrams for scattering amplitudes 101 The second example is the annihilation of an electron pair and creation of a muon pair. . Similarly. s2 )γµ u(k.s4 = 1 2s × v (k ′ . s2 ) ¯ ¯ 1 2s u(p. momenta and invariants are e− e+ → µ− µ+ scattering k p s t = = (k + k ′ )2 = (p + p′ )2 (k − p)2 = (k ′ − p′ )2 (k − p′ )2 = (k ′ − p)2 k’ p’ u = The scattering amplitude squared (spins summed and averaged) is given by |M |2 = 1 4s e4 s2 |¯(p. s4 ) u e2 v (k ′ . (b) Use the definition of the source-connected Green function Gsc and show that indeed sourceconnected diagrams survive. .s2 . The diagram. To lowest order (∝ α = e2 /4π) only one diagram contributes. (4) (4) . s3 )γ µ v(p′ .2. for instance Møller scattering (e− e− → e− e− ) and Bhabha scattering (e− e+ → e− e+ ) are related using crossing symmetry. Exercises Exercise 10.80) Note the similarity in the amplitudes for eµ scattering and e− e+ → µ− µ+ . s1 )|2 ¯ s 1 . Basically the same diagram is calculated and the result is the same after the interchange of s ↔ t. s1 ) u(k. (c) Do the same for the connected Green function Gc .

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

1 kinematics in scattering processes Phase space The 1-particle state is denoted |p .3) which is useful because the total 4-momentum of the final state usually is fixed by overall momentum conservation. .e. . + pn )2 . In the CM frame the threshold value for s obviously is n 2 sthreshold = i=1 mi . p1 . q). (11. . s is for obvious reasons known as the center of mass (CM) energy. . s = (p1 + . A physical state has positive energy. . Here s is the invariant mass of the n-particle system. for instance for determining the threshold energy for the production of a final state 1 + 2 + . (11.7) pb = cm (Eb . + n.1) = (2π)3 2E (2π)4 (proven in Chapter 2). . . . 103 . It is a useful quantity. pb we start with the four vectors pa = (Ea . pn ) = (2π)4 δ 4 (P − pi ) dR(p1 . . . (11. . . pb ) satisfying p2 = m2 and p2 = m2 . The phase space is determined by the weight factors assigned to each state in the summation or integration over states. pn ). the 1particle phase space is d3 p d4 p θ(p0 ) (2π)δ(p2 − m2 ). .4) For two particle states |pa . Its square root. .Chapter 11 Scattering theory 11. . a a b b the quantity s = P 2 = (pa + pb )2 . . −q). pa ) and pb = (Eb . For two particles.6) (11. In that case cm pa = (Ea . pn ) = and the reduced phase space element by dR(s. The first is the CM system. i. This is generalized to the multi-particle phase space dR(p1 . To be specific let us consider two frequently used frames.2) (11.5) √ is referred to as the invariant mass squared. p) which satisfies p2 = E 2 −p2 = m2 . (2π)3 2Ei i=1 (11. and the total momentum four-vector P = pa + pb . (11. It is determined by the energy-momentum four vector p = (E. i n d3 pi .

m2 . a (11. m2 . which in the specific case a b also can be expressed as λ(s. ptrf ). Prove that |q| = cm Ea cm Eb (s − m2 − m2 )2 − 4 m2 m2 a a b b = 4s s + m2 − m2 a b √ = . 4s (11. Prove that trf Ea = |ptrf | = a s − m2 − m2 a b .11) (11. 2 s s − m2 + m2 a b √ . m2 . p2 ) = = where λ12 denotes λ(s. m2 ) a b . p1 . In that case trf pa = (Ea .13) (11.8) (11.14) One can. = 2 s λ(s.15) Explicit calculation of the reduced two-body phase space element gives dR(s. a a b b The second frame considered explicitly is the socalled target rest frame in which one of the particles (called the target) is at rest. m2 ). m2 ) = 4(pa · pb )2 − 4p2 p2 . 4π s 4π 8π s 4π (11. mb and s). Also in this case one can express the energy and momentum in the invariants. trf Ea (threshold) = 1 2 mb ( i mi )2 − m2 − m2 a b . m2 .9) (11. (11. 2mb (11.12) pb = (mb .Scattering theory 104 It is straightforward to prove that the unknowns in the particular system can be expressed in the invariants (ma . m2 ) is a function symmetric in its three arguments.10) The function λ(s. m2 ) a b . p1 . 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 √ = .16) . 0). for instance. 2mb λ(s. use the first relation and the abovementioned threshold value for s to calculate the threshold for a specific n-particle final state in the target rest frame. 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. m2 .

18) (11. At high energies. the rapidity y is defined T T x y y= 1 ln 2 E + pz E − pz = ln E + pz mT = tanh−1 pz .21) (11. where the first is referred to as elastic channel and the set of all other channels as the inelastic channels. The minimum and maximum values for t correspond to θcm CM = m2 + m2 − 2 Ea Ec ± 2 qq ′ a c = m2 a + m2 c (s + m2 − m2 )(s + m2 − m2 ) a c b d − ± 2s √ λab λcd . The variable s is always larger than a c b d the minimal value (ma + mb )2 . The initial state. A specific reaction channel starts contributing at the threshold value (Eq. pT = (px . Instead of the scattering angle. for instance the momenta of incoming (colliding) hadrons and it is useful to use for the produced particle rapidity as a variable.4). Consider the process a + b → c + d. 2s (11.22) (11. (11. usually many particles are produced.25) dR(s. Of course there are not only 2-particle channels.. pd ) = 4π s 4π 8π s 4π dt dt √ √ .. 11. In inclusive measurements no particles are detected in the final state. however. there is usually a preferred direction. 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 .20) → π0 + n → π+ + π− + n → .19) (11. in exclusive measurements all particles are detected.27) with m2 = m2 − p2 = m2 + p2 + p2 . Writing E = mT cosh y. tmax min λcd /4s. s = (pa + pb )2 = (pc + pd )2 t = (pa − pc ) = (pb − pd ) u = (pa − pd )2 = (pb − pc )2 2 2 (11. An often used set of invariants are the Mandelstam variables. = (11. pz = mT sinh y. a c with q = λab /4s and q ′ = being 0 or 180 degrees. Consider the 1-particle inclusive case. while the final state often arises from a series of 2-particle processes combined with the decay of an intermediate particle (resonance). dt = 2 qq ′ d cos θcm and obtain for the two-body phase space element √ q ′ dΩcm λcd dΩcm √ = (11. pc .24) Using the relation between t and cos θcm it is straightforward to express dΩcm in dt. in which one particle is detected.Scattering theory 105 Kinematics of 2 → 2 scattering processes π− + p The simplest scattering process is 2 particles in and 2 particles out. .17) (11. py ).26) = 16π q s 8π λab Kinematics of inclusive hard scattering processes In high energy (hard) scattering processes. H1 + H2 → h + X.23) which are not independent as s + t + u = m2 + m2 + m2 + m2 . E . The various possibilities are referred to as different reaction channels. Examples appear in → π− + p (11. usually is a 2-particle state.

31) . t. t. (11. 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. u = (p1 − p4 )2 = (p2 − p3 )2 . u). t = (p1 − p3 )2 = (p2 − p4 )2 .2 Crossing symmetry In the previous chapter. u) with s = (p1 + p2 )2 = (p3 + p4 )2 . With the momenta defined as in the figures above one has for all these processes the same amplitude M (s.29) 3 2E 3 2 (2π) 2 (2π) 16π 2π E 16π 2 2π with at high energies the factor |p|/E ≃ 1. These variables are precisely the Mandelstam variables for the s-channel process (ab → cd). u). c bd Mad→c¯(su = u. uu = (pa − p¯)2 = s. For the t-channel process (a¯ → ¯ one has c bd) st = (pa + pc )2 = t.30) (11. For large energies and not too small scattering angles (θ ≫ 1/γ) y is approximately equal to the pseudo-rapidity η. ¯ while for the u-channel process (ad → c¯ one has b) su = (pa + pd )2 = u. This is known as crossing symmetry. 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. η = − ln(tan(θ/2)) = tanh−1 (cos θ). ut = u) = 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. ¯ b (11. (11. uu = s) = Mab→cd (s. ¯ tu = (pa − pc )2 = t. tu = s. which means that rapidity distributions dN/dy maintain their shape. b Analiticity of the field theoretically calculated result implies Ma¯→¯ (st = t. t-channel and u-channel processes respectively. t. The one-particle phase space in terms of these variables becomes d3 p dy d|pT |2 dφ dy d2 pT |p| dη d|pT |2 dφ = = = .28) which (only involving angles) is easier to determine. b ut = (pa − pd )2 = u. ¯ tt = (pa − p¯)2 = s. 11. tt = s.

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

From the solution N (t) = N (0) e−Γ t one knows that the decay time τ = 1/Γ. .T V. as expected.Scattering theory 108 Although this at first sight does not look covariant. For this we employ in essence Fermi’s Golden rule stating that when the S-matrix element is written as Sf i = δf i − (2π)4 δ 4 (Pi − Pf ) iMf i (11. 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 . (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.35) This quantity is not covariant.39) One of the δ functions can be rewritten as T · V (remember the normalization of plane waves).34) i.36) (11. N and T · V are covariant.e. T · V ρa (11. the amount of decaying particles is proportional to the number of particles with proportionality factor the em decay width Γ.38) (in which we can calculate −iMf i using Feynman diagrams). . σ= √ 2 λab T · V (11.33) Decay of particles For the decay of particle a one has macroscopically dN = −Γ N. .T d4 x = V · T (2π)4 δ 4 (Pi − Pf ). (11. the number of scattered or decayed particles is given by 2 N = (2π)4 δ 4 (Pi − Pf ) iMf i dR(p1 . For the (proper) decay width one thus has 1 N Γ0 = .37) 2 ma T · V Fermi’s Golden Rule In both the scattering cross section and the decay constant the quantity N/T V appears. it is. N 1 . (11. Microscopically one has Ndecay = Na · Γ T or Γ= N 1 . . pn ). 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 . dt (11. .

46) Partial wave expansion Often it is useful to make a partial wave expansion for the amplitude M (s.44) This can be used to get the full expression for dσ/dt for eµ and e+ e− scattering. 11.40) T ·V Combining this with the expressions for the width or the cross section one obtains for the decay width Γ = 1 2m = dR(m2 . q n ). The sign difference comes from the (conventional) sign in relation between S and quantummechanical and relativistic scattering amplitude. (11. . i M† M √ − √ 8π s 8π s = fi n dΩn ∗ Mnf qn Mni √ √ . . θ)|2 . θcm ) √ q 8π s 2 2 (11. 16π 2 s (11. p1 . 2 λab and for instance for two particles dσ = q ′ M (s. for which the √ amplitudes squared have been calculated in the previous chapter. The amplitude −M /8π s is the one to be compared with the quantum mechanical scattering amplitude f (E. √ (2ℓ + 1)Mℓ (s) Pℓ (cos θ). (11. (M † )f n (2π)4 δ 4 (Pi − Pn ) Mni . i.41) (11. . p1 .42) 2−body decay The differential cross section (final state not integrated over) is given by 1 dσ = √ |Mf i |2 dR(s. θ) or M (q i . (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 . .25) −i Mf i − (M † )f i = − ∗ dΩ(ˆn )Mnf (q f . q n ) q n qn √ Mni (q i . pn ). implies for the scattering matrix M . θ). . . . 8π s 2π 8π s . t) λab 4π dt.45) Since the amplitudes also depend on all momenta the full result for two-particle intermediate states is (in CM.e. q f ). . (11. .4 Unitarity condition (S † )f n Sni = δf i The unitarity of the S-matrix. θ) = −8π s ℓ (in analogy with the expansion for f (E. θ) in quantum mechanics. see 11. note the sign and cos θ = qi · qf ). ˆ ˆ Inserted in the unitarity condition for M .Scattering theory 109 (Using normalized wave packets these somewhat ill-defined manipulations can be made more rigorous). . . p1 . pn ). respectively. for which one has dσ/dΩ = |f (E. The result is N = |Mf i |2 dR(s.47) M (s. (11. pn ) |M |2 q 32π 2 m2 dΩ |M |2 . .43) dΩcm = π M (s.

49) or Im Mℓ = q |Mℓ |2 . Show that σT = = 4π q 2π q2 (2ℓ + 1) Im Mℓ (s) ℓ ℓ (2ℓ + 1) (1 − ηℓ cos 2δℓ ).51) and for the case that this is the only channel (purely elastic scattering.52) From the imaginary part of M (s. (11.Scattering theory 110 we obtain LHS = −i while for the RHS use is made of ℓ † (2ℓ + 1) Pℓ (ˆi · qf ) (Mℓ )f i − (Mℓ )f i . which allows writing Mℓ (s) = Sℓ (s) − 1 e2i δℓ (s) − 1 = . (11. † −i (Mℓ )f i − (Mℓ )f i = 2 † (Mℓ )f n qn (Mℓ )ni . the total cross section can be determined. 0). parametrized as Sℓ (s) = ηℓ (s) exp(2i δℓ (s)). n (11. Using σ= dΩ q ′ M (s.53) The difference is the inelastic cross section. η = 1) the result σel = 4π q2 (2ℓ + 1) sin2 δℓ . (11. 2i q 2i q (11.48) If only one channel is present this simplifies to ∗ ∗ −i (Mℓ − Mℓ ) = 2 qMℓ Mℓ . 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 .e. In general a given channel has |Sℓ (s)| ≤ 1. θcm ) √ q 8π s 2 .50) where Sℓ (s) satisfies |Sℓ (s)| = 1 and δℓ (s) is the phase shift. q ˆ i. σinel = π q2 2 (2ℓ + 1) (1 − ηℓ ). (11. in combination with the partial wave expansion for the amplitudes M and the orthogonality of the Legendre polynomials immediately gives for the elastic channel.54) ℓ . ℓ (11. σel = 4π ℓ (2ℓ + 1)|Mℓ (s)|2 (2ℓ + 1) ℓ = 4π q2 ηℓ e 2i δℓ − 1 2i 2 .

This is what we will do to discuss decay widths. but still real. however.57) This shows that Γ is the width of the resonance. For an unstable particle one expects that U (t.58) . 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. This is achieved with a propagator i∆R (k) = i k2 − M2 + iM Γ (11.56) (again disregarding spin). To check unitarity we need the amplitude at order e4 .5 Unstable particles i k 2 − M 2 + iǫ For a stable particle the propagator is i∆(k) = (11. it is straightforward to write down the partial wave amplitude. such that |U (t. .e. 11. The unitarity condition then states 2 Im ∆−1 (k) = R n † dR(p1 . We note that unitarity is generally broken in a finite order calculation. The quantity Γ is precisely the width for unstable particles. i. η = 0. . σel = σinel . .55) (note that we have disregarded spin). . 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. ∝ U (t. which is given by a sum of the partial widths into the different channels. 0)|2 = e−Γt . one has poles at k 0 = ±(Ek − iǫ) where Ek = + k2 + M 2 guarantees the correct behavior. i. The product of amplitudes is of order e4 .e. s − M 2 + iM Γ (11. the time-evolution operator.Scattering theory 111 Note that the total cross section is maximal in the case of full absorption. 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 . 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 . 0). For the amplitude in a scattering process going through a resonance. The prescription for the pole structure. For instance the tree-level calculation of Feynman diagrams at order e2 was real. (q Mℓ )ij (s) = −M Γi Γj . pn )MRn (2π)4 δ 4 (k − Pn ) MnR = 2M n Γn = 2M Γ. 0) ∝ e−i(E−iΓ/2)t . in which case. (11.

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

a quantity which we will ¯ ¯ encounter in the next chapter (section 12. Exercise 11. we have here already used unitarity to find a finite imaginary part that is proportional to the total decay width. cm |va − vb | in the center of mass system for two colliding particles a and b. writing N (s) M1 (s) = 2 + iM Γ . As discussed for unstable particles. 4. s − M∆ ∆ ∆ 2 calculate Re N (M∆ ) . ρcm ρcm a b 1 cm . Show that unitarity can be used one step further to fix also the real part of the residue of the pole in the amplitude.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∆ .Scattering theory 113 Exercise 11.3 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. Thus.6 × 10−8 s and the relative decay widths Γ(π − → µ− νµ ) = 0.2 Calculate the flux factor.000 123 Γ. Compare the results with the experimental π − lifetime τ = 2.999 877 Γ and Γ(π − → e− νe ) = 0.2). Exercise 11.8. Calculate now fπ . Compare the result with the factor trf 1/ρa ρb va that we used in the derivation of the basic expression for cross sections.

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

18) or considering a path xµ (s) from a fixed origin (0) to point x. What is this physical effect by which the ’connection’ Aµ can be observed? For this consider the phase around a closed loop. 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 . should be independent of such transformations.14) (12. (12. i. Dν ] ψ(x) 1 − ig dσ µν Gµν ψ(x). The relation in Eq. We note that local gauge invariance requires that we can modify all local systems with a (local) unitary transformation S(x).15) ψ(x). (12. (12. if two different paths to the same point give different phases the effects can be observed. -dx -dy µ µ dyµ µ dx For a vector. 12. = .19) This gives rise to a (path dependent) phase in each point. x+dx ψ(x) = = 1 + ig dxµ Aa (x)Ta µ 1 + ig dx Aµ (x) µ x x ψ(x) (12.The standard model the basis for x G and x+dx 117 G. ds s dxµ ds′ P exp ig A (s′ ) ψ(0). ds which is solved by dψ(s) ds = dxµ . A ’constant’ vector that only rotates because of the arbitrary definitions of local frames satisfies Dµ ψ(x) = 0. x x+dx ψ(x + dx) = (1 + dxµ ∂µ ) ψ(x). However. 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.16.e. x+dx ψ(x + dx) = 1 + dxµ Dµ x+dx ψ(x). In principle such a phase in a given point is not observable. requiring that the ’connection’ Aµ (x) is such that Dµ → S(x)D µ S −1 (x). ds′ µ 0 ig Aµ (s) ψ(s) dxµ Aµ (x) ψ(s) = which is the path-ordered integral denoted ψ(x) = P eig P ψ(0).16) which in the presence of the ’connection’ Aµ differs from the total change between x ψ(x) and x+dx ψ(x+ dx). but expresses them with respect to different bases. ∂µ − ig Aµ (x) ψ(x) = 0 dxµ ∂µ − ig Aµ (x(s)) ψ(x) = 0.e. (12. which connects two identical vectors. i.17) The three equations given so far immediately give Dµ = ∂µ − ig Aµ (x).

we can disregard those ’perturbations’ and look at the ideal situation. 2 2 4 −V (φ) The potential V (φ) is shown in fig. Consider as an example a ferromagnet with an interaction hamiltonian of the form H=− Jij Si · Sj . 12. which means that by an appropriate transformation S(x) it can be gauged away (see the example of Aharonov-Bohm effect).    φ1    φ . 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. Nevertheless. If it is zero one has dxµ Aµ (x) = 0. e. . Dν ]. This characterizes spontaneous symmetry breaking.2. Only if this quantity is nonzero a physically observable effect of Aµ exists. As there can be one and only one groundstate. and equivalently Aµ can be considered as a pure gauge effect.The standard model 118 where Gµν = (i/g)[Dµ .  φ= 2      φ3 (12. Nature will choose one. If the temperature is high enough the spins are oriented randomly and the (macroscopic) ground state is spherically symmetric. this means that there is more than one possibility for the groundstate. 12.20) with a lagrangian density of the form L = 1 1 1 ∂µ φ · ∂ µ φ − m2 φ · φ − λ(φ · φ)2 .2 Spontaneous symmetry breaking In this section we consider the situation that the groundstate of a physical system is degenerate. i>j which is rotationally invariant.e. In geometric language the effect on parallel transport of a vector depends on the ’curvature’ Gµν . 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). the groundstate itself appears degenerate. usually being (slightly) prejudiced by impurities.2: The symmetry-breaking ’potential’ in the lagrangian for the case that m2 < 0. a theory for a scalar degree of freedom (a scalar field) having three (real) components. external magnetic fields.g. i. In this case there are many possible groundstates (determined by a fixed direction in space). in reality a not perfectly symmetric situation. Classically the (time-independent) ground state is found for V(φ ) F |φ| Figure 12.

ϕ2 and η. has less symmetry. while the lagrangian including the nonzero groundstate expectation value chosen by nature. 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. The field along the third axis plays a special role because of the choice of the vacuum expectation value.25) 2 2 2 Therefore there are 2 massless scalar particles. i.e. 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 .) terms constitute interaction terms.   ϕc = 0|φ |0 =  (12. In this case the spectrum is described . .21) Stability of the action requires the classical groundstate ϕc to have a well-defined value (which can be nonzero). In the case of degeneracy. In order to see the consequences for the particle spectrum of the theory we construct the lagrangian in terms of the fields ϕ1 . . φ = ϕc + φquantum where for the (operator-valued) coefficients in the quantum field one wants 0|c† = c|0 = 0. It is sufficient to do this to second order in the fields as the higher (cubic. . Any constant field with ’length’ F is a possible groundstate. therefore a choice must be made. λ such that 0|ϕ1 |0 = 0|ϕ2 |0 = 0|η|0 = 0.23)     F +η the latter only forming a minimum for m2 < 0. It is appropriate to redefine the field as    ϕ1      φ =  ϕ2  . In this mode the lagrangian and the vacuum are both invariant under a set of symmetry transformations generated by Qa . The presence of a nonzero value for the classical groundstate value of the field will have an effect when the field is quantized. They are known as the Weyl mode or the Goldstone mode: Weyl mode. (12. (12. The η massless particles are called Goldstone bosons. so one has 0|φquantum |0 = 0 and hence 0|φ|0 = ϕc . It is only invariant under rotations around the 3-axis. The classical groundstate appears degenerate. The situation now is the following. (12.22)     F Realization of symmetries (model independent) In this section we want to discuss a bit more formal the two possible ways that a symmetry can be implemented. The original lagrangian contained an SO(3) invariance under (length conserving) rotations among the three fields. Writing the field φ as a sum of a classical and a quantum field.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 this situation one speaks of spontaneous symmetry breaking. say    0    0 . etc. A quantum field theory has only one nondegenerate groundstate |0 . for the vacuum Qa |0 = 0. while the quadratic terms must correspond with non-negative masses.24) 1 2 2 4 1 1 1 = (∂µ ϕ1 )2 + (∂µ ϕ2 )2 + (∂µ η)2 + m2 η 2 + .

(12. π (12. Neglecting at this point the local color symmetry. Taking the derivative. A bit more formal. Using the Dirac equation. The conserved currents corresponding to this symmetry transformation are found directly using Noether’s theorem (see chapter 6). = ψ(i/ − M )ψ. i. a massless Goldstone boson for each ’broken’ generator. (12. 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) (12.30) which in the case that we include only two flavors form an SU (2)V symmetry (isospin symmetry) generated by the Pauli matrices.e. strange. there must be a nonzero expectation value 0|Jµ (x)|π (k) . the lagrangian for the quarks consists of the free Dirac lagrangian for each of the types of quarks. ∂ ∂ (12. if the generators Qa generate a symmetry. In this mode the lagrangian is invariant but Qa |0 = 0 for a number of generators. down. This lagrangian density then is invariant under unitary (vector) transformations in the flavor space.  . [Qa . T = τ /2. T ] ψ. .      ψu   mu           md     ψ =  ψd  .    . called flavors. Known examples are rotational symmetry and the fact that the spectrum shows multiplets labeled by angular momentum ℓ (with members labeled by m).32) (12.31) .e. H] = 0.The standard model 120 by degenerate representations of the symmetry group. π Example: chiral symmetry An example of spontaneous symmetry breaking is chiral symmetry breaking in QCD. Including a sum over the different flavors (up. Note that for the fields π a (x) one would have a the relation 0|π a (x)|π a (k) = ei k·x .26) for all the states labeled by a corresponding to ’broken’ generators. irrespective of the fact if they annihilate the vacuum. a and a′ are degenerate states. i.e. 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′ . suggesting the stronger relation ∂ µ Jµ (x) = fπ m2 a π a (x).) one can write L = f ψ f (i/ − Mf )ψf . V µ = ψγ µ T ψ. ψ −→ ei α·T ψ. As the a generators for a symmetry are precisely the zero-components of a conserved current Jµ (x) integrated a a over space. This means that they are operators that create states from the vacuum. Lx and Ly or instead of the latter two L+ and L− ) are used to label the multiplet members or transform them into one another. The generators Qa (in that case the rotation operators Lz . i. a 0|∂ µ Jµ (x)|π b (k) = fπ k 2 ei k·x δab = fπ m2 a ei k·x δab .28) where ψ is extended to a vector in flavor space and M is a diagonal matrix in flavor space. (12.29) (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.    . and one must have mπa = 0.27) If the transformations in the lagrangian give rise to a symmetry the Noether currents are conserved. denoted |π a (k) . Goldstone mode.. etc. M =      . it is easy to see that one gets ∂µ V µ = i ψ [M.

M = m · 1. etc. That the massless theory has this symmetry can also be seen by writing it down for the socalled lefthanded and righthanded fermions. e.3 MeV/c2 and Mn = 939. ∂ ∂ (12. 12. The chiral fields ψR and ψL are transformed into each other under parity. . i. ψ −→ ei α·T γ5 ψ. including in Dirac space a γ5 matrix. 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. which is equivalent 2 to the V-A symmetry.35 is nonzero. as the quantity ∂µ Aµ (x). i. i. both are very small.e. This symmetry. parity would not play a role in the world. ψR/L = 1 (1 ± γ5 )ψ.e.6 MeV/c2 ). The currents corresponding to this symmetry transformation are again found using Noether’s theorem. (12. but also a triplet (isospin 1) of pions. however. behaves as a pseudoscalar particle (spin zero. it is easy to see that one gets ∂µ Aµ = i ψ {M. 12. T γ5 = τ γ5 /2. (12. which is approximately true for the up and down quarks. stated equivalently. Therefore realization in the Weyl mode would require that all particles come double with positive and negative parity. fπ = 0 remains.e.e. I. This situation. From the number of broken generators it is clear that one expects three massless Goldstone bosons. T } γ5 ψ. the lagrangian density is invariant under left (L) and right (R) rotations independently. Still the basic fact that the generators acting on the vacuum give a nonzero result (i. socalled axial vector transformations.27.e. while the groundstate is only invariant under isospin rotations (R = L). in terms of which the Dirac lagrangian 2 density looks like L = i ψL / ψL + i ψR / ψR − ψR M ψL − ψL M ψR .33) which in the case of two flavors form SU (2)A transformations generated by the Pauli matrices.The standard model 121 Furthermore ∂µ V µ = 0 ⇐⇒ [M. much smaller than any of the other mesons or baryons.g. a doublet (isospin 1/2) of nucleons. etc. if in flavor space M is proportional to the unit matrix. The Goldstone bosons of QCD are the pions for which fπ = 93 MeV and which have a mass of mπ ≈ 138 MeV/c2 . gives also rise to a nonzero mass for the Goldstone bosons according to Eq. (12.36) 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. parity minus). is what happens in the real world. for which the field (according to the discussion above) has the same behavior under parity. where the quark masses are not completely zero. generated by TR/L = 1 (1 ± γ5 )T . nucleons and atomic nuclei has not only an isospin or vector symmetry SU (2)V but also an axial vector symmetry SU (2)A . is by nature not realized in the Weyl mode. Note that these transformations also work on the spinor indices.34) Using the Dirac equation. (leaving out the flavor structure) the same as ψγ5 ψ.35) In this case ∂µ Aµ = 0 will be true if the quarks have zero mass. with almost degenerate masses (Mp = 938. Therefore the world of up and down quarks describing pions. T ] = 0. chiral symmetry is not perfect. or. We know that mesons and baryons (such as the nucleons) have a well-defined parity that is conserved. nature has choosen the path SU (2)L ⊗ SU (2)R =⇒ SU (2)V . There exists another set of symmetry transformations. SU (2)V (isospin) symmetry is good if the up and down quark masses are identical. but the fact that the symmetry is not perfect and the right hand side of Eq. Aµ = ψγ µ T γ5 ψ. In the real world. From group theory (Schur’s theorem) one knows that the latter can only be true. How can we see this. This symmetry is realized in the Weyl mode with the spectrum of QCD showing an almost perfect isospin symmetry. This combined symmetry is what one calls chiral symmetry. proton and neutron.

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

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

The standard model 124 3 3 where Wµ is a triplet of gauge bosons with TW = 1. 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. . −V (φ) L and (h2) = −Ge (LφR + Rφ L). the U (1)Q symmetry (generated by the charge operator). a complex Higgs field   +   1  φ   √2 (θ2 + iθ1 )    0 = 1   (12. For the classical field the choice θ4 = v is made.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. . 2 and 3 may be eliminated (the necessary SU (2)W rotation is precisely e−iθ(x)·τ ).58) (12.56)  v + h(x) 2 Dµ φ = (∂µ − and     Dµ φ =     = 1 √ 2 − ig 2 ∂µ h + 1 2 Wµ −iWµ √ (v + h) 2 3 gWµ −g′ Bµ i √ (v + 2 2 Up to cubic terms. this leads to the lagrangian L (h1)     .59) . .54) = (Dµ φ)† (Dµ φ) −m2 φ† φ − λ (φ† φ)2 . (12.   h)  (12.57) = 1 g2 v2 1 2 (Wµ )2 + (Wµ )2 (∂µ h)2 + m2 h2 + 2 8 v2 2 3 gWµ − g ′ Bµ + . (12.52) L L (f 1) (f 2) = = i Rγ µ (∂µ + ig ′ Bµ )R + i Lγ µ (∂µ + i ′ i g Bµ − g Wµ · τ )L 2 2 1 1 − (∂µ Wν − ∂ν Wµ + g Wµ × Wν )2 − (∂µ Bν − ∂ν Bµ )2 . + 8 g2 v2 1 + − (Wµ )2 + (Wµ )2 (∂µ h)2 + m2 h2 + 2 8 (g 2 + g ′2 ) v 2 (Zµ )2 + . Using local gauge invariance θi for i = 1. + 8 (12. 4 4 In order to break the symmetry to the symmetry of the physical world. . † i i g Wµ · τ − g ′ Bµ )φ. with the following lagrangian density consisting of a symmetry breaking piece and a coupling to the fermions. leading to the parametrization   1  0    φ(x) = √  (12. L where L (h1) (h) =L (h1) +L (h2) .53) φ=   √ (θ4 − iθ3 )  φ 2 with TW = 1/2 and YW = 1 is introduced. . Putting this in leads to L (f ) =L (f 1) +L (f 2) . which assures with the choice of YW for the Higgs field that Q generates the remaining U (1) symmetry.

. 2 Zµ (12.67) The coupling of electrons or muons to their respective neutrinos. 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 .69) .64) (12.61) (12. and correspond to three massive particle fields (W ± and Z 0 ) and one massless field (photon γ) with 2 MW 2 MZ 2 Mγ = = = g2 v2 .66) From the coupling to the photon. 2 + g ′2 g 3 g ′ Wµ + g Bµ (12.63) (12. 2 (12. 2θ 2θ 4 cos W cos W 0. (12. we can read off e = g sin θW = g ′ cos θW .68) g2 ρ 5 2 (νµ γρ (1 − γ5 )µ) (eγ (1 − γ )νe ) 8 MW (µ)† (jL )ρ (e) (jL )ρ GF (µ)† (e) ≡ i √ (jL )ρ (jL )ρ . σ (νµ γ ρ µL ) 2 2 (eL γ νe ) 2 k + MW (12. (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 . g ′ /g = tan θW .62) g 2 + g ′2 3 ≡ sin θW Wµ + cos θW Bµ .60) (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 + .65) The weak mixing angle is related to the ratio of coupling constants. 4 2 MW g2 v2 = . 2 3 g Wµ − g ′ Bµ 3 ≡ cos θW Wµ − sin θW Bµ .

(12. from the MW mass is about 250 GeV. g ′ and v determine a number of experimentally measurable quantities. Γe ≈ Γµ ≈ Γτ ≈ 83 MeV).19 GeV. Comparing this with the total width into (invisible!) channels. Γinvisible = 480 MeV one sees that three families of (light) neutrinos are allowed. .40 GeV.72) (12. The coupling to the Higgs particle is weak as the value for v calculated e. i.The standard model 126 the good old four-point interaction introduced by Fermi to explain the weak interactions.77) (12. 3 = TW . v (12.70) = 2 = 2 sin2 θ 8 MW 2 v2 8 MW 2 W In this way the parameters g.g. (12. 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.e. 1.5 MeV (exp. With the choosen vacuum expectation value for the Higgs field. (12.231 2. one obtains L (h2) Ge v Ge = − √ (eL eR + eR eL ) − √ (eL eR + eR eL ) h 2 2 me = −me ee − eeh.78) From this coupling it is straightforward to calculate the partial width for Z 0 into a fermion-antifermion pair. g2 MZ 2 2 (CV + CA ).71) GeV −2 = = = = . muon or tau.166 4 × 10 0. 80. one has the relation g2 e2 1 GF √ = .76) with CV CA 3 = TW − 2 sin2 θW Q. Finally we want to say something about the weak properties of the quarks.79) Γ(Z 0 → f f ) = 48π cos2 θW For the electron. as appear for instance in the decay of the neutron or the decay of the Λ (quark content uds).80) First. The masses of the fermions and the coupling to the Higgs particle are contained in L (h2) . −5 (12. (12. the mass of the electron comes from the spontaneous symmetry breaking but is not predicted (it is in the coupling Ge ). 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 . For each neutrino species (with CV = 1/2 and CA = 1/2 one expects Γ(νν) ≈ 155 MeV. 2 2 2 (12. i.05 and CA = −1/2 we calculate Γ(e+ e− ) ≈ 78.73) (12.e. me /v is extremely small. leptons with CV = −1/2 + 2 sin2 θW ≈ −0. 91.74) (12. giving rise to a running coupling constant after renormalization of the field theory. such as2 α = e2 /4π GF sin2 θW MW MZ ≈ 1/127.

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

(12.   φ =    −φ− 0 2 For the (squared) complex matrices we can find positive eigenvalues. d d (12.q) ′ weak † mass mass DL = DL = Vu Vd DL = VCKM DL . 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 . A ≈ 0. 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 12.227.e. † Λu Λ† = Vu G2 Vu .g. Thus one has † † † =⇒ −DL Vd Md Wd DR − DR Wd Md Vd† DL − UL Vu Mu Wu UR − UR Wu Mu Vu UL (12.The standard model 128 using the socalled Wolfenstein      + O(λ4 )   with λ ≈ 0. each of these containing   the three families. Note that we need the conjugate Higgs field to get a U (1)Y singlet in the case of the charge +2/3 quarks. the weak mass eigenstates are L (h2. allowing us to write † Λu = Vu Gu Wu † and Λd = Vd Gd Wd . (12. the three righthanded quarks with charge +2/3 in UR and the three righthanded quarks with charges −1/3 in DR .83) where Vu and Vd are unitary matrices.34.86) the unitary CKM-matrix introduced above in an ad hoc way.82 and ρ ≈ 0. The Higgs field is still limited to one complex doublet. respectively). for which we need the appropriate weak isospin doublet  0∗    1    φ c  = √  v +h . mc and mt ) and Md = Gd v/ 2 (diagonal matrix containing md . The Λu and Λd are complex matrices in the 3×3 family and Λd Λ† = Vd G2 Vd† . One then reads off that starting with the family basis as defined via the left doublets that the mass eigenstates (and states coupling to the Higgs field) involve the righthanded † mass † mass mass † mass states UR = Wu UR and DR = Wd DR and the lefthanded states UL = Vu UL and DL = † Vd DL . ms and mb ).82) space. The imaginary part i η gives rise to CP violation in decays of K and B-mesons (containing s and b quarks.84) with Gu and Gd being real and positive and Wu and Wd being different unitary matrices. UR = uR cR tR .22 and η ≈ 0.5 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 (h2.85) √ √ with Mu = Gu v/ 2 (diagonal matrix containing mu .q) = −QL φΛd DR − DR Λ† φ† QL − QL φc Λu UR − UR Λ† φc† QL u d (12. . e. i. u u where we include now the three lefthanded quark doublets in QL .

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

Thus (disregarding family structure) one has two Majorana neutrinos. (12. with for the righthanded sector a mass term MR .       0 0 1 L mass. however. The mass matrix can be written as   ML |MD | eiφ     M = (12. We use here the primes starting with the weak eigenstates. For the neutrino sector.and righthanded species then just form a Dirac fermion.94) M L NL NL + M L NL NL . 1 c ∗ c L mass. 12. Absorption of phases in the states is not possible for Majorana neutrinos.ν =− containing three (CP-violating) phases (α1 . Υc ≡ (ΥL )c = ΥR 1 2 L and Υc = ΥL ).c. The way to circumvent this is to introduce as in the previous section righthanded neutrinos. 2 although this option is not attractive as it violates the electroweak symmetry. a unitary matrix relating them to the weak eigenstates. couple the right. coupled by a Dirac mass term. For the leptons. This is called the see-saw mechanism (outlined below). 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. 1 ∗ c c (12.100)  |MD | eiφ MR 1 1 ∗ ∗ M L nc nL + M L nL nc − M R nR nc + M R nc nR L R L R 2 2 1 1 ∗ ∗ − M D nc nc + M D nL nR − M D nR nL + M D nc nc L R R L 2 2     1  nc nR   M L M D   nL   L   c  + h.99) .ν = − MR NR NR + MR NR NR .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. −    MD MR nR 2 (12.98) (12. Thus 2 Υ ′ = nc + nL .95) 2 In order to have more than a completely decoupled sector. For the charged leptons there cannot exist a Majorana mass term as this would break the U(1) electromagnetic symmetry.The standard model 130 The lepton sector (massive Majorana fields) An even simpler option than sterile righthanded Dirac neutrinos. The see-saw mechanism Consider (for one family N = n) the most general Lorentz invariant mass term for two independent Majorana spinors. 2 R ψ = nR + nL . For these light Majorana neutrinos one has. as above. hence the mixing matrix becomes   iα /2 0 0   e 1     (12. Υ′ and Υ′ (satisfying Υc = Υ and as discussed in chapter 6. If the Majorana mass MR ≫ MD one actually obtains in a natural way one Majorana neutrino with a very small mass.96) VPMNS = UPMNS K with K =  0 eiα2 /2 0  . one must for the neutrinos as well as charged leptons. α2 and δ). Actually. the massless and massive Majorana neutrinos. the left. one being massive. are equivalent to two decoupled Majorana neutrinos (see below). 1 L Υ ′ = nR + nc . (12.and lefthanded species through Dirac mass terms coming from the coupling to the Higgs sector as in the previous section. is to add in Eq.88 a Majorana mass term for the (lefthanded) neutrino mass eigenstates.

which implies U ∗ M † U † = U ∗ M ∗ U † = M0 and a ’normal’ diagonalization of the (hermitean) matrix M M † .103) for each of which one finds the lagrangians L = and we are left with two decoupled Majorana fields      Υ1L    = U ∗  nc  . Exercises Exercise 12.105) for i = 1.  (12. 2 with real masses Mi . i ψ {M.e.2 Derive for the vector and axial vector currents. V µ = ψγ µ T ψ and Aµ = ψγ µ γ5 T ψ ∂µ V µ ∂µ A µ = = i ψ [M.  L      Υ2L nR Υ1 and Υ2 . 2 U (M M † ) U † = M0 . Prove that if the generators Qa generate a symmetry. Exercise 12. T } γ5 ψ. (12.      Υ2R nR (12.e. i. . [Qa .104) ↔ 1 1 Υi i / Υi − Mi Υi Υi ∂ 4 2 (12. H] = 0. 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′ . For the situation ML = 0 and MR ≫ MD (taking MD real) one 2 finds M1 ≈ MD /MR and M2 ≈ MR .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)) . 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).The standard model 131 taking ML and MR real and non-negative.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 . (12. related via    c   Υ1R     = U  nL  . T ] ψ.3 (a) The coupling of the Z 0 particle to fermions is described by the vertex −i g 2 cos θW f f CV γ µ − CA γ µ γ5 . Exercise 12.101) Thus one obtains from  2 ML + |MD |2  † MM =   |MD | ML e+iφ + MR e−iφ the eigenvalues 2 M1/2 |MD | ML e−iφ + MR e+iφ 2 MR + |MD |2   . This is a mixing problem with a symmetric (complex) mass matrix leading to two (real) mass eigenstates. a and a′ are degenerate states. i.

The mass of the Z 0 is MZ = 91 GeV. to be precise one has for bosons (b) or fermions (f). at least up to a high (which?) order in λ. 0 ¯ Γ(Z 0 → f f ) = MZ g2 48π cos2 θW f f CV 2 + CA 2 . Γ(Z 0 → e+ e− ). Exercise 12. 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 +. 2 2 The masses are mt ≈ 175 GeV. Mf M2 and ghf f = . Neglect the masses of fermions and use the fact that the sum over polarizations is 3 pµ pν . hhh.The standard model 132 with CV CA = = 3 TW − 2Q sin2 θW . and the width to a pair of neutrino’s. In the standard model this coupling is described by the vertex −i g µ √ (γ − γ µ γ5 ) . Write down the matrix element squared (averaged over initial spins and summed over final spins) for the decay of the Z 0 . mb ≈ 5 GeV and MW ≈ 80 GeV. ǫ(λ) (p)ǫ(λ)∗ (p) = −gµν + µ ν M2 λ=1 ¯ to calculate the width Γ(Z → f f ). Exercise 12. 3 TW . Γ(Z 0 → νe νe ). ZZh.231. Note that you can find the answer without explicit construction of all interaction terms in the lagrangian.) are proportional to the mass squared for bosons or mass for fermions of the particles. Exercise 12.7 In this exercise two limits are investigated for the two-Majorana case. etc. .5 Show that the couplings to the Higgs (for W + W − h. Exercise 12.6 Check that the use of the Wolfenstein parametrization in the CKM matrix indeed gives a unitary matrix. (b) Calculate the width to electron-positron pair. e+ e− h. the weak mixing angle is given by ¯ sin2 θW = 0.

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

Sign up to vote on this title
UsefulNot useful