QUANTUM FIELD THEORY

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

November 2011 (version 6.04)

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

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

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

1

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

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

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

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

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

Schroeder.3 6.4 3.E. C.1 2.5 2.3 3.2. I: Foundations.2 4. Itzykson and J.2 2. M. Corresponding chapters in books of Ryder.5 4. Zuber. 1996.1 1. Cambridge University Press.3 2.2 2. The quantum theory of fields. L. M.1 2 2 3.5 Ryder Peskin & Schroeder Srednicki 1 1 1 1 3 2.2 36 22 . 3.3 3.1 2. Weinberg. Cambridge University Press. 6. but are rather compact. M.2 6.4 2. 2. Quantum Field Theory.3 References As most directly related books to these notes.2 1.5 5.4 4. Ryder. Diagrammatica.V. Cambridge University Press. Srednicki.3 2. 2007.1 5.-B. Peskin & Schroeder and Srednicki. 3.3. Other text books of Quantum Field Theory that are useful are given in refs [3-6]. 1995. Cambridge University Press. Cambridge University Press. 1980. Veltman. Vol.1 3.3 2.H.2 3.1 3. The notes contain all essential information. 1994. An introduction to Quantum Field Theory.4 3.2 3. Quantum Field Theory. 2. Quantum Field Theory.3 3. Peskin and D.4 2. I refer to the book of Srednicki [1] and Ryder [2].1 6.2 2. These notes Section 1.2 3. 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). S.1.3 2.2 3.5 2.3 3. 5.3 4.7 2. 1995. II: Modern Applications. 3.4 3.2 3. 1985.3 3.3 2.1 4. Vol. Addison-Wesly.4 6. McGraw-Hill. 1.2 6. 4.8 3.

4 7.6 3.6 6. 6.7.8 5 4.3.7 4.1 20.1 20.2.5 4. 6.6 8.5 6.3 7. 6.3 7.2 8.5 Ryder 4 4 4 4 4 4 Peskin & Schroeder 2.4.3 10.3 12. 2.4 These notes Section 7.5 7.5 6 7 6 6.6 9.3.5 2.1.1 20.2.1. 4.2 7.1 10.2 8. 4.2 9.3 20. 6.6 3.4 3.5.3 3.1 11. 4.8 6.2 4.1 9.5 7.4 10.2.4 12.4.3 2.4 12.1 12. 9. 6.2.5 9.1 7.5 .5 9.1 9.3 6 8 9 Srednicki 3 3 3 37 5.1 8. 20.4 8. 9.2 10.4.6 4. 8.2 5.3.4 2.4 4. 4.1.6 3.3 8.3 9.1 5.3 8.2 12.3 6.2 11.3 11.6 10.4 9.5 11. 2.7 5.10 8.5 4.

2 . • Slight rewriting of discussion of Majorana mass term in section 6.5 Corrections 2011/2012) • Extension of Exercise 3.3 • First two sentences of Chapter 5 have been modified.

054 571 68 (18) × 10−34 J s = 6.3) The uncertainty principle tells us that in this representation the momenta cannot be fully determined. Indeed. ˜ ψ(p) = i d3 r exp − p · r ψ(r).1 Quantum field theory In quantum field theory the theories of quantum mechanics and special relativity are united. It is given by pop ψ(r) = −i ∇ψ(r).5) One can also choose a representation in which the momenta of the particles are the dynamical variables. (1. ˜ The corresponding states are |p and the wave functions ψ(p) = p||ψ are the Fourier transforms of the coordinate space wave functions.2) In the limit that v ≪ c one reaches the non-relativistic domain. In quantum mechanics a special role is played by Planck’s constant h. In this coordinate representation the position operators r op simply acts as rop ψ(r) = r ψ(r).4) where δij is the Kronecker δ function. In quantum mechanics. (1. S ≫ . c = 299 792 458 m s−1 . Corresponding position and momentum operators do not commute. quantum effects do not play a role anymore and one is in the classical domain. pj ] = i δij .1) In the limit that the action S is much larger than .6) . 1 (1. the action of the momentum operator in the coordinate representation is not as simple as the position operator. One can talk about states |r and the wave function ψ(r) = r||ψ . the position can in principle be determined at any time with any accuracy.Chapter 1 Introduction 1. (1.582 119 15 (56) × 10−22 MeV s. In the framework of classical mechanics as well as quantum mechanics the position of a particle is a well-defined concept and the position coordinates can be used as dynamical variables in the description of the particles and their interactions. In special relativity a special role is played by the velocity of light c. being eigenvalues of the position operators. ≡ h/2π = 1. They satisfy the well-known (canonical) operator commutation relations [ri . usually given divided by 2π. (1. (1.

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.

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

= −24. gµ = gµρ g ρν and 0 1 2 3 is in essence a ’Kronecker delta’.1 (a) In the the Hydrogen atom (quantum system) the scale is set by the Bohr radius. ∂t (1.29) 2 = ∂ µ ∂µ = ∂t2 The value of the antisymmetric tensor ǫµνρσ is determined in the same way as for ǫijk . The derivative ∂µ is defined ∂µ = ∂/∂xµ and we have a four vector ∂ with components (∂0 . = ∂ .35) ρ ǫµνρσ ǫµν σ σ ǫµνρσ ǫµνρ ′ ′ = −2 g ρρ g σσ − g ρσ g σρ = −6g σσ . Examples are the scalar products above but also relations like pµ = m dxµ /dτ for the momentum four vector. . starting from ǫ0123 = 1. but more important has the same effect on lefthandside and righthandside. (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. ∂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 . for instance. ′ ′ . ∂3 ) = ∂ ∂ ∂ ∂ . explicitly written as (p0 . Note that in this equation one has on left. where m is called the mass of the system. .33) (1.34) (1. Each of these permutations leads to a minus sign. (1. because one has (1. It is an invariant tensor.31) ǫµνρσ ǫµν ′ ′ ρσ ′ = − (1. p) = (E. The length squared of ∂ is the d’Alembertian operator. a∞ = 4πǫ0 2 /me e2 . (1. . defined by ∂2 − ∇2 . not affected by Lorentz transformations. For the contraction of a vector with the antisymmetric tensor one often uses the shorthand notation ǫABCD = ǫµνρσ Aµ Bν Cρ Dσ .27) It is easy to convince oneself of the nature of the indices in the above equation. 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.30) (Note that there are different conventions around and sometimes the opposite sign is used). g0 = g1 = g2 = g3 = 1. p). constructed via the metric tensor itself. ∂t ∂x ∂y ∂z ν ∂µ xν = gµ .28) ν ν Note that gµ with one upper and lower index. 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 .32) . (1. (1.36) Exercises Exercise 1.Introduction 6 called covariant. 0123 0123 ′ ′ ǫ µνρσ ǫµνρσ The first identity.

ǫ0 . 1.3. ˆ ˆ . Note that ˆ ˆ ˆ ˆ a · nα = aα and derive from that the components nα of the four vector nα .4 Lightcone coordinates for a four vector a (which we will denote with square brackets as [a− . a+ . a2 . 1. n2 . . 0. n1 . c. 2. Use that to give the ˆ ˆµ ˆ components of the basis vectors n+ and n− . 0. This demonstrates how a careful use of units can save a lot of work. ρ. a1 . (b) The coordinates (a0 . 1) ˆ . n2 . a+ . 0) ˆ n2 = (0. n+ . 0) ˆ n3 = (0. Exercise 1. which is the Compton wavelength for the Planck mass. 0. a3 ) are the expansion coefficients using the basis vectors n0 .13 to give its actual value in kg. aT ]) are defined through √ a± ≡ (a0 ± a3 )/ 2.511 MeV. g+− and g−+ . ˆ ˆ ˆ ˆ These are n1 = (0. Exercise 1.(A · B) C using the properties of the tensor ǫijk given in section 1. How are a+ and a− related to a+ and a− . Also construct and calculate the Planck length Lpl . a2 ] we can find basis vectors n− . α and me c2 = 0. One does not need to know . 0. [Note: what is the MKS unit for V/T?] Exercise 1. (e) Note that /m = − c and use this to calculate in a simple way (avoiding putting in the value λ of e) the Bohr magneton µe = e /2me and the nuclear magneton µp = e /2mp in electronvolts per Tesla (eV/T).71 × 10−39 GeV−2 to construct a mass Mpl (Planck mass). a2 ] or [a− . by a simple few-line reasoning [For instance: If {µ.3 Prove the following relation ǫµνρσ g αβ = ǫανρσ g µβ + ǫµαρσ g νβ + ǫµνασ g ρβ + ǫµνρα g σβ . (d) Use the value of the gravitational constant GN / c5 = 6. 0. . a1 .Introduction 7 (b) Relate the classical radius of the electron (a relativistic concept). ]. (c) Calculate the Compton wavelength of the electron and the quantities under (a) and (b) using the value of c. g−− . e. n3 . a+ . Compare it with the proton mass and use Eq. σ} is a permutation of {0. but only appropriate combinations. ν. . 3} the index α can only be equal to one of the indices in ǫµνρσ .2 Prove the identity A × (B × C) = (A · C) B . 1. n1 . (a) Express the scalar product a · b in lightcone coordinates and deduce from this the values of g++ . me . re = e2 /4πǫ0 me c2 to the Compton wavelength. 1. 0) ˆ n0 = (1. a1 . Also for the coordinates [a− . 0.

5) with (k 0 )2 = k2 + M 2 . ∂t2 (2. ∂t 2M (2.2. t) = 0. φk (r. p). 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. Although it is straightforward to find the solutions of this equation.1) under the substitution (in coordinate representation) E −→ Eop = i ∂ . Thus a covariant equation can be obtained by starting with the (covariant) equation for the invariant length of the four vector (E. 2 + M 2 φ(r.7) k 0 = ± k2 + M 2 = ±Ek . t). namely plane waves characterized by a wave number k.3 are not covariant. ∂t p −→ pop = −i∇. t) = exp(−i k 0 t + i k · r).Chapter 2 Relativistic wave equations 2. (2. we just want to play a bit with covariant equations and study their behavior under Lorentz transformations. written as pµ −→ i∂µ is covariant (the same in every frame of reference).1 and 2. one obtains (2. E = p2 .3) Equations 2. t) = − ψ(r. But the replacement 2.2) Acting on the wave function one finds for a free particle. i ∂ ∇2 ψ(r.6) The energy spectrum is not bounded from below: considering the above stationary plane wave solutions. p2 = pµ pµ = E 2 − p2 = M 2 . The Schr¨dinger equation in quantum mechanics is the operator equation o corresponding to the non-relativistic expression for the energy.4) where M is the particle mass. Substitution of operators gives the Klein-Gordon (KG) equation for a real or complex function φ. 2M (2.1 The Klein-Gordon equation In this chapter. (2. t) = ∂2 − ∇2 + M 2 φ(r. (2. 8 .

2. (2.e.10) ∂t which follows directly from the Schr¨dinger equation. there are no longer fundamental objections to mix up space and time. 2M 2M (2. j) = ↔ i φ∗ ∂0 φ. is a second order equation and φ and ∂φ/∂t can be fixed arbitrarily at a given time.14) ˜ ˜ Introducing φ(Ek . This leads to the existence of negative densities. 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. having the KG equation as a space-time symmetric (classical) equation. The appropriate current corresponding to the KG equation (see Excercise 2. 2.2 Mode expansion of solutions of the KG equation Before quantizing fields.2) is j µ = i φ∗ ∂ µ φ or (ρ. Then. (2. j) of the four-current j. The KG equation. −k) . just as the dependence on time was in ordinary quantum mechanics. we want the most general solution. For this we note that an arbitrary solution for the field φ can always be written as a superposition of plane waves. there are solutions with negative energy. −k) ≡ b∗ (k) one has φ(x) = d3 k e−i k·x a(k) + ei k·x b∗ (k) = φ+ (x) + φ− (x). which is what Lorentz transformations do. φ(x) = d3 k (2π)3 2Ek ˜ ˜ e−i k·x φ(Ek .12) It is easy to see that this current is conserved if φ (and φ∗ ) satisfy the KG equation. k) ≡ a(k) and φ(−Ek . This operator has all possible solutions in it multiplied with creation (and annihilation) operators. ↔ ↔ (2. This continuity equation can be written down o covariantly using the components (ρ.13) ˜ with (in principle complex) coefficients φ(k). φ(x) = d4 k ˜ 2π δ(k 2 − M 2 ) e−i k·x φ(k) (2π)4 (2. rather than as a wave function. −i φ∗ ∇ φ . k) + ei k·x φ(−Ek . And. for which we need the interpretation of φ itself as an operator. it is simply a matter of being careful to find a consistent (covariant) theory. It is possible to rewrite it as an integration over positive energies only but this gives two terms (use the result of exercise 2.11) Therefore.3). relativistically the density is not a scalar quantity. ∂ρ = −∇ · j. As we will see later both problems are related and have to do with the existence of particles and antiparticles.Relativistic wave equations 9 i. however. (2π)3 2Ek (2. but rather the zero component of a four vector. (2. At that point the dependence on position r and time t is just a dependence on numbers/parameters on which the operator depends.8) (2. The integration over k-modes clearly is covariant and restricted to the ‘mass’-shell (as required by Eq. ∂µ j µ = 0.15) .9) They satisfy the continuity equation.5).

(2.22) showing that with each solution there is a corresponding charge conjugated solution φC (x) = φ∗ (x). 2. They are referred to as mode and anti-mode amplitudes (or because of their origin positive and negative energy modes). which implies (xµ ) = (t.14 and 2.18) (2. one thus must have that the function φ → φ′ such that φ′ (x′ ) = φ(x) or φ′ (x) = φ(Λ−1 x). after restricting the energies to be positive).19) implying that for each solution φ(x) there exists a corresponding solution with the same energy.23) .16) The consequence of this is discussed in Exercise 2. φP (x) ≡ φ(˜) (P for parity). The coefficients a(k) and b∗ (k) are the amplitudes of the two independent solutions (two.e. −x) ≡ (˜µ ).Relativistic wave equations 10 In Eqs 2. This shows how parity transforms k-modes into −k modes.6 We will explicitly discuss the example of a discrete symmetry.21) d3 k x x e−i k·˜ a(k) + ei k·˜ b∗ (k) (2π)3 2Ek d3 k ˜ ˜ e−i k·x a(k) + ei k·x b∗ (k) (2π)3 2Ek d3 k e−i k·x a(−k) + ei k·x b∗ (−k) . (2. i. x ∂µ ∂ µ + M 2 φ(˜) = 0.17) (2. for which we consider space inversion. Since a · b = a · ˜ it is easy to see that ˜ b. Another symmetry is found by complex conjugating the KG equation.15 one has elimated k 0 and in both equations k · x = Ek t − k · x.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. (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).20) one has for the mode amplitudes aP (k) = a(−k) and bP (k) = b(−k). (2π)3 2Ek (2. It is easy to show that x φP (x) = φ(˜) x = = = or since one can define φP (x) ≡ d3 k e−i k·x aP (k) + ei k·x bP ∗ (k) . x) → (t. (2π)3 2Ek (2. Performing some Lorentz transformation x → x′ = Λx. changing the sign of the spatial coordinates. x Transforming everywhere in the KG equation x → x one obtains ˜ ˜ ˜ x ∂µ ∂ µ + M 2 φ(˜) = 0. (2. It is trivial to see that ∂µ ∂ µ + M 2 φ∗ (x) = 0. 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) .

2 Show that if φ and φ∗ are solutions of the KG equation. Note that a(k) and b(k) are independent of t. t). t) = ei Ek t (i∂0 + Ek ) φ(k. ↔ ↔ Exercise 2. ˙ and thus for any normalized solution the full ‘charge’. d3 k F (Ek . This shows how charge conjugation transforms ’particle’ modes into ’antiparticle’ modes and vice versa. 2.4 ˜ Write down the 3-dimensional Fourier transform φ(k. t). k) = (2π)4 k2 + M 2 . Use these to show that ˜ ˜ a(k) = ei Ek t i ∂0 φ(k. (2π)3 2Ek where Ek = Exercise 2. QV ≡ satisfies ˙ QV = − ds · j.3 Show that1 d4 k 2π δ(k 2 − M 2 ) θ(k 0 ) F (k 0 . Q = 0.15. ˜ b(k) = ei Ek t i ∂0 φ∗ (−k. Exercise 2.e. For the real field one has aC (k) = a(k). 1 Use ↔ ↔ the following property of delta functions δ (f (x)) = zeros xn 1 δ(x − xn ). t) exp(−i k · x) and its time ˜ derivative i∂0 φ(k. Exercises Exercise 2. S V d3 x j 0 (x). t) for the mode expansion in Eq.1 Show that for a conserved current (∂µ j µ = 0) the charge in a finite volume. k). |f ′ (xn )| . is conserved. that j µ = i φ∗ ∂ µ φ is a conserved current Note: A ∂µ B ≡ A∂µ B − (∂µ A)B). for the mode amplitudes aC (k) = b(k) and bC (k) = a(k).Relativistic wave equations 11 i. t) ≡ d3 x φ(x. letting V → ∞.

Note that a{µ bν} indicates symmetrization. αj } = 2 δ ij I.7 To solve the problem with positive and negative energies and get a positive definite density.2) in terms of the a(k) and b(k) using the expansion in Eq. We will encounter these quantities later as energy and momentum. 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). 2.6 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). From this one concludes that α and β must be matrix-valued and ψ must be a multi-component wave function. d3 x (∂ {0 φ)∗ (∂ i} φ). {αi . and β 2 = I.Relativistic wave equations 12 Exercise 2. also using the result in (a). β} = 0.15. 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 .e. . 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). in terms of the a(k) and b(k). i (a) Show that relativistic invariance. 1 2 (b) Express now the full charge QV (t) (exercise 2. 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 . (c) Similarly. (b) Show that one has current conservation for the current 2 We denote the commutator as [A. One then finds that Q is expressed as an infinite sum over number operators a† a (particles) and b† b (anti-particles). express the quantities E(t) = P i (t) = d3 x (∂0 φ)∗ (∂0 φ) + ∇φ∗ · ∇φ + M 2 φ∗ φ . j 0 = ψ†ψ and j = ψ † αψ. t) = (−i α · ∇ + m β) ψ(r. B} = AB + BA. Exercise 2. Exercise 2. an attempt to construct a first order differential equation for the time evolution would be to write. B] = AB − BA and the anticommutator as {A.5 (a) As an introduction to parts (b) and (c). a{µ bν} ≡ aµ bν + aν bµ . t).1) for a complex scalar field current (exercise 2. ∂ ψ(r. ∂t where ψ is a wave function and the nature of α and β is left open for the moment. i.

z ) Lz = i ∂ϕ   0   = i   0  −i 0    0 0 .g. electrons. In particular. in which the parameter space forms locally a Euclidean space.g. z ) = 1 + i δϕ Lz with as generator ˆ 1 ∂R(ϕ.1 The rotation group and SU (2) The rotation groups SO(3) and SU (2) are examples of Lie groups. however. z )V . 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. The KG equation will actually remain valid.e. provided we identify the antipodes. the symmetry group describing rotations is embedded in the Lorentz group. n) ≡ R(π. 0 ≤ ϕ ≤ π. e. Particles with spin then will be described by certain spinors. For an infinitesimal rotation around the z-axis one has ˆ R(δϕ. In quantum mechanics spin is described by a vector. Since the KG equation expresses just the relativistic relation between energy and momentum.3) ϕ=0 13 .e. e. i. 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 (3.1) for a rotation around the z-axis or shorthand V ′ = R(ϕ. we also want it to hold for particles with spin. −n). in particular each component of these spinors will satisfy this equation. 3. Locally this parameter-space is 3-dimensional and correspondingly one has three generators. and we must study the representations of the Lorentz group.2) (3. we want to investigate if there exist objects other than just a scalar (real or complex) field φ. i. In this section we will investigate the requirements imposed by Poincar´ e invariance. Any rotation can be uniquely written as R(ϕ. R(π. that is groups characterized by a finite number of real parameters. it has 3 components that we know the behavior of under rotations. In a relativistic theory.   0 0 (3. The parameter-space of SO(3) is a ˆ ˆ ˆ sphere with radius π. Before proceeding with the Lorentz group we will first discuss the rotation group as an example of a Lie group with and a group we are familiar with in ordinary quantum mechanics. n) where n is a unit vector and ˆ ˆ ϕ is the rotation angle.Chapter 3 Groups and their representations The simple systems that we want to describe in a relativistically invariant way are free particles with spin.

namely that there exists a bilinear product [. These generators form a threedimensional Lie algebra SO(3). Next.2 π) (-n. y ∈ A ⇒ [x.Groups and their representations 14 (n.5) Rotations in general do not commute. with real a’s and 3 (ai )2 = 1. • [x. It is straightforward to check that any (finite) rotation can be obtained from a combination of infinitesimal rotations. [z. x] = 0 (thus [x. Locally this is a 3-dimensional Euclidean space and SU (2). 2 (3.4) R .6) Summarizing. y]] = 0 (Jacobi identity). In the same way we can consider rotations around the x.1: the parameter spaces of SO(3) (left) and SU (2) (right). n) = 1 cos ˆ 1 Note ϕ ϕ ˆ + i (σ · n) sin 2 2 = exp i ϕ ˆ σ·n .and y-axes     0 0 0   0 0    0 0 −i  . . 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. is a 3-dimensional Lie-group.8) ˆ that (σ · a)(σ · b) = a · b + i σ · (a × b). π) (n. π ) π -sphere (antipodes identified) 2 π -sphere (surface identified) Figure 3. z]] + [y. Ly and Lz . They satisfy the commutation relations [Li . z ) = lim N →∞ that are generated by  i    0 . [y. ˆ R(ϕ.   0 (3.    Lx =  Ly =  0 0       0 i 0 −i 0 ϕ ˆ . With matrix commutation this algebra satisfies the requirements for a Lie algebra. Lj ] = i ǫijk Lk . Writing a0 = cos(ϕ/2) and a = n sin(ϕ/2) we have1 A = A(ϕ. which reflects itself in the noncommutation of the generators. (3. the rotations in SO(3) can be generated from infinitesimal rotations that can be expressed in terms of a basis of three generators Lx . (3. we turn to the group SU (2) of special (det A = 1) unitary (A† = A−1 ) 2 × 2 matrices. These matrices can be defined as acting on 2-component spinors (χ → Aχ) or equivalently as acting on 2 × 2 matrices (B → ABA† ). ] that satisfies • ∀ x. x]). and thus (σ · n)2 = 1. One way of viewing the parameter space.z N N or (Lk )ij = −i ǫijk . for rotations around z for instance. y] = −[y. y] ∈ A. x]] + [z.7) = a0         0 −1 i 0 1 0 0 1 • [x. [x. thus is as the surface i=0 of a sphere in 4 Euclidean dimensions. ˆ therefore.

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

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

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

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 (3.21)  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 3.27) (3. Writing x as a column vector and the metric tensor in matrix form. 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.     0 0   g00 g01 g02 g03   1 0      g     g11 g12 g13   0 −1 0 0  = . it is easy to derive some properties of the matrices Λ: (i) det(Λ) = ±1. In this section we consider the Poincar´ group. We considered the generators and looked for representations in finite dimensional spaces. e To derive some of the properties of the Lorentz group. e. which as a matrix equation with (Λ)µν = Λµν and (G)µν = gµν gives (ΛT )ρµ (G)µν (Λ)νσ = (G)ρσ .22) (Note that xT is a row vector). explicitly x′µ = (Λ)µν xν + aµ or x′µ = Λµν xν + aµ . it is convenient to use a vector notation for the points in Minkowski space.g. Thus one has the (pseudo-orthogonality) relation ΛT GΛ = G ⇔ GΛT G = Λ−1 ⇔ ΛGΛT = G. det Λ = −1 is called improper). (ii) |Λ00 | ≥ 1. From this property. proof: det(ΛT GΛ) = det(G) → (det Λ)2 = 1.25) (3. Invariance of the length of a vector requires for the Lorentz transformations x′2 = gµν x′µ x′ν = gµν Λµρ xρ Λνσ xσ = x2 = gρσ xρ xσ or Λµρ gµν Λνσ = gρσ .26) (3. (det Λ = +1 is called proper.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.     G =  10 (3.24) L↑ + L↓ + L↑ − L↓ − proper orthochrone proper non-orthochrone improper orthochrone improper non-orthochrone . proof: (ΛT GΛ)00 = (G)00 = 1 → Λµ0 gµν Λν0 = 1 → (Λ00 )2 − i (Λi 0 )2 = 1. consisting of the Lorentz group and translations.23) The proper tensor structure of the matrix element (Λ)µν is a tensor with one upper and one lower index. (3.

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

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

Λµν pν ⇒ ωµν pν ⇒ ωµν = = = pµ + ω µν pν = pµ 0 ǫµνρσ pρ sσ . [K i .47) (3. For instance. We have here reinstated c. a) = Λ−1 µ P ν . . K j ] = −i ǫijk J k /c2 . Infinitesimally.48) . [J i . a)P µ U (Λ. the generators M αβ no longer exclusively act in Minkowski space. P j ] = [P i .38 could have been obtained in the same way. P ρ ] = 0. = −i (g µρ P ν − g νρ P µ ) . = pµ |p. ν (3. Taking any one of the states in an irreducible representation. (3. the generators J± in the case of the rotation group. (3. .Groups and their representations 21 infinitesimal form of any representation of the Poincar´ group (thus including the translations) is of e the form i U (I + ω. (3. . ǫ) = 1 − ωαβ M αβ + i ǫα P α . that commute e among themselves. P ν ] [M µν . To find other generators that commute with P µ we look for Lorentz transformations that leave the four vector pµ invariant.42) (to decide on where the inverse is. The generators of the translations are the (momentum) operators P µ . J j ] = i ǫijk J k . . In order to label the states in an irreducible representation we construct a maximal set of commuting operators. . Explicitly. H] = [J i . the tensor M µν in terms of boosts cK i = M i0 and rotations J i = 1 ǫijk M jk . known from non-relativistic quantum mechanics are obtained. These form a group called the little group associated with that four vector. [K i . The eigenvalues of these will be the four-momentum pµ of the state. (1 + iǫγ P γ − i i ωαβ M αβ )P µ (1 − iǫδ P δ + ωρσ M ρσ ) = P µ + ω µν P ν . Of the generators of the Poincar´ group we choose first of all the generators P µ .44) (3.43) giving the following commutation relations by equating the coefficients of ǫµ and ωµν . 2 2 (3. The requirement that P µ transforms as a four-vector (for which we know the explicit behavior from the defining four-dimensional representation). 3.45) Note that the commutation relations among the generators M µν in Eq. the generators J 2 and J 3 in the case of the rotation group. . For instance.46) [K i . . [P µ . H] = i P i . P µ |p. a)P µ U † (Λ. writing the generator P = (H/c. one obtains 2 [P i . They just state that M µν transforms as a tensor with two Lorentz indices. other states in that representation are obtained by the action of operators outside the maximal commuting set. K j ] = i ǫijk K k . because one then sees that by letting c → ∞ the commutation relations of the Galilei group. [J i . [J i .41) 2 At this point. These define states with specified quantum numbers that are eigenvalues of these generators. P j ] = i ǫijk P k . leads to U (Λ. as part of the set. check that U1 U2 corresponds with Λ1 Λ2 ). a) = Λµν P ν or U † (Λ. (3. P j ] = i δ ij H/c2 . P ) in terms of the Hamiltonian and the three-momentum operators. H] = 0. 3.5 Representations of the Poincar´ group e where pµ is a set of four arbitrary real numbers.

.54) This will enable us to pick a suitable commuting ’spin’ operator. i exp − ωµν M µν 2 i exp − ǫµνρσ M µν pρ sσ 2 (3. 2 (3. (3.Groups and their representations 22 where s is an arbitrary vector of which the length and the component along pσ are irrelevant.53) (3.52) (3. of which the components generate the little group of pµ or (for the third of the following relations) from explicit calculation [Mµν . . . while the others have no physical significance. Given the momentum four vector pµ we choose a tetrade consisting of three orthogonal spacelike unit vectors ni (p).55) commute with all generators and therefore are invariants under Poincar´ transformations. which can thus be chosen spacelike. (3. These e operators are the Casimir operators of the algebra and can be used to define the representations of ↑ P+ . .51) The following properties follow from the fact that W µ is a four-vector by construction. Massive particles: p2 = M 2 > 0.50) where the Pauli-Lubanski operators W µ are given by 1 Wµ = − ǫµνρσ M νρ P σ . M ǫijk . Wα ] = [Wµ . The elements of the little group thus are U (Λ(p)) = = and U (Λ(p))|p. the third case represents the vacuum. 0. for which we distinguish the following cases p 2 = m2 > 0 p2 = 0 pµ ≡ 0 p 2 = m2 > 0 p2 = 0 p2 < 0 p0 > 0 p0 > 0 p0 < 0 p0 < 0 Only the first two cases correspond to physical states. . 0. . i ǫµνρσ W ρ P σ . .56) . p0 > 0. Pν ] = [Wµ . 2 exp (−isµ W µ ) |p. satisfying gµν nµ (p)nν (p) = j i nµ (p) pµ i ǫµνρσ pµ nν nρ nσ i j k = = −δij .49) (3. = = i exp − ǫµνρσ M µν pρ sσ |p. . . Wν ] = −i (gµα Wν − gνα Wµ ) . From the algebra of the generators one finds that P 2 = Pµ P µ and W 2 = Wµ W µ (3.

ms . 0). n1 (p). S j (p)] = −i nµ (p)nν (p) j i ǫµνρσ W ρ P σ = i ǫijk S k (p). 0).59) (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 . (3. Having made a covariant decomposition. it is sufficient to choose a particular frame to investigate the coefficients in Eq.62) 1 Thus W 2 has the eigenvalues −M 2 s(s + 1) with s = 0. . 1 2 (3. the S i (p) form the generators of an SU (2) subgroup that belongs to the maximal set of commuting operators. Together with the four momentum states thus can be labeled as |M. n2 (p) and s(p) are obtained by a Lorentz boost from the reference frame one has the property ǫµνρσ sµ nν nρ pσ > 0. In that case the vectors ni (p) are just the space directions and W = (0. p0 > 0. 2 . 0. show that W /M can be interpreted as the intrinsic spin. The explicit construction of the wave function can be done using the D-functions (analogous as the rotation functions).57) (i. 0. (0. (1. The commutation relations [W i (p). . We will not do this but construct them as solutions of a wave equation to be discussed explicitly in chapter 4 Massless particles: p2 = M 2 = 0. M2 (3. i (3.58) i. 0. |p3 |). 1.e.64) such that in an arbitrary frame where the four vectors p. (3. M2 (3. .e.65) . 3. 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.63) where E = p2 + M 2 and ms is the z-component of the spin −s ≤ ms ≤ +s (in steps of one).61) one sees that S i (p) i =− W2 . 0.Groups and their representations 23 We can write W µ (p) = 3 W i (p) nµ (p) i i=1 (3. 0). s. 1. M J) with components W i = (M/2)ǫijk M jk . W i (p) = −W · ni (p)). 1.. (0. W j (p)] = i M ǫijk W k (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. p. Noting that S i (p) i 2 = 1 W µ (p)W ν (p) M2 niµ (p)niν (p). More generally (fully covariantly) one can proceed by defining S i (p) ≡ W i (p)/M and obtain [S i (p).

which can take any integer or half-integer value. Thus we have P 2 . 3. |p. A massless particle. 3. We don’t know of any physical relevance. Wν ].67) (3. P µ . W 2 and λ is derived from the general commutation relations for [Wµ .1 (a) Derive from the relation that defines the mapping of SU (2) matrices A into SO(3) matrices RA in Eq. thus. σ · RA a ≡ A(σ · a)A−1 ˆ (valid for any vector a). Exercises Exercise 3. and gives [W 1 (p). This state is in physical applications nondegenerate and is denoted by |0 .69) We reproduce the algebra for the massive case with W 3 (p) = M λ(p).71) which is called the helicity.66) where W · p = 0 implies that W s (p) = −(M 2 /p · s) λ(p). 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.67 has a vanishing right hand side). W 2 (p)] [λ(p). implying continuous values for the spin (actually for W i = M S i ). This algebra is (for instance) isomorphic to the one generated by rotations and translations in a 2-dimensional Euclidean plane (see exercises). 1 2 24 (3. where A = exp(i ϕ · σ/2) = exp(iϕ σ · n/2). i W 2 (p). U (Λ. The vacuum: pµ = 0. The algebra of W 1 . −i W 1 (p). λ . W 1 (p)] [λ(p).68) (3. (3. (3.73) (3.Groups and their representations The above does actually include the massive case.72) One has P µ |0 = W µ |0 = 0 and the state is invariant under Lorentz transformations.70) as a commuting set with zero eigenvalues for P 2 and W 2 . W 2 (3. But if M = 0 one has a different algebra (Eq. however.13. Thus 2 λ(p) = J ·p . 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). The eigenvalues of W 1 and W 2 (in that case corresponding to the translations) thus can take any continuous values. is characterized by its momentum and the helicity. the action of RA b · RA a = = ˆ ˆ ˆ b · a cos ϕ + n · (b × a) sin ϕ + 2(b · n)(a · n) sin2 (ϕ/2) ˆ ˆ b · a + b · (ˆ × n) sin ϕ − 2 (b · a − (b · n)(a · n)) sin2 (ϕ/2) a . W 2 (p)] = = = i M 2 λ(p). |p| (3. Note that angular momentum does not contribute to helicity as L · p = 0. 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). W µ (p) = λ(p) P µ .

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

e. they describe r −→ r′ = r − u t while t′ = t. = p.7 (optional) One might wonder if it is actually possible to write down a set of operators that generate the Poincar´ e transformations. although the presence of an interaction term V (r1 . rj ] = [pi .. . the others vanish. r2 ) inevitably leads to interaction terms in the boost operators.69 of the Pauli-Lubanski operators for massless particles. Exercise 3.5 In classical mechanics (Gallilean invariance) boosts describe invariance of physical laws when viewed from a moving frame. you might just check relations involving J or K by taking some (relevant) explicit components. sj ] = [pi . in particular those with the momentum operator pj and the Hamiltonian H = mc2 + p2 /2m. that’s why many-particle non-relativistic quantum mechanics is ’easy’. These do not matter in the non-relativistic limit (c → ∞). but it can be done. For a single free particle or for the free-moving center of mass system that is the quantity K = mr − tp. the operator U (u) that gives U (u) r U −1 (u) = r − u 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. (b) In quantum mechanics one has the Ehrenfest relation i dO = [O.6 Consider rotations and translations in a plane (two-dimensional Euclidean space). although that in this case is a relation that is found already from the Ehrenfest relation for O = ri . there is a one to one mapping) the algebra in Eqs 3. Comment: extending this to more particles is a highly non-trivial procedure. if you don’t want to do this in general.Groups and their representations 26 Exercise 3. derive the commutation relations and show that the algebra is isomorphic to (i.e. consistent with the (canonical) commutation relations of a quantum theory. pj ] = iδ ij and [si . (a) In classical mechanics. indeed assures d K /dt = 0.e. = r × p + s. = 2c2 H + mc2 satisfy the commutation relations of the Poincar´ group if the position. This is possible for a single particle.67 to 3. and U (u) p U −1 (u) = p − m u. boost invariance leads to a conserved quantity. From either of these. H] + i dt ∂O . [ri . show first the operator identity [r. Hint: for the Hamiltonian. sj ] = 0. (c) Give the unitary operator for boosts. Exercise 3. i. Do this by showing that the set of operators. f (p)] = i ∇p f (p). [ri . i. pj ] = [ri . H P J K = p 2 c 2 + m2 c4 . p×s 1 (rH + Hr) − t p + . momentum and spin operators e satisfy the canonical commutation relations. ∂t Show that boost invariance for a free-moving particle (the commutation relations under (a) being valid). sj ] = iǫijk sk . Determine the commutation relations for the corresponding operators Ki in quantum mechanics.

10) . K = i J (A = 0). The group SL(2. 0. K = −i . (j. becoming the defining representations of SL(2. M = exp i ϕ·σ 2 1 exp ± η · σ 2 = U (ϕ) H(η) . With this choice of parameter-spaces the plus and minus signs are actually relevant. This tells us how to find the representations of the group. (4. similarly as the homeomorphism between SO(3) and SU (2). j ′ ). 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 . B j ] = [Ai . They precisely correspond to the two types of representations that we have seen before. j) K = −i J (B = 0). B j ] = i ǫijk Ak .3) (4. C). i ǫijk B k . They will be labeled by two angular momenta corresponding to the A and B groups. 2 1 (J − i K). U (ϕ) H(η) (4. C) can be written as a product of a unitary + matrix U and a hermitean matrix H. it also follows directly that the Lorentz group is homeomorphic with the group SL(2.6) (4. It is simply connected and forms the covering group of L↑ .1 The Lorentz group and SL(2. C) is the group of complex 2 × 2 matrices with determinant one.2) This shows that the Lie algebra of the Lorentz group is identical to that of SU (2) ⊗ SU (2).8) ˆ ˆ ˆ with η = η n and ϕ = ϕn.1) (4. 2 2 (4.4) (4. Special cases are the following representations: Type I : (j.5) = 1 (J + i K). 2 2 σ σ J = .Chapter 4 The Dirac equation 4. K = +i . Aj ] = [B i . respectively.7) From the considerations above. (4. 0) Type II : (0. C): Type I (denoted M ): Type II (denoted M ): 27 J= σ σ .9) (4. where we restrict (for fixed n) the parameters 0 ≤ ϕ ≤ 2π and 0 ≤ η < ∞. 2 (4. The matrices in SL(2.

One defines spinors ξ and η transforming similarly under unitary rotations (U † = U −1 . ˜ = σ µ Λ B aµ . C) and.14) satisfying Tr(σ µ σ ν ) = −2 gµν and Tr(σ µ σ ν ) = 2gµν = 2 δ µν (the matrices. thus. (4. σ).15) (4. U σ µ aµ U ˜ −1 H σ µ aµ H −1 = σ µ ΛB aµ H σ µ aµ H ˜ −1 (4. H ≡ (H † )−1 = H −1 ). Eqs 4. C).16) = σ µ Λ R aµ .The Dirac equation 28 Let us investigate the defining (two-dimensional) representations of SL(2. With E = M γ = M cosh(η) and p = M βγ n = M sinh(η) n. cannot be connected by a Lorentz transformation belonging to L↑ . Note that M ∗ = ǫ† M ǫ has its equivalent for the Lorentz transformations Λ(M ) and Λ(M ∗ ).e. 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. 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). ˜ Non-equivalence of M and M One might wonder if type I and type II representations are not equivalent. η → Hη. ǫ−1 = ǫ† = ǫT = −ǫ). In fact. there exists the matrix ǫ = iσ 2 . one can use a boost to the frame with momentum ˆ ˆ p. namely H(η) = exp(η · σ/2).19) . Λ(ǫ) ∈ L+ . 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. U σ µ aµ U −1 = σ µ ΛR aµ . One has + −1 Λ(M ∗ ) = I2 Λ(M ) I2 . (4.18) As already discussed for SU (2). if there does not exist a unitary matrix S. σ µ ≡ (1. while ±ǫη ∗ transforms as a type-I + spinor.1). U ≡ (U † )−1 = U ) U (ϕ) = exp(i ϕ · σ/2) but differently under hermitean boosts (H † = H. ξ −→ Hξ.12) ξ → U ξ.18 show that ±ǫξ ∗ transforms as a type-II (η) spinor. ˜ The sets of four operators defined by definitions σ µ ≡ (1. ¯ ¯ H ∗ = ǫ† Hǫ = ǫ−1 Hǫ. (4. ˜ (4. Also useful is the relation H 2 (p) = σ µ pµ /M = (E + σ · p)/M . C) or L↑ . i. C). The Lorentz transformations Λ(M ) and Λ(M ).13) 2M (E + M ) (exercise 4. however. (4. The following relations for + rotations and boosts are useful. such that M = SM S −1 . −σ). This shows that M ∗ ≃ M . η → U η.11) Considering ξ and η as spin states in the rest-frame. (4. (ǫ∗ = ǫ. are not covariant!) ˜ can be used to establish the homeomorphism between L↑ and SL(2. H(η) = exp(−η · σ/2).17) that can be used to relate σ µ∗ = ǫ† σ µ ǫ = ǫ−1 σ µ ǫ and hence ˜ ¯ U ∗ = ǫ† U ǫ = ǫ† U ǫ. This is not true for the two-component spinors in SL(2. thus.

but by a transformation belonging ↑ ↑ to the class P− . Examples (with between brackets the Euclidean assigments) are r t p H J λ(p) K −→ −λ(p) −→ −K −→ H −→ J −→ t (scalar). as we have seen in section 4. ) −→ ( . a) and ˜ a = (a0 .e.g.13. 0) of SL(2. Thus M ∗ and M are not equivalent within SL(2.      (4. C). we have seen that in the rest frame W i (p)/M → J i . and the angular momentum in the rest frame is what we are familiar with as the spin of a particle. the aij elements of a tensor will not change sign. (vector). 2 2 (4. where a = (a0 .      0   χm   uR   H(p)      . Although these operators cannot be used to label the representations. (scalar). 1 ) and ( 1 . It has the same effect as lowering the indices. the boost in Eq.1. they can be connected. i. 1 ) and ( 1 . 0). Within the Lorentz group. A parity transformation changes aµ into aµ . ǫµνρσ will change sign.21) In nature parity turns (often) out to be a good quantum number for elementary particle states. m) = η(0. that have the correct commutation relations. must combine the repe resentations. 0) (via operators J and K) one sees that under parity 2 2 1 1 (0. etc. we obtain for the two components of u which we will refer to as chiral right and chiral left components. (axial vector). 0) are inequivalent. e. The representing member of the class P− is the parity or space inversion operator Is . −→ −r (vector). i. consider the four component spinor that transforms under a Lorentz transformation as      0  ξ   ξ  . Taking M (Λ) = H(p).22) u=     η η M (Λ) 0 where M (Λ) = ǫM ∗ (Λ)ǫ−1 with ǫ given in Eq. 4. 4.   −→  M (Λ)   (4. but rather the operators W i (p) should be used.20) 4.23) u(p. For the spin 1/2 representations of the Poincar´ group including parity we. (pseudoscalar). From the definition of the representations (0. the well-known two-component spinor for a spin 1/2 particle. 2 ) and ( 2 . + −   1 0    0 1  I2 =    0 0   0 0 0 0 −1 0 0 0 0 1      . The i angular momentum operators J are represented by σ i /2.       = (4. but to L↑ . For a particle at rest only angular momentum is important and we can choose ξ(0. The behavior is the same for classical quantities. m) = χm .e.17. C) are suitable for representing spin 1/2 particles. This provides the easiest way of seeing ˜ what is happening. they cannot be connected by a unitary 2 2 transformation.The Dirac equation 29 where The matrix I2 does not belong to L↑ . therefore. We also have seen that the representations (0. m) =            χm uL H(p) 0 .2 Spin 1/2 representations of the Lorentz group 1 1 Both representations (0. −a). −→ −p (vector). generators.

Since ∂µ ∂ν is symmetric. . one must require that for ψ(x) the Klein-Gordon relation 2 + M 2 ψ(x) = 0 is valid for each component separately). which in general is a linear equation in pµ .33) define for a four vector a the contraction a = aµ γµ / . To see this consider the equation for the hermitean † † conjugate spinor ψ † (noting that γ0 = γ0 and γi = −γi ).32) suppressing on the RHS the identity matrix in Dirac space. The explicit realization appearing in Eq.The Dirac equation 30 with H(p) = ¯ H(p) = 2M (E + M ) E+M −σ·p 2M (E + M ) E+M +σ·p .7. The general requirements for the γ matrices are thus easily obtained. Applying (iγ µ ∂µ + M ) from the left gives (4. We will discuss another explicit realization of this algebra in the next section. From this one obtains the Clifford algebra for the Dirac matrices.28) which is a covariant (linear) first order differential equation.25) It is straightforward to eliminate χm and obtain the following constraint on the components of u. (4. ψ † −iγ 0 ∂0 +iγ i ∂i −M 1 We ← ← = 0.      0 H 2 (p)   uR   uR        . We first want to investigate if the Dirac equation solves the problem with the negative densities and negative energies.27) where γ µ are 4 × 4 matrices called the Dirac matrices1 . 4. (4. 2 (4. γ ν } = 2 g µν . {γ µ .24) (4. (iγ µ ∂µ − M ) ψ(x) = 0.31) To achieve also that the energy-momentum relation p2 = M 2 is satisfied for u(p). = H −1 (p). (4. As in section 2 we can use Pµ = i∂µ as a representation for the momenta (translation operators) in function space.27 is known as the Weyl representation.   (4. this can be rewritten 1 µ ν {γ . (pµ γ µ − M ) u(p) ≡ (/ − M ) u(p) = 0.29) which is an explicit realization of the (momentum space) Dirac equation. γ } ∂µ ∂ν + M 2 ψ(x) = 0. (4. It is of a form that we also played with in Exercise 2.30) γ µ γ ν ∂µ ∂ν + M 2 ψ(x) = 0.  =       (4.26)        −2    uL uL H (p) 0 or explicitly in the socalled Weyl representation  −M E+σ·p       E−σ·p −M    uR      uL      = 0. The first indeed is solved. This leads to the Dirac equation for ψ(x) = u(p)e−i p·x in coordinate space. p (4.

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

K 3 = M 30 =    0 iσ 3 2 4 or in general one has the transformation i ψ → Lψ = exp − ωρσ S ρσ ψ.10 are   1  σ3 0  i   J 3 = M 12 =   = [γ 1 . Another matrix which is often used is γ5 defined as γ5 = i γ 0 γ 1 γ 2 γ 3 = −i γ0 γ1 γ2 γ3 = It satisfies {γ5 . We note that ψ → Lψ and ψ → ψ L−1 .   0 γ k = −iρ2 ⊗ σ k =  k  σ  −σ k  . 1 −1 2 For all representations one has † γµ = γ0 γµ γ0 .R.  =ρ ⊗1=  0 −1 3 (4. the rotation and boost generators in Weyl representation in Eqs 4. it is easy to see that PR/L = 1 (1 ± γ5 ) 2    1 0  .R. −→ Sγµ S −1 .48) are projection operators. γ 2 ]. is   1  1 1    (4. = ρ ⊗ 1 =   1 0 i ǫµνρσ µ ν ρ σ γ γ γ γ .The Dirac equation 32 The Weyl (or chiral) representation:    0 1   . 2 (4.41) (4.  (γ5 )S.47) (4. −→ Sψ.  0 (4. which in the case of the Weyl representation are just the upper and lower components. (4. 4! (4.R. For instance in the Weyl representation (but valid generally). Lorentz invariance The Lorentz transformations can also be written in terms of Dirac matrices. S −1 . that project out chiral right/left states. γ 0 ].43) S=√  . ν The latter assures Lorentz invariance of the Dirac equation (see Exercise 4.40) (4.R.46) (4.44) and an adjoint spinor defined by ψ = ψ † γ0 . For example. γ µ } = 0 and explicitly one has    0 1  1 . γ 0 = ρ1 ⊗ 1 =   1 0 γµ ψ Different representations can be related to each other by unitary transformations.9 and 4.42) We note that the explicit matrix taking us from the Weyl representation to the standard representation. γ σ ]. (γµ )S.3) . 0 σ3 2 4   1  −iσ 3 0   = − i [γ 3 .49) i 1 with −i S ρσ = 2 σ ρσ = − 4 [γ ρ . while L−1 γ µ L = Λµ γ ν .45) (γ5 )W. = S(γµ )W.

   −iσ 2   0 −ǫ  = . x) → x = (t. for instance. is a left-handed spinor.R. in Weyl representation we find in Dirac space     T C  ξ   −ǫ η ∗     . usually to be taken unity.R. Note that we have (as expected) explicitly in Weyl x representation in Dirac space     P  ξ    −→ ψ p = γ 0 ψ =  η  .57) (4. γ5 ] = 0. T (4.) and all representations connected via a real (up to an overall phase) matrix S. (or W. = =   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. ψ= (4.    0 −ǫ 0    0   −ǫ 0  .g. −x) and via a few manipulations obtain again the ˜ Dirac equation (i/ − M ) ψ p (x) = 0. C is real and one has C −1 = C † = C T = −C c and [C. e. As with parity we look for a transformation. (C)W. reverse the charge of the particle.51) The existence of positive and negative energy solutions implies another symmetry in the Dirac equation.54) e.58) where ηc is an arbitrary (unobservable) phase. One has ψ c = −ψ T C −1 .56) (i/ − M ) ψ c (x) = 0. In S. ∂ (4. =    0 ǫ iσ 2 (4. (i/ − M ) ψ(x) = 0. In any representation a matrix C exist. but it does. called charge conjugation. The latter implies that the conjugate of a right-handed spinor.R. x = (t. It is easy to convince oneself that if ψ(x) is a solution of the Dirac equation. ∂ (4.g. Explicitly. 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. parity. (C)S.52)    η ξ one can apply space inversion. Some properties of C are C −1 = C † and C T = −C. which is again a solution of the Dirac equation. such that T Cγµ C −1 = −γµ . that brings ψ → ψ c .50) Charge conjugation but with ψ p (x) ≡ ηp γ0 ψ(˜) (Exercise 4.53) (4. This symmetry does not change the spin 1/2 nature. ∂ with the solution ψ c (x) = ηc Cψ (x).R.55) (4.   ψ= (4.The Dirac equation 33 Parity There are a number of symmetries in the Dirac equation.59)  −→ ψ c = C ψ =    η ǫ ξ∗ . ψR . T (4.6).

ψ(x) = u±s (p) e−i p·x . where v satisfies   σ·p  −(Ep + M )      v(p) = 0. s) and C v T (p. the spin polarization vector in the rest frame is the starting ˆ point. It is easy to check that   1  1±σ·s 1 ± γ5 / s ˆ 0    (4. s′ ) = 2Ep δss′ . s).67) It is straightforward to find projection operators for the positive and negative energy states P+ P− = s p /+M u(p. s ) = v † (p. 2M 2M −/ + M p v(p.65) (4. in which γ 0 is diagonal (see discussion of negative energy states in section 4. ψ(x) = v ±s (p) ei p·x . s)v(p.     ∂ −iσ · ∇ −i ∂t − M (4.66) An arbitrary spin 1/2 field can be expanded in the independent solutions.64) (4. be obtained by a Lorentz transformation. s)v(p. Note that the spinors in the negative energy modes (antiparticles) could be two different spinors. s) v = .62) where χs are two independent (s = ±) two-component spinors. v (p. s)u(p. (2π)3 2Ek (4.   ⇔ (/ + M ) v(p) = 0.68) (4. s ) = v (p. s) + v(k. s). In an arbitrary frame one has s · p = 0 and s(p) can e. s) = Ep + M      σ ·p   Ep +M χs     . p     −σ · p (Ep − M )     Ep + M     σ ·p ¯ Ep +M χs χs ¯ For a free massive particle. where u satisfies   −σ · p  Ep − M      u(p) = 0. After separating positive and negative energy solutions as done in the case of the scalar field one has ψ(x) = s d3 k u(k. s)v(p.69) = − s In order to project out spin states. ¯ ¯ † ′ u (p.63) u(p. s)¯(p.60)   ψ(x) = 0. 2M 2M (4. In that frame is a spacelike unit vector sµ = (0. s) u = . The ¯ ¯ solutions are normalized to u(p. s) ei k·x d∗ (k. p (4.    (4. The explicit Dirac equation in the standard representation reads  ∂  iσ · ∇   i ∂t − M       (4. s) . the best representation to describe particles at rest is the standard representation. the spinors satisfy C uT (p.g. s) e−i k·x b(k. s)u(p. s) = v(p. with E = Ep = + p2 + M 2 .. s)¯(p.70) =  Ps =  ˆ 0 1∓σ·s 2 2 . ¯ ′ ′ Explicit solutions in the standard representation are   χs       . s ) = 0. v(p. ¯ (4. s′ ) = −2M δss′ .61)     σ·p −(Ep + M ) There are also two negative energy solutions.   ⇔ (/ − M ) u(p) = 0. s)u(p. Choosing χ = −ǫχ∗ (the equivalent spin ¯ 1/2 conjugate representation). s) = u(p. s′ ) = 2M δss′ .The Dirac equation 34 4.1).4 Plane wave solutions Looking for positive energy solutions ∝ exp(−iEt) one finds two solutions. s) = u(p.

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

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

77) Exercises Excercise 4. γ σ ] = σ ρσ . Show that the equality holds.2 Show that j µ = ψγ µ ψ is a conserved current if ψ en ψ are solutions of the Dirac equation Exercise 4.4 Show that in P+ ≡ s=±1/2 u(p. 1 i −i S ρσ = − [γ ρ . by proving that H 2 (p) = exp (η · σ) = (E + σ · p)/M = σ µ pµ /M.g. ν µ µ Look at the infinitesimal forms Λµ = gν + ων and L = 1 − (i/2)ωρσ S ρσ to obtain ν [γ µ . 4.5 Show starting from the relation {γµ . s) u p /+M = 2M 2M both sides are projection operators.3 Imposing Lorentz invariance on the Dirac equation can be achieved by requiring L−1 γ µ L = Λµ γ ν . 0 if n is odd. s)¯(p. where x = (t. of which only the traces of four and two gamma-matrices are nonzero. a 4 a · b. γν } = 2gµν that γµ / γ µ a = Tr(//) = ab µn µ1 Tr(γ · · · γ ) = −2 /. x → x. ˜ Exercise 4. x ˜ . (4. Do this e. 2M (E + M ) ˆ ˆ where E = M γ = M cosh(η) and p = M βγ n = M sinh(η) n. H(p) = exp η·σ 2 = M +E+σ·p . Exercise 4.13. −x) is also a solution of the Dirac equation. e. 4 2 Exercise 4.6 Apply space-inversion.1 Prove Eq.g.The Dirac equation 37 The trace is linear and can be split up in parts containing traces with up to four gamma-matrices. to the Dirac equation and use this to show that the spinor ψ p (x) = ˜ γ0 ψ(˜). s) and v(p. They lead to Lµν 1 Tr [(/ ′ + m)γ µ (/ + m)γ ν ] k k 2 = 2 k µ k ′ν + k ν k ′µ − g µν (k · k ′ − m2 ) = = 2 k µ k ′ν + 2 k ν k ′µ + (k − k ′ )2 g µν . s). Excercise 4. i S ρσ ] = g ρµ γ σ − g σµ γ ρ . by letting them act on the four basis spinors u(p.

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. r). (d) Plot for these cases the vector densities ψ † ψ and the scalar densities ψψ as a function of r/R. n (c) Apply the confinement condition to find the value of k for the lowest eigenmode in the spherical ˆ cavity. A condition for confining the fermion in the cavity is i / ψ = ψ (where n is the surface normal). .e. which is the most important factor needed to find the decay width (inverse lifetime). the lepton and neutrino have spin 1/2 and are ν described by momentum-space spinors.The Dirac equation 38 Exercise 4. The conditions for ψ and ψ imply that nµ j µ = ψ/ ψ = 0. |M (π − → µ− νµ )|2 ¯ Actually if the lepton would be massless.    is a stationary solution of the (free) Dirac equation. 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 ). i. We will not pursue this chiral freedom at this point and take α = 0. Excercise 4. Calculate for now just |M |2 .5 and M → ∞. Calculate the value of k for M R = 0. me = 0. M R = 1.8 A negatively charged pion (π − ) decays in a lepton ℓ (ℓ can be an electron or a muon) and a corresponding anti-neutrino (¯ℓ ). ν p Note: how such parametrizations work will be explained in chapter 8. 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. the decay would even be forbidden.8 × 10−6 . Note: the outcome shows that the decay into muons is much more likely that the decay into electrons (mπ = 139. t) = N     k ˆ i E+M σ · r j1 (kr)χm    −i Et   e .511 MeV). For such a cavity nµ = (0. Note: To prove this. n It turns out that there is actually a whole family of conditions that provide confinement. we want to confine this solution to a spherical cavity with radius R. there is no current flowing through the surface of the cavity.57 MeV. namely i / ψ = eiαγ5 ψ. (b) Next. Show that the ratio is |M (π − → e− νe )|2 ¯ ≈ 54.7 (a) Show that (in standard representation)  j0 (kr)χm    ψ(x. mµ = 105.66 MeV. you need to determine the corresponding confinement condition for ψ. Do you understand why? The calculation of the life-time is completed in chapter 11. Show that this n condition is sufficient to guarantee that nµ j µ = 0. The pion has spin 0.

± 5. µ (2π)3 2E µ (5. The vectorfield is therefore (λ) suited to describe spin 1.2) (5. In an arbitrary frame ǫµ (k) with λ = 0. ±(1) are obtained by boosting the restframe vectors. namely ǫµ (k0 ) = (0. just as the two-component Dirac spinors did for spin 1/2. The covariant generalization is to describe them with Vµ (x) = ǫµ (k) e±ik·x . In the restframe. These are solutions of the Klein-Gordon equation with in addition a condition.Chapter 5 Vector fields and Maxwell equations 5. one has ǫ(λ) (k) ǫ(λ)∗ (k) = −gµν + µ ν kµ kν . These can describe spin 1 states in the rest-frame. there are then indeed three independent possibilities. λ) .± (λ) d3 k ǫ(λ) (k) e−ik·x c(k. A). where k = k0 = (M. with ∂µ V µ = 0 (Lorentz condition).1 Fields for spin 1 In section 3.2 we have discussed the vectors ǫλ (with λ = 0 and ±1) as basis states of the spin 1 representation. A real vector field Vµ (x) then can be expanded in modes as Vµ (x) = λ=0. 0).2 The electromagnetic field F µν = = ∂ µ Aν − ∂ ν Aµ  1 −E 2  0 −E  1  E  0 −B 3   2  E 3  B 0   3 E −B 2 B 1 Aµ = (φ. λ) + ǫ(λ)∗ (k) eik·x c∗ (k.  −B 1    0 3 With the electromagnetic field tensor (5. (2 + M 2 )Vµ = 0.4) λ=0.3) Since the vectors ǫµ (k) together with the momentum k µ form a complete set of four-vectors.7) . M2 (5. 39 (5.6) where Aµ is the four-vector potential (5. (5. ǫλ ).5)  −E    B2   .1) The Lorentz condition implies that k µ ǫµ (k) = 0.

15 the d’Alembertian is replaced by 2 + M 2 . only two independent vectors remain ǫ(±) µ (k) = (0. ˜ x Note that via parity transformation one obtains another solution. (5. for M = 0. AP (x) = Aµ (˜). in which case one has 2Aµ = jµ .18) corresponding to the helicity states λ = ±1 of the massless photon field. ∂µ Aµ = 0. The equation in vacuum. Fµν −→ Fµν + (∂µ ∂ν χ − ∂ν ∂µ χ) = Fµν . One possible choice is the gauge A0 = 0 (5. shows that the electromagnetic fields correspond to the case of a massless spin 1 field. . (5.5. 0.e. Furthermore. It states the absence of magnetic charges and Faraday’s law. The equations of motion for the fields Aµ become 2Aµ − ∂µ (∂ν Aν ) = jµ . Taking the divergence of this equiation.11) This equation is not affected by a gauge transformation. the antisymmetry of F µν implies that ˜ ∂µ F µν = 0 or ∂λ Fµν + ∂µ Fνλ + ∂ν Fλµ = 0. 40 (5. The gauge freedom can first of all be used to impose the Lorentz condition for the electromagnetic field. while the fact µ C ∗ that the Minkowski space is real implies that Aµ (x) = Aµ (x) = Aµ (x). For electrodynamics one has furthermore the freedom of gauge transformations.8) (5. it will vanish at all times. (5.14) of which the solutions give the Li´nard-Wiechert potentials.15) (5. The Lorenz condition can actually be incorporated by changing the equations of motion to 2Aµ + (λ − 1)∂µ (∂ν Aν ) = jµ . if in Eq. As remarked above.16) (5.9) 1 ˜ where F µν ≡ − 2 ǫµνρσ Fρσ is the tensor where E → B and B → −E. Under a gauge transformation Aµ −→ Aµ + ∂µ χ. (5. Note that the incorporation of the Lorentz condition can also be done for the case of the massive vector field. Although the Lorenz condition is a constraint.17) (radiation. |k|). it does not eliminate the freedom of gauge transformations but they are now restricted to Aµ → Aµ + ∂µ χ with 2χ = 0. moree over. 0). 5. one finds for a conserved current and λ = 0 2∂µ Aµ = 0. j) the Maxwell equations read ∂µ F µν = j ν . 0. i. Therefore if k µ = (|k|.12) (5. if ∂µ Aµ = 0 for large times |t|. in accordance with what we found in section 3. (5. transverse or Coulomb gauge). In that case one also has ∇·A = 0 or ki ǫi (k) = 0. ∓1. there is still a gauge freedom left (Aµ → Aµ + ∂µ χ with 2χ = 0).Vector fields and Maxwell equations and the (conserved) current j µ = (ρ.10) the electric and magnetic fields are unchanged. −i.13) Hence. 2Aµ = 0.

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

Below the critical temperature the magnetic field is squeezed out of the superconducting material. organizing itself in tiny flux tubes (Abrikosov strings). the number of times they go around the defect) cannot be continuously deformed into one another. 5. is multivalued as ϕ = 0 and ϕ = 2π are the same point in space. The AharonovBohm experiment shows a realization of this possibility. If this would be be happening in empty space one would be in trouble. Therefore space outside the solenoid doesn’t care that χ is multi-valued.e. 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. minimizing the space occupied by B fields.2: The topology of the space in the Aharonov-Bohm experiment. By going around an arbitrary loop a different number of times the electron wave function would acquire different phases or it would acquire an arbitrary phase in a point by contracting a loop continuously into that point. nonobservable phases ϕ = 2π n. The difference here is that we are working in a space with a ’defect’ (the infinitely long solenoid). forming a simple (connected) space. implying that it must be single-valued1 . it can be obtained from the situation A = 0 by a gauge transformation.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.e. Another situation in which the topology of space can be used is in the case of superconductors. In such a space loops with different winding number (i. Consider a superconductor. A = ∇χ with χ= BR2 ϕ 2 (5.21. ϕ=2π The function χ in Eq. Now back to the Aharonov-Bohm experiment. The observable phase going around in general is ∆δ = e A · dr = e ∇χ · dr = e χ|ϕ=0 . i. however. 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. The only way for flux tubes to be formed and move around without global consequences is when each flux tube contains a flux Φ such that ∆δ = (e/ )Φ = n · 2π. 1 The occurrence of nontrivial possibilities. Therefore the gauge transformation must be uniquely defined in the space one is working in. giving no observable phase2 .21) (ϕ being the azimuthal angle). This situation that A can be written as ∇χ is always true when ∇ × A = 0.

1 Show that Eqs. 5.8 and 5.9 explicitly give the equations ∇ · E = ρ. ∂t ∇ · B = 0. .Vector fields and Maxwell equations 43 Exercises Exercise 5. ∇×B =j+ ∇×E+ ∂E . ∂t ∂B = 0.

x) ˙ (6. Dirac fields (Dirac equation) and massless vector fields (Maxwell equations) and corresponding to these fields conserved currents describing the probability and probability current. thus t′ = t + δτ. and the total change x′ (t′ ) = x(t) + ∆x(t).Chapter 6 Classical lagrangian field theory 6.8) 44 . ∂x (6.3) (6. These equations can be obtained from a lagrangian using the action principle. x (t) = x(t) + δx(t).1) and as an example the lagrangian L(x. The requirement δS = 0 with fixed boundaries x(t1 ) = x1 and x(t2 ) = ˙ x2 leads to t2 ′ (6. x) = K − V = ˙ 1 mx2 − V (x).7) The quantity ∂L/∂ x plays a special role and is known as the canonical momentum. (6.4) δS = t1 t2 dt dt t1 ∂L ∂L δx + δ x + L δτ ˙ ∂x ∂x ˙ d ∂L − ∂x dt ∂L ∂x ˙ δx + t2 t1 = ∂L δx + L δτ ∂x ˙ t2 .1 Euler-Lagrange equations In the previous chapter we have seen the equations for scalar fields (Klein-Gordon equation). ˙ p= ∂L .2) The principle of minimal action looks for a stationary action under variations in the coordinates and time. As an example. recall classical mechanics with the action t2 S= t1 dt L(x. ∂x ˙ (6. t1 (6. ˙ 2 (6.5) with ∆x(t) = δx(t) + x(t) δτ .6) The first term leads to the Euler-Lagrange equations. d dt ∂L ∂x ˙ = ∂L .

∂µ φ(x)).9) which is done because the first term (multiplying ∆x. 6. . ∂µ φ. with ∆φ(x) = δφ(x) + (∂µ φ)δx . L (φ(x). (6. The resulting variation of the action is δS = R µ ′ = xµ + δxµ .14) (6. indicated as x′µ φ (x) or combined φ′ (x′ ) = φ(x) + ∆φ(x). = φ(x) + δφ(x) (6.15) d4 x′ L (φ′ . . In general any continuous transformation. .13) (6. We will come back to it in a bit more formal way in section 9. Variations in the action can come from the coordinates or the fields. x′ ) − d4 x L (φ. also indicated by ∂R (a more general volume in four-dimensional space-time with some boundary ∂R can also be considered). t1 (6. ˙ x The second term in Eq. . Consider a lagrangian density L which depends on these functions. . . their derivatives and possibly on the position. δ(∂µ φ) (6.17) Furthermore the variations δφ and δ∂µ φ contribute to δS.Classical lagrangian field theory 45 For the lagrangian specified above this leads directly to Newton’s law p = m¨ = −∂V /∂x. ∂R Note for the specific choice of the surface for constant times t1 and t2 . but which in 3 dimensions includes conserved quantities related to rotations and boosts. t2 ). − d3 x .t2 ) d3 x . dσµ . + ∂L ∆x − H δτ ∂x ˙ t2 . t1 ) and R3 . H is a conserved quantity. which in classical mechanics vanishes at the boundary) does not play a role. i. = ∂R (R3 . ∆x = (dx/dλ)λ=0 δλ and δt = (dt/dλ)λ=0 δλ. x) ≡ p x − L . 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 . ∂µ φ(x). think for instance of a temperature or density distribution or of an electromagnetic field). λ=0 (6.11) λ=0 of which p and H are the simplest examples related to space and time translations.16) The change in variables x → x′ in the integration volume involves a surface variation of the form dσµ L δxµ . . x) and an action t2 S[φ] = t1 dt L = dt d3 x L (φ(x). x). .18) 1 Taking a functional derivative. R (6.e.6 can be rewritten as δS = . ∂µ φ′ . gives rise to a conserved quantity (at least if H is invariant!) Q=p dx dλ −H dt dλ .t1 ) (6.12) Here R indicates a space-time volume bounded by (R3 . In classical field theory one proceeds in complete analogy but using functions depending on space and time (classical fields. ˙ (6. .10) One sees that invariance under time translations requires that H(t1 ) = H(t2 ). ∂µ φ(x)) = R d4 x L (φ(x).2. (R3 . The hamiltonian H is defined by H(p. indicated with δF [φ]/δφ should pose no problems. .

δL δ(∂µ ψ) δL δψ one obtains immediately i / −M ∂ → i = − γ µψ 2 i → = ∂ / ψ − M ψ. 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).Classical lagrangian field theory 46 With for the situation of classical fields all variations of the fields and coordinates at the surface vanishing. L = 1 1 ∂µ φ∂ µ φ − M 2 φ2 2 2 1 µ = − φ ∂µ ∂ + M 2 φ. leading to the Euler-Lagrange equations. The scalar field It is straightforward to derive the equations of motion for a real scalar field φ from the lagrangian densities. 2 ψ = 0. 2 ∗ (6.27) (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. 2 (6.24) (6. Using the variations in ψ (in the symmetric form). the second term is irrelevant.25) (6. (6.2 Lagrangians for spin 0.28) where the second line is not symmetric but in the action only differs from the symmetric version by a surface term (partial integration). the Dirac field (spin 1/2) and the vector field (spin 1) can be found.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. leading to (2 + M 2 )φ(x) = 0. The integrand of the first term must vanish.23) (6.29) . ∂ (6. For the complex scalar field one conventionally uses L = = ∂µ φ∗ ∂ µ φ − M 2 φ∗ φ −φ∗ ∂µ ∂ µ + M 2 φ. (6. (6.21) which differ only by surface terms. One easily obtains (2 + M 2 )φ(x) = 0. δL δL ∂µ = .20) (6. (2 + M )φ (x) = 0.19) δ(∂µ φ) δφ 6.

31) showing e. .36) (6.    ∗  0    ⇒ Υ ≡  −ǫ η  ΥL = ψL =    η η ∗ for which η0 = ǫ η0 .34)  (6. ∂ ∂ 2 2 (6. since a lagrangian with both Dirac and Majorana mass terms can be rewritten as the sum of two Majorana lagrangians after redefining the fields (See e. (6.31 is given by LM (Dirac) = −M ξ † η + η † ξ .32) There exists another possibility to write down a mass term with only one kind of fields. 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 . (6. 2 αβ = −βα.33) We now introduce a ’real’ (four-component) spinor satisfying Υc = Υ with ψL . Υi / Υ − ∂ 4 2   T  −ǫ η ∗  c  = ΥR . namely2 LM (Majorana) = + (6.7).g.and lefthanded ones.30) It is often useful to link to the two-component spinors ξ and η which we started with in chapter 4. and thus (β ∗ α)∗ = α∗ β = −βα∗ . In that case one finds trivially L = ↔ ↔ 1 1 ψR i / ψR + ψL i / ψL − M (ψR ψL + ψL ψR ).Classical lagrangian field theory 47 and similarly from the variation with respect to ψ ψ i / +M ∂ ← = 0. 1 M η † ǫη ∗ − M ∗ η T ǫη .g. that the lagrangian separates into two independent parts for M = 0. The reasons for Grassmann variables will become clear in the next chapter. 2 (6. but note the factor 1/2 as compared to the Dirac lagrangian. the mass term in the Dirac lagrangian 6. ψL ≡ (ψL )c = C ψL =    0 (6. Peshkin and Schroeder or Exercise 12. L = = ↔ 1 1 c c M ψL ψL + M ∗ ψL ψL ψL i / ψL − ∂ 2 2 ↔ 1 1 M ΥR ΥL + M ∗ ΥL ΥR .37) This gives an expression with a mass term that is actually of the same form as the Dirac mass term.39) 2 This term is written down with η being an anticommuting Grassmann number for which (αβ)∗ = β ∗ α∗ = −α∗ β ∗ . (6. We note that of which the left part coincides   .35) Since the kinetic term in L separates naturally in left and right parts (or ξ and η). Using the twospinors ξ and η. The Majorana case is in fact more general. which comes because we in essence use ’real’ spinors. 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.38) (6.

δ(∂µ φ) For continuous transformations. ∂µ J µ (x) = 0. 6. (6. these conserved quantities become operators.44) δL ∆φ + Θµν (x)δxν .+ R 4 δL δL ∆φ − ∂ ν φ − L g µν δxν δ(∂µ φ) δ(∂µ φ) δL ∆φ − Θµν (x)δxν . the presence of a symmetry that leaves the lagrangian invariant.41) (6. (6. . Q(t1 ) = Q(t2 ). These do not affect the dynamics and will give rise to conserved currents. .3 Symmetries and conserved (Noether) currents Next. As an example for the classical field. (6. ψ(x) → ei eΛ ψ(x) or ∆ψ(x) = i eΛ ψ(x). + ∂R dσµ dσµ ∂R ≡ where d x .Classical lagrangian field theory 48 where Fµν = ∂µ Vν − ∂ν Vµ . t).40) 6. 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 ) = 0.46) 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 . consider U (1) transformations of the Dirac field proportional to the charge e. requires the presence of a conserved quantity. Returning to δS the surface term is rewritten to δS = R d4 x .45) J µ (x) = − δ(∂µ φ) dλ λ=0 dλ λ=0 Thus considering ∂R = σ2 . we get (omitting the parameter δλ) dxν dφ δL + Θµν (x) . we consider the second term in Eq. In particular when the lagrangian is invariant under symmetries. ∆φ = (dφ/dλ)λ=0 δλ and δxµ = (dxµ /dλ)λ=0 δλ.. δ(∂µ φ) (6.σ1 with σ = (R3 . which is the space-integral over the zero-component of a conserved current. δ(∂µ φ) (6. which in a consistent picture precisely generate the symmetries. t)..47) . In this case. Q= d3 x J 0 (x. implying ∂µ F µν + M 2 V ν = 0 and ∂µ V µ = 0.42) Θµν = The variation δS thus can be expressed as δS with J µ (x) = − = δL ∂ ν φ − L g µν .18.43) F (σ1 ) − F (σ2 ) = dσµ J µ (x) = ∂R R d4 x ∂µ J µ (x). (6. (6. discussed in the next chapter. one can consider the variations at the initial or final surface.

g.51) (6.49) These currents are conserved as discussed already in chapter 2. ↔ (6. Therefore there are four conserved currents Θµν (ν labeling the currents!) and four conserved quantities Pν = dσµ Θµν = d3 x Θ0ν . ˙ Q = d3 x j 0 (x. = 0. 2 (6.53) These are the energy and momentum.4 Space-time symmetries One kind of symmetries that leave the lagrangian invariant are the Poincar´ transformations. Summarizing. (6. (6. 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.3) j µ = i e φ∗ ∂µ φ. The integral over the zero-component. (6. we obtain (since δxµ = 0). including e the space-time translations and the Lorentz transformations.56) δxµ = ωρσ (aρσ )µν xν 2 with (aρσ )µν = aρσ µν = g ρµ g σν − g σµ g ρν . (ω µν antisymmetric) 1 ωρσ (−i S ρσ )ij φj (x). = −(∂µ φ)δxµ = −ǫµ ∂µ φ(x). Translations µ Under translations (generated by Pop ) we have δxµ ∆φ(x) δφ(x) = ǫµ . (6.57) For the fields the behavior under Lorentz transformations has been the subject of previous chapters. 1 (6. From Noether’s theorem one sees that the current Θµν ǫν is conserved. Lorentz transformations In this case the transformation of the coordinates and fields are written as δxµ ∆φi (x) = = ω µν xν .Classical lagrangian field theory From the lagrangian for the Dirac field. t) is the conserved charge (Q = 0). 6. In the next section we will see the quantity show up as the charge operator. For quantized fields this will become the expressions of the hamiltonian and the momentum operators in terms of the fields. .55) Note that the coordinate transformations can be written in a form similar to that for the fields (matrices in internal space). H = d3 x Θ00 (x).48) For the complex scalar field the U (1) transformations leave the lagrangian invariant and lead to the conserved current (see Exercise 6.54) (6. e.52) The behavior of the field under translations (δφ) is governed by the translational behavior of the argument in such a way that ∆φ = 0.50) (6.

∂µ φ(x) + i e ∂µ Λ(x) e (6. in which the U (1) transformation involves an angle e Λ. • Vector field Aµ : −i Sρσ = aρσ .63) µν = ∂µ Θ .3). In some applications (specifically coupling to gravity) it is advantageous to have an equivalent current that is symmetric.66) In section 6.60) Therefore there are six conserved currents M µρσ labeled by ρ and σ (antisymmetric) and corresponding to it there exist conserved quantities M ρσ = d3 x M 0ρσ .68) (6.5 (Abelian) gauge theories φ(x) → ei eΛ φ(x).e. the angle of the transformation depends on the space-time point x. where H ρµν is the quantity appearing in M ρµν . γσ ] This result for the Dirac field has been discussed in chapter 3 (see also Exercise 4. (6. ∂µ T µν (6. −i eΛ(x) ∗ i eΛ(x) (6. is not invariant under local gauge transformations. φ(x) φ (x) ∂µ φ(x) ∗ → e → e → ei eΛ(x) φ(x).6) T µν = T νµ . In general this tensor is not symmetric.70) . one has a tensor that satisfies (exercise 6.64) (6. • Dirac field ψ: −i Sρσ = −(1/4)[γρ .58) (6. φ(x). The lagrangians which we have considered sofar are invariant under global gauge transformations and corresponding to this there exist a conserved Noether current.g. independent of x.67) i. 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 .69) i eΛ(x) φ (x). (6. (6. however. Any lagrangian containing derivatives. Defining T µν = Θµν − ∂ρ Gρµν .61) A final note concernes the symmetry properties of Θµν .62) with Gρµν = (H ρµν + H µνρ + H νµρ )/2. (6.Classical lagrangian field theory 50 • Scalar field φ: −i Sρσ = 0.59) (6. 2 (6. e. Gauge transformations of the second kind or local gauge transformations are transformations of the type φ(x) → ei eΛ(x) φ(x). 6.65) M µρσ = T µρ xσ − T µσ xρ . (6.3 we have seen global transformations or gauge transformations of the first kind.

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 .75) The field φ is used here in a general sense standing for any possible field. −e j µ Aµ = −e ρφ + e j · A. ∂ 2 4 (6.74) 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 equation of motion for the fermion follow from δL δ(∂µ ψ) δL δψ = = i − γ µ ψ. 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.Classical lagrangian field theory 51 where the last term spoils gauge invariance. 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). As an example consider the Dirac lagrangian. 1 L (φ.78) . (6. 4 (6. Dµ φ(x) ≡ (∂µ + i eAµ (x))φ(x). 2 i ∂ / ψ − M ψ − e/ ψ A 2 giving the Dirac equation in an electromagnetic field. (6.72) (6. (i/ − M ) ψ = (i/ − e/ − M ) ψ = 0. i L = ψγ µ ∂µ ψ − (∂µ ψ)γ µ ψ − M ψψ. (6. ∂µ φ) =⇒ L (φ. For this purpose it is necessary to introduce a vector field Aµ .73) (6. Replacing derivatives by covariant derivatives and adding the (free) part for the massless vector fields to the original lagrangian therefore produces a gauge invariant lagrangian. D ∂ A (6.77) 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).71) Thus the covariant derivative has the correct transformation behavior provided Aµ → Aµ − ∂µ Λ.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 . Dµ φ) − Fµν F µν .

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

A) (see also exercise 6. t). . in order to show that Θρσ − Θσρ = ∂µ H µρσ . It shows that for a spin 1/2 particle the g-factor in the coupling of spin to a magnetic field takes the value gs = 2.4) 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 .5 Show that the Dirac equation for an electron with charge −e in an external electromagnetic field Aµ = (A0 . 2me Exercise 6.Classical lagrangian field theory 53 Exercise 6. Hint ≡ −gs µe · B = −gs Qe s · B. use ∂µ T µν = 0 and ∂µ M µρσ = 0.6 (optional) Prove the properties of the tensor T µν in section 6. t) = Enr ψu (r. 2M 2M with Enr ≡ E − M ≪ M . Hints: it may be useful to realize that Gµρσ is antisymmetric in first two indices. This relation is known as the Pauli equation.4. finally note that it is sufficient to show the last equation for dσµ M µρσ . Comment: this relation is known as the Pauli equation. e 1 (−i∇ + eA)2 + σ · B − eA0 ψu (r.

q) also considered in the previous chapter. This is nothing else as representing the classical Poisson algebra as operators in the Hilbert space.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.1) The Hamiltonian (also corresponding to a conserved quantity because of time translation invariance) is given by p2 + V (q). (7.3) For variations δO(p. H]P dt and dO = [O. q) ≈ (dO/dλ)λ=0 δλ under symmetry transformations one has dO = [O. Actually the conserved quantities satisfy a similar algebra as the Lie algebras for continuous symmetries that they are obtained from. 6. namely [q. (7. q) = p q − L(q. in our example p = mq. however. Quantization of the theory is achieved by imposing canonical commutation relations between q and p. labeled by the position. p] = i. dq The basic Poisson bracket is actually the one between q and its canonical momentum p. An immediate generalization for fields can be obtained by considering them as coordinates. 1 qx (t) = 3 d3 x φ(x. q(q. q) = ˙ 1 mq 2 − V (q). (7. there are conserved ˙ ˙ quantities Q (Eq. t). B]P = dB dA dA dB − .11) found through Noether’s theorem. starting from the lagrangian L(q. p]P = 1. ˙ 2 (7.Chapter 7 Quantization of fields 7. Q]P . p)) = ˙ ˙ 2m where the (canonical) momentum p = ∂L/∂ q. The bilinear product for the conserved quantities.1 Canonical quantization We will first recall the example of classical mechanics for one coordinate q(t). [q.5) ∆ x ∆3 x 54 .2) H(p. p]P . Further. is the Poisson bracket [A. dλ Examples are dO = [O. dq dp dq dp (7. ˙ L(q.

Quantization of fields 55 etc.7) d3 x L (φ(x). ∂µ φ(x)) = ˙ d3 x L (φ. qx+∆x ) . t). a) φi (x)U (Λ. φ. The hamiltonian then is H= x px qx − L = ˙ = x where H (x) = δL ˙ φ(x) − L (x) = Θ00 (x). for symmetry transformations. py (t)] = i δxy or for the fields φ(x) and Π(x) the socalled equal time commutation relations [φ(x. t).6) ∆3 x L x (qx . the discretization procedure above is an explicit example.2. (7. • Canonical commutation relations The quantization of the theory is achieved by imposing the canonical quantization condition [qx (t). albeit with ’sharp’ functions.10) x ∆3 x [Πx qx − L x ] ˙ ∆3 x Hx (7. Π(y. formally.11) (7. for the scalar field theory. δ∂0 φ) 1 1 1 = Θ00 (x) = (∂0 φ)2 + (∇φ)2 + M 2 φ2 . t)] = i δ 3 (x − y). a) = Rj (Λ−1 ) φj (Λ−1 x − a). discussed in the section 6. the fields φ(x) can be considered as regular operators in the Hilbert space after smearing with a test function f . qx .15) with furthermore the relations [φ(x. This implies specific transformation properties in the Hilbert space for the quantum fields φ. 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. ∇φ) (7. ˙ δ φ(x) (7. t)] = [Π(x. ∂ qx ˙ ∂ qx ˙ (7. Sometimes it is useful to keep in mind that.16) .13) Π(x) H (x) For the quantization procedure we can formulate a number of basic axioms of quantum field theory.9) Note that this indeed corresponds to the zero-zero component (Θ00 ) of the conserved energy-momentum stress tensor Θµν . (7.14) In fact.12) (7. (7. φ(y. As an example. ∆φ. i U † (Λ. The lagrangian is given by L = = x In order to construct the hamiltonian it is necessary to find the canonical momenta. Φ(f ) = d4 x φ(x) f (x). Π(y. px (t) = ∂L ∂Lx = ∆3 x = ∆3 x Πx (t). 2 2 2 (7. t). t)] = 0.8) (7. The essential conditions to consistently quantize a theory then are the following. • Poincar´ invariance e The full variation of the fields. ˙ where Πx (t) is obtained from the continuous field Π(x) ≡ δL /δ(∂0 φ).

4. j 2 one has [φi (x). [N.18) with S µν given in 6. 2 i ωµν (S µν )i . (7. φ(y)] = 0 if (x − y)2 < 0. P ] = i. a] = [a† . H = = ω ω P P Q−i √ Q+i√ 2 2 2ω 2ω 1 1 =ω N+ . a† ] = a† . 2 2 (7.2 Creation and annihilation operators Before discussing (real and complex) scalar fields and Dirac fields we recall the analogy with the wellknown harmonic oscillator as an example of quantization using creation and annihilation operators. j (7. a] = −a.17) (Sµν )i φj (x). Q] = [P. ω a† a + 2 2 ω i − [Q. a† ] = 0. The measurements cannot influence each other or [Φ(f ).22) it is straightforward to check that the commutation relations between Q and P are equivalent with the commutation relations [a. 1 Q = √ (a + a† ) 2ω and P = −i ω (a − a† ) 2 (7. (7. In simplified form the hamiltonian is given by H= 1 2 1 2 2 P + ω Q . [Q. and [N. [φ(x). • Causality Operators Φ(f ) and Φ(g) for which the test functions are space-like separated can be measured simultaneously (macroscopic causality). P ] 2 (7. Φ(g)] = 0. i(xµ ∂ν − xν ∂µ )φ (x) + i (7. Microscopic causality implies local commutativity. Mµν ] = i i∂µ φi (x).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† . P ] = 0 (7.21) Writing Q and P in terms of creation (a† ) and annihilation (a) operators. a) = i Rj (Λ−1 ) = 1 + iǫµ P µ − 1− i ωµν M µν . [a.19) 7.23) The hamiltonian in this case can be expressed in the number operator N = a† a.Quantization of fields 56 or if U (Λ. 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. sometimes referred to as second quantization. Pµ ] = [φ (x). (7.25) . a† ] = 1.

and we see that a state |0 must exist for which N |0 = a|0 = 0. From the normalizations one obtains a |n = √ n + 1 |n + 1 and a|n = n |n − 1 . (2π)3 2Ek (7. [φ(x). Π(x )]x0 =x′0 = 0. † √ a and a act as raising and lowering operators.e. a† (k ′ )] = (2π)3 2Ek δ 3 (k − k′ ) [a(k).27) 2 (2π)3 It is easy to check that these equations can be inverted (see Exercise 2. i. is equivalent with [a(k). a† (k ′ )] = 0.4 for the classical field) a(k) a† (k ′ ) = = d3 x ei k·x i ∂0 φ(x). e. H = = = d3 x 1 1 1 (∂0 φ)2 + (∇φ)2 + M 2 φ2 2 2 2 d3 k Ek † a (k)a(k) + a(k)a† (k) (2π)3 2Ek 2 d3 k Ek N (k) + Evac . Note that we will often write a(k) or a† (k). d3 x φ(x) i ∂0 e−i k ·x ↔ ′ ↔ (7. the real scalar field φ(x) becomes d3 k (7. φ(x) = (2π)3 2Ek where the Fourier coefficients a(k) and a† (k) are now operators.like’ commutation relations between a(k) and a† (k ′ ). φ(x )]x0 =x′0 = [Π(x). N |n = n|n one finds N a† |n N a|n = = (n + 1) a† |n .32) (7. The quantization of the field is achieved by quantizing the coefficients in the Fourier expansion.g. (n − 1) a|n † i.e. The canonical momentum becomes −i d3 k ˙ Π(x) = φ(x) = a(k) e−i k·x − a† (k) ei k·x .29) It is straightforward to prove that the equal time commutation relations between φ(x) and Π(x′ ) are equivalent with ’harmonic oscillator . 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.30) [φ(x).33) . which represents the ’number of particles’ with momentum k. (7. a(k ′ )] = [a† (k). but one needs to realize that in that case k 0 = Ek = k2 + M 2 . 7. and (7.26) a(k) e−i k·x + a† (k) ei k·x .Quantization of fields 57 Defining states |n as eigenstates of N with eigenvalue n.3 The real scalar field We have expanded the (classical) field in plane wave solutions. which we have split into positive and negative energy pieces with (complex) coefficients a(k) and a∗ (k) multiplying them.28) (7. Π(x′ )]x0 =x′0 = i δ 3 (x − x′ ) ′ ′ and (7.31) The hamiltonian can be rewritten in terms of a number operator N (k) = N (k) = a† (k)a(k).

35) where the vacuum contribution disappears because of rotational symmetry. 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. (2π)3 2Ek (7.44) p|φ− (x)|0 = ei p·x . d3 k |k k|. . is solved by subtracting it as an (infinite) constant. |k = a† (k)|0 . and multiparticle states |(k1 )n1 (k2 )n2 . 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 (7.38) d4 k 2π δ(k 2 − M 2 ) θ(k 0 ) |k k| = (2π)4 (7. φ(x) φ+ (x) φ− (x) = = = φ+ (x) + φ− (x).46) (7. in field theory also referred to as vacuum energy Evac = 1 V 2 d3 k Ek . defining particle states |k = |k (with positive energy. For the purpose of normal ordering it is convenient to decompose the field in positive and negative frequency parts. |0 .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. assuring that the vacuum (by definition) has eigenvalue 0!. (2π)3 2Ek d3 k a† (k) ei k·x .45) (7.40) This procedure is known as normal ordering. = normalized as (a† (k1 ))n1 (a† (k2 ))n2 √ √ . This term will be adressed below. (2π)3 2Ek (7.e. (2π)3 2Ek (7. .42) (7. .36) and satisfying the completeness condition 1 = k|k ′ = (2π)3 2Ek δ 3 (k − k′ ). 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). (2π)3 2Ek (7. n1 ! n2 ! (7. a(k)|0 = 0. k 0 = Ek = k2 + M 2 ). i.41) (7.39) The problem with the vacuum or zero-point energy.34) where V = (2π)3 δ 3 (0) is the space-volume. writing all annihilation operators to the right of the creation operators. . (2π)3 (7. which amounts to redefining H as H= d3 x : H (x) := d3 k Ek N (k). d3 k a(k) e−i k·x .37) (7. The rest of the states are then obtained from the groundstate via the creation operator.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). .

47) ′ d3 k ′ e−i k·x+i k ·y [a(k).e.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.53) t=0 . (iii) ∆(0.49) (7. (iv) The equal time commutation relations follow from ∂ ∆(x) ∂t (v) For M = 0.56) = −δ 3 (x). The last item to be checked for the scalar field are the causality condition. φ+ (y)] d3 k e−i k·(x−y) .50) or as integrals over d4 k.52) The result for the invariant commutator function is [φ(x). a† (k ′ )] 3 2E ′ (2π) k [φ− (x). In order to calculate [φ(x). (ii) ∆(x) is a solution of the homogeneous Klein-Gordon equation. (2π)3 2Ek ≡ i∆− (x − y). i.48) (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 (7.54) = i∆(x − y) = i (∆+ (x − y) + ∆− (x − y)) . (7. i∆+ (x) i∆− (x) = = = d4 k θ(k 0 ) 2π δ(k 2 − M 2 ) e−i k·x . φ(y)] which has the properties (i) i∆(x) = i∆+ (x) + i∆− (x) can be expressed as i∆(x) = where ǫ(k 0 ) = θ(k 0 ) − θ(−k 0 ). ∆(x) = − ǫ(x0 ) δ(x2 ). (2π)4 (7.55) d4 k ǫ(k 0 ) 2π δ(k 2 − M 2 ) e−i k·x . x) = 0 and hence ∆(x) = 0 for x2 < 0. 2π (7. = − (2π)3 2Ek (7.18).17) and (7. φ(y)] consider µν [φ+ (x). (7. d3 k = (2π)3 2Ek = (7.51) (7. that they satisfy the required commutation relations in Eqs (7. d3 k ei k·(x−y) = −i∆+ (y − x).

particles (created by a† ) and antiparticles (created by b† ) with the same momentum contribute equally to the energy.61) (7.64) The commutator of φ and φ† is as for the real field given by [φ(x). (2π)3 2Ek (7.65) 7. b† (k ′ )] = (2π)3 2Ek δ 3 (k − k′ ). ∂0 φ† (y)]x0 =y0 = i δ 3 (x − y).62) (7.e.59) (7. 2 From the lagrangian density (7. (2π)3 2Ek d3 k b(k) e−i k·x + a† (k) ei k·x . (2π)3 2Ek (7. ↔ (7. φ† (y)] = i∆(x − y).66) . (7.63) i. (2π)3 2Ek (7.4 The complex scalar field In spite of the similarity with the case of the real field.Quantization of fields 60 7. = ∂{µ φ∗ ∂ν} φ − L gµν .57) (7. we will consider it as a repetition of the quantization procedure.5 The Dirac field L = ↔ i ψγ µ ∂µ ψ − M ψψ. which is equivalent to the relations (only nonzero ones) [a(k).60) = i φ∗ ∂µ φ.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. a† (k ′ )] = [b(k). Also the charge operator requires normal ordering (in order to give the vacuum eigenvalue zero). 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 . 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) . 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) .

(2π)3 2Ek (7. (7. s).69) (7. {b(k. The quantized fields are written ψ(x) = s d3 k b(k.75) (7.78) (7. s)¯(k. s) ei k·(x−y) x0 =y 0 . (2π)3 2Ek d3 k b† (k. s)ei k·x .71) (7. interchanging creation and annihilation operators. s)vj (k.74) (7. s)v(k. ˙ δ ψ(x) ↔ i Θ00 (x) = − ψγ i ∂i ψ + M ψψ 2 = iψγ 0 ∂0 ψ = iψ † ∂0 ψ. s)e−i k·x . s) e−i k·(x−y) j s † + vi (k. s).67) i ψ / ψ − M ψψ gµν . ↔ i ψγµ ∂ν ψ − 2 (7.Quantization of fields 61 the conserved density and energy-momentum currents are easily obtained. jµ Θµν = = ψγµ ψ.72) ψ(x) = s In terms of the operators for the b and d quanta the hamiltonian and charge operators are (omitting mostly the spin summation in the rest of this section) H = = Q = = d3 x : ψ † (x) i∂0 ψ(x) : d3 k Ek : b† (k)b(k) − d(k)d† (k) : (2π)3 2Ek d3 x : ψ † ψ : d3 k : b† (k)b(k) + d(k)d† (k) : . s)u† (k. s)u(k. then leads to additional minus signs and H Q = = d3 k Ek b† (k)b(k) + d† (k)d(k) (2π)3 2Ek d3 k b† (k)b(k) − d† (k)d(k) .77) Note that achieving normal ordering. i. s)e−i k·x + d† (k. (7. b† (k ′ . u v (2π)3 2Ek (7.70) where the last line is obtained by using the Dirac equation.73) (7. ∂ 2 ↔ (7. s′ )} = (2π)3 2Ek δ 3 (k − k′ ) δss′ . ψj (y)}x0 =y0 = d3 k (2π)3 2Ek ui (k. (2π)3 2Ek (7. d† (k ′ .e. s)ei k·x + d(k. s′ )} = {d(k. † {ψi (x). s)¯(k.79) Also for the field and the canonical conjugate momentum anticommutation relations are considered.68) The canonical momentum and the hamiltonian are given by Π(x) H (x) = = δL = i ψ † (x).76) which seems to cause problems as the antiparticles (d-quanta) contribute negatively to the energy and the charges of particles (b-quanta) and antiparticles (d-quanta) are the same. The solution is the introduction of anticommutation relations. .

4 Π0 Πi δL = 0. 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). (7.82) where i∆1 = i∆+ −i∆− .80) 3 = δ 3 (x − y) δij .81) where i∆ = i∆+ + i∆− is the same invariant commutator function as encountered before. but we must restrict ourselves to the weaker condition B|∂µ Aµ |A = 0. −λ(∂ρ Aρ ). 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. When we would have started with commutation relations for the field ψ and the canonical momentum. (7. ˙ δ A0 δL = F i0 = E i .6 The electromagnetic field 1 L = − Fµν F µν . 7.83) From the lagrangian density the canonical momenta are = = (7. 4 2 This gives the equations of motion discussed before. For the scalar combination {ψi (x). one has † {ψi (x). ∂ (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. 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) .Quantization of fields 62 Using the positive and negative energy projection operators discussed in section 4. as the vanishing of Π0 induces a constraint. ∂ (7. ψ j (y)} at arbitrary times one has {ψi (x). but has the problem of being noncovariant. ψ j (y)] = (i/x + M )ij i∆1 (x − y). ˙ δ Ai (7.85) which reflects the gauge freedom. it implies the Lorentz constraint and leads to the canonical momenta Π0 Π i = = E i. we would have obtained [ψi (x).86) (7.84) (7.

k 3 ) this reads |k 3 |c(0) (k) − k 3 c(3) (k) |A = 0 c(0) (k) ∓ c(3) (k) |A = 0. 0)c(k. The quantized field is expanded as Aµ (x) = d3 k (2π)3 2Ek (λ) 3 ǫ(λ) (k) c(k. The canonical equal time commutation relations are [Aµ (x).90) where Πν = F ν0 − λ g ν0 (∂ρ Aρ ) and we have furthermore [Aµ (x). c† (k ′ . In fact the commutation relations imply ˙ [Aµ (x). i. µ λ=0 (7. and are equivalent with [c(k.89) with four independent vectors ǫµ . pj ]P .3) satisfy the conditions of a Lie algebra. 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. pj ]P . It gives 3 k µ ǫ(λ) (k)c(k. (b) Proof for a quantity O(p. µ λ=0 (7. which reproduce the Hamilton equations. Aν (y)]x0 =y0 = i gµν δ 3 (x − y). λ). one time-like photon by itself is not allowed! This solves the problems with the normalization and the negative energies. j = 1.91) (7. (d) Use Poisson brackets to calculate q and p. rj ]P . 6. where Q is the conserved quantity in Eq. λ)ei k·x . ′ (7. q) that dO/dλ = [O. λ)|A = 0.e. 0. what are [ri .11. where Aµ + + is the part of the vector field containing the annihilation operators. λ)e−i k·x + 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. (7. Πν (y)] = 0. 7.94) Choosing k µ = (|k 3 |. a longitudinal photon and two transverse photons. [ℓi .1 (optional) If you haven’t seen Poisson brackets in classical mechanics you should (a) Proof that Poisson brackets (Eq. λ)c(k. (7. Q]P . (c) Calculate the Poisson brackets for the coordinates x and the standard conserved quantities p and ℓ.Quantization of fields 63 for physical states |A and |B . λ′ )] = −g λλ 2Ek (2π)3 δ 3 (k − k′ ). containing a time-like photon.93) which does exhibit problems with the time-like photon. 0. Πν (y)]x0 =y0 = i gµν δ 3 (x − y). 0) . and [ℓi . 2. for which it is sufficient that ∂µ Aµ |A = 0. ˙ ˙ . Aν (y)] = [Πµ (x). These problems are solved by the Lorentz constraint between physical states given above. λ) − c† (k. (7.e.95) Exercises Exercise 7. i.

(Similarly one can show that for the advanced Green’s function ∆A (x) = 0 if x0 > 0). (e) Show that ∆R (x) = 0 if x0 < 0. Is the subtraction of ’zero point’ contributions essential in this check? µ Pµ φ(x) ei a µ Pµ = φ(x − a) Exercise 7.4 (Green’s functions) (a) Show that ∆hom (x) = 1 (2π)4 d4 k e−ik·x δ(k 2 − M 2 )f (k) is a solution of the homogeneous Klein-Gordon equation. (2 + M 2 ) ∆hom (x) = 0. e−i a requires [φ(x). (2 + M 2 ) ∆inhom (x) = −δ 4 (x). we can write for ∆R an integral where the k 0 integration remains along the real axis.3 (a) Show that the correct implimentation of symmetries in the Hilbert space. (b) Show that ∆inhom (x) = 1 (2π)4 d4 k e−ikx 1 k2 − M 2 is a solution of the inhomogeneous Klein-Gordon equation. Pµ ] = i∂µ φ(x). Pµ ] for the (real) scalar field using the expressions for the fields and momentum operator in terms of creation and annihilation operators. ∆hom (Λx) = ∆hom (x) if f is an invariant function. It is invariant. (b) Check the above commutation relation [φ(x). ∆R = 1 (2π)4 d4 k e−ikx k2 1 − M 2 + iǫk 0 (with the implicit prescription to take ǫ → 0 after integration). Give also the expression for ∆C .Quantization of fields 64 Exercise 7. .2 Prove that the equal time commutation relations between quantum fields φ(x) and Π(x′ ) are equivalent with the commutation relations between a(k) and a† (k ′ ) Exercise 7. f (Λk) = f (k) for Lorentz transformations Λ. (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.

or ǫ(k 0 ) = θ(k 0 ) − θ(−k 0 ). In that case one actually needs discontinuous functions f (k) such as the step function.Quantization of fields 65 (f) By performing the k0 integration over any closed path C (encircling one or both of the ples) show that 1 1 d4 k e−ikx 2 ∆(x) = − 4 (2π) C k − M2 are homogeneous solutions for the contours: k0 −E ∆_ ∆+ E ∆ One way of showing this is to write these contours as integrals of the form under (a). . (h) Show that the causal Green’s function ∆C (x) = θ(x0 ) ∆+ (x) − θ(−x0 ) ∆− (x). (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). x) = 0 and argue that one can use Lorentz invariance to show that ∆(x) = 0 for x2 < 0. f (k) ∝ θ(k 0 ).

1) The parity operator transforms We will consider the transformation properties for a fermion field ψ(x). The consequences of P. Table 8. x (8.2) where ηP is the intrinsic parity of the field and A is a 4 × 4 matrix acting in the spinor space. (iγ µ ∂µ − M ) ψ(x) = 0. −r). ˜ (8. time reversal (T) and charge conjugation (C). Both ψ p and ψ satisfy the Dirac equation. r) −→ xµ ≡ xµ = (t. starting with the Dirac equation for ψ(x).1 Parity xµ = (t.1: The behavior of classical quantities under P. parity (P). T. We can determine A. writing −1 ψ(x) −→ Pop ψ(x)Pop = ηP Aψ(˜) ≡ ψ p (x). T and C for classical quantities is shown in the table 1. 8.Chapter 8 Discrete symmetries In this chapter we discuss the discrete symmetries. 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 .

e. −λ) eik·x (2π)3 2Ek ˜ d3 k ˜ ˜ ˜ ˜ ηP b(k. −λ)v(k. . m) = = ˜ u(k. λ)γ0 v(k.4) (8. the above operation reverses the sign of λ). λ)v(k. if helicity λ is used instead of the z-component of the spin m. −λ) eik·x 3 2E (2π) ˜ k d3 k ˜ ˜ ηP b(k. (i˜ µ ∂µ − M ) ψ(˜) = 0. −λ).8) 3 = λ = λ = λ From this one sees immediately that −1 ˜ Pop b(k. λ) e−ik·x − d† (k. In the same way as the Fermion field. choosing ηP is real (ηP = ±1) particle and antiparticle have opposite parity. x (8.Discrete symmetries 67 After parity transforming x to x the Dirac equation becomes after some manipulations ˜ ˜ x iγ µ ∂µ − M ψ(˜) = 0.5) (8. ˜ Pop d(k.3) (8. m) γ0 v(k. x It is straightforward to apply this to the explicit field operator ψ(x) using γ0 u(k. λ) eik·˜ (2π)3 2Ek d3 k ˜ ˜ ˜ ˜ ηP b(k. For the scalar field we have seen −1 φ(x) −→ Pop φ(x) Pop = ηP φ(˜). γ x (8.10) i.7) = λ d k x x ηP b(k. λ)u(k. iγ µ† ∂µ − M ψ(˜) = 0. λ) eik·x . op P (8.6) (check this for the standard representation. x Therefore γ0 ψ(˜) is again a solution of the Dirac equation and we have x ψ p (x) = γ0 ψ(˜). ˜ −v(k. The result is ψ p (x) −1 = Pop ψ(x) Pop = ηP γ0 ψ(˜) x (8. λ) P −1 = −η ∗ d(k. λ) Pop = ηP b(k. λ)γ0 u(k.9) (8.2 Charge conjugation We have already seen the particle-antiparticle symmetry with under what we will call charge conjugation the behavior T (8.12) and for the vector field −1 Aµ (x) −→ Pop Aµ (x) Pop = −Aµ (˜). x µ (iγ ∂µ − M ) γ0 ψ(˜) = 0. λ) e−ik·˜ + d† (k. −λ). λ)v(k. m). −λ) e−ik·x − d† (k. −λ) e−ik·x − d† (k. m). λ)u(k. 8. (2π)3 2Ek (8.13) ψ(x) −→ ψ c (x) = ηC C ψ (x). x The latter behavior of the vector field will be discussed further below. one can also consider the scalar field and vector fields.11) (8. −λ)u(k.

Norm conservation requires Top to be anti-unitary2 . (2π)3 2Ek (8. antilinear operator is anti-unitary if A† = A−1 . m) ¯ C v T (k. Top |φ = φt | = (|φt )∗ . λ)C v T (k. 8. r). λ)v(k. λ) −1 Cop = ∗ ηC b(k.20) The time reversal operator transforms We will again consider the transformation properties for a fermion field ψ(x). to find ψ t (−˜) that is a solution of the Dirac equation.18) (8. One has Aφ|Aψ = φ|ψ ∗ = ψ|A† Aφ = ψ|φ . As time reversal will transform ’bra’ into ’ket’.19) Cop d(k. m) ¯ = = v(k.15) (where one must be aware of the choice of spinors made in the expansion.21) where A is a 4 × 4 matrix acting in the spinor space. (8. u(k.16) d k ηC d(k. x (8. λ) e−ik·x + b† (k. x 1A 2 An −iC −1 γ µ C∂µ − M ψ(−˜) = 0. γ x i(γ5 C)−1 γ µ γ5 C∂µ − M ψ(−˜) = 0. x x (8. λ). ((iγ µ )∗ ∂µ − M ) ψ(x) = 0. The action on the spinors (using C = iγ 2 γ 0 = −iρ1 σ 2 in standard representation) gives C uT (k. λ) eik·x .3 Time reversal xµ = (t.17) 3 = λ This shows that −1 Cop b(k. λ)u(k. m). λ). as discussed in section 4). r) −→ −˜µ ≡ −xµ = (−t. writing −1 ψ(x) −→ Top ψ(x)Top = ηT Aψ(−˜) ≡ ψ t (−˜). we start with the complex conjugated x Dirac equation for ψ. i. The (time-reversed) Dirac equation becomes. .22) = Aψ|Aφ ∗ is antilinear if A(λ|φ + µ|ψ ) = λ∗ A|φ + µ∗ A|ψ . x µ (iγ ∂µ − M ) γ5 Cψ(−˜) = 0. Therefore ψ c (x) = = λ −1 ¯ Cop ψ(x) Cop = ηC C ψ T (x) (8. For a quantized field one has −1 ∗ −1 Top fk (x)bk Top = fk (x)Top bk Top . λ)C uT (k. (i˜ µ∗ ∂µ − M ) ψ(−˜) = 0. The same relations hold for helicity states. ˜ x −(iγ µ )∗ ∂µ − M ψ(−˜) = 0. (8. it is antilinear1 . x iγ µT ∂µ − M ψ(−˜) = 0. λ) eik·x ¯ ¯ (2π)3 2Ek d3 k ηC d(k.e. m).14) (8. λ) e−ik·x + b† (k. λ) Cop = ηC d(k.Discrete symmetries 68 the latter being also a solution of the Dirac equation. x (8.

λ) e−ik·x (2π)3 2Ek (8. λ)u∗ (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. ˜ Top d(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(−˜).Discrete symmetries 69 Table 8. λ)u∗ (k. λ)v ∗ (k. λ)iγ5 Cv(k. λ)iγ5 Cu(k. λ |¯. λ) e−ik·x 3 2E (2π) k d3 k ˜ ˜ ηT b(k. ˜ = 0|b(k)|A 8. λ a P |a.27) 3 = λ = λ = λ From this one obtains −1 ˜ Top b(k. λ)v ∗ (k. there are 16 independent of these bilinear combinations. op T (8. −p.2: The transformation properties of physical states for particles (a) and antiparticles (¯).25) (check this for the standard representation). p. ∗ ˜ v (k. λ)u∗ (k. λ| ¯ C |¯. λ). λ) e−ik·˜ (2π)3 2Ek d3 k ˜ ˜ ˜ ˜ ηT b(k. λ) i γ5 C v(k. a state |a. Therefore one has A|k = A|b† (k)|0 = A|T † T b† (k) T † T |0 ∗ = At |b† (k)|0 ∗ ˜ t = k|At . −p.4 Bi-linear combinations In quantities such as currents and lagrangians often bilinear combinations of spinor fields are encountered. λ) eik·˜ + d† (k. λ a |a. λ) eik·x + d† (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. −λ a T a. −λ |¯.29) In table 2 the behavior of particle states under the various transformations has been summarized. λ) eik·x + d† (k. (8.30) (8. λ). λ) ψ t (x) −1 Top ψ(x) Top = i ηT γ5 Cψ(−˜) x = = ˜ u∗ (k. p. λ). λ) T −1 = η ∗ d(k. p. −p.28) (8. λ) e−ik·x 3 2E (2π) k ˜ d3 k ˜ ˜ ˜ ˜ ηT b(k. p. λ). λ) Top = ηT b(k.34) . −p. The result is = = λ (8.33) (8. Since there are 16 independent 4 × 4 matrices.32) (8. Note that applying an anti-unitary transformation such as Top one must take for the matrix element ˜ the complex conjugate.24) (8. λ)v ∗ (k. λ) eik·x + d† (k. x (8.31) (8.26) d k x x ηT b(k. λ| a.

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 . of which the expectation value between momentum eigenstates can be simply found. F. C. F. p′ |V µ (0)|p = u(p′ )γ µ u(p). σ µν . Applying the results from the previous sections it is straightforward to determine the behavior of the combinations under P.g. (q 2 ). p) = γµ F1 (q 2 ) + iσµν q ν F2 (q 2 ).38) (8. This implies p′ |V µ (x)|p = e−i (p−p )·x p′ |V µ (0)|p . e.40) where the coefficients are functions of invariants constructed out of the momenta. called form factors. (q 2 ).Discrete symmetries 70 Table 8. ¯ V µ (x) =: ψ(x)γ µ ψ(x) :. γ5 γ µ or γ5 ). For instance for nucleons one has Γµ (p′ . ¯ In general the expectation value between momentum states can be more complicated.36) (8. as well as under the combined operation Θ = PCT (see Table 3). 2M (8.35) The x-dependence can be accounted for straightforwardly using translation invariance. p′ .5 Form factors Currents play an important role in field theory. s . (8.39) where Γµ (p′ . In many applications the expectation values of currents are needed. γ ν ]. for the vector current V µ (x). As an example consider the vector current for a point fermion. p′ |V µ (x)|p = u(p′ )Γµ (p′ . s′ |V µ (x)|p. momenta (pµ or p′µ ) or constant tensors (ǫµνρσ or gµν ). 8. C. (8. Note that the lagrangian density L (x) → L (−x) under Θ. 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). ¯ ′ ′ (8.. p)u(p) e−i (p−p )·x . and T.37) (8.3: The behavior of the independent bilinear combinations of fermi fields under P. in this case only q 2 = (p′ − p)2 . p) can be built from any combination of Dirac matrices (1. γ µ . The 16 combinations of Dirac matrices appearing above are linearly independent. V µ (x) = ei Pop ·x V µ (0)e−i Pop ·x . 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 )..41) 2M 2M .

p)γ0 u(p). p) that Γµ (p′ . C and T invariance. or relations that follow from the equations of motion. p′ )γ0 u(p). p) that γ0 Γµ† (p. p). e. ¯ p ˜ Therefore parity invariance implies for the structure of Γµ (p′ . p)u∗ (˜) up p µ p ˜ = ei q·x u(p′ ) (iγ5 C)Γ∗ (˜′ .Discrete symmetries 71 but that are eliminated because of relations between Dirac matrices.42) where q = p′ − p. ¯ 2M (8. p′ )γ0 = Γµ (p′ . p)(iγ5 C)u(p). • Hermiticity p′ |V µ (0)|p = p′ |V µ† (0)|p = u(p′ )Γµ (p′ .44) = p′ |Vµ (−˜)|˜ ˜ x p ∗ ˜x = ei q ·˜ [¯(˜′ )]∗ Γ∗ (˜′ . the Gordon decomposition u(p′)γ µ u(p) = ¯ 1 u(p′ ) [(p′ + p)µ + iσ µν qν ] u(p).1 The matrix element of the electromagnetic current between nucleon states is written as < p′ |Jµ (x)|p >= e−i(p−p )·x u(p′ )Γµ u(p) ¯ Here Γµ iσµν q ν F2 (q 2 ) 2M pµ + pµ ′ = γµ H1 (q 2 ) − H2 (q 2 ) 2M = γµ F1 (q 2 ) + ′ . p) = γ0 Γµ (˜′ . e. p)(i γ5 C). ¯ µ p ˜ Therefore time reversal invariance implies for the structure of Γµ (p′ . p′ )u(p′ ) = u† (p′ )Γµ† (p. p) = (i γ5 C)Γ∗ (˜′ . p)γ0 .45) Exercises Exercise 8.43) = ei q·x u(p′ )γ0 Γµ (˜′ . p)u(˜) ˜ x p ¯ p p ˜ p (8. Therefore hermiticity implies for the structure of Γµ (p′ .g. • Parity p′ |V µ (x)|p = ei q·x u(p′ )Γµ (p′ . µ p ˜ (8. γ5 σµν = (1/2)ǫµνρσ γ ρ γ σ . p)u(p) ¯ = p|V µ (0)|p′ ∗ = (¯(p)Γµ (p. p) that Γµ (p′ .g. or relations based on hermiticity of the operator or P . p)u(p) ¯ † † = p′ |Pop Pop V µ (x)Pop Pop |p ˜x = p′ |Vµ (˜)|˜ = ei q ·˜ u(˜′ )Γµ (˜′ . p)u(p) ¯ † † = p′ |Top Top V µ (x)Top Top |p ∗ (8. p′ )u(p′ )) u = ∗ † u† (p)γ0 Γµ (p. p ˜ • Time reversal p′ |V µ (x)|p = ei q·x u(p′ )Γµ (p′ .

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

t′ ≡ q ′ |q ′ . Consider the two (time-independent) Heisenberg states: |q. t0 )|ψS (t0 ) . t0 ) = e−i(t−t0 )H or i Two situations can be distinguished: (i) Schr¨dinger picture. t ′ QH (t′ )|q ′ . |ψH (t) = |ψH .5) (9. t′ . in which the states are time-independent. QH (t)|q. |ψS (t) = U (t. AH (t) = U −1 (t. and the operators are time-dependent. t ≡ q|q.4) ∂ U (t. [AH . t0 ) = H U (t. i ∂ |ψH ∂t ∂ i AH ∂t ≡ = 0. t0 ) AH (t0 ) U (t. in which the operators are time-independent. (9.6) 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.2) (ii) Heisenberg picture. t . ≡ 0. (9.3) (9. AS (t) = AS and the states o are time dependent.1) The time evolution from t0 → t of a quantum mechanical system is generated by the Hamiltonian. i ∂ |ψS ∂t ∂ i AS ∂t = H |ψS . t0 ). t0 ) ∂t (9. (9. t ′ |q .7) 73 .Chapter 9 Path integrals and quantum mechanics 9. H]. (9.1 Time evolution as path integral U (t.

qj ) e 2π dpj exp (i ∆τ [pj qj − H(pj .14) . . The purpose of this is to calculate the evolution for an infinitesimal time interval. eA eB = eC with C = A+B+ (9. qj ). tj+1 |qj . H(P. B]. t and QH (t) = QS and study the evolution of the system by calculating the quantum mechanical overlap amplitude q ′ . Q) expressed in terms of the operators P and Q. ′ (9.13) At least for a simple hamiltonian such as consisting of a kinetic energy term and a local potential.12) where the transformation between coordinate and momentum space involves q|p = ei p. . 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 . The hamiltonian is an operator H = H(P. QS |q ′ ≡ q ′ |q ′ .q . Then qj+1 . 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 . tj+1 |qj . 2 12 12 (9. These can be written in coordinate or momentum representation as Q = P = dq |q q q| dq |q −i ∂ ∂q q| = dp |p p p|. t′ |q. 2π (9.16) this.Path integrals and quantum mechanics 74 and Schr¨dinger states o |q |q ′ QS |q ≡ q|q . with the correction1 being of order (∆τ )2 . use the Campbell-Baker-Hausdorff formula. 2π (9.11) (9. t′ |q. |q1 q1 |e−i H∆τ |q .. qj )]) . which is a hamiltonian function in which the operators are replaced by real-numbered variables. . t = q ′ |e−i H(t −t) |q . B] + [A.qj )∆τ e e 2π dpj i pj (qj+1 −qj )−i ∆τ 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 ′ . [A. n |q (9. (9. t = = q ′ |e−i n H∆τ |q = q ′ | e−i H∆τ dq1 ..8) Choose t as the starting point with |q = |q. ˙ 2π 1 1 1 [A. tj = = = 1 For dpj i pj (qj+1 −qj ) −i H(pj .10) dqn−1 q ′ |e−i H∆τ |qn−1 qn−1 | .15) By letting the kinetic and potential parts act to left and right respectively one can express the expectation values in terms of integrals containing H(pj . B]] + [[A. Combining the two terms gives qj+1 . B] + . . (9.

F3 [α. ˙ where Dq and Dp indicate functional integrals. qN ) ˙ . while Dα F [α] = x dαx N F (αx ) (9. dx α2 (x). y) in the above examples can be (hermitean) operators acting on the functions. α(x) → αx and F [α] = F (αx ) changes into a multivariable function. q)] . .20) becomes a multidimensional integral. . ′ ′ q1 . including differential operators. qN . Before proceeding we also give the straightforward extension to more than one degree of freedom. . ≡ exp − 1 2 dx α2 (x) . . . (9. Two working approaches for functional integration are the following: (i) the heuristic division of the space on which α acts into cells. .18) t ′ = n=1 Dqn D pn 2π exp i t dτ n pn qn − H(p1 . t N (9. dx dy α(x) K(x. . . pN . . . t′ |q1 . t′ |q. What one is after is the meaning of Dα F [α].2 Functional integrals In this section I want to give a fairly heuristic discussion of functional integrals. i. 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. etc.17)  = Dq D dτ [pq − H(p. qN . . . i. t = τ  dq(τ ) dp(τ ) exp i 2π p exp i 2π t′ t t′ t ∆τ [pq − H(p. y) α(y) .e. Examples are F1 [α] = α ≡ F2 [α] = α2 ≡ dx α(x). β] = α K β ≡ 1 F4 [α] = exp − α2 2 F5 [α.Path integrals and quantum mechanics 75 or for the full interval q ′ .e. dx dy α∗ (x) K(x. 9. . q)] ˙ (9.19) 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). q1 . α∗ ] = exp (−α∗ K α) ≡ exp − The kernels K(x. F [α] ∈ R.

21) again a multidimensional integral.Path integrals and quantum mechanics 76 (ii) Write α in terms of a sum of orthonormal basis functions. A useful property of functional integration is the translation invariance. N (9.28) Dα Dα∗ exp (−α∗ α − α∗ ω − ω ∗ α) = exp (ω ∗ ω) . 1 = √ det K 1 Dα Dα∗ exp (−α∗ K α) = det K 1 Dα exp − α K α 2 (α real-valued) (α complex-valued) (9. (9.27) (9. Note that procedure (i) is an example of the more general procedure under (ii). with Dα F [α] = n dαn N F (αn ) (9.22) = x The last equality is obtained by defining the right measure (normalization N ) in the integration. (9.25) (see Exercises).n fm (x) Kmn fn (y) and consider F [α] = F (αn ) as a multivariable function. 2 (9.n dx fm (x)fn (x) = n = n dαn 1 exp − α2 N 2 n √ 2π ≡ 1. 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]. . y) = ∗ m. Consider the gaussian functional as an example. Physical answers will usually come out as the ratio of two functional integrals and are thus independent of the chosen measure.23) Having defined the Gaussian integral. Using method 1 one writes 1 Dα exp − α2 2 = x dαx N exp − 1 2 ∆x α2 x x = x dαx 1 exp − ∆x α2 x N 2 1 N 2π ∆x ≡ 1. α(x) = n αn fn and K(x. the following integrals can be derived for a real symmetric or complex hermitean kernel K.24) (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.26) As an important application of translation invariance we mention the identities 1 Dα exp − α2 − ωα 2 = exp 1 2 ω .

33) (9. (9.Path integrals and quantum mechanics 77 Functional differentiation is defined as δ .30) (9. y) θ(y) . F [θ.g.35) For applications to fermion fields.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. θ∗ ] = exp (−θ∗ K θ) ≡ exp − Integration for Grassmann variables is defined as dθ 1 = 0 and dθ θ = 1. αy = δxy . (9. It is now easy to check that Dθ∗ Dθ exp (−θ∗ K θ) = det K.34) (9. α(y) = δ(x − y). dx dy θ∗ (x) K(x. (9. θ2 = 0. products of the same pair appear.29) (9. we need to consider Grassmann-valued functions involving anticommuting Grassmann variables.31) δ δ α = dy α(y) = dy δ(x − y) = 1. −→ δα(x) ∆x ∂αx ∂αx δ ∂ ∂ ∗ with . δα(x) 2 2 δ exp (−α β) = −β(x) exp (−α β) . which vanish. Examples of functional differentiation are (9. In principle the definitions of functionals is the same. y) β(y) δα(x) δ 1 1 exp − α2 = −α(x) exp − α2 . n θn fn (x) dx θ∗ (x) θ(x) . the Gaussian-type functionals. δα(x) δα(x) δ αKβ = dy K(x. θη = −ηθ.38) . −→ fn (x) δα(x) ∂αn ∂αm n Note that δ(x − y) = n ∗ fn (x) fn (y).37) We note that from the first to the second line one needs to realize that a pair of different Grassmann variables behaves as ordinary complex variables. e. however.32) (9. i. δα(x) (9. θ∗ ] = exp (−θ∗ θ) ≡ exp − F [θ. In the expansion of the exponential (from second to third line).e. δα(x) or in discretized form δ 1 ∂ ∂ with . αn = δmn .

Path integrals and quantum mechanics

78

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

= δ(x − y),

(9.39)

(9.40) (9.41)

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

9.3

Time ordered products of operators and path integrals

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

q ′ , t′ |q, t =

Dq exp i
t

dτ L(q, q) , ˙

(9.44)

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

Dq exp i
t i 2

dτ Leff (q, q) , ˙

(9.45)

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

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

δ(0) ln f (q).

dqi q ′ , t′ |qn , tn . . . (9.46)

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

dτ [pq − H(p, q)] ˙

(9.47) (9.48)

Dq q(s) exp i
t

dτ L(q, q) . ˙

Path integrals and quantum mechanics

79

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

(9.49)

dτ L(q, q) . ˙

(9.50)

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

9.4

An application: time-dependent perturbation theory

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

(9.51) (9.52) (9.53)

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

= HI (t) ψI (t), = [AI , H0 ],

(9.54) (9.55)

(9.56) (9.57)

dτ HI (τ ) U (τ, t),
t

(9.58)

U (n) (t′ , t),

(9.59)

1−i (−i)

t t′ t

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

Path integrals and quantum mechanics
t′ t t′ t

80

U

(2)

(t , t) = 1 − i = (−i) . . .

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

t′

U

(n)

(t , t) = (−i)
t

dτ HI (τ ) U (n−1) (τ, t)
t′ τ1 τn−1

= (−i)n
t

dτ1
t t′ t′

dτ2 . . .
t t′

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

=

(−i)n n!

dτ1
t t

dτ2 . . .

(9.60)

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

dτ1
t t

dτ2 HI (τ1 ) HI (τ2 )

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

dτ1
t t

dτ2 HI (τ1 ) HI (τ2 )
t′ t′

+

1 2

dτ2
t τ2

dτ1 HI (τ1 ) HI (τ2 ),

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

dτ1
t t

dτ2 HI (τ1 ) HI (τ2 )θ(τ1 − τ2 )
t′

+

1 2

t′

dτ2
t t

dτ1 HI (τ1 ) HI (τ2 )θ(τ1 − τ2 ),

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

dτ1
t t

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

the desired result.

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

dτ1
t t′ t

dτ2 . . .
t

dτn T HI (τ1 ) . . . HI (τn ) (9.61)

≡ T exp −i

dτ HI (τ ) .
t

t is related to the Schr¨dinger state |q by |q. ˙ The combination of exponentials in the last step is allowed because we are simply dealing with classical quantities (not operators!). it is possible to switch on an interaction. t|n = q|n e−iEn t with q|n the time-independent wave function. Before and after these processes there is only the vacuum or ground-state |0 . The Heisenberg state |q. q)] . i. t|Q.63) for plane waves. T = n J t′ = = Dq exp i t dτ [L(q. one has q ′ .Path integrals and quantum mechanics source experiment detector 81 T L t L+Jq t’ L T’ Figure 9. a set |n of physical eigenstates. T ′ |q ′ .e. t) φ∗ (Q) e+i En T . t′ |q. for example x|n x|n = φn (x) = ei kn x = e −ω 2 x2 /2 (9.66) . t |q.5 The generating functional for time ordered products By introducing a source-term. Considering the source to be present between times t and t′ .e. t = ei Ht |q . t′ q ′ . t t′ V =0 Dq D dτ HI (q) exp i t t′ t dτ [pq − H0 (p. t|Q. say. n (9. T .62) p Dq D exp i 2π t dτ [pq − H(p. t Q′ . i. We can again rewrite the time-ordered products as functional integrals as derived in the previous section. φn (q. L(q. ˙ ˙ physically pictured as. q) → L(q. T = n q. Consider. harmonic oscillator. t q. furthermore.64) (9. t = = ′ ′ V = q .s. ˙ J (9. making the electron) and the absorption of an electron (think of a detector). t |T exp −i p exp −i 2π t′ ′ ′ t′ t dτ HI (Q(τ )) |q. q . T < t < t′ < T ′ . T with q. t′ |q. for the physical states o q. q) + J(t) · q. q)] ˙ (9. for g. which in turn are embedded between an early time T and a future time T ′ . the creation of an electron (think of a radio-tube. 9. q) + J(τ )q] . t|n n|Q.65) J dq ′ dq Q′ . T ′ |Q.1: Physical picture of vacuum to vacuum amplitude in presence of a source term This is not a surprising result.

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

g. .Path integrals and quantum mechanics 83 In terms of the imaginary time we can write the Euclidean generating functional. (9. τ dq ˜ dt ≡ −L q.77) where the expressions have been divided by Z[J] and ZE [J] in order to get rid of dependence on multiplicative factors.2 Check the examples of functional derivation for real-. . 1 = √ . dq ˜ dt = 1 2 dq ˜ dt dq ˜ dt 2 − V (q).1 Convince yourself by choosing an appropriate (orthonormal) set of functions that 1 Dα exp − α K α 2 for real-valued functions.72) (9. . Exercises Exercise 9. dq dt = 1 2 1 2 dq dt 2 Dq exp Dq exp − ∞ d˜ L q. As discussed in the previous section the interesting quantities are obtained from functional differentiation with respect to the sources. ZE [J] = = where LE q. δ n Z[J] 1 Z[J] δJ(t1 ) . det K Exercise 9. when L q. δJ(tn ) J = 0 ˜ t = it (9.and Grassmann-valued functions. + V (q). .75) = − LE q. In all of these cases the determinant of the (hermitean) operator K is defined as det K ≡ p Kp . ˜ dt (9. complex. (9.76) which is a positive definite quantity. ˜ ˜ ZE [J] δJ(t1 ) . ensuring convergence for the functional integral ZE [J]. i τ −∞ ∞ −∞ dq d˜ τ + J(˜)q(˜) τ τ dq d˜ τ − J(˜)q(˜) τ τ .73) d˜ LE q. i dq . where Kp are the eigenvalues of the matrix Kmn in K(x. for Grassmann valued functions.74) − V (q) 2 (9. δJ(tn ) = (i)n J=0 1 δ n ZE [J] . y) = ∗ m.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. The differentiations in Z[J] and ZE [J] are related. . det K 1 . e.

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

Π) (10. δJ(xn ) 85 ≡ G(n) (x1 . .2) = ∞ −∞ d4 xE [LE (φ.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.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. ∂µ φ) + Jφ] (10. i. 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 ) . .3) The Euclidean formulation is as before. x′ . .Chapter 10 Feynman diagrams for scattering amplitudes 10. t′ |φ. . . t = = Dφ DΠ exp i t σ′ ˙ dτ d3 x Π(x)φ(x) − H (φ.e. x. .4) J=0 . t) is considered as a set of quantum operators φx (t) in the Heisenberg picture and t′ φ′ . ∂µ φ) − Jφ] (10. φ(x. φ(xn )|0 = (−i)n δ n Z[J] δJ(x1 ) . xn ). (10. . .

8)  1 d4 x  φ (∂µ ∂ µ + M 2 ) φ − Jφ 2 K  (10.9) i. . .3) one finds 1 Z0 [J] = exp − i 2 d4 x d4 y J(x) ∆F (x − y) J(y) ≡ exp (i W0 [J]) . . (10.5) Z0 [J] for the free scalar field For the (free) scalar field lagrangian L = ← 1 1 ∂µ φ∂ µ φ − M 2 φ2 .13) . (2π)4 (k 0 + iǫ)2 − E 2 (10. .7) where we made the choice Z0 [0] = 1 and ∂µ ∂ µ + M 2 ∆F (x − y) = −δ 4 (x − y). with Z0 [0] = 1. . These Green functions appear in the expansion Z[J] = n in n! d4 x1 . As discussed in the previous section (exercise 9.Feynman diagrams for scattering amplitudes 86 for which we introduce the picture x1 G(n) (x1 .11) Depending on the path choosen in the complex k 0 plane (see exercise 7. xn ) J(x1 ) . . xn ) = x2 xn . . (10. which can be written as ∆(x) = ˜ with (k 2 − M 2 )∆(k) = 1 or ˜ ∆(k) = 1 1 1 = 0 2 . (10. . . 2 2 → (10. one immediately sees that i∆F (x1 − x2 ) = δ 2 Z0 [J] (−i)2 Z0 [0] δJ(x1 ) δJ(x2 ) = G0 (x1 − x2 ) ≡ (2) x1 x2 (10. Furthermore. . = 0 2 2 2 k2 − M 2 (k ) − E 2 (k ) − k − M (10. . ∆F (the socalled Feynman propagator) is the Green’s function of the Klein-Gordon equation.e. .4) one distinguishes the retarded Green’s function ∆R (x) = d4 k e−i k·x .10) J=0 In order to determine ∆F . d4 xn G(n) (x1 .12) d4 k −i k·x ˜ e ∆(k) (2π)4 (10. consider the general solution of ∂µ ∂ µ + M 2 ∆(x) = −δ 4 (x). J(xn ).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).

∆(x) = − d4 k e−i k·x . the advanced Green’s function ∆A (x) = satisfying ∆A (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 .21) ∆F (x) = = −i 0 kE =−∞ d4 kE e−i kE ·xE 2 (2π)4 kE + M 2 (10.22) (10.17) d4 k e−i k·x . µ 1 1 • Firstly.4). 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 .14) They are solutions of the inhomogeneous equation. 4 k 2 − M 2 + iǫ (2π) (10. t) = 0 for t < 0.e. (2π)4 (k 0 − iǫ)2 − E 2 (10. (2π)4 k 2 − M 2 (10. (2π)4 k 2 − M 2 where the latter contour can be deformed to the contour for ∆C . 10. ∆F = ∆C . the second is to explicitly consider 0|T φ(x1 ) φ(x2 )|0 . 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). 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. i. in the (well-defined) Euclidean formulation starting with LE = 2 (∂E φ)2 + 2 M 2 φ2 µ 2 2 Jφ with the equation of motion (∂E ) − M φ(x) = −J(x). (10. φ+ (0)] = (2π)3 2E are also shown in exercise (7. . (2π)4 k 2 − M 2 (10. the generating functional can be written 1 µ µ Z0 [J] = Dφ exp d4 xE (10.20) = exp − 2 where or (see fig.19) φ ∂E ∂E − M 2 φ + Jφ 2 1 d4 xE d4 yE J(x) (−i ∆F (x − y)) J(y) . The first possibility is to consider the well-defined Euclidean formulation. φ(0)] = i∆+ (x) + i∆− (x) d3 k e−i k·x = [φ+ (x).Feynman diagrams for scattering amplitudes 87 satisfying ∆R (x. t) = 0 for t > 0.18) Thus we see various solutions. φ− (0)] = (2π)3 2E d3 k e−i k·x = [φ− (x).15) e−i k·x d4 k . The solution of the homogeneous equation ∂µ ∂ µ + M 2 ∆(x) = 0 can also be written as an integral in k-space. (10.1 for countours) µ (∂E )2 − M 2 (i ∆F (x − y)) = −δ 4 (x − y) 0 kE =∞ (10. The contours for i ∆(x) i ∆+ (x) i ∆− (x) = [φ(x).

(4) (2) It should be clear that G0 can be expressed in terms of G0 = i ∆F . . φ+ (y)]|0 = θ(x0 − y 0 ) i ∆+ (x − y) − θ(y 0 − x0 ) i ∆− (x − y) (10. δJ(x4 ) 2 J=0 = − [∆F (x1 − x2 ) ∆F (x3 − x4 ) + ∆F (x1 − x3 ) ∆F (x2 − x4 ) + ∆F (x1 − x4 ) ∆F (x2 − x3 )] . Neglecting multiplicative factors or equivalently assuming that Z0 [0] = 1. because this is the only quantity entering Z0 [J]. we have G0 (x1 .24) Knowing the explicit form of Z0 [J] it is straightforward to calculate the 4-points Green’s function. = θ(x0 − y 0 ) 0|[φ+ (x). . φ− (y)]|0 − θ(y 0 − x0 ) 0|[φ− (x). x2 . 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. the same result as above. x4 ) = 0|T φ(x1 ) φ(x2 ) φ(x3 ) φ(x4 )|0 1 δ4 exp − i J ∆F J = (−i)4 δJ(x1 ) . or diagrammatically (4) (10.Feynman diagrams for scattering amplitudes 0 1 1 0 0 0 1 1 1 0 1 0 0 0 1 0 1 0 11 00 k 1 0 1 0 k 11 00 E 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 -E 1 0 1 0 1 0 1111111111111111111111111111111111111 0000000000000000000000000000000000000 1111111 1 0000000 0 11111111111111111 00000000000000000 11111111111111 00000000000000 11 00 11 00 1 0 1 0 1 0 11111111 1 00000000 0 1 0 1 0 E 1 0 1 0 1 0 1 0 11 00 1 0 1 0 1 0 1 0 1 0 1 0 0 1 1 0 1 0 1 0 1 0 1 0 88 Figure 10. We find (see also section 7.3) 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).25) x1 x2 x3 = x4 + + .1: Contour in the k 0 plane for the Feynman propagator in Euclidean and Minkowski space • As the second possibility. x3 .

34) Introducing a vertex point to which four propagators are connected.g ) (x1 . the generating functional can be written. x3 . x2 ) = G0 (x1 .29) This expression will be the one from which Feynman rules will be derived.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 . ≡ −i g d4 z (10.30) and in zeroth order G(0. x4 ) = To first order in g one has Z (g ) [J] = = 1 0 0 x1 + + x2 (10. with propagators (i ∆F ) being connected to vertices (i LI ) according to the above expression for the functional Z[J].Feynman diagrams for scattering amplitudes 89 10.g 0 ) = G0 = 1.32) (10.e. (2) (0) (10. L (φ) = L0 (φ) + LI (φ). (10. (10. To zeroth order in the coupling one has 0 1 Z (g ) [J] = Z0 [J] = exp − i 2 J ∆F J .g ) (x1 .2 Generating functionals for interacting scalar fields d4 x [L0 (φ) + LI (φ) + Jφ] 1 δ i δJ(z) d4 x [L0 (φ) + Jφ] (10.28) 1 exp − i 2 = exp 1 2 d4 x d4 y exp i δ δ i ∆F (x − y) δφ(x) δφ(y) d4 z [LI (φ) + J(z)φ(z)] φ=0 .35) z .27) When interactions are present. x2 . x2 ) = i ∆F (x1 − x2 ) = G(4.26) Dφ exp i = exp i d4 z LI 1 δ i δJ(z) d4 x d4 y J(x) ∆F (x − y) J(y) (10.31) G(2. (10. Z[J] = = Dφ exp i exp i d4 z LI (10. i. x4 ) = G0 (x1 . x3 . x2 . Consider as an example the interaction LI (φ) = − g 4 φ 4! in the scalar field theory discussed sofar.

This remains also true for the interacting case. i.42) x1 G(n) (x1 . (10. .41) in terms of socalled connected Green’s functions. did not contain parts that could be written as products of simpler Green’s functions. d4 xn G(n) (x1 . .e. . x2 ) = 1 z x1 x2 + 1 2 x1 z x2 . . . . . (10. . . d4 xn G(n) (x1 . . . (10. .g ) (x1 . .40) in−1 n! d4 x1 . .37) G(2. xn ) = (−i)n−1 c which are denoted δ n W [J] δJ(x1 ) . J(xn ). In the free case we have seen that the exponent of Z[J] contained all essential information. . xn ) J(x1 ) . . . which also was the only Green’s function for which the diagram was connected. J(xn ) c (10. Z[J] = n in n! d4 x1 . . .Feynman diagrams for scattering amplitudes 90 one has J Z (g1 ) [J] = 1 8 1 − 4 J J 1 + 24 J Z0 [J]. Connected Green’s functions We have now seen the Green’s functions as the quantities appearing in the expansion of the generating functional Z[J]. (10.38) The result for the 4-points Green’s function is left as an exercise.g 1 ) = 1 8 z 1 8 . . (10. J=0 (10. To see this write Z[J] = exp (i W [J]) with (by definition) the expansion W [J] = n or i W [J] = ln Z[J]. . xn ) = x2 c xn Specific examples are G(0) = c G(1) (x1 ) = c G(2) (x1 . G(n) (x1 . δJ(xn ) .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. . . x2 ) = c (vacuum bubble) (tadpole) (connected propagator) . .36) J J from which one obtains G(0. xn ) J(x1 ) .

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

pn . out| and S † |α. (2) The one-particle state is invariant (conservation of energy and momentum. Sβα ≡ β. In line with the consideration of the generating functional representing the vacuum to vacuum amplitude we consider fields φin . . out = p|p′ . out| = eiϕ0 0. one finds for the connected 4-point Green’s function at order g. out . thus β.e. (3) S is unitary (it conserves the scalar product from initial to final state). (10.43) implying as expected as the only nonzero connected Green’s function G(2) (x1 − x2 ) = (−i) c δ 2 W [J] δJ(x1 ) δJ(x2 ) = G0 (x1 − x2 ) = i ∆F (x1 − x2 ).45) x2 as the only surviving diagram (see exercises). in|p′ . x4 ) = c 1 x3 z (10. translation invariance).g. Also for free field (Heisenberg) operators a distinction is made between φin and φout .g ) (x1 . out = δαβ ↔ SS † = 1. e. x4 10. Next. in|S = β. in|S|p′ . in = p. in = β. (10. p′ .44) J=0 For the interacting theory. out (suppressing except momenta all other 1 m quantum numbers). out|α. this will be translated to the action on fields. out . out|p′ . . .46) The properties of the S-matrix are (1) The vacuum is invariant or |S00 | = 1. out = |p′ .3 Interactions and the S-matrix The S-matrix The S-matrix transforms initial state free particle states (in-states) |α. x2 . ∂µ ∂ µ + M 2 φin (x) = 0. in|S = α. . in|SS † |α. in = p. t = −∞ φin (x) t = +∞ φout (x) φ(x) where φin and φout transform under the Poincar´ group as scalar fields and satisfy the homogeneous e Klein-Gordon equation with the physical mass M . φout and the interpolating field φ(x).Feynman diagrams for scattering amplitudes 92 i. in = p. in|S = 0. in| (choose ϕ0 = 0).47) . in = S|α. out| ⇔ |α. the vacuum expectation value of the field φ(x) is zero. out|p′ . (2) (10. Proof: 0. considered explicitly. out|α. we have W0 [J] = − 1 2 d4 x d4 y J(x) ∆F (x − y) J(y). in = |α. . then (n) (n) Gsc = Gc for n ≤ 3. x1 G(4. x3 . . p. in into final state particle states (out-states) |β. For the free scalar theory. in ⇔ β. Proof: α. in = |p1 . implying the absence of tadpoles. (10.

52) where Z is a constant. a) = S. β.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). as it stands. φ(y)] = Z [φin (x). 2 ∂µ ∂ µ + M0 φ(x) = J(x).62) is then T′ UT [J] T′ = T exp i T d4 x J(x)φ(x) .54) .and out-fields is: φin (x) = S φout (x)S −1 . 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.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 ). implies the strong (operator) convergence φ(x) → Zφin (x). The time evolution operator (see Eq.50) = φ(x) = √ Zφin (x) − √ Zφout (x) − d4 y d4 z ∆R (x − y) K(y. φin (y)] = iZ ∆(x − y).g. 2 ˜ ∂µ ∂ µ + M 2 φ(x) = J(x) + (M 2 − M0 )φ(x) = J(x) √ ˜ φ(x) = Zφin (x) − d4 y ∆R (x − y) J(y) (10. (10. (2π)3 2E (10. (10.e. Proof: β. i. e Proof: φin (Λx + a) = U φin (x)U −1 = U Sφout (x)S −1 U −1 = U S U −1 U φout (x) U −1 U S −1 U −1 = U SU −1 φout (Λx + a) U S −1 U −1 ⇔ U SU −1 = S. The convergence therefore must be weakened to t→−∞ √ (10. z) φ(z). in|φin S → φin S = S φout (x). which in a later stage (renormalization) will become more important. out|φout = β. The relation between S-matrix and in. this can actually not be used as it would imply [φ(x).49) The field φ can be expressed in φin/out using retarded or advanced Green’s functions. φin (x) = d3 k ∗ a(k) fk (x) + a† (k) fk (x) . S is invariant under e Poincar´ transformations: U (Λ. in|Sφout Finally we check that as expected S does not spoil Poincar´ invariance. a causality condition that can be proven to imply the absence of interactions.48) The fact that φin and φout satisfy the homogeneous Klein-Gordon equation with the physical mass M implies that they create particles and antiparticles as discussed. a)SU −1 (Λ. e. 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. (10.51) (10. √ Although the above. 9. The relation between S and Z[J] To establish this relation.

one has [A.63) .58) In order to find a solution to this equation note that. t i Ut∞ [J] φ(x)U−∞ [J] (10. Noting that 0| : eA : |0 = 1 it follows that F [J] = 0|S U [J]|0 = 0|U [J]|0 = 0out |0in ∝ Z[J]. C] are c-numbers. B]eB .Feynman diagrams for scattering amplitudes 94 and satisfies the property T ′ δUT [J] t = i UtT [J] φ(x)UT [J].60) e d4 y φ+ (y) f (y) d4 z φ− (z) f (z) d4 z φ+ (z) f (z) d4 y [φ(x). z) δU [J] . e = i.56) Since δU [J]/δJ(x) satisfies the same equations as φ(x) and we know the limits. : =i d4 y ∆(x − y) f (y) : e d4 z φ(z) f (z) :. φ− f (10. which is equal to the expression e in which the creation operators are placed left of annihilation operators. φ(x). : e d4 z φ(z) f (z) d4 x φ− (x) f (x) B C B C (10. Thus S U [J] =: exp 1 √ Z d4 x d4 y φin (x) K(x. φ(x). (10. just as we did for φ(x). δJ(x) ∞ or for U [J] ≡ U−∞ [J] ′ (10.59) (10. [A. eB ] = [A. y) δ δJ(y) : F [J]. (10. B] and [A. provided [A. B] is a c-number). Thus applied to the field φ(x) = φ+ (x) + φ− (x). with the use of the Baker-Campbell-Hausdorff formula e−B A e+B = A + [A.57) i [φin (x). z) . B] (for the case that [A. S U [J]] = √ Z δ S U [J]. φ(y)] f (y) e e . 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.62) where F [J] is some (arbitrary) functional.e. we can.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). z) δU [J] δJ(z) δU [J] d4 y d4 z ∆A (x − y) K(y. δJ(z) (10. e e ] = [A. δJ(z) (10. express it in terms of advanced and retarded Green’s functions. δU [J] δJ(x) = = √ i Z U [J] φin (x) − √ i Z φout (x) U [J] − d4 y d4 z ∆R (x − y) K(y. B + C]e e .61) e φ+ f where the normal ordered expression : e φ f : is used. z) d4 y d4 z ∆(x − y)K(y.

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

= −i g . vertices and external particle wave functions in that diagram. 3! 4! The vertices are: = −i f Consider first the lowest order self-energy diagram. the factor 2 corresponds to the two-points Green’s function having two identical ends). ’t Hooft and M.66 and 10.Feynman diagrams for scattering amplitudes 96 (in fact the inverse of the operator found in the quadratic term is for real scalar fields also multiplied by a factor 2. but overall momentum conservation at a vertex is understood. At these vertices each line can be assigned a momentum. Note that in calculating amputated Green’s functions external lines are neglected. = −i g (multiplied with 4! corresponding to the allowed number of permutations of identical particles).27. 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. For the interaction terms in the lagrangian. (Rule 3) Carry out the integration over all internal momenta (keeping track of momentum conservation at all vertices!) (Rule 4) Add a symmetry factor 1/S corresponding to permutation of internal lines and vertices (keeping external lines fixed). For the symmetry factor consider the examples (given in G. Veltman. 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.69) LI (φ) = − φ3 − φ4 . to be precise iLI vertices in momentum space are introduced. If problems arise go back to the defining expression for the generating function in 10.27. Corresponding to external particles wave functions are introduced k k 1 1 In order to calculate the connected amplitude −i Mf i appearing in the S-matrix element these ingredients are combined using Eqs 10. Diagrammar) in the case of the interaction terms g f (10. which cancels the factor 1/2 in the quadratic piece. or calculating full Green’s functions external lines are treated as propagators.

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

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 . η] = Dψ Dψ exp i d4 z LI exp − d4 x ψ (i/ − M ) ψ + LI (ψ) + ηψ + ψη ∂ 1 δ 1 δ . (10.73) = i p / − M + iǫ = ji i(/ + M )ji p p2 − M 2 + iǫ The time ordered functions are obtained by functional derivatives from Z(η. namely (Rule 5) Feynman diagrams which only differ by exchanging identical fermions in initial or final state have a relative minus sign.72) (10. = + . but the anticommutating properties of Grassmann variables imply some additional minus sign in Feynman diagrams.Feynman diagrams for scattering amplitudes 98 Dirac fields For fermions the generating functional is given by Z[η. which arises from the quadratic term in exp i d4 z LI .e. which is the solution of (i/ − M )SF (x) = −δ 4 (x) ∂ (i. (Rule 6) Each closed fermion loop gets a sign −1. in e− e− → e− e− scattering (Møller scattering) the lowest order contribution is (see next section for e-e-γ vertex). .. − g2 2 dz dz ′ δ2 δ2 δη(z) δη(z) δη(z ′ ) δη(z ′ ) and the quadratic term in Z0 [η. η). 4 2 − M 2 + iǫ (2π) p and the (oriented) propagator in Feynman diagrams involving fermions is i p j (10. e.− i δη(z) i δη(z) d4 x d4 y η(x) iSF (x − y) η(y) . The two-points Green’s function 0|T φ(x) φ(y)|0 contains a fermionic loop contribution. 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..g. η]. − 1 2 dx dy dx′ dy ′ η(x)S(x − y)η(y) η(x′ )S(x′ − y ′ )η(y ′ )..71) = exp i where iSF is the Feynman propagator for fermions.

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

s3 )γ µ u(k. The diagram and momenta and the commonly used invariants (Mandelstam variables) for a 2 → 2 scattering process are 1 k 3 k’ s t u = (k + p)2 = m2 + M 2 + 2 k · p = (k ′ + p′ )2 = m2 + M 2 + 2 k ′ · p′ = (k − k ′ )2 = q 2 = 2 m2 − 2 k · k ′ = (p − p′ )2 = 2 M 2 − 2 p · p′ p 2 p’ 4 s = (k ′ − p)2 = m2 + M 2 − 2 k ′ · p + t+u= m2 = 2 m2 + 2 M 2 i = (k − p′ )2 = m2 + M 2 − 2 k · p′ The scattering amplitude is given by −iM = u(k ′ . s)¯(k. s4 )(−ie)γ ν u(p.76) 1 .s2 ) which can be written as (see also chapter 4) |M |2 = = 1 4s |¯(k ′ .s4 e4 (m) µν (M) L L .s3 . t2 (10. To lowest order (∝ α = e2 /4π) only one diagram contributes.5 Some examples eµ scattering The first example is the electromagnetic scattering of an electron and a muon. s3 )(−ie)γ µ u(k. s4 )γµ u(p.s4 ) and averaged over spins in the initial state (1/2 × 1/2 × s1 . s2 ). s2 )|2 ¯ q2 (10. s1 ) ¯ −igµν u(p′ . s′ )γµ u(k. s1 ) u e2 u(p′ . s′ ) ¯ u s.s2 . s)γν u(k ′ . where u ¯ L(m) µν = = 1 2 u(k ′ . ¯ q2 (10.79) .Feynman diagrams for scattering amplitudes 100 10.75) Note that the q µ q ν term in the photon propagator are irrelevant because the photon couples to a conserved current. 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 .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. q 4 µν [using that (¯(k ′ )γµ u(k))∗ = u(k)γµ u(k ′ )].s′ ′ ′ = 2 kµ kν + 2 kν kµ + q 2 gµν 1 Tr [(/ ′ + m)γµ (/ + m)γν ] k k 2 ′ ′ = 2 kµ kν + kν kµ − gµν (k · k ′ − m2 ) (10.78) = 2 α2 2 s + u2 + 4t(M 2 + m2 ) − 2(M 2 + m2 )2 .

s4 ) v (p′ . . s2 )γµ u(k.2. (b) Use the definition of the source-connected Green function Gsc and show that indeed sourceconnected diagrams survive. the masses. s3 ) ¯ ¯ 1 . s4 )γ ν u(p. for instance Møller scattering (e− e− → e− e− ) and Bhabha scattering (e− e+ → e− e+ ) are related using crossing symmetry. Basically the same diagram is calculated and the result is the same after the interchange of s ↔ t. s1 )|2 ¯ s 1 . (c) Do the same for the connected Green function Gc . (4) (4) . s2 ) ¯ ¯ 1 2s u(p. x4 ) for the interacting case in φ4 -theory to first order in the coupling constant g as obtained from the full expression for Z[J] in section 10. s1 ) u(k. Exercises Exercise 10. To lowest order (∝ α = e2 /4π) only one diagram contributes. s3 )γ µ v(p′ .Feynman diagrams for scattering amplitudes 101 The second example is the annihilation of an electron pair and creation of a muon pair. .80) Note the similarity in the amplitudes for eµ scattering and e− e+ → µ− µ+ . 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 .s2 3 . . This is known as crossing symmetry. s4 ) u e2 v (k ′ .s4 = 1 2s × v (k ′ . s1 )γν v(k ′ . s2 (10. s2 )γµ u(k.1 (a) Give in diagrammatic notation the full Green functions G(4) (x1 . s3 )γ µ v(p′ . Similarly.s2 .s4 = 1 1 e Tr(/ − m)γµ (/ ′ + m)γν k k Tr(/ + M )γ µ (/′ − M )γ ν p p s2 2 2 1 e4 1 ′ ′ = 4 2 kµ kν + kν kµ − gµν s pµ p′ν + pν p′µ − g µν s s 2 2 M 4π 2 4 and = 2 α2 2 t + u2 + 4s(M 2 + m2 ) − 2(M 2 + m2 )2 . The diagram.

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

2) (11.5) √ is referred to as the invariant mass squared. pa ) and pb = (Eb .e.1 kinematics in scattering processes Phase space The 1-particle state is denoted |p .7) pb = cm (Eb . i n d3 pi . In that case cm pa = (Ea . It is determined by the energy-momentum four vector p = (E. −q). 103 . (11. s is for obvious reasons known as the center of mass (CM) energy.4) For two particle states |pa . A physical state has positive energy. . p1 . For two particles. p) which satisfies p2 = E 2 −p2 = m2 . . q). . . . (11. pn ) = (2π)4 δ 4 (P − pi ) dR(p1 . + n. . Its square root. pn ) = and the reduced phase space element by dR(s. the 1particle phase space is d3 p d4 p θ(p0 ) (2π)δ(p2 − m2 ). The first is the CM system. In the CM frame the threshold value for s obviously is n 2 sthreshold = i=1 mi . . The phase space is determined by the weight factors assigned to each state in the summation or integration over states. . . It is a useful quantity. (11. . and the total momentum four-vector P = pa + pb . (2π)3 2Ei i=1 (11. . a a b b the quantity s = P 2 = (pa + pb )2 . for instance for determining the threshold energy for the production of a final state 1 + 2 + . pb we start with the four vectors pa = (Ea . . This is generalized to the multi-particle phase space dR(p1 .6) (11. To be specific let us consider two frequently used frames. . (11. Here s is the invariant mass of the n-particle system.1) = (2π)3 2E (2π)4 (proven in Chapter 2).Chapter 11 Scattering theory 11. s = (p1 + .3) which is useful because the total 4-momentum of the final state usually is fixed by overall momentum conservation. i. pb ) satisfying p2 = m2 and p2 = m2 . . + pn )2 . pn ). . .

for instance.8) (11. ptrf ). p2 ) = = where λ12 denotes λ(s. p1 . m2 . m2 ). 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. Prove that trf Ea = |ptrf | = a s − m2 − m2 a b .11) (11. 2 s s − m2 + m2 a b √ . m2 ) a b . a (11. mb and s).Scattering theory 104 It is straightforward to prove that the unknowns in the particular system can be expressed in the invariants (ma . m2 ) is a function symmetric in its three arguments.15) Explicit calculation of the reduced two-body phase space element gives dR(s. 2mb (11. Also in this case one can express the energy and momentum in the invariants. 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. = 2 s λ(s. 4s (11. m2 . p1 . 0). In that case trf pa = (Ea . 4π s 4π 8π s 4π (11. 2mb λ(s.13) (11. which in the specific case a b also can be expressed as λ(s. m2 . (11.16) . m2 . 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 . 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 √ = .14) One can. m2 ) a b .10) The function λ(s.9) (11. p2 ) = CM = = 1 (2π)2 1 (2π)2 1 (2π)2 d3 p1 d3 p2 4 δ (P − p1 − p2 ) 2E1 2E2 √ d3 q δ( s − E1 − E2 ) 4 E1 E2 q 2 d|q| √ dΩ(ˆ) q δ( s − E1 − E2 ) 4 E1 E2 which using |q| d|q| = (E1 E2 /(E1 + E2 )) d(E1 + E2 ) gives dR(s. trf Ea (threshold) = 1 2 mb ( i mi )2 − m2 − m2 a b .12) pb = (mb .

Of course there are not only 2-particle channels. H1 + H2 → h + X. there is usually a preferred direction. where the first is referred to as elastic channel and the set of all other channels as the inelastic channels. dt = 2 qq ′ d cos θcm and obtain for the two-body phase space element √ q ′ dΩcm λcd dΩcm √ = (11. At high energies. tmax min λcd /4s.Scattering theory 105 Kinematics of 2 → 2 scattering processes π− + p The simplest scattering process is 2 particles in and 2 particles out. pT = (px . 11.17) (11.23) which are not independent as s + t + u = m2 + m2 + m2 + m2 . = (11. Writing E = mT cosh y. (11. in exclusive measurements all particles are detected.24) Using the relation between t and cos θcm it is straightforward to express dΩcm in dt. pd ) = 4π s 4π 8π s 4π dt dt √ √ . 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 ..26) = 16π q s 8π λab Kinematics of inclusive hard scattering processes In high energy (hard) scattering processes. while the final state often arises from a series of 2-particle processes combined with the decay of an intermediate particle (resonance).22) (11. Instead of the scattering angle.25) dR(s. The various possibilities are referred to as different reaction channels.20) → π0 + n → π+ + π− + n → . for instance the momenta of incoming (colliding) hadrons and it is useful to use for the produced particle rapidity as a variable.4). pc . An often used set of invariants are the Mandelstam variables. In inclusive measurements no particles are detected in the final state. usually many particles are produced. The initial state. The minimum and maximum values for t correspond to θcm CM = m2 + m2 − 2 Ea Ec ± 2 qq ′ a c = m2 a + m2 c (s + m2 − m2 )(s + m2 − m2 ) a c b d − ± 2s √ λab λcd . py ). A specific reaction channel starts contributing at the threshold value (Eq. usually is a 2-particle state. Consider the 1-particle inclusive case.21) (11. .18) (11.. 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. E . in which one particle is detected. 2s (11.27) with m2 = m2 − p2 = m2 + p2 + p2 . however. a c with q = λab /4s and q ′ = being 0 or 180 degrees. Examples appear in → π− + p (11. The variable s is always larger than a c b d the minimal value (ma + mb )2 . the rapidity y is defined T T x y y= 1 ln 2 E + pz E − pz = ln E + pz mT = tanh−1 pz .19) (11. pz = mT sinh y.

29) 3 2E 3 2 (2π) 2 (2π) 16π 2π E 16π 2 2π with at high energies the factor |p|/E ≃ 1. With the momenta defined as in the figures above one has for all these processes the same amplitude M (s. These variables are precisely the Mandelstam variables for the s-channel process (ab → cd). t.Scattering theory 106 It is convenient because under a boost (along z) with velocity β it changes as y → y − tanh−1 β. ¯ tu = (pa − pc )2 = t. ¯ while for the u-channel process (ad → c¯ one has b) su = (pa + pd )2 = u.30) (11.31) . For the t-channel process (a¯ → ¯ one has c bd) st = (pa + pc )2 = t. 11. u). ¯ tt = (pa − p¯)2 = s.28) which (only involving angles) is easier to determine. ut = u) = Mab→cd (s. c bd Mad→c¯(su = u. For large energies and not too small scattering angles (θ ≫ 1/γ) y is approximately equal to the pseudo-rapidity η.2 Crossing symmetry In the previous chapter. tt = s. u) with s = (p1 + p2 )2 = (p3 + p4 )2 . t. u). b ut = (pa − pd )2 = u. b Analiticity of the field theoretically calculated result implies Ma¯→¯ (st = t. t-channel and u-channel processes respectively. which means that rapidity distributions dN/dy maintain their shape. t = (p1 − p3 )2 = (p2 − p4 )2 . ¯ b (11. u = (p1 − p4 )2 = (p2 − p3 )2 . 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 = = = . uu = (pa − p¯)2 = s. η = − ln(tan(θ/2)) = tanh−1 (cos θ). Given a two-to-two scattering process ab → cd one can relate the processes b p p a 2 1 d p 4 _ c −p p 1 3 d p 4 _ d −p p a 1 4 _ b −p 2 p 3 c ab cd a _ ac −p 2 _ _ b bd _ ad p 3 _ c cb They are referred to as s-channel. uu = s) = Mab→cd (s. tu = s. (11. we have seen that the amplitudes for the processes e− µ− → e− µ− and the process e− e+ → µ+ µ− are simply related by an interchange of the variables s and t. t. This is known as crossing symmetry. (11.

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

(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. the amount of decaying particles is proportional to the number of particles with proportionality factor the em decay width Γ. N 1 .35) This quantity is not covariant. (11.37) 2 ma T · V Fermi’s Golden Rule In both the scattering cross section and the decay constant the quantity N/T V appears. σ= √ 2 λab T · V (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 . . N and T · V are covariant. For this we employ in essence Fermi’s Golden rule stating that when the S-matrix element is written as Sf i = δf i − (2π)4 δ 4 (Pi − Pf ) iMf i (11.34) i. the number of scattered or decayed particles is given by 2 N = (2π)4 δ 4 (Pi − Pf ) iMf i dR(p1 . . pn ). . From the solution N (t) = N (0) e−Γ t one knows that the decay time τ = 1/Γ. it is.38) (in which we can calculate −iMf i using Feynman diagrams). T · V ρa (11. (11. . dt (11. Microscopically one has Ndecay = Na · Γ T or Γ= N 1 .e.33) Decay of particles For the decay of particle a one has macroscopically dN = −Γ N.Scattering theory 108 Although this at first sight does not look covariant. as expected.T d4 x = V · T (2π)4 δ 4 (Pi − Pf ). For the (proper) decay width one thus has 1 N Γ0 = .T V. 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 .36) (11. .39) One of the δ functions can be rewritten as T · V (remember the normalization of plane waves).

e.4 Unitarity condition (S † )f n Sni = δf i The unitarity of the S-matrix. (11. p1 . t) λab 4π dt. . . . (11. . . q n ) q n qn √ Mni (q i . pn ) |M |2 q 32π 2 m2 dΩ |M |2 .42) 2−body decay The differential cross section (final state not integrated over) is given by 1 dσ = √ |Mf i |2 dR(s. √ (2ℓ + 1)Mℓ (s) Pℓ (cos θ). q n ). 2 λab and for instance for two particles dσ = q ′ M (s. see 11. θ) in quantum mechanics. pn ). 16π 2 s (11. (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 .Scattering theory 109 (Using normalized wave packets these somewhat ill-defined manipulations can be made more rigorous). θ). implies for the scattering matrix M . The result is N = |Mf i |2 dR(s. (11. i M† M √ − √ 8π s 8π s = fi n dΩn ∗ Mnf qn Mni √ √ . 8π s 2π 8π s .44) This can be used to get the full expression for dσ/dt for eµ and e+ e− scattering.43) dΩcm = π M (s. note the sign and cos θ = qi · qf ). θ)|2 . (11. θcm ) √ q 8π s 2 2 (11. .25) −i Mf i − (M † )f i = − ∗ dΩ(ˆn )Mnf (q f . . θ) = −8π s ℓ (in analogy with the expansion for f (E. . 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 . . for which one has dσ/dΩ = |f (E. ˆ ˆ Inserted in the unitarity condition for M . . for which the √ amplitudes squared have been calculated in the previous chapter. . 11.45) Since the amplitudes also depend on all momenta the full result for two-particle intermediate states is (in CM. pn ). q f ). θ) or M (q i . The sign difference comes from the (conventional) sign in relation between S and quantummechanical and relativistic scattering amplitude. The amplitude −M /8π s is the one to be compared with the quantum mechanical scattering amplitude f (E. p1 .47) M (s. (M † )f n (2π)4 δ 4 (Pi − Pn ) Mni . .46) Partial wave expansion Often it is useful to make a partial wave expansion for the amplitude M (s.41) (11. i. respectively. p1 .

(11.48) If only one channel is present this simplifies to ∗ ∗ −i (Mℓ − Mℓ ) = 2 qMℓ Mℓ .52) From the imaginary part of M (s.49) or Im Mℓ = q |Mℓ |2 . In general a given channel has |Sℓ (s)| ≤ 1. 0). Show that σT = = 4π q 2π q2 (2ℓ + 1) Im Mℓ (s) ℓ ℓ (2ℓ + 1) (1 − ηℓ cos 2δℓ ). σinel = π q2 2 (2ℓ + 1) (1 − ηℓ ).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 . the total cross section can be determined. ℓ (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 . θcm ) √ q 8π s 2 . η = 1) the result σel = 4π q2 (2ℓ + 1) sin2 δℓ . Using σ= dΩ q ′ M (s. σel = 4π ℓ (2ℓ + 1)|Mℓ (s)|2 (2ℓ + 1) ℓ = 4π q2 ηℓ e 2i δℓ − 1 2i 2 . n (11.54) ℓ .53) The difference is the inelastic cross section. which allows writing Mℓ (s) = Sℓ (s) − 1 e2i δℓ (s) − 1 = . in combination with the partial wave expansion for the amplitudes M and the orthogonality of the Legendre polynomials immediately gives for the elastic channel. parametrized as Sℓ (s) = ηℓ (s) exp(2i δℓ (s)).e. (11. (11. 2i q 2i q (11.50) where Sℓ (s) satisfies |Sℓ (s)| = 1 and δℓ (s) is the phase shift. (11. † −i (Mℓ )f i − (Mℓ )f i = 2 † (Mℓ )f n qn (Mℓ )ni .51) and for the case that this is the only channel (purely elastic scattering. q ˆ i.

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

the partial width into e+ e− is Γee = 5. t)} . u) = g(t. Three neutrinos explain the resonance shape.2 GeV. Γ = 2. 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). dt s(s − 4m2 ) with f (t.26 keV. • The Z 0 resonance in e+ e− scattering with M = 91. where furthermore σinel = 0. The half-width of the resonance is M Γ. The cross section σT (π − p) also shows a resonance with the same width with a value of about 70 mb.Scattering theory 112 (Prove this using the unitarity condition for partial waves). • The J/ψ resonance in e+ e− scattering.e. Note that because of the presence of a background the phase shift at resonance may actually be shifted.59) reaching the unitarity limit for s = M 2 . the full width is Γ = 88 keV. This is a narrow resonance discovered in 1974. Knowing that each neutrino contributes about 160 MeV (see next chapter). Exercises Exercise 11. one can reconstruct the resonance shape for different numbers of neutrino species.88 MeV. i. u) + g(t. The phase shift in the resonating channel near the resonance is given by tan δℓ (s) = MΓ . From this one sees that a resonance has the same shape in all channels but different strength. which is attributed to neutrinos. Essentially this resonance can decay into quark-antiquark pairs or into pairs of charged leptons. a fast change in the phase shift for a narrow resonance.60) showing that the phase shift at resonance rises through δ = π/2 with a ’velocity’ ∂δ/∂s = 1/M Γ. its width Γ = 120 MeV. M2 − s (11. At resonance the cross section σT (π + p) is about 210 mb. u) = 1 1 2 (s + u2 ) + 4m2 (t − m2 ) t2 2 2 1 1 ( s − m2 )( s − 3m2 ) tu 2 2 . t) + g(u. The cross section at resonance is about 30 nb. Its mass is M = 1232 MeV. 2 q (s − M 2 )2 + M 2 Γ2 (11.49 GeV. Three famous resonances are: • The ∆-resonance seen in pion-nucleon scattering. All these decays can be seen and leave an ’invisible’ width of 498 MeV. Its mass is M = 3096.1 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) . This characteristic shape of a resonance is called the Breit-Wigner shape. u) + f (u. Limiting ourselves to a resonance in one channel.4) can be written as 4πα2 dσ = {f (t.

Compare the result with the factor trf 1/ρa ρb va that we used in the derivation of the basic expression for cross sections.Scattering theory 113 Exercise 11.2). ρcm ρcm a b 1 cm . we have here already used unitarity to find a finite imaginary part that is proportional to the total decay width. s − M∆ ∆ ∆ 2 calculate Re N (M∆ ) . writing N (s) M1 (s) = 2 + iM Γ . Calculate now fπ . Exercise 11. As discussed for unstable particles. Exercise 11.3 Complete the calculation of the decay width and lifetime of the pion by including the appropriate phase space factors and using the calculation of the squared matrix element in Exc.8. Compare the results with the experimental π − lifetime τ = 2.6 × 10−8 s and the relative decay widths Γ(π − → µ− νµ ) = 0. a quantity which we will ¯ ¯ encounter in the next chapter (section 12. 4.999 877 Γ and Γ(π − → e− νe ) = 0.2 Calculate the flux factor.000 123 Γ. Thus. Show that unitarity can be used one step further to fix also the real part of the residue of the pole in the amplitude.4 The ∆ resonance in N π scattering appears in just one partial wave (a P-wave) as a pole at the position 2 s = s0 = M∆ − iM∆ Γ∆ . cm |va − vb | in the center of mass system for two colliding particles a and b.

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

requiring that the ’connection’ Aµ (x) is such that Dµ → S(x)D µ S −1 (x).18) or considering a path xµ (s) from a fixed origin (0) to point x. x+dx ψ(x + dx) = 1 + dxµ Dµ x+dx ψ(x). 12.16. We note that local gauge invariance requires that we can modify all local systems with a (local) unitary transformation S(x). ∂µ − ig Aµ (x) ψ(x) = 0 dxµ ∂µ − ig Aµ (x(s)) ψ(x) = 0. i.e. ds s dxµ ds′ P exp ig A (s′ ) ψ(0). However. = .17) The three equations given so far immediately give Dµ = ∂µ − ig Aµ (x). A ’constant’ vector that only rotates because of the arbitrary definitions of local frames satisfies Dµ ψ(x) = 0. which connects two identical vectors.15) ψ(x).14) (12. 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). Dν ] ψ(x) 1 − ig dσ µν Gµν ψ(x). In principle such a phase in a given point is not observable.The standard model the basis for x G and x+dx 117 G. i. If there is a ’true’ difference in the vector ψ(x) and ψ(x + dx) it is denoted with the covariant derivative connecting the vectors expressed with respect to the same basis. the real change thus is given by ψ(x) = (1 − dy ρ Dρ )(1 − dxσ Dσ )(1 + dy ν Dν )(1 + dxµ Dµ )ψ(x) = 1 + dxµ dy ν [Dµ . (12. (12.19) This gives rise to a (path dependent) phase in each point. x x+dx ψ(x + dx) = (1 + dxµ ∂µ ) ψ(x). should be independent of such transformations.16) which in the presence of the ’connection’ Aµ differs from the total change between x ψ(x) and x+dx ψ(x+ dx). ds which is solved by dψ(s) ds = dxµ . (12.e. -dx -dy µ µ dyµ µ dx For a vector. if two different paths to the same point give different phases the effects can be observed. What is this physical effect by which the ’connection’ Aµ can be observed? For this consider the phase around a closed loop. x+dx ψ(x) = = 1 + ig dxµ Aa (x)Ta µ 1 + ig dx Aµ (x) µ x x ψ(x) (12. but expresses them with respect to different bases. (12. The relation in Eq.

If the temperature is below a certain critical temperature (T < Tc ) the kinetic energy is no longer dominant and the above hamiltonian prefers a lowest energy configuration in which all spins are parallel. 12.e. external magnetic fields. 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. Classically the (time-independent) ground state is found for V(φ ) F |φ| Figure 12.2: The symmetry-breaking ’potential’ in the lagrangian for the case that m2 < 0. in reality a not perfectly symmetric situation. In this case there are many possible groundstates (determined by a fixed direction in space). In geometric language the effect on parallel transport of a vector depends on the ’curvature’ Gµν . we can disregard those ’perturbations’ and look at the ideal situation. Dν ]. . i>j which is rotationally invariant. which means that by an appropriate transformation S(x) it can be gauged away (see the example of Aharonov-Bohm effect). 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).    φ1    φ . This characterizes spontaneous symmetry breaking. If the temperature is high enough the spins are oriented randomly and the (macroscopic) ground state is spherically symmetric.20) with a lagrangian density of the form L = 1 1 1 ∂µ φ · ∂ µ φ − m2 φ · φ − λ(φ · φ)2 . 12. the groundstate itself appears degenerate. i.g. and equivalently Aµ can be considered as a pure gauge effect. As there can be one and only one groundstate. Nevertheless. usually being (slightly) prejudiced by impurities. this means that there is more than one possibility for the groundstate.2. a theory for a scalar degree of freedom (a scalar field) having three (real) components.The standard model 118 where Gµν = (i/g)[Dµ . Nature will choose one. e.  φ= 2      φ3 (12. Only if this quantity is nonzero a physically observable effect of Aµ exists. 2 2 4 −V (φ) The potential V (φ) is shown in fig. If it is zero one has dxµ Aµ (x) = 0.

  ϕc = 0|φ |0 =  (12. λ such that 0|ϕ1 |0 = 0|ϕ2 |0 = 0|η|0 = 0. Any constant field with ’length’ F is a possible groundstate.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. They are known as the Weyl mode or the Goldstone mode: Weyl mode. In order to see the consequences for the particle spectrum of the theory we construct the lagrangian in terms of the fields ϕ1 . (12. while the quadratic terms must correspond with non-negative masses. A quantum field theory has only one nondegenerate groundstate |0 . The situation now is the following. (12. so one has 0|φquantum |0 = 0 and hence 0|φ|0 = ϕc . say    0    0 . In this situation one speaks of spontaneous symmetry breaking. Writing the field φ as a sum of a classical and a quantum field.e. The field along the third axis plays a special role because of the choice of the vacuum expectation value. for the vacuum Qa |0 = 0.22)     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.24) 1 2 2 4 1 1 1 = (∂µ ϕ1 )2 + (∂µ ϕ2 )2 + (∂µ η)2 + m2 η 2 + . The presence of a nonzero value for the classical groundstate value of the field will have an effect when the field is quantized. The η massless particles are called Goldstone bosons. It is sufficient to do this to second order in the fields as the higher (cubic. 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. In this mode the lagrangian and the vacuum are both invariant under a set of symmetry transformations generated by Qa . etc. 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 . (12. ϕ2 and η. .23)     F +η the latter only forming a minimum for m2 < 0.21) Stability of the action requires the classical groundstate ϕc to have a well-defined value (which can be nonzero). It is only invariant under rotations around the 3-axis. . The classical groundstate appears degenerate. has less symmetry. while the lagrangian including the nonzero groundstate expectation value chosen by nature. It is appropriate to redefine the field as    ϕ1      φ =  ϕ2  . The original lagrangian contained an SO(3) invariance under (length conserving) rotations among the three fields. i. In the case of degeneracy. .25) 2 2 2 Therefore there are 2 massless scalar particles.) terms constitute interaction terms. 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. In this case the spectrum is described .

. i. (12. i.27) If the transformations in the lagrangian give rise to a symmetry the Noether currents are conserved. = ψ(i/ − M )ψ.. (12.32) (12. called flavors. down. T ] ψ. Using the Dirac equation. a massless Goldstone boson for each ’broken’ generator. ψ −→ ei α·T ψ. This means that they are operators that create states from the vacuum. Note that for the fields π a (x) one would have a the relation 0|π a (x)|π a (k) = ei k·x . V µ = ψγ µ T ψ.30) which in the case that we include only two flavors form an SU (2)V symmetry (isospin symmetry) generated by the Pauli matrices. 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. A bit more formal.e. Neglecting at this point the local color symmetry.31) . ∂ ∂ (12. it is easy to see that one gets ∂µ V µ = i ψ [M. Including a sum over the different flavors (up.e.26) for all the states labeled by a corresponding to ’broken’ generators. Known examples are rotational symmetry and the fact that the spectrum shows multiplets labeled by angular momentum ℓ (with members labeled by m). a 0|∂ µ Jµ (x)|π b (k) = fπ k 2 ei k·x δab = fπ m2 a ei k·x δab . irrespective of the fact if they annihilate the vacuum. Goldstone mode. This lagrangian density then is invariant under unitary (vector) transformations in the flavor space. In this mode the lagrangian is invariant but Qa |0 = 0 for a number of generators.28) where ψ is extended to a vector in flavor space and M is a diagonal matrix in flavor space.The standard model 120 by degenerate representations of the symmetry group.      ψu   mu           md     ψ =  ψd  . H] = 0. Taking the derivative. The conserved currents corresponding to this symmetry transformation are found directly using Noether’s theorem (see chapter 6). there must be a nonzero expectation value 0|Jµ (x)|π (k) .e.29) (Note that each of the entries in the vector for ψ in the above equation is still a 4-component Dirac spinor). the lagrangian for the quarks consists of the free Dirac lagrangian for each of the types of quarks. 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′ .  . a ∂ µ Jµ = 0. π Example: chiral symmetry An example of spontaneous symmetry breaking is chiral symmetry breaking in QCD. π (12. The generators Qa (in that case the rotation operators Lz . 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. a and a′ are degenerate states. As the a generators for a symmetry are precisely the zero-components of a conserved current Jµ (x) integrated a a over space. M =      . T = τ /2. if the generators Qa generate a symmetry. (12. strange. i. etc. [Qa . suggesting the stronger relation ∂ µ Jµ (x) = fπ m2 a π a (x). and one must have mπa = 0.    .) one can write L = f ψ f (i/ − Mf )ψf .    . denoted |π a (k) .

Therefore realization in the Weyl mode would require that all particles come double with positive and negative parity. ψ −→ ei α·T γ5 ψ. as the quantity ∂µ Aµ (x). This situation. proton and neutron. Aµ = ψγ µ T γ5 ψ. with almost degenerate masses (Mp = 938. is by nature not realized in the Weyl mode. That the massless theory has this symmetry can also be seen by writing it down for the socalled lefthanded and righthanded fermions. From the number of broken generators it is clear that one expects three massless Goldstone bosons.e. . but the fact that the symmetry is not perfect and the right hand side of Eq.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. or. The chiral fields ψR and ψL are transformed into each other under parity. which is equivalent 2 to the V-A symmetry. in terms of which the Dirac lagrangian 2 density looks like L = i ψL / ψL + i ψR / ψR − ψR M ψL − ψL M ψR . e. The currents corresponding to this symmetry transformation are again found using Noether’s theorem. How can we see this.35) In this case ∂µ Aµ = 0 will be true if the quarks have zero mass. Still the basic fact that the generators acting on the vacuum give a nonzero result (i. SU (2)V (isospin) symmetry is good if the up and down quark masses are identical.34) Using the Dirac equation. Note that these transformations also work on the spinor indices. (12. both are very small. etc.33) which in the case of two flavors form SU (2)A transformations generated by the Pauli matrices. behaves as a pseudoscalar particle (spin zero. socalled axial vector transformations. M = m · 1.35 is nonzero. 12. while the groundstate is only invariant under isospin rotations (R = L). T } γ5 ψ. ∂ ∂ (12. ψR/L = 1 (1 ± γ5 )ψ.e. (12.e. the lagrangian density is invariant under left (L) and right (R) rotations independently. T γ5 = τ γ5 /2. i. This symmetry is realized in the Weyl mode with the spectrum of QCD showing an almost perfect isospin symmetry. much smaller than any of the other mesons or baryons. T ] = 0. but also a triplet (isospin 1) of pions.The standard model 121 Furthermore ∂µ V µ = 0 ⇐⇒ [M. From group theory (Schur’s theorem) one knows that the latter can only be true. for which the field (according to the discussion above) has the same behavior under parity. The Goldstone bosons of QCD are the pions for which fπ = 93 MeV and which have a mass of mπ ≈ 138 MeV/c2 .e. fπ = 0 remains. is what happens in the real world.e. a doublet (isospin 1/2) of nucleons. etc.6 MeV/c2 ). chiral symmetry is not perfect.g. This symmetry. parity minus). i. if in flavor space M is proportional to the unit matrix. stated equivalently. nature has choosen the path SU (2)L ⊗ SU (2)R =⇒ SU (2)V . 12. where the quark masses are not completely zero. 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. This combined symmetry is what one calls chiral symmetry. Therefore the world of up and down quarks describing pions. In the real world. (leaving out the flavor structure) the same as ψγ5 ψ. We know that mesons and baryons (such as the nucleons) have a well-defined parity that is conserved. I. (12. which is approximately true for the up and down quarks.3 MeV/c2 and Mn = 939. i. The conclusion is that the original symmetry of the lagrangian is spontaneously broken and as the vector part of the symmetry is the well-known isospin symmetry. however. generated by TR/L = 1 (1 ± γ5 )T . parity would not play a role in the world. it is easy to see that one gets ∂µ Aµ = i ψ {M. gives also rise to a nonzero mass for the Goldstone bosons according to Eq. including in Dirac space a γ5 matrix. There exists another set of symmetry transformations.

39) (12. . thus 9 independent degrees of freedom.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.     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 + .38) ∂µ Wν − ∂ν Wµ + g Wµ × Wν . is made. The difference comes when we reparametrize the field φ. Initially there are 3 massless gauge fields (each. in this case three-dimensional) representation reads (La )ij = −i ǫaij . The latter is a spin 0 particle (real scalar field) called a Higgs particle.37) (12.   The symmetry is broken in the same way as before and the same choice for the vacuum.    0      ϕc = 0|φ |0 =  0  . . while the number of massive gauge bosons coincides with the number of ’broken’ generators. 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.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     . two massive bosons (Wµ and Wµ with MW = gF ) and a massive scalar particle (η with m2 = η 2 −2 m . like a photon. having two independent spin components) and three scalar fields (one degree of freedom each). 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. Dµ φ = Gµν = ∂µ φ + g Wµ × φ. After symmetry breaking the same number (as expected) comes out. . i. one sees that the fields Wµ and Gµν also can be represented as three-component fields.     0  0         0 = .40) a Dµ φ = ∂µ φ − ig Wµ La φ.    0 φ= (12.42) 2 from which one reads off that the particle content of the theory consists of one massless gauge boson 3 1 2 (Wµ ). again 9 degrees of freedom. One may wonder about the degrees of freedom. but one has 1 massless gauge field (2). (12. made into a gauge theory. as in this case there are no massless Goldstone bosons. (12. 4 2 where Since the explicit (adjoint. 1 1 L = − Gµν · Gµν + Dµ φ · Dµ φ − V (φ).The standard model 122 12.e. 2 massive vector fields or spin 1 bosons (2 × 3) and one scalar field (1). Considering the same example with rotational symmetry (SO(3)) as for spontaneous symmetry breaking of a scalar field (Higgs field) with three components.

R. The lagrangian for the leptons consists of three families each containing an elementary fermion (electron e− . R −→ e−i β R. 1 As they are massless.The standard model 123 12. 2 (12. that gives a good description of the physical world. Thus the lagrangian density is L (f ) ∂ ∂ = i L/ L + i R/ R.44) which has an SU (2)W symmetry under transformations eiα·TW . (12. explicitly L R SU(2)W SU(2)W −→ ei α·τ /2 L. its corresponding neutrino (νe . For the neutrino only a lefthanded particle (and righthanded antiparticle) exist. the leptons (elektrons.) and the gauge bosons responsible for the strong. U(1)Y U(1)Y (12. νµ and ντ ) and their antiparticles. while the lefthanded particles form a doublet. 2 2 = ∂µ R + i g ′ Bµ R. muons. explicitly Dµ L Dµ R i i g Wµ · τ L + g ′ Bµ L. This is written as 3 Q = TW + YW . (12. τ ). ∂ ∂ ∂ (12. = ∂µ L − (12. electromagnetic and weak forces.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 /2.45) (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.50) (12. the quarks (up.). left.43) and One introduces a (weak) SU (2)W symmetry under which eR forms a singlet. neutrinos.48) (12. down.51) . under which the lefthanded and righthanded particles with YW (L) = −1 and YW (R) = −2 transform with eiβYW /2 .   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 = 0.47) and YW is considered as an operator that generates a U (1)Y symmetry. ψR/L = 2 (1 ± γ5 )ψ decouple. The procedure.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.e. By introducing gauge fields and breaking the symmetry a more complex lagrangian is obtained. however. Thus L (f ) = i eR / eR + i eL / eL + i νeL / νeL + (µ. etc. implies certain nontrivial relations between masses and mixing angles that can be tested experimentally and sofar are in excellent agreement with experiment. i.and righthanded particles. In the standard model one starts with a very simple basic lagrangian for (massless) fermions which exhibits more symmetry than observed in nature. explicitly L −→ e−i β/2 L. muon µ− or tau τ − ). etc.

+ 8 (12.58) (12.54) = (Dµ φ)† (Dµ φ) −m2 φ† φ − λ (φ† φ)2 .The standard model 124 3 3 where Wµ is a triplet of gauge bosons with TW = 1. −V (φ) L and (h2) = −Ge (LφR + Rφ L). a complex Higgs field   +   1  φ   √2 (θ2 + iθ1 )    0 = 1   (12.53) φ=   √ (θ4 − iθ3 )  φ 2 with TW = 1/2 and YW = 1 is introduced. . † i i g Wµ · τ − g ′ Bµ )φ.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. Putting this in leads to L (f ) =L (f 1) +L (f 2) . 2 and 3 may be eliminated (the necessary SU (2)W rotation is precisely e−iθ(x)·τ ). this leads to the lagrangian L (h1)     .   h)  (12. (12. . + 8 g2 v2 1 + − (Wµ )2 + (Wµ )2 (∂µ h)2 + m2 h2 + 2 8 (g 2 + g ′2 ) v 2 (Zµ )2 + . the U (1)Q symmetry (generated by the charge operator). leading to the parametrization   1  0    φ(x) = √  (12. .56)  v + h(x) 2 Dµ φ = (∂µ − and     Dµ φ =     = 1 √ 2 − ig 2 ∂µ h + 1 2 Wµ −iWµ √ (v + h) 2 3 gWµ −g′ Bµ i √ (v + 2 2 Up to cubic terms.57) = 1 g2 v2 1 2 (Wµ )2 + (Wµ )2 (∂µ h)2 + m2 h2 + 2 8 v2 2 3 gWµ − g ′ Bµ + . which assures with the choice of YW for the Higgs field that Q generates the remaining U (1) symmetry. . with the following lagrangian density consisting of a symmetry breaking piece and a coupling to the fermions.52) L L (f 1) (f 2) = = i Rγ µ (∂µ + ig ′ Bµ )R + i Lγ µ (∂µ + i ′ i g Bµ − g Wµ · τ )L 2 2 1 1 − (∂µ Wν − ∂ν Wµ + g Wµ × Wν )2 − (∂µ Bν − ∂ν Bµ )2 . Using local gauge invariance θi for i = 1. 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. 4 4 In order to break the symmetry to the symmetry of the physical world. (12. For the classical field the choice θ4 = v is made. (12. L where L (h1) (h) =L (h1) +L (h2) . .59) .

(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 .65) The weak mixing angle is related to the ratio of coupling constants. σ (νµ γ ρ µL ) 2 2 (eL γ νe ) 2 k + MW (12.The standard model 125 where the quadratically appearing gauge fields that are furthermore eigenstates of the charge operator are ± Wµ = = = Zµ Aµ 1 1 2 √ Wµ ± i Wµ . we can read off e = g sin θW = g ′ cos θW .66) From the coupling to the photon. . 2 3 g Wµ − g ′ Bµ 3 ≡ cos θW Wµ − sin θW Bµ . . 2 Zµ (12.60) (12. 2 + g ′2 g 3 g ′ Wµ + g Bµ (12.68) g2 ρ 5 2 (νµ γρ (1 − γ5 )µ) (eγ (1 − γ )νe ) 8 MW (µ)† (jL )ρ (e) (jL )ρ GF (µ)† (e) ≡ i √ (jL )ρ (jL )ρ .61) (12. for instance in the amplitude for the decay of the muon νµ µ− W − e− νµ = µ− e− νe νe is given by −i M = − ≈ i g2 −i gρσ + . 2θ 2θ 4 cos W cos W 0.64) (12. g ′ /g = tan θW .69) .62) g 2 + g ′2 3 ≡ sin θW Wµ + cos θW Bµ .67) The coupling of electrons or muons to their respective neutrinos. 2 (12.63) (12. and correspond to three massive particle fields (W ± and Z 0 ) and one massless field (photon γ) with 2 MW 2 MZ 2 Mγ = = = g2 v2 . 4 2 MW g2 v2 = . (12.

19 GeV. g ′ and v determine a number of experimentally measurable quantities.231 2. from the MW mass is about 250 GeV. 2 The value of α deviates from the known value α ≈ 1/137 because of higher order contributions.e. leptons with CV = −1/2 + 2 sin2 θW ≈ −0. With the choosen vacuum expectation value for the Higgs field. Comparing this with the total width into (invisible!) channels.40 GeV. one obtains L (h2) Ge v Ge = − √ (eL eR + eR eL ) − √ (eL eR + eR eL ) h 2 2 me = −me ee − eeh. giving rise to a running coupling constant after renormalization of the field theory.73) (12. 3 = TW .g. (12. 91. the mass of the electron comes from the spontaneous symmetry breaking but is not predicted (it is in the coupling Ge ). . For each neutrino species (with CV = 1/2 and CA = 1/2 one expects Γ(νν) ≈ 155 MeV. (12. The coupling to the Higgs particle is weak as the value for v calculated e.80) First. (12. i.74) (12. (12.5 MeV (exp.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 . such as2 α = e2 /4π GF sin2 θW MW MZ ≈ 1/127.72) (12. The masses of the fermions and the coupling to the Higgs particle are contained in L (h2) . g2 MZ 2 2 (CV + CA ). Γe ≈ Γµ ≈ Γτ ≈ 83 MeV).79) Γ(Z 0 → f f ) = 48π cos2 θW For the electron.71) GeV −2 = = = = .The standard model 126 the good old four-point interaction introduced by Fermi to explain the weak interactions. Finally we want to say something about the weak properties of the quarks.70) = 2 = 2 sin2 θ 8 MW 2 v2 8 MW 2 W In this way the parameters g. one has the relation g2 e2 1 GF √ = . i.77) (12. v (12.05 and CA = −1/2 we calculate Γ(e+ e− ) ≈ 78. 2 2 2 (12. as appear for instance in the decay of the neutron or the decay of the Λ (quark content uds). me /v is extremely small. muon or tau.166 4 × 10 0. Γinvisible = 480 MeV one sees that three families of (light) neutrinos are allowed. 80. −5 (12.76) with CV CA 3 = TW − 2 sin2 θW Q.e.78) From this coupling it is straightforward to calculate the partial width for Z 0 into a fermion-antifermion pair. 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. 1.

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

u u where we include now the three lefthanded quark doublets in QL .The standard model 128 using the socalled Wolfenstein      + O(λ4 )   with λ ≈ 0.83) where Vu and Vd are unitary matrices.   φ =    −φ− 0 2 For the (squared) complex matrices we can find positive eigenvalues. † Λu Λ† = Vu G2 Vu . i. The imaginary part i η gives rise to CP violation in decays of K and B-mesons (containing s and b quarks. A ≈ 0. (12. The Higgs field is still limited to one complex doublet. each of these containing   the three families.84) with Gu and Gd being real and positive and Wu and Wd being different unitary matrices.22 and η ≈ 0. respectively).85) √ √ with Mu = Gu v/ 2 (diagonal matrix containing mu . mc and mt ) and Md = Gd v/ 2 (diagonal matrix containing md .g. 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. d d (12. allowing us to write † Λu = Vu Gu Wu † and Λd = Vd Gd Wd . Thus one has † † † =⇒ −DL Vd Md Wd DR − DR Wd Md Vd† DL − UL Vu Mu Wu UR − UR Wu Mu Vu UL (12. UR = uR cR tR . 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 .q) ′ weak † mass mass DL = DL = Vu Vd DL = VCKM DL . (12.227. e.q) = −QL φΛd DR − DR Λ† φ† QL − QL φc Λu UR − UR Λ† φc† QL u d (12.e. . Note that we need the conjugate Higgs field to get a U (1)Y singlet in the case of the charge +2/3 quarks. ms and mb ). the three righthanded quarks with charge +2/3 in UR and the three righthanded quarks with charges −1/3 in DR . 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 .82) space. for which we need the appropriate weak isospin doublet  0∗    1    φ c  = √  v +h . the weak mass eigenstates are L (h2.34.86) the unitary CKM-matrix introduced above in an ad hoc way. The Λu and Λd are complex matrices in the 3×3 family and Λd Λ† = Vd G2 Vd† .5 Family mixing in the Higgs sector and neutrino masses The quark sector Allowing for the most general (Dirac) mass generating term in the lagrangian one starts with L (h2.82 and ρ ≈ 0.

90) mass † mass mass † mass † We note that there are mass fields ER = We ER . (12. (12. For neutrino’s this matrix is parametrized in terms of three angles θij with cij = cos θij and sij = sin θij and one angle δ. decoupling from all known interactions. one can write Λe = Ve Ge We and we find † L (h2.93)   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    . The lepton sector (massive Dirac neutrinos) In principle a massive Dirac neutrino could be accounted for by a lagrangian of the type L (h2. The mass fields ER † mass † mass = We ER . 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 . † † 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 ). The W -current then becomes mass mass E L γ µ NL = E L γ µ NL . (12.ℓ) =⇒ −Me EL Ve We ER − ER We Ve† EL .ℓ) † † † =⇒ −EL Ve Me We ER − ER We Me Ve† EL − NL Vn Mn Wn NR − NR Wn Mn Vn NL . while θ23 then is determined by atmospheric neutrino oscillations requiring |∆m2 | ≈ 2.87) is a weak doublet containing the three families of neutrinos (NL ) and charged leptons (EL ) and ER is † a three-family weak singlet.The standard model 129 The lepton sector (massless neutrinos) For a lepton sector with a lagrangian density of the form L in which (h2. In this case.ℓ) = −LφΛe ER − ER Λ† φ† L − Lφc Λn NR − NR Λ† φc† L e n (12. i. EL = Ve EL . UPMNS a parametrization that in principle also could have been used for quarks.91) † with UPMNS = Vn Ve known as the Pontecorvo-Maki-Nakagawa-Sakata mixing matrix. since the weak current is the only place where they show up. NL = Vn NL and NR = Wn NR µ mass γ µ V † V N mass .ℓ) = −LφΛe ER − ER Λ† φ† L. there is no family mixing for massless neutrinos. we find massless neutrinos. EL = Ve† EL .88) √ mass with Me = Ge v/ 2 the diagonal mass matrix with masses me .5×10−3 eV2 .92) .89) with three righthanded neutrinos added to the previous case.  =  − 1/6 2/3 0           1/2 1/2 0 0 1 0 1/3 1/2 − 1/6 (12.e. As before. and obtain L (h2. The mixing is intriguingly close to the Harrison-Perkins-Scott tri-bimaximal 23 mixing matrix      0 0 2/3 1/3 0 2/3 1/3 0     1                 1/2 − 1/2   − 1/3 UHPS =  0 1/3 − 1/2  .    (12. For the (massless) neutrino fields we just can redefine fields into NL = Ve† NL . e   N L= L  EL     (12. mµ and mτ .

99) . is to add in Eq.95) 2 In order to have more than a completely decoupled sector. the massless and massive Majorana neutrinos. however. 1 ∗ c c (12. We use here the primes starting with the weak eigenstates.and lefthanded species through Dirac mass terms coming from the coupling to the Higgs sector as in the previous section. 12. Υc ≡ (ΥL )c = ΥR 1 2 L and Υc = ΥL ). Υ′ and Υ′ (satisfying Υc = Υ and as discussed in chapter 6.c. For the leptons. Absorption of phases in the states is not possible for Majorana neutrinos.The standard model 130 The lepton sector (massive Majorana fields) An even simpler option than sterile righthanded Dirac neutrinos. This is called the see-saw mechanism (outlined below). coupled by a Dirac mass term.and righthanded species then just form a Dirac fermion.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. with for the righthanded sector a mass term MR . −    MD MR nR 2 (12. one must for the neutrinos as well as charged leptons. 1 c ∗ c L mass. 1 L Υ ′ = nR + nc . If the Majorana mass MR ≫ MD one actually obtains in a natural way one Majorana neutrino with a very small mass. (12. a unitary matrix relating them to the weak eigenstates. α2 and δ). The mass matrix can be written as   ML |MD | eiφ     M = (12. as above.96) VPMNS = UPMNS K with K =  0 eiα2 /2 0  . Thus 2 Υ ′ = nc + nL .94) M L NL NL + M L NL NL .98) (12. For these light Majorana neutrinos one has. For the charged leptons there cannot exist a Majorana mass term as this would break the U(1) electromagnetic symmetry. hence the mixing matrix becomes   iα /2 0 0   e 1     (12. The way to circumvent this is to introduce as in the previous section righthanded neutrinos. 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 =− 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. couple the right.ν = − MR NR NR + MR NR NR . Actually. the left.       0 0 1 L mass. 2 R ψ = nR + nL . (12. The see-saw mechanism Consider (for one family N = n) the most general Lorentz invariant mass term for two independent Majorana spinors. 2 although this option is not attractive as it violates the electroweak symmetry. Thus (disregarding family structure) one has two Majorana neutrinos. are equivalent to two decoupled Majorana neutrinos (see below).88 a Majorana mass term for the (lefthanded) neutrino mass eigenstates. For the neutrino sector. one being massive.

a and a′ are degenerate states.  L      Υ2L nR Υ1 and Υ2 . This is a mixing problem with a symmetric (complex) mass matrix leading to two (real) mass eigenstates. 2 with real masses Mi .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   . i. T ] ψ.2 Derive for the vector and axial vector currents.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)) . i ψ {M. . V µ = ψγ µ T ψ and Aµ = ψγ µ γ5 T ψ ∂µ V µ ∂µ A µ = = i ψ [M. (12.  (12. T } γ5 ψ.The standard model 131 taking ML and MR real and non-negative.e. Prove that if the generators Qa generate a symmetry.      Υ2R nR (12. [Qa . (12. i.e. Exercise 12.105) for i = 1. This choice is possible without loss of generality because the phases can be absorbed into Υ′ and Υ′ (real must be replaced by hermitean if one includes 1 2 families). The diagonalization is analogous to what was done for the Λ-matrices and one finds U M U T = M0 with a (unitary) matrix U . related via    c   Υ1R     = U  nL  . Exercise 12.1 Consider the case of the Weyl mode for symmetries. 2 U (M M † ) U † = M0 . H] = 0. Exercises Exercise 12.3 (a) The coupling of the Z 0 particle to fermions is described by the vertex −i g 2 cos θW f f CV γ µ − CA γ µ γ5 . which implies U ∗ M † U † = U ∗ M ∗ U † = M0 and a ’normal’ diagonalization of the (hermitean) matrix M M † . For the situation ML = 0 and MR ≫ MD (taking MD real) one 2 finds M1 ≈ MD /MR and M2 ≈ MR .103) for each of which one finds the lagrangians L = and we are left with two decoupled Majorana fields      Υ1L    = U ∗  nc  . 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′ .104) ↔ 1 1 Υi i / Υi − Mi Υi Υi ∂ 4 2 (12.

to be precise one has for bosons (b) or fermions (f). Neglect the masses of fermions and use the fact that the sum over polarizations is 3 pµ pν . ghbb = b v v where v is the vacuum expectation value of the Higgs field. Γ(Z 0 → e+ e− ).) are proportional to the mass squared for bosons or mass for fermions of the particles. and the width to a pair of neutrino’s. (b) Calculate the width to electron-positron pair.231. In the standard model this coupling is described by the vertex −i g µ √ (γ − γ µ γ5 ) . 3 TW . mb ≈ 5 GeV and MW ≈ 80 GeV. Exercise 12. hhh. the weak mixing angle is given by ¯ sin2 θW = 0. e+ e− h. Exercise 12. Note that you can find the answer without explicit construction of all interaction terms in the lagrangian. etc.4 Calculate the lifetime τ = 1/Γ for the top quark (t) assuming that the dominant decay mode is t → b + W +. Γ(Z 0 → νe νe ). Exercise 12.The standard model 132 with CV CA = = 3 TW − 2Q sin2 θW .6 Check that the use of the Wolfenstein parametrization in the CKM matrix indeed gives a unitary matrix. . 0 ¯ Γ(Z 0 → f f ) = MZ g2 48π cos2 θW f f CV 2 + CA 2 . Mf M2 and ghf f = .7 In this exercise two limits are investigated for the two-Majorana case. 2 2 The masses are mt ≈ 175 GeV. Write down the matrix element squared (averaged over initial spins and summed over final spins) for the decay of the Z 0 . Exercise 12. 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. ZZh.5 Show that the couplings to the Higgs (for W + W − h. at least up to a high (which?) order in λ.

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

Sign up to vote on this title
UsefulNot useful

Master Your Semester with Scribd & The New York Times

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

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