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. 1995.3 2.3 3. These notes Section 1. An introduction to Quantum Field Theory. M. Quantum Field Theory.4 2.3 3. Cambridge University Press. 1995.4 3. Zuber. 1994. Peskin and D.2 1. S.1 2.4 3. but are rather compact.5 Ryder Peskin & Schroeder Srednicki 1 1 1 1 3 2.2 2.3 References As most directly related books to these notes. Cambridge University Press.1 1.-B. 1. Diagrammatica.1.4 6. II: Modern Applications. Veltman. 2.3 6. Quantum Field Theory. Cambridge University Press. Corresponding chapters in books of Ryder. Srednicki. 2.2 3.3 2.2 3. I refer to the book of Srednicki [1] and Ryder [2]. M.2 36 22 .5 2.1 3. 4. Weinberg.8 3. 1996.2 6.H. 6.2 3.1 5.3 3. 5.2 3. Cambridge University Press.7 2.3.3 2. I: Foundations. Peskin & Schroeder and Srednicki. 2007.E. The notes contain all essential information.1 6.2 6. Schroeder.1 2 2 3.3 2. 1985.3 2.1 3.5 2. 1980. 3.4 4. C. Addison-Wesly.3 4. McGraw-Hill. 3. Itzykson and J.2.2 4.4 3.2 2. Vol. M. The quantum theory of fields. Vol.1 2.3 3. L.3 3.4 2. Other text books of Quantum Field Theory that are useful are given in refs [3-6]. Quantum Field Theory.2 3.5 4. Ryder.V.5 5. Cambridge University Press.2 2.1 4. The schedule in the Fall of 2011 is Chapters 1 through 10 in period 2 (7 weeks in November and December 2011) and Chapters 11 and 12 in period 3 (January 2012).

2 8.5 7.2 5.2.1 7.5 7.4 3.1 20.1 20.7 4.5 11.3 6 8 9 Srednicki 3 3 3 37 5.1 9.4 8.2.3 8.5 4. 20.5 .7 5.1 12.1 8.4 These notes Section 7.5 9.8 5 4.2 11.1 10.1 9. 6. 2.3 8. 6. 9.4.3 11. 2.5. 4.6 4.1 20.3.3 2.6 6. 6.3 7.6 3.3 20.3.7.4 12.3.5 Ryder 4 4 4 4 4 4 Peskin & Schroeder 2.6 9.3 10.2 8.4 12.4 4.5 6. 4.6 3.1.1.5 6 7 6 6.4 10.3 3.2.6 3.2.4 9.6 10.5 4.8 6.2 9. 6.4.4 2. 6.5 2. 4.1 5.1 11. 4.2 10.5 9.10 8.4 7. 9.2 12.4.1.6 8.3 6.2 4.2 7.3 9. 8.3 12.3 7.

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

(1. ≡ h/2π = 1.6) . In quantum mechanics. the position can in principle be determined at any time with any accuracy.5) One can also choose a representation in which the momenta of the particles are the dynamical variables. (1. S ≫ .1) In the limit that the action S is much larger than .1 Quantum field theory In quantum field theory the theories of quantum mechanics and special relativity are united. pj ] = i δij . ˜ ψ(p) = i d3 r exp − p · r ψ(r). One can talk about states |r and the wave function ψ(r) = r||ψ . 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. It is given by pop ψ(r) = −i ∇ψ(r). 1 (1. (1. In this coordinate representation the position operators r op simply acts as rop ψ(r) = r ψ(r). In quantum mechanics a special role is played by Planck’s constant h. (1.3) The uncertainty principle tells us that in this representation the momenta cannot be fully determined. (1.2) In the limit that v ≪ c one reaches the non-relativistic domain. In special relativity a special role is played by the velocity of light c. ˜ The corresponding states are |p and the wave functions ψ(p) = p||ψ are the Fourier transforms of the coordinate space wave functions.582 119 15 (56) × 10−22 MeV s.054 571 68 (18) × 10−34 J s = 6.Chapter 1 Introduction 1. the action of the momentum operator in the coordinate representation is not as simple as the position operator. Corresponding position and momentum operators do not commute. Indeed.4) where δij is the Kronecker δ function. quantum effects do not play a role anymore and one is in the classical domain. They satisfy the well-known (canonical) operator commutation relations [ri . c = 299 792 458 m s−1 . being eigenvalues of the position operators. 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.

Vectors are denoted x = xµ nµ . The (invariant) lengths often have special names. . x3 ). it is convenient to distinguish lower indices from upper indices. The scalar product of two different vectors x and y is denoted x · y = xµ y ν gµν = xµ yµ = xµ y µ = x0 y 0 − x · y. (1. Useful relations are ǫijk ǫℓmn ǫijk ǫimn ǫijk ǫijl = = = δiℓ δjℓ δkℓ 2 δkl . Unlike in Euclidean space.νn can be conν structed. x2 = x · x = xµ xν nµ · nν = xµ xν gµν . We note that for Euclidean spaces (with a positive definite metric) vectors and tensors there is only one type of indices.. One can distinguish: • ds2 > 0 (timelike intervals).2. linear transformations that leave the length of a vector invariant are called (homogeneous) Lorentz transformations. xi are the three space components. are . ˆ ˆ (1.. • ds2 < 0 (spacelike intervals). the points lie on the lightcone and they can be connected by a light signal.µn = Λµ1ν1 .26) (1. δim δjm δkm δin δjn δkn .25) Within Minkowski space the real. one must distinguish tensors with upper or lower indices and one can have mixed tensors. denoted as V ′µ = Λµν V ν . In Minkowski space. such as eigentime τ for the position vector τ 2 ≡ x2 = t2 − x2 . The length (squared) of a vector is obtained from the scalar ˆ product. x1 . In special relativity we start with a four-dimensional real vector space E(1. given by g00 = −g11 = −g22 = −g33 = 1 (the ˆ ˆ other components are zero). Many other four vectors and tensors transforming like T ′ µ1 .21) (1.3) with basis nµ (µ = ˆ 0. .. in which case zi = ǫijk xj yk . Relations relating tensor expressions. where the coordinate t = x0 is referred to as the time component. One has x2 = xµ xµ = t2 − x2 . x3 ) = (t. (1. independent of a coordinate system. When 3-dimensional space is considered as part of Minkowski space. No difference is made between upper or lower. x2 . we will use upper indices for the three-vectors. The real. x2 . The lower indices are constructed in the following way.22) (1. For four-vectors in Minkowski space we will use the notation with upper indices and write x = (t. and are given by (x0 .3). x) = (x0 ..Introduction 5 that can be used in the cross product of two vectors z = x × y. x1 . however. Because of the different signs occurring in gµν .24) The quantity gµν ≡ nµ · nν is the metric tensor. linear transformations that do not change the length of a four-vector are called the Lorentz transformations. −x). • ds2 = 0 (lightlike or null intervals). 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 . These transformations do change the components of a vector. The Poincar´ transformations include e e Lorentz transformations and translations.1. The invariant distance between two points x and y in Minkowski space is determined from the length dsµ = (x − y)µ . So we could have used all upper indices in the above equations. xµ = gµν xν . 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 .23) δjm δkn − δjn δkm . 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. the invariant length or distance (squared) is not positive definite. Λµn n T ν1 .

32) . (1. p) = (E. ∂1 .∇ .34) (1. Each of these permutations leads to a minus sign. (1. explicitly written as (p0 . a∞ = 4πǫ0 2 /me e2 . Examples are the scalar products above but also relations like pµ = m dxµ /dτ for the momentum four vector. starting from ǫ0123 = 1. g0 = g1 = g2 = g3 = 1. It is an invariant tensor.30) (Note that there are different conventions around and sometimes the opposite sign is used). . (1. = ∂ . ∂3 ) = ∂ ∂ ∂ ∂ . gµ = gµρ g ρν and 0 1 2 3 is in essence a ’Kronecker delta’. The length squared of ∂ is the d’Alembertian operator.35) ρ ǫµνρσ ǫµν σ σ ǫµνρσ ǫµνρ ′ ′ = −2 g ρρ g σσ − g ρσ g σρ = −6g σσ . 0123 0123 ′ ′ ǫ µνρσ ǫµνρσ The first identity. For the contraction of a vector with the antisymmetric tensor one often uses the shorthand notation ǫABCD = ǫµνρσ Aµ Bν Cρ Dσ . . ∂t (1. ′ ′ .31) ǫµνρσ ǫµν ′ ′ ρσ ′ = − (1. is timelike with invariant length (squared) p2 = p · p = pµ pµ = E 2 − p2 = m2 . Relate this quantity to the electron Compton wavelength −e via the dimensionless λ fine structure constant α. p). 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. The derivative ∂µ is defined ∂µ = ∂/∂xµ and we have a four vector ∂ with components (∂0 .36) Exercises Exercise 1. where m is called the mass of the system. constructed via the metric tensor itself.29) 2 = ∂ µ ∂µ = ∂t2 The value of the antisymmetric tensor ǫµνρσ is determined in the same way as for ǫijk .33) (1. not affected by Lorentz transformations. defined by ∂2 − ∇2 . ∂2 .and righthandside a four vector because τ is a scalar quantity! The equation with t = x0 instead of τ simply would not make sense! The momentum four vector. (1.1 (a) In the the Hydrogen atom (quantum system) the scale is set by the Bohr radius. ∂t ∂x ∂y ∂z ν ∂µ xν = gµ . (1. 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.27) It is easy to convince oneself of the nature of the indices in the above equation. but more important has the same effect on lefthandside and righthandside.28) ν ν Note that gµ with one upper and lower index. for instance. because one has (1.Introduction 6 called covariant. . = −24.

a2 . How are a+ and a− related to a+ and a− . n2 . 0) ˆ n2 = (0.13 to give its actual value in kg. c. Also construct and calculate the Planck length Lpl . Use that to give the ˆ ˆµ ˆ components of the basis vectors n+ and n− .2 Prove the identity A × (B × C) = (A · C) B . ǫ0 . n2 . 1. Compare it with the proton mass and use Eq. ]. e. Exercise 1. (b) The coordinates (a0 . Also for the coordinates [a− . g−− . σ} is a permutation of {0. 0. ν. a+ . but only appropriate combinations.3 Prove the following relation ǫµνρσ g αβ = ǫανρσ g µβ + ǫµαρσ g νβ + ǫµνασ g ρβ + ǫµνρα g σβ . 0. a2 ] we can find basis vectors n− . ρ.71 × 10−39 GeV−2 to construct a mass Mpl (Planck mass). 1. 1. One does not need to know . . (d) Use the value of the gravitational constant GN / c5 = 6. 1. 0) ˆ n0 = (1. g+− and g−+ . n+ . This demonstrates how a careful use of units can save a lot of work. 0. α and me c2 = 0. Note that ˆ ˆ ˆ ˆ a · nα = aα and derive from that the components nα of the four vector nα . a+ . n3 . 1) ˆ . Exercise 1. (a) Express the scalar product a · b in lightcone coordinates and deduce from this the values of g++ . a1 . me . [Note: what is the MKS unit for V/T?] Exercise 1.4 Lightcone coordinates for a four vector a (which we will denote with square brackets as [a− . 0) ˆ n3 = (0. a1 . (c) Calculate the Compton wavelength of the electron and the quantities under (a) and (b) using the value of c.511 MeV. . n1 .3. re = e2 /4πǫ0 me c2 to the Compton wavelength.(A · B) C using the properties of the tensor ǫijk given in section 1.Introduction 7 (b) Relate the classical radius of the electron (a relativistic concept). (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). a2 ] or [a− . aT ]) are defined through √ a± ≡ (a0 ± a3 )/ 2. 0. ˆ ˆ ˆ ˆ These are n1 = (0. 0. by a simple few-line reasoning [For instance: If {µ. 2. 3} the index α can only be equal to one of the indices in ǫµνρσ . 0. a+ . . n1 . which is the Compton wavelength for the Planck mass. a3 ) are the expansion coefficients using the basis vectors n0 . a1 . ˆ ˆ .

t) = 0. 2M (2. namely plane waves characterized by a wave number k.3 are not covariant.1 The Klein-Gordon equation In this chapter. (2.5) with (k 0 )2 = k2 + M 2 . But the replacement 2. ∂t p −→ pop = −i∇. one obtains (2.2) Acting on the wave function one finds for a free particle.2. i ∂ ∇2 ψ(r. φk (r. t) = exp(−i k 0 t + i k · r). The Schr¨dinger equation in quantum mechanics is the operator equation o corresponding to the non-relativistic expression for the energy. E = p2 . p). 8 . 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. t).7) k 0 = ± k2 + M 2 = ±Ek .4) where M is the particle mass. (2. written as pµ −→ i∂µ is covariant (the same in every frame of reference). t) = ∂2 − ∇2 + M 2 φ(r. ∂t2 (2.6) The energy spectrum is not bounded from below: considering the above stationary plane wave solutions. (2. ∂t 2M (2. Substitution of operators gives the Klein-Gordon (KG) equation for a real or complex function φ.1) under the substitution (in coordinate representation) E −→ Eop = i ∂ . Although it is straightforward to find the solutions of this equation. 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. 2 + M 2 φ(r. t) = − ψ(r. p2 = pµ pµ = E 2 − p2 = M 2 .Chapter 2 Relativistic wave equations 2. we just want to play a bit with covariant equations and study their behavior under Lorentz transformations.3) Equations 2.

however.2) is j µ = i φ∗ ∂ µ φ or (ρ. ∂ρ = −∇ · j. j) of the four-current j. 2. As we will see later both problems are related and have to do with the existence of particles and antiparticles. ∂µ j µ = 0.2 Mode expansion of solutions of the KG equation Before quantizing fields. there are no longer fundamental objections to mix up space and time. Then. k) + ei k·x φ(−Ek .11) Therefore. is a second order equation and φ and ∂φ/∂t can be fixed arbitrarily at a given time. At that point the dependence on position r and time t is just a dependence on numbers/parameters on which the operator depends. (2.14) ˜ ˜ Introducing φ(Ek . ↔ ↔ (2. −i φ∗ ∇ φ . just as the dependence on time was in ordinary quantum mechanics. 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. 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.9) They satisfy the continuity equation. 2M 2M (2. And.Relativistic wave equations 9 i. k) ≡ a(k) and φ(−Ek . for which we need the interpretation of φ itself as an operator. (2π)3 2Ek (2.5). −k) .12) It is easy to see that this current is conserved if φ (and φ∗ ) satisfy the KG equation.3). This operator has all possible solutions in it multiplied with creation (and annihilation) operators.8) (2. there are solutions with negative energy.e. having the KG equation as a space-time symmetric (classical) equation. −k) ≡ b∗ (k) one has φ(x) = d3 k e−i k·x a(k) + ei k·x b∗ (k) = φ+ (x) + φ− (x). φ(x) = d4 k ˜ 2π δ(k 2 − M 2 ) e−i k·x φ(k) (2π)4 (2. but rather the zero component of a four vector. The appropriate current corresponding to the KG equation (see Excercise 2. rather than as a wave function. j) = ↔ i φ∗ ∂0 φ. which is what Lorentz transformations do. we want the most general solution. φ(x) = d3 k (2π)3 2Ek ˜ ˜ e−i k·x φ(Ek . This leads to the existence of negative densities. This continuity equation can be written down o covariantly using the components (ρ. (2. The integration over k-modes clearly is covariant and restricted to the ‘mass’-shell (as required by Eq. It is possible to rewrite it as an integration over positive energies only but this gives two terms (use the result of exercise 2. (2. relativistically the density is not a scalar quantity.13) ˜ with (in principle complex) coefficients φ(k).10) ∂t which follows directly from the Schr¨dinger equation. 2.15) . it is simply a matter of being careful to find a consistent (covariant) theory.

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. 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).22) showing that with each solution there is a corresponding charge conjugated solution φC (x) = φ∗ (x).Relativistic wave equations 10 In Eqs 2. for which we consider space inversion. 2. i. (2. x ∂µ ∂ µ + M 2 φ(˜) = 0. They are referred to as mode and anti-mode amplitudes (or because of their origin positive and negative energy modes). (2.23) . −x) ≡ (˜µ ).6 We will explicitly discuss the example of a discrete symmetry.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.18) (2. Since a · b = a · ˜ it is easy to see that ˜ b.16) The consequence of this is discussed in Exercise 2. (2.21) 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) . x) → (t. (2π)3 2Ek (2.19) implying that for each solution φ(x) there exists a corresponding solution with the same energy. 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) .15 one has elimated k 0 and in both equations k · x = Ek t − k · x. x Transforming everywhere in the KG equation x → x one obtains ˜ ˜ ˜ x ∂µ ∂ µ + M 2 φ(˜) = 0. after restricting the energies to be positive). It is trivial to see that ∂µ ∂ µ + M 2 φ∗ (x) = 0.14 and 2.e. Another symmetry is found by complex conjugating the KG equation. Performing some Lorentz transformation x → x′ = Λx. This shows how parity transforms k-modes into −k modes.20) one has for the mode amplitudes aP (k) = a(−k) and bP (k) = b(−k). In terms of modes one has φC (x) = φ∗ (x) = ≡ d3 k e−i k·x b(k) + ei k·x a∗ (k) (2π)3 2Ek d3 k e−i k·x aC (k) + ei k·x bC∗ (k) .17) (2. which implies (xµ ) = (t. φP (x) ≡ φ(˜) (P for parity). The coefficients a(k) and b∗ (k) are the amplitudes of the two independent solutions (two. changing the sign of the spatial coordinates. (2π)3 2Ek (2.

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

Relativistic wave equations 12 Exercise 2. One then finds that Q is expressed as an infinite sum over number operators a† a (particles) and b† b (anti-particles). (c) Similarly. d3 x (∂ {0 φ)∗ (∂ i} φ). Note that a{µ bν} indicates symmetrization. Exercise 2. i (a) Show that relativistic invariance. B] = AB − BA and the anticommutator as {A. B} = AB + BA. 1 2 (b) Express now the full charge QV (t) (exercise 2. (b) Show that one has current conservation for the current 2 We denote the commutator as [A. i. 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 . . j 0 = ψ†ψ and j = ψ † αψ. ∂t where ψ is a wave function and the nature of α and β is left open for the moment.5 (a) As an introduction to parts (b) and (c). express the quantities E(t) = P i (t) = d3 x (∂0 φ)∗ (∂0 φ) + ∇φ∗ · ∇φ + M 2 φ∗ φ . Exercise 2. From this one concludes that α and β must be matrix-valued and ψ must be a multi-component wave function. ∂ ψ(r. also using the result in (a). and β 2 = I. We will encounter these quantities later as energy and momentum. t). making sure that each component of ψ satisfies the KleinGordon equation (in terms of differential operators ∂µ ∂ µ +m2 = 0) requires the anticommutation relation2 {αi .7 To solve the problem with positive and negative energies and get a positive definite density. in terms of the a(k) and b(k). t) = (−i α · ∇ + m β) ψ(r.15. β} = 0. Is the time dependence of the result as expected? Note that the mode amplitudes will become creation and annihilation operators after quantization of the fields (Chapter 7). We will 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). a{µ bν} ≡ aµ bν + aν bµ . 2. {αi . αj } = 2 δ ij I. an attempt to construct a first order differential equation for the time evolution would be to write.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).1) for a complex scalar field current (exercise 2.2) in terms of the a(k) and b(k) using the expansion in Eq.e.

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

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

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

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

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

    G =  10 (3.27) (3. 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.g. proof: det(ΛT GΛ) = det(G) → (det Λ)2 = 1. which as a matrix equation with (Λ)µν = Λµν and (G)µν = gµν gives (ΛT )ρµ (G)µν (Λ)νσ = (G)ρσ .     0 0   g00 g01 g02 g03   1 0      g     g11 g12 g13   0 −1 0 0  = . (ii) |Λ00 | ≥ 1. it is convenient to use a vector notation for the points in Minkowski space. (3.21)  g  20 g21 g22 g23   0 0 −1 0             g30 g31 g32 g33 0 0 0 −1 the scalar product can be written as x2 = xT G x (3.Groups and their representations 18 3. explicitly x′µ = (Λ)µν xν + aµ or x′µ = Λµν xν + aµ . det Λ = −1 is called improper). (det Λ = +1 is called proper.22) (Note that xT is a row vector). σ/2 in a two-dimensional (spin 1/2) case. Denoted in terms of column vectors and 4-dimensional matrices one writes for the poincar´ transe formations x′ = Λ x + a.24) L↑ + L↓ + L↑ − L↓ − proper orthochrone proper non-orthochrone improper orthochrone improper non-orthochrone . e To derive some of the properties of the Lorentz group.26) (3. From this property. e. Thus one has the (pseudo-orthogonality) relation ΛT GΛ = G ⇔ GΛT G = Λ−1 ⇔ ΛGΛT = G. proof: (ΛT GΛ)00 = (G)00 = 1 → Λµ0 gµν Λν0 = 1 → (Λ00 )2 − i (Λi 0 )2 = 1. Writing x as a column vector and the metric tensor in matrix form.23) The proper tensor structure of the matrix element (Λ)µν is a tensor with one upper and one lower index. We considered the generators and looked for representations in finite dimensional spaces. In this section we consider the Poincar´ group. consisting of the Lorentz group and translations.25) (3. it is easy to derive some properties of the matrices Λ: (i) det(Λ) = ±1. 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 .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.

since I ∈ L↑ and + + det Λ and Λ0 change continuously along a path from the identity. z ) = exp(i ϕ J 3 ). K 3 ). J = (J 1 . Note that the velocity β = v = v/c and the Lorentz contraction factor γ = (1 − β 2 )−1/2 corresponding to the boost are related to η as γ = cosh η. it is easy to find the following typical examples from each of the four classes. one distinguishes rotations 0 + and boosts. (3.   0 0   +1 0    0 +1 0   0   ∈ L↑ (identity)   I =  (3. K 2 . Rotations around the z-axis are given by ΛR (ϕ. This is the origin of the Thomas precession. Using these explicit transformations. infinitesimally given by ˆ ΛR (ϕ.32)  0 −   0 −1 0      0 0 0 −1 . From the commutation relations one sees that the boosts (pure Lorentz transformations) do not form a group. 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  . J j ] = i ǫijk J k . and those of the boosts.Groups and their representations = i=1 (Λ0 )2 . z ) ≈ I + i ϕ J 3 . since the generators K do not form a closed algebra. βγ = sinh η. [J i . 3 i 2 i=1 (Λ0 ) 3 19 (iii) Boosts along the z-direction are given by ΛB (η. Thus ˆ  0′       0 0 0  V 0   V   1  0 0 0 0   1′    1    V   0 cos ϕ sin ϕ 0   V   0 0 −i 0              2′  =     2    −→ J3 =   V   0 − sin ϕ cos ϕ 0   V   0 i 0 0  . J 2 . K j ] = −i ǫijk J k . infinitesimally given by ΛB (η. K = (K 1 .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↑ . The commutator of two boosts in different directions (e. (3.31)  0 −   0 +1 0      0 0 0 +1   0 0   +1 0    0 −1 0   0   ∈ L↑ (space inversion)   Is =  (3. J 3 ). Returning to the global group.28)                 3′    3    V 0 0 0 1 V 0 0 0 0 The parameter η runs from −∞ < η < ∞. z ) ≈ ˆ ˆ I − i η K 3 . i proof: use ΛT GΛ = G and ΛGΛT = G. z ) = exp(−i ηK 3 ). K j ] = i ǫijk K k . which satisfy the commutation relations (check!) [J i . [K i . The first two sets of commutation relations exhibit the rotational behavior of J and K as vectors in E(3) under rotations. In L↑ . we have found the generators of rotations.30)  0 +   0 +1 0      0 0 0 +1   0 0   −1 0    0 +1 0   0   ∈ L↓ (time inversion)   It =  (3.g. 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).

37) The algebra of the generators of the Lorentz transformations can be obtained by an explicit calculation2 . with ϕi = − 1 ǫijk ω jk and J i = 2 z-axis). [M µν . Summarizing.40) ΛB = I + i η 3 K 3 = I − i ω30 M 30 . M ρσ ] = −i (g µρ M νσ − g νρ M µσ ) − i (g µσ M ρν − g νσ M ρµ ) . a). We now want to find a covariant form of the six generators of the Lorentz transformations. 2 (M αβ )µν = i g αµ g βν − g α g βµ ν (3.g.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. One can easily convince oneself that i (ω)µν = ω µν = − ωαβ (M αβ )µν . (3. three of which only involve spatial coordinates (rotations) and three others involving time components (boosts). We therefore find (again) that there are six generators for the Lorentz group. In order to find the commutation relations including the translation generators.35) thus (ω)ij = −(ω)ji and (ω)0i = (ω)i0 . we have for the (infinitesimal) rotations (around z-axis) ΛR = I + i ϕ3 J 3 = I − i ω12 M 12 . The e 2 These commutation relations are the same as those for the ’quantummechanics’ operators i(xµ ∂ ν − xν ∂ µ ) . Also this group can be divided into four parts. J 3 = M 12 ) and for the (infinitesimal) boosts (along (3. ǫ). This is a + normal subgroup and the factor group is the Vierer group. inf (3.36) (3. Thus the two vector operators J and K form under Lorentz transformations an antisymmetric tensor M µν . There are six generators.34) The condition ΛT GΛ = G yields g ρµ + ω ρµ gρσ (g σν + ω σν ) = gµν =⇒ ωνµ + ωµν = 0. (3. i i0 i i0 1 ijk M jk 2ǫ (3. 3 30 with η = ω and K = M (e. The extension to the Poincar´ group is e ↑ straightforward. Multiplying the proper orthochrone transformations with one of them gives all Lorentz transformations. Of the four parts only L↑ forms a group. infinitesimally approximated by (I + ω. P+ etc.39) (e.g. 3. explicitly reading x′µ = = (Λ)µν xν + aµ = ((I)µν + (ω)µν ) xν + ǫµ Λµν xν + aµ = (g µν + ω µν ) xν + ǫµ . we continue using covariance arguments. We will just require Poincar´ invariance for the generators themselves.38) Explicitly.33) These four transformations form the Vierer group (group of Klein). the Lorentz transformations form a (Lie) group (Λ1 Λ2 and Λ−1 are again Lorentz transformations. which are obtained by writing the infinitesimal parameters (ω)µν in terms of six antisymmetric matrices (M αβ )µν .4 The generators of the Poincar´ group e ↑ The transformations belonging to P+ are denoted as (Λ. K = M ).

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

The elements of the little group thus are U (Λ(p)) = = and U (Λ(p))|p. Wα ] = [Wµ .Groups and their representations 22 where s is an arbitrary vector of which the length and the component along pσ are irrelevant. Given the momentum four vector pµ we choose a tetrade consisting of three orthogonal spacelike unit vectors ni (p). 2 exp (−isµ W µ ) |p. From the algebra of the generators one finds that P 2 = Pµ P µ and W 2 = Wµ W µ (3. i ǫµνρσ W ρ P σ . M ǫijk . (3. which can thus be chosen spacelike.53) (3. Pν ] = [Wµ . i exp − ωµν M µν 2 i exp − ǫµνρσ M µν pρ sσ 2 (3. satisfying gµν nµ (p)nν (p) = j i nµ (p) pµ i ǫµνρσ pµ nν nρ nσ i j k = = −δij . 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. of which the components generate the little group of pµ or (for the third of the following relations) from explicit calculation [Mµν . .56) . 2 (3. Massive particles: p2 = M 2 > 0. p0 > 0.54) This will enable us to pick a suitable commuting ’spin’ operator. These e operators are the Casimir operators of the algebra and can be used to define the representations of ↑ P+ . while the others have no physical significance. Wν ] = −i (gµα Wν − gνα Wµ ) . . 0. .55) commute with all generators and therefore are invariants under Poincar´ transformations. (3. = = i exp − ǫµνρσ M µν pρ sσ |p. the third case represents the vacuum. . . .49) (3. .50) where the Pauli-Lubanski operators W µ are given by 1 Wµ = − ǫµνρσ M νρ P σ . .52) (3. 0. .51) The following properties follow from the fact that W µ is a four-vector by construction.

1. 1 2 (3.58) i. . 0. 0).64) such that in an arbitrary frame where the four vectors p. n2 (p) and s(p) are obtained by a Lorentz boost from the reference frame one has the property ǫµνρσ sµ nν nρ pσ > 0. (3. Having made a covariant decomposition. The explicit construction of the wave function can be done using the D-functions (analogous as the rotation functions). the S i (p) form the generators of an SU (2) subgroup that belongs to the maximal set of commuting operators. S j (p)] = −i nµ (p)nν (p) j i ǫµνρσ W ρ P σ = i ǫijk S k (p). (1.e.62) 1 Thus W 2 has the eigenvalues −M 2 s(s + 1) with s = 0. 0. (3.Groups and their representations 23 We can write W µ (p) = 3 W i (p) nµ (p) i i=1 (3. W j (p)] = i M ǫijk W k (p). The commutation relations [W i (p).e. i (3. 1. 0.59) (3. 0. . show that W /M can be interpreted as the intrinsic spin. |p3 |). In that case the vectors ni (p) are just the space directions and W = (0. it is sufficient to choose a particular frame to investigate the coefficients in Eq. More generally (fully covariantly) one can proceed by defining S i (p) ≡ W i (p)/M and obtain [S i (p). 3. In this case a set of four independent vectors is chosen starting with a reference frame in which pµ = pµ 0 in the following way: p0 = (p0 . 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. n1 (p).60) and using the completeness relation nµ (p)nν (p) = − g µν − i i 2 i pµ pν M2 . −1) n1 (p0 ) ≡ n2 (p0 ) ≡ s(p0 ) ≡ (3. 0). p0 > 0.57) (i. Noting that S i (p) i 2 = 1 W µ (p)W ν (p) M2 niµ (p)niν (p).63) where E = p2 + M 2 and ms is the z-component of the spin −s ≤ ms ≤ +s (in steps of one). ms . W i (p) = −W · ni (p)). 0. M J) with components W i = (M/2)ǫijk M jk . (0. s. 2 . Together with the four momentum states thus can be labeled as |M. 0. p. . 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. M2 (3.61) one sees that S i (p) i =− W2 . M2 (3. 1.65) . (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 . Wν ]. (3. a)|0 = |0 . The meaning of λ(p) is most easily seen by comparing for a momentum eigenstate W 0 (p) = λ(p) |p| with W 0 (p) = − 1 ǫ0ijk Mij pk = J · p. The algebra of W 1 . Thus we have P 2 . 1 2 24 (3. −i W 1 (p). |p| (3. λ . W 1 (p)] [λ(p). W 2 and λ is derived from the general commutation relations for [Wµ . |p.70) as a commuting set with zero eigenvalues for P 2 and W 2 . 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 ). and gives [W 1 (p). This state is in physical applications nondegenerate and is denoted by |0 . This algebra is (for instance) isomorphic to the one generated by rotations and translations in a 2-dimensional Euclidean plane (see exercises). is characterized by its momentum and the helicity.68) (3. (3.69) We reproduce the algebra for the massive case with W 3 (p) = M λ(p). W 2 (3. The eigenvalues of W 1 and W 2 (in that case corresponding to the translations) thus can take any continuous values. 3. however. i W 2 (p). W 2 (p)] [λ(p).67 has a vanishing right hand side). W 2 (p)] = = = i M 2 λ(p).73) (3. P µ . 3. where A = exp(i ϕ · σ/2) = exp(iϕ σ · n/2). We don’t know of any physical relevance. W µ (p) = λ(p) P µ . Note that angular momentum does not contribute to helicity as L · p = 0.67) (3.1 (a) Derive from the relation that defines the mapping of SU (2) matrices A into SO(3) matrices RA in Eq. Thus 2 λ(p) = J ·p . The vacuum: pµ = 0. which can take any integer or half-integer value.72) One has P µ |0 = W µ |0 = 0 and the state is invariant under Lorentz transformations.Groups and their representations The above does actually include the massive case.66) where W · p = 0 implies that W s (p) = −(M 2 /p · s) λ(p).13.71) which is called the helicity. A massless particle. Exercises Exercise 3. U (Λ. But if M = 0 one has a different algebra (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). thus. σ · RA a ≡ A(σ · a)A−1 ˆ (valid for any vector a).

show that one matrix ǫ exist such that J c = ǫ J ǫ† . the pions and the isospin 0 meson state. Exercise 3. Tr(σi σj σk ) = 2i ǫijk . Tr(σi σj σk σl ) = 2 (δij δkl − δik δjl + δil δjk ) . Writing A∗ = exp(i α · J c ). the trace of the operator Jk is zero (Convince yourself that det(eA ) = eTr A by diagonalizing A.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. d) doublet representation for SU (2)isospin or the Higgs (φ+ . Is this easy to understand. Now write down the three isospin 1 meson quark-antiquark (meson) states. Both particle and antiparticle have ’spin 1/2’. Note: Recall (or derive) that the Pauli matrices σ satisfy σi σj = δij + i ǫijk σk and thus Tr(σi σj ) = 2 δij . This is true for SU (2) but it is. what are the matrices J c in the Lie-algebra representation describing the spin rotations for two-component ’anti-spinors’. For a spin 1/2 antiparticle the representation A∗ is needed.Groups and their representations 25 for any vectors a and b. for instance. Note the difference with the (usual) spin 0 and spin 1 states composed from two spin 1/2 states. triplet and anti-triplet representation. . and calculate U (a) p U −1 (a) . the (non-strange) η-meson state. u). Excercise 3. The appropriate isospin doublet to be used for antiquarks then is (−d. Show actually that ǫ = 2 eiπJ (up to a phase). (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 if det(U ) = 1. O] + . the operators Jk must be hermitean. In both of these cases the symmetry is broken by the charges ¯¯ of the particles. .e. (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). Examples are the quark (u. not true for SU (3) (flavor or color properties of particles know two distinct defining 3-dimensional representation.2 Show that for a unitary operator U = exp[iαk Jk ] with real coefficients αk . (b) This establishes that for spin 1/2 one doesn’t need to talk about ’anti-spin’ for the antiparticle.4 (a) If the states in a Hilbert space (unitarily) transform as φ′ = U φ = eiJ φ and O is an operator acting on φ in that Hilbert space. . show that the transformation for the operators is given by O′ = U O U −1 = O + i [J. respectively). Show that the Lie-algebra representations J and J c are equivalent. Exercise 3. i. (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. one has to explicitly include this rotation. φ0 ) doublet representation for SU (2)weak .

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

It is simply connected and forms the covering group of L↑ .4) (4. 2 2 (4. U (ϕ) H(η) (4.6) (4. K = i J (A = 0). K = −i . Aj ] = [B i . (4.10) .1 The Lorentz group and SL(2. B j ] = [Ai . j) K = −i J (B = 0). This tells us how to find the representations of the group. 2 1 (J − i K).5) = 1 (J + i K). Special cases are the following representations: Type I : (j. They precisely correspond to the two types of representations that we have seen before.1) (4.3) (4. C).7) From the considerations above. With this choice of parameter-spaces the plus and minus signs are actually relevant. (4. C) can be written as a product of a unitary + matrix U and a hermitean matrix H.8) ˆ ˆ ˆ with η = η n and ϕ = ϕn. K = +i . 0) Type II : (0. where we restrict (for fixed n) the parameters 0 ≤ ϕ ≤ 2π and 0 ≤ η < ∞.9) (4. 2 2 σ σ J = . The group SL(2. respectively. 2 (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 . becoming the defining representations of SL(2. They will be labeled by two angular momenta corresponding to the A and B groups. 0. i ǫijk B k . it also follows directly that the Lorentz group is homeomorphic with the group SL(2. C): Type I (denoted M ): Type II (denoted M ): 27 J= σ σ . j ′ ).Chapter 4 The Dirac equation 4. (j. similarly as the homeomorphism between SO(3) and SU (2). M = exp i ϕ·σ 2 1 exp ± η · σ 2 = U (ϕ) H(η) .2) This shows that the Lie algebra of the Lorentz group is identical to that of SU (2) ⊗ SU (2). The matrices in SL(2. C) is the group of complex 2 × 2 matrices with determinant one. B j ] = i ǫijk Ak .

thus. (4. η → Hη.19) .12) ξ → U ξ.The Dirac equation 28 Let us investigate the defining (two-dimensional) representations of SL(2. The Lorentz transformations Λ(M ) and Λ(M ).18 show that ±ǫξ ∗ transforms as a type-II (η) spinor.13) 2M (E + M ) (exercise 4.16) = σ µ Λ R aµ .15) (4. −σ). are not covariant!) ˜ can be used to establish the homeomorphism between L↑ and SL(2. ˜ = σ µ Λ B aµ . while ±ǫη ∗ transforms as a type-I + spinor. there exists the matrix ǫ = iσ 2 . if there does not exist a unitary matrix S. ξ −→ Hξ. U σ µ aµ U ˜ −1 H σ µ aµ H −1 = σ µ ΛB aµ H σ µ aµ H ˜ −1 (4. however. Λ(ǫ) ∈ L+ . One has + −1 Λ(M ∗ ) = I2 Λ(M ) I2 . (4. In fact. C).1). (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. H ≡ (H † )−1 = H −1 ).14) satisfying Tr(σ µ σ ν ) = −2 gµν and Tr(σ µ σ ν ) = 2gµν = 2 δ µν (the matrices. ǫ−1 = ǫ† = ǫT = −ǫ). (4.18) As already discussed for SU (2). 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. such that M = SM S −1 . C) or L↑ . This is not true for the two-component spinors in SL(2. U σ µ aµ U −1 = σ µ ΛR aµ . (4.e. (ǫ∗ = ǫ. Note that M ∗ = ǫ† M ǫ has its equivalent for the Lorentz transformations Λ(M ) and Λ(M ∗ ). U ≡ (U † )−1 = U ) U (ϕ) = exp(i ϕ · σ/2) but differently under hermitean boosts (H † = H. With E = M γ = M cosh(η) and p = M βγ n = M sinh(η) n. σ µ ≡ (1. cannot be connected by a Lorentz transformation belonging to L↑ . Also useful is the relation H 2 (p) = σ µ pµ /M = (E + σ · p)/M . C). The following relations for + rotations and boosts are useful. η → U η. namely H(η) = exp(η · σ/2). One defines spinors ξ and η transforming similarly under unitary rotations (U † = U −1 . ˜ The sets of four operators defined by definitions σ µ ≡ (1. H(η) = exp(−η · σ/2). ˜ Non-equivalence of M and M One might wonder if type I and type II representations are not equivalent. ¯ ¯ H ∗ = ǫ† Hǫ = ǫ−1 Hǫ. This shows that M ∗ ≃ M . i. one can use a boost to the frame with momentum ˆ ˆ p. They are connected via Λ(M ∗ ) = Λ−1 (ǫ)Λ(M )Λ(ǫ) within L↑ because iσ 2 = U (πˆ ) showing that y + ↑ ∗ ǫ ∈ SL(2. C) and. 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). Eqs 4. thus.11) Considering ξ and η as spin states in the rest-frame. σ). ˜ (4.

0) of SL(2.17. they cannot be connected by a unitary 2 2 transformation. consider the four component spinor that transforms under a Lorentz transformation as      0  ξ   ξ  . we have seen that in the rest frame W i (p)/M → J i . that have the correct commutation relations. −→ −r (vector). (scalar). 4. m) =            χm uL H(p) 0 .13. i.1. m) = χm .23) u(p.The Dirac equation 29 where The matrix I2 does not belong to L↑ . Taking M (Λ) = H(p). We also have seen that the representations (0. The i angular momentum operators J are represented by σ i /2. The representing member of the class P− is the parity or space inversion operator Is . ) −→ ( . and the angular momentum in the rest frame is what we are familiar with as the spin of a particle.2 Spin 1/2 representations of the Lorentz group 1 1 Both representations (0. Although these operators cannot be used to label the representations.e. + −   1 0    0 1  I2 =    0 0   0 0 0 0 −1 0 0 0 0 1      . as we have seen in section 4. Examples (with between brackets the Euclidean assigments) are r t p H J λ(p) K −→ −λ(p) −→ −K −→ H −→ J −→ t (scalar). From the definition of the representations (0.g. e. 1 ) and ( 1 .       = (4.22) u=     η η M (Λ) 0 where M (Λ) = ǫM ∗ (Λ)ǫ−1 with ǫ given in Eq. etc. the well-known two-component spinor for a spin 1/2 particle. 0). 0) (via operators J and K) one sees that under parity 2 2 1 1 (0. 2 2 (4. C). (axial vector). It has the same effect as lowering the indices. −a). For the spin 1/2 representations of the Poincar´ group including parity we. (pseudoscalar). but rather the operators W i (p) should be used. 1 ) and ( 1 . C) are suitable for representing spin 1/2 particles. generators. we obtain for the two components of u which we will refer to as chiral right and chiral left components.      0   χm   uR   H(p)      . 4. but to L↑ . the boost in Eq.      (4. 2 ) and ( 2 . For a particle at rest only angular momentum is important and we can choose ξ(0. −→ −p (vector). 0) are inequivalent. a) and ˜ a = (a0 . The behavior is the same for classical quantities. therefore.20) 4.   −→  M (Λ)   (4. the aij elements of a tensor will not change sign. must combine the repe resentations. i. they can be connected.e. Thus M ∗ and M are not equivalent within SL(2. A parity transformation changes aµ into aµ . Within the Lorentz group. but by a transformation belonging ↑ ↑ to the class P− . ǫµνρσ will change sign. m) = η(0. This provides the easiest way of seeing ˜ what is happening. where a = (a0 . (vector).21) In nature parity turns (often) out to be a good quantum number for elementary particle states.

30) γ µ γ ν ∂µ ∂ν + M 2 ψ(x) = 0. From this one obtains the Clifford algebra for the Dirac matrices. 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. (4. (4.24) (4. γ } ∂µ ∂ν + M 2 ψ(x) = 0.27 is known as the Weyl representation. This leads to the Dirac equation for ψ(x) = u(p)e−i p·x in coordinate space. {γ µ . We first want to investigate if the Dirac equation solves the problem with the negative densities and negative energies.7. We will discuss another explicit realization of this algebra in the next section. this can be rewritten 1 µ ν {γ . .26)        −2    uL uL H (p) 0 or explicitly in the socalled Weyl representation  −M E+σ·p       E−σ·p −M    uR      uL      = 0.28) which is a covariant (linear) first order differential equation. Applying (iγ µ ∂µ + M ) from the left gives (4.33) define for a four vector a the contraction a = aµ γµ / . The explicit realization appearing in Eq.   (4.27) where γ µ are 4 × 4 matrices called the Dirac matrices1 .29) which is an explicit realization of the (momentum space) Dirac equation. γ ν } = 2 g µν . As in section 2 we can use Pµ = i∂µ as a representation for the momenta (translation operators) in function space.The Dirac equation 30 with H(p) = ¯ H(p) = 2M (E + M ) E+M −σ·p 2M (E + M ) E+M +σ·p . which in general is a linear equation in pµ .      0 H 2 (p)   uR   uR        . = H −1 (p). The general requirements for the γ matrices are thus easily obtained. p (4. (iγ µ ∂µ − M ) ψ(x) = 0. To see this consider the equation for the hermitean † † conjugate spinor ψ † (noting that γ0 = γ0 and γi = −γi ).25) It is straightforward to eliminate χm and obtain the following constraint on the components of u. (4. 4.32) suppressing on the RHS the identity matrix in Dirac space.31) To achieve also that the energy-momentum relation p2 = M 2 is satisfied for u(p). It is of a form that we also played with in Exercise 2. (pµ γ µ − M ) u(p) ≡ (/ − M ) u(p) = 0. Since ∂µ ∂ν is symmetric. ψ † −iγ 0 ∂0 +iγ i ∂i −M 1 We ← ← = 0. The first indeed is solved. 2 (4.

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

For example. γ σ ].  0 (4. K 3 = M 30 =    0 iσ 3 2 4 or in general one has the transformation i ψ → Lψ = exp − ωρσ S ρσ ψ. it is easy to see that PR/L = 1 (1 ± γ5 ) 2    1 0  .42) We note that the explicit matrix taking us from the Weyl representation to the standard representation. is   1  1 1    (4. 1 −1 2 For all representations one has † γµ = γ0 γµ γ0 .9 and 4.41) (4.3) . γ 0 ]. ν The latter assures Lorentz invariance of the Dirac equation (see Exercise 4. Lorentz invariance The Lorentz transformations can also be written in terms of Dirac matrices. the rotation and boost generators in Weyl representation in Eqs 4. 0 σ3 2 4   1  −iσ 3 0   = − i [γ 3 . = S(γµ )W. (4.  =ρ ⊗1=  0 −1 3 (4. 2 (4.The Dirac equation 32 The Weyl (or chiral) representation:    0 1   . 4! (4.R. Another matrix which is often used is γ5 defined as γ5 = i γ 0 γ 1 γ 2 γ 3 = −i γ0 γ1 γ2 γ3 = It satisfies {γ5 . γ 2 ]. We note that ψ → Lψ and ψ → ψ L−1 .R. γ 0 = ρ1 ⊗ 1 =   1 0 γµ ψ Different representations can be related to each other by unitary transformations. −→ Sγµ S −1 . that project out chiral right/left states.R.46) (4.10 are   1  σ3 0  i   J 3 = M 12 =   = [γ 1 . while L−1 γ µ L = Λµ γ ν .49) i 1 with −i S ρσ = 2 σ ρσ = − 4 [γ ρ .48) are projection operators.R. For instance in the Weyl representation (but valid generally).45) (γ5 )W.   0 γ k = −iρ2 ⊗ σ k =  k  σ  −σ k  . γ µ } = 0 and explicitly one has    0 1  1 . which in the case of the Weyl representation are just the upper and lower components.40) (4. S −1 . −→ Sψ. (γµ )S. = ρ ⊗ 1 =   1 0 i ǫµνρσ µ ν ρ σ γ γ γ γ .43) S=√  .47) (4.  (γ5 )S.44) and an adjoint spinor defined by ψ = ψ † γ0 .

∂ with the solution ψ c (x) = ηc Cψ (x). T (4. = =   0 iγ γ = −iρ ⊗ σ =   −iσ 2   −iσ 2 iγ 2 γ 0 = −iρ3 ⊗ σ 2 =   0 2 0 1 2 Thus we find back the Dirac equation. (C)S.R.54) e. It is easy to convince oneself that if ψ(x) is a solution of the Dirac equation. One has ψ c = −ψ T C −1 . ψR . (i/ − M ) ψ(x) = 0. usually to be taken unity. 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) = 0.52)    η ξ one can apply space inversion. As with parity we look for a transformation. (or W. called charge conjugation. ∂ (4. C is real and one has C −1 = C † = C T = −C c and [C. for instance. =    0 ǫ iσ 2 (4. Note that we have (as expected) explicitly in Weyl x representation in Dirac space     P  ξ    −→ ψ p = γ 0 ψ =  η  .57) (4. ψ= (4.53) (4. parity. −x) and via a few manipulations obtain again the ˜ Dirac equation (i/ − M ) ψ p (x) = 0.56) (i/ − M ) ψ c (x) = 0.R. In any representation a matrix C exist. Some properties of C are C −1 = C † and C T = −C. This symmetry does not change the spin 1/2 nature.51) The existence of positive and negative energy solutions implies another symmetry in the Dirac equation.59)  −→ ψ c = C ψ =    η ǫ ξ∗ .6).58) where ηc is an arbitrary (unobservable) phase.R. is a left-handed spinor. ∂ (4.The Dirac equation 33 Parity There are a number of symmetries in the Dirac equation. e. In S. reverse the charge of the particle.    −iσ 2   0 −ǫ  = .) and all representations connected via a real (up to an overall phase) matrix S. such that T Cγµ C −1 = −γµ .R. The latter implies that the conjugate of a right-handed spinor.   ψ= (4.g. in Weyl representation we find in Dirac space     T C  ξ   −ǫ η ∗     . γ5 ] = 0. x = (t. that brings ψ → ψ c . which is again a solution of the Dirac equation. (C)W.    0 −ǫ 0    0   −ǫ 0  . Explicitly.g.55) (4. T (4. but it does. x) → x = (t.50) Charge conjugation but with ψ p (x) ≡ ηp γ0 ψ(˜) (Exercise 4.

2M 2M −/ + M p v(p. 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) . s ) = 0. Choosing χ = −ǫχ∗ (the equivalent spin ¯ 1/2 conjugate representation). s)v(p. p (4. (2π)3 2Ek (4. s) = Ep + M      σ ·p   Ep +M χs     .69) = − s In order to project out spin states. s) v = .The Dirac equation 34 4.1). Note that the spinors in the negative energy modes (antiparticles) could be two different spinors.63) u(p. s) ei k·x d∗ (k. s). ¯ ¯ † ′ u (p.    (4. s ) = v † (p. s)u(p.65) (4. s) = v(p. 2M 2M (4. The ¯ ¯ solutions are normalized to u(p.66) An arbitrary spin 1/2 field can be expanded in the independent solutions. In an arbitrary frame one has s · p = 0 and s(p) can e.     ∂ −iσ · ∇ −i ∂t − M (4.70) =  Ps =  ˆ 0 1∓σ·s 2 2 . the best representation to describe particles at rest is the standard representation. v (p.   ⇔ (/ + M ) v(p) = 0.68) (4.. in which γ 0 is diagonal (see discussion of negative energy states in section 4. s ) = v (p. ¯ (4. ¯ ′ ′ Explicit solutions in the standard representation are   χs       . s)u(p. with E = Ep = + p2 + M 2 . s)¯(p. s′ ) = −2M δss′ . s)v(p. s′ ) = 2Ep δss′ . p     −σ · p (Ep − M )     Ep + M     σ ·p ¯ Ep +M χs χs ¯ For a free massive particle.62) where χs are two independent (s = ±) two-component spinors. where u satisfies   −σ · p  Ep − M      u(p) = 0.67) It is straightforward to find projection operators for the positive and negative energy states P+ P− = s p /+M u(p.g.61)     σ·p −(Ep + M ) There are also two negative energy solutions.   ⇔ (/ − M ) u(p) = 0. s). v(p. s) + v(k. s)¯(p. s)v(p. It is easy to check that   1  1±σ·s 1 ± γ5 / s ˆ 0    (4. In that frame is a spacelike unit vector sµ = (0. The explicit Dirac equation in the standard representation reads  ∂  iσ · ∇   i ∂t − M       (4. s) u = . s′ ) = 2M δss′ . s) and C v T (p. where v satisfies   σ·p  −(Ep + M )      v(p) = 0. s) = u(p. the spin polarization vector in the rest frame is the starting ˆ point. ψ(x) = v ±s (p) ei p·x .4 Plane wave solutions Looking for positive energy solutions ∝ exp(−iEt) one finds two solutions. s) e−i k·x b(k. be obtained by a Lorentz transformation.60)   ψ(x) = 0. the spinors satisfy C uT (p. s) = u(p.64) (4. ψ(x) = u±s (p) e−i p·x . s)u(p.

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

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

6 Apply space-inversion. to the Dirac equation and use this to show that the spinor ψ p (x) = ˜ γ0 ψ(˜). γν } = 2gµν that γµ / γ µ a = Tr(//) = ab µn µ1 Tr(γ · · · γ ) = −2 /. where x = (t. Show that the equality holds. s) u p /+M = 2M 2M both sides are projection operators. ˜ Exercise 4. 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 .5 Show starting from the relation {γµ . Exercise 4. γ σ ] = σ ρσ . 4. i S ρσ ] = g ρµ γ σ − g σµ γ ρ . s). H(p) = exp η·σ 2 = M +E+σ·p . by letting them act on the four basis spinors u(p. e.The Dirac equation 37 The trace is linear and can be split up in parts containing traces with up to four gamma-matrices. 0 if n is odd.2 Show that j µ = ψγ µ ψ is a conserved current if ψ en ψ are solutions of the Dirac equation Exercise 4. x ˜ . 4 2 Exercise 4. s) and v(p.4 Show that in P+ ≡ s=±1/2 u(p.77) Exercises Excercise 4.1 Prove Eq.g.3 Imposing Lorentz invariance on the Dirac equation can be achieved by requiring L−1 γ µ L = Λµ γ ν . −x) is also a solution of the Dirac equation. of which only the traces of four and two gamma-matrices are nonzero. 2M (E + M ) ˆ ˆ where E = M γ = M cosh(η) and p = M βγ n = M sinh(η) n. (4.g. 1 i −i S ρσ = − [γ ρ . by proving that H 2 (p) = exp (η · σ) = (E + σ · p)/M = σ µ pµ /M.13. Do this e. s)¯(p. Excercise 4. a 4 a · b. ν µ µ Look at the infinitesimal forms Λµ = gν + ων and L = 1 − (i/2)ωρσ S ρσ to obtain ν [γ µ . x → x.

A condition for confining the fermion in the cavity is i / ψ = ψ (where n is the surface normal). which is the most important factor needed to find the decay width (inverse lifetime).8 × 10−6 . Calculate the value of k for M R = 0. For such a cavity nµ = (0.e.66 MeV. Do you understand why? The calculation of the life-time is completed in chapter 11. M R = 1. . Note: the outcome shows that the decay into muons is much more likely that the decay into electrons (mπ = 139.5 and M → ∞.7 (a) Show that (in standard representation)  j0 (kr)χm    ψ(x. Show that this n condition is sufficient to guarantee that nµ j µ = 0.511 MeV). ν p Note: how such parametrizations work will be explained in chapter 8. The conditions for ψ and ψ imply that nµ j µ = ψ/ ψ = 0. the lepton and neutrino have spin 1/2 and are ν described by momentum-space spinors. namely i / ψ = eiαγ5 ψ. you need to determine the corresponding confinement condition for ψ. we want to confine this solution to a spherical cavity with radius R. there is no current flowing through the surface of the cavity. Note: use first the appropriate equation (which?) that is obeyed by each individual component of the Dirac equation to show that the upper two components are allowed solutions. In this solution the wave number k = |k| √ = E 2 − M 2 and χm a two-component spinor while j0 and j1 are spherical Bessel functions.57 MeV. the decay would even be forbidden. Show that the ratio is |M (π − → e− νe )|2 ¯ ≈ 54. n (c) Apply the confinement condition to find the value of k for the lowest eigenmode in the spherical ˆ cavity. |M (π − → µ− νµ )|2 ¯ Actually if the lepton would be massless.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 ). (d) Plot for these cases the vector densities ψ † ψ and the scalar densities ψψ as a function of r/R. mµ = 105.    is a stationary solution of the (free) Dirac equation. i. Excercise 4. The pion has spin 0. 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 ). me = 0. Note: To prove this. Calculate for now just |M |2 .The Dirac equation 38 Exercise 4. We will not pursue this chiral freedom at this point and take α = 0. r). t) = N     k ˆ i E+M σ · r j1 (kr)χm    −i Et   e . n It turns out that there is actually a whole family of conditions that provide confinement. (b) Next.

± 5. In the restframe. λ) + ǫ(λ)∗ (k) eik·x c∗ (k. 0).5)  −E    B2   . µ (2π)3 2E µ (5. just as the two-component Dirac spinors did for spin 1/2. ±(1) are obtained by boosting the restframe vectors. A). ǫλ ). one has ǫ(λ) (k) ǫ(λ)∗ (k) = −gµν + µ ν kµ kν .± (λ) d3 k ǫ(λ) (k) e−ik·x c(k. A real vector field Vµ (x) then can be expanded in modes as Vµ (x) = λ=0. These can describe spin 1 states in the rest-frame.7) .1 Fields for spin 1 In section 3. with ∂µ V µ = 0 (Lorentz condition).3) Since the vectors ǫµ (k) together with the momentum k µ form a complete set of four-vectors.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. where k = k0 = (M.Chapter 5 Vector fields and Maxwell equations 5.2) (5. 39 (5. The covariant generalization is to describe them with Vµ (x) = ǫµ (k) e±ik·x . M2 (5. namely ǫµ (k0 ) = (0. there are then indeed three independent possibilities.4) λ=0. (5.  −B 1    0 3 With the electromagnetic field tensor (5. In an arbitrary frame ǫµ (k) with λ = 0. These are solutions of the Klein-Gordon equation with in addition a condition. (2 + M 2 )Vµ = 0.1) The Lorentz condition implies that k µ ǫµ (k) = 0. λ) . The vectorfield is therefore (λ) suited to describe spin 1.2 we have discussed the vectors ǫλ (with λ = 0 and ±1) as basis states of the spin 1 representation.6) where Aµ is the four-vector potential (5.

17) (radiation. (5. Under a gauge transformation Aµ −→ Aµ + ∂µ χ.5. the antisymmetry of F µν implies that ˜ ∂µ F µν = 0 or ∂λ Fµν + ∂µ Fνλ + ∂ν Fλµ = 0. it will vanish at all times. (5. In that case one also has ∇·A = 0 or ki ǫi (k) = 0. ∓1. j) the Maxwell equations read ∂µ F µν = j ν .15 the d’Alembertian is replaced by 2 + M 2 . The equation in vacuum. while the fact µ C ∗ that the Minkowski space is real implies that Aµ (x) = Aµ (x) = Aµ (x). |k|). it does not eliminate the freedom of gauge transformations but they are now restricted to Aµ → Aµ + ∂µ χ with 2χ = 0. One possible choice is the gauge A0 = 0 (5. (5. transverse or Coulomb gauge). for M = 0.8) (5.16) (5. ˜ x Note that via parity transformation one obtains another solution.9) 1 ˜ where F µν ≡ − 2 ǫµνρσ Fρσ is the tensor where E → B and B → −E. Taking the divergence of this equiation.e. Therefore if k µ = (|k|. Note that the incorporation of the Lorentz condition can also be done for the case of the massive vector field. one finds for a conserved current and λ = 0 2∂µ Aµ = 0.14) of which the solutions give the Li´nard-Wiechert potentials. only two independent vectors remain ǫ(±) µ (k) = (0. shows that the electromagnetic fields correspond to the case of a massless spin 1 field. Furthermore.13) Hence. For electrodynamics one has furthermore the freedom of gauge transformations. 2Aµ = 0. i. . 0). 40 (5. (5. in which case one has 2Aµ = jµ . 0.15) (5. The Lorenz condition can actually be incorporated by changing the equations of motion to 2Aµ + (λ − 1)∂µ (∂ν Aν ) = jµ . moree over. −i. 0. The equations of motion for the fields Aµ become 2Aµ − ∂µ (∂ν Aν ) = jµ . 5. Although the Lorenz condition is a constraint. if ∂µ Aµ = 0 for large times |t|.18) corresponding to the helicity states λ = ±1 of the massless photon field. (5. The gauge freedom can first of all be used to impose the Lorentz condition for the electromagnetic field. ∂µ Aµ = 0. in accordance with what we found in section 3. It states the absence of magnetic charges and Faraday’s law.Vector fields and Maxwell equations and the (conserved) current j µ = (ρ.12) (5.10) the electric and magnetic fields are unchanged. if in Eq.11) This equation is not affected by a gauge transformation. Fµν −→ Fµν + (∂µ ∂ν χ − ∂ν ∂µ χ) = Fµν . As remarked above. there is still a gauge freedom left (Aµ → Aµ + ∂µ χ with 2χ = 0). AP (x) = Aµ (˜).

Inside solenoid: Outside solenoid: B = B z.19) are gauge invariant. = exp −i The phase difference between the two paths of the electron then are given by ∆δ = − = e 1 A · dr + e 2 A · dr = e e A · dr e Φ. This phase difference is actually observed. where Aµ = 0.3 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. F = e E + ev × B (5. This causes an interference pattern as a function of ∆x. may look a bit akward. nothing more than a manisfes1 tation of a nontrivial vacuum (a la the Dirac sea for fermions). while the electromagnetic field Aµ is not. but A = 0 (hence there is structure in the vacuum). . The situation. sin ϕ. A= By Bx Br ϕ= − ˆ . ϕ= − ˆ 2r 2r2 2r2 where r = (cos ϕ.1. 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 .0 . 5. . λ L − λ λ or ∆x = (L− λ/d)δ. 0). nevertheless.1: The Aharonov-Bohm experiment 5. illustrated in fig. Nevertheless the significance of Aµ is shown in the Aharonov-Bohm experiment. 2 2 2 BR2 BR2 y BR2 x A= . but where E = B = 0. . 0) and ϕ = (− sin ϕ.Vector fields and Maxwell equations 41 L ∆x d a (infinitely) long solenoid Figure 5. The energy density 2 (E 2 + B 2 ) = 0 (hence a vacuum). ˆ B = 0. cos ϕ.0 . The phase difference between the electrons travelling two different paths is given by δ = 2π a a ∆x d = −≈ . To be precise.20) e (∇ × A) · ds = B · ds = where Φ is the flux through the solenoid. In this experiment an observable phase difference is measured for electrons moving through field-free space (E = B = 0 but Aµ = 0). This occurs in the region outside the solenoid. (5. In the presence of an electromagnetic field Aµ an additional phase difference is observed.

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

∇×B =j+ ∇×E+ ∂E .Vector fields and Maxwell equations 43 Exercises Exercise 5. ∂t ∂B = 0.8 and 5. 5.9 explicitly give the equations ∇ · E = ρ. ∂t ∇ · B = 0. .1 Show that Eqs.

The requirement δS = 0 with fixed boundaries x(t1 ) = x1 and x(t2 ) = ˙ x2 leads to t2 ′ (6.3) (6.Chapter 6 Classical lagrangian field theory 6.1 Euler-Lagrange equations In the previous chapter we have seen the equations for scalar fields (Klein-Gordon equation). ∂x (6.8) 44 . Dirac fields (Dirac equation) and massless vector fields (Maxwell equations) and corresponding to these fields conserved currents describing the probability and probability current. and the total change x′ (t′ ) = x(t) + ∆x(t). x (t) = x(t) + δx(t). These equations can be obtained from a lagrangian using the action principle. (6.7) The quantity ∂L/∂ x plays a special role and is known as the canonical momentum.6) The first term leads to the Euler-Lagrange equations. x) ˙ (6. As an example. ∂x ˙ (6.2) The principle of minimal action looks for a stationary action under variations in the coordinates and time. recall classical mechanics with the action t2 S= t1 dt L(x. ˙ 2 (6. t1 (6. ˙ p= ∂L . x) = K − V = ˙ 1 mx2 − V (x). d dt ∂L ∂x ˙ = ∂L .4) δS = t1 t2 dt dt t1 ∂L ∂L δx + δ x + L δτ ˙ ∂x ∂x ˙ d ∂L − ∂x dt ∂L ∂x ˙ δx + t2 t1 = ∂L δx + L δτ ∂x ˙ t2 .5) with ∆x(t) = δx(t) + x(t) δτ .1) and as an example the lagrangian L(x. thus t′ = t + δτ.

Consider a lagrangian density L which depends on these functions. gives rise to a conserved quantity (at least if H is invariant!) Q=p dx dλ −H dt dλ . t1 ) and R3 . ∂µ φ(x)).18) 1 Taking a functional derivative.11) λ=0 of which p and H are the simplest examples related to space and time translations. t1 (6.Classical lagrangian field theory 45 For the lagrangian specified above this leads directly to Newton’s law p = m¨ = −∂V /∂x. H is a conserved quantity.15) d4 x′ L (φ′ . In general any continuous transformation. x) and an action t2 S[φ] = t1 dt L = dt d3 x L (φ(x). 6. = ∂R (R3 .9) which is done because the first term (multiplying ∆x.6 can be rewritten as δS = .2. δ(∂µ φ) (6. x) ≡ p x − L .t1 ) (6.13) (6. with ∆φ(x) = δφ(x) + (∂µ φ)δx . (6. .10) One sees that invariance under time translations requires that H(t1 ) = H(t2 ). indicated with δF [φ]/δφ should pose no problems. think for instance of a temperature or density distribution or of an electromagnetic field). ∆x = (dx/dλ)λ=0 δλ and δt = (dt/dλ)λ=0 δλ. . The hamiltonian H is defined by H(p. . x).17) Furthermore the variations δφ and δ∂µ φ contribute to δS. .e. + ∂L ∆x − H δτ ∂x ˙ t2 . .t2 ) d3 x .16) The change in variables x → x′ in the integration volume involves a surface variation of the form dσµ L δxµ . t2 ). ∂µ φ(x). λ=0 (6. which in classical mechanics vanishes at the boundary) does not play a role. The resulting variation of the action is δS = R µ ′ = xµ + δxµ . their derivatives and possibly on the position. L (φ(x). ˙ x The second term in Eq. .14) (6. = φ(x) + δφ(x) (6. . ˙ (6. ∂µ φ. R (6.12) Here R indicates a space-time volume bounded by (R3 . − d3 x . (R3 . ∂µ φ′ . also indicated by ∂R (a more general volume in four-dimensional space-time with some boundary ∂R can also be considered). i. Variations in the action can come from the coordinates or the fields. . In classical field theory one proceeds in complete analogy but using functions depending on space and time (classical fields. ∂µ φ(x)) = R d4 x L (φ(x). 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 . . We will come back to it in a bit more formal way in section 9. but which in 3 dimensions includes conserved quantities related to rotations and boosts. . indicated as x′µ φ (x) or combined φ′ (x′ ) = φ(x) + ∆φ(x). dσµ . ∂R Note for the specific choice of the surface for constant times t1 and t2 . x′ ) − d4 x L (φ.

Using the variations in ψ (in the symmetric form).20) (6.25) (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. (6.2 Lagrangians for spin 0. For the complex scalar field one conventionally uses L = = ∂µ φ∗ ∂ µ φ − M 2 φ∗ φ −φ∗ ∂µ ∂ µ + M 2 φ. ∂ (6. L = 1 1 ∂µ φ∂ µ φ − M 2 φ2 2 2 1 µ = − φ ∂µ ∂ + M 2 φ.24) (6. 2 ∗ (6. One easily obtains (2 + M 2 )φ(x) = 0. leading to (2 + M 2 )φ(x) = 0. (2 + M )φ (x) = 0.27) (6. 2 (6.23) (6.21) which differ only by surface terms. δL δL ∂µ = . 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).26) The Dirac field The appropriate lagrangian from which to derive the equations of motion is L = i → i ← i ↔ ψ / ψ − M ψψ = ψ / ψ − ψ / ψ − M ψψ ∂ ∂ ∂ 2 2 2 = ψ (i/ − M ) ψ. the Dirac field (spin 1/2) and the vector field (spin 1) can be found. (6. (6.19) δ(∂µ φ) δφ 6. the second term is irrelevant.28) where the second line is not symmetric but in the action only differs from the symmetric version by a surface term (partial integration). 2 ψ = 0.Classical lagrangian field theory 46 With for the situation of classical fields all variations of the fields and coordinates at the surface vanishing.29) . The scalar field It is straightforward to derive the equations of motion for a real scalar field φ from the lagrangian densities. The integrand of the first term must vanish. δL δ(∂µ ψ) δL δψ one obtains immediately i / −M ∂ → i = − γ µψ 2 i → = ∂ / ψ − M ψ. leading to the Euler-Lagrange equations.

2 (6. Using the twospinors ξ and η. In that case one finds trivially L = ↔ ↔ 1 1 ψR i / ψR + ψL i / ψL − M (ψR ψL + ψL ψR ).35) Since the kinetic term in L separates naturally in left and right parts (or ξ and η). ∂ ∂ 2 2 (6. 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.7). Peshkin and Schroeder or Exercise 12. and thus (β ∗ α)∗ = α∗ β = −βα∗ .34)  (6. that the lagrangian separates into two independent parts for M = 0.    ∗  0    ⇒ Υ ≡  −ǫ η  ΥL = ψL =    η η ∗ for which η0 = ǫ η0 .30) It is often useful to link to the two-component spinors ξ and η which we started with in chapter 4.g.38) (6.and lefthanded ones.39) 2 This term is written down with η being an anticommuting Grassmann number for which (αβ)∗ = β ∗ α∗ = −α∗ β ∗ . 1 M η † ǫη ∗ − M ∗ η T ǫη . ψL ≡ (ψL )c = C ψL =    0 (6.g. 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. . 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. (6. the mass term in the Dirac lagrangian 6.32) There exists another possibility to write down a mass term with only one kind of fields.Classical lagrangian field theory 47 and similarly from the variation with respect to ψ ψ i / +M ∂ ← = 0. 2 αβ = −βα.31) showing e. namely2 LM (Majorana) = + (6. L = = ↔ 1 1 c c M ψL ψL + M ∗ ψL ψL ψL i / ψL − ∂ 2 2 ↔ 1 1 M ΥR ΥL + M ∗ ΥL ΥR . The Majorana case is in fact more general. The reasons for Grassmann variables will become clear in the next chapter. (6. but note the factor 1/2 as compared to the Dirac lagrangian.36) (6.33) We now introduce a ’real’ (four-component) spinor satisfying Υc = Υ with ψL .31 is given by LM (Dirac) = −M ξ † η + η † ξ . Υi / Υ − ∂ 4 2   T  −ǫ η ∗  c  = ΥR . which comes because we in essence use ’real’ spinors. (6. or equivalently separate the field into right. 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 .

In particular when the lagrangian is invariant under symmetries. Q(t1 ) = Q(t2 ). implying ∂µ F µν + M 2 V ν = 0 and ∂µ V µ = 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 . we consider the second term in Eq.3 Symmetries and conserved (Noether) currents Next. (6. + ∂R dσµ dσµ ∂R ≡ where d x .45) J µ (x) = − δ(∂µ φ) dλ λ=0 dλ λ=0 Thus considering ∂R = σ2 .18. (6. These do not affect the dynamics and will give rise to conserved currents. discussed in the next chapter.43) F (σ1 ) − F (σ2 ) = dσµ J µ (x) = ∂R R d4 x ∂µ J µ (x). t). . 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.. which in a consistent picture precisely generate the symmetries. In this case. Q= d3 x J 0 (x. δ(∂µ φ) (6. ∂µ J µ (x) = 0. we get (omitting the parameter δλ) dxν dφ δL + Θµν (x) . (6.σ1 with σ = (R3 .42) Θµν = The variation δS thus can be expressed as δS with J µ (x) = − = δL ∂ ν φ − L g µν . 6. Returning to δS the surface term is rewritten to δS = R d4 x . .+ R 4 δL δL ∆φ − ∂ ν φ − L g µν δxν δ(∂µ φ) δ(∂µ φ) δL ∆φ − Θµν (x)δxν . (6..44) δL ∆φ + Θµν (x)δxν . these conserved quantities become operators.40) 6. requires the presence of a conserved quantity. one can consider the variations at the initial or final surface. t). (6. δ(∂µ φ) (6. As an example for the classical field.47) .41) (6.Classical lagrangian field theory 48 where Fµν = ∂µ Vν − ∂ν Vµ . δ(∂µ φ) For continuous transformations. ψ(x) → ei eΛ ψ(x) or ∆ψ(x) = i eΛ ψ(x). consider U (1) transformations of the Dirac field proportional to the charge e. the presence of a symmetry that leaves the lagrangian invariant. which is the space-integral over the zero-component of a conserved current. ∆φ = (dφ/dλ)λ=0 δλ and δxµ = (dxµ /dλ)λ=0 δλ.

(6.57) For the fields the behavior under Lorentz transformations has been the subject of previous chapters.49) These currents are conserved as discussed already in chapter 2. Lorentz transformations In this case the transformation of the coordinates and fields are written as δxµ ∆φi (x) = = ω µν xν . = −(∂µ φ)δxµ = −ǫµ ∂µ φ(x).48) For the complex scalar field the U (1) transformations leave the lagrangian invariant and lead to the conserved current (see Exercise 6. For quantized fields this will become the expressions of the hamiltonian and the momentum operators in terms of the fields. 1 (6.Classical lagrangian field theory From the lagrangian for the Dirac field. = 0.3) j µ = i e φ∗ ∂µ φ. (6. 2 (6. (6.55) Note that the coordinate transformations can be written in a form similar to that for the fields (matrices in internal space). t) is the conserved charge (Q = 0). H = d3 x Θ00 (x). 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. 6. ˙ Q = d3 x j 0 (x. From Noether’s theorem one sees that the current Θµν ǫν is conserved.g.56) δxµ = ωρσ (aρσ )µν xν 2 with (aρσ )µν = aρσ µν = g ρµ g σν − g σµ g ρν . Translations µ Under translations (generated by Pop ) we have δxµ ∆φ(x) δφ(x) = ǫµ .53) These are the energy and momentum. e. including e the space-time translations and the Lorentz transformations. Summarizing.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. (ω µν antisymmetric) 1 ωρσ (−i S ρσ )ij φj (x). The integral over the zero-component. . ↔ (6.50) (6. In the next section we will see the quantity show up as the charge operator. (6.54) (6. we obtain (since δxµ = 0). Therefore there are four conserved currents Θµν (ν labeling the currents!) and four conserved quantities Pν = dσµ Θµν = d3 x Θ0ν .51) (6.4 Space-time symmetries One kind of symmetries that leave the lagrangian invariant are the Poincar´ transformations.

Gauge transformations of the second kind or local gauge transformations are transformations of the type φ(x) → ei eΛ(x) φ(x). is not invariant under local gauge transformations.58) (6. Any lagrangian containing derivatives. (6. (6. the angle of the transformation depends on the space-time point x.64) (6. φ(x) φ (x) ∂µ φ(x) ∗ → e → e → ei eΛ(x) φ(x).6) T µν = T νµ .69) i eΛ(x) φ (x). In general this tensor is not symmetric.68) (6.70) .61) A final note concernes the symmetry properties of Θµν . (6.e. φ(x). 6. −i eΛ(x) ∗ i eΛ(x) (6. e. • Vector field Aµ : −i Sρσ = aρσ .g. in which the U (1) transformation involves an angle e Λ. The current deduced from Noether’s theorem is Jµ = = = = − δL ∆φi + Θµν δxν δ(∂µ φi ) δL 1 ωρσ − (−i S ρσ )ij φj + Θµρ xσ − Θµσ xρ 2 δ(∂µ φi ) 1 ωρσ {H µρσ + Θµρ xσ − Θµσ xρ } 2 1 ωρσ M µρσ .63) µν = ∂µ Θ .62) with Gρµν = (H ρµν + H µνρ + H νµρ )/2. ∂µ T µν (6. The lagrangians which we have considered sofar are invariant under global gauge transformations and corresponding to this there exist a conserved Noether current. where H ρµν is the quantity appearing in M ρµν . In some applications (specifically coupling to gravity) it is advantageous to have an equivalent current that is symmetric.59) (6. γσ ] This result for the Dirac field has been discussed in chapter 3 (see also Exercise 4. ∂µ φ(x) + i e ∂µ Λ(x) e (6.67) i.5 (Abelian) gauge theories φ(x) → ei eΛ φ(x).65) M µρσ = T µρ xσ − T µσ xρ . independent of x. • Dirac field ψ: −i Sρσ = −(1/4)[γρ .Classical lagrangian field theory 50 • Scalar field φ: −i Sρσ = 0. one has a tensor that satisfies (exercise 6. (6. (6. however. Defining T µν = Θµν − ∂ρ Gρµν .60) Therefore there are six conserved currents M µρσ labeled by ρ and σ (antisymmetric) and corresponding to it there exist conserved quantities M ρσ = d3 x M 0ρσ .3 we have seen global transformations or gauge transformations of the first kind. 2 (6.66) In section 6.3).

(6. i L = ψγ µ ∂µ ψ − (∂µ ψ)γ µ ψ − M ψψ.71) Thus the covariant derivative has the correct transformation behavior provided Aµ → Aµ − ∂µ Λ. ∂µ φ) =⇒ L (φ. Dµ φ) − Fµν F µν . (i/ − M ) ψ = (i/ − e/ − M ) ψ = 0. −e j µ Aµ = −e ρφ + e j · A.73) (6.72) (6.78) . D ∂ A (6. The equation of motion for the fermion follow from δL δ(∂µ ψ) δL δψ = = i − γ µ ψ.Classical lagrangian field theory 51 where the last term spoils gauge invariance. ∂ 2 4 (6. (6. 2 Minimal substitution ∂µ ψ → (∂µ + i eAµ )ψ leads to the gauge invariant lagrangian L = 1 i ↔ ψ / ψ − M ψψ − e ψγ µ ψ Aµ − Fµν F µν .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). As an example consider the Dirac lagrangian.74) the behavior which we have encountered before as a gauge freedom for massless vector fields with the (free) lagrangian density L = −(1/4)Fµν F µν . 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. 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). 4 (6. 1 L (φ. 2 i ∂ / ψ − M ψ − e/ ψ A 2 giving the Dirac equation in an electromagnetic field. Replacing derivatives by covariant derivatives and adding the (free) part for the massless vector fields to the original lagrangian therefore produces a gauge invariant lagrangian. (6.75) The field φ is used here in a general sense standing for any possible field. Dµ φ(x) ≡ (∂µ + i eAµ (x))φ(x).76) We note first of all that the coupling of the Dirac field (electron) to the vector field (photon) can be written in the familiar interaction form Lint = −e ψγ µ ψ Aµ = −e j µ Aµ . For this purpose it is necessary to introduce a vector field Aµ .

2 (a) Show that the homogeneous Maxwell equations can be derived from the lagrangian density 1 L = − Fµν F µν .4 Given the Dirac equation for a (negatively charged) spin-1/2 particle in an external electromagnetic T field. ∂µ F µν = j ν . where j µ = e ψγ µ ψ.e. . −eψγ ν ψ. (c) Show that the interaction term is invariant under gauge transformations only if the current jµ is conserved. φ → eiΛ φ. Exercise 6. giving the Maxwell equation coupling to the electromagnetic current. [iγ µ (∂µ − i e Aµ ) − M ]ψ = 0.79) Exercises Exercise 6.Classical lagrangian field theory 52 For the photon the equations of motion follow from δL δ(∂µ Aν ) δL δAν = = −F µν . Exercise 6. ∂µ F µν = j ν . 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.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 . i. (6. ∂µ j µ = 0 (Note that the addition of a total derivative to the lagrangian density does not modify the equations of motion). and deduce what is the charge of the antiparticle as compared to a particle.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. 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. give the equation which is satisfied by ψ c = Cψ .

Comment: this relation is known as the Pauli equation. 2M 2M with Enr ≡ E − M ≪ M . . t).5 Show that the Dirac equation for an electron with charge −e in an external electromagnetic field Aµ = (A0 . Hints: it may be useful to realize that Gµρσ is antisymmetric in first two indices. t) = Enr ψu (r. finally note that it is sufficient to show the last equation for dσµ M µρσ .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 . 2me Exercise 6. It shows that for a spin 1/2 particle the g-factor in the coupling of spin to a magnetic field takes the value gs = 2.Classical lagrangian field theory 53 Exercise 6.4. use ∂µ T µν = 0 and ∂µ M µρσ = 0. in order to show that Θρσ − Θσρ = ∂µ H µρσ . A) (see also exercise 6. e 1 (−i∇ + eA)2 + σ · B − eA0 ψu (r.6 (optional) Prove the properties of the tensor T µν in section 6. Hint ≡ −gs µe · B = −gs Qe s · B. This relation is known as the Pauli equation.

(7.3) For variations δO(p. ˙ 2 (7. dq dp dq dp (7. q) = p q − L(q. ˙ L(q. dλ Examples are dO = [O. namely [q. starting from the lagrangian L(q. An immediate generalization for fields can be obtained by considering them as coordinates. (7. Q]P .4) with a possible realization as operators in the Hilbert space of (coordinate space) wave functions through qop ψ(q) = qψ(q) and pop ψ(q) = −idψ/dq. (7. The bilinear product for the conserved quantities. dq The basic Poisson bracket is actually the one between q and its canonical momentum p. is the Poisson bracket [A.1 Canonical quantization We will first recall the example of classical mechanics for one coordinate q(t). q) = ˙ 1 mq 2 − V (q). 6.11) found through Noether’s theorem. t).Chapter 7 Quantization of fields 7. p)) = ˙ ˙ 2m where the (canonical) momentum p = ∂L/∂ q. there are conserved ˙ ˙ quantities Q (Eq.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). p]P = 1. Further. q) ≈ (dO/dλ)λ=0 δλ under symmetry transformations one has dO = [O. p]P . [q. 1 qx (t) = 3 d3 x φ(x. q) also considered in the previous chapter. q(q. labeled by the position.2) H(p. B]P = dB dA dA dB − . in our example p = mq. Quantization of the theory is achieved by imposing canonical commutation relations between q and p. however. This is nothing else as representing the classical Poisson algebra as operators in the Hilbert space. Actually the conserved quantities satisfy a similar algebra as the Lie algebras for continuous symmetries that they are obtained from. H]P dt and dO = [O. p] = i.

a) = Rj (Λ−1 ) φj (Λ−1 x − a). This implies specific transformation properties in the Hilbert space for the quantum fields φ.15) with furthermore the relations [φ(x. py (t)] = i δxy or for the fields φ(x) and Π(x) the socalled equal time commutation relations [φ(x.7) d3 x L (φ(x). for symmetry transformations. ∂ qx ˙ ∂ qx ˙ (7.16) . t). • Canonical commutation relations The quantization of the theory is achieved by imposing the canonical quantization condition [qx (t).9) Note that this indeed corresponds to the zero-zero component (Θ00 ) of the conserved energy-momentum stress tensor Θµν . qx . the discretization procedure above is an explicit example. t)] = i δ 3 (x − y). 2 2 2 (7. we have L (x) = 1 1 ∂µ φ∂ µ φ − M 2 φ2 2 2 1 1 1 2 (∂0 φ) − (∇φ)2 − M 2 φ2 = 2 2 2 δL = = ∂0 φ. qx+∆x ) . t).14) In fact. for the scalar field theory. a) φi (x)U (Λ. ∂µ φ(x)) = ˙ d3 x L (φ. ∆φ. ˙ where Πx (t) is obtained from the continuous field Π(x) ≡ δL /δ(∂0 φ). φ(y.11) (7. The hamiltonian then is H= x px qx − L = ˙ = x where H (x) = δL ˙ φ(x) − L (x) = Θ00 (x). the fields φ(x) can be considered as regular operators in the Hilbert space after smearing with a test function f . As an example.6) ∆3 x L x (qx . (7. Sometimes it is useful to keep in mind that. i U † (Λ. • Poincar´ invariance e The full variation of the fields. t). δ∂0 φ) 1 1 1 = Θ00 (x) = (∂0 φ)2 + (∇φ)2 + M 2 φ2 . t)] = [Π(x. The lagrangian is given by L = = x In order to construct the hamiltonian it is necessary to find the canonical momenta.Quantization of fields 55 etc.8) (7. px (t) = ∂L ∂Lx = ∆3 x = ∆3 x Πx (t). φ.12) (7. formally. Φ(f ) = d4 x φ(x) f (x). The essential conditions to consistently quantize a theory then are the following. albeit with ’sharp’ functions.13) Π(x) H (x) For the quantization procedure we can formulate a number of basic axioms of quantum field theory. discussed in the section 6. ∇φ) (7. t)] = 0. Π(y. Π(y.10) x ∆3 x [Πx qx − L x ] ˙ ∆3 x Hx (7.2. ˙ δ φ(x) (7. (7. (7.

sometimes referred to as second quantization. Pµ ] = [φ (x).2 Creation and annihilation operators Before discussing (real and complex) scalar fields and Dirac fields we recall the analogy with the wellknown harmonic oscillator as an example of quantization using creation and annihilation operators. P ] 2 (7.22) it is straightforward to check that the commutation relations between Q and P are equivalent with the commutation relations [a.20) where the coordinate Q and the momentum P satisfy the canonical commutation relations [Q.24) It is straightforward to find the commutation relations between N and a and a† . Φ(g)] = 0.Quantization of fields 56 or if U (Λ. 2 2 (7. In simplified form the hamiltonian is given by H= 1 2 1 2 2 P + ω Q . P ] = i. P ] = 0 (7. 2 i ωµν (S µν )i . [a.25) . [Q. ω a† a + 2 2 ω i − [Q.23) The hamiltonian in this case can be expressed in the number operator N = a† a. a] = [a† . The measurements cannot influence each other or [Φ(f ). 1 Q = √ (a + a† ) 2ω and P = −i ω (a − a† ) 2 (7. [φ(x).17) (Sµν )i φj (x). [N. a) = i Rj (Λ−1 ) = 1 + iǫµ P µ − 1− i ωµν M µν . (7. • Causality Operators Φ(f ) and Φ(g) for which the test functions are space-like separated can be measured simultaneously (macroscopic causality).18) with S µν given in 6. H = = ω ω P P Q−i √ Q+i√ 2 2 2ω 2ω 1 1 =ω N+ .21) Writing Q and P in terms of creation (a† ) and annihilation (a) operators.19) 7. a† ] = a† . (7. and [N. Q] = [P. a] = −a. j 2 one has [φi (x). (7. Microscopic causality implies local commutativity. a† ] = 0. i(xµ ∂ν − xν ∂µ )φ (x) + i (7. 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. φ(y)] = 0 if (x − y)2 < 0.4. a† ] = 1. Mµν ] = i i∂µ φi (x). j (7.

30) [φ(x). which represents the ’number of particles’ with momentum k. but one needs to realize that in that case k 0 = Ek = k2 + M 2 . a† (k ′ )] = 0.27) 2 (2π)3 It is easy to check that these equations can be inverted (see Exercise 2. 7. and (7. e. Π(x )]x0 =x′0 = 0. N |n = n|n one finds N a† |n N a|n = = (n + 1) a† |n . 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 .3 The real scalar field We have expanded the (classical) field in plane wave solutions. φ(x) = (2π)3 2Ek where the Fourier coefficients a(k) and a† (k) are now operators. is equivalent with [a(k).4 for the classical field) a(k) a† (k ′ ) = = d3 x ei k·x i ∂0 φ(x).e. Π(x′ )]x0 =x′0 = i δ 3 (x − x′ ) ′ ′ and (7. The quantization of the field is achieved by quantizing the coefficients in the Fourier expansion.g. d3 x φ(x) i ∂0 e−i k ·x ↔ ′ ↔ (7. From the normalizations one obtains a |n = √ n + 1 |n + 1 and a|n = n |n − 1 . (7.like’ commutation relations between a(k) and a† (k ′ ). a(k ′ )] = [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.26) a(k) e−i k·x + a† (k) ei k·x . (2π)3 2Ek (7. † √ a and a act as raising and lowering operators. φ(x )]x0 =x′0 = [Π(x). Note that we will often write a(k) or a† (k).Quantization of fields 57 Defining states |n as eigenstates of N with eigenvalue n.31) The hamiltonian can be rewritten in terms of a number operator N (k) = N (k) = a† (k)a(k).28) (7. which we have split into positive and negative energy pieces with (complex) coefficients a(k) and a∗ (k) multiplying them.32) (7. The canonical momentum becomes −i d3 k ˙ Π(x) = φ(x) = a(k) e−i k·x − a† (k) ei k·x .33) . the real scalar field φ(x) becomes d3 k (7. a† (k ′ )] = (2π)3 2Ek δ 3 (k − k′ ) [a(k). i. (n − 1) a|n † i.e.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). and we see that a state |0 must exist for which N |0 = a|0 = 0.

.46) (7. writing all annihilation operators to the right of the creation operators. 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). . which now contains an infinite number of oscillators.e. .41) (7. d3 k |k k|. assuring that the vacuum (by definition) has eigenvalue 0!. . φ(x) φ+ (x) φ− (x) = = = φ+ (x) + φ− (x). (2π)3 2Ek d3 k a† (k) ei k·x .35) where the vacuum contribution disappears because of rotational symmetry. = normalized as (a† (k1 ))n1 (a† (k2 ))n2 √ √ .40) This procedure is known as normal ordering. |0 . Just as in the case of the harmonic oscillator it is essential (axiom) that there exists a ground state |0 that is annihilated by a(k). d3 k a(k) e−i k·x .42) (7. For the purpose of normal ordering it is convenient to decompose the field in positive and negative frequency parts. which amounts to redefining H as H= d3 x : H (x) := d3 k Ek N (k).34) where V = (2π)3 δ 3 (0) is the space-volume. (2π)3 2Ek (7. a(k)|0 = 0. (2π)3 2Ek (7. . The 1-particle wave function e−ip·x is obtained via 0|φ(x)|p p|φ(x)|0 = = 0|φ+ (x)|p = d3 k 0|a(k)a† (p)|0 e−i k·x = e−i p·x (2π)3 2Ek (7. |k = a† (k)|0 .44) p|φ− (x)|0 = ei p·x .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.38) d4 k 2π δ(k 2 − M 2 ) θ(k 0 ) |k k| = (2π)4 (7. (2π)3 2Ek (7.37) (7. (2π)3 (7.45) (7.36) and satisfying the completeness condition 1 = k|k ′ = (2π)3 2Ek δ 3 (k − k′ ). i.39) The problem with the vacuum or zero-point energy. and multiparticle states |(k1 )n1 (k2 )n2 .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). in field theory also referred to as vacuum energy Evac = 1 V 2 d3 k Ek . n1 ! n2 ! (7. (2π)3 2Ek (7. k 0 = Ek = k2 + M 2 ). defining particle states |k = |k (with positive energy. is solved by subtracting it as an (infinite) constant. The rest of the states are then obtained from the groundstate via the creation operator. This term will be adressed below.

54) = i∆(x − y) = i (∆+ (x − y) + ∆− (x − y)) . (7. (iv) The equal time commutation relations follow from ∂ ∆(x) ∂t (v) For M = 0.47) ′ d3 k ′ e−i k·x+i k ·y [a(k). (iii) ∆(0. x) = 0 and hence ∆(x) = 0 for x2 < 0.53) t=0 .51) (7.49) (7. 2π (7. a† (k ′ )] 3 2E ′ (2π) k [φ− (x). d3 k ei k·(x−y) = −i∆+ (y − x).56) = −δ 3 (x). i. (2π)3 2Ek ≡ i∆− (x − y). (2π)4 (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. = − (2π)3 2Ek (7.55) d4 k ǫ(k 0 ) 2π δ(k 2 − M 2 ) e−i k·x . (ii) ∆(x) is a solution of the homogeneous Klein-Gordon equation. The last item to be checked for the scalar field are the causality condition. i∆+ (x) i∆− (x) = = = d4 k θ(k 0 ) 2π δ(k 2 − M 2 ) e−i k·x . φ+ (y)] d3 k e−i k·(x−y) .48) (7. In order to calculate [φ(x).50) or as integrals over d4 k.52) The result for the invariant commutator function is [φ(x). d3 k = (2π)3 2Ek = (7. ∆(x) = − ǫ(x0 ) δ(x2 ). that they satisfy the required commutation relations in Eqs (7. (7. φ− (y)] ≡ i∆+ (x − y). (2π)4 d4 k − θ(k 0 ) 2π δ(k 2 − M 2 ) ei k·x (2π)4 d4 k − θ(−k 0 ) 2π δ(k 2 − M 2 ) e−i k·x = −i∆+ (−x) (2π)4 (7. φ(y)] consider µν [φ+ (x). φ(y)] which has the properties (i) i∆(x) = i∆+ (x) + i∆− (x) can be expressed as i∆(x) = where ǫ(k 0 ) = θ(k 0 ) − θ(−k 0 ).18).17) and (7.e.

particles (created by a† ) and antiparticles (created by b† ) with the same momentum contribute equally to the energy.58) and satisfy the equal time commutation relation (only nonzero ones) [φ(x). extending it with the charge operator and the introduction of particle and antiparticle operators. (7.62) (7.4 The complex scalar field In spite of the similarity with the case of the real field.e. φ† (y)] = i∆(x − y). The field satisfies the Klein-Gordon equation and the density current (U (1) transformations) and the energy-momentum tensor are jµ Θµν The quantized fields are written as φ(x) = φ† (x) = d3 k a(k) e−i k·x + b† (k) ei k·x .65) 7. which is equivalent to the relations (only nonzero ones) [a(k). = ∂{µ φ∗ ∂ν} φ − L gµν . 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) . 2 From the lagrangian density (7. we will consider it as a repetition of the quantization procedure.5 The Dirac field L = ↔ i ψγ µ ∂µ ψ − M ψψ. (2π)3 2Ek (7.64) The commutator of φ and φ† is as for the real field given by [φ(x).63) i.Quantization of fields 60 7. (2π)3 2Ek (7.61) (7. (2π)3 2Ek d3 k b(k) e−i k·x + a† (k) ei k·x . (2π)3 2Ek (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) . Also the charge operator requires normal ordering (in order to give the vacuum eigenvalue zero). b† (k ′ )] = (2π)3 2Ek δ 3 (k − k′ ). a† (k ′ )] = [b(k). ∂0 φ† (y)]x0 =y0 = i δ 3 (x − y).57) (7. ↔ (7.66) .60) = i φ∗ ∂µ φ.59) (7.

s)e−i k·x . s)ei k·x + d(k. ↔ i ψγµ ∂ν ψ − 2 (7.78) (7. {b(k. s)v(k. ˙ δ ψ(x) ↔ i Θ00 (x) = − ψγ i ∂i ψ + M ψψ 2 = iψγ 0 ∂0 ψ = iψ † ∂0 ψ. s′ )} = {d(k.79) Also for the field and the canonical conjugate momentum anticommutation relations are considered. d† (k ′ .71) (7. s) e−i k·(x−y) j s † + vi (k.73) (7. (7. s)ei k·x . u v (2π)3 2Ek (7. (2π)3 2Ek (7. s)vj (k. interchanging creation and annihilation operators. The solution is the introduction of anticommutation relations.74) (7. s′ )} = (2π)3 2Ek δ 3 (k − k′ ) δss′ .68) The canonical momentum and the hamiltonian are given by Π(x) H (x) = = δL = i ψ † (x).70) where the last line is obtained by using the Dirac equation.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. † {ψi (x). s)¯(k.Quantization of fields 61 the conserved density and energy-momentum currents are easily obtained. ψj (y)}x0 =y0 = d3 k (2π)3 2Ek ui (k. b† (k ′ .e. s)e−i k·x + d† (k. ∂ 2 ↔ (7. i. (2π)3 2Ek (7. s)¯(k. jµ Θµν = = ψγµ ψ. The quantized fields are written ψ(x) = s d3 k b(k. s). s)u† (k. (2π)3 2Ek d3 k b† (k.75) (7. s).67) i ψ / ψ − M ψψ gµν . s)u(k.77) Note that achieving normal ordering. .69) (7. (7.72) ψ(x) = s In terms of the operators for the b and d quanta the hamiltonian and charge operators are (omitting mostly the spin summation in the rest of this section) H = = Q = = d3 x : ψ † (x) i∂0 ψ(x) : d3 k Ek : b† (k)b(k) − d(k)d† (k) : (2π)3 2Ek d3 x : ψ † ψ : d3 k : b† (k)b(k) + d(k)d† (k) : . s) ei k·(x−y) x0 =y 0 . 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) .

6 The electromagnetic field 1 L = − Fµν F µν . as the vanishing of Π0 induces a constraint. It is possible to continue in a covariant way with the lagrangian 1 λ L = − Fµν F µν − (∂µ Aµ )2 .87) If one wants to impose canonical commutation relations ∂µ Aµ = 0 cannot hold as an operator identity. 4 Π0 Πi δL = 0. 4 2 This gives the equations of motion discussed before. it implies the Lorentz constraint and leads to the canonical momenta Π0 Π i = = E i. but has the problem of being noncovariant.84) (7. ∂ (7. we would have obtained [ψi (x). ∂ (7. −λ(∂ρ Aρ ). ψ j (y)] = (i/x + M )ij i∆1 (x − y). ψ j (y)} = d3 k (/ + M )ij e−i k·(x−y) + (/ − M )ij ei k·(x−y) k k (2π)3 2Ek d3 k e−i k·(x−y) + −ei k·(x−y) = (i/x + M )ij ∂ (2π)3 2Ek = (i/x + M )ij i∆(x − y).83) From the lagrangian density the canonical momenta are = = (7. 7. ˙ δ Ai (7. ψj (y)}x0 =y0 = d3 k (/ + M )il (γ0 )lj e−i k·(x−y) k (2π)3 2Ek + (/ − M )il (γ0 )lj ei k·(x−y) k x0 =y 0 = d k 2Ek (γ0 γ0 )ij ei k·(x−y ) (2π)3 2Ek (7. (7. When we would have started with commutation relations for the field ψ and the canonical momentum.Quantization of fields 62 Using the positive and negative energy projection operators discussed in section 4.80) 3 = δ 3 (x − y) δij . For the scalar combination {ψi (x). one has † {ψi (x). ψ j (y)} at arbitrary times one has {ψi (x).85) which reflects the gauge freedom.81) where i∆ = i∆+ + i∆− is the same invariant commutator function as encountered before.86) (7. but we must restrict ourselves to the weaker condition B|∂µ Aµ |A = 0. which however has wrong causality properties! Therefore the relation between spin and statistics is required to get micro-causality (which is also known as the spin statistics theorem).88) .82) where i∆1 = i∆+ −i∆− . (7. ˙ δ A0 δL = F i0 = E i .

Aν (y)] = [Πµ (x).93) which does exhibit problems with the time-like photon.1 (optional) If you haven’t seen Poisson brackets in classical mechanics you should (a) Proof that Poisson brackets (Eq. ′ (7. i. 7. one time-like photon by itself is not allowed! This solves the problems with the normalization and the negative energies. µ λ=0 (7.3) satisfy the conditions of a Lie algebra.e. In fact the commutation relations imply ˙ [Aµ (x). λ) − c† (k. The quantized field is expanded as Aµ (x) = d3 k (2π)3 2Ek (λ) 3 ǫ(λ) (k) c(k.92) Note that for the transverse states there are no problems with the normalization and the statistics of the states as −g ij = δ ij for i. λ)ei k·x . 0) . pj ]P . λ)e−i k·x + c† (k. 0. rj ]P . where Aµ + + is the part of the vector field containing the annihilation operators. Q]P . a longitudinal photon and two transverse photons. which reproduce the Hamilton equations. and [ℓi . Πν (y)] = 0. k 3 ) this reads |k 3 |c(0) (k) − k 3 c(3) (k) |A = 0 c(0) (k) ∓ c(3) (k) |A = 0. for which it is sufficient that ∂µ Aµ |A = 0. (d) Use Poisson brackets to calculate q and p. (c) Calculate the Poisson brackets for the coordinates x and the standard conserved quantities p and ℓ. 0)c(k.89) with four independent vectors ǫµ . (7. 6. The canonical equal time commutation relations are [Aµ (x). [ℓi . pj ]P . Πν (y)]x0 =y0 = i gµν δ 3 (x − y). q) that dO/dλ = [O. λ′ )] = −g λλ 2Ek (2π)3 δ 3 (k − k′ ). ˙ ˙ . where Q is the conserved quantity in Eq.94) Choosing k µ = (|k 3 |. 0. (b) Proof for a quantity O(p. c† (k ′ .11. i. 2. λ)c(k.91) (7. (7.95) Exercises Exercise 7. λ).e. j = 1. µ λ=0 (7. The hamiltonian in terms of the creation and annihilation operators is (after normal ordering) given by H= d3 k Ek (2π)3 2Ek 3 λ=1 c† (k. what are [ri . These problems are solved by the Lorentz constraint between physical states given above. It gives 3 k µ ǫ(λ) (k)c(k.Quantization of fields 63 for physical states |A and |B . (7. Aν (y)]x0 =y0 = i gµν δ 3 (x − y).90) where Πν = F ν0 − λ g ν0 (∂ρ Aρ ) and we have furthermore [Aµ (x). containing a time-like photon. λ)|A = 0. and are equivalent with [c(k.

Is the subtraction of ’zero point’ contributions essential in this check? µ Pµ φ(x) ei a µ Pµ = φ(x − a) Exercise 7. f (Λk) = f (k) for Lorentz transformations Λ.3 (a) Show that the correct implimentation of symmetries in the Hilbert space.Quantization of fields 64 Exercise 7. ∆R = 1 (2π)4 d4 k e−ikx k2 1 − M 2 + iǫk 0 (with the implicit prescription to take ǫ → 0 after integration).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. Pµ ] for the (real) scalar field using the expressions for the fields and momentum operator in terms of creation and annihilation operators. . (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. Give also the expression for ∆C . (2 + M 2 ) ∆inhom (x) = −δ 4 (x). (2 + M 2 ) ∆hom (x) = 0. e−i a requires [φ(x). (e) Show that ∆R (x) = 0 if x0 < 0. (b) Check the above commutation relation [φ(x). we can write for ∆R an integral where the k 0 integration remains along the real axis. (Similarly one can show that for the advanced Green’s function ∆A (x) = 0 if x0 > 0). Pµ ] = i∂µ φ(x). ∆hom (Λx) = ∆hom (x) if f is an invariant function. (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. It is invariant.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.

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

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

λ)γ0 v(k.12) and for the vector field −1 Aµ (x) −→ Pop Aµ (x) Pop = −Aµ (˜). λ)v(k. op P (8.5) (8.4) (8. −λ) eik·x 3 2E (2π) ˜ k d3 k ˜ ˜ ηP b(k. if helicity λ is used instead of the z-component of the spin m. The result is ψ p (x) −1 = Pop ψ(x) Pop = ηP γ0 ψ(˜) x (8. λ)u(k. m) = = ˜ u(k. x The latter behavior of the vector field will be discussed further below. λ) Pop = ηP b(k. λ)γ0 u(k. −λ) eik·x (2π)3 2Ek ˜ d3 k ˜ ˜ ˜ ˜ ηP b(k. −λ).Discrete symmetries 67 After parity transforming x to x the Dirac equation becomes after some manipulations ˜ ˜ x iγ µ ∂µ − M ψ(˜) = 0. −λ). −λ) e−ik·x − d† (k.7) = λ d k x x ηP b(k. m). In the same way as the Fermion field. For the scalar field we have seen −1 φ(x) −→ Pop φ(x) Pop = ηP φ(˜). γ x (8. λ) eik·x .8) 3 = λ = λ = λ From this one sees immediately that −1 ˜ Pop b(k.10) i. choosing ηP is real (ηP = ±1) particle and antiparticle have opposite parity. one can also consider the scalar field and vector fields. (2π)3 2Ek (8. λ) P −1 = −η ∗ d(k. λ)u(k. iγ µ† ∂µ − M ψ(˜) = 0. (i˜ µ ∂µ − M ) ψ(˜) = 0.2 Charge conjugation We have already seen the particle-antiparticle symmetry with under what we will call charge conjugation the behavior T (8. λ) eik·˜ (2π)3 2Ek d3 k ˜ ˜ ˜ ˜ ηP b(k. ˜ Pop d(k.13) ψ(x) −→ ψ c (x) = ηC C ψ (x). λ) e−ik·˜ + d† (k. m) γ0 v(k. x It is straightforward to apply this to the explicit field operator ψ(x) using γ0 u(k. x µ (iγ ∂µ − M ) γ0 ψ(˜) = 0. x (8. m). −λ)u(k.6) (check this for the standard representation.9) (8.3) (8. −λ)v(k.e. λ)v(k. the above operation reverses the sign of λ).11) (8. . 8. x Therefore γ0 ψ(˜) is again a solution of the Dirac equation and we have x ψ p (x) = γ0 ψ(˜). λ) e−ik·x − d† (k. −λ) e−ik·x − d† (k. ˜ −v(k.

(8. x x (8.e. writing −1 ψ(x) −→ Top ψ(x)Top = ηT Aψ(−˜) ≡ ψ t (−˜). As time reversal will transform ’bra’ into ’ket’. λ) Cop = ηC d(k. λ) e−ik·x + b† (k. x (8.19) Cop d(k. m).15) (where one must be aware of the choice of spinors made in the expansion.16) d k ηC d(k. x (8.Discrete symmetries 68 the latter being also a solution of the Dirac equation. r). λ). antilinear operator is anti-unitary if A† = A−1 .20) The time reversal operator transforms We will again consider the transformation properties for a fermion field ψ(x). m) ¯ = = v(k. we start with the complex conjugated x Dirac equation for ψ. (8. λ) eik·x ¯ ¯ (2π)3 2Ek d3 k ηC d(k. The (time-reversed) Dirac equation becomes. λ). x µ (iγ ∂µ − M ) γ5 Cψ(−˜) = 0. ˜ x −(iγ µ )∗ ∂µ − M ψ(−˜) = 0. The action on the spinors (using C = iγ 2 γ 0 = −iρ1 σ 2 in standard representation) gives C uT (k. One has Aφ|Aψ = φ|ψ ∗ = ψ|A† Aφ = ψ|φ . λ)C uT (k.21) where A is a 4 × 4 matrix acting in the spinor space. λ) eik·x .22) = Aψ|Aφ ∗ is antilinear if A(λ|φ + µ|ψ ) = λ∗ A|φ + µ∗ A|ψ . The same relations hold for helicity states. x iγ µT ∂µ − M ψ(−˜) = 0. (2π)3 2Ek (8.3 Time reversal xµ = (t. r) −→ −˜µ ≡ −xµ = (−t.18) (8. Therefore ψ c (x) = = λ −1 ¯ Cop ψ(x) Cop = ηC C ψ T (x) (8.14) (8. λ)u(k. (i˜ µ∗ ∂µ − M ) ψ(−˜) = 0. . λ) −1 Cop = ∗ ηC b(k. Norm conservation requires Top to be anti-unitary2 . For a quantized field one has −1 ∗ −1 Top fk (x)bk Top = fk (x)Top bk Top . i. m) ¯ C v T (k. ((iγ µ )∗ ∂µ − M ) ψ(x) = 0. u(k. 8. it is antilinear1 . λ)v(k. λ)C v T (k. λ) e−ik·x + b† (k. Top |φ = φt | = (|φt )∗ . m).17) 3 = λ This shows that −1 Cop b(k. γ x i(γ5 C)−1 γ µ γ5 C∂µ − M ψ(−˜) = 0. to find ψ t (−˜) that is a solution of the Dirac equation. as discussed in section 4). x 1A 2 An −iC −1 γ µ C∂µ − M ψ(−˜) = 0.

˜ = 0|b(k)|A 8. −p. λ) eik·x + d† (k. op T (8. −λ a T a. λ) e−ik·x 3 2E (2π) k d3 k ˜ ˜ ηT b(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. −p. λ). Therefore one has A|k = A|b† (k)|0 = A|T † T b† (k) T † T |0 ∗ = At |b† (k)|0 ∗ ˜ t = k|At . λ) eik·x + d† (k.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. 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(−˜).32) (8.34) .26) d k x x ηT b(k. λ| ¯ C |¯. p.Discrete symmetries 69 Table 8. λ a |a. λ) i γ5 C v(k. λ| a.2: The transformation properties of physical states for particles (a) and antiparticles (¯). −p. ∗ ˜ v (k. Since there are 16 independent 4 × 4 matrices.4 Bi-linear combinations In quantities such as currents and lagrangians often bilinear combinations of spinor fields are encountered. λ) e−ik·x 3 2E (2π) k ˜ d3 k ˜ ˜ ˜ ˜ ηT b(k. λ). λ)u∗ (k. a state |a.24) (8.30) (8. λ) Top = ηT b(k. λ)u∗ (k. λ) e−ik·˜ (2π)3 2Ek d3 k ˜ ˜ ˜ ˜ ηT b(k. ˜ Top d(k. p.25) (check this for the standard representation). (8. λ) T −1 = η ∗ d(k.31) (8. Note that applying an anti-unitary transformation such as Top one must take for the matrix element ˜ the complex conjugate. λ).27) 3 = λ = λ = λ From this one obtains −1 ˜ Top b(k. λ)v ∗ (k. λ a P |a. λ).28) (8. −p. p. λ |¯. λ)v ∗ (k. λ)iγ5 Cv(k. λ)u∗ (k. λ)iγ5 Cu(k. The result is = = λ (8. λ)v ∗ (k.33) (8.29) In table 2 the behavior of particle states under the various transformations has been summarized. −λ |¯. λ) eik·x + d† (k. λ) ψ t (x) −1 Top ψ(x) Top = i ηT γ5 Cψ(−˜) x = = ˜ u∗ (k. λ) eik·˜ + d† (k. λ) e−ik·x (2π)3 2Ek (8. p. x (8. there are 16 independent of these bilinear combinations.

γ5 γ µ or γ5 ). in this case only q 2 = (p′ − p)2 . Note that the lagrangian density L (x) → L (−x) under Θ. For instance for nucleons one has Γµ (p′ . (8.Discrete symmetries 70 Table 8. (8.37) (8.40) where the coefficients are functions of invariants constructed out of the momenta. 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 ). 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 . In many applications the expectation values of currents are needed. s . and T. s′ |V µ (x)|p. σ µν . e. called form factors. ¯ ′ ′ (8. V µ (x) = ei Pop ·x V µ (0)e−i Pop ·x . for the vector current V µ (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).3: The behavior of the independent bilinear combinations of fermi fields under P. F.36) (8.. (q 2 ). ¯ V µ (x) =: ψ(x)γ µ ψ(x) :. p) = γµ F1 (q 2 ) + iσµν q ν F2 (q 2 ). as well as under the combined operation Θ = PCT (see Table 3). (q 2 ). 8. As an example consider the vector current for a point fermion.g. p) can be built from any combination of Dirac matrices (1. of which the expectation value between momentum eigenstates can be simply found. C. γ µ . The 16 combinations of Dirac matrices appearing above are linearly independent. γ ν ]. 2M (8.38) (8. Applying the results from the previous sections it is straightforward to determine the behavior of the combinations under P. p′ |V µ (x)|p = u(p′ )Γµ (p′ . C. p′ |V µ (0)|p = u(p′ )γ µ u(p).39) where Γµ (p′ . ¯ In general the expectation value between momentum states can be more complicated..5 Form factors Currents play an important role in field theory.41) 2M 2M . momenta (pµ or p′µ ) or constant tensors (ǫµνρσ or gµν ). p′ . This implies p′ |V µ (x)|p = e−i (p−p )·x p′ |V µ (0)|p . p)u(p) e−i (p−p )·x . F.35) The x-dependence can be accounted for straightforwardly using translation invariance.

45) Exercises Exercise 8. p)u∗ (˜) up p µ p ˜ = ei q·x u(p′ ) (iγ5 C)Γ∗ (˜′ . p′ )γ0 u(p). p) that γ0 Γµ† (p. p′ )u(p′ ) = u† (p′ )Γµ† (p. p) that Γµ (p′ .Discrete symmetries 71 but that are eliminated because of relations between Dirac matrices. p)γ0 . ¯ µ p ˜ Therefore time reversal invariance implies for the structure of Γµ (p′ . p)u(˜) ˜ x p ¯ p p ˜ p (8. p) = γ0 Γµ (˜′ .g. ¯ 2M (8. p).42) where q = p′ − p. Therefore hermiticity implies for the structure of Γµ (p′ . or relations based on hermiticity of the operator or P .43) = ei q·x u(p′ )γ0 Γµ (˜′ . p) = (i γ5 C)Γ∗ (˜′ .g. p)u(p) ¯ † † = p′ |Top Top V µ (x)Top Top |p ∗ (8. e. µ p ˜ (8.44) = p′ |Vµ (−˜)|˜ ˜ x p ∗ ˜x = ei q ·˜ [¯(˜′ )]∗ Γ∗ (˜′ .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′ |Pop Pop V µ (x)Pop Pop |p ˜x = p′ |Vµ (˜)|˜ = ei q ·˜ u(˜′ )Γµ (˜′ . p′ )u(p′ )) u = ∗ † u† (p)γ0 Γµ (p. p)γ0 u(p). p′ )γ0 = Γµ (p′ . p ˜ • Time reversal p′ |V µ (x)|p = ei q·x u(p′ )Γµ (p′ . p)u(p) ¯ = p|V µ (0)|p′ ∗ = (¯(p)Γµ (p. • Parity p′ |V µ (x)|p = ei q·x u(p′ )Γµ (p′ . γ5 σµν = (1/2)ǫµνρσ γ ρ γ σ . or relations that follow from the equations of motion. p) that Γµ (p′ . e. p)(iγ5 C)u(p). ¯ p ˜ Therefore parity invariance implies for the structure of Γµ (p′ . the Gordon decomposition u(p′)γ µ u(p) = ¯ 1 u(p′ ) [(p′ + p)µ + iσ µν qν ] u(p). • Hermiticity p′ |V µ (0)|p = p′ |V µ† (0)|p = u(p′ )Γµ (p′ . C and T invariance. p)(i γ5 C).

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

≡ 0. t′ ≡ q ′ |q ′ . in which the operators are time-independent. i ∂ |ψS ∂t ∂ i AS ∂t = H |ψS .1) The time evolution from t0 → t of a quantum mechanical system is generated by the Hamiltonian. in which the states are time-independent. t0 ). QH (t)|q.5) (9. t ′ QH (t′ )|q ′ .1 Time evolution as path integral U (t. t0 ) AH (t0 ) U (t. and the operators are time-dependent. (9.2) (ii) Heisenberg picture. |ψS (t) = U (t. [AH .Chapter 9 Path integrals and quantum mechanics 9. t ≡ q|q. t′ .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. AS (t) = AS and the states o are time dependent. |ψH (t) = |ψH . (9. H]. Consider the two (time-independent) Heisenberg states: |q. AH (t) = U −1 (t.7) 73 . (9. t0 ) ∂t (9. t0 ) = H U (t. t . (9. t0 ) = e−i(t−t0 )H or i Two situations can be distinguished: (i) Schr¨dinger picture. t0 )|ψS (t0 ) . t ′ |q .3) (9. i ∂ |ψH ∂t ∂ i AH ∂t ≡ = 0.4) ∂ U (t.

tj+1 |qj . 2π (9. qj )]) . The hamiltonian is an operator H = H(P. n |q (9. ˙ 2π 1 1 1 [A. (9.qj )∆τ e e 2π dpj i pj (qj+1 −qj )−i ∆τ H(pj .Path integrals and quantum mechanics 74 and Schr¨dinger states o |q |q ′ QS |q ≡ q|q . . ′ (9.. eA eB = eC with C = A+B+ (9.q . .14) . t = q ′ |e−i H(t −t) |q . (9. B] + . |q1 q1 |e−i H∆τ |q . with the correction1 being of order (∆τ )2 . tj+1 |qj . B] + [A. t = = q ′ |e−i n H∆τ |q = q ′ | e−i H∆τ dq1 .13) At least for a simple hamiltonian such as consisting of a kinetic energy term and a local potential. B]. 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 . 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|. t′ |q.11) (9. which is a hamiltonian function in which the operators are replaced by real-numbered variables. B]] + [[A. [A. Combining the two terms gives qj+1 . . 2π (9.15) By letting the kinetic and potential parts act to left and right respectively one can express the expectation values in terms of integrals containing H(pj .9) 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 ′ . 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 . qj ). 2 12 12 (9.qj ) e 2π dpj exp (i ∆τ [pj qj − H(pj .8) Choose t as the starting point with |q = |q..16) this. t and QH (t) = QS and study the evolution of the system by calculating the quantum mechanical overlap amplitude q ′ . . The purpose of this is to calculate the evolution for an infinitesimal time interval. use the Campbell-Baker-Hausdorff formula.10) dqn−1 q ′ |e−i H∆τ |qn−1 qn−1 | .12) where the transformation between coordinate and momentum space involves q|p = ei p. Q) expressed in terms of the operators P and Q. QS |q ′ ≡ q ′ |q ′ . tj = = = 1 For dpj i pj (qj+1 −qj ) −i H(pj . t′ |q. Then qj+1 . H(P.

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

24) (9. Note that procedure (i) is an example of the more general procedure under (ii). Physical answers will usually come out as the ratio of two functional integrals and are thus independent of the chosen measure.n fm (x) Kmn fn (y) and consider F [α] = F (αn ) as a multivariable function.Path integrals and quantum mechanics 76 (ii) Write α in terms of a sum of orthonormal basis functions. 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. the following integrals can be derived for a real symmetric or complex hermitean kernel K.22) = x The last equality is obtained by defining the right measure (normalization N ) in the integration. 2 (9. N (9.21) again a multidimensional integral. (9. In both cases N is an appropriately choosen normalization constant in such a way that the integral is finite. Dα F [α] = Dα F [α + β].23) Having defined the Gaussian integral.27) (9. . A useful property of functional integration is the translation invariance. (9. Consider the gaussian functional as an example. y) = ∗ m.26) As an important application of translation invariance we mention the identities 1 Dα exp − α2 − ωα 2 = exp 1 2 ω . α(x) = n αn fn and K(x.25) (see Exercises). with Dα F [α] = n dαn N F (αn ) (9.28) Dα Dα∗ exp (−α∗ α − α∗ ω − ω ∗ α) = exp (ω ∗ ω) .n dx fm (x)fn (x) = n = n dαn 1 exp − α2 N 2 n √ 2π ≡ 1. 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. 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.

F [θ. i. y) θ(y) . In principle the definitions of functionals is the same. δα(x) (9. −→ fn (x) δα(x) ∂αn ∂αm n Note that δ(x − y) = n ∗ fn (x) fn (y).32) (9. n θn fn (x) dx θ∗ (x) θ(x) .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. (9. e. however. α(y) = δ(x − y). θ2 = 0.38) . δα(x) 2 2 δ exp (−α β) = −β(x) exp (−α β) .30) (9. products of the same pair appear.34) (9. which vanish. −→ δα(x) ∆x ∂αx ∂αx δ ∂ ∂ ∗ with . the Gaussian-type functionals. we need to consider Grassmann-valued functions involving anticommuting Grassmann variables. (9. θ∗ ] = exp (−θ∗ K θ) ≡ exp − Integration for Grassmann variables is defined as dθ 1 = 0 and dθ θ = 1. Examples of functional differentiation are (9. θ∗ ] = exp (−θ∗ θ) ≡ exp − F [θ. δα(x) δα(x) δ αKβ = dy K(x. dx dy θ∗ (x) K(x.33) (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. y) β(y) δα(x) δ 1 1 exp − α2 = −α(x) exp − α2 . In the expansion of the exponential (from second to third line). It is now easy to check that Dθ∗ Dθ exp (−θ∗ K θ) = det K. θη = −ηθ. αn = δmn .31) δ δ α = dy α(y) = dy δ(x − y) = 1. (9.35) For applications to fermion fields.29) (9. δα(x) or in discretized form δ 1 ∂ ∂ with .Path integrals and quantum mechanics 77 Functional differentiation is defined as δ .g.e. αy = δxy .

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

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

9.6 Euclidean formulation The above expression (Eq. T′ Z[J] = lim T ′ → −i∞ T → i∞ Dq exp i T dτ L q. t|Q. . . J=0 (9. The importance of Z[J] is that the time ordered product of operators can be obtained from it.70) one immediately finds δ n Z[J] δJ(t1 ) .71) 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 J lim −i E0 (T ′ −T ) φ (Q′ ) φ∗ (Q) 0 0 T ′ → −i∞ e T → i∞ dq ′ dq φ∗ (q ′ . t 0 0out |0in J . 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. Better is the use ˜ of imaginary time t = −it. t′ |q. Q(tn )|0 . δJ(tn ) hence the name generating functional. T = φ0 (q.67) = = φ0 (q.70) for the generating functional is in fact ill-defined. t) (9.69) The factor that has been divided out in the first line of this equation is a numerical factor depending on the boundaries of the space-time volume (T and T ′ ). . T ′ |Q. ˜ t′ → ∞. t). . J (9. thus T →i∞ lim q. T → i ∞ and T ′ → −i ∞.Path integrals and quantum mechanics t 82 T t t’ T’ Figure 9.68) (9. = (i)n 0|T Q(t1 ) . (9. such that t′ → −i∞ t → i∞ ↔ ↔ ˜ t → −∞. . t′ ) q ′ . Since we have (neglecting multiplicative factors). dq dτ + J(τ )q(τ ) . in which case eiEn T → e−En ·∞ and the term eiE0 T is dominant.

. . i τ −∞ ∞ −∞ dq d˜ τ + J(˜)q(˜) τ τ dq d˜ τ − J(˜)q(˜) τ τ .76) which is a positive definite quantity. for Grassmann valued functions.72) (9.2 Check the examples of functional derivation for real-. ˜ ˜ ZE [J] δJ(t1 ) .77) where the expressions have been divided by Z[J] and ZE [J] in order to get rid of dependence on multiplicative factors. i dq . + V (q).75) = − LE q.1 Convince yourself by choosing an appropriate (orthonormal) set of functions that 1 Dα exp − α K α 2 for real-valued functions. where Kp are the eigenvalues of the matrix Kmn in K(x. δJ(tn ) = (i)n J=0 1 δ n ZE [J] . dq ˜ dt = 1 2 dq ˜ dt dq ˜ dt 2 − V (q). (9. e. dq dt = 1 2 1 2 dq dt 2 Dq exp Dq exp − ∞ d˜ L q. . 1 = √ .and Grassmann-valued functions. The differentiations in Z[J] and ZE [J] are related. δJ(tn ) J = 0 ˜ t = it (9. when L q. det K Exercise 9.74) − V (q) 2 (9.g. ZE [J] = = where LE q.73) d˜ LE q. det K 1 .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. complex. Exercises Exercise 9. (9. . ensuring convergence for the functional integral ZE [J].Path integrals and quantum mechanics 83 In terms of the imaginary time we can write the Euclidean generating functional. ˜ dt (9. y) = ∗ m. δ n Z[J] 1 Z[J] δJ(t1 ) . . In all of these cases the determinant of the (hermitean) operator K is defined as det K ≡ p Kp . As discussed in the previous section the interesting quantities are obtained from functional differentiation with respect to the sources. τ dq ˜ dt ≡ −L q.

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. where dy K(x. 2 (9.80) Dα∗ Dα exp (−α∗ K α − α∗ ω − ω ∗ α) = 1 exp ω ∗ K −1 ω .79) (9. y) K −1 (y.3 Prove the following relations 1 Dα exp − α K α − αω 2 1 = √ exp det K 1 ω K −1 ω . z) = δ(x − z).78) (9.

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

(2π)4 (k 0 + iǫ)2 − E 2 (10.9) i. with Z0 [0] = 1. . As discussed in the previous section (exercise 9. .6) the generating functional is (using ∂µ = − ∂µ ) given by   Z0 [J] =  Dφ exp −i where Kxy = (∂µ ∂ µ + M 2 ) δ 4 (x − y). . .11) Depending on the path choosen in the complex k 0 plane (see exercise 7. = 0 2 2 2 k2 − M 2 (k ) − E 2 (k ) − k − M (10. (10.8)  1 d4 x  φ (∂µ ∂ µ + M 2 ) φ − Jφ 2 K  (10.4) one distinguishes the retarded Green’s function ∆R (x) = d4 k e−i k·x . These Green functions appear in the expansion Z[J] = n in n! d4 x1 . J(xn ). .10) J=0 In order to determine ∆F . . xn ) = x2 xn . which can be written as ∆(x) = ˜ with (k 2 − M 2 )∆(k) = 1 or ˜ ∆(k) = 1 1 1 = 0 2 . .12) d4 k −i k·x ˜ e ∆(k) (2π)4 (10. ∆F (the socalled Feynman propagator) is the Green’s function of the Klein-Gordon equation. d4 xn G(n) (x1 .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 . 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. 2 2 → (10. xn ) J(x1 ) . Furthermore. (10.3) one finds 1 Z0 [J] = exp − i 2 d4 x d4 y J(x) ∆F (x − y) J(y) ≡ exp (i W0 [J]) . .7) where we made the choice Z0 [0] = 1 and ∂µ ∂ µ + M 2 ∆F (x − y) = −δ 4 (x − y). .e. consider the general solution of ∂µ ∂ µ + M 2 ∆(x) = −δ 4 (x). . . .Feynman diagrams for scattering amplitudes 86 for which we introduce the picture x1 G(n) (x1 .

The first possibility is to consider the well-defined Euclidean formulation. the advanced Green’s function ∆A (x) = satisfying ∆A (x.18) Thus we see various solutions. µ 1 1 • Firstly.21) ∆F (x) = = −i 0 kE =−∞ d4 kE e−i kE ·xE 2 (2π)4 kE + M 2 (10. (10. i. φ(0)] = i∆+ (x) + i∆− (x) d3 k e−i k·x = [φ+ (x).e. φ+ (0)] = (2π)3 2E are also shown in exercise (7. 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. (2π)4 k 2 − M 2 (10. φ− (0)] = (2π)3 2E d3 k e−i k·x = [φ− (x). 4 k 2 − M 2 + iǫ (2π) (10. 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).14) They are solutions of the inhomogeneous equation. t) = 0 for t < 0. ∆(x) = − d4 k e−i k·x .20) = exp − 2 where or (see fig. (2π)4 k 2 − M 2 (10. t) = 0 for t > 0. 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).15) e−i k·x d4 k .4). the second is to explicitly consider 0|T φ(x1 ) φ(x2 )|0 .17) d4 k e−i k·x . . 10.1 for countours) µ (∂E )2 − M 2 (i ∆F (x − y)) = −δ 4 (x − y) 0 kE =∞ (10.Feynman diagrams for scattering amplitudes 87 satisfying ∆R (x.19) φ ∂E ∂E − M 2 φ + Jφ 2 1 d4 xE d4 yE J(x) (−i ∆F (x − y)) J(y) . (2π)4 (k 0 − iǫ)2 − E 2 (10. (10.22) (10. The contours for i ∆(x) i ∆+ (x) i ∆− (x) = [φ(x). the causal Green’s function ∆C (x) = and finally the Green’s function d4 k e−i k·x = 4 (k 0 )2 − E 2 + iǫ (2π) ∆(x) = PV d4 k e−i k·x .16) C where C is a closed contour in the complex k 0 -plane.23) k0 =i∞ k0 =−i∞ d4 k e−i k·x . (2π)4 k 2 − M 2 where the latter contour can be deformed to the contour for ∆C . ∆F = ∆C . The solution of the homogeneous equation ∂µ ∂ µ + M 2 ∆(x) = 0 can also be written as an integral in k-space. the generating functional can be written 1 µ µ Z0 [J] = Dφ exp d4 xE (10.

25) x1 x2 x3 = x4 + + . because this is the only quantity entering Z0 [J]. we have G0 (x1 . (4) (2) It should be clear that G0 can be expressed in terms of G0 = i ∆F . or diagrammatically (4) (10. Neglecting multiplicative factors or equivalently assuming that Z0 [0] = 1. the same result as above. φ− (y)]|0 − θ(y 0 − x0 ) 0|[φ− (x). . We find (see also section 7. x2 .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. . φ+ (y)]|0 = θ(x0 − y 0 ) i ∆+ (x − y) − θ(y 0 − x0 ) i ∆− (x − y) (10.3) 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). 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. δJ(x4 ) 2 J=0 = − [∆F (x1 − x2 ) ∆F (x3 − x4 ) + ∆F (x1 − x3 ) ∆F (x2 − x4 ) + ∆F (x1 − x4 ) ∆F (x2 − x3 )] . = θ(x0 − y 0 ) 0|[φ+ (x).1: Contour in the k 0 plane for the Feynman propagator in Euclidean and Minkowski space • As the second possibility. x4 ) = 0|T φ(x1 ) φ(x2 ) φ(x3 ) φ(x4 )|0 1 δ4 exp − i J ∆F J = (−i)4 δJ(x1 ) .24) Knowing the explicit form of Z0 [J] it is straightforward to calculate the 4-points Green’s function.

e. x3 . x4 ) = G0 (x1 . To zeroth order in the coupling one has 0 1 Z (g ) [J] = Z0 [J] = exp − i 2 J ∆F J .g 0 ) = G0 = 1.35) z . Z[J] = = Dφ exp i exp i d4 z LI (10. x4 ) = To first order in g one has Z (g ) [J] = = 1 0 0 x1 + + x2 (10.33) (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 .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) (10.32) (10. ≡ −i g d4 z (10.30) and in zeroth order G(0. (2) (0) (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 .g ) (x1 . the generating functional can be written. x2 ) = G0 (x1 . x3 . x2 ) = i ∆F (x1 − x2 ) = G(4. (10. L (φ) = L0 (φ) + LI (φ).27) When interactions are present.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.Feynman diagrams for scattering amplitudes 89 10.34) Introducing a vertex point to which four propagators are connected.g ) (x1 . (10. (10.31) G(2. x2 .29) This expression will be the one from which Feynman rules will be derived. with propagators (i ∆F ) being connected to vertices (i LI ) according to the above expression for the functional Z[J]. Consider as an example the interaction LI (φ) = − g 4 φ 4! in the scalar field theory discussed sofar. i. x2 .

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

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

(10. in = S|α. . out . out|α. . p. in|p′ . (3) S is unitary (it conserves the scalar product from initial to final state).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 ). Sβα ≡ β. x3 . in into final state particle states (out-states) |β. out = δαβ ↔ SS † = 1. pn . . out = |p′ . in|S = α. in = β. out (suppressing except momenta all other 1 m quantum numbers). (2) (10. implying the absence of tadpoles. Next. one finds for the connected 4-point Green’s function at order g.3 Interactions and the S-matrix The S-matrix The S-matrix transforms initial state free particle states (in-states) |α. in = |p1 . ∂µ ∂ µ + M 2 φin (x) = 0. p′ . in|S = 0. out . in = p. considered explicitly. out| ⇔ |α. Proof: α.e.45) x2 as the only surviving diagram (see exercises). In line with the consideration of the generating functional representing the vacuum to vacuum amplitude we consider fields φin . x1 G(4. out|p′ . x2 . in|S|p′ . .44) J=0 For the interacting theory. the vacuum expectation value of the field φ(x) is zero. in = p. φout and the interpolating field φ(x). then (n) (n) Gsc = Gc for n ≤ 3. . (2) The one-particle state is invariant (conservation of energy and momentum. in = |α. in ⇔ β. in = p. (10. we have W0 [J] = − 1 2 d4 x d4 y J(x) ∆F (x − y) J(y). this will be translated to the action on fields. (10.g ) (x1 . in|SS † |α.g.46) The properties of the S-matrix are (1) The vacuum is invariant or |S00 | = 1.47) . x4 ) = c 1 x3 z (10. 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 . e. translation invariance). in|S = β. in| (choose ϕ0 = 0).Feynman diagrams for scattering amplitudes 92 i. Proof: 0. x4 10. out| and S † |α. out|p′ . out| = eiϕ0 0. out = p|p′ . For the free scalar theory. thus β. . Also for free field (Heisenberg) operators a distinction is made between φin and φout . out|α.

φin (y)] = iZ ∆(x − y).49) The field φ can be expressed in φin/out using retarded or advanced Green’s functions. a)SU −1 (Λ. The relation between S-matrix and in. 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. this can actually not be used as it would imply [φ(x). The time evolution operator (see Eq. 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. The relation between S and Z[J] To establish this relation. z) φ(z). 9.54) . (2π)3 2E (10. √ Although the above. a) = S.51) (10. z) φ(z) d4 y d4 z ∆A (x − y) K(y. the source term Jφ in the lagrangian density is treated in the interaction picture. The convergence therefore must be weakened to t→−∞ √ (10. implies the strong (operator) convergence φ(x) → Zφin (x). in|φin S → φin S = S φout (x). β. a causality condition that can be proven to imply the absence of interactions. as it stands.50) = φ(x) = √ Zφin (x) − √ Zφout (x) − d4 y d4 z ∆R (x − y) K(y.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 ).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). (10. φin (x) = d3 k ∗ a(k) fk (x) + a† (k) fk (x) .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.g. which in a later stage (renormalization) will become more important. S is invariant under e Poincar´ transformations: U (Λ.62) is then T′ UT [J] T′ = T exp i T d4 x J(x)φ(x) .and out-fields is: φin (x) = S φout (x)S −1 . 2 ∂µ ∂ µ + M0 φ(x) = J(x).e.52) where Z is a constant. out|φout = β. Proof: β. e Proof: φin (Λx + a) = U φin (x)U −1 = U Sφout (x)S −1 U −1 = U S U −1 U φout (x) U −1 U S −1 U −1 = U SU −1 φout (Λx + a) U S −1 U −1 ⇔ U SU −1 = S. e. (10. i. (10. φ(y)] = Z [φin (x). in|Sφout Finally we check that as expected S does not spoil Poincar´ invariance.

express it in terms of advanced and retarded Green’s functions. Noting that 0| : eA : |0 = 1 it follows that F [J] = 0|S U [J]|0 = 0|U [J]|0 = 0out |0in ∝ Z[J]. one has [A. φ− f (10. eB ] = [A. B + C]e e . t i Ut∞ [J] φ(x)U−∞ [J] (10. we can.57) i [φin (x).Feynman diagrams for scattering amplitudes 94 and satisfies the property T ′ δUT [J] t = i UtT [J] φ(x)UT [J]. 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. C] are c-numbers. just as we did for φ(x). z) . e e ] = [A. 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.e.61) e φ+ f where the normal ordered expression : e φ f : is used. [A. : =i d4 y ∆(x − y) f (y) : e d4 z φ(z) f (z) :. z) δU [J] . S U [J]] = √ Z δ S U [J].63) . B]eB . (10. z) δU [J] δJ(z) δU [J] d4 y d4 z ∆A (x − y) K(y. B] (for the case that [A.58) In order to find a solution to this equation note that.60) e d4 y φ+ (y) f (y) d4 z φ− (z) f (z) d4 z φ+ (z) f (z) d4 y [φ(x).62) where F [J] is some (arbitrary) functional. φ(x). : e d4 z φ(z) f (z) d4 x φ− (x) f (x) B C B C (10. e = i. Thus S U [J] =: exp 1 √ Z d4 x d4 y φin (x) K(x. which is equal to the expression e in which the creation operators are placed left of annihilation operators. y) δ δJ(y) : F [J]. provided [A. z) d4 y d4 z ∆(x − y)K(y. with the use of the Baker-Campbell-Hausdorff formula e−B A e+B = A + [A. Thus applied to the field φ(x) = φ+ (x) + φ− (x). φ(x). (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).59) (10. δJ(x) ∞ or for U [J] ≡ U−∞ [J] ′ (10. δJ(z) (10. φ(y)] f (y) e e . B] and [A. B] is a c-number). δJ(z) (10.56) Since δU [J]/δJ(x) satisfies the same equations as φ(x) and we know the limits.

. k = k2 i − M 2 + iǫ . They start with the propagator (i∆F (k)) in momentum space. . .64) Therefore. using that φin (x) = d3 k ∗ a(k) fk (x) + a† (k) fk (x) . Usually we are interested in the part of the S-matrix describing the scattering. .e. . i. . = . . . c (n) (10. . . . y) δ δJ(y) : Z[J] Z[0] .2). . S =: exp 1 √ Z d4 x d4 y φin (x) K(x. Explicitly. .e. . Note that the Green’s function connecting the external point xi with the bubble is annihilated by K(x. i.65) we get the Lehman-Symanzik-Zimmerman (LSZ) reduction formula. . u(p) and v (p) for fermions and ǫµ (p) for vector fields). . see exercise 10. fp (x) . 1 for scalar case. . . . . .4 Feynman rules The real scalar field The procedure to obtain the matrix element is commonly summarized by a set of rules known as Feynman rules. It is straightforward to check that the S-matrix element now precisely is given by the amputated Green’s function multiplied with the momentum space wave functions of the particles in initial and final state (by which we refer to the quantities multiplying the plane wave e±i p·x in the field expansion. ∆(pj ) j=1 (10. which is obtained considering only connected diagrams. ←− − → × G(n) (x′ . + pn ) G(n) (p1 . fp′ (x′ ) .. . . . d4 x . iKx′ . ¯ ¯ and the amputated or one-particle irreducible (1PI) Green’s functions   n −i  (n) Γ(n) (p1 . pn ). . . . p′ . . xn ). n (2π)4 δ 4 (p1 + . . (2π)3 2E n (10... J=0 (10.67) where ∆(p) is the Fourier transform of the (full) propagator ∆F (x). . . which is determined by the inverse of the operator found in the quadratic term in the lagrangian i. . (10. pn ) =  G (p1 . pn ) = i=1 d4 xi ei pi ·xi G(n) (x1 . (λ) v(p). iKx′ = (2x′ + M 2 ) precisely annihilating an external propagator i∆F in the Green’s function. Next we introduce the Fourier transform (after extracting a momentum conserving delta function coming from translation invariance. . x. y). Sf i = δf i − i (2π)4 δ 4 (Pi − Pf ) Mf i .+ 1 √ Z ∗ d4 x′ .e. . .Feynman diagrams for scattering amplitudes 95 while for J = 0 one has U [0] = 1. ..δ. u(p).66) where Gc is the connected Green’s function. .68) 10. . . . an S-matrix element between momentum eigenstates in initial and final states is found by considering those source-connected Green’s functions (action of δ/δJ on Z[J]/Z[0]) where the external sources J(xi ) are replaced by the particle wave functions (which are the result of acting with φin (xi ) on momentum eigenstates). . . . .) iKx . |S|p.

69) LI (φ) = − φ3 − φ4 . (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). If problems arise go back to the defining expression for the generating function in 10. Corresponding to external particles wave functions are introduced k k 1 1 In order to calculate the connected amplitude −i Mf i appearing in the S-matrix element these ingredients are combined using Eqs 10. Diagrammar) in the case of the interaction terms g f (10.27. At these vertices each line can be assigned a momentum.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. = −i g (multiplied with 4! corresponding to the allowed number of permutations of identical particles). vertices and external particle wave functions in that diagram. which cancels the factor 1/2 in the quadratic piece. For the interaction terms in the lagrangian. ’t Hooft and M. Veltman. to be precise iLI vertices in momentum space are introduced. or calculating full Green’s functions external lines are treated as propagators. which is summarized in the following rules: (Rule 1) Start with external legs (incoming particles/outgoing particles) and draw all possible topologically different connected diagrams. For the symmetry factor consider the examples (given in G. Note that in calculating amputated Green’s functions external lines are neglected. = −i g . for example up to order g 2 the scattering of two neutral spin 0 particles (real scalar field) is described by + + + (Rule 2) The contribution of each diagram is obtained by multiplying the contributions from propagators. but overall momentum conservation at a vertex is understood. 3! 4! The vertices are: = −i f Consider first the lowest order self-energy diagram.27. the factor 2 corresponds to the two-points Green’s function having two identical ends).66 and 10.

Divide by the permutational factors of the vertices. 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. The total result is 6×3×2 1 1 = = . S 3! 3! 4! 2! 2 Complex scalar fields The case of complex scalar fields can be considered as two independent fields. denoted k = i k 2 − M 2 + iǫ Note that in this case the propagator does not have identical ends. (10. External line 1 can be attached in six. but it connects a source with its complex conjugate and therefore is oriented. 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). the result is 1 6×4×6×3×2 1 = = . S 3! 3! 2! 2 As a second example consider the diagram 1 2 There are three vertices.e. 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 . ∗ (z) i δJ(z) i δJ d4 x d4 y J ∗ (x) i∆F (x − y) J(y) . Then there are two ways to connect the remaining lines such that the desired diagram results. after that line 2 in three ways. i. which have been included in the definition of vertices (here 3! for each vertex). 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. Dividing by vertex factors and permutations of identical vertices.Feynman diagrams for scattering amplitudes 97 Draw two points corresponding to the two vertices and draw in each of these points the lines coming out of the vertices: Now count in how many ways the lines can be connected with the same topological result. or equivalently as independent fields φ and φ∗ . .

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

λ (∂µ Aµ )2 2 1 For a massive vector boson inversion of the quadratic term including the Lorentz constraint ∝ the propagator kµ kν 1 −gµν + 1− i Dµν (k) = i . Vector fields and Quantum Electrodynamics As the most important example of vector fields consider the lagrangian density for quantum electrodynamics (QED). λ 1 ∂ (10. 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. + 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. 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 . Particular choices of λ are λ = 1 (Feynman propagator or Feynman gauge) and λ = ∞ (Landau gauge). It is necessary to start at the end of a fermion line (with the above arrow convention an outgoing fermion wave function u or an incoming antifermion ¯ wave function v ) and keep on following that line. 1 k 2 − M 2 + iǫ λ (k 2 − M 2 + iǫ)(k 2 − λ M 2 + iǫ) leads to . The wave functions for fermions are given by ∝ g 2 S(z − z ′ ) S(z ′ − z) = −g 2 iS(z − z ′ ) iS(z ′ − z). 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 . 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.

75) Note that the q µ q ν term in the photon propagator are irrelevant because the photon couples to a conserved current. s3 )γ µ u(k. s′ ) ¯ u s. s4 )(−ie)γ ν u(p.s4 ) and averaged over spins in the initial state (1/2 × 1/2 × s1 . s2 ).s2 ) which can be written as (see also chapter 4) |M |2 = = 1 4s |¯(k ′ .76) 1 .78) = 2 α2 2 s + u2 + 4t(M 2 + m2 ) − 2(M 2 + m2 )2 .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. t2 (10. To lowest order (∝ α = e2 /4π) only one diagram contributes. s)γν u(k ′ .s4 e4 (m) µν (M) L L . The diagram and momenta and the commonly used invariants (Mandelstam variables) for a 2 → 2 scattering process are 1 k 3 k’ s t u = (k + p)2 = m2 + M 2 + 2 k · p = (k ′ + p′ )2 = m2 + M 2 + 2 k ′ · p′ = (k − k ′ )2 = q 2 = 2 m2 − 2 k · k ′ = (p − p′ )2 = 2 M 2 − 2 p · p′ p 2 p’ 4 s = (k ′ − p)2 = m2 + M 2 − 2 k ′ · p + t+u= m2 = 2 m2 + 2 M 2 i = (k − p′ )2 = m2 + M 2 − 2 k · p′ The scattering amplitude is given by −iM = u(k ′ . s1 ) u e2 u(p′ . s′ )γµ u(k. s1 ) ¯ −igµν u(p′ .79) .s2 . s2 )|2 ¯ q2 (10. s)¯(k.Feynman diagrams for scattering amplitudes 100 10. s3 )(−ie)γ µ u(k.s3 . where u ¯ L(m) µν = = 1 2 u(k ′ . q 4 µν [using that (¯(k ′ )γµ u(k))∗ = u(k)γµ u(k ′ )].77) Combining Lµν and Lµν (M) one obtains L(m) Lµν (M) = 2 s2 + u2 + 4t(M 2 + m2 ) − 2(M 2 + m2 )2 µν and M 4π 2 (m) (10.5 Some examples eµ scattering The first example is the electromagnetic scattering of an electron and a muon. 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 . ¯ q2 (10. s4 )γµ u(p.

s2 3 . s1 )γν v(k ′ . Basically the same diagram is calculated and the result is the same after the interchange of s ↔ t. s4 ) u e2 v (k ′ . . 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. s3 )γ µ v(p′ .80) Note the similarity in the amplitudes for eµ scattering and e− e+ → µ− µ+ . s4 ) v (p′ . This is known as crossing symmetry. Exercises Exercise 10. s2 (10. . s3 ) ¯ ¯ 1 . (c) Do the same for the connected Green function Gc .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 . Similarly.s3 . s4 )γ ν u(p. To lowest order (∝ α = e2 /4π) only one diagram contributes. the masses. The diagram. s3 )γ µ v(p′ . s2 )γµ u(k.2. .s2 . (4) (4) . (b) Use the definition of the source-connected Green function Gsc and show that indeed sourceconnected diagrams survive. s2 ) ¯ ¯ 1 2s u(p. s1 )|2 ¯ s 1 . for instance Møller scattering (e− e− → e− e− ) and Bhabha scattering (e− e+ → e− e+ ) are related using crossing symmetry.Feynman diagrams for scattering amplitudes 101 The second example is the annihilation of an electron pair and creation of a muon pair. s2 )γµ u(k. s1 ) u(k. x4 ) for the interacting case in φ4 -theory to first order in the coupling constant g as obtained from the full expression for Z[J] in section 10.s4 = 1 2s × v (k ′ .1 (a) Give in diagrammatic notation the full Green functions G(4) (x1 .

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

For two particles. (11.1 kinematics in scattering processes Phase space The 1-particle state is denoted |p . . q).2) (11.4) For two particle states |pa . (11. for instance for determining the threshold energy for the production of a final state 1 + 2 + . The first is the CM system. It is determined by the energy-momentum four vector p = (E. the 1particle phase space is d3 p d4 p θ(p0 ) (2π)δ(p2 − m2 ).e. . (11. a a b b the quantity s = P 2 = (pa + pb )2 . pn ) = and the reduced phase space element by dR(s. pn ) = (2π)4 δ 4 (P − pi ) dR(p1 . To be specific let us consider two frequently used frames. (11. . i n d3 pi . + pn )2 .7) pb = cm (Eb . (2π)3 2Ei i=1 (11.6) (11.Chapter 11 Scattering theory 11. pa ) and pb = (Eb . The phase space is determined by the weight factors assigned to each state in the summation or integration over states. 103 . Here s is the invariant mass of the n-particle system. . This is generalized to the multi-particle phase space dR(p1 . −q). . .1) = (2π)3 2E (2π)4 (proven in Chapter 2). s is for obvious reasons known as the center of mass (CM) energy. . It is a useful quantity. . . In the CM frame the threshold value for s obviously is n 2 sthreshold = i=1 mi . pn ). p1 . A physical state has positive energy. i. . + n. p) which satisfies p2 = E 2 −p2 = m2 . .3) which is useful because the total 4-momentum of the final state usually is fixed by overall momentum conservation. pb ) satisfying p2 = m2 and p2 = m2 . . In that case cm pa = (Ea . . Its square root.5) √ is referred to as the invariant mass squared. . . . pb we start with the four vectors pa = (Ea . and the total momentum four-vector P = pa + pb . s = (p1 + .

which in the specific case a b also can be expressed as λ(s. 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. 4s (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 √ = . m2 ). 4π s 4π 8π s 4π (11.11) (11. (11. 2mb (11. p2 ) = = where λ12 denotes λ(s.15) Explicit calculation of the reduced two-body phase space element gives dR(s. 2 s s − m2 + m2 a b √ . m2 ) is a function symmetric in its three arguments. for instance. p1 . m2 ) a b . p1 .13) (11. mb and s). 2mb λ(s. trf Ea (threshold) = 1 2 mb ( i mi )2 − m2 − m2 a b . Prove that trf Ea = |ptrf | = a s − m2 − m2 a b .10) The function λ(s.Scattering theory 104 It is straightforward to prove that the unknowns in the particular system can be expressed in the invariants (ma . 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.8) (11. m2 . = 2 s λ(s. m2 . m2 ) a b . 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 . 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. a (11.12) pb = (mb . 0). m2 .14) One can. m2 . Also in this case one can express the energy and momentum in the invariants. m2 .9) (11.16) . In that case trf pa = (Ea . ptrf ).

the rapidity y is defined T T x y y= 1 ln 2 E + pz E − pz = ln E + pz mT = tanh−1 pz .23) which are not independent as s + t + u = m2 + m2 + m2 + m2 .22) (11. while the final state often arises from a series of 2-particle processes combined with the decay of an intermediate particle (resonance).24) Using the relation between t and cos θcm it is straightforward to express dΩcm in dt.. Consider the 1-particle inclusive case. In inclusive measurements no particles are detected in the final state. s = (pa + pb )2 = (pc + pd )2 t = (pa − pc ) = (pb − pd ) u = (pa − pd )2 = (pb − pc )2 2 2 (11. Consider the process a + b → c + d. 11. E . 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 . usually is a 2-particle state.21) (11. py ).20) → π0 + n → π+ + π− + n → . a c with q = λab /4s and q ′ = being 0 or 180 degrees.17) (11.4).25) dR(s.27) with m2 = m2 − p2 = m2 + p2 + p2 . . for instance the momenta of incoming (colliding) hadrons and it is useful to use for the produced particle rapidity as a variable. pd ) = 4π s 4π 8π s 4π dt dt √ √ . = (11. Instead of the scattering angle. tmax min λcd /4s. The variable s is always larger than a c b d the minimal value (ma + mb )2 . 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 . At high energies. Examples appear in → π− + p (11. in exclusive measurements all particles are detected. A specific reaction channel starts contributing at the threshold value (Eq. pc . pz = mT sinh y. (11.26) = 16π q s 8π λab Kinematics of inclusive hard scattering processes In high energy (hard) scattering processes.Scattering theory 105 Kinematics of 2 → 2 scattering processes π− + p The simplest scattering process is 2 particles in and 2 particles out. The various possibilities are referred to as different reaction channels. 2s (11. usually many particles are produced. The initial state. however.19) (11. dt = 2 qq ′ d cos θcm and obtain for the two-body phase space element √ q ′ dΩcm λcd dΩcm √ = (11. where the first is referred to as elastic channel and the set of all other channels as the inelastic channels. in which one particle is detected. there is usually a preferred direction.. pT = (px . Writing E = mT cosh y. An often used set of invariants are the Mandelstam variables.18) (11. Of course there are not only 2-particle channels. H1 + H2 → h + X.

u). which means that rapidity distributions dN/dy maintain their shape. u). t. ¯ tu = (pa − pc )2 = t. This is known as crossing symmetry. t-channel and u-channel processes respectively. tu = s. 11.30) (11. tt = s. c bd Mad→c¯(su = u.Scattering theory 106 It is convenient because under a boost (along z) with velocity β it changes as y → y − tanh−1 β. uu = (pa − p¯)2 = s. we have seen that the amplitudes for the processes e− µ− → e− µ− and the process e− e+ → µ+ µ− are simply related by an interchange of the variables s and t. (11.28) which (only involving angles) is easier to determine. η = − ln(tan(θ/2)) = tanh−1 (cos θ). b ut = (pa − pd )2 = u. u) with s = (p1 + p2 )2 = (p3 + p4 )2 .2 Crossing symmetry In the previous chapter. (11. With the momenta defined as in the figures above one has for all these processes the same amplitude M (s.29) 3 2E 3 2 (2π) 2 (2π) 16π 2π E 16π 2 2π with at high energies the factor |p|/E ≃ 1. For large energies and not too small scattering angles (θ ≫ 1/γ) y is approximately equal to the pseudo-rapidity η. ¯ tt = (pa − p¯)2 = s. These variables are precisely the Mandelstam variables for the s-channel process (ab → cd). 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. ¯ b (11. b Analiticity of the field theoretically calculated result implies Ma¯→¯ (st = t. The one-particle phase space in terms of these variables becomes d3 p dy d|pT |2 dφ dy d2 pT |p| dη d|pT |2 dφ = = = .31) . t. ¯ while for the u-channel process (ad → c¯ one has b) su = (pa + pd )2 = u. ut = u) = Mab→cd (s. t. t = (p1 − p3 )2 = (p2 − p4 )2 . u = (p1 − p4 )2 = (p2 − p3 )2 . For the t-channel process (a¯ → ¯ one has c bd) st = (pa + pc )2 = t. uu = s) = Mab→cd (s.

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

e. 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. dt (11. the number of scattered or decayed particles is given by 2 N = (2π)4 δ 4 (Pi − Pf ) iMf i dR(p1 .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. as expected. σ= √ 2 λab T · V (11.33) Decay of particles For the decay of particle a one has macroscopically dN = −Γ N.39) One of the δ functions can be rewritten as T · V (remember the normalization of plane waves). (11. N 1 . (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 . From the solution N (t) = N (0) e−Γ t one knows that the decay time τ = 1/Γ. For the (proper) decay width one thus has 1 N Γ0 = .Scattering theory 108 Although this at first sight does not look covariant. . . 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 . . .34) i. the amount of decaying particles is proportional to the number of particles with proportionality factor the em decay width Γ.T V. it is. . (11. The decay time for moving particles τ is related to the decay time in the rest frame of that particle (the proper decay time τ0 ) by τ = γ τ0 .35) This quantity is not covariant. T · V ρa (11.38) (in which we can calculate −iMf i using Feynman diagrams).T d4 x = V · T (2π)4 δ 4 (Pi − Pf ). pn ). N and T · V are covariant.36) (11.

note the sign and cos θ = qi · qf ). p1 .45) Since the amplitudes also depend on all momenta the full result for two-particle intermediate states is (in CM. t) λab 4π dt.e. . 16π 2 s (11. . p1 .4 Unitarity condition (S † )f n Sni = δf i The unitarity of the S-matrix. . θ) or M (q i . i M† M √ − √ 8π s 8π s = fi n dΩn ∗ Mnf qn Mni √ √ . θ) = −8π s ℓ (in analogy with the expansion for f (E. (11. θ) in quantum mechanics. θcm ) √ q 8π s 2 2 (11. The sign difference comes from the (conventional) sign in relation between S and quantummechanical and relativistic scattering amplitude. respectively. θ). θ)|2 . √ (2ℓ + 1)Mℓ (s) Pℓ (cos θ). . 8π s 2π 8π s . . The result is N = |Mf i |2 dR(s. implies for the scattering matrix M . 11. The amplitude −M /8π s is the one to be compared with the quantum mechanical scattering amplitude f (E.46) Partial wave expansion Often it is useful to make a partial wave expansion for the amplitude M (s.Scattering theory 109 (Using normalized wave packets these somewhat ill-defined manipulations can be made more rigorous). .41) (11. 2 λab and for instance for two particles dσ = q ′ M (s. for which one has dσ/dΩ = |f (E. ˆ ˆ Inserted in the unitarity condition for M . q f ). (M † )f n (2π)4 δ 4 (Pi − Pn ) Mni . . (11.43) dΩcm = π M (s. q n ) q n qn √ Mni (q i .42) 2−body decay The differential cross section (final state not integrated over) is given by 1 dσ = √ |Mf i |2 dR(s. . i.44) This can be used to get the full expression for dσ/dt for eµ and e+ e− scattering. for which the √ amplitudes squared have been calculated in the previous chapter. (11. pn ). p1 . . pn ).47) M (s. . . q n ). (11.25) −i Mf i − (M † )f i = − ∗ dΩ(ˆn )Mnf (q f . see 11. pn ) |M |2 q 32π 2 m2 dΩ |M |2 . . (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 . 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 .

In general a given channel has |Sℓ (s)| ≤ 1.52) From the imaginary part of M (s. q ˆ i. n (11. (11. † −i (Mℓ )f i − (Mℓ )f i = 2 † (Mℓ )f n qn (Mℓ )ni . Using σ= dΩ q ′ M (s.49) or Im Mℓ = q |Mℓ |2 . the total cross section can be determined.54) ℓ . Show that σT = = 4π q 2π q2 (2ℓ + 1) Im Mℓ (s) ℓ ℓ (2ℓ + 1) (1 − ηℓ cos 2δℓ ). 0). (11. 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 . η = 1) the result σel = 4π q2 (2ℓ + 1) sin2 δℓ . parametrized as Sℓ (s) = ηℓ (s) exp(2i δℓ (s)). ℓ (11. (11. which allows writing Mℓ (s) = Sℓ (s) − 1 e2i δℓ (s) − 1 = . σel = 4π ℓ (2ℓ + 1)|Mℓ (s)|2 (2ℓ + 1) ℓ = 4π q2 ηℓ e 2i δℓ − 1 2i 2 .51) and for the case that this is the only channel (purely elastic scattering.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 .53) The difference is the inelastic cross section. θcm ) √ q 8π s 2 .48) If only one channel is present this simplifies to ∗ ∗ −i (Mℓ − Mℓ ) = 2 qMℓ Mℓ . 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.e. (11. σinel = π q2 2 (2ℓ + 1) (1 − ηℓ ). 2i q 2i q (11.

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

The cross section σT (π − p) also shows a resonance with the same width with a value of about 70 mb. Its mass is M = 1232 MeV.49 GeV. Exercises Exercise 11. where furthermore σinel = 0. Essentially this resonance can decay into quark-antiquark pairs or into pairs of charged leptons. Its mass is M = 3096. Γ = 2. 2 q (s − M 2 )2 + M 2 Γ2 (11. which is attributed to neutrinos. u) = 1 1 2 (s + u2 ) + 4m2 (t − m2 ) t2 2 2 1 1 ( s − m2 )( s − 3m2 ) tu 2 2 . The cross section at resonance is about 30 nb. The phase shift in the resonating channel near the resonance is given by tan δℓ (s) = MΓ . Three famous resonances are: • The ∆-resonance seen in pion-nucleon scattering. the full width is Γ = 88 keV. a fast change in the phase shift for a narrow resonance. At resonance the cross section σT (π + p) is about 210 mb. i. All these decays can be seen and leave an ’invisible’ width of 498 MeV.1 Show that the cross section for electron-electron scattering (exercise 10. 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.2 GeV. dt s(s − 4m2 ) with f (t.Scattering theory 112 (Prove this using the unitarity condition for partial waves). • The J/ψ resonance in e+ e− scattering. one can reconstruct the resonance shape for different numbers of neutrino species. u) + g(t. the partial width into e+ e− is Γee = 5. its width Γ = 120 MeV.60) showing that the phase shift at resonance rises through δ = π/2 with a ’velocity’ ∂δ/∂s = 1/M Γ. 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).59) reaching the unitarity limit for s = M 2 . u) = g(t. Knowing that each neutrino contributes about 160 MeV (see next chapter).4) can be written as 4πα2 dσ = {f (t. Note that because of the presence of a background the phase shift at resonance may actually be shifted. • The Z 0 resonance in e+ e− scattering with M = 91.26 keV.88 MeV. This characteristic shape of a resonance is called the Breit-Wigner shape. t) + g(u. u) + f (u.e. This is a narrow resonance discovered in 1974. The half-width of the resonance is M Γ. t)} . M2 − s (11. From this one sees that a resonance has the same shape in all channels but different strength.

ρcm ρcm a b 1 cm . writing N (s) M1 (s) = 2 + iM Γ .Scattering theory 113 Exercise 11. 4.3 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. Exercise 11. Exercise 11. cm |va − vb | in the center of mass system for two colliding particles a and b. Calculate now fπ .8. Compare the result with the factor trf 1/ρa ρb va that we used in the derivation of the basic expression for cross sections. Thus.999 877 Γ and Γ(π − → e− νe ) = 0. 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. a quantity which we will ¯ ¯ encounter in the next chapter (section 12.4 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∆ . s − M∆ ∆ ∆ 2 calculate Re N (M∆ ) . Show that unitarity can be used one step further to fix also the real part of the residue of the pole in the amplitude.2).000 123 Γ.6 × 10−8 s and the relative decay widths Γ(π − → µ− νµ ) = 0. As discussed for unstable particles.

Chapter 12

The standard model
12.1 Non-abelian gauge theories

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

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

La

φ(x) = (1 + i θa La ) φ(x)

inf

(12.2)

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

(12.3)

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

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

The standard model

115

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

i (∂µ U (θ)) U −1 (θ), g

(12.7)

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

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

(12.8)

(12.9) (12.10)

The gauge-invariant lagrangian density is now constructed as

Gµν → U (θ) Gµν U −1 (θ).

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

(12.11)

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

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

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

The standard model
xψ x+dx

116

(x)

ψ(x)

x+dx ψ(x+dx)

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

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

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

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

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

    λ7 =   

i

    −i   

  1  1  λ8 = √   3

−2

However.16) which in the presence of the ’connection’ Aµ differs from the total change between x ψ(x) and x+dx ψ(x+ dx). We note that local gauge invariance requires that we can modify all local systems with a (local) unitary transformation S(x). A ’constant’ vector that only rotates because of the arbitrary definitions of local frames satisfies Dµ ψ(x) = 0. Dν ] ψ(x) 1 − ig dσ µν Gµν ψ(x). ds′ µ 0 ig Aµ (s) ψ(s) dxµ Aµ (x) ψ(s) = which is the path-ordered integral denoted ψ(x) = P eig P ψ(0). (12. should be independent of such transformations. x+dx ψ(x) = = 1 + ig dxµ Aa (x)Ta µ 1 + ig dx Aµ (x) µ x x ψ(x) (12.16. 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.e.17) The three equations given so far immediately give Dµ = ∂µ − ig Aµ (x). 12. requiring that the ’connection’ Aµ (x) is such that Dµ → S(x)D µ S −1 (x). (12. x+dx ψ(x + dx) = 1 + dxµ Dµ x+dx ψ(x).18) or considering a path xµ (s) from a fixed origin (0) to point x. (12. but expresses them with respect to different bases. The relation in Eq. i. if two different paths to the same point give different phases the effects can be observed. -dx -dy µ µ dyµ µ dx For a vector. = .19) This gives rise to a (path dependent) phase in each point. 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 .The standard model the basis for x G and x+dx 117 G. (12.e. What is this physical effect by which the ’connection’ Aµ can be observed? For this consider the phase around a closed loop. 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) = 0.14) (12.15) ψ(x). ds s dxµ ds′ P exp ig A (s′ ) ψ(0). which connects two identical vectors. ds which is solved by dψ(s) ds = dxµ . In principle such a phase in a given point is not observable. i.

20) with a lagrangian density of the form L = 1 1 1 ∂µ φ · ∂ µ φ − m2 φ · φ − λ(φ · φ)2 . If the temperature is high enough the spins are oriented randomly and the (macroscopic) ground state is spherically symmetric. usually being (slightly) prejudiced by impurities. 12. In this case there are many possible groundstates (determined by a fixed direction in space). 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. in reality a not perfectly symmetric situation. Consider as an example a ferromagnet with an interaction hamiltonian of the form H=− Jij Si · Sj .2 Spontaneous symmetry breaking In this section we consider the situation that the groundstate of a physical system is degenerate. and equivalently Aµ can be considered as a pure gauge effect. a theory for a scalar degree of freedom (a scalar field) having three (real) components. If it is zero one has dxµ Aµ (x) = 0. Classically the (time-independent) ground state is found for V(φ ) F |φ| Figure 12. In geometric language the effect on parallel transport of a vector depends on the ’curvature’ Gµν .    φ1    φ .2. 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).  φ= 2      φ3 (12. .The standard model 118 where Gµν = (i/g)[Dµ . 2 2 4 −V (φ) The potential V (φ) is shown in fig. i. e.g. 12.2: The symmetry-breaking ’potential’ in the lagrangian for the case that m2 < 0. Only if this quantity is nonzero a physically observable effect of Aµ exists. i>j which is rotationally invariant. this means that there is more than one possibility for the groundstate. This characterizes spontaneous symmetry breaking. we can disregard those ’perturbations’ and look at the ideal situation. As there can be one and only one groundstate. which means that by an appropriate transformation S(x) it can be gauged away (see the example of Aharonov-Bohm effect). the groundstate itself appears degenerate. external magnetic fields. Nevertheless. Nature will choose one.e. Dν ].

(12. It is appropriate to redefine the field as    ϕ1      φ =  ϕ2  . In this case the spectrum is described .24) 1 2 2 4 1 1 1 = (∂µ ϕ1 )2 + (∂µ ϕ2 )2 + (∂µ η)2 + m2 η 2 + . . (12. In order to see the consequences for the particle spectrum of the theory we construct the lagrangian in terms of the fields ϕ1 . Any constant field with ’length’ F is a possible groundstate. 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 . λ such that 0|ϕ1 |0 = 0|ϕ2 |0 = 0|η|0 = 0. The field along the third axis plays a special role because of the choice of the vacuum expectation value. say    0    0 .) terms constitute interaction terms. ϕ2 and η. .   ϕc = 0|φ |0 =  (12. The original lagrangian contained an SO(3) invariance under (length conserving) rotations among the three fields.25) 2 2 2 Therefore there are 2 massless scalar particles. They are known as the Weyl mode or the Goldstone mode: Weyl mode. has less symmetry. etc.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. In this mode the lagrangian and the vacuum are both invariant under a set of symmetry transformations generated by Qa .21) Stability of the action requires the classical groundstate ϕc to have a well-defined value (which can be nonzero).e. (12. It is only invariant under rotations around the 3-axis.23)     F +η the latter only forming a minimum for m2 < 0. The situation now is the following. so one has 0|φquantum |0 = 0 and hence 0|φ|0 = ϕc . . In the case of degeneracy. The classical groundstate appears degenerate. while the quadratic terms must correspond with non-negative masses. for the vacuum Qa |0 = 0. In this situation one speaks of spontaneous symmetry breaking. The presence of a nonzero value for the classical groundstate value of the field will have an effect when the field is quantized. Writing the field φ as a sum of a classical and a quantum field. i. 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. It is sufficient to do this to second order in the fields as the higher (cubic. therefore a choice must be made. φ = ϕc + φquantum where for the (operator-valued) coefficients in the quantum field one wants 0|c† = c|0 = 0.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. while the lagrangian including the nonzero groundstate expectation value chosen by nature. A quantum field theory has only one nondegenerate groundstate |0 . The η massless particles are called Goldstone bosons.

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

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

Initially there are 3 massless gauge fields (each.    0 φ= (12. made into a gauge theory.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. . One may wonder about the degrees of freedom. After symmetry breaking the same number (as expected) comes out.38) ∂µ Wν − ∂ν Wµ + g Wµ × Wν . having two independent spin components) and three scalar fields (one degree of freedom each).e. 1 1 L = − Gµν · Gµν + Dµ φ · Dµ φ − V (φ). The latter is a spin 0 particle (real scalar field) called a Higgs particle. (12.   The symmetry is broken in the same way as before and the same choice for the vacuum. The difference comes when we reparametrize the field φ. as in this case there are no massless Goldstone bosons. in this case three-dimensional) representation reads (La )ij = −i ǫaij . while the number of massive gauge bosons coincides with the number of ’broken’ generators. 2 massive vector fields or spin 1 bosons (2 × 3) and one scalar field (1).    0      ϕc = 0|φ |0 =  0  . . Therefore we can always rotate the field φ into the 3-direction in order to simplify the calculation. We have the possibility to perform local gauge transformations. Considering the same example with rotational symmetry (SO(3)) as for spontaneous symmetry breaking of a scalar field (Higgs field) with three components.     0  0         0 = . (12. Note that the number of massless gauge bosons (in this case one) coincides with the number of generators corresponding to the remaining symmetry (in this case rotations around the 3-axis). (12.39) (12. is made.41)         φ3 F +η Explicitly one then has  2 2  gF Wµ + g Wµ η   1 1  −gF Wµ − g Wµ η Dµ φ = ∂µ φ + g Wµ × φ =   ∂µ η     . one sees that the fields Wµ and Gµν also can be represented as three-component fields. again 9 degrees of freedom. like a photon.     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 + . two massive bosons (Wµ and Wµ with MW = gF ) and a massive scalar particle (η with m2 = η 2 −2 m . but one has 1 massless gauge field (2). . . thus 9 independent degrees of freedom. i.42) 2 from which one reads off that the particle content of the theory consists of one massless gauge boson 3 1 2 (Wµ ).The standard model 122 12. 4 2 where Since the explicit (adjoint.37) (12. Dµ φ = Gµν = ∂µ φ + g Wµ × φ.40) a Dµ φ = ∂µ φ − ig Wµ La φ.

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

4 4 In order to break the symmetry to the symmetry of the physical world. (12.The standard model 124 3 3 where Wµ is a triplet of gauge bosons with TW = 1. Putting this in leads to L (f ) =L (f 1) +L (f 2) . (12.   h)  (12. .54) = (Dµ φ)† (Dµ φ) −m2 φ† φ − λ (φ† φ)2 . leading to the parametrization   1  0    φ(x) = √  (12. .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. . 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. For the classical field the choice θ4 = v is made. the U (1)Q symmetry (generated by the charge operator). 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 ).53) φ=   √ (θ4 − iθ3 )  φ 2 with TW = 1/2 and YW = 1 is introduced. † i i g Wµ · τ − g ′ Bµ )φ. which assures with the choice of YW for the Higgs field that Q generates the remaining U (1) symmetry. −V (φ) L and (h2) = −Ge (LφR + Rφ L). (12.59) . . + 8 (12. with the following lagrangian density consisting of a symmetry breaking piece and a coupling to the fermions. this leads to the lagrangian L (h1)     .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.58) (12. + 8 g2 v2 1 + − (Wµ )2 + (Wµ )2 (∂µ h)2 + m2 h2 + 2 8 (g 2 + g ′2 ) v 2 (Zµ )2 + .57) = 1 g2 v2 1 2 (Wµ )2 + (Wµ )2 (∂µ h)2 + m2 h2 + 2 8 v2 2 3 gWµ − g ′ Bµ + . .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 . L where L (h1) (h) =L (h1) +L (h2) . a complex Higgs field   +   1  φ   √2 (θ2 + iθ1 )    0 = 1   (12.

σ (νµ γ ρ µL ) 2 2 (eL γ νe ) 2 k + MW (12.66) From the coupling to the photon.69) .62) g 2 + g ′2 3 ≡ sin θW Wµ + cos θW Bµ . 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 . 2θ 2θ 4 cos W cos W 0. (12. (12.64) (12. .60) (12. The coupling of the fermions to the physical gauge bosons are contained in L (f 1) giving L (f 1) = i eγ µ ∂µ e + i νe γ µ ∂µ νe − g sin θW eγ µ e Aµ g 1 1 + sin2 θW eR γ µ eR − cos 2θW eL γ µ eL + νe γ µ νe cos θW 2 2 g µ + µ − + √ νe γ eL Wµ + eL γ νe Wµ .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 . 2 Zµ (12. g ′ /g = tan θW . 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 + . we can read off e = g sin θW = g ′ cos θW . 2 + g ′2 g 3 g ′ Wµ + g Bµ (12.65) The weak mixing angle is related to the ratio of coupling constants. . 2 3 g Wµ − g ′ Bµ 3 ≡ cos θW Wµ − sin θW Bµ .67) The coupling of electrons or muons to their respective neutrinos. 2 (12.61) (12.63) (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 . 4 2 MW g2 v2 = .

as appear for instance in the decay of the neutron or the decay of the Λ (quark content uds). 2 The value of α deviates from the known value α ≈ 1/137 because of higher order contributions. leptons with CV = −1/2 + 2 sin2 θW ≈ −0.19 GeV. 91. g2 MZ 2 2 (CV + CA ). Finally we want to say something about the weak properties of the quarks.05 and CA = −1/2 we calculate Γ(e+ e− ) ≈ 78. For each neutrino species (with CV = 1/2 and CA = 1/2 one expects Γ(νν) ≈ 155 MeV.75) The coupling of the Z 0 to fermions is given by (g/ cos θW )γ µ multiplied with 3 TW 1 1 1 (1 − γ5 ) − sin2 θW Q ≡ CV − CA γ5 . (12. 3 = TW .70) = 2 = 2 sin2 θ 8 MW 2 v2 8 MW 2 W In this way the parameters g. such as2 α = e2 /4π GF sin2 θW MW MZ ≈ 1/127. Comparing this with the total width into (invisible!) channels. The masses of the fermions and the coupling to the Higgs particle are contained in L (h2) . muon or tau.76) with CV CA 3 = TW − 2 sin2 θW Q. i.40 GeV. 80.5 MeV (exp. one has the relation g2 e2 1 GF √ = . With the choosen vacuum expectation value for the Higgs field.g. 2 2 2 (12.e.80) First.79) Γ(Z 0 → f f ) = 48π cos2 θW For the electron.73) (12. Actually including corrections corresponding to higher order diagrams the agreement for the decay width into electrons can be calculated much more accurately and the number of allowed (light) neutrinos turns to be even closer to three.231 2. Γe ≈ Γµ ≈ Γτ ≈ 83 MeV). (12.77) (12. v (12. The coupling to the Higgs particle is weak as the value for v calculated e.e.78) From this coupling it is straightforward to calculate the partial width for Z 0 into a fermion-antifermion pair. i. g ′ and v determine a number of experimentally measurable quantities.74) (12. 1. −5 (12.71) GeV −2 = = = = . .72) (12. (12. giving rise to a running coupling constant after renormalization of the field theory.The standard model 126 the good old four-point interaction introduced by Fermi to explain the weak interactions. from the MW mass is about 250 GeV. the mass of the electron comes from the spontaneous symmetry breaking but is not predicted (it is in the coupling Ge ). (12.166 4 × 10 0. me /v is extremely small. Γinvisible = 480 MeV one sees that three families of (light) neutrinos are allowed. one obtains L (h2) Ge v Ge = − √ (eL eR + eR eL ) − √ (eL eR + eR eL ) h 2 2 me = −me ee − eeh.

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

g. † Λu Λ† = Vu G2 Vu .q) ′ weak † mass mass DL = DL = Vu Vd DL = VCKM DL . 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 imaginary part i η gives rise to CP violation in decays of K and B-mesons (containing s and b quarks. ms and mb ). (12. magnitudes of the entries in the CKM matrix are nicely represented parametrization  1 − 1 λ2 λ λ3 A(ρ − i η)  2   1 2  −λ 1− 2λ λ2 A V =  3 2 λ A(1 − ρ − i η) −λ A 1 12. 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 . UR = uR cR tR .83) where Vu and Vd are unitary matrices. .e. for which we need the appropriate weak isospin doublet  0∗    1    φ c  = √  v +h . respectively). The Higgs field is still limited to one complex doublet. A ≈ 0.86) the unitary CKM-matrix introduced above in an ad hoc way.   φ =    −φ− 0 2 For the (squared) complex matrices we can find positive eigenvalues.The standard model 128 using the socalled Wolfenstein      + O(λ4 )   with λ ≈ 0. i. allowing us to write † Λu = Vu Gu Wu † and Λd = Vd Gd Wd . each of these containing   the three families. Note that we need the conjugate Higgs field to get a U (1)Y singlet in the case of the charge +2/3 quarks. the weak mass eigenstates are L (h2.84) with Gu and Gd being real and positive and Wu and Wd being different unitary matrices.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 (h2.22 and η ≈ 0.85) √ √ with Mu = Gu v/ 2 (diagonal matrix containing mu .82 and ρ ≈ 0. mc and mt ) and Md = Gd v/ 2 (diagonal matrix containing md . the three righthanded quarks with charge +2/3 in UR and the three righthanded quarks with charges −1/3 in DR .q) = −QL φΛd DR − DR Λ† φ† QL − QL φc Λu UR − UR Λ† φc† QL u d (12.227. u u where we include now the three lefthanded quark doublets in QL . (12. The Λu and Λd are complex matrices in the 3×3 family and Λd Λ† = Vd G2 Vd† .82) space. d d (12. Thus one has † † † =⇒ −DL Vd Md Wd DR − DR Wd Md Vd† DL − UL Vu Mu Wu UR − UR Wu Mu Vu UL (12.34. e.

EL = Ve EL . since the weak current is the only place where they show up.5×10−3 eV2 . UPMNS a parametrization that in principle also could have been used for quarks.    (12. In this case. mµ and mτ . one can write Λe = Ve Ge We and we find † L (h2. decoupling from all known interactions. (12.ℓ) =⇒ −Me EL Ve We ER − ER We Ve† EL . For neutrino’s this matrix is parametrized in terms of three angles θij with cij = cos θij and sij = sin θij and one angle δ. † † Again we continue as before now with matrices Λe = Ve Ge We and Λn = Vn Gn Wn . it is particularly useful because θ12 is essentially determined by solar neutrino oscillations requiring ∆m2 ≈ 8 × 10−5 12 eV2 (convention m2 > m1 ).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 − Lφc Λn NR − NR Λ† φc† L e n (12. The W -current then becomes mass mass E L γ µ NL = E L γ µ NL . EL = Ve† EL .ℓ) † † † =⇒ −EL Ve Me We ER − ER We Me Ve† EL − NL Vn Mn Wn NR − NR Wn Mn Vn NL . we find massless neutrinos.88) √ mass with Me = Ge v/ 2 the diagonal mass matrix with masses me . 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  .e. and obtain L (h2.87) is a weak doublet containing the three families of neutrinos (NL ) and charged leptons (EL ) and ER is † a three-family weak singlet.ℓ) = −LφΛe ER − ER Λ† φ† L.91) † with UPMNS = Vn Ve known as the Pontecorvo-Maki-Nakagawa-Sakata mixing matrix. (12. while θ23 then is determined by atmospheric neutrino oscillations requiring |∆m2 | ≈ 2.92) . there is no family mixing for massless neutrinos.  =  − 1/6 2/3 0           1/2 1/2 0 0 1 0 1/3 1/2 − 1/6 (12. For the (massless) neutrino fields we just can redefine fields into NL = Ve† NL . Working with the mass eigenstates and the weak current becomes EL γ NL = EL e n L weak for the charged leptons we see that the weak eigenstates for the neutrinos are NL = Ve† NL with the relation to the mass eigenstates for the lefthanded neutrinos given by † ′ weak mass mass NL = NL = Ve† Vn NL = UPMNS NL . The lepton sector (massive Dirac neutrinos) In principle a massive Dirac neutrino could be accounted for by a lagrangian of the type L (h2. As before.The standard model 129 The lepton sector (massless neutrinos) For a lepton sector with a lagrangian density of the form L in which (h2. NL = Vn NL and NR = Wn NR µ mass γ µ V † V N mass .90) mass † mass mass † mass † We note that there are mass fields ER = We ER . (12. i.89) with three righthanded neutrinos added to the previous case. e   N L= L  EL     (12. The mass fields ER † mass † mass = We ER .

12.94) M L NL NL + M L NL NL .98) (12. −    MD MR nR 2 (12.       0 0 1 L mass. (12. 2 although this option is not attractive as it violates the electroweak symmetry. one must for the neutrinos as well as charged leptons.ν =− containing three (CP-violating) phases (α1 .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.and righthanded species then just form a Dirac fermion. 1 c ∗ c L mass. (12. For the neutrino sector. however. is to add in Eq. For the charged leptons there cannot exist a Majorana mass term as this would break the U(1) electromagnetic symmetry. 1 L Υ ′ = nR + nc . are equivalent to two decoupled Majorana neutrinos (see below). This is called the see-saw mechanism (outlined below).88 a Majorana mass term for the (lefthanded) neutrino mass eigenstates.c. For these light Majorana neutrinos one has.99) . Υc ≡ (ΥL )c = ΥR 1 2 L and Υc = ΥL ). the massless and massive Majorana neutrinos. The way to circumvent this is to introduce as in the previous section righthanded neutrinos. coupled by a Dirac mass term. We use here the primes starting with the weak eigenstates. Actually. a unitary matrix relating them to the weak eigenstates. Thus 2 Υ ′ = nc + nL . For the leptons. The mass matrix can be written as   ML |MD | eiφ     M = (12. If the Majorana mass MR ≫ MD one actually obtains in a natural way one Majorana neutrino with a very small mass. 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.ν = − MR NR NR + MR NR NR . with for the righthanded sector a mass term MR . Thus (disregarding family structure) one has two Majorana neutrinos. Υ′ and Υ′ (satisfying Υc = Υ and as discussed in chapter 6.96) VPMNS = UPMNS K with K =  0 eiα2 /2 0  .The standard model 130 The lepton sector (massive Majorana fields) An even simpler option than sterile righthanded Dirac neutrinos. the left. couple the right. Absorption of phases in the states is not possible for Majorana neutrinos.95) 2 In order to have more than a completely decoupled sector.and lefthanded species through Dirac mass terms coming from the coupling to the Higgs sector as in the previous section. The see-saw mechanism Consider (for one family N = n) the most general Lorentz invariant mass term for two independent Majorana spinors. one being massive. as above. 1 ∗ c c (12. α2 and δ). hence the mixing matrix becomes   iα /2 0 0   e 1     (12. 2 R ψ = nR + nL .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.

 L      Υ2L nR Υ1 and Υ2 . (12. 2 with real masses Mi . which implies U ∗ M † U † = U ∗ M ∗ U † = M0 and a ’normal’ diagonalization of the (hermitean) matrix M M † .101) Thus one obtains from  2 ML + |MD |2  † MM =   |MD | ML e+iφ + MR e−iφ the eigenvalues 2 M1/2 |MD | ML e−iφ + MR e+iφ 2 MR + |MD |2   . This 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). related via    c   Υ1R     = U  nL  . This is a mixing problem with a symmetric (complex) mass matrix leading to two (real) mass eigenstates. T ] ψ. Exercises Exercise 12. For the situation ML = 0 and MR ≫ MD (taking MD real) one 2 finds M1 ≈ MD /MR and M2 ≈ MR . Exercise 12. . i ψ {M. Prove that if the generators Qa generate a symmetry.1 Consider the case of the Weyl mode for symmetries. 2 U (M M † ) U † = M0 .The standard model 131 taking ML and MR real and non-negative. Exercise 12. i.e.103) for each of which one finds the lagrangians L = and we are left with two decoupled Majorana fields      Υ1L    = U ∗  nc  .2 Derive for the vector and axial vector currents.e. H] = 0.      Υ2R nR (12. T } γ5 ψ. i.104) ↔ 1 1 Υi i / Υi − Mi Υi Υi ∂ 4 2 (12.  (12. a and a′ are degenerate states. V µ = ψγ µ T ψ and Aµ = ψγ µ γ5 T ψ ∂µ V µ ∂µ A µ = = i ψ [M.105) for i = 1. (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 . [Qa . 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′ .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)) . 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 .

hhh. Note that you can find the answer without explicit construction of all interaction terms in the lagrangian.6 Check that the use of the Wolfenstein parametrization in the CKM matrix indeed gives a unitary matrix. Neglect the masses of fermions and use the fact that the sum over polarizations is 3 pµ pν . to be precise one has for bosons (b) or fermions (f). In the standard model this coupling is described by the vertex −i g µ √ (γ − γ µ γ5 ) .) are proportional to the mass squared for bosons or mass for fermions of the particles. Γ(Z 0 → e+ e− ). (b) Calculate the width to electron-positron pair. the weak mixing angle is given by ¯ sin2 θW = 0. 3 TW . Exercise 12.The standard model 132 with CV CA = = 3 TW − 2Q sin2 θW . ghbb = b v v where v is the vacuum expectation value of the Higgs field. 0 ¯ Γ(Z 0 → f f ) = MZ g2 48π cos2 θW f f CV 2 + CA 2 . at least up to a high (which?) order in λ.4 Calculate the lifetime τ = 1/Γ for the top quark (t) assuming that the dominant decay mode is t → b + W +. 2 2 The masses are mt ≈ 175 GeV. Γ(Z 0 → νe νe ). Mf M2 and ghf f = . e+ e− h. Write down the matrix element squared (averaged over initial spins and summed over final spins) for the decay of the Z 0 . Exercise 12.231.5 Show that the couplings to the Higgs (for W + W − h. ǫ(λ) (p)ǫ(λ)∗ (p) = −gµν + µ ν M2 λ=1 ¯ to calculate the width Γ(Z → f f ). . The mass of the Z 0 is MZ = 91 GeV. Exercise 12. etc. Exercise 12. mb ≈ 5 GeV and MW ≈ 80 GeV.7 In this exercise two limits are investigated for the two-Majorana case. and the width to a pair of neutrino’s. ZZh.

what is the mass MX if we take for MD the electroweak symmetry breaking scale v (about 250 GeV).The standard model 133 (a) Calculate for the special choice ML = MR = 0 and MD real. 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. (b) A more interesting situation is 0 = ML < |MD | ≪ MR . Given that neutrino masses are of the order of 0.05 eV. which leads to the socalled see-saw mechanism. Give the explicit expressions that relate ψ and ψ c with Υ1 and Υ2 . . Calculate the eigenvalues ML = 0 and MR = MX .

Sign up to vote on this title
UsefulNot useful