QUANTUM FIELD THEORY

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

November 2011 (version 6.04)

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

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

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

1

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

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

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

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

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

3 2. The quantum theory of fields.4 6. Quantum Field Theory. but are rather compact. M.3 2.3 2. 3.2 2. 2007.3 3.2 6. The notes contain all essential information.3 6. These notes Section 1.3 3. I refer to the book of Srednicki [1] and Ryder [2]. 2.E.5 Ryder Peskin & Schroeder Srednicki 1 1 1 1 3 2.2 4. Itzykson and J. 1980.2 3. Cambridge University Press. Quantum Field Theory. Corresponding chapters in books of Ryder.3 4.2 2.3 3. 1.-B.8 3.1 6. 1985.2 3. 3. S. Peskin and D.1 4. Cambridge University Press.2 6.3. 4.4 3.2. 2.1 1.2 36 22 . 5.7 2. I: Foundations. 1995. Cambridge University Press.1. Cambridge University Press.2 1.3 2. An introduction to Quantum Field Theory.1 3.2 3. Ryder.4 2. M. 6. Veltman. McGraw-Hill.5 5.2 2. The schedule in the Fall of 2011 is Chapters 1 through 10 in period 2 (7 weeks in November and December 2011) and Chapters 11 and 12 in period 3 (January 2012). Other text books of Quantum Field Theory that are useful are given in refs [3-6]. Zuber. Vol. 1994. 1995.H.5 2. Addison-Wesly.1 2 2 3. 3.3 2. Weinberg. Quantum Field Theory.V. Vol. M.1 2.3 References As most directly related books to these notes. Peskin & Schroeder and Srednicki. Diagrammatica. C.2 3. II: Modern Applications.5 2.1 5.1 3.3 3. L. Cambridge University Press. 1996.3 3. Srednicki.5 4. Schroeder.4 4.4 3.4 3.2 3.4 2.1 2.

5 9. 4.5 .6 8.1 11.4 These notes Section 7.2.5 9.5 4.1 7.1.5 6.4 3. 6.3 7.1 12.3 12.3 8.3 9.4 12.7 5.4. 2.8 6.3 6. 4.5 Ryder 4 4 4 4 4 4 Peskin & Schroeder 2.1 9.1 10. 2.2 4.5 7.1 5.5 2.2 5. 9.6 3. 6.3 2. 4.4 9.2 11.1 9.3 20.3 3.7 4.5 6 7 6 6.5 11.2.6 10.1.6 3.3 6 8 9 Srednicki 3 3 3 37 5.3 11.5 4.5 7.2.10 8. 9. 6.4 2.2 8.4 10.1 20.1 20.3 7.1.7.4.5.3 8.4 4.2.1 8.2 12.6 3.6 9.3. 20.4 8.3.4.1 20.2 10.3.8 5 4.2 9.2 8. 8.4 12.6 6. 6.6 4. 6.3 10. 4.2 7.4 7.

2 . • Slight rewriting of discussion of Majorana mass term in section 6.3 • First two sentences of Chapter 5 have been modified.5 Corrections 2011/2012) • Extension of Exercise 3.

quantum effects do not play a role anymore and one is in the classical domain. the action of the momentum operator in the coordinate representation is not as simple as the position operator. Indeed. In quantum mechanics. being eigenvalues of the position operators. (1.582 119 15 (56) × 10−22 MeV s. c = 299 792 458 m s−1 . (1.4) where δij is the Kronecker δ function. It is given by pop ψ(r) = −i ∇ψ(r). They satisfy the well-known (canonical) operator commutation relations [ri . In special relativity a special role is played by the velocity of light c.5) One can also choose a representation in which the momenta of the particles are the dynamical variables. In the framework of classical mechanics as well as quantum mechanics the position of a particle is a well-defined concept and the position coordinates can be used as dynamical variables in the description of the particles and their interactions. 1 (1. (1. One can talk about states |r and the wave function ψ(r) = r||ψ . pj ] = i δij . (1. Corresponding position and momentum operators do not commute. the position can in principle be determined at any time with any accuracy.6) .054 571 68 (18) × 10−34 J s = 6.3) The uncertainty principle tells us that in this representation the momenta cannot be fully determined. ˜ ψ(p) = i d3 r exp − p · r ψ(r).1) In the limit that the action S is much larger than .2) In the limit that v ≪ c one reaches the non-relativistic domain. S ≫ . ≡ h/2π = 1.1 Quantum field theory In quantum field theory the theories of quantum mechanics and special relativity are united. (1. In quantum mechanics a special role is played by Planck’s constant h. In this coordinate representation the position operators r op simply acts as rop ψ(r) = r ψ(r).Chapter 1 Introduction 1. ˜ The corresponding states are |p and the wave functions ψ(p) = p||ψ are the Fourier transforms of the coordinate space wave functions. usually given divided by 2π.

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.

25) Within Minkowski space the real. In Minkowski space. No difference is made between upper or lower. (1.3).Introduction 5 that can be used in the cross product of two vectors z = x × y. xi are the three space components.26) (1.22) (1.νn can be conν structed. So we could have used all upper indices in the above equations.2. denoted as V ′µ = Λµν V ν . x3 ) = (t. x1 . Unlike in Euclidean space.3) with basis nµ (µ = ˆ 0. ˆ ˆ (1.23) δjm δkn − δjn δkm . For four-vectors in Minkowski space we will use the notation with upper indices and write x = (t. 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 . δim δjm δkm δin δjn δkn . • ds2 = 0 (lightlike or null intervals). the invariant length or distance (squared) is not positive definite.. the points lie on the lightcone and they can be connected by a light signal. When 3-dimensional space is considered as part of Minkowski space. The transformations that leave invariant the distance ds2 = dt2 − (dx2 + dy 2 + dz 2 ) between two points are called inhomogeneous Lorentz transformations or Poincar´ transformations. x2 = x · x = xµ xν nµ · nν = xµ xν gµν . xµ = gµν xν . x2 .1.. such as eigentime τ for the position vector τ 2 ≡ x2 = t2 − x2 . −x). in which case zi = ǫijk xj yk . and are given by (x0 . Relations relating tensor expressions. x) = (x0 . One has x2 = xµ xµ = t2 − x2 . x3 ). The invariant distance between two points x and y in Minkowski space is determined from the length dsµ = (x − y)µ .21) (1. we will use upper indices for the three-vectors. The real. linear transformations that leave the length of a vector invariant are called (homogeneous) Lorentz transformations. it is convenient to distinguish lower indices from upper indices.µn = Λµ1ν1 . however. linear transformations that do not change the length of a four-vector are called the Lorentz transformations. 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.24) The quantity gµν ≡ nµ · nν is the metric tensor. Λµn n T ν1 . x1 . . Vectors are denoted x = xµ nµ . We note that for Euclidean spaces (with a positive definite metric) vectors and tensors there is only one type of indices.. These transformations do change the components of a vector. Useful relations are ǫijk ǫℓmn ǫijk ǫimn ǫijk ǫijl = = = δiℓ δjℓ δkℓ 2 δkl . In special relativity we start with a four-dimensional real vector space E(1. one must distinguish tensors with upper or lower indices and one can have mixed tensors. Because of the different signs occurring in gµν . The Poincar´ transformations include e e Lorentz transformations and translations. 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 . given by g00 = −g11 = −g22 = −g33 = 1 (the ˆ ˆ other components are zero). The length (squared) of a vector is obtained from the scalar ˆ product. . are . • ds2 < 0 (spacelike intervals). x2 . One can distinguish: • ds2 > 0 (timelike intervals). Many other four vectors and tensors transforming like T ′ µ1 . (1. The (invariant) lengths often have special names. where the coordinate t = x0 is referred to as the time component.. The lower indices are constructed in the following way. independent of a coordinate system.

29) 2 = ∂ µ ∂µ = ∂t2 The value of the antisymmetric tensor ǫµνρσ is determined in the same way as for ǫijk . g0 = g1 = g2 = g3 = 1.28) ν ν Note that gµ with one upper and lower index.36) Exercises Exercise 1. . gµ = gµρ g ρν and 0 1 2 3 is in essence a ’Kronecker delta’. (1. 0123 0123 ′ ′ ǫ µνρσ ǫµνρσ The first identity. ∂3 ) = ∂ ∂ ∂ ∂ .∇ . The derivative ∂µ is defined ∂µ = ∂/∂xµ and we have a four vector ∂ with components (∂0 .33) (1. but more important has the same effect on lefthandside and righthandside.Introduction 6 called covariant. Relate this quantity to the electron Compton wavelength −e via the dimensionless λ fine structure constant α. (1. (1. It is an invariant tensor.32) . not affected by Lorentz transformations. ∂2 .30) (Note that there are different conventions around and sometimes the opposite sign is used).1 (a) In the the Hydrogen atom (quantum system) the scale is set by the Bohr radius. . p) = (E. For the contraction of a vector with the antisymmetric tensor one often uses the shorthand notation ǫABCD = ǫµνρσ Aµ Bν Cρ Dσ . Each of these permutations leads to a minus sign. = ∂ .27) It is easy to convince oneself of the nature of the indices in the above equation. ′ ′ . 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 . Note that in this equation one has on left.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. where m is called the mass of the system. The length squared of ∂ is the d’Alembertian operator. . ∂t ∂x ∂y ∂z ν ∂µ xν = gµ . Examples are the scalar products above but also relations like pµ = m dxµ /dτ for the momentum four vector. p). for instance. starting from ǫ0123 = 1. (1. ∂t (1. a∞ = 4πǫ0 2 /me e2 . = −24. constructed via the metric tensor itself. defined by ∂2 − ∇2 . ∂1 . is timelike with invariant length (squared) p2 = p · p = pµ pµ = E 2 − p2 = m2 . is easily proven for ǫ ǫ from which the general case can be obtained by making permutations of indices on the lefthandside and permutations of rows or columns on the righthandside.31) ǫµνρσ ǫµν ′ ′ ρσ ′ = − (1.34) (1. because one has (1. explicitly written as (p0 . (1.35) ρ ǫµνρσ ǫµν σ σ ǫµνρσ ǫµνρ ′ ′ = −2 g ρρ g σσ − g ρσ g σρ = −6g σσ .

[Note: what is the MKS unit for V/T?] Exercise 1. 0. n1 . aT ]) are defined through √ a± ≡ (a0 ± a3 )/ 2. a2 ] we can find basis vectors n− . ν. 0. ˆ ˆ . g+− and g−+ . (c) Calculate the Compton wavelength of the electron and the quantities under (a) and (b) using the value of c. re = e2 /4πǫ0 me c2 to the Compton wavelength.Introduction 7 (b) Relate the classical radius of the electron (a relativistic concept). ǫ0 . n2 . Note that ˆ ˆ ˆ ˆ a · nα = aα and derive from that the components nα of the four vector nα . ˆ ˆ ˆ ˆ These are n1 = (0. σ} is a permutation of {0. which is the Compton wavelength for the Planck mass.4 Lightcone coordinates for a four vector a (which we will denote with square brackets as [a− . c. (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). n3 . e. ].71 × 10−39 GeV−2 to construct a mass Mpl (Planck mass). How are a+ and a− related to a+ and a− .13 to give its actual value in kg. 0) ˆ n3 = (0. This demonstrates how a careful use of units can save a lot of work. . One does not need to know . (d) Use the value of the gravitational constant GN / c5 = 6. 1. Use that to give the ˆ ˆµ ˆ components of the basis vectors n+ and n− . (a) Express the scalar product a · b in lightcone coordinates and deduce from this the values of g++ . but only appropriate combinations. a+ . 0. a2 . a+ . by a simple few-line reasoning [For instance: If {µ. 0) ˆ n2 = (0. Also for the coordinates [a− . . 0) ˆ n0 = (1. Exercise 1. (b) The coordinates (a0 . α and me c2 = 0. 1.3 Prove the following relation ǫµνρσ g αβ = ǫανρσ g µβ + ǫµαρσ g νβ + ǫµνασ g ρβ + ǫµνρα g σβ . a1 . . 0. a2 ] or [a− . 0. Compare it with the proton mass and use Eq. 1. a1 . Exercise 1. 1) ˆ . 2.3. a1 . a+ . ρ.(A · B) C using the properties of the tensor ǫijk given in section 1. me .2 Prove the identity A × (B × C) = (A · C) B . 1. a3 ) are the expansion coefficients using the basis vectors n0 . n+ . g−− . n1 . Also construct and calculate the Planck length Lpl . n2 . 3} the index α can only be equal to one of the indices in ǫµνρσ . 0.511 MeV.

∂t 2M (2.2. t) = exp(−i k 0 t + i k · r).3 are not covariant. t) = ∂2 − ∇2 + M 2 φ(r. 2M (2.1) under the substitution (in coordinate representation) E −→ Eop = i ∂ . p). p2 = pµ pµ = E 2 − p2 = M 2 .7) k 0 = ± k2 + M 2 = ±Ek .6) The energy spectrum is not bounded from below: considering the above stationary plane wave solutions. (2.1 The Klein-Gordon equation In this chapter. Substitution of operators gives the Klein-Gordon (KG) equation for a real or complex function φ. (2.5) with (k 0 )2 = k2 + M 2 . t).3) Equations 2. φk (r. Thus a covariant equation can be obtained by starting with the (covariant) equation for the invariant length of the four vector (E.1 and 2. we just want to play a bit with covariant equations and study their behavior under Lorentz transformations. But the replacement 2.4) where M is the particle mass. (2. one obtains (2. t) = − ψ(r. namely plane waves characterized by a wave number k. Although it is straightforward to find the solutions of this equation. ∂t p −→ pop = −i∇. E = p2 . 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. 8 .Chapter 2 Relativistic wave equations 2. t) = 0. i ∂ ∇2 ψ(r.2) Acting on the wave function one finds for a free particle. The Schr¨dinger equation in quantum mechanics is the operator equation o corresponding to the non-relativistic expression for the energy. 2 + M 2 φ(r. written as pµ −→ i∂µ is covariant (the same in every frame of reference). ∂t2 (2.

just as the dependence on time was in ordinary quantum mechanics. (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).e. it is simply a matter of being careful to find a consistent (covariant) theory. Then. φ(x) = d3 k (2π)3 2Ek ˜ ˜ e−i k·x φ(Ek .11) Therefore. k) ≡ a(k) and φ(−Ek . however. we want the most general solution. k) + ei k·x φ(−Ek . The integration over k-modes clearly is covariant and restricted to the ‘mass’-shell (as required by Eq.8) (2. having the KG equation as a space-time symmetric (classical) equation. rather than as a wave function. ∂µ j µ = 0. (2.2 Mode expansion of solutions of the KG equation Before quantizing fields. The appropriate current corresponding to the KG equation (see Excercise 2. This continuity equation can be written down o covariantly using the components (ρ. This operator has all possible solutions in it multiplied with creation (and annihilation) operators. As we will see later both problems are related and have to do with the existence of particles and antiparticles.3). for which we need the interpretation of φ itself as an operator. 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. 2. This leads to the existence of negative densities. ↔ ↔ (2.Relativistic wave equations 9 i. there are solutions with negative energy.5). At that point the dependence on position r and time t is just a dependence on numbers/parameters on which the operator depends. j) of the four-current j. j) = ↔ i φ∗ ∂0 φ. −i φ∗ ∇ φ . ∂ρ = −∇ · j. relativistically the density is not a scalar quantity. φ(x) = d4 k ˜ 2π δ(k 2 − M 2 ) e−i k·x φ(k) (2π)4 (2. is a second order equation and φ and ∂φ/∂t can be fixed arbitrarily at a given time. And. (2. but rather the zero component of a four vector. 2. −k) .14) ˜ ˜ Introducing φ(Ek . (2π)3 2Ek (2. For this we note that an arbitrary solution for the field φ can always be written as a superposition of plane waves. The KG equation.15) .9) They satisfy the continuity equation. there are no longer fundamental objections to mix up space and time. 2M 2M (2.2) is j µ = i φ∗ ∂ µ φ or (ρ.13) ˜ with (in principle complex) coefficients φ(k). which is what Lorentz transformations do.12) It is easy to see that this current is conserved if φ (and φ∗ ) satisfy the KG equation. It is possible to rewrite it as an integration over positive energies only but this gives two terms (use the result of exercise 2.10) ∂t which follows directly from the Schr¨dinger equation.

Relativistic wave equations 10 In Eqs 2.17) (2. which implies (xµ ) = (t. Since a · b = a · ˜ it is easy to see that ˜ b. 2.18) (2. x ∂µ ∂ µ + M 2 φ(˜) = 0.e. (2π)3 2Ek (2.23) . It is trivial to see that ∂µ ∂ µ + M 2 φ∗ (x) = 0. The choice of a and b∗ allows an easier distinction between the cases that φ is real (a = b) or complex (a and b are independent amplitudes). x Transforming everywhere in the KG equation x → x one obtains ˜ ˜ ˜ x ∂µ ∂ µ + M 2 φ(˜) = 0. −x) ≡ (˜µ ).6 We will explicitly discuss the example of a discrete symmetry. one thus must have that the function φ → φ′ such that φ′ (x′ ) = φ(x) or φ′ (x) = φ(Λ−1 x). (2π)3 2Ek (2.16) The consequence of this is discussed in Exercise 2. (2.14 and 2. x) → (t. φP (x) ≡ φ(˜) (P for parity). They are referred to as mode and anti-mode amplitudes (or because of their origin positive and negative energy modes).15 one has elimated k 0 and in both equations k · x = Ek t − k · x. Performing some Lorentz transformation x → x′ = Λx. The coefficients a(k) and b∗ (k) are the amplitudes of the two independent solutions (two. changing the sign of the spatial coordinates. 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) . This shows how parity transforms k-modes into −k modes. (2.22) showing that with each solution there is a corresponding charge conjugated solution φC (x) = φ∗ (x). 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) .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) .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. for which we consider space inversion. (2π)3 2Ek (2.19) implying that for each solution φ(x) there exists a corresponding solution with the same energy. i.20) one has for the mode amplitudes aP (k) = a(−k) and bP (k) = b(−k). Another symmetry is found by complex conjugating the KG equation. after restricting the energies to be positive). (2.

Exercises Exercise 2.1 Show that for a conserved current (∂µ j µ = 0) the charge in a finite volume. t) for the mode expansion in Eq. |f ′ (xn )| . QV ≡ satisfies ˙ QV = − ds · j. for the mode amplitudes aC (k) = b(k) and bC (k) = a(k).e. ˙ and thus for any normalized solution the full ‘charge’. (2π)3 2Ek where Ek = Exercise 2. t). letting V → ∞.3 Show that1 d4 k 2π δ(k 2 − M 2 ) θ(k 0 ) F (k 0 .15. S V d3 x j 0 (x). Note that a(k) and b(k) are independent of t. k). is conserved. t) = ei Ek t (i∂0 + Ek ) φ(k. k) = (2π)4 k2 + M 2 . ˜ b(k) = ei Ek t i ∂0 φ∗ (−k. Q = 0. Exercise 2.2 Show that if φ and φ∗ are solutions of the KG equation. For the real field one has aC (k) = a(k). that j µ = i φ∗ ∂ µ φ is a conserved current Note: A ∂µ B ≡ A∂µ B − (∂µ A)B).4 ˜ Write down the 3-dimensional Fourier transform φ(k.Relativistic wave equations 11 i. d3 k F (Ek . t) exp(−i k · x) and its time ˜ derivative i∂0 φ(k. Use these to show that ˜ ˜ a(k) = ei Ek t i ∂0 φ(k. 2. This shows how charge conjugation transforms ’particle’ modes into ’antiparticle’ modes and vice versa. t) ≡ d3 x φ(x. t). 1 Use ↔ ↔ the following property of delta functions δ (f (x)) = zeros xn 1 δ(x − xn ). ↔ ↔ Exercise 2.

15. 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).Relativistic wave equations 12 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). i (a) Show that relativistic invariance. i.1) for a complex scalar field current (exercise 2. One then finds that Q is expressed as an infinite sum over number operators a† a (particles) and b† b (anti-particles). (b) Show that one has current conservation for the current 2 We denote the commutator as [A. t) = (−i α · ∇ + m β) ψ(r. in terms of the a(k) and b(k). and β 2 = 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). Exercise 2. B} = AB + BA. also using the result in (a). 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 . (c) Similarly. d3 x (∂ {0 φ)∗ (∂ i} φ). B] = AB − BA and the anticommutator as {A.5 (a) As an introduction to parts (b) and (c). making sure that each component of ψ satisfies the KleinGordon equation (in terms of differential operators ∂µ ∂ µ +m2 = 0) requires the anticommutation relation2 {αi . ∂t where ψ is a wave function and the nature of α and β is left open for the moment. Exercise 2. {αi . a{µ bν} ≡ aµ bν + aν bµ . ∂ ψ(r. Note that a{µ bν} indicates symmetrization. an attempt to construct a first order differential equation for the time evolution would be to write. From this one concludes that α and β must be matrix-valued and ψ must be a multi-component wave function. 1 2 (b) Express now the full charge QV (t) (exercise 2.e.7 To solve the problem with positive and negative energies and get a positive definite density. express the quantities E(t) = P i (t) = d3 x (∂0 φ)∗ (∂0 φ) + ∇φ∗ · ∇φ + M 2 φ∗ φ .2) in terms of the a(k) and b(k) using the expansion in Eq. β} = 0. t). j 0 = ψ†ψ and j = ψ † αψ. 2. αj } = 2 δ ij I. We will encounter these quantities later as energy and momentum. .

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

y ∈ A ⇒ [x.Groups and their representations 14 (n.6) Summarizing. with real a’s and 3 (ai )2 = 1. • [x. One way of viewing the parameter space. With matrix commutation this algebra satisfies the requirements for a Lie algebra. thus is as the surface i=0 of a sphere in 4 Euclidean dimensions. x]] + [z.7) = a0         0 −1 i 0 1 0 0 1 • [x. Locally this is a 3-dimensional Euclidean space and SU (2).4) R . It is straightforward to check that any (finite) rotation can be obtained from a combination of infinitesimal rotations.    Lx =  Ly =  0 0       0 i 0 −i 0 ϕ ˆ . z ) = lim N →∞ that are generated by  i    0 . is a 3-dimensional Lie-group. y] ∈ A. They satisfy the commutation relations [Li . n) = 1 cos ˆ 1 Note ϕ ϕ ˆ + i (σ · n) sin 2 2 = exp i ϕ ˆ σ·n . ˆ therefore. It is straightforward to check that the conditions require    a0 + i a3 +i a1 + a2    = a0 1 + i a · σ A =   +i a1 − a2 a0 − i a3          1 0          + i a1  0 1  + i a2  0 −i  + i a3  1 0  (3. namely that there exists a bilinear product [.8) ˆ that (σ · a)(σ · b) = a · b + i σ · (a × b). (3. [y. (3. These matrices can be defined as acting on 2-component spinors (χ → Aχ) or equivalently as acting on 2 × 2 matrices (B → ABA† ). Next.z N N or (Lk )ij = −i ǫijk . x]). Writing a0 = cos(ϕ/2) and a = n sin(ϕ/2) we have1 A = A(ϕ.2 π) (-n. π) (n. 2 (3. In the same way we can consider rotations around the x. [x. for rotations around z for instance. which reflects itself in the noncommutation of the generators. z]] + [y.1: the parameter spaces of SO(3) (left) and SU (2) (right). ˆ R(ϕ.   0 (3. y]] = 0 (Jacobi identity). Ly and Lz . and thus (σ · n)2 = 1. we turn to the group SU (2) of special (det A = 1) unitary (A† = A−1 ) 2 × 2 matrices.and y-axes     0 0 0   0 0    0 0 −i  . y] = −[y. . the rotations in SO(3) can be generated from infinitesimal rotations that can be expressed in terms of a basis of three generators Lx . These generators form a threedimensional Lie algebra SO(3). ] that satisfies • ∀ x. x] = 0 (thus [x. π ) π -sphere (antipodes identified) 2 π -sphere (surface identified) Figure 3. Lj ] = i ǫijk Lk .5) Rotations in general do not commute. [z.

immediately sees that the Lie algebras are identical. (3. n) −→ R(ϕ. 3. The infinitesimal generators of SU (2) are ˆ obtained by considering infinitesimal transformations. thus. .12) One. one has a Lie algebra isomorphism that is linear and preserves the bilinear product.9) The parameter-space of SU (2). In the full parameter space. now with radius 2π and with all points at the surface identified (see figure). . . Suppose we have a system described by a Hamiltonian H. 2! . (3. ˆ ˆ A(ϕ.2 Representations of symmetry groups The presence of symmetries simplifies the description of a physical system and is at the heart of physics.15) [g.e. i. They satisfy σi σj σk = i ǫijk . H] = 0 for g ∈ G. There exists a corresponding mapping of the groups given by µ: SU (2) −→ SO(3) ˆ ˆ A(ϕ.11) Thus σx /2. also can be considered as a filled 3-sphere. H] = 0 for g ∈ G. n) ˆ −→ R(2π − ϕ. . n) ≈ 1 + i ϕ J · n. The next issue is to find the appropriate symmetry operators in the Hilbert space. n! 2! n=0 ∞ (3. For instance in a function space translations. SU (2) ≃ SO(3). sometimes in more than one way (but this will be discussed later). since any finite transformation can be constructed from the infinitesimal ones. Near the identity. for fixed n. n) i ∂ϕ = ϕ=0 (3. the above mapping corresponds to the trivial mapping of the Lie algebras. i. it is sufficient that the generators commute with H. n) π ≤ ϕ ≤ 2π.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.10) σ ˆ · n. i.Groups and their representations 15 where we have used the (for operators new!) definition eA ≡ An 1 = 1 + A + A2 + . the SU (2) → SO(3) mapping is a 2 : 1 mapping where both A = ±1 are mapped into R = I. 2 2 2 (3.14) For a Lie group. [g. thus. in particular that AB → RAB = RA RB ). σy /2 and σz /2 form the basis of the Lie-algebra SU (2). however. The existence of symmetries means that there are operators g belonging to a symmetry group G that commute with the Hamiltonian. . φ′ (r) = φ(r + a) = φ(r) + a · ∇ φ + 1 (a · ∇)2 φ + . with ˆ J ·n ≡ ˆ 1 ∂A(ϕ.e. . (3. 2 (3. .e.

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

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

it is convenient to use a vector notation for the points in Minkowski space. which as a matrix equation with (Λ)µν = Λµν and (G)µν = gµν gives (ΛT )ρµ (G)µν (Λ)νσ = (G)ρσ . det Λ = −1 is called improper). e To derive some of the properties of the Lorentz group. explicitly x′µ = (Λ)µν xν + aµ or x′µ = Λµν xν + aµ . Using (i) and (ii) the Lorentz transformations can be divided into 4 classes (with disconnected parameter spaces) det Λ +1 +1 -1 -1 Λ00 ≥1 ≤ -1 ≥1 ≤ -1 (3. 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 .24) L↑ + L↓ + L↑ − L↓ − proper orthochrone proper non-orthochrone improper orthochrone improper non-orthochrone .22) (Note that xT is a row vector).g. (ii) |Λ00 | ≥ 1.21)  g  20 g21 g22 g23   0 0 −1 0             g30 g31 g32 g33 0 0 0 −1 the scalar product can be written as x2 = xT G x (3.25) (3. proof: (ΛT GΛ)00 = (G)00 = 1 → Λµ0 gµν Λν0 = 1 → (Λ00 )2 − i (Λi 0 )2 = 1.23) The proper tensor structure of the matrix element (Λ)µν is a tensor with one upper and one lower index. it is easy to derive some properties of the matrices Λ: (i) det(Λ) = ±1. Denoted in terms of column vectors and 4-dimensional matrices one writes for the poincar´ transe formations x′ = Λ x + a.26) (3. proof: det(ΛT GΛ) = det(G) → (det Λ)2 = 1. (3.27) (3. consisting of the Lorentz group and translations. In this section we consider the Poincar´ group. (det Λ = +1 is called proper.     0 0   g00 g01 g02 g03   1 0      g     g11 g12 g13   0 −1 0 0  = .Groups and their representations 18 3. e. Writing x as a column vector and the metric tensor in matrix form.     G =  10 (3. σ/2 in a two-dimensional (spin 1/2) case. From this property. We considered the generators and looked for representations in finite dimensional spaces. Thus one has the (pseudo-orthogonality) relation ΛT GΛ = G ⇔ GΛT G = Λ−1 ⇔ ΛGΛT = G.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.

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

We will just require Poincar´ invariance for the generators themselves. three of which only involve spatial coordinates (rotations) and three others involving time components (boosts). Also this group can be divided into four parts.35) thus (ω)ij = −(ω)ji and (ω)0i = (ω)i0 .40) ΛB = I + i η 3 K 3 = I − i ω30 M 30 . J 3 = M 12 ) and for the (infinitesimal) boosts (along (3. ǫ).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) (3.38) Explicitly. we continue using covariance arguments. (3. In order to find the commutation relations including the translation generators.36) (3. the Lorentz transformations form a (Lie) group (Λ1 Λ2 and Λ−1 are again Lorentz transformations. Thus the two vector operators J and K form under Lorentz transformations an antisymmetric tensor M µν . One can easily convince oneself that i (ω)µν = ω µν = − ωαβ (M αβ )µν . we have for the (infinitesimal) rotations (around z-axis) ΛR = I + i ϕ3 J 3 = I − i ω12 M 12 . which are obtained by writing the infinitesimal parameters (ω)µν in terms of six antisymmetric matrices (M αβ )µν . infinitesimally approximated by (I + ω. (3.37) The algebra of the generators of the Lorentz transformations can be obtained by an explicit calculation2 . There are six generators. Multiplying the proper orthochrone transformations with one of them gives all Lorentz transformations. M ρσ ] = −i (g µρ M νσ − g νρ M µσ ) − i (g µσ M ρν − g νσ M ρµ ) .33) These four transformations form the Vierer group (group of Klein). 3 30 with η = ω and K = M (e. This is a + normal subgroup and the factor group is the Vierer group. i i0 i i0 1 ijk M jk 2ǫ (3. with ϕi = − 1 ǫijk ω jk and J i = 2 z-axis). explicitly reading x′µ = = (Λ)µν xν + aµ = ((I)µν + (ω)µν ) xν + ǫµ Λµν xν + aµ = (g µν + ω µν ) xν + ǫµ .34) The condition ΛT GΛ = G yields g ρµ + ω ρµ gρσ (g σν + ω σν ) = gµν =⇒ ωνµ + ωµν = 0. Of the four parts only L↑ forms a group. We now want to find a covariant form of the six generators of the Lorentz transformations.g.g. Summarizing. K = M ). 3.4 The generators of the Poincar´ group e ↑ The transformations belonging to P+ are denoted as (Λ. The extension to the Poincar´ group is e ↑ straightforward. a).39) (e. 2 (M αβ )µν = i g αµ g βν − g α g βµ ν (3. We therefore find (again) that there are six generators for the Lorentz group. inf (3. P+ etc. The e 2 These commutation relations are the same as those for the ’quantummechanics’ operators i(xµ ∂ ν − xν ∂ µ ) . [M µν .

writing the generator P = (H/c. These define states with specified quantum numbers that are eigenvalues of these generators. [K i . a) = Λ−1 µ P ν . because one then sees that by letting c → ∞ the commutation relations of the Galilei group. . K j ] = i ǫijk K k . (3. a)P µ U (Λ. H] = [J i .46) [K i . [P µ . as part of the set. . (3. = −i (g µρ P ν − g νρ P µ ) . 2 2 (3. . a)P µ U † (Λ.43) giving the following commutation relations by equating the coefficients of ǫµ and ωµν . P ν ] [M µν .45) Note that the commutation relations among the generators M µν in Eq. Infinitesimally. The requirement that P µ transforms as a four-vector (for which we know the explicit behavior from the defining four-dimensional representation). J j ] = i ǫijk J k . [J i . Explicitly. P µ |p. a) = Λµν P ν or U † (Λ. To find other generators that commute with P µ we look for Lorentz transformations that leave the four vector pµ invariant. For instance. In order to label the states in an irreducible representation we construct a maximal set of commuting operators.38 could have been obtained in the same way. ν (3. H] = i P i . P j ] = i δ ij H/c2 . check that U1 U2 corresponds with Λ1 Λ2 ). [K i .47) (3. (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 ) in terms of the Hamiltonian and the three-momentum operators.44) (3. known from non-relativistic quantum mechanics are obtained. P j ] = i ǫijk P k . other states in that representation are obtained by the action of operators outside the maximal commuting set. leads to U (Λ. We have here reinstated c.42) (to decide on where the inverse is. 3. that commute e among themselves. . The eigenvalues of these will be the four-momentum pµ of the state.48) .5 Representations of the Poincar´ group e where pµ is a set of four arbitrary real numbers. 3.41) 2 At this point. the tensor M µν in terms of boosts cK i = M i0 and rotations J i = 1 ǫijk M jk . [J i . one obtains 2 [P i .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. Of the generators of the Poincar´ group we choose first of all the generators P µ . . Taking any one of the states in an irreducible representation. P j ] = [P i . For instance. Λµν pν ⇒ ωµν pν ⇒ ωµν = = = pµ + ω µν pν = pµ 0 ǫµνρσ pρ sσ . (3. = pµ |p. the generators J 2 and J 3 in the case of the rotation group. P ρ ] = 0. . [J i . K j ] = −i ǫijk J k /c2 . ǫ) = 1 − ωαβ M αβ + i ǫα P α . They just state that M µν transforms as a tensor with two Lorentz indices. the generators J± in the case of the rotation group. the generators M αβ no longer exclusively act in Minkowski space. . These form a group called the little group associated with that four vector. (3. The generators of the translations are the (momentum) operators P µ . H] = 0.

These e operators are the Casimir operators of the algebra and can be used to define the representations of ↑ P+ . = = i exp − ǫµνρσ M µν pρ sσ |p. 0. Wν ] = −i (gµα Wν − gνα Wµ ) . . .50) where the Pauli-Lubanski operators W µ are given by 1 Wµ = − ǫµνρσ M νρ P σ . M ǫijk .52) (3. .53) (3. satisfying gµν nµ (p)nν (p) = j i nµ (p) pµ i ǫµνρσ pµ nν nρ nσ i j k = = −δij . (3. while the others have no physical significance. The elements of the little group thus are U (Λ(p)) = = and U (Λ(p))|p. . 2 exp (−isµ W µ ) |p.55) commute with all generators and therefore are invariants under Poincar´ transformations. . p0 > 0.54) This will enable us to pick a suitable commuting ’spin’ operator. 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. 0. From the algebra of the generators one finds that P 2 = Pµ P µ and W 2 = Wµ W µ (3. .49) (3. Given the momentum four vector pµ we choose a tetrade consisting of three orthogonal spacelike unit vectors ni (p). i exp − ωµν M µν 2 i exp − ǫµνρσ M µν pρ sσ 2 (3. of which the components generate the little group of pµ or (for the third of the following relations) from explicit calculation [Mµν .51) The following properties follow from the fact that W µ is a four-vector by construction. Pν ] = [Wµ . which can thus be chosen spacelike. 2 (3. the third case represents the vacuum. Wα ] = [Wµ .56) . . . .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 . Massive particles: p2 = M 2 > 0. (3.

Having made a covariant decomposition.63) where E = p2 + M 2 and ms is the z-component of the spin −s ≤ ms ≤ +s (in steps of one).60) and using the completeness relation nµ (p)nν (p) = − g µν − i i 2 i pµ pν M2 .Groups and their representations 23 We can write W µ (p) = 3 W i (p) nµ (p) i i=1 (3. The commutation relations [W i (p). (0. −1) n1 (p0 ) ≡ n2 (p0 ) ≡ s(p0 ) ≡ (3. ms . M2 (3. Together with the four momentum states thus can be labeled as |M. (0. M2 (3. 0). 1. 2 . 1 2 (3.62) 1 Thus W 2 has the eigenvalues −M 2 s(s + 1) with s = 0. More generally (fully covariantly) one can proceed by defining S i (p) ≡ W i (p)/M and obtain [S i (p).65) . i (3. it is sufficient to choose a particular frame to investigate the coefficients in Eq. Noting that S i (p) i 2 = 1 W µ (p)W ν (p) M2 niµ (p)niν (p). show that W /M can be interpreted as the intrinsic spin.e. 0). p.58) i. the S i (p) form the generators of an SU (2) subgroup that belongs to the maximal set of commuting operators. 3.. s. S j (p)] = −i nµ (p)nν (p) j i ǫµνρσ W ρ P σ = i ǫijk S k (p). (3. 0).57) (i.59) (3. In that case the vectors ni (p) are just the space directions and W = (0. In this case a set of four independent vectors is chosen starting with a reference frame in which pµ = pµ 0 in the following way: p0 = (p0 . |p3 |). (3.e. 0. n2 (p) and s(p) are obtained by a Lorentz boost from the reference frame one has the property ǫµνρσ sµ nν nρ pσ > 0.57 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. 0.64) such that in an arbitrary frame where the four vectors p. M J) with components W i = (M/2)ǫijk M jk . 1. The explicit construction of the wave function can be done using the D-functions (analogous as the rotation functions). 1. p0 > 0. (1. 0. n1 (p). 0. . 0. W i (p) = −W · ni (p)). . 0. W j (p)] = i M ǫijk W k (p). 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. .61) one sees that S i (p) i =− W2 .

thus.72) One has P µ |0 = W µ |0 = 0 and the state is invariant under Lorentz transformations. Exercises Exercise 3. which can take any integer or half-integer value. A massless particle.66) where W · p = 0 implies that W s (p) = −(M 2 /p · s) λ(p).71) which is called the helicity.1 (a) Derive from the relation that defines the mapping of SU (2) matrices A into SO(3) matrices RA in Eq. The vector W µ (p) now can be expanded W µ (p) = W 1 (p) nµ (p) + W 2 (p) nµ (p) + λ(p) pµ + W s (p) sµ (p). 1 2 24 (3.73) (3. The meaning of λ(p) is most easily seen by comparing for a momentum eigenstate W 0 (p) = λ(p) |p| with W 0 (p) = − 1 ǫ0ijk Mij pk = J · p. W 2 (p)] = = = i M 2 λ(p).69) We reproduce the algebra for the massive case with W 3 (p) = M λ(p). and gives [W 1 (p). Wν ]. But if M = 0 one has a different algebra (Eq. a)|0 = |0 . is characterized by its momentum and the helicity. 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). implying continuous values for the spin (actually for W i = M S i ). W 2 (3. This algebra is (for instance) isomorphic to the one generated by rotations and translations in a 2-dimensional Euclidean plane (see exercises). U (Λ. W 1 (p)] [λ(p).67 has a vanishing right hand side). The algebra of W 1 . however. σ · RA a ≡ A(σ · a)A−1 ˆ (valid for any vector a). W 2 (p)] [λ(p). 3. i W 2 (p). λ .13. W µ (p) = λ(p) P µ . (3. where A = exp(i ϕ · σ/2) = exp(iϕ σ · n/2).68) (3. Thus 2 λ(p) = J ·p . The vacuum: pµ = 0. −i W 1 (p). This state is in physical applications nondegenerate and is denoted by |0 . The eigenvalues of W 1 and W 2 (in that case corresponding to the translations) thus can take any continuous values. |p. |p| (3.70) as a commuting set with zero eigenvalues for P 2 and W 2 . Thus we have P 2 .Groups and their representations The above does actually include the massive case. Note that angular momentum does not contribute to helicity as L · p = 0. the action of RA b · RA a = = ˆ ˆ ˆ b · a cos ϕ + n · (b × a) sin ϕ + 2(b · n)(a · n) sin2 (ϕ/2) ˆ ˆ b · a + b · (ˆ × n) sin ϕ − 2 (b · a − (b · n)(a · n)) sin2 (ϕ/2) a . We don’t know of any physical relevance. W 2 and λ is derived from the general commutation relations for [Wµ . (3.67) (3. P µ . 3.

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

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

6) (4.10) . respectively. 0. It is simply connected and forms the covering group of L↑ . 2 (4. (4. K = +i . With this choice of parameter-spaces the plus and minus signs are actually relevant. C) is the group of complex 2 × 2 matrices with determinant one. j ′ ). The group SL(2. where we restrict (for fixed n) the parameters 0 ≤ ϕ ≤ 2π and 0 ≤ η < ∞.2) This shows that the Lie algebra of the Lorentz group is identical to that of SU (2) ⊗ SU (2). i ǫijk B k .4) (4. They will be labeled by two angular momenta corresponding to the A and B groups.5) = 1 (J + i K).9) (4. similarly as the homeomorphism between SO(3) and SU (2). 2 2 σ σ J = .1) (4. (4. U (ϕ) H(η) (4. 2 1 (J − i K). 0) Type II : (0. 2 2 (4.Chapter 4 The Dirac equation 4. 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 .8) ˆ ˆ ˆ with η = η n and ϕ = ϕn. (j. K = i J (A = 0). it also follows directly that the Lorentz group is homeomorphic with the group SL(2. C). M = exp i ϕ·σ 2 1 exp ± η · σ 2 = U (ϕ) H(η) . They precisely correspond to the two types of representations that we have seen before. B j ] = [Ai . K = −i .1 The Lorentz group and SL(2. B j ] = i ǫijk Ak . Aj ] = [B i . The matrices in SL(2. becoming the defining representations of SL(2.3) (4. C) can be written as a product of a unitary + matrix U and a hermitean matrix H. Special cases are the following representations: Type I : (j.7) From the considerations above. j) K = −i J (B = 0). C): Type I (denoted M ): Type II (denoted M ): 27 J= σ σ . This tells us how to find the representations of the group.

η → U η.13) 2M (E + M ) (exercise 4. are not covariant!) ˜ can be used to establish the homeomorphism between L↑ and SL(2. ˜ (4. −σ). one can use a boost to the frame with momentum ˆ ˆ p. ǫ−1 = ǫ† = ǫT = −ǫ). σ). ˜ Non-equivalence of M and M One might wonder if type I and type II representations are not equivalent. ˜ = σ µ Λ B aµ .15) (4. 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). such that M = SM S −1 . (ǫ∗ = ǫ.18 show that ±ǫξ ∗ transforms as a type-II (η) spinor. however. there exists the matrix ǫ = iσ 2 . cannot be connected by a Lorentz transformation belonging to L↑ . This boost then is given by H(p) = exp η·σ 2 = cosh η η ˆ + σ · n sinh 2 2 = M +E+σ·p (4. H(η) = exp(−η · σ/2). This shows that M ∗ ≃ M . i. The following relations for + rotations and boosts are useful. ¯ ¯ H ∗ = ǫ† Hǫ = ǫ−1 Hǫ. One has + −1 Λ(M ∗ ) = I2 Λ(M ) I2 .19) . η → Hη. σ µ ≡ (1. (4. The Lorentz transformations Λ(M ) and Λ(M ). C). This is not true for the two-component spinors in SL(2. One defines spinors ξ and η transforming similarly under unitary rotations (U † = U −1 . C) or L↑ . With E = M γ = M cosh(η) and p = M βγ n = M sinh(η) n. while ±ǫη ∗ transforms as a type-I + spinor. 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. Note that M ∗ = ǫ† M ǫ has its equivalent for the Lorentz transformations Λ(M ) and Λ(M ∗ ). namely H(η) = exp(η · σ/2). U σ µ aµ U −1 = σ µ ΛR aµ .12) ξ → U ξ. C) and. Also useful is the relation H 2 (p) = σ µ pµ /M = (E + σ · p)/M .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. U ≡ (U † )−1 = U ) U (ϕ) = exp(i ϕ · σ/2) but differently under hermitean boosts (H † = H. ξ −→ Hξ.18) As already discussed for SU (2). U σ µ aµ U ˜ −1 H σ µ aµ H −1 = σ µ ΛB aµ H σ µ aµ H ˜ −1 (4. if there does not exist a unitary matrix S.16) = σ µ Λ R aµ . Λ(ǫ) ∈ L+ . thus. (4. C). thus. ˜ The sets of four operators defined by definitions σ µ ≡ (1. H ≡ (H † )−1 = H −1 ). (4.1). Eqs 4. In fact. (4. (4.17) 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.14) satisfying Tr(σ µ σ ν ) = −2 gµν and Tr(σ µ σ ν ) = 2gµν = 2 δ µν (the matrices.e.

0). Although these operators cannot be used to label the representations.The Dirac equation 29 where The matrix I2 does not belong to L↑ . 1 ) and ( 1 .20) 4. a) and ˜ a = (a0 .       = (4. 0) are inequivalent.1. but to L↑ . We also have seen that the representations (0. Within the Lorentz group. (scalar). ǫµνρσ will change sign. must combine the repe resentations. (pseudoscalar). For a particle at rest only angular momentum is important and we can choose ξ(0. −a). 4.22) u=     η η M (Λ) 0 where M (Λ) = ǫM ∗ (Λ)ǫ−1 with ǫ given in Eq. therefore.g.21) In nature parity turns (often) out to be a good quantum number for elementary particle states. e. 2 2 (4. From the definition of the representations (0. the boost in Eq. This provides the easiest way of seeing ˜ what is happening. the aij elements of a tensor will not change sign. we obtain for the two components of u which we will refer to as chiral right and chiral left components. we have seen that in the rest frame W i (p)/M → J i . C) are suitable for representing spin 1/2 particles. that have the correct commutation relations. i.e. m) = χm . For the spin 1/2 representations of the Poincar´ group including parity we. and the angular momentum in the rest frame is what we are familiar with as the spin of a particle. m) = η(0. 0) (via operators J and K) one sees that under parity 2 2 1 1 (0. Thus M ∗ and M are not equivalent within SL(2. as we have seen in section 4. (axial vector).      (4.13. Examples (with between brackets the Euclidean assigments) are r t p H J λ(p) K −→ −λ(p) −→ −K −→ H −→ J −→ t (scalar).   −→  M (Λ)   (4. The behavior is the same for classical quantities. m) =            χm uL H(p) 0 . consider the four component spinor that transforms under a Lorentz transformation as      0  ξ   ξ  . they can be connected. Taking M (Λ) = H(p). The representing member of the class P− is the parity or space inversion operator Is .17. where a = (a0 . 0) of SL(2.2 Spin 1/2 representations of the Lorentz group 1 1 Both representations (0. 2 ) and ( 2 . + −   1 0    0 1  I2 =    0 0   0 0 0 0 −1 0 0 0 0 1      . −→ −r (vector).      0   χm   uR   H(p)      . etc.23) u(p. 1 ) and ( 1 .e. The i angular momentum operators J are represented by σ i /2. ) −→ ( . generators. i. they cannot be connected by a unitary 2 2 transformation. (vector). C). −→ −p (vector). the well-known two-component spinor for a spin 1/2 particle. It has the same effect as lowering the indices. 4. but by a transformation belonging ↑ ↑ to the class P− . but rather the operators W i (p) should be used. A parity transformation changes aµ into aµ .

(4.   (4. ψ † −iγ 0 ∂0 +iγ i ∂i −M 1 We ← ← = 0.30) γ µ γ ν ∂µ ∂ν + M 2 ψ(x) = 0. γ ν } = 2 g µν . It is of a form that we also played with in Exercise 2. The general requirements for the γ matrices are thus easily obtained.33) define for a four vector a the contraction a = aµ γµ / .28) which is a covariant (linear) first order differential equation.25) It is straightforward to eliminate χm and obtain the following constraint on the components of u. From this one obtains the Clifford algebra for the Dirac matrices. (pµ γ µ − M ) u(p) ≡ (/ − M ) u(p) = 0.32) suppressing on the RHS the identity matrix in Dirac space. γ } ∂µ ∂ν + M 2 ψ(x) = 0. (4. one must require that for ψ(x) the Klein-Gordon relation 2 + M 2 ψ(x) = 0 is valid for each component separately). 4. (4. p (4. This leads to the Dirac equation for ψ(x) = u(p)e−i p·x in coordinate space.The Dirac equation 30 with H(p) = ¯ H(p) = 2M (E + M ) E+M −σ·p 2M (E + M ) E+M +σ·p .24) (4. Since ∂µ ∂ν is symmetric. 2 (4. To see this consider the equation for the hermitean † † conjugate spinor ψ † (noting that γ0 = γ0 and γi = −γi ).7. As in section 2 we can use Pµ = i∂µ as a representation for the momenta (translation operators) in function space. We will discuss another explicit realization of this algebra in the next section.27) where γ µ are 4 × 4 matrices called the Dirac matrices1 .      0 H 2 (p)   uR   uR        . (iγ µ ∂µ − M ) ψ(x) = 0. which in general is a linear equation in pµ .  =       (4. We first want to investigate if the Dirac equation solves the problem with the negative densities and negative energies. Applying (iγ µ ∂µ + M ) from the left gives (4.29) which is an explicit realization of the (momentum space) Dirac equation. {γ µ . The first indeed is solved.31) To achieve also that the energy-momentum relation p2 = M 2 is satisfied for u(p).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. .27 is known as the Weyl representation. this can be rewritten 1 µ ν {γ . = H −1 (p). The explicit realization appearing in Eq.

Relying on the Pauli exclusion principle for spin 1/2 particles.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  .3 General representations of γ matrices and Dirac spinors {γ µ . This can most easily be seen in the particle rest frame. . the one of the negative energy states.37) Using the explicit form of γ in the Weyl representation (see Eq. remains. Multiplying with γ 0 on the right and pulling it through one finds a covariant equation ψ(x) iγ µ ∂µ +M ← = 0.39) use ρi ⊗ σj with both ρ and σ being the standard 2 × 2 Pauli matrices. Dirac supposed that the negative energy states were already completely filled and the exclusion principle prevents any more particles to enter the sea of negative energy states.The Dirac equation 31 E empty levels M -M filled levels E particle M -M hole Figure 4.34) for the adjoint spinor ψ ≡ ψ † γ0 . (4. We will see how this is properly implemented (also for bosons!) when quantizing fields. (4.27). one sees that there are two positive and two negative eigenvalues. 4.1: The Dirac sea of negative energy states and antiparticles. From the vacuum a particle can be removed. From the equations for ψ and ψ one immediately sees that j µ = ψγ µ ψ is a conserved probability current (exercise!). (4.36) (4. opposite charge). The second problem encountered before. E = +M (twice) and E = −M (twice). 4. but with properties such that it can be annihilated by the particle (e. The density j 0 = ψγ 0 ψ = ψ † ψ (4.  . where γ 0 p0 ψ = M ψ 0 −→ Eψ = M γ 0 ψ.g.35) is indeed positive and j 0 can serve as a probability density. This hole forms an antiparticle with the same mass. This problem was solved by Dirac through the introduction of a negative energy sea. The Dirac sea forms the vacuum. γ =ρ ⊗1= γ = iρ ⊗ σ =    0 −1 −σ k 0 2 We (4. γ ν } = 2 g µν .

while L−1 γ µ L = Λµ γ ν . 2 (4.  (γ5 )S.41) (4. −→ Sψ.10 are   1  σ3 0  i   J 3 = M 12 =   = [γ 1 . (4.R. γ 2 ].   0 γ k = −iρ2 ⊗ σ k =  k  σ  −σ k  .49) i 1 with −i S ρσ = 2 σ ρσ = − 4 [γ ρ . For example. which in the case of the Weyl representation are just the upper and lower components. 4! (4. it is easy to see that PR/L = 1 (1 ± γ5 ) 2    1 0  .46) (4. γ 0 = ρ1 ⊗ 1 =   1 0 γµ ψ Different representations can be related to each other by unitary transformations.R. the rotation and boost generators in Weyl representation in Eqs 4. 1 −1 2 For all representations one has † γµ = γ0 γµ γ0 . ν The latter assures Lorentz invariance of the Dirac equation (see Exercise 4.40) (4. γ σ ]. −→ Sγµ S −1 .47) (4. γ 0 ].R. γ µ } = 0 and explicitly one has    0 1  1 . 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 .43) S=√  . 0 σ3 2 4   1  −iσ 3 0   = − i [γ 3 . = ρ ⊗ 1 =   1 0 i ǫµνρσ µ ν ρ σ γ γ γ γ .42) We note that the explicit matrix taking us from the Weyl representation to the standard representation.The Dirac equation 32 The Weyl (or chiral) representation:    0 1   .45) (γ5 )W. Lorentz invariance The Lorentz transformations can also be written in terms of Dirac matrices.  =ρ ⊗1=  0 −1 3 (4. We note that ψ → Lψ and ψ → ψ L−1 .9 and 4.48) are projection operators. that project out chiral right/left states. K 3 = M 30 =    0 iσ 3 2 4 or in general one has the transformation i ψ → Lψ = exp − ωρσ S ρσ ψ. S −1 . (γµ )S. For instance in the Weyl representation (but valid generally). = S(γµ )W.  0 (4.44) and an adjoint spinor defined by ψ = ψ † γ0 .R.3) . is   1  1 1    (4.

52)    η ξ one can apply space inversion. As with parity we look for a transformation.57) (4.53) (4.   ψ= (4. such that T Cγµ C −1 = −γµ . Note that we have (as expected) explicitly in Weyl x representation in Dirac space     P  ξ    −→ ψ p = γ 0 ψ =  η  . called charge conjugation. e. One has ψ c = −ψ T C −1 . usually to be taken unity. which is again a solution of the Dirac equation. In any representation a matrix C exist. The latter implies that the conjugate of a right-handed spinor. x = (t.56) (i/ − M ) ψ c (x) = 0.59)  −→ ψ c = C ψ =    η ǫ ξ∗ . 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.g. for instance. ψR . Some properties of C are C −1 = C † and C T = −C. =    0 ǫ iσ 2 (4.R. It is easy to convince oneself that if ψ(x) is a solution of the Dirac equation. ∂ (4. Explicitly. but it does.50) Charge conjugation but with ψ p (x) ≡ ηp γ0 ψ(˜) (Exercise 4. γ5 ] = 0.58) where ηc is an arbitrary (unobservable) phase.R.R. C is real and one has C −1 = C † = C T = −C c and [C. in Weyl representation we find in Dirac space     T C  ξ   −ǫ η ∗     .    0 −ǫ 0    0   −ǫ 0  . −x) and via a few manipulations obtain again the ˜ Dirac equation (i/ − M ) ψ p (x) = 0. T (4. ψ= (4.g.55) (4.) and all representations connected via a real (up to an overall phase) matrix S. x) → x = (t.51) The existence of positive and negative energy solutions implies another symmetry in the Dirac equation. This symmetry does not change the spin 1/2 nature. parity.6). (C)S. (or W.54) e. is a left-handed spinor. that brings ψ → ψ c . ∂ (4. (i/ − M ) ψ(x) = 0. In S. reverse the charge of the particle.R.    −iσ 2   0 −ǫ  = . (C)W. ∂ with the solution ψ c (x) = ηc Cψ (x). = =   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. T (4.The Dirac equation 33 Parity There are a number of symmetries in the Dirac equation.

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 ) = v † (p.64) (4.65) (4.The Dirac equation 34 4. 2M 2M −/ + M p v(p. ¯ ′ ′ Explicit solutions in the standard representation are   χs       . s) = u(p. Note that the spinors in the negative energy modes (antiparticles) could be two different spinors.69) = − s In order to project out spin states. s)¯(p.70) =  Ps =  ˆ 0 1∓σ·s 2 2 . s′ ) = 2Ep δss′ . s ) = 0.68) (4. p     −σ · p (Ep − M )     Ep + M     σ ·p ¯ Ep +M χs χs ¯ For a free massive particle. ¯ (4. the spinors satisfy C uT (p. where u satisfies   −σ · p  Ep − M      u(p) = 0. s)u(p. be obtained by a Lorentz transformation. s) v = .    (4. s). ψ(x) = v ±s (p) ei p·x . s)¯(p. s) and C v T (p. s) = v(p.63) u(p. 2M 2M (4. v (p.60)   ψ(x) = 0. s)v(p. The explicit Dirac equation in the standard representation reads  ∂  iσ · ∇   i ∂t − M       (4. the best representation to describe particles at rest is the standard representation. In that frame is a spacelike unit vector sµ = (0. s ) = v (p.66) An arbitrary spin 1/2 field can be expanded in the independent solutions. s) . v(p. Choosing χ = −ǫχ∗ (the equivalent spin ¯ 1/2 conjugate representation).62) where χs are two independent (s = ±) two-component spinors. p (4. ¯ ¯ † ′ u (p. s) = u(p. s) = Ep + M      σ ·p   Ep +M χs     . with E = Ep = + p2 + M 2 . s) ei k·x d∗ (k.. s). It is easy to check that   1  1±σ·s 1 ± γ5 / s ˆ 0    (4. s)v(p. (2π)3 2Ek (4. s) + v(k. s) e−i k·x b(k. s)v(p.g. s′ ) = 2M δss′ .61)     σ·p −(Ep + M ) There are also two negative energy solutions. ψ(x) = u±s (p) e−i p·x .   ⇔ (/ − M ) u(p) = 0.4 Plane wave solutions Looking for positive energy solutions ∝ exp(−iEt) one finds two solutions. s)u(p. in which γ 0 is diagonal (see discussion of negative energy states in section 4. The ¯ ¯ solutions are normalized to u(p. s′ ) = −2M δss′ .1). In an arbitrary frame one has s · p = 0 and s(p) can e. s) u = . the spin polarization vector in the rest frame is the starting ˆ point. where v satisfies   σ·p  −(Ep + M )      v(p) = 0.   ⇔ (/ + M ) v(p) = 0.     ∂ −iσ · ∇ −i ∂t − M (4.67) It is straightforward to find projection operators for the positive and negative energy states P+ P− = s p /+M u(p. s)u(p.

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

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

4. s)¯(p.g. 0 if n is odd. −x) is also a solution of the Dirac equation. They lead to Lµν 1 Tr [(/ ′ + m)γ µ (/ + m)γ ν ] k k 2 = 2 k µ k ′ν + k ν k ′µ − g µν (k · k ′ − m2 ) = = 2 k µ k ′ν + 2 k ν k ′µ + (k − k ′ )2 g µν . ˜ Exercise 4. γ σ ] = σ ρσ .3 Imposing Lorentz invariance on the Dirac equation can be achieved by requiring L−1 γ µ L = Λµ γ ν .5 Show starting from the relation {γµ . s). e. i S ρσ ] = g ρµ γ σ − g σµ γ ρ . H(p) = exp η·σ 2 = M +E+σ·p . Exercise 4. by proving that H 2 (p) = exp (η · σ) = (E + σ · p)/M = σ µ pµ /M. (4.77) Exercises Excercise 4.g. 1 i −i S ρσ = − [γ ρ . ν µ µ Look at the infinitesimal forms Λµ = gν + ων and L = 1 − (i/2)ωρσ S ρσ to obtain ν [γ µ . γν } = 2gµν that γµ / γ µ a = Tr(//) = ab µn µ1 Tr(γ · · · γ ) = −2 /. 4 2 Exercise 4. x → x.6 Apply space-inversion. x ˜ . by letting them act on the four basis spinors u(p. Excercise 4. 2M (E + M ) ˆ ˆ where E = M γ = M cosh(η) and p = M βγ n = M sinh(η) n. Do this e. to the Dirac equation and use this to show that the spinor ψ p (x) = ˜ γ0 ψ(˜).13.4 Show that in P+ ≡ s=±1/2 u(p. s) u p /+M = 2M 2M both sides are projection operators.The Dirac equation 37 The trace is linear and can be split up in parts containing traces with up to four gamma-matrices. a 4 a · b. of which only the traces of four and two gamma-matrices are nonzero.1 Prove Eq. s) and v(p.2 Show that j µ = ψγ µ ψ is a conserved current if ψ en ψ are solutions of the Dirac equation Exercise 4. Show that the equality holds. where x = (t.

Show that this n condition is sufficient to guarantee that nµ j µ = 0. . We will not pursue this chiral freedom at this point and take α = 0. M R = 1.8 A negatively charged pion (π − ) decays in a lepton ℓ (ℓ can be an electron or a muon) and a corresponding anti-neutrino (¯ℓ ). we want to confine this solution to a spherical cavity with radius R.7 (a) Show that (in standard representation)  j0 (kr)χm    ψ(x. Note: the outcome shows that the decay into muons is much more likely that the decay into electrons (mπ = 139. 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. For such a cavity nµ = (0. me = 0. n (c) Apply the confinement condition to find the value of k for the lowest eigenmode in the spherical ˆ cavity. (d) Plot for these cases the vector densities ψ † ψ and the scalar densities ψψ as a function of r/R. The conditions for ψ and ψ imply that nµ j µ = ψ/ ψ = 0. i. The pion has spin 0. r). Note: use first the appropriate equation (which?) that is obeyed by each individual component of the Dirac equation to show that the upper two components are allowed solutions. which is the most important factor needed to find the decay width (inverse lifetime). (b) Next.66 MeV. Show that the ratio is |M (π − → e− νe )|2 ¯ ≈ 54.511 MeV). you need to determine the corresponding confinement condition for ψ. t) = N     k ˆ i E+M σ · r j1 (kr)χm    −i Et   e . mµ = 105.8 × 10−6 .57 MeV. 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 ). n It turns out that there is actually a whole family of conditions that provide confinement. the lepton and neutrino have spin 1/2 and are ν described by momentum-space spinors.The Dirac equation 38 Exercise 4. Note: To prove this.e. A condition for confining the fermion in the cavity is i / ψ = ψ (where n is the surface normal).    is a stationary solution of the (free) Dirac equation. namely i / ψ = eiαγ5 ψ. Calculate the value of k for M R = 0. there is no current flowing through the surface of the cavity. Calculate for now just |M |2 . ν p Note: how such parametrizations work will be explained in chapter 8.5 and M → ∞. Do you understand why? The calculation of the life-time is completed in chapter 11. |M (π − → µ− νµ )|2 ¯ Actually if the lepton would be massless. Excercise 4. the decay would even be forbidden.

λ) + ǫ(λ)∗ (k) eik·x c∗ (k. there are then indeed three independent possibilities.± 5. ±(1) are obtained by boosting the restframe vectors. A real vector field Vµ (x) then can be expanded in modes as Vµ (x) = λ=0. µ (2π)3 2E µ (5.7) . (5.  −B 1    0 3 With the electromagnetic field tensor (5. The vectorfield is therefore (λ) suited to describe spin 1. (2 + M 2 )Vµ = 0. just as the two-component Dirac spinors did for spin 1/2. namely ǫµ (k0 ) = (0.3) Since the vectors ǫµ (k) together with the momentum k µ form a complete set of four-vectors.2 we have discussed the vectors ǫλ (with λ = 0 and ±1) as basis states of the spin 1 representation. with ∂µ V µ = 0 (Lorentz condition). M2 (5.2 The electromagnetic field F µν = = ∂ µ Aν − ∂ ν Aµ  1 −E 2  0 −E  1  E  0 −B 3   2  E 3  B 0   3 E −B 2 B 1 Aµ = (φ. These are solutions of the Klein-Gordon equation with in addition a condition. A). λ) . In an arbitrary frame ǫµ (k) with λ = 0.4) λ=0. These can describe spin 1 states in the rest-frame. In the restframe.2) (5.1) The Lorentz condition implies that k µ ǫµ (k) = 0. where k = k0 = (M. ǫλ ). 39 (5.6) where Aµ is the four-vector potential (5. 0).5)  −E    B2   .± (λ) d3 k ǫ(λ) (k) e−ik·x c(k. The covariant generalization is to describe them with Vµ (x) = ǫµ (k) e±ik·x .Chapter 5 Vector fields and Maxwell equations 5.1 Fields for spin 1 In section 3. one has ǫ(λ) (k) ǫ(λ)∗ (k) = −gµν + µ ν kµ kν .

5. The gauge freedom can first of all be used to impose the Lorentz condition for the electromagnetic field.18) corresponding to the helicity states λ = ±1 of the massless photon field. The Lorenz condition can actually be incorporated by changing the equations of motion to 2Aµ + (λ − 1)∂µ (∂ν Aν ) = jµ .10) the electric and magnetic fields are unchanged. For electrodynamics one has furthermore the freedom of gauge transformations. Fµν −→ Fµν + (∂µ ∂ν χ − ∂ν ∂µ χ) = Fµν .15 the d’Alembertian is replaced by 2 + M 2 .e. j) the Maxwell equations read ∂µ F µν = j ν . shows that the electromagnetic fields correspond to the case of a massless spin 1 field. in accordance with what we found in section 3. One possible choice is the gauge A0 = 0 (5. 0). (5. Note that the incorporation of the Lorentz condition can also be done for the case of the massive vector field. it will vanish at all times. if in Eq. it does not eliminate the freedom of gauge transformations but they are now restricted to Aµ → Aµ + ∂µ χ with 2χ = 0. for M = 0. Furthermore. 40 (5. . 0. Under a gauge transformation Aµ −→ Aµ + ∂µ χ. while the fact µ C ∗ that the Minkowski space is real implies that Aµ (x) = Aµ (x) = Aµ (x). in which case one has 2Aµ = jµ .15) (5. only two independent vectors remain ǫ(±) µ (k) = (0.Vector fields and Maxwell equations and the (conserved) current j µ = (ρ. The equations of motion for the fields Aµ become 2Aµ − ∂µ (∂ν Aν ) = jµ .14) of which the solutions give the Li´nard-Wiechert potentials. i. |k|). one finds for a conserved current and λ = 0 2∂µ Aµ = 0.11) This equation is not affected by a gauge transformation. the antisymmetry of F µν implies that ˜ ∂µ F µν = 0 or ∂λ Fµν + ∂µ Fνλ + ∂ν Fλµ = 0. ∓1.9) 1 ˜ where F µν ≡ − 2 ǫµνρσ Fρσ is the tensor where E → B and B → −E. (5. It states the absence of magnetic charges and Faraday’s law. AP (x) = Aµ (˜). Taking the divergence of this equiation. ∂µ Aµ = 0. (5. 2Aµ = 0. if ∂µ Aµ = 0 for large times |t|.16) (5. As remarked above. In that case one also has ∇·A = 0 or ki ǫi (k) = 0. moree over. ˜ x Note that via parity transformation one obtains another solution. Therefore if k µ = (|k|. Although the Lorenz condition is a constraint.8) (5.17) (radiation. −i. there is still a gauge freedom left (Aµ → Aµ + ∂µ χ with 2χ = 0). 0. The equation in vacuum. (5. transverse or Coulomb gauge).13) Hence. (5. 5.12) (5.

The situation. .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. It is however. nothing more than a manisfes1 tation of a nontrivial vacuum (a la the Dirac sea for fermions). where Aµ = 0. The phase difference between the electrons travelling two different paths is given by δ = 2π a a ∆x d = −≈ . but A = 0 (hence there is structure in the vacuum). This causes an interference pattern as a function of ∆x. illustrated in fig. nevertheless. 5. λ L − λ λ or ∆x = (L− λ/d)δ.0 . ϕ= − ˆ 2r 2r2 2r2 where r = (cos ϕ. (5. . sin ϕ. but where E = B = 0.20) e (∇ × A) · ds = B · ds = where Φ is the flux through the solenoid. cos ϕ. may look a bit akward. This occurs in the region outside the solenoid. . 0). In the presence of an electromagnetic field Aµ an additional phase difference is observed.1.0 . To be precise.1: The Aharonov-Bohm experiment 5. F = e E + ev × B (5. = 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. The energy density 2 (E 2 + B 2 ) = 0 (hence a vacuum). while the electromagnetic field Aµ is not. 2 2 2 BR2 BR2 y BR2 x A= . Inside solenoid: Outside solenoid: B = B z. Nevertheless the significance of Aµ is shown in the Aharonov-Bohm experiment.19) are gauge invariant. In this experiment an observable phase difference is measured for electrons moving through field-free space (E = B = 0 but Aµ = 0). ˆ B = 0.Vector fields and Maxwell equations 41 L ∆x d a (infinitely) long solenoid Figure 5. The explanation of the significance of the vector ˆ ˆ potential A for the phase of the electron wave function is found in minimal substitution ψ ∝ exp i p·r −→ exp i (p − eA) · r e A · r exp i p·r . A= By Bx Br ϕ= − ˆ . This phase difference is actually observed. 0) and ϕ = (− sin ϕ.

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

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

Chapter 6 Classical lagrangian field theory 6.2) The principle of minimal action looks for a stationary action under variations in the coordinates and time. d dt ∂L ∂x ˙ = ∂L .5) with ∆x(t) = δx(t) + x(t) δτ . Dirac fields (Dirac equation) and massless vector fields (Maxwell equations) and corresponding to these fields conserved currents describing the probability and probability current. As an example. t1 (6.7) The quantity ∂L/∂ x plays a special role and is known as the canonical momentum. x) ˙ (6. ∂x ˙ (6. recall classical mechanics with the action t2 S= t1 dt L(x. x (t) = x(t) + δx(t). ˙ p= ∂L .1 Euler-Lagrange equations In the previous chapter we have seen the equations for scalar fields (Klein-Gordon equation). (6. ˙ 2 (6. x) = K − V = ˙ 1 mx2 − V (x).4) δS = t1 t2 dt dt t1 ∂L ∂L δx + δ x + L δτ ˙ ∂x ∂x ˙ d ∂L − ∂x dt ∂L ∂x ˙ δx + t2 t1 = ∂L δx + L δτ ∂x ˙ t2 .6) The first term leads to the Euler-Lagrange equations.8) 44 . ∂x (6.1) and as an example the lagrangian L(x. and the total change x′ (t′ ) = x(t) + ∆x(t). The requirement δS = 0 with fixed boundaries x(t1 ) = x1 and x(t2 ) = ˙ x2 leads to t2 ′ (6. thus t′ = t + δτ. These equations can be obtained from a lagrangian using the action principle.3) (6.

14) (6. . x′ ) − d4 x L (φ. . ∆x = (dx/dλ)λ=0 δλ and δt = (dt/dλ)λ=0 δλ. but which in 3 dimensions includes conserved quantities related to rotations and boosts. t2 ).15) d4 x′ L (φ′ . . indicated with δF [φ]/δφ should pose no problems. gives rise to a conserved quantity (at least if H is invariant!) Q=p dx dλ −H dt dλ . ∂µ φ(x)). = φ(x) + δφ(x) (6. giving1 δS = = δL δL δ(∂µ φ) + δφ + δ(∂µ φ) δφ R δL δL δφ + d4 x − ∂µ δφ δ(∂µ φ) R d4 x dσµ L δxµ ∂R dσµ ∂R δL δφ + L δxµ . δ(∂µ φ) (6. ∂R Note for the specific choice of the surface for constant times t1 and t2 .18) 1 Taking a functional derivative.12) Here R indicates a space-time volume bounded by (R3 .2. In classical field theory one proceeds in complete analogy but using functions depending on space and time (classical fields. 6.10) One sees that invariance under time translations requires that H(t1 ) = H(t2 ).e. ˙ (6. ˙ x The second term in Eq. λ=0 (6. Variations in the action can come from the coordinates or the fields. . also indicated by ∂R (a more general volume in four-dimensional space-time with some boundary ∂R can also be considered).13) (6. Consider a lagrangian density L which depends on these functions. their derivatives and possibly on the position. H is a conserved quantity.16) The change in variables x → x′ in the integration volume involves a surface variation of the form dσµ L δxµ . x). think for instance of a temperature or density distribution or of an electromagnetic field). indicated as x′µ φ (x) or combined φ′ (x′ ) = φ(x) + ∆φ(x).11) λ=0 of which p and H are the simplest examples related to space and time translations. . In general any continuous transformation.t2 ) d3 x . + ∂L ∆x − H δτ ∂x ˙ t2 . L (φ(x). The resulting variation of the action is δS = R µ ′ = xµ + δxµ . . . ∂µ φ(x)) = R d4 x L (φ(x). ∂µ φ. The hamiltonian H is defined by H(p.6 can be rewritten as δS = . − d3 x . ∂µ φ′ . dσµ . with ∆φ(x) = δφ(x) + (∂µ φ)δx . .17) Furthermore the variations δφ and δ∂µ φ contribute to δS. . (6.t1 ) (6.9) which is done because the first term (multiplying ∆x. = ∂R (R3 . R (6. . ∂µ φ(x). i.Classical lagrangian field theory 45 For the lagrangian specified above this leads directly to Newton’s law p = m¨ = −∂V /∂x. x) and an action t2 S[φ] = t1 dt L = dt d3 x L (φ(x). t1 ) and R3 . t1 (6. which in classical mechanics vanishes at the boundary) does not play a role. (R3 . We will come back to it in a bit more formal way in section 9. x) ≡ p x − L .

The integrand of the first term must vanish.19) δ(∂µ φ) δφ 6.Classical lagrangian field theory 46 With for the situation of classical fields all variations of the fields and coordinates at the surface vanishing. 2 ψ = 0. (6. δL δL ∂µ = . ∂ (6.20) (6.24) (6. The scalar field It is straightforward to derive the equations of motion for a real scalar field φ from the lagrangian densities. 2 (6. (6. the Dirac field (spin 1/2) and the vector field (spin 1) can be found. (2 + M )φ (x) = 0. 2 ∗ (6. For the complex scalar field one conventionally uses L = = ∂µ φ∗ ∂ µ φ − M 2 φ∗ φ −φ∗ ∂µ ∂ µ + M 2 φ.28) where the second line is not symmetric but in the action only differs from the symmetric version by a surface term (partial integration). leading to (2 + M 2 )φ(x) = 0.29) .21) which differ only by surface terms. (6.2 Lagrangians for spin 0. the second term is irrelevant. Using the variations in ψ (in the symmetric form).23) (6. δL δ(∂µ ψ) δL δψ one obtains immediately i / −M ∂ → i = − γ µψ 2 i → = ∂ / ψ − M ψ.27) (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. L = 1 1 ∂µ φ∂ µ φ − M 2 φ2 2 2 1 µ = − φ ∂µ ∂ + M 2 φ. leading to the Euler-Lagrange equations. 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). One easily obtains (2 + M 2 )φ(x) = 0.25) (6.22) which can be considered as the sum of the lagrangian densities for two real scalar fields φ1 and φ2 √ with φ = (φ1 + iφ2 )/ 2.

1 M η † ǫη ∗ − M ∗ η T ǫη .31 is given by LM (Dirac) = −M ξ † η + η † ξ .34)  (6. it is in the absence c of a Dirac mass term possible to introduce a lagrangian in which only left fields ψL and ψL appear or one can work with ’real’ spinors or Majorana spinors.32) There exists another possibility to write down a mass term with only one kind of fields. Peshkin and Schroeder or Exercise 12. (6. The reasons for Grassmann variables will become clear in the next chapter. 2 αβ = −βα.38) (6. or equivalently separate the field into right. the mass term in the Dirac lagrangian 6. . 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 . 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. that the lagrangian separates into two independent parts for M = 0. but note the factor 1/2 as compared to the Dirac lagrangian. We note that of which the left part coincides   .37) This gives an expression with a mass term that is actually of the same form as the Dirac mass term.g.39) 2 This term is written down with η being an anticommuting Grassmann number for which (αβ)∗ = β ∗ α∗ = −α∗ β ∗ .and lefthanded ones. Υi / Υ − ∂ 4 2   T  −ǫ η ∗  c  = ΥR .g. and thus (β ∗ α)∗ = α∗ β = −βα∗ . ψL ≡ (ψL )c = C ψL =    0 (6.31) showing e.33) We now introduce a ’real’ (four-component) spinor satisfying Υc = Υ with ψL . ∂ ∂ 2 2 (6.36) (6.    ∗  0    ⇒ Υ ≡  −ǫ η  ΥL = ψL =    η η ∗ for which η0 = ǫ η0 .35) Since the kinetic term in L separates naturally in left and right parts (or ξ and η). which comes because we in essence use ’real’ spinors.Classical lagrangian field theory 47 and similarly from the variation with respect to ψ ψ i / +M ∂ ← = 0.7). (6. namely2 LM (Majorana) = + (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. The Majorana case is in fact more general. 2 (6. In that case one finds trivially L = ↔ ↔ 1 1 ψR i / ψR + ψL i / ψL − M (ψR ψL + ψL ψR ). Using the twospinors 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.

(6. one can consider the variations at the initial or final surface. In particular when the lagrangian is invariant under symmetries.18. (6.42) Θµν = The variation δS thus can be expressed as δS with J µ (x) = − = δL ∂ ν φ − L g µν . As an example for the classical field. (6. discussed in the next chapter.43) F (σ1 ) − F (σ2 ) = dσµ J µ (x) = ∂R R d4 x ∂µ J µ (x). (6. the presence of a symmetry that leaves the lagrangian invariant.44) δL ∆φ + Θµν (x)δxν .41) (6. implying ∂µ F µν + M 2 V ν = 0 and ∂µ V µ = 0. 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 ) = 0.45) J µ (x) = − δ(∂µ φ) dλ λ=0 dλ λ=0 Thus considering ∂R = σ2 . δ(∂µ φ) For continuous transformations. δ(∂µ φ) (6. These do not affect the dynamics and will give rise to conserved currents.+ R 4 δL δL ∆φ − ∂ ν φ − L g µν δxν δ(∂µ φ) δ(∂µ φ) δL ∆φ − Θµν (x)δxν . consider U (1) transformations of the Dirac field proportional to the charge e. which is the space-integral over the zero-component of a conserved current. we consider the second term in Eq.σ1 with σ = (R3 . 6. requires the presence of a conserved quantity. (6. δ(∂µ φ) (6. t).Classical lagrangian field theory 48 where Fµν = ∂µ Vν − ∂ν Vµ . Q= d3 x J 0 (x. which in a consistent picture precisely generate the symmetries. we get (omitting the parameter δλ) dxν dφ δL + Θµν (x) . . these conserved quantities become operators. In this case..3 Symmetries and conserved (Noether) currents Next. Returning to δS the surface term is rewritten to δS = R d4 x .. . ∂µ J µ (x) = 0.46) 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 . Q(t1 ) = Q(t2 ).47) .40) 6. ψ(x) → ei eΛ ψ(x) or ∆ψ(x) = i eΛ ψ(x). t). ∆φ = (dφ/dλ)λ=0 δλ and δxµ = (dxµ /dλ)λ=0 δλ. + ∂R dσµ dσµ ∂R ≡ where d x .

.54) (6. From Noether’s theorem one sees that the current Θµν ǫν is conserved. ↔ (6. (6.48) For the complex scalar field the U (1) transformations leave the lagrangian invariant and lead to the conserved current (see Exercise 6. In the next section we will see the quantity show up as the charge operator.56) δxµ = ωρσ (aρσ )µν xν 2 with (aρσ )µν = aρσ µν = g ρµ g σν − g σµ g ρν . (6. t) is the conserved charge (Q = 0).Classical lagrangian field theory From the lagrangian for the Dirac field. = −(∂µ φ)δxµ = −ǫµ ∂µ φ(x). (6. we obtain (since δxµ = 0). 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. Summarizing. including e the space-time translations and the Lorentz transformations. e.3) j µ = i e φ∗ ∂µ φ. Lorentz transformations In this case the transformation of the coordinates and fields are written as δxµ ∆φi (x) = = ω µν xν .53) These are the energy and momentum.49) These currents are conserved as discussed already in chapter 2.57) For the fields the behavior under Lorentz transformations has been the subject of previous chapters. 2 (6. (6. Therefore there are four conserved currents Θµν (ν labeling the currents!) and four conserved quantities Pν = dσµ Θµν = d3 x Θ0ν . The integral over the zero-component. ˙ Q = d3 x j 0 (x.51) (6. 6.55) Note that the coordinate transformations can be written in a form similar to that for the fields (matrices in internal space).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.g. For quantized fields this will become the expressions of the hamiltonian and the momentum operators in terms of the fields.4 Space-time symmetries One kind of symmetries that leave the lagrangian invariant are the Poincar´ transformations. Translations µ Under translations (generated by Pop ) we have δxµ ∆φ(x) δφ(x) = ǫµ .50) (6. 1 (6. H = d3 x Θ00 (x). = 0. (ω µν antisymmetric) 1 ωρσ (−i S ρσ )ij φj (x).

one has a tensor that satisfies (exercise 6.60) Therefore there are six conserved currents M µρσ labeled by ρ and σ (antisymmetric) and corresponding to it there exist conserved quantities M ρσ = d3 x M 0ρσ . the angle of the transformation depends on the space-time point x. The current deduced from Noether’s theorem is Jµ = = = = − δL ∆φi + Θµν δxν δ(∂µ φi ) δL 1 ωρσ − (−i S ρσ )ij φj + Θµρ xσ − Θµσ xρ 2 δ(∂µ φi ) 1 ωρσ {H µρσ + Θµρ xσ − Θµσ xρ } 2 1 ωρσ M µρσ . In general this tensor is not symmetric. −i eΛ(x) ∗ i eΛ(x) (6.67) i. γσ ] This result for the Dirac field has been discussed in chapter 3 (see also Exercise 4.68) (6.61) A final note concernes the symmetry properties of Θµν . 2 (6. e. Gauge transformations of the second kind or local gauge transformations are transformations of the type φ(x) → ei eΛ(x) φ(x).62) with Gρµν = (H ρµν + H µνρ + H νµρ )/2. in which the U (1) transformation involves an angle e Λ. φ(x) φ (x) ∂µ φ(x) ∗ → e → e → ei eΛ(x) φ(x). In some applications (specifically coupling to gravity) it is advantageous to have an equivalent current that is symmetric. 6. The lagrangians which we have considered sofar are invariant under global gauge transformations and corresponding to this there exist a conserved Noether current. however. ∂µ T µν (6.70) .g. • Dirac field ψ: −i Sρσ = −(1/4)[γρ .3). (6.Classical lagrangian field theory 50 • Scalar field φ: −i Sρσ = 0.3 we have seen global transformations or gauge transformations of the first kind.6) T µν = T νµ .63) µν = ∂µ Θ . (6.69) i eΛ(x) φ (x). Defining T µν = Θµν − ∂ρ Gρµν .59) (6.58) (6. independent of x. where H ρµν is the quantity appearing in M ρµν . • Vector field Aµ : −i Sρσ = aρσ . Any lagrangian containing derivatives.65) M µρσ = T µρ xσ − T µσ xρ .66) In section 6. ∂µ φ(x) + i e ∂µ Λ(x) e (6. φ(x). is not invariant under local gauge transformations. (6.64) (6. (6.e.5 (Abelian) gauge theories φ(x) → ei eΛ φ(x). (6.

A solution is the one known as minimal substitution in which the derivative ∂µ is replaced by a covariant derivative Dµ which satisfies Dµ φ(x) → ei eΛ(x) Dµ φ(x). Dµ φ) − Fµν F µν . 1 L (φ.73) (6. The equation of motion for the fermion follow from δL δ(∂µ ψ) δL δψ = = i − γ µ ψ.72) (6. D ∂ A (6. Replacing derivatives by covariant derivatives and adding the (free) part for the massless vector fields to the original lagrangian therefore produces a gauge invariant lagrangian.75) The field φ is used here in a general sense standing for any possible field. For this purpose it is necessary to introduce a vector field Aµ .Classical lagrangian field theory 51 where the last term spoils gauge invariance.78) .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 .74) the behavior which we have encountered before as a gauge freedom for massless vector fields with the (free) lagrangian density L = −(1/4)Fµν F µν . (i/ − M ) ψ = (i/ − e/ − M ) ψ = 0.71) Thus the covariant derivative has the correct transformation behavior provided Aµ → Aµ − ∂µ Λ. (6. 4 (6. As an example consider the Dirac lagrangian. ∂µ φ) =⇒ L (φ. (6. ∂ 2 4 (6. (6. 2 i ∂ / ψ − M ψ − e/ ψ A 2 giving the Dirac equation in an electromagnetic field. −e j µ Aµ = −e ρφ + e j · A. i L = ψγ µ ∂µ ψ − (∂µ ψ)γ µ ψ − M ψψ. Dµ φ(x) ≡ (∂µ + i eAµ (x))φ(x).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). 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 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 .

[iγ µ (∂µ − i e Aµ ) − M ]ψ = 0. giving the Maxwell equation coupling to the electromagnetic current. .79) Exercises Exercise 6. and deduce what is the charge of the antiparticle as compared to a particle.4 Given the Dirac equation for a (negatively charged) spin-1/2 particle in an external electromagnetic T field. (6. ∂µ F µν = j ν . i.1 (a) Show that the Klein-Gordon equation for the real scalar field can be derived from the lagrangian density 1 1 L = ∂µ φ∂ µ φ − M 2 φ2 .2 (a) Show that the homogeneous Maxwell equations can be derived from the lagrangian density 1 L = − Fµν F µν .e. (c) Show that the interaction term is invariant under gauge transformations only if the current jµ is conserved. ∂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). Exercise 6. −eψγ ν ψ. give the equation which is satisfied by ψ c = Cψ . where j µ = e ψγ µ ψ. Exercise 6. 2 2 (b) Show that the Klein-Gordon equation for the complex scalar field (considering φ and φ∗ as independent fields can be derived from the lagrangian density L = ∂µ φ∗ ∂ µ φ − M 2 φ∗ φ.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. φ → eiΛ φ.Classical lagrangian field theory 52 For the photon the equations of motion follow from δL δ(∂µ Aν ) δL δAν = = −F µν .

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. in order to show that Θρσ − Θσρ = ∂µ H µρσ . use ∂µ T µν = 0 and ∂µ M µρσ = 0.Classical lagrangian field theory 53 Exercise 6.4.6 (optional) Prove the properties of the tensor T µν in section 6. Comment: this relation is known as the Pauli equation. t). A) (see also exercise 6.5 Show that the Dirac equation for an electron with charge −e in an external electromagnetic field Aµ = (A0 . 2M 2M with Enr ≡ E − M ≪ M . Hints: it may be useful to realize that Gµρσ is antisymmetric in first two indices. This relation is known as the Pauli equation.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 . finally note that it is sufficient to show the last equation for dσµ M µρσ . e 1 (−i∇ + eA)2 + σ · B − eA0 ψu (r. t) = Enr ψu (r. Hint ≡ −gs µe · B = −gs Qe s · B. . 2me Exercise 6.

dq dp dq dp (7. p]P . (7. ˙ L(q.2) H(p.1 Canonical quantization We will first recall the example of classical mechanics for one coordinate q(t). dq The basic Poisson bracket is actually the one between q and its canonical momentum p. H]P dt and dO = [O. q) also considered in the previous chapter. dλ Examples are dO = [O. 1 qx (t) = 3 d3 x φ(x. An immediate generalization for fields can be obtained by considering them as coordinates. B]P = dB dA dA dB − . in our example p = mq. Further. (7. Q]P . is the Poisson bracket [A. [q. (7. p)) = ˙ ˙ 2m where the (canonical) momentum p = ∂L/∂ q. labeled by the position. p]P = 1. q) = ˙ 1 mq 2 − V (q). however. namely [q.5) ∆ x ∆3 x 54 .1) The Hamiltonian (also corresponding to a conserved quantity because of time translation invariance) is given by p2 + V (q).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. q) = p q − L(q.3) For variations δO(p. The bilinear product for the conserved quantities. t). p] = i. there are conserved ˙ ˙ quantities Q (Eq. Actually the conserved quantities satisfy a similar algebra as the Lie algebras for continuous symmetries that they are obtained from. Quantization of the theory is achieved by imposing canonical commutation relations between q and p. 6.11) found through Noether’s theorem. starting from the lagrangian L(q. q(q. q) ≈ (dO/dλ)λ=0 δλ under symmetry transformations one has dO = [O. ˙ 2 (7.Chapter 7 Quantization of fields 7. This is nothing else as representing the classical Poisson algebra as operators in the Hilbert space.

Quantization of fields 55 etc. ∆φ. Φ(f ) = d4 x φ(x) f (x).10) x ∆3 x [Πx qx − L x ] ˙ ∆3 x Hx (7. Π(y. ∇φ) (7.7) d3 x L (φ(x). for symmetry transformations. albeit with ’sharp’ functions. φ. φ(y.15) with furthermore the relations [φ(x. t)] = i δ 3 (x − y).16) .14) In fact.12) (7. The hamiltonian then is H= x px qx − L = ˙ = x where H (x) = δL ˙ φ(x) − L (x) = Θ00 (x). the discretization procedure above is an explicit example. This implies specific transformation properties in the Hilbert space for the quantum fields φ. (7. t)] = [Π(x. for the scalar field theory. a) φi (x)U (Λ.6) ∆3 x L x (qx . ∂µ φ(x)) = ˙ d3 x L (φ. t). 2 2 2 (7. Π(y.11) (7. t).13) Π(x) H (x) For the quantization procedure we can formulate a number of basic axioms of quantum field theory. py (t)] = i δxy or for the fields φ(x) and Π(x) the socalled equal time commutation relations [φ(x. ˙ δ φ(x) (7. The lagrangian is given by L = = x In order to construct the hamiltonian it is necessary to find the canonical momenta. i U † (Λ. As an example.9) Note that this indeed corresponds to the zero-zero component (Θ00 ) of the conserved energy-momentum stress tensor Θµν . ∂ qx ˙ ∂ qx ˙ (7. δ∂0 φ) 1 1 1 = Θ00 (x) = (∂0 φ)2 + (∇φ)2 + M 2 φ2 . discussed in the section 6. a) = Rj (Λ−1 ) φj (Λ−1 x − a). (7. formally. 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. (7. qx+∆x ) . Sometimes it is useful to keep in mind that. • Canonical commutation relations The quantization of the theory is achieved by imposing the canonical quantization condition [qx (t). The essential conditions to consistently quantize a theory then are the following. ˙ where Πx (t) is obtained from the continuous field Π(x) ≡ δL /δ(∂0 φ). the fields φ(x) can be considered as regular operators in the Hilbert space after smearing with a test function f . • Poincar´ invariance e The full variation of the fields. qx . t).2.8) (7. t)] = 0. px (t) = ∂L ∂Lx = ∆3 x = ∆3 x Πx (t).

j 2 one has [φi (x). H = = ω ω P P Q−i √ Q+i√ 2 2 2ω 2ω 1 1 =ω N+ .18) with S µν given in 6. (7.22) it is straightforward to check that the commutation relations between Q and P are equivalent with the commutation relations [a. φ(y)] = 0 if (x − y)2 < 0. a) = i Rj (Λ−1 ) = 1 + iǫµ P µ − 1− i ωµν M µν . (7. j (7.2 Creation and annihilation operators Before discussing (real and complex) scalar fields and Dirac fields we recall the analogy with the wellknown harmonic oscillator as an example of quantization using creation and annihilation operators. In simplified form the hamiltonian is given by H= 1 2 1 2 2 P + ω Q . Pµ ] = [φ (x).21) Writing Q and P in terms of creation (a† ) and annihilation (a) operators.20) where the coordinate Q and the momentum P satisfy the canonical commutation relations [Q. P ] = i. i(xµ ∂ν − xν ∂µ )φ (x) + i (7. Q] = [P. P ] = 0 (7.25) . • Causality Operators Φ(f ) and Φ(g) for which the test functions are space-like separated can be measured simultaneously (macroscopic causality). Microscopic causality implies local commutativity. sometimes referred to as second quantization. a] = [a† . a] = −a. a† ] = a† . 2 2 (7. [Q.4. The measurements cannot influence each other or [Φ(f ).19) 7. [a.23) The hamiltonian in this case can be expressed in the number operator N = a† a. a† ] = 1. Φ(g)] = 0. ω a† a + 2 2 ω i − [Q. 1 Q = √ (a + a† ) 2ω and P = −i ω (a − a† ) 2 (7.Quantization of fields 56 or if U (Λ. 2 i ωµν (S µν )i . Mµν ] = i i∂µ φi (x). a† ] = 0. and [N.17) (Sµν )i φj (x).24) It is straightforward to find the commutation relations between N and a and a† . [N. [φ(x). 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. (7. P ] 2 (7.

7. d3 x φ(x) i ∂0 e−i k ·x ↔ ′ ↔ (7.e. e. a(k ′ )] = [a† (k). The quantization of the field is achieved by quantizing the coefficients in the Fourier expansion.27) 2 (2π)3 It is easy to check that these equations can be inverted (see Exercise 2.g.28) (7.32) (7. and (7. which represents the ’number of particles’ with momentum k.4 for the classical field) a(k) a† (k ′ ) = = d3 x ei k·x i ∂0 φ(x). the real scalar field φ(x) becomes d3 k (7. (7. [φ(x).26) a(k) e−i k·x + a† (k) ei k·x .like’ commutation relations between a(k) and a† (k ′ ). 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. is equivalent with [a(k). and we see that a state |0 must exist for which N |0 = a|0 = 0. † √ a and a act as raising and lowering operators. i.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. The canonical momentum becomes −i d3 k ˙ Π(x) = φ(x) = a(k) e−i k·x − a† (k) ei k·x . (n − 1) a|n † i. but one needs to realize that in that case k 0 = Ek = k2 + M 2 .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). Π(x )]x0 =x′0 = 0.e. a† (k ′ )] = 0. N |n = n|n one finds N a† |n N a|n = = (n + 1) a† |n .30) [φ(x). a† (k ′ )] = (2π)3 2Ek δ 3 (k − k′ ) [a(k). (2π)3 2Ek (7. φ(x )]x0 =x′0 = [Π(x).33) .3 The real scalar field We have expanded the (classical) field in plane wave solutions. From the normalizations one obtains a |n = √ n + 1 |n + 1 and a|n = n |n − 1 .Quantization of fields 57 Defining states |n as eigenstates of N with eigenvalue n. Π(x′ )]x0 =x′0 = i δ 3 (x − x′ ) ′ ′ and (7. H = = = d3 x 1 1 1 (∂0 φ)2 + (∇φ)2 + M 2 φ2 2 2 2 d3 k Ek † a (k)a(k) + a(k)a† (k) (2π)3 2Ek 2 d3 k Ek N (k) + Evac . which we have split into positive and negative energy pieces with (complex) coefficients a(k) and a∗ (k) multiplying them.

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.44) p|φ− (x)|0 = ei p·x . and multiparticle states |(k1 )n1 (k2 )n2 . writing all annihilation operators to the right of the creation operators. This term will be adressed below. 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 d3 k a† (k) ei k·x .42) (7. |0 .40) This procedure is known as normal ordering. . is solved by subtracting it as an (infinite) constant. = normalized as (a† (k1 ))n1 (a† (k2 ))n2 √ √ .38) d4 k 2π δ(k 2 − M 2 ) θ(k 0 ) |k k| = (2π)4 (7. |k = a† (k)|0 . n1 ! n2 ! (7. d3 k |k k|. .37) (7. (2π)3 2Ek (7.35) where the vacuum contribution disappears because of rotational symmetry. (2π)3 (7. Just as in the case of the harmonic oscillator it is essential (axiom) that there exists a ground state |0 that is annihilated by a(k). a(k)|0 = 0.e.39) The problem with the vacuum or zero-point energy. in field theory also referred to as vacuum energy Evac = 1 V 2 d3 k Ek . . which now contains an infinite number of oscillators. defining particle states |k = |k (with positive energy.36) and satisfying the completeness condition 1 = k|k ′ = (2π)3 2Ek δ 3 (k − k′ ). (2π)3 2Ek (7.45) (7. . (2π)3 2Ek (7. which amounts to redefining H as H= d3 x : H (x) := d3 k Ek N (k). i.34) where V = (2π)3 δ 3 (0) is the space-volume. assuring that the vacuum (by definition) has eigenvalue 0!. k 0 = Ek = k2 + M 2 ). (2π)3 2Ek (7. The rest of the states are then obtained from the groundstate via the creation operator. d3 k a(k) e−i k·x . .46) (7. φ(x) φ+ (x) φ− (x) = = = φ+ (x) + φ− (x).43) The normal ordered product can be expressed as : φ(x)φ(y) := φ+ (x)φ+ (y) + φ− (x)φ+ (y) + φ− (y)φ+ (x) + φ− (x)φ− (y).41) (7. For the purpose of normal ordering it is convenient to decompose the field in positive and negative frequency parts. 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).

x) = 0 and hence ∆(x) = 0 for x2 < 0. (2π)4 (7. φ(y)] consider µν [φ+ (x). a† (k ′ )] 3 2E ′ (2π) k [φ− (x).55) d4 k ǫ(k 0 ) 2π δ(k 2 − M 2 ) e−i k·x .17) and (7.Quantization of fields 59 In order to ensure the consistency of the theory it is necessary to check that the operators P ν and M obtained from the conserved currents Θρσ and M µρσ are indeed the generators of the Poincar´ e group.56) = −δ 3 (x). (7. d3 k = (2π)3 2Ek = (7. i∆+ (x) i∆− (x) = = = d4 k θ(k 0 ) 2π δ(k 2 − M 2 ) e−i k·x . (2π)3 2Ek ≡ i∆− (x − y).54) = i∆(x − y) = i (∆+ (x − y) + ∆− (x − y)) .47) ′ d3 k ′ e−i k·x+i k ·y [a(k).53) t=0 .50) or as integrals over d4 k. d3 k ei k·(x−y) = −i∆+ (y − x).51) (7. The last item to be checked for the scalar field are the causality condition. = − (2π)3 2Ek (7. 2π (7. (7. φ+ (y)] d3 k e−i k·(x−y) . i.52) The result for the invariant commutator function is [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 (7.18).49) (7. ∆(x) = − ǫ(x0 ) δ(x2 ). (ii) ∆(x) is a solution of the homogeneous Klein-Gordon equation.e. (iv) The equal time commutation relations follow from ∂ ∆(x) ∂t (v) For M = 0. φ(y)] which has the properties (i) i∆(x) = i∆+ (x) + i∆− (x) can be expressed as i∆(x) = where ǫ(k 0 ) = θ(k 0 ) − θ(−k 0 ).48) (7. φ− (y)] ≡ i∆+ (x − y). that they satisfy the required commutation relations in Eqs (7. (iii) ∆(0. In order to calculate [φ(x).

59) (7. we will consider it as a repetition of the quantization procedure.e. (2π)3 2Ek (7. particles (created by a† ) and antiparticles (created by b† ) with the same momentum contribute equally to the energy. which is equivalent to the relations (only nonzero ones) [a(k). b† (k ′ )] = (2π)3 2Ek δ 3 (k − k′ ).61) (7. Also the charge operator requires normal ordering (in order to give the vacuum eigenvalue zero).63) i. 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) .65) 7. Q = = = i d3 x : φ† ∂0 φ − ∂0 φ† φ(x) : d3 k : a† (k)a(k) − b(k)b† (k) : (2π)3 2Ek d3 k a† (k)a(k) − b† (k)b(k) .5 The Dirac field L = ↔ i ψγ µ ∂µ ψ − M ψψ.66) . ↔ (7. The field satisfies the Klein-Gordon equation and the density current (U (1) transformations) and the energy-momentum tensor are jµ Θµν The quantized fields are written as φ(x) = φ† (x) = d3 k a(k) e−i k·x + b† (k) ei k·x . (7. = ∂{µ φ∗ ∂ν} φ − L gµν . a† (k ′ )] = [b(k). 2 From the lagrangian density (7. (2π)3 2Ek (7. extending it with the charge operator and the introduction of particle and antiparticle operators. (2π)3 2Ek (7. φ† (y)] = i∆(x − y).62) (7.Quantization of fields 60 7.60) = i φ∗ ∂µ φ.4 The complex scalar field In spite of the similarity with the case of the real field.58) and satisfy the equal time commutation relation (only nonzero ones) [φ(x).64) The commutator of φ and φ† is as for the real field given by [φ(x). ∂0 φ† (y)]x0 =y0 = i δ 3 (x − y).57) (7. (2π)3 2Ek d3 k b(k) e−i k·x + a† (k) ei k·x .

(7. s′ )} = {d(k. (2π)3 2Ek (7. s) ei k·(x−y) x0 =y 0 . s)e−i k·x + d† (k. b† (k ′ . (2π)3 2Ek d3 k b† (k. † {ψi (x). s)ei k·x + d(k. The solution is the introduction of anticommutation relations. s).73) (7. ψj (y)}x0 =y0 = d3 k (2π)3 2Ek ui (k.78) (7. d† (k ′ .75) (7.67) i ψ / ψ − M ψψ gµν . s)v(k.e. s)vj (k. ˙ δ ψ(x) ↔ i Θ00 (x) = − ψγ i ∂i ψ + M ψψ 2 = iψγ 0 ∂0 ψ = iψ † ∂0 ψ. s)u(k.72) ψ(x) = s In terms of the operators for the b and d quanta the hamiltonian and charge operators are (omitting mostly the spin summation in the rest of this section) H = = Q = = d3 x : ψ † (x) i∂0 ψ(x) : d3 k Ek : b† (k)b(k) − d(k)d† (k) : (2π)3 2Ek d3 x : ψ † ψ : d3 k : b† (k)b(k) + d(k)d† (k) : . The quantized fields are written ψ(x) = s d3 k b(k. (2π)3 2Ek (7.68) The canonical momentum and the hamiltonian are given by Π(x) H (x) = = δL = i ψ † (x). {b(k. s)¯(k. s).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.77) Note that achieving normal ordering. (7. s) e−i k·(x−y) j s † + vi (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) .69) (7.79) Also for the field and the canonical conjugate momentum anticommutation relations are considered. ∂ 2 ↔ (7. jµ Θµν = = ψγµ ψ. . u v (2π)3 2Ek (7. s′ )} = (2π)3 2Ek δ 3 (k − k′ ) δss′ .71) (7.Quantization of fields 61 the conserved density and energy-momentum currents are easily obtained. s)¯(k. s)u† (k. ↔ i ψγµ ∂ν ψ − 2 (7. s)ei k·x .70) where the last line is obtained by using the Dirac equation. i.74) (7. s)e−i k·x . interchanging creation and annihilation operators.

˙ δ A0 δL = F i0 = E i . ∂ (7. −λ(∂ρ Aρ ). ˙ δ Ai (7.85) which reflects the gauge freedom. ψ j (y)] = (i/x + M )ij i∆1 (x − y).80) 3 = δ 3 (x − y) δij .82) where i∆1 = i∆+ −i∆− .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. It is possible to continue in a covariant way with the lagrangian 1 λ L = − Fµν F µν − (∂µ Aµ )2 . ∂ (7. When we would have started with commutation relations for the field ψ and the canonical momentum.84) (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). 4 Π0 Πi δL = 0. it implies the Lorentz constraint and leads to the canonical momenta Π0 Π i = = E i. one has † {ψi (x).87) If one wants to impose canonical commutation relations ∂µ Aµ = 0 cannot hold as an operator identity. 4 2 This gives the equations of motion discussed before.88) . (7. (7.Quantization of fields 62 Using the positive and negative energy projection operators discussed in section 4.81) where i∆ = i∆+ + i∆− is the same invariant commutator function as encountered before. ψ j (y)} at arbitrary times one has {ψi (x). but has the problem of being noncovariant.83) From the lagrangian density the canonical momenta are = = (7. For the scalar combination {ψi (x). as the vanishing of Π0 induces a constraint. 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). we would have obtained [ψi (x). but we must restrict ourselves to the weaker condition B|∂µ Aµ |A = 0.

3) satisfy the conditions of a Lie algebra. µ λ=0 (7.93) which does exhibit problems with the time-like photon. Q]P . pj ]P . for which it is sufficient that ∂µ Aµ |A = 0.94) Choosing k µ = (|k 3 |. The quantized field is expanded as Aµ (x) = d3 k (2π)3 2Ek (λ) 3 ǫ(λ) (k) c(k. Aν (y)]x0 =y0 = i gµν δ 3 (x − y). a longitudinal photon and two transverse photons. λ′ )] = −g λλ 2Ek (2π)3 δ 3 (k − k′ ). λ)ei k·x . q) that dO/dλ = [O. one time-like photon by itself is not allowed! This solves the problems with the normalization and the negative energies. The hamiltonian in terms of the creation and annihilation operators is (after normal ordering) given by H= d3 k Ek (2π)3 2Ek 3 λ=1 c† (k. 0) . Πν (y)] = 0. These problems are solved by the Lorentz constraint between physical states given above. 6. and [ℓi . Πν (y)]x0 =y0 = i gµν δ 3 (x − y). 0.1 (optional) If you haven’t seen Poisson brackets in classical mechanics you should (a) Proof that Poisson brackets (Eq. (c) Calculate the Poisson brackets for the coordinates x and the standard conserved quantities p and ℓ. what are [ri . j = 1. λ) − c† (k. (7. where Q is the conserved quantity in Eq. containing a time-like photon. λ). and are equivalent with [c(k. c† (k ′ .Quantization of fields 63 for physical states |A and |B . k 3 ) this reads |k 3 |c(0) (k) − k 3 c(3) (k) |A = 0 c(0) (k) ∓ c(3) (k) |A = 0. [ℓi . ˙ ˙ . λ)e−i k·x + c† (k. (7. λ)|A = 0. λ)c(k.91) (7. pj ]P . i.90) where Πν = F ν0 − λ g ν0 (∂ρ Aρ ) and we have furthermore [Aµ (x). 0. The canonical equal time commutation relations are [Aµ (x).95) Exercises Exercise 7. which reproduce the Hamilton equations. It gives 3 k µ ǫ(λ) (k)c(k. rj ]P . µ λ=0 (7. where Aµ + + is the part of the vector field containing the annihilation operators. i. In fact the commutation relations imply ˙ [Aµ (x).e. ′ (7.89) with four independent vectors ǫµ .11.92) 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. (d) Use Poisson brackets to calculate q and p. (7. 2.e. (b) Proof for a quantity O(p. Aν (y)] = [Πµ (x). 0)c(k. 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.4 (Green’s functions) (a) Show that ∆hom (x) = 1 (2π)4 d4 k e−ik·x δ(k 2 − M 2 )f (k) is a solution of the homogeneous Klein-Gordon equation. . we can write for ∆R an integral where the k 0 integration remains along the real axis.2 Prove that the equal time commutation relations between quantum fields φ(x) and Π(x′ ) are equivalent with the commutation relations between a(k) and a† (k ′ ) Exercise 7. (2 + M 2 ) ∆hom (x) = 0. ∆R = 1 (2π)4 d4 k e−ikx k2 1 − M 2 + iǫk 0 (with the implicit prescription to take ǫ → 0 after integration). (2 + M 2 ) ∆inhom (x) = −δ 4 (x). It is invariant.Quantization of fields 64 Exercise 7. 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).3 (a) Show that the correct implimentation of symmetries in the Hilbert space. Pµ ] = i∂µ φ(x). (c) What are the poles in the integral under (b) in the (complex) k0 plane? (d) Depending on the paths in the k0 plane going from k0 = −∞ to k0 = +∞ one can distinguish four different Green’s functions (solutions to the above inhomogeneous Klein-Gordon equation): k0 -E E ∆ R (retarded) ∆ C (causal) ∆ ∆A (advanced) Show that by appropriate shifts of the poles into the complex plane. ∆hom (Λx) = ∆hom (x) if f is an invariant function. Is the subtraction of ’zero point’ contributions essential in this check? µ Pµ φ(x) ei a µ Pµ = φ(x − a) Exercise 7. Give also the expression for ∆C . (Similarly one can show that for the advanced Green’s function ∆A (x) = 0 if x0 > 0). (e) Show that ∆R (x) = 0 if x0 < 0. e−i a requires [φ(x).

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

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

5) (8. λ)u(k. m) = = ˜ u(k.11) (8. 8. iγ µ† ∂µ − M ψ(˜) = 0. x µ (iγ ∂µ − M ) γ0 ψ(˜) = 0. λ)v(k. −λ) e−ik·x − d† (k. x The latter behavior of the vector field will be discussed further below. .6) (check this for the standard representation. (2π)3 2Ek (8. In the same way as the Fermion field.10) i.7) = λ d k x x ηP b(k. choosing ηP is real (ηP = ±1) particle and antiparticle have opposite parity. x (8. The result is ψ p (x) −1 = Pop ψ(x) Pop = ηP γ0 ψ(˜) x (8. λ)u(k. the above operation reverses the sign of λ). λ) Pop = ηP b(k. λ) eik·˜ (2π)3 2Ek d3 k ˜ ˜ ˜ ˜ ηP b(k.8) 3 = λ = λ = λ From this one sees immediately that −1 ˜ Pop b(k. −λ). −λ) e−ik·x − d† (k. m) γ0 v(k. m). x Therefore γ0 ψ(˜) is again a solution of the Dirac equation and we have x ψ p (x) = γ0 ψ(˜). if helicity λ is used instead of the z-component of the spin m. op P (8. λ) e−ik·˜ + d† (k. λ)γ0 v(k. m).e. −λ) eik·x (2π)3 2Ek ˜ d3 k ˜ ˜ ˜ ˜ ηP b(k. −λ). one can also consider the scalar field and vector fields. λ)v(k. λ) e−ik·x − d† (k.3) (8.9) (8. λ)γ0 u(k. γ x (8.2 Charge conjugation We have already seen the particle-antiparticle symmetry with under what we will call charge conjugation the behavior T (8.4) (8. −λ)v(k. For the scalar field we have seen −1 φ(x) −→ Pop φ(x) Pop = ηP φ(˜).Discrete symmetries 67 After parity transforming x to x the Dirac equation becomes after some manipulations ˜ ˜ x iγ µ ∂µ − M ψ(˜) = 0. ˜ −v(k. λ) eik·x .12) and for the vector field −1 Aµ (x) −→ Pop Aµ (x) Pop = −Aµ (˜). x It is straightforward to apply this to the explicit field operator ψ(x) using γ0 u(k.13) ψ(x) −→ ψ c (x) = ηC C ψ (x). (i˜ µ ∂µ − M ) ψ(˜) = 0. −λ)u(k. λ) P −1 = −η ∗ d(k. −λ) eik·x 3 2E (2π) ˜ k d3 k ˜ ˜ ηP b(k. ˜ Pop d(k.

For a quantized field one has −1 ∗ −1 Top fk (x)bk Top = fk (x)Top bk Top . x µ (iγ ∂µ − M ) γ5 Cψ(−˜) = 0. (2π)3 2Ek (8. γ x i(γ5 C)−1 γ µ γ5 C∂µ − M ψ(−˜) = 0. m) ¯ = = v(k. Norm conservation requires Top to be anti-unitary2 .e. x 1A 2 An −iC −1 γ µ C∂µ − M ψ(−˜) = 0. ((iγ µ )∗ ∂µ − M ) ψ(x) = 0. λ). writing −1 ψ(x) −→ Top ψ(x)Top = ηT Aψ(−˜) ≡ ψ t (−˜). m). x (8. λ)C v T (k. (8.3 Time reversal xµ = (t. m). As time reversal will transform ’bra’ into ’ket’. x x (8. λ) e−ik·x + b† (k. λ)u(k.22) = Aψ|Aφ ∗ is antilinear if A(λ|φ + µ|ψ ) = λ∗ A|φ + µ∗ A|ψ .17) 3 = λ This shows that −1 Cop b(k. 8. i. r). x (8. as discussed in section 4).14) (8. we start with the complex conjugated x Dirac equation for ψ. Therefore ψ c (x) = = λ −1 ¯ Cop ψ(x) Cop = ηC C ψ T (x) (8. λ)C uT (k. λ). to find ψ t (−˜) that is a solution of the Dirac equation. λ) e−ik·x + b† (k. it is antilinear1 . λ)v(k.18) (8.16) d k ηC d(k. The action on the spinors (using C = iγ 2 γ 0 = −iρ1 σ 2 in standard representation) gives C uT (k. Top |φ = φt | = (|φt )∗ . x iγ µT ∂µ − M ψ(−˜) = 0. λ) eik·x ¯ ¯ (2π)3 2Ek d3 k ηC d(k. . The same relations hold for helicity states. The (time-reversed) Dirac equation becomes.Discrete symmetries 68 the latter being also a solution of the Dirac equation.19) Cop d(k. ˜ x −(iγ µ )∗ ∂µ − M ψ(−˜) = 0. (8.21) where A is a 4 × 4 matrix acting in the spinor space. λ) −1 Cop = ∗ ηC b(k. m) ¯ C v T (k. One has Aφ|Aψ = φ|ψ ∗ = ψ|A† Aφ = ψ|φ . λ) eik·x . u(k.20) The time reversal operator transforms We will again consider the transformation properties for a fermion field ψ(x). λ) Cop = ηC d(k.15) (where one must be aware of the choice of spinors made in the expansion. antilinear operator is anti-unitary if A† = A−1 . r) −→ −˜µ ≡ −xµ = (−t. (i˜ µ∗ ∂µ − M ) ψ(−˜) = 0.

λ) e−ik·x 3 2E (2π) k d3 k ˜ ˜ ηT b(k. λ |¯. λ). λ)v ∗ (k. λ) Top = ηT b(k. Therefore one has A|k = A|b† (k)|0 = A|T † T b† (k) T † T |0 ∗ = At |b† (k)|0 ∗ ˜ t = k|At . −p.29) In table 2 the behavior of particle states under the various transformations has been summarized.4 Bi-linear combinations In quantities such as currents and lagrangians often bilinear combinations of spinor fields are encountered. The result is = = λ (8. ∗ ˜ v (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)u∗ (k. λ) eik·x + d† (k. λ a P |a.30) (8. x (8.24) (8.33) (8. there are 16 independent of these bilinear combinations. λ) e−ik·x 3 2E (2π) k ˜ d3 k ˜ ˜ ˜ ˜ ηT b(k. λ) eik·x + d† (k. λ) eik·˜ + d† (k. λ). λ). −λ a T a. λ)v ∗ (k. They are the following S(x) V µ (x) T µν (x) Aµ (x) P (x) ¯ = ψ(x)ψ(x) (scalar) ¯ = ψ(x)γ µ ψ(x) (vector) µν ¯ = ψ(x)σ ψ(x) (tensor) µ ¯ = ψ(x)γ5 γ ψ(x) (axial vector) ¯ = iψ(x)γ5 ψ(x) (pseudoscalar) (8.34) . λ) e−ik·˜ (2π)3 2Ek d3 k ˜ ˜ ˜ ˜ ηT b(k. λ) eik·x + d† (k.2: The transformation properties of physical states for particles (a) and antiparticles (¯). (8. λ)u∗ (k. −p. λ) ψ t (x) −1 Top ψ(x) Top = i ηT γ5 Cψ(−˜) x = = ˜ u∗ (k. ˜ = 0|b(k)|A 8. Since there are 16 independent 4 × 4 matrices. −p.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.28) (8. p.Discrete symmetries 69 Table 8. λ) e−ik·x (2π)3 2Ek (8. λ)v ∗ (k. Note that applying an anti-unitary transformation such as Top one must take for the matrix element ˜ the complex conjugate.27) 3 = λ = λ = λ From this one obtains −1 ˜ Top b(k. a state |a. λ) i γ5 C v(k. −p. λ). λ| a. p.26) d k x x ηT b(k. λ a |a. −λ |¯. λ) T −1 = η ∗ d(k.31) (8. p. op T (8. p. ˜ Top d(k.25) (check this for the standard representation). λ| ¯ C |¯.32) (8. λ)iγ5 Cu(k. λ)iγ5 Cv(k. λ)u∗ (k.

¯ ′ ′ (8. p) can be built from any combination of Dirac matrices (1. for the vector current V µ (x). p)u(p) e−i (p−p )·x . As the coupling of the photon field to fermions is given by an interaction term in the ¯ lagrangian of the form : ψ(x)γ µ ψ(x) : Aµ (x) and behaves as a scalar one sees immediately that the ¯ photon field Aµ (x) behaves in the same way as the vector combination ψ(x)γ µ ψ(x). as well as under the combined operation Θ = PCT (see Table 3). s .. C.Discrete symmetries 70 Table 8. (q 2 ). e. F. p′ |V µ (x)|p = u(p′ )Γµ (p′ . (8. 8.38) (8. ¯ V µ (x) =: ψ(x)γ µ ψ(x) :.40) where the coefficients are functions of invariants constructed out of the momenta. (q 2 ).. There are several other terms that also have the correct tensorial behavior such as (pµ + p′µ ) γ5 σ µν qν q µ F3 (q 2 ). called form factors. σ µν . p′ .36) (8.39) where Γµ (p′ . γ ν ]. of which the expectation value between momentum eigenstates can be simply found. p) = γµ F1 (q 2 ) + iσµν q ν F2 (q 2 ).37) (8.35) The x-dependence can be accounted for straightforwardly using translation invariance. 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 .41) 2M 2M . γ5 γ µ or γ5 ). V µ (x) = ei Pop ·x V µ (0)e−i Pop ·x .3: The behavior of the independent bilinear combinations of fermi fields under P. 2M (8.g. This implies p′ |V µ (x)|p = e−i (p−p )·x p′ |V µ (0)|p . γ µ . The 16 combinations of Dirac matrices appearing above are linearly independent. and T. In many applications the expectation values of currents are needed. For instance for nucleons one has Γµ (p′ . p′ |V µ (0)|p = u(p′ )γ µ u(p).5 Form factors Currents play an important role in field theory. As an example consider the vector current for a point fermion. Note that the lagrangian density L (x) → L (−x) under Θ. Applying the results from the previous sections it is straightforward to determine the behavior of the combinations under P. C. in this case only q 2 = (p′ − p)2 . s′ |V µ (x)|p. ¯ In general the expectation value between momentum states can be more complicated. F. (8. momenta (pµ or p′µ ) or constant tensors (ǫµνρσ or gµν ).

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

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

|ψS (t) = U (t.5) (9.2) (ii) Heisenberg picture. i ∂ |ψS ∂t ∂ i AS ∂t = H |ψS . QH (t)|q. and the operators are time-dependent. (9. AS (t) = AS and the states o are time dependent. t ′ |q .3) (9. t ′ QH (t′ )|q ′ . [AH . AH (t) = U −1 (t.7) 73 . |ψH (t) = |ψH .1) The time evolution from t0 → t of a quantum mechanical system is generated by the Hamiltonian. Consider the two (time-independent) Heisenberg states: |q. (9. t′ ≡ q ′ |q ′ .1 Time evolution as path integral U (t. t0 ) ∂t (9. in which the operators are time-independent.6) Of these the Heisenberg picture is most appropriate for quantum field theory since the field operators do depend on the position and one would like to have position and time on the same footing. t0 )|ψS (t0 ) . i ∂ |ψH ∂t ∂ i AH ∂t ≡ = 0. t′ . t0 ). t0 ) = H U (t. H]. t . t ≡ q|q.4) ∂ U (t. (9. ≡ 0.Chapter 9 Path integrals and quantum mechanics 9. t0 ) AH (t0 ) U (t. in which the states are time-independent. t0 ) = e−i(t−t0 )H or i Two situations can be distinguished: (i) Schr¨dinger picture. (9.

˙ 2π 1 1 1 [A. t = q ′ |e−i H(t −t) |q . t = = q ′ |e−i n H∆τ |q = q ′ | e−i H∆τ dq1 . |q1 q1 |e−i H∆τ |q . (9. (9. use the Campbell-Baker-Hausdorff formula. B] + . The purpose of this is to calculate the evolution for an infinitesimal time interval.8) Choose t as the starting point with |q = |q.. 2 12 12 (9. tj+1 |qj . .11) (9. ′ (9. tj ≈ = qj+1 |e−i K∆τ e−i V ∆τ |qj dpj qj+1 |pj pj |e−i K∆τ e−i V ∆τ |qj ..14) . Q) expressed in terms of the operators P and Q.Path integrals and quantum mechanics 74 and Schr¨dinger states o |q |q ′ QS |q ≡ q|q . eA eB = eC with C = A+B+ (9. Combining the two terms gives qj+1 . t′ |q. qj ). 2π (9. B] + [A. 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 .10) dqn−1 q ′ |e−i H∆τ |qn−1 qn−1 | . [A. t′ |q.16) this. qj )]) .12) where the transformation between coordinate and momentum space involves q|p = ei p. These can be written in coordinate or momentum representation as Q = P = dq |q q q| dq |q −i ∂ ∂q q| = dp |p p p|.q . . tj = = = 1 For dpj i pj (qj+1 −qj ) −i H(pj . n |q (9. Then qj+1 . tj+1 |qj . H(P.qj )∆τ e e 2π dpj i pj (qj+1 −qj )−i ∆τ H(pj .13) At least for a simple hamiltonian such as consisting of a kinetic energy term and a local potential. .qj ) e 2π dpj exp (i ∆τ [pj qj − H(pj . .9) Dividing the interval from t ≡ t0 to t′ ≡ tn into n pieces of length ∆τ and using completeness (at each time ti ) one writes q ′ . 2π (9. B]. QS |q ′ ≡ q ′ |q ′ . The hamiltonian is an operator H = H(P.15) By letting the kinetic and potential parts act to left and right respectively one can express the expectation values in terms of integrals containing H(pj . t and QH (t) = QS and study the evolution of the system by calculating the quantum mechanical overlap amplitude q ′ . which is a hamiltonian function in which the operators are replaced by real-numbered variables. B]] + [[A. with the correction1 being of order (∆τ )2 .

. . t′ |q. q1 . qN . Before proceeding we also give the straightforward extension to more than one degree of freedom. 9. F3 [α. while Dα F [α] = x dαx N F (αx ) (9. t′ |q1 . α∗ ] = exp (−α∗ K α) ≡ exp − The kernels K(x. dx dy α(x) K(x. ′ ′ q1 .e. i. dx α2 (x).18) t ′ = n=1 Dqn D pn 2π exp i t dτ n pn qn − H(p1 . β] = α K β ≡ 1 F4 [α] = exp − α2 2 F5 [α. q)] ˙ (9. .Path integrals and quantum mechanics 75 or for the full interval q ′ . including differential operators. q)] . . i. . ˙ where Dq and Dp indicate functional integrals. . ≡ exp − 1 2 dx α2 (x) . F [α] ∈ R. Examples are F1 [α] = α ≡ F2 [α] = α2 ≡ dx α(x). y) α(y) . t = τ  dq(τ ) dp(τ ) exp i 2π p exp i 2π t′ t t′ t ∆τ [pq − H(p. qN . pN . 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.2 Functional integrals In this section I want to give a fairly heuristic discussion of functional integrals. . t N (9. dx dy α∗ (x) K(x. . y) β(y). (9. . . qN ) ˙ .20) becomes a multidimensional integral.19) where F [α] is a functional that represents a mapping from a function space F of (real) functions α (R → R) into the real numbers R. . . α(x) → αx and F [α] = F (αx ) changes into a multivariable function.17)  = Dq D dτ [pq − H(p.e. etc. y) in the above examples can be (hermitean) operators acting on the functions. What one is after is the meaning of Dα F [α]. . Two working approaches for functional integration are the following: (i) the heuristic division of the space on which α acts into cells.

23) Having defined the Gaussian integral.24) (9. 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. Physical answers will usually come out as the ratio of two functional integrals and are thus independent of the chosen measure. . Note that procedure (i) is an example of the more general procedure under (ii). In both cases N is an appropriately choosen normalization constant in such a way that the integral is finite.Path integrals and quantum mechanics 76 (ii) Write α in terms of a sum of orthonormal basis functions. (9.28) Dα Dα∗ exp (−α∗ α − α∗ ω − ω ∗ α) = exp (ω ∗ ω) . Consider the gaussian functional as an example. (9.n dx fm (x)fn (x) = n = n dαn 1 exp − α2 N 2 n √ 2π ≡ 1.25) (see Exercises). y) = ∗ m.21) again a multidimensional integral. α(x) = n αn fn and K(x. the following integrals can be derived for a real symmetric or complex hermitean kernel K. with Dα F [α] = n dαn N F (αn ) (9. Using the second method and expanding in a basis set of functions one obtains for the (real) gaussian functional 1 Dα exp − α2 2 = n dαn N exp − 1 αm αn 2 m.n fm (x) Kmn fn (y) and consider F [α] = F (αn ) as a multivariable function. A useful property of functional integration is the translation invariance.27) (9.22) = x The last equality is obtained by defining the right measure (normalization N ) in the integration. Dα F [α] = Dα F [α + β]. N (9.26) As an important application of translation invariance we mention the identities 1 Dα exp − α2 − ωα 2 = exp 1 2 ω . 2 (9. 1 = √ det K 1 Dα Dα∗ exp (−α∗ K α) = det K 1 Dα exp − α K α 2 (α real-valued) (α complex-valued) (9.

(9. Examples of functional differentiation are (9. θ∗ ] = exp (−θ∗ K θ) ≡ exp − Integration for Grassmann variables is defined as dθ 1 = 0 and dθ θ = 1. It is now easy to check that Dθ∗ Dθ exp (−θ∗ K θ) = det K. αn = δmn . products of the same pair appear. −→ fn (x) δα(x) ∂αn ∂αm n Note that δ(x − y) = n ∗ fn (x) fn (y).35) For applications to fermion fields. n θn fn (x) dx θ∗ (x) θ(x) .31) δ δ α = dy α(y) = dy δ(x − y) = 1.32) (9. θη = −ηθ. y) θ(y) . y) β(y) δα(x) δ 1 1 exp − α2 = −α(x) exp − α2 . −→ δα(x) ∆x ∂αx ∂αx δ ∂ ∂ ∗ with . α(y) = δ(x − y). (9.37) We note that from the first to the second line one needs to realize that a pair of different Grassmann variables behaves as ordinary complex variables. δα(x) (9. αy = δxy . θ2 = 0. which vanish. F [θ.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. the Gaussian-type functionals.38) .g. In the expansion of the exponential (from second to third line). however. (9.Path integrals and quantum mechanics 77 Functional differentiation is defined as δ . δα(x) δα(x) δ αKβ = dy K(x.34) (9. dx dy θ∗ (x) K(x.30) (9. i. we need to consider Grassmann-valued functions involving anticommuting Grassmann variables. e. δα(x) 2 2 δ exp (−α β) = −β(x) exp (−α β) .33) (9. δα(x) or in discretized form δ 1 ∂ ∂ with .29) (9. θ∗ ] = exp (−θ∗ θ) ≡ exp − F [θ.e. In principle the definitions of functionals is the same.

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

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

T → i ∞ and T ′ → −i ∞. . T = φ0 (q. T′ Z[J] = lim T ′ → −i∞ T → i∞ Dq exp i T dτ L q. t 0 0out |0in J . δJ(tn ) hence the name generating functional.67) = = φ0 (q. Better is the use ˜ of imaginary time t = −it. J=0 (9. eiE0 T φ∗ (Q) 0 We define the generating functional Z[J] as Z[J] = Q′ .2: Analytic continuation of boundaries T and T ′ We can project out the groundstate by an analytic continuation in the time.71) 9. thus T →i∞ lim q. T J lim −i E0 (T ′ −T ) φ (Q′ ) φ∗ (Q) 0 0 T ′ → −i∞ e T → i∞ dq ′ dq φ∗ (q ′ . . ˜ t′ → ∞.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 ′ ).6 Euclidean formulation The above expression (Eq. t). t′ ) q ′ . . such that t′ → −i∞ t → i∞ ↔ ↔ ˜ t → −∞. t′ |q. . The importance of Z[J] is that the time ordered product of operators can be obtained from it.Path integrals and quantum mechanics t 82 T t t’ T’ Figure 9. . The generating functional precisely represents the vacuum to vacuum amplitude from initial (’in’) to final(’out’) situation in the presence of a source. T ′ |Q.70) one immediately finds δ n Z[J] δJ(t1 ) . t|Q.68) (9.70) for the generating functional is in fact ill-defined. J (9. in which case eiEn T → e−En ·∞ and the term eiE0 T is dominant. 9. Since we have (neglecting multiplicative factors). (9. t) (9. dq dτ + J(τ )q(τ ) . = (i)n 0|T Q(t1 ) . Q(tn )|0 .

. when L q. det K 1 . . i τ −∞ ∞ −∞ dq d˜ τ + J(˜)q(˜) τ τ dq d˜ τ − J(˜)q(˜) τ τ .76) which is a positive definite quantity. dq dt = 1 2 1 2 dq dt 2 Dq exp Dq exp − ∞ d˜ L q. δJ(tn ) = (i)n J=0 1 δ n ZE [J] . ensuring convergence for the functional integral ZE [J]. + V (q). i dq .77) where the expressions have been divided by Z[J] and ZE [J] in order to get rid of dependence on multiplicative factors. . ZE [J] = = where LE q.and Grassmann-valued functions. As discussed in the previous section the interesting quantities are obtained from functional differentiation with respect to the sources. where Kp are the eigenvalues of the matrix Kmn in K(x. dq ˜ dt = 1 2 dq ˜ dt dq ˜ dt 2 − V (q). (9.Path integrals and quantum mechanics 83 In terms of the imaginary time we can write the Euclidean generating functional.75) = − LE q. complex. .74) − V (q) 2 (9. δ n Z[J] 1 Z[J] δJ(t1 ) . ˜ ˜ ZE [J] δJ(t1 ) . τ dq ˜ dt ≡ −L q. det K Exercise 9. (9. e. The differentiations in Z[J] and ZE [J] are related. y) = ∗ m.2 Check the examples of functional derivation for real-. Exercises Exercise 9. δJ(tn ) J = 0 ˜ t = it (9. 1 = √ .72) (9.73) d˜ LE q.n fm (x) Kmn fn (y). Dα Dα∗ exp (−α∗ K α) = for complex-valued functions and Dθ∗ Dθ exp (−θ∗ K θ) = det K. ˜ dt (9.1 Convince yourself by choosing an appropriate (orthonormal) set of functions that 1 Dα exp − α K α 2 for real-valued functions. for Grassmann valued functions. . In all of these cases the determinant of the (hermitean) operator K is defined as det K ≡ p Kp .g.

where dy K(x.80) Dα∗ Dα exp (−α∗ K α − α∗ ω − ω ∗ α) = 1 exp ω ∗ K −1 ω .78) (9. z) = δ(x − z).3 Prove the following relations 1 Dα exp − α K α − αω 2 1 = √ exp det K 1 ω K −1 ω . . y) K −1 (y. 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. 2 (9.79) (9.

e. . i. .3) The Euclidean formulation is as before. . δJ(xn ) 85 ≡ G(n) (x1 .4) J=0 . t′ |φ.1) Dφ exp i σ d4 x L (x) and T′ Z[J] = lim T → i∞ T ′ → −i∞ Dφ exp − Dφ exp i T d4 x [L (φ. x. t) is considered as a set of quantum operators φx (t) in the Heisenberg picture and t′ φ′ . . 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 ). . t = = Dφ DΠ exp i t σ′ ˙ dτ d3 x Π(x)φ(x) − H (φ. φ(x. Π) (10. ∂µ φ) + Jφ] (10. (10. .2) = ∞ −∞ d4 xE [LE (φ. x′ .1 Generating functionals for free scalar fields The generating functional for quantum fields is a generalization of the results in the previous section to a system with more degrees of freedom. φ(xn )|0 = (−i)n δ n Z[J] δJ(x1 ) . ∂µ φ) − Jφ] (10. . .Chapter 10 Feynman diagrams for scattering amplitudes 10.

4) one distinguishes the retarded Green’s function ∆R (x) = d4 k e−i k·x . . xn ) = x2 xn . consider the general solution of ∂µ ∂ µ + M 2 ∆(x) = −δ 4 (x). .e. . .8)  1 d4 x  φ (∂µ ∂ µ + M 2 ) φ − Jφ 2 K  (10. . one immediately sees that i∆F (x1 − x2 ) = δ 2 Z0 [J] (−i)2 Z0 [0] δJ(x1 ) δJ(x2 ) = G0 (x1 − x2 ) ≡ (2) x1 x2 (10. (10. ∆F (the socalled Feynman propagator) is the Green’s function of the Klein-Gordon equation. d4 xn G(n) (x1 . (10.9) i. . 2 2 → (10.11) Depending on the path choosen in the complex k 0 plane (see exercise 7.12) d4 k −i k·x ˜ e ∆(k) (2π)4 (10. J(xn ).3) one finds 1 Z0 [J] = exp − i 2 d4 x d4 y J(x) ∆F (x − y) J(y) ≡ exp (i W0 [J]) . Furthermore. . which can be written as ∆(x) = ˜ with (k 2 − M 2 )∆(k) = 1 or ˜ ∆(k) = 1 1 1 = 0 2 . These Green functions appear in the expansion Z[J] = n in n! d4 x1 .7) where we made the choice Z0 [0] = 1 and ∂µ ∂ µ + M 2 ∆F (x − y) = −δ 4 (x − y).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).13) .5) Z0 [J] for the free scalar field For the (free) scalar field lagrangian L = ← 1 1 ∂µ φ∂ µ φ − M 2 φ2 . . As discussed in the previous section (exercise 9. . xn ) J(x1 ) .10) J=0 In order to determine ∆F . (2π)4 (k 0 + iǫ)2 − E 2 (10. = 0 2 2 2 k2 − M 2 (k ) − E 2 (k ) − k − M (10.Feynman diagrams for scattering amplitudes 86 for which we introduce the picture x1 G(n) (x1 . (10. . . . with Z0 [0] = 1.

the second is to explicitly consider 0|T φ(x1 ) φ(x2 )|0 .4).14) They are solutions of the inhomogeneous equation.15) e−i k·x d4 k . t) = 0 for t > 0. φ(0)] = i∆+ (x) + i∆− (x) d3 k e−i k·x = [φ+ (x).19) φ ∂E ∂E − M 2 φ + Jφ 2 1 d4 xE d4 yE J(x) (−i ∆F (x − y)) J(y) . 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).18) Thus we see various solutions. the generating functional can be written 1 µ µ Z0 [J] = Dφ exp d4 xE (10.17) d4 k e−i k·x . t) = 0 for t < 0.23) k0 =i∞ k0 =−i∞ d4 k e−i k·x . µ 1 1 • Firstly. 4 k 2 − M 2 + iǫ (2π) (10.e. (2π)4 k 2 − M 2 (10. (10.22) (10. The contours for i ∆(x) i ∆+ (x) i ∆− (x) = [φ(x). 10. ∆F = ∆C .16) C where C is a closed contour in the complex k 0 -plane. i. 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. The first possibility is to consider the well-defined Euclidean formulation. . It is straightforward (closing contours in the appropriate half of the complex plane) to prove for instance that i ∆C (x) = θ(x0 ) i ∆+ (x) − θ(−x0 ) i ∆− (x).Feynman diagrams for scattering amplitudes 87 satisfying ∆R (x. φ+ (0)] = (2π)3 2E are also shown in exercise (7.21) ∆F (x) = = −i 0 kE =−∞ d4 kE e−i kE ·xE 2 (2π)4 kE + M 2 (10. the advanced Green’s function ∆A (x) = satisfying ∆A (x. the causal Green’s function ∆C (x) = and finally the Green’s function d4 k e−i k·x = 4 (k 0 )2 − E 2 + iǫ (2π) ∆(x) = PV d4 k e−i k·x . (2π)4 k 2 − M 2 (10. (2π)4 (k 0 − iǫ)2 − E 2 (10. (10. ∆(x) = − d4 k e−i k·x . (2π)4 k 2 − M 2 where the latter contour can be deformed to the contour for ∆C .1 for countours) µ (∂E )2 − M 2 (i ∆F (x − y)) = −δ 4 (x − y) 0 kE =∞ (10. φ− (0)] = (2π)3 2E d3 k e−i k·x = [φ− (x).20) = exp − 2 where or (see fig. The solution of the homogeneous equation ∂µ ∂ µ + M 2 ∆(x) = 0 can also be written as an integral in k-space.

because this is the only quantity entering Z0 [J]. δJ(x4 ) 2 J=0 = − [∆F (x1 − x2 ) ∆F (x3 − x4 ) + ∆F (x1 − x3 ) ∆F (x2 − x4 ) + ∆F (x1 − x4 ) ∆F (x2 − x3 )] . . we use the fact that i∆F (x − y) = G(2) (x − y) = 0|T φ(x)φ(y)|0 to calculate ∆F via the time-ordered product. . = θ(x0 − y 0 ) 0|[φ+ (x). Neglecting multiplicative factors or equivalently assuming that Z0 [0] = 1. x3 .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). φ− (y)]|0 − θ(y 0 − x0 ) 0|[φ− (x). φ+ (y)]|0 = θ(x0 − y 0 ) i ∆+ (x − y) − θ(y 0 − x0 ) i ∆− (x − y) (10.25) x1 x2 x3 = x4 + + . We find (see also section 7. (4) (2) It should be clear that G0 can be expressed in terms of G0 = i ∆F . or diagrammatically (4) (10. x4 ) = 0|T φ(x1 ) φ(x2 ) φ(x3 ) φ(x4 )|0 1 δ4 exp − i J ∆F J = (−i)4 δJ(x1 ) . x2 .24) Knowing the explicit form of Z0 [J] it is straightforward to calculate the 4-points Green’s function.1: Contour in the k 0 plane for the Feynman propagator in Euclidean and Minkowski space • As the second possibility. the same result as above. we have G0 (x1 .Feynman diagrams for scattering amplitudes 0 1 1 0 0 0 1 1 1 0 1 0 0 0 1 0 1 0 11 00 k 1 0 1 0 k 11 00 E 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 -E 1 0 1 0 1 0 1111111111111111111111111111111111111 0000000000000000000000000000000000000 1111111 1 0000000 0 11111111111111111 00000000000000000 11111111111111 00000000000000 11 00 11 00 1 0 1 0 1 0 11111111 1 00000000 0 1 0 1 0 E 1 0 1 0 1 0 1 0 11 00 1 0 1 0 1 0 1 0 1 0 1 0 0 1 1 0 1 0 1 0 1 0 1 0 88 Figure 10.

To zeroth order in the coupling one has 0 1 Z (g ) [J] = Z0 [J] = exp − i 2 J ∆F J .27) When interactions are present.31) G(2. with propagators (i ∆F ) being connected to vertices (i LI ) according to the above expression for the functional Z[J].33) (4) i −i d4 z LI g 4! d4 z 1 δ i δJ(z) 1 exp − i 2 J ∆F J 2 −3 ∆2 (0) + 6i ∆F (0) F + d4 x ∆F (z − x) J(x) 4 d4 x ∆F (z − x) J(x) 1 exp − i 2 J ∆F J . (10. x2 . (2) (0) (10. x4 ) = To first order in g one has Z (g ) [J] = = 1 0 0 x1 + + x2 (10. (10.26) Dφ exp i = exp i d4 z LI 1 δ i δJ(z) d4 x d4 y J(x) ∆F (x − y) J(y) (10.34) Introducing a vertex point to which four propagators are connected. (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 . x2 ) = G0 (x1 . the generating functional can be written. x2 .e.29) This expression will be the one from which Feynman rules will be derived.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. L (φ) = L0 (φ) + LI (φ).Feynman diagrams for scattering amplitudes 89 10.g 0 ) = G0 = 1.32) (10. x3 . x3 .35) z . ≡ −i g d4 z (10. Consider as an example the interaction LI (φ) = − g 4 φ 4! in the scalar field theory discussed sofar. x4 ) = G0 (x1 .g ) (x1 . x2 ) = i ∆F (x1 − x2 ) = G(4.g ) (x1 . Z[J] = = Dφ exp i exp i d4 z LI (10.30) and in zeroth order G(0. i.

. (10. did not contain parts that could be written as products of simpler Green’s functions. x2 ) = c (vacuum bubble) (tadpole) (connected propagator) . . . . J(xn ) c (10. . .41) in terms of socalled connected Green’s functions. (10. d4 xn G(n) (x1 . In the free case we have seen that the exponent of Z[J] contained all essential information. . J=0 (10. . which also was the only Green’s function for which the diagram was connected.42) x1 G(n) (x1 .Feynman diagrams for scattering amplitudes 90 one has J Z (g1 ) [J] = 1 8 1 − 4 J J 1 + 24 J Z0 [J]. 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].e. . . x2 ) = 1 z x1 x2 + 1 2 x1 z x2 . . (10. .36) J J from which one obtains G(0. . .37) G(2.38) The result for the 4-points Green’s function is left as an exercise. .g ) (x1 . (10. J(xn ).39) We have calculated the function up to first order in the coupling constant for φ4 -theroy. . xn ) J(x1 ) . xn ) = x2 c xn Specific examples are G(0) = c G(1) (x1 ) = c G(2) (x1 . i.40) in−1 n! d4 x1 . d4 xn G(n) (x1 . xn ) = (−i)n−1 c which are denoted δ n W [J] δJ(x1 ) . Z[J] = n in n! d4 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]. . G(n) (x1 . . . .g 1 ) = 1 8 z 1 8 . (10. . Diagrammatically this exponent only contained the two-point function. xn ) J(x1 ) . . This remains also true for the interacting case. . . δJ(xn ) . .

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

thus β. in|S|p′ . then (n) (n) Gsc = Gc for n ≤ 3. (10. (2) (10. . In line with the consideration of the generating functional representing the vacuum to vacuum amplitude we consider fields φin . Next. in|S = α. (10. . out = δαβ ↔ SS † = 1. . in = p. out . p′ . φout and the interpolating field φ(x). in = β. . this will be translated to the action on fields. Proof: 0. in = p. (2) The one-particle state is invariant (conservation of energy and momentum. in = |p1 . Proof: α. x4 ) = c 1 x3 z (10. out|p′ .47) . we have W0 [J] = − 1 2 d4 x d4 y J(x) ∆F (x − y) J(y). 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 . out = |p′ .e. . in|SS † |α. x3 . out|α.g ) (x1 . in| (choose ϕ0 = 0). p. For the free scalar theory. Also for free field (Heisenberg) operators a distinction is made between φin and φout . in ⇔ β. out . x1 G(4.44) J=0 For the interacting theory.g. . out|p′ . x2 . in = |α. in|S = β. x4 10. in into final state particle states (out-states) |β. ∂µ ∂ µ + M 2 φin (x) = 0.3 Interactions and the S-matrix The S-matrix The S-matrix transforms initial state free particle states (in-states) |α. one finds for the connected 4-point Green’s function at order g. translation invariance). in|S = 0. (10. out|α.Feynman diagrams for scattering amplitudes 92 i. out (suppressing except momenta all other 1 m quantum numbers). in|p′ . out = p|p′ .43) implying as expected as the only nonzero connected Green’s function G(2) (x1 − x2 ) = (−i) c δ 2 W [J] δJ(x1 ) δJ(x2 ) = G0 (x1 − x2 ) = i ∆F (x1 − x2 ). in = S|α. Sβα ≡ β. pn . the vacuum expectation value of the field φ(x) is zero. out| ⇔ |α.46) The properties of the S-matrix are (1) The vacuum is invariant or |S00 | = 1. e. in = p. out| = eiϕ0 0. (3) S is unitary (it conserves the scalar product from initial to final state). implying the absence of tadpoles. considered explicitly.45) x2 as the only surviving diagram (see exercises). out| and S † |α.

52) where Z is a constant.Feynman diagrams for scattering amplitudes 93 while φ satisfies the inhomogeneous Klein-Gordon equation with the bare mass M0 (this is the mass appearing in the lagrangian L0 ). The time evolution operator (see Eq.50) = φ(x) = √ Zφin (x) − √ Zφout (x) − d4 y d4 z ∆R (x − y) K(y. 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. φin (y)] = iZ ∆(x − y). as it stands.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.62) is then T′ UT [J] T′ = T exp i T d4 x J(x)φ(x) . The relation between S-matrix and in.49) The field φ can be expressed in φin/out using retarded or advanced Green’s functions. (10. The convergence therefore must be weakened to t→−∞ √ (10. 2 ∂µ ∂ µ + M0 φ(x) = J(x). in|Sφout Finally we check that as expected S does not spoil Poincar´ invariance. (2π)3 2E (10. √ Although the above. e. z) φ(z). a)SU −1 (Λ. S is invariant under e Poincar´ transformations: U (Λ.54) .51) (10.and out-fields is: φin (x) = S φout (x)S −1 . The relation between S and Z[J] To establish this relation. a) = S. β. in|φin S → φin S = S φout (x). which in a later stage (renormalization) will become more important. implies the strong (operator) convergence φ(x) → Zφin (x). out|φout = β. 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. (10. i. 9. the source term Jφ in the lagrangian density is treated in the interaction picture. φ(y)] = Z [φin (x). this can actually not be used as it would imply [φ(x).g. (10. Considering plane waves one sees from √ √ 0|φ(x)|p = lim Z 0|φout (x)|p = Z e−i p·x t→∞ √ √ = lim Z 0|φin (x)|p = Z e−i p·x t→−∞ ↔ that identical normalization of (single-particle) plane waves in initial and final state implies the same wave function normalization Z for in and out fields. z) φ(z) d4 y d4 z ∆A (x − y) K(y. φin (x) = d3 k ∗ a(k) fk (x) + a† (k) fk (x) .53) α|φf (t)β −→ Z α|φf (t)β in for normalizable states |α and |β and φf (t) ≡ d3 (x) f ∗ (x) i ∂ 0 φ(x) with f a normalizable solution of the Klein-Gordon equation (wave packet). a causality condition that can be proven to imply the absence of interactions.e. Proof: β.

one has [A. t i Ut∞ [J] φ(x)U−∞ [J] (10. B] is a c-number).61) e φ+ f where the normal ordered expression : e φ f : is used. 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. eB ] = [A. B] and [A. we can. just as we did for φ(x).56) Since δU [J]/δJ(x) satisfies the same equations as φ(x) and we know the limits. S U [J]] = √ Z δ S U [J].e. [A. B] (for the case that [A.Feynman diagrams for scattering amplitudes 94 and satisfies the property T ′ δUT [J] t = i UtT [J] φ(x)UT [J].58) In order to find a solution to this equation note that. with the use of the Baker-Campbell-Hausdorff formula e−B A e+B = A + [A. : e d4 z φ(z) f (z) d4 x φ− (x) f (x) B C B C (10.60) e d4 y φ+ (y) f (y) d4 z φ− (z) f (z) d4 z φ+ (z) f (z) d4 y [φ(x). 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. (10. φ(x). Thus applied to the field φ(x) = φ+ (x) + 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]. provided [A. (10.55) 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). δJ(x) ∞ or for U [J] ≡ U−∞ [J] ′ (10. e e ] = [A. φ(x).57) i [φin (x). : =i d4 y ∆(x − y) f (y) : e d4 z φ(z) f (z) :. φ− f (10. δJ(z) (10. express it in terms of advanced and retarded Green’s functions. z) δU [J] . z) . which is equal to the expression e in which the creation operators are placed left of annihilation operators. z) d4 y d4 z ∆(x − y)K(y. Thus S U [J] =: exp 1 √ Z d4 x d4 y φin (x) K(x. B + C]e e .63) .62) where F [J] is some (arbitrary) functional.59) (10. φ(y)] f (y) e e . y) δ δJ(y) : F [J]. C] are c-numbers. δJ(z) (10. z) δU [J] δJ(z) δU [J] d4 y d4 z ∆A (x − y) K(y. e = i. B]eB .

p′ . . .e. . see exercise 10.66) where Gc is the connected Green’s function. . . fp (x) . c (n) (10. 1 for scalar case.δ. . u(p) and v (p) for fermions and ǫµ (p) for vector fields). . . which is obtained considering only connected diagrams. ¯ ¯ and the amputated or one-particle irreducible (1PI) Green’s functions   n −i  (n) Γ(n) (p1 . k = k2 i − M 2 + iǫ .. (λ) v(p).67) where ∆(p) is the Fourier transform of the (full) propagator ∆F (x).68) 10. . |S|p. fp′ (x′ ) . using that φin (x) = d3 k ∗ a(k) fk (x) + a† (k) fk (x) . . (10.2). ∆(pj ) j=1 (10.. .+ 1 √ Z ∗ d4 x′ . .e. . i. . Explicitly. u(p).. .. . Next we introduce the Fourier transform (after extracting a momentum conserving delta function coming from translation invariance. . .65) we get the Lehman-Symanzik-Zimmerman (LSZ) reduction formula. . . .) iKx . which is determined by the inverse of the operator found in the quadratic term in the lagrangian i. i. J=0 (10. . pn ) =  G (p1 . . . (2π)3 2E n (10. pn ). . iKx′ = (2x′ + M 2 ) precisely annihilating an external propagator i∆F in the Green’s function. iKx′ . y) δ δJ(y) : Z[J] Z[0] . . . . . 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). = . .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. . n (2π)4 δ 4 (p1 + . Sf i = δf i − i (2π)4 δ 4 (Pi − Pf ) Mf i . + pn ) G(n) (p1 . y). x. . . . .e. S =: exp 1 √ Z d4 x d4 y φin (x) K(x. . 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. . . d4 x . Usually we are interested in the part of the S-matrix describing the scattering. . . ←− − → × G(n) (x′ . . pn ) = i=1 d4 xi ei pi ·xi G(n) (x1 . . They start with the propagator (i∆F (k)) in momentum space. .Feynman diagrams for scattering amplitudes 95 while for J = 0 one has U [0] = 1. xn ). Note that the Green’s function connecting the external point xi with the bubble is annihilated by K(x.64) Therefore. .

= −i g . or calculating full Green’s functions external lines are treated as propagators. but overall momentum conservation at a vertex is understood. to be precise iLI vertices in momentum space are introduced. ’t Hooft and M. For the interaction terms in the lagrangian.27. vertices and external particle wave functions in that diagram. For the symmetry factor consider the examples (given in G. Veltman. (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). which is summarized in the following rules: (Rule 1) Start with external legs (incoming particles/outgoing particles) and draw all possible topologically different connected diagrams. Corresponding to external particles wave functions are introduced k k 1 1 In order to calculate the connected amplitude −i Mf i appearing in the S-matrix element these ingredients are combined using Eqs 10. 3! 4! The vertices are: = −i f Consider first the lowest order self-energy diagram. Diagrammar) in the case of the interaction terms g f (10. At these vertices each line can be assigned a momentum. Note that in calculating amputated Green’s functions external lines are neglected.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. which cancels the factor 1/2 in the quadratic piece. = −i g (multiplied with 4! corresponding to the allowed number of permutations of identical particles). for 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. If problems arise go back to the defining expression for the generating function in 10. the factor 2 corresponds to the two-points Green’s function having two identical ends).69) LI (φ) = − φ3 − φ4 .27.

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

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

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

s2 . s3 )γ µ u(k. where u ¯ L(m) µν = = 1 2 u(k ′ . ¯ q2 (10.75) Note that the q µ q ν term in the photon propagator are irrelevant because the photon couples to a conserved current. To lowest order (∝ α = e2 /4π) only one diagram contributes. s′ ) ¯ u s. s1 ) u e2 u(p′ . s′ )γµ u(k. s)γν u(k ′ . The diagram and momenta and the commonly used invariants (Mandelstam variables) for a 2 → 2 scattering process are 1 k 3 k’ s t u = (k + p)2 = m2 + M 2 + 2 k · p = (k ′ + p′ )2 = m2 + M 2 + 2 k ′ · p′ = (k − k ′ )2 = q 2 = 2 m2 − 2 k · k ′ = (p − p′ )2 = 2 M 2 − 2 p · p′ p 2 p’ 4 s = (k ′ − p)2 = m2 + M 2 − 2 k ′ · p + t+u= m2 = 2 m2 + 2 M 2 i = (k − p′ )2 = m2 + M 2 − 2 k · p′ The scattering amplitude is given by −iM = u(k ′ . If we are interested in the scattering process of an unpolarized initial state and we are not interested in the spins in the final state we need |M |2 summed over spins in the final state ( s3 .s4 ) and averaged over spins in the initial state (1/2 × 1/2 × s1 . q 4 µν [using that (¯(k ′ )γµ u(k))∗ = u(k)γµ u(k ′ )].s2 ) which can be written as (see also chapter 4) |M |2 = = 1 4s |¯(k ′ . s2 ).5 Some examples eµ scattering The first example is the electromagnetic scattering of an electron and a muon.Feynman diagrams for scattering amplitudes 100 10. t2 (10. s3 )(−ie)γ µ u(k.78) = 2 α2 2 s + u2 + 4t(M 2 + m2 ) − 2(M 2 + m2 )2 .76) 1 .s4 e4 (m) µν (M) L L . s2 )|2 ¯ q2 (10.77) Combining Lµν and Lµν (M) one obtains L(m) Lµν (M) = 2 s2 + u2 + 4t(M 2 + m2 ) − 2(M 2 + m2 )2 µν and M 4π 2 (m) (10. s)¯(k.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 ) (10.79) . s4 )(−ie)γ ν u(p. s1 ) ¯ −igµν u(p′ .s3 . s4 )γµ u(p.

s3 )γ µ v(p′ . The diagram.s2 3 . (c) Do the same for the connected Green function Gc . s2 ) ¯ ¯ 1 2s u(p. s2 )γµ u(k. . (4) (4) . for instance Møller scattering (e− e− → e− e− ) and Bhabha scattering (e− e+ → e− e+ ) are related using crossing symmetry. To lowest order (∝ α = e2 /4π) only one diagram contributes. s3 ) ¯ ¯ 1 . 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. This is known as crossing symmetry. Exercises Exercise 10.80) Note the similarity in the amplitudes for eµ scattering and e− e+ → µ− µ+ .s4 = 1 2s × v (k ′ .2. . Similarly. s2 )γµ u(k. s4 ) v (p′ . s1 ) u(k.s4 = 1 1 e Tr(/ − m)γµ (/ ′ + m)γν k k Tr(/ + M )γ µ (/′ − M )γ ν p p s2 2 2 1 e4 1 ′ ′ = 4 2 kµ kν + kν kµ − gµν s pµ p′ν + pν p′µ − g µν s s 2 2 M 4π 2 4 and = 2 α2 2 t + u2 + 4s(M 2 + m2 ) − 2(M 2 + m2 )2 . the masses. Basically the same diagram is calculated and the result is the same after the interchange of s ↔ t.Feynman diagrams for scattering amplitudes 101 The second example is the annihilation of an electron pair and creation of a muon pair. s4 ) u e2 v (k ′ .1 (a) Give in diagrammatic notation the full Green functions G(4) (x1 . . x4 ) for the interacting case in φ4 -theory to first order in the coupling constant g as obtained from the full expression for Z[J] in section 10.s2 .s3 . s3 )γ µ v(p′ . s4 )γ ν u(p. s1 )γν v(k ′ . s2 (10. (b) Use the definition of the source-connected Green function Gsc and show that indeed sourceconnected diagrams survive. s1 )|2 ¯ s 1 .

. . (b) Show (by using x1 as shift-varible) that this implies that n i=1 d4 xi ei pi ·xi G(n) (x1 . their sum is zero. xn ). . Exercise 10. . . in the amplitude (Tij = A∗ Aj ). . . xn ) ∝ (2π)4 δ 4 (p1 + .Feynman diagrams for scattering amplitudes 102 Exercise 10.3 Show that the combinatorial factors found using the rules given in section 10. . xn + a) = G(n) (x1 . . . . . or if we in general would consider all momenta as incoming. 1 2 (b) Calculate the quadratic pieces and interference terms.37 and 10. . .2 (a) Show that the translation properties of the fields and the vacuum imply G(n) (x1 + a. . .38. . hence we can write n i=1 d4 xi ei pi ·xi G(n) (x1 . . . . . which means overall momentum conservation in Green functions in momentum space. It is of the form −i M = A1 − A2 . t and u. e(p1 ) + e(p2 ) → e(p′ ) + e(p′ ) in lowest order in α. pn ). + pn ). . . . + pn ) G(n) (p1 . pn ) only has n − 1 independent variables in it. 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). . Exercise 10. |M |2 = T11 + T22 − T12 − T21 . xn ) ≡ (2π)4 δ 4 (p1 + . but the incoming and outgoing momenta are identical. Express the contributions in invariants s.4 (a) Write down the Feynman diagrams contributing to electron-electron scattering. Note that the function G(n) (p1 . . Pictorially this implies. . .3 reproduce for the diagrams precisely the combinatorial factors that appear in Eqs 10. for instance.

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

(11. m2 ).14) One can. ptrf ). 0). a (11. 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. 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.13) (11. 4π s 4π 8π s 4π (11. m2 ) a b . m2 . 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. p1 . p2 ) = = where λ12 denotes λ(s.8) (11. for instance. m2 .10) The function λ(s. Prove that trf Ea = |ptrf | = a s − m2 − m2 a b . mb and s).12) pb = (mb . m2 . 4s (11. Prove that |q| = cm Ea cm Eb (s − m2 − m2 )2 − 4 m2 m2 a a b b = 4s s + m2 − m2 a b √ = . m2 ) = 4(pa · pb )2 − 4p2 p2 . = 2 s λ(s.11) (11. which in the specific case a b also can be expressed as λ(s.16) . m2 . m2 ) is a function symmetric in its three arguments. 2mb λ(s. m2 ) a b . p1 . 2 s s − m2 + m2 a b √ . m2 . Also in this case one can express the energy and momentum in the invariants. In that case trf pa = (Ea . 2mb (11. 1 2 d(E1 + E2 ) √ |q| dΩ(ˆ) q δ( s − E1 − E2 ) 2 (2π) 4(E1 + E2 ) √ q |q| dΩ(ˆ) λ12 dΩ(ˆ) q √ = .9) (11. trf Ea (threshold) = 1 2 mb ( i mi )2 − m2 − m2 a b .15) Explicit calculation of the reduced two-body phase space element gives dR(s.Scattering theory 104 It is straightforward to prove that the unknowns in the particular system can be expressed in the invariants (ma .

pc . where the first is referred to as elastic channel and the set of all other channels as the inelastic channels. (11. pz = mT sinh y. for instance the momenta of incoming (colliding) hadrons and it is useful to use for the produced particle rapidity as a variable. = (11. A specific reaction channel starts contributing at the threshold value (Eq. The variable s is always larger than a c b d the minimal value (ma + mb )2 .19) (11.18) (11. the rapidity y is defined T T x y y= 1 ln 2 E + pz E − pz = ln E + pz mT = tanh−1 pz . 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 .17) (11. E . Examples appear in → π− + p (11.20) → π0 + n → π+ + π− + n → . dt = 2 qq ′ d cos θcm and obtain for the two-body phase space element √ q ′ dΩcm λcd dΩcm √ = (11.27) with m2 = m2 − p2 = m2 + p2 + p2 .. usually many particles are produced.25) dR(s.4). At high energies. Writing E = mT cosh y.Scattering theory 105 Kinematics of 2 → 2 scattering processes π− + p The simplest scattering process is 2 particles in and 2 particles out. there is usually a preferred direction. py ). Consider the 1-particle inclusive case. H1 + H2 → h + X.26) = 16π q s 8π λab Kinematics of inclusive hard scattering processes In high energy (hard) scattering processes. in exclusive measurements all particles are detected. 11.21) (11. Of course there are not only 2-particle channels. while the final state often arises from a series of 2-particle processes combined with the decay of an intermediate particle (resonance). Instead of the scattering angle. The various possibilities are referred to as different reaction 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 . in which one particle is detected. An often used set of invariants are the Mandelstam variables. a c with q = λab /4s and q ′ = being 0 or 180 degrees. 2s (11. . Consider the process a + b → c + d. In inclusive measurements no particles are detected in the final state.23) which are not independent as s + t + u = m2 + m2 + m2 + m2 . pd ) = 4π s 4π 8π s 4π dt dt √ √ . pT = (px .22) (11. usually is a 2-particle state.24) Using the relation between t and cos θcm it is straightforward to express dΩcm in dt. s = (pa + pb )2 = (pc + pd )2 t = (pa − pc ) = (pb − pd ) u = (pa − pd )2 = (pb − pc )2 2 2 (11. tmax min λcd /4s.. however. The initial state.

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

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

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

42) 2−body decay The differential cross section (final state not integrated over) is given by 1 dσ = √ |Mf i |2 dR(s.Scattering theory 109 (Using normalized wave packets these somewhat ill-defined manipulations can be made more rigorous). . p1 . implies for the scattering matrix M . ˆ ˆ Inserted in the unitarity condition for M .46) Partial wave expansion Often it is useful to make a partial wave expansion for the amplitude M (s. (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 . q n ) q n qn √ Mni (q i . t) λab 4π dt. for which one has dσ/dΩ = |f (E. p1 . 11. θ) = −8π s ℓ (in analogy with the expansion for f (E.4 Unitarity condition (S † )f n Sni = δf i The unitarity of the S-matrix. note the sign and cos θ = qi · qf ). . respectively. q f ). pn ). see 11.e. for which the √ amplitudes squared have been calculated in the previous chapter. . p1 . i M† M √ − √ 8π s 8π s = fi n dΩn ∗ Mnf qn Mni √ √ . pn ).47) M (s. i. . . . pn ) |M |2 q 32π 2 m2 dΩ |M |2 . . (11. √ (2ℓ + 1)Mℓ (s) Pℓ (cos θ). The result is N = |Mf i |2 dR(s. . (11. . .43) dΩcm = π M (s.25) −i Mf i − (M † )f i = − ∗ dΩ(ˆn )Mnf (q f .44) This can be used to get the full expression for dσ/dt for eµ and e+ e− scattering. (11. . θ) or M (q i . θcm ) √ q 8π s 2 2 (11. The amplitude −M /8π s is the one to be compared with the quantum mechanical scattering amplitude f (E. θ). 8π s 2π 8π s . The sign difference comes from the (conventional) sign in relation between S and quantummechanical and relativistic scattering amplitude. θ)|2 . (M † )f n (2π)4 δ 4 (Pi − Pn ) Mni . 2 λab and for instance for two particles dσ = q ′ M (s. . q n ). (11. θ) in quantum mechanics. 16π 2 s (11.41) (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 .45) Since the amplitudes also depend on all momenta the full result for two-particle intermediate states is (in CM.

ℓ (11.49) or Im Mℓ = q |Mℓ |2 .52) From the imaginary part of M (s.53) The difference is the inelastic cross section.54) ℓ . the total cross section can be determined.Scattering theory 110 we obtain LHS = −i while for the RHS use is made of ℓ † (2ℓ + 1) Pℓ (ˆi · qf ) (Mℓ )f i − (Mℓ )f i . n (11. which allows writing Mℓ (s) = Sℓ (s) − 1 e2i δℓ (s) − 1 = . (11. θcm ) √ q 8π s 2 .e.51) and for the case that this is the only channel (purely elastic scattering. (11. 2i q 2i q (11. q ˆ i. Using σ= dΩ q ′ M (s. 0). parametrized as Sℓ (s) = ηℓ (s) exp(2i δℓ (s)). In general a given channel has |Sℓ (s)| ≤ 1. (11. Show that σT = = 4π q 2π q2 (2ℓ + 1) Im Mℓ (s) ℓ ℓ (2ℓ + 1) (1 − ηℓ cos 2δℓ ). † −i (Mℓ )f i − (Mℓ )f i = 2 † (Mℓ )f n qn (Mℓ )ni . (11. σinel = π q2 2 (2ℓ + 1) (1 − ηℓ ). in combination with the partial wave expansion for the amplitudes M and the orthogonality of the Legendre polynomials immediately gives for the elastic channel.50) where Sℓ (s) satisfies |Sℓ (s)| = 1 and δℓ (s) is the phase shift. σel = 4π ℓ (2ℓ + 1)|Mℓ (s)|2 (2ℓ + 1) ℓ = 4π q2 ηℓ e 2i δℓ − 1 2i 2 .48) If only one channel is present this simplifies to ∗ ∗ −i (Mℓ − Mℓ ) = 2 qMℓ Mℓ . η = 1) the result σel = 4π q2 (2ℓ + 1) sin2 δℓ . 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 .

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

Essentially this resonance can decay into quark-antiquark pairs or into pairs of charged leptons. Exercises Exercise 11. its width Γ = 120 MeV. it is furthermore easy to prove that the cross section is given by σel = M 2 Γ2 4π (2ℓ + 1) . Limiting ourselves to a resonance in one channel.e. Note that because of the presence of a background the phase shift at resonance may actually be shifted. the full width is Γ = 88 keV. t) + g(u. Its mass is M = 1232 MeV. a fast change in the phase shift for a narrow resonance. u) + f (u.49 GeV. u) = 1 1 2 (s + u2 ) + 4m2 (t − m2 ) t2 2 2 1 1 ( s − m2 )( s − 3m2 ) tu 2 2 . the partial width into e+ e− is Γee = 5.4) can be written as 4πα2 dσ = {f (t. Three famous resonances are: • The ∆-resonance seen in pion-nucleon scattering.26 keV.2 GeV. All these decays can be seen and leave an ’invisible’ width of 498 MeV. At resonance the cross section σT (π + p) is about 210 mb. u) = g(t. The cross section σT (π − p) also shows a resonance with the same width with a value of about 70 mb. • The J/ψ resonance in e+ e− scattering. u) + g(t.Scattering theory 112 (Prove this using the unitarity condition for partial waves).60) showing that the phase shift at resonance rises through δ = π/2 with a ’velocity’ ∂δ/∂s = 1/M Γ. The cross section at resonance is about 30 nb. Three neutrinos explain the resonance shape. 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).1 Show that the cross section for electron-electron scattering (exercise 10. This characteristic shape of a resonance is called the Breit-Wigner shape. Its mass is M = 3096. The phase shift in the resonating channel near the resonance is given by tan δℓ (s) = MΓ . The half-width of the resonance is M Γ. M2 − s (11. where furthermore σinel = 0. which is attributed to neutrinos. t)} .88 MeV. From this one sees that a resonance has the same shape in all channels but different strength. • The Z 0 resonance in e+ e− scattering with M = 91.59) reaching the unitarity limit for s = M 2 . i. one can reconstruct the resonance shape for different numbers of neutrino species. 2 q (s − M 2 )2 + M 2 Γ2 (11. dt s(s − 4m2 ) with f (t. Knowing that each neutrino contributes about 160 MeV (see next chapter). This is a narrow resonance discovered in 1974. Γ = 2.

Calculate now fπ .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∆ . Compare the result with the factor trf 1/ρa ρb va that we used in the derivation of the basic expression for cross sections. As discussed for unstable particles. Compare the results with the experimental π − lifetime τ = 2.2 Calculate the flux factor. we have here already used unitarity to find a finite imaginary part that is proportional to the total decay width. writing N (s) M1 (s) = 2 + iM Γ . Exercise 11. Thus.2).Scattering theory 113 Exercise 11.8. Exercise 11. cm |va − vb | in the center of mass system for two colliding particles a and b.999 877 Γ and Γ(π − → e− νe ) = 0. s − M∆ ∆ ∆ 2 calculate Re N (M∆ ) .000 123 Γ.6 × 10−8 s and the relative decay widths Γ(π − → µ− νµ ) = 0. a quantity which we will ¯ ¯ encounter in the next chapter (section 12. 4. Show that unitarity can be used one step further to fix also the real part of the residue of the pole in the amplitude.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. ρcm ρcm a b 1 cm .

Chapter 12

The standard model
12.1 Non-abelian gauge theories

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

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

La

φ(x) = (1 + i θa La ) φ(x)

inf

(12.2)

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

(12.3)

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

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

The standard model

115

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

i (∂µ U (θ)) U −1 (θ), g

(12.7)

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

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

(12.8)

(12.9) (12.10)

The gauge-invariant lagrangian density is now constructed as

Gµν → U (θ) Gµν U −1 (θ).

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

(12.11)

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

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

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

The standard model
xψ x+dx

116

(x)

ψ(x)

x+dx ψ(x+dx)

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

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

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

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

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

    λ7 =   

i

    −i   

  1  1  λ8 = √   3

−2

A ’constant’ vector that only rotates because of the arbitrary definitions of local frames satisfies Dµ ψ(x) = 0. However. = . In principle such a phase in a given point is not observable. We note that local gauge invariance requires that we can modify all local systems with a (local) unitary transformation S(x). ds s dxµ ds′ P exp ig A (s′ ) ψ(0). 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. (12. What is this physical effect by which the ’connection’ Aµ can be observed? For this consider the phase around a closed loop. requiring that the ’connection’ Aµ (x) is such that Dµ → S(x)D µ S −1 (x).18) or considering a path xµ (s) from a fixed origin (0) to point x. if two different paths to the same point give different phases the effects can be observed. 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). x+dx ψ(x + dx) = 1 + dxµ Dµ x+dx ψ(x). ds which is solved by dψ(s) ds = dxµ .19) This gives rise to a (path dependent) phase in each point.17) The three equations given so far immediately give Dµ = ∂µ − ig Aµ (x). -dx -dy µ µ dyµ µ dx For a vector.e. x x+dx ψ(x + dx) = (1 + dxµ ∂µ ) ψ(x). ∂µ − ig Aµ (x) ψ(x) = 0 dxµ ∂µ − ig Aµ (x(s)) ψ(x) = 0. which connects two identical vectors. (12. i. 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 . i. Dν ] ψ(x) 1 − ig dσ µν Gµν ψ(x).14) (12. x+dx ψ(x) = = 1 + ig dxµ Aa (x)Ta µ 1 + ig dx Aµ (x) µ x x ψ(x) (12. (12.16) which in the presence of the ’connection’ Aµ differs from the total change between x ψ(x) and x+dx ψ(x+ dx). but expresses them with respect to different bases. (12. The relation in Eq.The standard model the basis for x G and x+dx 117 G. should be independent of such transformations.16. 12.e.15) ψ(x).

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

(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. . Any constant field with ’length’ F is a possible groundstate. It is sufficient to do this to second order in the fields as the higher (cubic. Writing the field φ as a sum of a classical and a quantum field. etc. i. while the lagrangian including the nonzero groundstate expectation value chosen by nature. It is only invariant under rotations around the 3-axis. has less symmetry. λ such that 0|ϕ1 |0 = 0|ϕ2 |0 = 0|η|0 = 0.21) Stability of the action requires the classical groundstate ϕc to have a well-defined value (which can be nonzero). .   ϕc = 0|φ |0 =  (12. . It is appropriate to redefine the field as    ϕ1      φ =  ϕ2  . The presence of a nonzero value for the classical groundstate value of the field will have an effect when the field is quantized. The classical groundstate appears degenerate. In order to see the consequences for the particle spectrum of the theory we construct the lagrangian in terms of the fields ϕ1 . They are known as the Weyl mode or the Goldstone mode: Weyl mode. corresponding to the number of broken generators (in this case rotations around 1 and 2 axis) and 1 massive scalar particle with mass m2 = −2m2 . The original lagrangian contained an SO(3) invariance under (length conserving) rotations among the three fields.) terms constitute interaction terms. In this situation one speaks of spontaneous symmetry breaking. for the vacuum Qa |0 = 0. therefore a choice must be made. say    0    0 . φ = ϕc + φquantum where for the (operator-valued) coefficients in the quantum field one wants 0|c† = c|0 = 0.24) 1 2 2 4 1 1 1 = (∂µ ϕ1 )2 + (∂µ ϕ2 )2 + (∂µ η)2 + m2 η 2 + . The field along the third axis plays a special role because of the choice of the vacuum expectation value. 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. while the quadratic terms must correspond with non-negative masses. ϕ2 and η. In this case the spectrum is described .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. The η massless particles are called Goldstone bosons. A quantum field theory has only one nondegenerate groundstate |0 .25) 2 2 2 Therefore there are 2 massless scalar particles.e. so one has 0|φquantum |0 = 0 and hence 0|φ|0 = ϕc . (12. (12. In this mode the lagrangian and the vacuum are both invariant under a set of symmetry transformations generated by Qa .23)     F +η the latter only forming a minimum for m2 < 0. The situation now is the following. In the case of degeneracy.

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

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

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. . .    0      ϕc = 0|φ |0 =  0  .    0 φ= (12. thus 9 independent degrees of freedom.The standard model 122 12.38) ∂µ Wν − ∂ν Wµ + g Wµ × Wν . as in this case there are no massless Goldstone bosons. 1 1 L = − Gµν · Gµν + Dµ φ · Dµ φ − V (φ). having two independent spin components) and three scalar fields (one degree of freedom each).40) a Dµ φ = ∂µ φ − ig Wµ La φ. 2 massive vector fields or spin 1 bosons (2 × 3) and one scalar field (1). The difference comes when we reparametrize the field φ. After symmetry breaking the same number (as expected) comes out. made into a gauge theory.41)         φ3 F +η Explicitly one then has  2 2  gF Wµ + g Wµ η   1 1  −gF Wµ − g Wµ η Dµ φ = ∂µ φ + g Wµ × φ =   ∂µ η     .     0  0         0 = . while the number of massive gauge bosons coincides with the number of ’broken’ generators. like a photon. Initially there are 3 massless gauge fields (each. i.     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 + . (12. Dµ φ = Gµν = ∂µ φ + g Wµ × φ. Therefore we can always rotate the field φ into the 3-direction in order to simplify the calculation.39) (12. One may wonder about the degrees of freedom. 4 2 where Since the explicit (adjoint. Considering the same example with rotational symmetry (SO(3)) as for spontaneous symmetry breaking of a scalar field (Higgs field) with three components. . . We have the possibility to perform local gauge transformations. two massive bosons (Wµ and Wµ with MW = gF ) and a massive scalar particle (η with m2 = η 2 −2 m . 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). but one has 1 massless gauge field (2). in this case three-dimensional) representation reads (La )ij = −i ǫaij . one sees that the fields Wµ and Gµν also can be represented as three-component fields.37) (12.   The symmetry is broken in the same way as before and the same choice for the vacuum.42) 2 from which one reads off that the particle content of the theory consists of one massless gauge boson 3 1 2 (Wµ ).e. The latter is a spin 0 particle (real scalar field) called a Higgs particle. (12. again 9 degrees of freedom. is made. (12.

(12.) and the gauge bosons responsible for the strong. etc. that gives a good description of the physical world. the leptons (elektrons. muons. down. etc. In the standard model one starts with a very simple basic lagrangian for (massless) fermions which exhibits more symmetry than observed in nature. U(1)Y U(1)Y (12.and righthanded particles.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. R. = ∂µ L − (12. Thus the lagrangian density is L (f ) ∂ ∂ = i L/ L + i R/ R. under which the lefthanded and righthanded particles with YW (L) = −1 and YW (R) = −2 transform with eiβYW /2 . implies certain nontrivial relations between masses and mixing angles that can be tested experimentally and sofar are in excellent agreement with experiment. explicitly L R SU(2)W SU(2)W −→ ei α·τ /2 L. neutrinos. ∂ ∂ ∂ (12.47) and YW is considered as an operator that generates a U (1)Y symmetry.44) which has an SU (2)W symmetry under transformations eiα·TW . For the neutrino only a lefthanded particle (and righthanded antiparticle) exist.48) (12. νµ and ντ ) and their antiparticles. ψR/L = 2 (1 ± γ5 )ψ decouple. 2 2 = ∂µ R + i g ′ Bµ R.45) (12. explicitly Dµ L Dµ R i i g Wµ · τ L + g ′ Bµ L.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.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. τ ).The standard model 123 12. however.).e. (12. 2 (12. By introducing gauge fields and breaking the symmetry a more complex lagrangian is obtained.51) . its corresponding neutrino (νe . 1 As they are massless.50) (12. The lagrangian for the leptons consists of three families each containing an elementary fermion (electron e− . explicitly L −→ e−i β/2 L. muon µ− or tau τ − ). Thus L (f ) = i eR / eR + i eL / eL + i νeL / νeL + (µ. This is written as 3 Q = TW + YW . i. R −→ e−i β R.   0   1 +1/2  L   νe  3   − = with TW = and TW = L=    −1/2 eL L 2 R = R − = eR 3 with TW = 0 and TW = 0. while the lefthanded particles form a doublet. left.43) and One introduces a (weak) SU (2)W symmetry under which eR forms a singlet. electromagnetic and weak forces. The procedure. the quarks (up.

  h)  (12. + 8 (12. .The standard model 124 3 3 where Wµ is a triplet of gauge bosons with TW = 1. . Using local gauge invariance θi for i = 1. 2 and 3 may be eliminated (the necessary SU (2)W rotation is precisely e−iθ(x)·τ ).55) 2 2 The Higgs potential V (φ) is choosen such that it gives rise to spontaneous symmetry breaking with ϕ† ϕ = −m2 /2λ ≡ v 2 /2. leading to the parametrization   1  0    φ(x) = √  (12.52) 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 . which assures with the choice of YW for the Higgs field that Q generates the remaining U (1) symmetry. . the U (1)Q symmetry (generated by the charge operator). .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. −V (φ) L and (h2) = −Ge (LφR + Rφ L). (12. 4 4 In order to break the symmetry to the symmetry of the physical world. this leads to the lagrangian L (h1)     . 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. with the following lagrangian density consisting of a symmetry breaking piece and a coupling to the fermions. + 8 g2 v2 1 + − (Wµ )2 + (Wµ )2 (∂µ h)2 + m2 h2 + 2 8 (g 2 + g ′2 ) v 2 (Zµ )2 + .53) φ=   √ (θ4 − iθ3 )  φ 2 with TW = 1/2 and YW = 1 is introduced.54) = (Dµ φ)† (Dµ φ) −m2 φ† φ − λ (φ† φ)2 . (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. † i i g Wµ · τ − g ′ Bµ )φ. . Putting this in leads to L (f ) =L (f 1) +L (f 2) . L where L (h1) (h) =L (h1) +L (h2) .58) (12. a complex Higgs field   +   1  φ   √2 (θ2 + iθ1 )    0 = 1   (12. For the classical field the choice θ4 = v is made.59) .

63) (12.68) g2 ρ 5 2 (νµ γρ (1 − γ5 )µ) (eγ (1 − γ )νe ) 8 MW (µ)† (jL )ρ (e) (jL )ρ GF (µ)† (e) ≡ i √ (jL )ρ (jL )ρ . g ′ /g = tan θW . 2 Zµ (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 + . 2θ 2θ 4 cos W cos W 0. 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 .62) g 2 + g ′2 3 ≡ sin θW Wµ + cos θW Bµ . 2 (12. we can read off e = g sin θW = g ′ cos θW .The standard model 125 where the quadratically appearing gauge fields that are furthermore eigenstates of the charge operator are ± Wµ = = = Zµ Aµ 1 1 2 √ Wµ ± i Wµ . 4 2 MW g2 v2 = . 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 .66) From the coupling to the photon. .60) (12.69) . 2 3 g Wµ − g ′ Bµ 3 ≡ cos θW Wµ − sin θW Bµ . (12.61) (12.64) (12.67) The coupling of electrons or muons to their respective neutrinos. . σ (νµ γ ρ µL ) 2 2 (eL γ νe ) 2 k + MW (12. (12. 2 + g ′2 g 3 g ′ Wµ + g Bµ (12.65) The weak mixing angle is related to the ratio of coupling constants.

leptons with CV = −1/2 + 2 sin2 θW ≈ −0.71) GeV −2 = = = = .05 and CA = −1/2 we calculate Γ(e+ e− ) ≈ 78.5 MeV (exp.79) Γ(Z 0 → f f ) = 48π cos2 θW For the electron.74) (12.75) The coupling of the Z 0 to fermions is given by (g/ cos θW )γ µ multiplied with 3 TW 1 1 1 (1 − γ5 ) − sin2 θW Q ≡ CV − CA γ5 .73) (12.e. me /v is extremely small.76) with CV CA 3 = TW − 2 sin2 θW Q. g ′ and v determine a number of experimentally measurable quantities. 2 The value of α deviates from the known value α ≈ 1/137 because of higher order contributions.166 4 × 10 0.The standard model 126 the good old four-point interaction introduced by Fermi to explain the weak interactions.231 2. Γinvisible = 480 MeV one sees that three families of (light) neutrinos are allowed. muon or tau. one obtains L (h2) Ge v Ge = − √ (eL eR + eR eL ) − √ (eL eR + eR eL ) h 2 2 me = −me ee − eeh. i.40 GeV. the mass of the electron comes from the spontaneous symmetry breaking but is not predicted (it is in the coupling Ge ). one has the relation g2 e2 1 GF √ = . from the MW mass is about 250 GeV.70) = 2 = 2 sin2 θ 8 MW 2 v2 8 MW 2 W In this way the parameters g. 3 = TW .77) (12. With the choosen vacuum expectation value for the Higgs field. (12. The coupling to the Higgs particle is weak as the value for v calculated e. 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. such as2 α = e2 /4π GF sin2 θW MW MZ ≈ 1/127. giving rise to a running coupling constant after renormalization of the field theory. i. g2 MZ 2 2 (CV + CA ).e. For each neutrino species (with CV = 1/2 and CA = 1/2 one expects Γ(νν) ≈ 155 MeV. 1. v (12. The masses of the fermions and the coupling to the Higgs particle are contained in L (h2) . (12. −5 (12. 2 2 2 (12.78) From this coupling it is straightforward to calculate the partial width for Z 0 into a fermion-antifermion pair.80) First. (12. . Γe ≈ Γµ ≈ Γτ ≈ 83 MeV). as appear for instance in the decay of the neutron or the decay of the Λ (quark content uds). 91.72) (12. Comparing this with the total width into (invisible!) channels. 80.g. (12.19 GeV. Finally we want to say something about the weak properties of the quarks.

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

84) with Gu and Gd being real and positive and Wu and Wd being different unitary matrices.83) where Vu and Vd are unitary matrices.q) ′ weak † mass mass DL = DL = Vu Vd DL = VCKM DL .227.85) √ √ with Mu = Gu v/ 2 (diagonal matrix containing mu . . (12. A ≈ 0. e. Note that we need the conjugate Higgs field to get a U (1)Y singlet in the case of the charge +2/3 quarks.g. for which we need the appropriate weak isospin doublet  0∗    1    φ c  = √  v +h . magnitudes of the entries in the CKM matrix are nicely represented parametrization  1 − 1 λ2 λ λ3 A(ρ − i η)  2   1 2  −λ 1− 2λ λ2 A V =  3 2 λ A(1 − ρ − i η) −λ A 1 12. The Λu and Λd are complex matrices in the 3×3 family and Λd Λ† = Vd G2 Vd† .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. i. (12. The Higgs field is still limited to one complex doublet. mc and mt ) and Md = Gd v/ 2 (diagonal matrix containing md . UR = uR cR tR . One then reads off that starting with the family basis as defined via the left doublets that the mass eigenstates (and states coupling to the Higgs field) involve the righthanded † mass † mass mass † mass states UR = Wu UR and DR = Wd DR and the lefthanded states UL = Vu UL and DL = † Vd DL . the three righthanded quarks with charge +2/3 in UR and the three righthanded quarks with charges −1/3 in DR . respectively). the weak mass eigenstates are L (h2.The standard model 128 using the socalled Wolfenstein      + O(λ4 )   with λ ≈ 0. Thus one has † † † =⇒ −DL Vd Md Wd DR − DR Wd Md Vd† DL − UL Vu Mu Wu UR − UR Wu Mu Vu UL (12.   φ =    −φ− 0 2 For the (squared) complex matrices we can find positive eigenvalues.q) = −QL φΛd DR − DR Λ† φ† QL − QL φc Λu UR − UR Λ† φc† QL u d (12. u u where we include now the three lefthanded quark doublets in QL . 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 . allowing us to write † Λu = Vu Gu Wu † and Λd = Vd Gd Wd .86) the unitary CKM-matrix introduced above in an ad hoc way.34. ms and mb ). each of these containing   the three families. d d (12.22 and η ≈ 0.82 and ρ ≈ 0.82) space. † Λu Λ† = Vu G2 Vu .e. The imaginary part i η gives rise to CP violation in decays of K and B-mesons (containing s and b quarks.

† † Again we continue as before now with matrices Λe = Ve Ge We and Λn = Vn Gn Wn . As before. while θ23 then is determined by atmospheric neutrino oscillations requiring |∆m2 | ≈ 2.ℓ) † † † =⇒ −EL Ve Me We ER − ER We Me Ve† EL − NL Vn Mn Wn NR − NR Wn Mn Vn NL . (12. it is particularly useful because θ12 is essentially determined by solar neutrino oscillations requiring ∆m2 ≈ 8 × 10−5 12 eV2 (convention m2 > m1 ).87) is a weak doublet containing the three families of neutrinos (NL ) and charged leptons (EL ) and ER is † a three-family weak singlet. The mass fields ER † mass † mass = We ER . e   N L= L  EL     (12. For the (massless) neutrino fields we just can redefine fields into NL = Ve† NL .The standard model 129 The lepton sector (massless neutrinos) For a lepton sector with a lagrangian density of the form L in which (h2. The W -current then becomes mass mass E L γ µ NL = E L γ µ NL . EL = Ve EL . The lepton sector (massive Dirac neutrinos) In principle a massive Dirac neutrino could be accounted for by a lagrangian of the type L (h2. NL = Vn NL and NR = Wn NR µ mass γ µ V † V N mass . decoupling from all known interactions.91) † with UPMNS = Vn Ve known as the Pontecorvo-Maki-Nakagawa-Sakata mixing matrix. we find massless neutrinos. i.89) with three righthanded neutrinos added to the previous case.e.5×10−3 eV2 .90) mass † mass mass † mass † We note that there are mass fields ER = We ER . EL = Ve† EL . In this case. 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 . (12.  =  − 1/6 2/3 0           1/2 1/2 0 0 1 0 1/3 1/2 − 1/6 (12. The mixing is intriguingly close to the Harrison-Perkins-Scott tri-bimaximal 23 mixing matrix      0 0 2/3 1/3 0 2/3 1/3 0     1                 1/2 − 1/2   − 1/3 UHPS =  0 1/3 − 1/2  . (12. and obtain L (h2. UPMNS a parametrization that in principle also could have been used for quarks.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    .ℓ) = −LφΛe ER − ER Λ† φ† L. there is no family mixing for massless neutrinos. since the weak current is the only place where they show up. mµ and mτ . one can write Λe = Ve Ge We and we find † L (h2.    (12.ℓ) =⇒ −Me EL Ve We ER − ER We Ve† EL .92) .ℓ) = −LφΛe ER − ER Λ† φ† L − Lφc Λn NR − NR Λ† φc† L e n (12. For neutrino’s this matrix is parametrized in terms of three angles θij with cij = cos θij and sij = sin θij and one angle δ.88) √ mass with Me = Ge v/ 2 the diagonal mass matrix with masses me .

α2 and δ).96) VPMNS = UPMNS K with K =  0 eiα2 /2 0  . Actually. 2 R ψ = nR + nL . This is called the see-saw mechanism (outlined below). The see-saw mechanism Consider (for one family N = n) the most general Lorentz invariant mass term for two independent Majorana spinors. a unitary matrix relating them to the weak eigenstates. For these light Majorana neutrinos one has. Absorption of phases in the states is not possible for Majorana neutrinos. 1 c ∗ c L mass. For the charged leptons there cannot exist a Majorana mass term as this would break the U(1) electromagnetic symmetry. (12.The standard model 130 The lepton sector (massive Majorana fields) An even simpler option than sterile righthanded Dirac neutrinos. however.and lefthanded species through Dirac mass terms coming from the coupling to the Higgs sector as in the previous section. are equivalent to two decoupled Majorana neutrinos (see below). one being massive. For the leptons. 12. 1 ∗ c c (12.and righthanded species then just form a Dirac fermion. The mass matrix can be written as   ML |MD | eiφ     M = (12.95) 2 In order to have more than a completely decoupled sector.100)  |MD | eiφ MR 1 1 ∗ ∗ M L nc nL + M L nL nc − M R nR nc + M R nc nR L R L R 2 2 1 1 ∗ ∗ − M D nc nc + M D nL nR − M D nR nL + M D nc nc L R R L 2 2     1  nc nR   M L M D   nL   L   c  + h. Υ′ and Υ′ (satisfying Υc = Υ and as discussed in chapter 6.99) . Υc ≡ (ΥL )c = ΥR 1 2 L and Υc = ΥL ). For the neutrino sector.ν =− containing three (CP-violating) phases (α1 . 2 although this option is not attractive as it violates the electroweak symmetry. one must for the neutrinos as well as charged leptons. hence the mixing matrix becomes   iα /2 0 0   e 1     (12. coupled by a Dirac mass term.88 a Majorana mass term for the (lefthanded) neutrino mass eigenstates.ν = − MR NR NR + MR NR NR . Thus 2 Υ ′ = nc + nL . couple the right. Thus (disregarding family structure) one has two Majorana neutrinos. with for the righthanded sector a mass term MR . as above. 1 L Υ ′ = nR + nc .c. the massless and massive Majorana neutrinos.97) As the most general mass term in the lagrangian density we have L mass = − = which for MD = 0 is a pure Majorana lagrangian and for ML = MR = 0 and real MD represents the Dirac case. We use here the primes starting with the weak eigenstates. If the Majorana mass MR ≫ MD one actually obtains in a natural way one Majorana neutrino with a very small mass. −    MD MR nR 2 (12. The way to circumvent this is to introduce as in the previous section righthanded neutrinos.94) M L NL NL + M L NL NL .98) (12. it is easy to see R that this incorporates the Dirac case by considering the lefthanded part of Υ′ and the righthanded 1 part of Υ′ as a Dirac spinor ψ. the left. (12. is to add in Eq.       0 0 1 L mass.

related via    c   Υ1R     = U  nL  . The diagonalization is analogous to what was done for the Λ-matrices and one finds U M U T = M0 with a (unitary) matrix U . T ] ψ. 2 U (M M † ) U † = M0 . which implies U ∗ M † U † = U ∗ M ∗ U † = M0 and a ’normal’ diagonalization of the (hermitean) matrix M M † .  (12. i ψ {M. Exercise 12.3 (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 .103) for each of which one finds the lagrangians L = and we are left with two decoupled Majorana fields      Υ1L    = U ∗  nc  . a and a′ are degenerate states.1 Consider the case of the Weyl mode for symmetries. This is a mixing problem with a symmetric (complex) mass matrix leading to two (real) mass eigenstates.  L      Υ2L nR Υ1 and Υ2 . Prove that if the generators Qa generate a symmetry.2 Derive for the vector and axial vector currents.105) for i = 1.The standard model 131 taking ML and MR real and non-negative. (12. and |a and |a′ belong to the same multiplet (there is a Qa such that |a′ = Qa |a ) then H|a = Ea |a implies that H|a′ = Ea |a′ .e. For the situation ML = 0 and MR ≫ MD (taking MD real) one 2 finds M1 ≈ MD /MR and M2 ≈ MR . i. i.e. V µ = ψγ µ T ψ and Aµ = ψγ µ γ5 T ψ ∂µ V µ ∂µ A µ = = i ψ [M.104) ↔ 1 1 Υi i / Υi − Mi Υi Υi ∂ 4 2 (12. 2 with real masses Mi . H] = 0. [Qa .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   . (12. This choice is possible without loss of generality because the phases can be absorbed into Υ′ and Υ′ (real must be replaced by hermitean if one includes 1 2 families).      Υ2R nR (12. Exercises Exercise 12. T } γ5 ψ. Exercise 12.102) = 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)) . .

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

The standard model 133 (a) Calculate for the special choice ML = MR = 0 and MD real. 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 . . Give the explicit expressions that relate ψ and ψ c with Υ1 and Υ2 . (b) A more interesting situation is 0 = ML < |MD | ≪ MR .05 eV. Given that neutrino masses are of the order of 0. which leads to the socalled see-saw mechanism.

Sign up to vote on this title
UsefulNot useful